Tracking Animal Migration With Stable Isotopes

TRACKING ANIMAL MIGRATION WITH
STABLE ISOTOPES
This page intentionally left blank
TRACKING ANIMAL
MIGRATION WITH
STABLE ISOTOPES
SECOND EDITION
Edited by
Keith A. Hobson
Environment and Climate Change Canada, Saskatoon, SK, Canada
Leonard I. Wassenaar
International Atomic Energy Agency, Vienna, Austria
Academic Press is an imprint of Elsevier
125 London Wall, London EC2Y 5AS, United Kingdom
525 B Street, Suite 1650, San Diego, CA 92101, United States
50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom
Copyright r 2019 Elsevier Inc. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopying, recording, or any information storage and retrieval system, without permission in writing from
the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our
arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be
found at our website: www.elsevier.com/permissions.
This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as
may be noted herein).
Notices
Knowledge and best practice in this field are constantly changing. As new research and experience broaden our
understanding, changes in research methods, professional practices, or medical treatment may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any
information, methods, compounds, or experiments described herein. In using such information or methods they should
be mindful of their own safety and the safety of others, including parties for whom they have a professional
responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any
injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or
operation of any methods, products, instructions, or ideas contained in the material herein.
British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress
ISBN: 978-0-12-814723-8
For Information on all Academic Press publications

visit our website at https://www.elsevier.com/books-and-journals
Publisher: Andre Gerhard Wolff

Acquisition Editor: Anna Valutkevich
Editorial Project Manager: Ruby Smith
Production Project Manager: Sreejith Viswanathan
Cover Designer: Ewaryst Izewski
Typeset by MPS Limited, Chennai, India
Contents
List of Contributors vii 3.2 Process 54

3.3 Pattern 58
Preface ix
3.4 Mapping Isoscapes 63
Preface to the First Edition xi 3.5 Isoscapes for Terrestrial Migration Research 69
Acknowledgments xiii 3.6 Summary and Look Forward 76
Acknowledgment 78
References 78
1. Animal Migration: A Context for Using
New Techniques and Approaches 4. Application of Isotopic
KEITH A. HOBSON, D. RYAN NORRIS, KEVIN J. KARDYNAL
AND ELIZABETH YOHANNES
Methods to Tracking Animal Movements
KEITH A. HOBSON
1.1 Introduction 1
1.2 Migratory Populations, Connectivity, 4.1 Introduction 85
and Conservation 5 4.2 Toward Isotopic Assignment of Origins 86
1.3 Scientific Tools Used to Study Migration 8 4.3 Movements Inferred Without Isoscapes 95
1.4 Technical Advances and Outlook 16 4.4 Using Isoscapes 98
References 17 4.5 Challenges 105
Further Reading 23 4.6 Summary 107
Acknowledgment 108
References 108
2. Introduction to Conducting Stable Further Reading 115
Isotope Measurements for Animal Migration
Studies 5. Tracking of Movements
LEONARD I. WASSENAAR of Terrestrial Mammals Using
Stable Isotopes
2.1 Introduction 25
CHRISTIAN C. VOIGT AND LINN S. LEHNERT
2.2 Sample Collection and Preparative
Methods 30 5.1 Introduction 117
2.3 Global-Spatial Isotopes 38 5.2 Stable Isotopes and Movements of Terrestrial
2.4 Local-Spatial Isotopes 44 Mammals 121
2.5 Conclusions 48 5.3 Isotopic Retention in Tissues: The Retrospective
Acknowledgments 48 Moving Time Window 123
References 48 5.4 Application of Stable Isotopes to the Study
of Migratory Movements 126
3. Isoscapes for Terrestrial Migration 5.5 Future Directions 131
Research 5.6 Summary 132
GABRIEL J. BOWEN AND JASON B. WEST Acknowledgments 132
References 132
3.1 Introduction 53 Further Reading 135
v
vi CONTENTS
6. Isotopic Tracking of Marine Animal 8.5 Assignment Model Types 203

Movement 8.6 Conclusion 205
References 205
CLIVE N. TRUEMAN AND KATIE ST JOHN GLEW
6.1 Introduction 137 9. Isoscape Computation and Inference

6.2 Part 1: Mechanism, Spatial Structure, and of Spatial Origins With Mixed Models
Isoscapes 139 Using the R package IsoriX
6.3 Part 2: Examples of Isotopes and Isoscapes for
ALEXANDRE COURTIOL, FRANÇOIS ROUSSET,
Marine Migration Research 156 MARIE-SOPHIE ROHWÄDER, DAVID X. SOTO,
6.4 Computer Modeling and Simulation 162 LINN S. LEHNERT, CHRISTIAN C. VOIGT, KEITH A. HOBSON,
6.5 Conclusion 166 LEONARD I. WASSENAAR AND STEPHANIE KRAMER-SCHADT
References 167
9.1 Introduction 207
9.2 What Is IsoriX? 208
7. Amino Acid Isotope Analysis:
9.3 The IsoriX Package 210
A New Frontier in Studies of Animal 9.4 The IsoriX Workflow 214
Migration and Foraging Ecology 9.5 The Future of IsoriX 230
KELTON W. MCMAHON AND SETH D. NEWSOME Acknowledgments 232
References 232
7.1 Introduction 173 Appendix: Statistical Framework 234
7.2 Primer on Amino Acid Biochemistry
and Isotope Discrimination 174 10. Outlook for Using Stable Isotopes
7.3 Accounting for Consumer Physiology 178
7.4 Case Studies in Movement and Foraging
in Animal Migration Studies
Ecology Using CSIA-AA 180 KEITH A. HOBSON, LEONARD I. WASSENAAR,
GABRIEL J. BOWEN, ALEXANDRE COURTIOL,
7.5 CSIA-AA Methodology 184 CLIVE N. TRUEMAN, CHRISTIAN C. VOIGT,
7.6 Summary and Future Work 185 JASON B. WEST, KELTON W. MCMAHON
Acknowledgment 186 AND SETH D. NEWSOME
References 187
10.1 Samples and Isotopic Analyses 238
8. Design and Analysis for Isotope-Based 10.2 Migratory Systems 239
10.3 Isoscapes and Assignment 241
Studies of Migratory Animals 10.4 Linking Stable Isotopes to Other Spatial
MICHAEL B. WUNDER AND D. RYAN NORRIS Markers 242
10.5 Research in Support of Conservation 242
8.1 Introduction 191
References 243
8.2 Planning Your Study 192
8.3 Sampling Considerations 195
8.4 Data Analysis and Modeling Considerations 200 Index 245
List of Contributors
Gabriel J. Bowen University of Utah, Salt Lake Marie-Sophie Rohwäder Leibniz Institute for Zoo
City, UT, United States and Wildlife Research, Berlin, Germany
Alexandre Courtiol Leibniz Institute for Zoo and François Rousset Université de Montpellier,
Wildlife Research, Berlin, Germany Montpellier, France
Keith A. Hobson Environment and Climate Change David X. Soto KU Leuven, Leuven, Belgium
Canada, Saskatoon, SK, Canada; University of Katie St John Glew University of Southampton,
Western Ontario, London, ON, Canada Southampton, United Kingdom
Kevin J. Kardynal Environment and Climate Clive N. Trueman University of Southampton,
Change Canada, Saskatoon, SK, Canada Southampton, United Kingdom
Stephanie Kramer-Schadt Leibniz Institute for Christian C. Voigt Leibniz Institute for Zoo and
Zoo and Wildlife Research, Berlin, Germany Wildlife Research, Berlin, Germany; Freie
Linn S. Lehnert Leibniz Institute for Zoo and Universität Berlin, Berlin, Germany
Wildlife Research, Berlin, Germany; Freie Leonard I. Wassenaar International Atomic
Universität Berlin, Berlin, Germany Energy Agency, Vienna, Austria
Kelton W. McMahon University of Rhode Island, Jason B. West Texas A&M University, College
Narragansett, RI, United States Station, TX, United States
Seth D. Newsome University of New Mexico, Michael B. Wunder University of Colorado
Albuquerque, NM, United States Denver, Denver, CO, United States
D. Ryan Norris University of Guelph, Guelph, Elizabeth Yohannes University of Konstanz,
ON, United States; University of Guelph, Konstanz, Germany
Guelph, ON, Canada
vii
Preface
The use of stable isotopes in ecological and advance our understanding of animal migra-
biological sciences is nowadays routine and tion phenomenon.
often an integral component of several large- Key changes in this second edition include:
scale ecological and interdisciplinary environ-
• Review of current and new isotopic
mental studies. Stable isotopic methods are
methods and remote analyses (Chapter 1:
well established and invaluable for investigat-
Animal Migration: A Context for Using
ing the ecology of migrant individuals and
New Techniques and Approaches,
populations.
Chapter 2: Introduction to Conducting
Animal migration is a compelling area of
Stable Isotope Measurements for Animal
evolutionary and ecological research, and
Migration Studies, and Chapter 7: Amino
understanding the movement patterns in ani-
Acid Isotope Analysis: A New Frontier in
mals has become a serious topic of concern as
Studies of Animal Migration and Foraging
we struggle to conserve threatened and other
Ecology)
species that move across geopolitical bound-
• New computerized isoscape and isotope
aries. Traditional applications of tagging tech-
assignment approaches (Chapter 3:
niques have met with little success except for
Isoscapes for Terrestrial Migration Research
only a few larger and conspicuous species that
and Chapter 9: Isoscape Computation and
can be intercepted with high probability. For
Inference of Spatial Origins With Mixed
most migratory birds, mammals, fishes, and
Models Using the R package IsoriX)
insects, fundamental internal markers are
• New chapter on marine animal migration
required, and among these, stable isotopes
(Chapter 6: Isotopic Tracking of Marine
show great promise if used appropriately, and
Animal Movement)
with other tools.
• New chapter on terrestrial mammal
In this completely revised second edition of
migration (Chapter 5: Tracking of
Tracking Animal Migration With Stable Isotopes,
Movements of Terrestrial Mammals Using
we focus on the topic of using stable isotopes
Stable Isotopes)
in unraveling terrestrial animal migration
• Updated case studies and comprehensive
at various spatial scales around the globe.
review of contemporary literature (all the
However, since the first edition 10 years ago,
chapters)
there have been numerous technological
advances and field studies that have provided The aim of this volume is to provide a foun-
a growing and rich body of knowledge to dational handbook that will serve to introduce
ix
x PREFACE
students, professional ecologists, and biologists Ecological Research (1989). In addition, some
interested in terrestrial animal migration to websites are also of interest, primarily via elec-
the key aspects of measuring, applying, and tronic forum for researchers using isotopes in
interpreting stable isotopes for terrestrial and migration. These include the biannual
marine migratory systems. Applications in Stable Isotope Ecology interna-
From an educational perspective, there are tional scientific meetings (www.isoecol.org)
still few textbooks available for biologists and where ecologists often present findings on
ecologists to provide an introductory overview migratory research, as well as the discussion
of stable isotopes. Several textbooks include B. list sever Isogeochem (www.isogeochem.com).
Fry, Stable Isotope Ecology (2006); R. Michener Our hope is that the foundational aspects of
and K. Lajtha, Stable Isotopes in Ecology and migration and isotopes reviewed here provide a
Environmental Science (2007); Dawson and compelling research stimulus for spawning
Seigwolf, Stable Isotopes as Indicators of new ideas and research into the fascinating topic
Ecological Change (2007), and outdated bench- and mysteries of animal migration in terrestrial
mark books like P.W. Rundell, Stable Isotope in and aquatic environments into the future.
Preface to the First Edition
The use of stable isotopes in the ecological migratory systems. The reader will very
and biological sciences is a rapidly growing quickly appreciate that this topic has not
area of research that was historically con- recently developed in isolation, but draws
strained within the domains of the earth and from many decades of foundational
geological sciences. Today, stable isotope mea- stable isotope research in the geological,
surements are becoming a routine, wide- hydrological, biological, and statistical
spread, and integral component of many sciences. A volume like this would not have
large-scale ecological and interdisciplinary been possible a few decades ago.
environmental studies. Isotopic methods are From an educational perspective, until
invaluable for investigating the ecology of recently there were no textbooks available for
individuals and populations. biologists and ecologists that provided a good
Here we focus on the very specific topic of overview of stable isotopes. Fortunately sev-
using stable isotopes in unraveling terrestrial eral textbooks have now appeared, and newco-
animal migration at various spatial scales mers to the field of stable isotopes are also
around the globe. Migration is a compelling encouraged to consider reading books by B.
area of evolutionary and ecological research Fry, Stable Isotope Ecology (2006); R. Michener
and understanding the movement patterns in and K. Lajtha, Stable Isotopes in Ecology and
animals has become a topic of concern as we Environmental Science (2007), T. Dawson and R.
struggle to conserve threatened and other spe- Seigwolf, Stable Isotopes as Indicators of
cies that move across geopolitical boundaries. Ecological Change (2007) in addition to older
Previous traditional applications of tagging benchmark books like P.W. Rundell,
techniques have met with little success except Stable Isotope in Ecological Research (1989). In
for only a few larger and conspicuous species addition, some internet websites are also of
that can be intercepted with high probability. interest, primarily as an electronic forum for
For the vast majority of migratory birds, mam- researchers using isotopes in migration. These
mals, and insects, many fundamental internal include 8220Migrate8221 (www.migrate.ou.
markers are required, and among these, edu) and the biannual 8220Applications in
stable isotopes show great promise if used Stable Isotope Ecology8221 international scien-
appropriately. tific meetings (www.isoecol.org).
The aim of this volume is to provide a col- Our hope is that the foundations laid here
lege- and graduate-level handbook that will will provide a research stimulus for spawning
serve to introduce ecologists and biologists new ideas and research into the fascinating
interested in terrestrial animal migration to the topic and mysteries of animal migration in ter-
key elements of measuring, applying, and restrial and aquatic environments into the
interpreting stable isotopes in terrestrial future.
xi
Acknowledgments
The second edition book cover art is and excellent reviews. We thank Stefan Terzer
designed by Ewaryst Izewski, and conveys the and David Soto who conducted external chap-
concept of global spatial migratory move- ter reviews for us.
ments, isoscapes, and animal diversity in ter-
restrial and aquatic ecosystems. We thank all Keith A. Hobson, London, Canada
the chapter authors for their comprehensive Leonard I. Wassenaar, Vienna, Austria
xiii
C H A P T E R
1
Animal Migration: A Context for Using
New Techniques and Approaches
Keith A. Hobson1, D. Ryan Norris2, Kevin J. Kardynal1
and Elizabeth Yohannes3
1
Environment and Climate Change Canada, Saskatoon, SK, Canada, 2University of Guelph, Guelph,
ON, Canada, 3University of Konstanz, Konstanz, Germany
1.1 INTRODUCTION Some key proximate questions related to

animal migratory movement include: How
The spatiotemporal movement of organisms do these individuals know where they are
determines their interaction with their environ- going? How do they cope with the tremendous
ments and, therefore, comprises a fundamental physical demands of travel? How do they
aspect of their ecology and evolutionary life adjust to unfamiliar or changing environments
history. The degree to which organisms move along the way to their destinations? What are
also characterizes the range of resources they the costs and benefits of migration? How do
encounter, the array of hazards they experi- migratory patterns influence long-term popu-
ence from predators to hurricanes, and the lation dynamics? Ultimately, the evolution of
number of organisms with which they may migratory movements remains among the
interact. For animals, movement is very much most fascinating and challenging to understand.
a question of geospatial scale. While some spe- Finding answers to these fundamental ques-
cies may wander nomadically over a landscape tions has proven to be an immense scientific
of a few square meters for their entire lives, challenge. A large part of this challenge is
others travel across thousands of kilometers in related to the limited tools available to research-
regular movements that constitute some of the ers to infer or determine large- and small-scale
most spectacular natural phenomena on the animal movements. This volume explores recent
planet. These migrations are the movements developments in stable isotope techniques
that have captured the interest of researchers, which have contributed tremendously to the
citizen scientists, and laypersons alike and field of understanding animal movement, dis-
leave us with a true sense of wonderment. persal and long-distance migration in terrestrial
Tracking Animal Migration with Stable Isotopes

DOI: https://doi.org/10.1016/B978-0-12-814723-8.00001-5 1 © 2019 Elsevier Inc. All rights reserved.
2 1. ANIMAL MIGRATION: A CONTEXT FOR USING NEW TECHNIQUES AND APPROACHES
and marine ecosystems. As we will demon- especially birds, over vast distances but
strate, stable isotope tools are not only a critical involves many taxa (Fig. 1.1). The Arctic Tern
component of the larger toolbox but can typi- (Sterna paradisaea) migrates from breeding
cally augment other sources and forms of grounds in the Arctic to wintering grounds in
information. the Antarctic, an annual round trip of a stag-
The term migration often evokes images of gering 40,000 km (Egevang et al., 2010), equiv-
spectacular seasonal movements of animals, alent to 2 years of day-to-day driving by the
Jan
(A) (B) Dec Feb
Nov Mar
Oct Apr
Sep May
Aug Jun
Jul
Jan
Dec Feb
Nov Mar
Oct Apr
Sep May
Aug Jun
Jul
Migration
Breeding
(C) Nonbreeding
Jan
Dec Feb
Nov Mar (D)
Oct Apr
Sep May
Aug Jun
Jul
Year 1
Year 4 Year 2
Year 3
FIGURE 1.1 Examples of migration patterns in different taxa in North America: (A) the Porcupine caribou herd
(Rangifer tarandus), (B) songbirds (American Redstart Setophaga ruticilla), (C) insects (Monarch Butterfly Danaus plexippus),
and (D) fish (Pacific salmon). Migratory pathways depicted by arrows are only generalizations.
TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

1.1 INTRODUCTION 3
average North American motorist. Salmon when only part of the population migrates
return by the millions to natal streams at the (Boyle, Norris, & Guglielmo, 2010; Chapman,
end of their lives after spending years moving Brönmark, Nilsson, & Hansson, 2011; Jahn,
thousands of kilometers in the open ocean Levey, Hostetler, & Mamani, 2010). Differential
(Quinn, 2005). In the first year of their lives, migration describes those situations when
eastern populations of Monarch Butterflies migration patterns differ between sexes, age
(Danaus plexippus) in North America travel to groups, or morphs within a population
overwintering roosts in the Transvolcanic (Åkesson, 2016; Ketterson & Nolan, 1976;
Mountain range of central Mexico, a trip that Woodworth et al., 2016).
can be over 4000 km for an insect that weighs Some insect species move from the location
only 0.5 grams (Flockhart et al., 2013). For they were produced to another location where
those of us living in temperate (and in sub- they reproduce the next generation and die.
tropical) environments, the annual spring and That generation may then move on, repeating
autumn movements of billions of migratory the process. For Monarch butterflies, this
birds, from warblers to waterfowl, likely pro- occurs in a series of one-way steps involving
vide the most familiar examples of migration multiple generations before a final cohort
(Dingle, 2014; Greenberg & Marra, 2005). returns to their starting point, the discrete
Indeed these movements are spectacular, overwinter roosts in the mountains of central
yet much confusion still exists as to what Mexico (Flockhart et al., 2013). Many salmon
exactly we mean by migration. Clearly, migra- species also undertake spectacular migrations
tion can include both to-and-fro and one-way from the sea to freshwater rivers where they
movements. A to-and-fro or round-trip migra- spawn and die. Although this is often consid-
tion can be characterized by animals either ered one-way movement, it may be more
returning on the same path or by individuals accurately thought of as a “fatal round-trip
following a loop migration pattern. Various migration” because offspring first have to
other patterns of long- and short-distance migrate downriver where they spend several
movement have been described between origin years at sea and then return to freshwater to
and destination, especially for birds (Berthold, spawn and then die (Quinn, 2005).
1999). For example, leap-frog migration Altitudinal migration is a short-distance
involves individuals at the northern limits of movement between lower and higher elevation
their breeding range traveling the farthest habitat. It is commonly thought to happen in
south (in northern hemispheric animals) to response to resource availability (Levey &
most distant wintering grounds, whereas those Stiles, 1992) and weather or climate conditions
from more southern breeding regions migrate (Boyle et al., 2010). These movements are more
the least distance to more northern wintering common in the tropics where frugivory or nec-
grounds (e.g., Bell, 1997; Kelly, Atudorei, tarivory is well developed and in more tem-
Sharp, & Finch, 2002; Ramos et al., 2015). perate areas of high relief (Boyle, 2017).
Longitudinal migration involves all individuals Migration within the tropics over large dis-
migrating the same approximate distance in a tances, known as intratropical migration, occurs
chain pattern or in parallel (Norris et al., 2006; in many tropical-breeding bird species and has
Salomonsen, 1955). Animal migration can be more recently been observed in several
obligate whereby all members of a population temperate-breeding long-distance migrants
move or facultative whereby resource avail- (Heckscher, Halley, & Stampul, 2015; Levey &
ability may act to determine if migration Stiles, 1992; Morton, 1977). Such movements
occurs. Partial migration refers to a situation are potentially in response to changes in

resource availability and abundance, competi- their holistic view of migration, Dingle and
tion habitat change, or weather events (e.g., coworkers (Dingle, 2014; Dingle & Drake,
flooding; Boyle et al., 2010; Kardynal & 2007) see the arena as the environment to
Hobson, 2017; Stutchbury et al., 2016). which migrants are adapted and the migration
Although it may not seem like a form of syndrome is the suite of traits enabling migra-
migration, nomadism, invasion, irruption, where tory activity that involves both locomotory
animals move in irregular patterns and breed- capabilities and a set of responses to environ-
ing sites are established opportunistically mental cues that schedule and steer the loco-
where conditions are favorable, is a strategy motory activity. In this respect, migratory
employed by some animals. Several species of movements can be differentiated from move-
temperate forest seed-eating birds like cross- ments within a home range or dispersal. Their
bills, grosbeaks, and siskins are nomadic conceptual model also includes a genetic com-
because of their reliance on cone crops or plex that underlies the syndrome and a popu-
mast, which varies considerably in time and lation trajectory comprising the route followed
space (Newton, 2006; Strong, Zuckerberg, by the migrants, the timing of travel and stops
Betancourt, & Koenig, 2015). along it, and the periods and locations corre-
However, as Dingle (2014) noted, defining sponding to breeding and other key life-
migration based solely on the outcomes (or history events. The migration system involves
patterns) such as to-and-fro, one-way, altitudi- a species’ specific migration syndrome that
nal, and nomadic movements can be some- determines how the animal responds to exter-
what problematic. For example, many animals nal cues so that preemptive movements can
also show to-and-fro or nomadic movements occur prior to resource collapse. A fascinating
at a local scale within their home range. Regular area of study involves understanding those
movements within a home range can involve physiological adaptations to migration, includ-
commuting between resource patches such as ing the incorporation of reproduction and
hummingbirds moving among sources of nec- other tasks (Ramenofsky & Wingfield, 2007)
tar or between resource patches and reproduc- and the underlying genetic processes that
tive sites as seen in wolves during the denning result in selection for specific migratory traits
period and in many species of colonial nesting (Bazzi et al., 2016; Pulido, 2007; Roff &
seabirds. These types of movements are not Fairbairn, 2007). Although the overall control
considered migration because they occur of migration, its timing and response are
within the home range of the species and are known to be genetically determined, strikingly
irregular with no consistent or strict annual or this “unique” trait appears to be a primitive
seasonal basis (Dingle, 2014). Migration also characteristic also exhibited even in strict resi-
requires special physiological and behavioral dent and nonmigratory avian species (Helm &
adaptations during the period of movement, Gwinner, 2006). However, the ability to navi-
whereas movements within a home range usu- gate and orient during migration is a much
ally require few such demanding changes. more complex phenomenon that may require
Migration can be considered an adaptation both endogenous programing and acquired by
specific to arenas in which changes in habitat learning.
quality in different regions occur asynchro- As Dingle (2014) points out, the definition
nously so that movement allows a succession of migration must be based on the behavioral
of temporary resources to be exploited as they and physiological processes involved with the
arise. In this sense, migration involves escape movement rather than the spatial pattern or
and colonization (Dingle & Drake, 2007). In outcome. Thus, we consider migration as the

1.2 MIGRATORY POPULATIONS, CONNECTIVITY, AND CONSERVATION 5
movement away from the home range that Faaborg, 1995). In this sense, migratory organ-
does not cease (at least not initially when isms are best considered as a network that is
suitable resources are encountered) and which made up of a series of populations (that may or
requires a set of behavioral and physiological may not mix between different periods of the
adaptations for sustained movement that are year) rather than as a single species per se
unique from day-to-day adaptations related to (Taylor & Norris, 2010). For some animals,
self-maintenance and breeding. populations breeding in one area may follow
similar migration routes and winter in the same
general wintering region. In this case, we
1.2 MIGRATORY POPULATIONS, would consider the subpopulations to show
CONNECTIVITY, AND strong connectivity. Other breeding populations
CONSERVATION may disperse widely on the wintering grounds
(or vice versa), which can be considered a case
Although selection for adaptive traits related of weak connectivity.
to migration is thought to occur at the individ- The concept of migratory connectivity is
ual level, migration systems and patterns are important to conservation. Many forms of con-
more often described at the population level, nectivity are possible (Boulet & Norris, 2006;
typically as the migration of populations or Webster & Marra, 2005) ranging from the
subpopulations within a species. We usually extreme cases where individuals of breeding
deal with population units when engaging con- populations show high or low philopatry to
servation efforts and the conservation of popu- the same wintering location. Among birds, the
lations that move over large distances presents Kirtland’s Warbler (Setophaga kirtlandii) is a
considerable challenges (Marra, Norris, Haig, good example because the small breeding pop-
Webster, & Royle, 2006; Webster, Marra, Haig, ulation in Michigan, USA, winters exclusively
Bensch, & Holmes, 2002). Populations of ani- in the Bahamas. Weak connectivity may
mals might be well protected and managed at involve individuals from many breeding popu-
one location, but then suffer no protection once lations mixing among many wintering popula-
they leave that area. For long-distance migrants, tions or vice versa. However, such concepts
their travels may take them to new geopolitical have generally lacked quantification. The first
jurisdictions where conservation measures are attempt at quantifying connectivity was by
absent or inadequate. It makes little sense to Ambrosini, Moller, and Saino (2009) who
expect migratory populations to respond posi- applied Mantel correlation tests on distances
tively to local conservation measures if key between pairs of birds on breeding and winter-
aspects of individual fitness or overall popula- ing grounds based on ring recovery data of
tion health are being influenced at another Barn Swallow (Hirundo rustica) breeding in
location. Europe and wintering in Africa. More current
The problem of conserving species whose approaches to measuring connectivity consider
life cycles cross geopolitical borders has become both spatial and temporal dynamics (Bauer,
a profound issue in the 21st century as habitats Lisovski, & Hahn, 2016) and sampling bias
everywhere are being reduced in size or qual- (Cohen, Hostetler, Royle, & Marra, 2014; Cohen
ity, and migrants are forced to move among et al., 2017). A recent metaanalysis by Finch,
a myriad of patches, each with its own level Butler, Franco, and Cresswell (2017) showed
of quality, safety, and prospects for future that weak connectivity was the norm for long-
existence (Moore, Smith, & Sandberg, 2005; distance migratory birds in the Old and New
Robinson, Thompson, Donovan, Whitehead, & World.

Despite a general lack of quantification, the In terms of conservation, climatic and envi-
concept of migratory connectivity can be useful ronmental changes, and land use progression,
for conservation priority setting. Populations the timing of breeding has important implica-
with strong connectivity are hypothesized to be tions for avian productivity and is one of the
those that are most vulnerable to declines traits most noticeably affected by spring
because they have little chance to be “rescued” weather conditions (Dunn, 2004; Dunn &
by peripherally connected populations (Marra Winkler, 1999; Winkler, Dunn, & McCulloch,
et al., 2006; Taylor & Norris, 2010). On the other 2002). Birds that nest earlier generally produce
hand, it is the strongly connected populations more offspring that are ultimately recruited
that may benefit the most from conservation into breeding populations (Dunn, 2004;
efforts because the connections between two or Perrins, 1970). It remains uncertain whether
more areas are well established. Although migratory birds and other animals will be able
populations of weakly connected individuals to continue to synchronize the timing of their
may be inherently “safer,” in the sense that risk breeding with optimal conditions for repro-
is spread among more locations, these popula- duction on breeding areas in light of environ-
tions will understandably be harder to manage mental changes (Both & Visser, 2001;
unless measures are enacted over large regions Charmantier & Gienapp, 2014; Mayor et al.,
that encompass a significant percentage of the 2017; Visser, van Noordwijk, Tinbergen, &
subpopulation of interest. It may also be more Lessells, 1998).
difficult to identify the causes of population In their long-distance travels that cross
declines in subpopulations that are weakly con- international borders, migratory animals can
nected. For example, when migratory popula- also act as vectors of parasites and pathogens.
tions are mixed between two periods, habitat There are a number of ornithophilic mosqui-
loss in one subpopulation may produce toes that can infect migratory birds that in turn
synchronous declines in subpopulations in the act as introductory or amplifying hosts
following season (Hewson, Thorup, Pearce- (Rappole, Derrickson, & Hubalek, 2000;
Higgins, & Atkinson, 2016; Marra et al., 2006; Reed, Meece, Henkel, & Shukla, 2003;
Taylor & Norris, 2010). Yohannes, Križanauskienė, Valcu, Bensch, &
Knowledge of migratory connectivity is also Kempenaers, 2009). Recent interest in the
important for being able to predict how con- spread of avian influenza viruses around the
servation measures in one season will influ- globe has increased the interest in determining
ence populations the following season. For the migratory origins of infected and non-
example, Martin et al. (2007) demonstrated infected wild birds (Bodewes & Kuiken, 2018;
that the protection of nonbreeding habitat for Liu et al., 2005). This level of interest will only
American Redstarts (Setophaga ruticilla) based increase as the distributions of suitable habitats
on rate of habitat loss, density, and land cost available to these pathogens change over time.
resulted in the almost complete extinction of Effective management of species also
one breeding region. This was because the cri- requires identifying where populations are
teria for deciding where and when to conserve most productive. We are all familiar with field
nonbreeding habitat did not consider where guides that show the absolute spatial ranges
individuals were spending the remainder of occupied by a given species, but where within
the annual cycle. Consideration of how popu- that vast space are the most young being pro-
lations are connected can, therefore, result in duced and recruited into the autumn migra-
radically different decisions on how to allocate tory population and where are individuals
resources for conserving species. experiencing the highest survival rates during

1.2 MIGRATORY POPULATIONS, CONNECTIVITY, AND CONSERVATION 7
the nonbreeding period? Understanding where Marra, Hobson, and Holmes (1998) who, using
hot spots of productivity and survivorship stable isotopes, first demonstrated in American
occur helps us protect those places and better Redstarts a clear link between spring arrival
understand factors that may be limiting popu- times of males and the quality of the habitat
lations. Nor are species necessarily more pro- they occupied in winter. The quality of a habi-
ductive where they are most abundant, and tat where an animal winters will determine the
numerous examples exist of how animals can level and speed with which the necessary
be lured into ecological traps where they suffer body condition for migration can be reached,
reduced success. For migratory game birds, and so can determine when and in what condi-
understanding how zones of productivity tion that individual arrives on the breeding
relate to protected areas and the timing of har- grounds. Several studies of birds have shown
vest is particularly crucial to their conservation that earlier arriving birds can obtain better ter-
(Asante, Jardine, Van Wilgenburg, & Hobson, ritories, initiate clutches sooner, and generally
2017; Hobson, Van Wilgenburg, Ferrand, fledge more young than those arriving later
Gossman, & Bastat, 2013). (e.g., Bêty, Gauthier, & Giroux, 2003; Gill et al.,
For many years, our understanding of the 2001; Norris, Marra, Kyser, Sherry, & Ratcliffe,
ecology of migrant organisms was influenced 2004). How costs associated with reproduction
by their activities during the breeding season might influence an individual’s ability to com-
in locations where they were accessible to pete during the nonbreeding season remains a
scientists working in the temperate regions. mystery. Nonetheless, seasonal interactions
Such interest was also driven by an assump- between the wintering grounds and the breed-
tion that decisions made by individual animals ing grounds, as well as during migration itself,
related to breeding were of paramount impor- should ultimately play a large role in deter-
tance to an individual’s fitness. Aspects of mining lifetime fitness and population abun-
breeding success related to habitat choice, dance (Harrison, Blount, Inger, Norris, &
mate choice, timing of breeding, and the qual- Bearhop, 2011; Norris & Marra, 2007; Rushing
ity of individuals were topics that were consid- et al., 2017; Sillett & Holmes, 2002; Yohannes,
ered largely as they played out on the Mihai Valcu, Lee, & Kempenaers, 2010).
breeding grounds, with little regard to events Establishing details of migratory programs
that preceded and followed them on the win- in animals is fundamental to understanding
tering grounds and during migration. This sit- key aspects of their evolution, life history, and
uation changed in the 1980s when waterfowl conservation. The field has witnessed a renais-
researchers began to discover that female body sance in recent years primarily because the
condition upon arrival on the breeding development of new analytical techniques has
grounds in spring was linked to conditions she provided some true breakthroughs.
experienced in winter, and that these could There is also a more urgent sense that we
have profound influence on the number of need to more quickly understand key migra-
young recruited into the autumn population tory linkages for animal populations in order
(Heitmeyer & Fredrickson, 1981; Kaminski & to help conserve them in a changing world. It
Gleusing, 1987). is in this spirit that we developed this volume
In the late 1990s, it became more widely on stable isotope techniques because that
accepted that events occurring outside the approach has and will continue to contribute
breeding area can profoundly affect the fitness immensely to the study of animal migration.
of migratory individuals. Much of this recent We also recognize that, as with all techniques
interest has been generated from the work of and approaches, there is a real need to

provide a firm foundation for practitioners 30 % include 24 subspecies, and only 6 %
and to point to ways in which the field might include 5 or more subspecies.
best progress. By far the most widespread approach to
track migrant animals is through the applica-
tion of passive extrinsic markers. For birds,
1.3 SCIENTIFIC TOOLS USED these have overwhelmingly involved leg
TO STUDY MIGRATION bands or rings carrying a unique number
combination and some instruction on where
Tracking migratory terrestrial animals has to report the band if it is recovered. Other
involved numerous techniques over the years markers such as patagial tags, numbered neck
(Table 1.1). Until very recently, all approaches collars, streamers, or color dyes have also
involved the use of extrinsic markers applied been used. Insects have proven to be more of
to individuals with the hope of relocating a challenge due to their small size, but num-
those same individual elsewhere, or on the use bered labels have been successfully affixed to
of recognized phenotypic or morphological the wings of butterflies and later recovered
traits that showed known geographic varia- (Urquhart & Urquhart, 1978; www.monarch-
tion. For birds, there is a rich literature on geo- watch.org). In the relatively short historical
graphic variation in plumage and morphology period (i.e., 100 years) of banding birds, many
and these traits have been used to describe millions of individuals have been individually
migratory connectivity. A good example is the tagged. For a number of species with small
Yellow-rumped Warbler (Setophaga coronata) global populations and restricted ranges,
that shows plumage variation between eastern some very impressive recovery rates have
(coronata) and western (audubon) races in North been achieved (e.g., Owen & Black, 1989) and
America and that also segregates largely on some key insights into migratory connectivity
the wintering grounds. Similarly, the established (Cohen et al., 2014; Gill et al.,
Swainson’s Thrush (Catharus ustulatus) occurs 2001; Moore & Kremenetz, 2017). However,
as a reddish backed phenotype on the Pacific for the vast majority of species, extremely low
coast of North America and a more common recovery rates (i.e., ,0.01 %) are the norm
olive-backed form elsewhere. Birds captured (Hobson, 2003).
during migration can readily be placed into
these types without the need for mark-
recapture techniques. Most current field guides
provide numerous examples of geographic
1.3.1 Transmitters, Radar, and Satellites
variation in plumages (Sibley, 2014). An excel- Active extrinsic markers are those that send
lent example of how plumage and morpho- out signals that can be intercepted with a
metrics have been used to describe migration suitable receiver device. Advances in transmit-
patterns in NearcticNeotropical migrant ter technology have allowed the placement of
birds that moved through Mexico was that of devices on migratory animals. If a receiver is
Ramos and Warner (1980) and Ramos (1983). within the range of the transmitter, then the
However, typically the geographic resolution location of the animal can be inferred either by
provided by this approach is both poor and tracking down the individual or by triangula-
highly variable among taxonomic groups. For tion. Radio frequency transmitters can be made
example, as noted by Boulet and Norris (2006), small enough (B0.4 g) to place on small passer-
within the 50 species of migratory wood war- ines and bats. However, with miniaturization
blers of the New World, 65 % are monotypic, comes a loss of range and battery life so that

1.3 SCIENTIFIC TOOLS USED TO STUDY MIGRATION 9
TABLE 1.1 Summary of Various Techniques Used to Track Migration in Animals
Technique Advantages Disadvantages References
Extrinsic 1. Can apply to a broad 1. Requires initial capture and

range of animals. then recapture.
2. High spatial resolution. 2. Biased toward initial capture

population.
Phenotypic 1. Inexpensive. 1. Low resolution. Ramos (1983), Ramos and Warner
variation (1980), Bell (1997), Boulet and
(e.g., morphology 2. Can be applied to 2. Not applicable to all species. Norris (2006), Conklin, Battley,
historical specimens with
and plumage) 3. Provides estimate of natal Potter, and Ruthrauff (2011)
high degree of
origin only.
confidence.
Banding/marking 1. Inexpensive. 1. Many species have low Brewer, Diamond, Woodsworth,
recovery rates (,0.5%). Collins, and Dunn (2000), Bairlein
2. Provides exact (2001), Cohen et al. (2014), Moore
information on start and 2. Can take many years to get and Kremenetz (2017)
end of movements. adequate data.
3. Still only a small number of
major banding stations across
globe.
4. Both marking and recovery
(start and end points) are
biased toward locations of
major banding stations.
Radio 1. Produces precise 1. Low range. Cochran et al. (2004), Barron et al.
transmitters locations (relative to (2010), Taylor et al. (2017)
extrinsic markers through 2. Can obtain precise trajectory
triangulation). if within range of
transmission.
2. Can obtain precise
trajectory if within range 3. Expensive relative to banding
or morphological methods.
of transmission.
4. Some evidence suggests
transmitters can have adverse
effect on behavior.
Satellite tags 1. Precise animal trajectory. 1. Expensive (severely Hays, Akesson, Godley, Luschi, and
constrains sample size). Santidrian (2001), Hallworth and
Marra (2015), Kays et al. (2015)
2. Transmitters constrained to
large animals only (B100 g).
3. Archival tags provide
relatively few positions for
smaller models, and animals
have to be recaptured.
4. Transmitters can have
adverse effect on behavior.
(Continued)

TABLE 1.1 (Continued)

ICARUS Initiative 1. Small and light 1. Huge start-up investment. Wikelski et al. (2007)
transmitters allow many
types of species to be 2. Not yet proven technology.
tracked. 3. Like above, some evidence
2. Similar to satellite tags, suggests that transmitters can
have adverse effect on
can track individuals
over large distances behavior.
(global).
3. Relatively inexpensive
following start-up
investment.
Remote sensing/ 1. Coverage over large 1. Coverage only from existing Gauthreaux and Belser (2003),
passive radar geographic area. stations or Chilson et al. (2012), Zaugg,
portable instruments. Saporta, van Loon, Schmaljohann,
2. Inexpensive. and Liechti (2008), Kirkeby et al.
2. Poor ability to determine (2016)
3. Individuals do not have species- and individual-
to be captured. specific movements.
Transponders 1. Small transponder size. 1. Requires external (radar/ Riley et al. (1996), Chapman,
microwave) activation. Reynolds, and Smith (2004)
2. Low range.
3. Coverage only from existing
stations or
portable instruments.
Archival 1. Produces an animal 1. Individuals must be Shaffer et al. (2005), Stutchbury
geolocation tags trajectory; interval recaptured to download data. et al. (2009)
between recorded
locations can be 2. Accuracy relative to satellite
customized. tags still low ( 6 200 km).
2. Light weight (,1 g).

3. Inexpensive relative to
satellite tags (but more
expensive than some
radio transmitters).
Intrinsic Because only requires one 1. Biased to final captured Hobson (1999), Webster et al.
capture, it is: population (can be overcome (2002), Rubenstein and Hobson
1. Not biased to initial by comprehensive sampling (2004)
capture population. coverage).
2. Less labor intensive than 2. Typically lower resolution
most extrinsic methods. than extrinsic markers.
(Continued)

Contaminants 1. Potentially high spatial 1. Lack of a priori maps of Ochoas-Acuna et al. (2002), Braune
resolution (e.g., Mirex). distribution and relative and Simon (2003), Sanpera, Ruiz,
abundance. Moreno, Jover, and Waldron (2007)
2. Distribution of contaminants
may not vary predictably
over geographic areas.
3. Potential transport of
contaminants may dampen or
provide unreliable geographic
signal.
Parasites 1. Potentially high spatial 1. Potentially expensive. Fallon, Fleisher, and Graves (2006),
resolution. Ricklefs et al. (2005), von Rönn
et al. (2015)
Pollen 1. Potentially high spatial 1. Potentially expensive. Mikkola (1971), Hendrix, William,
resolution and Showers (1992)
2. Generally provides
information of recent plant
use only.
Genetics 1. Several markers possible. 1. Species specific Kelly, Ruegg, and Smith (2005),
Smith et al. (2005), Boulet et al.
2. Longitudinal resolution 2. Typically low resolution (2006), Ruegg et al. (2014)
of migratory fauna in
North America. 3. Provides estimate of natal
origin only
Trace elements 1. Simultaneous 1. Can be expensive. Parrish, Rogers, and Ward (1983),
measurement of a large Kelsall (1984), Szep et al. (2003),
number of elements. 2. 2. Cannot extrapolate to large Norris et al. (2007), Kaimal,
Potentially high spatial scales (i.e. site specific). Johmson, and Hannigan (2009)
resolution.
Stable isotopes 1. Inexpensive. 1. Relatively low resolution. Chamberlain et al. (1997), Hobson
and Wassenaar (1997), Hobson
2. Not species or taxon (1999, 2005), Norris et al. (2006)
specific.
3. Multiple isotopes can be
combined to increase
spatial resolution.
optimally these devices provide location infor- Wikelski, 2004; Holland, Thorup, Vonhof,
mation up to a few kilometers. Nonetheless, Cochran, & Wikelski, 2006; Wikelski et al.,
adventurous researchers have attempted to fol- 2003). Using a combination of aerial and
low migrating birds, bats, and dragonflies ground tracking receivers, Wikelski et al. (2006)
equipped with transmitters over portions of followed the southward migration of dragon-
their flight paths (Cochran, Mouritsen, & flies (Common Green Darner Anax junius)

equipped with 300 mg radio transmitters. Cell a recovery point of individuals if they pass near
phone technology has also provided new a tower but no inferences can be made for
possibilities for tracking animals but the minia- those individuals not detected. Nonetheless, the
turization and development required for animal system to date has produced some impressive
tracking remains a hurdle. As well, cell phone results documenting timing of migration over
network coverage is limited internationally and continental scales (Gómez et al., 2017).
locational accuracy is low (Stokely, 2005). A final Remote sensing technologies such as radar
concern is that fastening of markers or trans- have made great contributions to our under-
mitters may alter flight or movement behavior standing of migration because they can pro-
(Barron, Brawn, & Weatherhead, 2010). vide information of flying animal movements
A major advance in the way migratory over considerable distance (Chilson, Bridge,
wildlife, especially birds and bats, can be Frick, Chapman, & Kelly, 2012; Gauthreaux &
tracked using Very High Frequency (VHF) Belser, 2003). However, like automated recei-
technology has been the MOTUS system in vers, radar installations are usually at fixed
North America (https://motus.org). This locations, and mobile radar systems are gener-
involves a broad-scale automated telemetry ally impractical to follow animal movements
array (Taylor, Crewe, Mackenzie, Lepage, & over long migratory distances. Crossband
Aubry, 2017) whereby collaborators employ transponders placed on migratory organisms
receivers tuned to a single radio frequency can be used to elicit a detectable radio fre-
across receiving stations over a broad geo- quency signal after the transponder is inter-
graphic scale, allowing individuals to be cepted by radar. The transponder is dormant
detected at distal sites maintained by others. until such an event and so can last for a much
Each individual VHF transmitter can be coded longer period than conventional “always trans-
to a specific signal burst rate. Motus also coor- mitting” radio frequency tags. However, such
dinates, disseminates, and archives detections active radar systems face a number of limita-
and associated metadata in a central reposi- tions, again related to size and weight of
tory. Combined with the ability to track many instrumentation and the need to intercept the
individuals simultaneously, Motus has organism of interest within range of the radar
expanded the scope and spatial scale of system (see www.earthspan.org). Further
research questions that can be addressed using advances in remote sensing like light in detec-
radio telemetry from local to regional and tion and ranging will undoubtedly play a larger
even hemispheric scales. Since its inception in role in advancing the study of animal migra-
2012, more than 9000 individuals of over 87 tion in the future (Kirkeby, Wellenreuther, &
species of birds, bats, and insects have been Brydegaard, 2016).
tracked, resulting in more than 250 million Satellite transmitters have provided a major
detections. This rich and comprehensive data- advance in methods to track migratory animals
set includes detection of individuals during all because they provide extremely accurate posi-
phases of the annual cycle (breeding, migrations of individuals remotely. Precise trajecto-
tion, and nonbreeding), and at a variety of spa- ries of individuals are possible over vast
tial scales, resulting in novel insights into the portions of the globe that have suitable satellite
movement behavior of small flying animals. coverage. Satellite tracking systems such as
The value of the Motus network will grow as ARGOS (www.argosinc.com) have been placed
spatial coverage of stations and number of on different satellites from the US National
partners and collaborators increases. Current Oceanic and Atmospheric Administration, The
drawbacks to this technology are cost and lim- Japanese Space Agency, and the European
itations of coverage. Thus it is possible to define Meteorological Satellite Organization. The

ARGOS system collects data from platform ter- signals obtained by the ICARUS receiver and
minal transmitters and delivers telemetry data data streaming. ICARUS foresee its onboard
back to the user. Unlike other extrinsic computer to be active and readily operational
techniques, this approach does not require the to user interface on Earth in 2018.
physical capture of individuals once they are A relatively new application for tracking
marked (although many researchers attempt migratory animals is the use of archival geolo-
this to recover the expensive transmitters). One cation tags (Delong, Stewart, & Hill, 1992; Hill,
of the most spectacular examples of this 1994). These tags estimate longitude and lati-
approach was provided by Jouventin and tude based on light levels (and in some cases,
Weimerskirch (1990) who tracked Wandering sea-surface temperatures), and long-distance
Albatrosses (Diomedea exulans) on foraging movement data have been reported on many
flights up to 15,000 km. Future prospects for large migratory species such as elephant seals
this technology are encouraging and there is (Delong et al., 1992) and seabirds (Shaffer
strong interest in diminishing the size of trans- et al., 2005; Tuck et al., 1999). However, recent
mitters and batteries so that smaller species can miniaturization has allowed their use in the
be monitored. Currently, the smallest satellite study of smaller species (Stutchbury et al.,
tags available are about 3.5 g which excludes 2009). The advantage of these tags is that
many of the smaller migratory birds, bats, and they are significantly lighter (e.g. , 1g)than
almost all insects. The use of satellites may satellite transmitters. However, the disadvan-
potentially provide a major breakthrough in tage is that their accuracy is only 6 200 km
tracking migratory animals down to the size of that restricts application of these tags to ques-
large insects. tions pertaining only to large-scale movements
Wikelski et al. (2007) proposed that a new although recent mathematical developments
satellite equipped with radio receivers could have increased the reliability of light-level geo-
track radio tags with radiated power as low as location data (Bindoff, Wotherspoon, Guinet,
1 mW with an accuracy of a few kilometers & Hindell, 2017; Rakhimberdiev et al., 2016).
under favorable conditions. This power can be With the exception of satellite transmitters,
achieved from existing radio frequency tags as all extrinsic markers require that the individ-
small as B1 g. This approach is similar to solu- ual be recaptured, resighted, or at least move
tions derived by space researchers who need within detection distance at a later time. The
to detect very weak signals from distant galax- probability of this recapture will clearly be the
ies and so the technology exists to modify the product of a number of individual probabili-
solution to detecting weak radio signals from ties related to the number of observers, the
Earth amidst a background of much stronger likelihood of an observer reporting the infor-
radio frequency noise. A group known as mation, the regions and habitats used by the
International Cooperation for Animal Research animal, and so on. As we have seen, combin-
Using Space (ICARUS, www.icarusinitiative. ing these probabilities can result in vanishingly
org) is promoting this idea as a means of track- small chances of obtaining information on any
ing small animals around the globe. It is esti- given individual.
mated that it will take about US$50100 Apart from the burden of recovering a
million in order to build and launch a satellite. marked individual, the use of passive extrinsic
Thus far, the central computer unit of ICARUS markers suffers another fundamental
has been installed in the cabin of the ISS inter- flaw—they provide information only on the
national space station’s Russian module. In the movement of marked individuals. The possibil-
near future, it is expected that the onboard ity of extrapolating the findings based on a
computer will be responsible for decoding the small marked cohort to the population or

species level depends on how representative the northeast Atlantic. In the Great Lakes region of
marked individuals are among the population. North America, Mirex could be used as a
Our confidence in this approach increases with marker to distinguish among those waterbirds
the number of independent recaptures and breeding and wintering in different regions
some band recovery data show unequivocal pat- because the upper lakes have much lower con-
terns that are clearly representative of popula- centrations of this contaminant than the lower
tions. A single band recovery or satellite track lakes. The ratio of DDE (dichlorodiphenyldi-
while interesting and ultimately useful may tell chloroethylene) to PCB (polychlorinated biphe-
us very little about what populations are doing. nyl) can also be used to illustrate which birds
have likely overwintered in agricultural areas
of South/Central America versus coastal areas
1.3.2 Intrinsic Markers (Braune, unpublished data). Brominated flame
retardants are another material that occurs in
The primary advantage of intrinsic markers
greater concentrations in food webs used by
is that initial marking of individuals is not nec-
animals exposed to European air masses than
essary and that every capture provides infor-
those exposed to North American air masses
mation on origin. In this sense, every capture
(de Wit, 2002). Finally, methyl mercury and
becomes a recapture. The sampling scheme is
other heavy metals are another potential
biased then, only by the limitations of where
marker in migratory animals because exposure
animals are ultimately located and this typi-
to these varies considerably throughout the
cally represents a much less serious form of
world (Hario & Uuksulainen, 1993; Janssens,
bias compared to where individuals can be
Dauwe, Bervoets, & Eens, 2002; Ochoas-
marked initially using extrinsic markers.
Acuna, Sepulveda, & Gross, 2002).
Similar to contaminants, parasite and patho-
1.3.2.1 Contaminants, Parasites, gen exposure experienced by migratory animals
and Pathogens varies geographically and there is interest in
There are several forms of potential intrinsic using these markers to examine movements of
markers that can be used to infer origins of individuals (Ricklefs, Fallon, Latta, Swanson, &
migratory animals. If the spatial distribution of Bermingham, 2005; von Rönn, Harrod, Bensch,
a suite of contaminants were known, then pre- & Wolf, 2015). Very little research has been
sumably the occurrence of contaminants in conducted on these sorts of tools to assist
animals could provide some information of the with deciphering animal movements, perhaps
past distribution of the animals being sampled. because we have little information to begin
Although this approach has not yet been used with on the spatial distributions of the potential
to any significant degree, there is some inter- markers themselves. On the other hand, the use
esting potential here. For example, Braune and of genetics has generated more interest because
Simon (2003) noted that in contrast to Thick- it is possible to assay and describe genetic vari-
billed Murres (Uria lomvia) and Black-legged ation across breeding populations and then to
Kittiwakes (Rissa tridactyla), the pattern and use this information to assign a probability that
concentration of dioxin/furan congeners as a given individual came from a given (known)
well as the ratio of total PBDEs (polybrome- subpopulation (Ruegg et al., 2014; Smith et al.,
nated diphenyl ethers) to HBCD (hexabromo- 2005). The identification of population structure
cyclododecane) in Northern Fulmars (Fulmarus using genetic markers has included the use of
glacialis) breeding in the Canadian high Arctic allozymes, mitochondrial DNA sequences, and
was suggestive of them wintering in the DNA fragment analyses such as microsatellites

and amplified fragment length polymorphism. places the previous winter (Szep, Moller,
These markers can yield different scales of pop- Vallner, Kovacs, & Norman, 2003).
ulation structure because they evolve at differ- However, because we have little idea of how
ent rates. In North America, genetic markers trace elements vary over the landscape, the
have been particularly useful in differentiating ability of this technique to estimate the precise
between eastern and western breeding origins origin of individuals seems, at the moment,
of wintering Neotropical migrant songbirds limited. To determine how trace elements vary
(Boulet, Gibbs, & Hobson, 2006; Ruegg et al., across the landscape, Norris et al. (2007) mea-
2014; Smith et al., 2005), reflecting patterns of sured 42 trace elements in feathers of western
rapid demographic expansions following glaci- sandpipers (Calidris mauri) at 5 different loca-
ation events on that continent. The combination tions on their tropical wintering grounds.
of marker techniques that can provide informa- Feathers were grown during the winter periods
tion on northsouth origins in North America and were, therefore, assumed to provide a sig-
would clearly augment the resolution of genetic nature of known-origin. Elemental profiles suc-
approaches. Genetic analyses also hold great cessfully distinguished between birds wintering
potential for developing parasite markers in at all five locations. However, two locations
migrant organisms because PCR-based assays were less than 3 km apart suggesting that trace
of blood parasites can now identify pathogens elements are likely very specific to the location
to species and haplotype (Ricklefs et al., 2005). or origin in which they were sampled. Trace
element approaches are probably best suited to
1.3.2.2 Trace Elements species that are aggregated over only a few
Trace elements are another set of intrinsic breeding or wintering sites so the majority of
markers that can potentially be used to track populations can be sampled over the entire
individual movements over both small and range (Donavan et al., 2006). Studies several
large geographic distances. Similar to years ago also show that some trace elements
stable isotopes, the idea with trace elements is in feathers may be acquired after growth
that individuals acquire distinctive chemical (Bortolotti & Barlow, 1988; Bortolotti, Szuba,
profiles at one geographic location and then Naylor, & Bendell, 1988), implying that this
carry that profile with them to another area may not be a reliable approach for tracking
where they can be sampled to estimate their long-distance movements. Further studies are
geographic origin. In the past, one of the needed to test the generality of these results
limitations of using trace elements was the before trace elements can be widely used as a
amount of sample required for analysis. marker of geographic location.
However, the use of inductively coupled
plasma mass spectrometers has allowed smal- 1.3.2.3 Stable Isotope Approaches
ler quantities of samples (Donavan, Buzas, The intent of this volume is to provide the
Jones, & Gibbs, 2006; Norris et al., 2007) to be reader with a comprehensive background
measured with relatively high precision. This needed to understand the application of
technological advance has allowed research- stable isotope tools to the study of animal
ers to focus on nondestructive tissues, such as migration. However, animal ecologists have
metabolically inactive keratin, that are grown really only recently become aware of using
during specific periods of the migratory cycle. stable isotope measurements in ecological
Trace element profiles have been used to infer studies in general. The world of stable isotopes
whether individuals sampled in the same is a multidisciplinary one that has a very rich
breeding population originate from different history involving physics, earth sciences,

biogeochemistry, animal and plant physiology, sensors integrated into tracking tags has strik-
and even anthropology and archeology. The ing effect on animal movement ecology. It has
1988 publication of the book Stable Isotopes created opportunities to advance science most
in Ecological Research edited by Rundel, rapidly but has also created challenges with
Ehleringer, and Nagy (1988) marked a signifi- the generation of massive datasets applicable
cant pivot point that informed ecologists of the in several fields (Kays, Crofoot, Jetz, &
immense potential of stable isotope measure- Wikelski, 2015).
ments in natural ecosystems. This was quickly Elsewhere, analytical chemistry laboratories
followed by Stable Isotopes in Ecology and have improved instrumentation establishing
Environmental Science edited by Lajtha and both the identity of newly synthesized physio-
Michener (1994), which is now updated to logical material, or probing the properties of
Michener and Lajtha (2007). Most recently, bioactive compounds. Today’s analytical
Fry (2006) has produced a useful text that chemistry offers a broad array of equipment
provides some of the necessary background to with a an alphabet soup of options that range
understanding fundamental concepts in from gas chromatographymass spectrometry,
stable isotope ecology and Karasov and gas chromatographyisotope ratio mass spec-
Martı́nez del Rio (2007) provide a very useful trometry and nuclear magnetic resonance
overview of key linkages between animal (NMR) to ICP (inductively coupled plasma)
physiology and isotopic ecology. and matrix-assisted laser desorption electro-
Compound-specific isotope analysis refers to spray ionization.
measurement of the isotopic signatures (typi- Indeed, the type of hardware is growing,
cally, carbon, hydrogen, oxygen, nitrogen, or with new developments appearing rapidly.
sulfur stable isotopes) of individual compounds Many of these devices are increasingly used in
(such as amino acids, fatty acids, and sterols) the biological world to create tools capable of
from a complex environmental or biological detecting molecular signatures in biological tis-
mixture. The methodology offers information sues. Smaller molecular signatures may thus
on dietary and trophic source information even be applied not only to track movement but
in the absence of food collected from the study also to understand physiological processes
subject (e.g., birds on migration route). It also during migration, movement, breeding, and
provides data on source differentiation, and can molt. And of course, researchers are constantly
be used as quantitative means to delineate reac- discovering ever more creative ways to apply
tion pathways. Moreover, it can provide evi- NMR and mass spectrometry to identify com-
dence for contaminant monitoring of breeding, plexes of physiological/endogenous molecules
migration, or wintering sites with pollutants that can be applicable to migration ecology
such as hydrocarbons. and easily coupled to tracking sciences
(Mitchell, Guglielmo, & Hobson, 2015).
New analytical methods in isotopic and
1.4 TECHNICAL ADVANCES molecular ecology, DNA barcoding, approaches,
AND OUTLOOK integrated with remote sensing and theoretical
ecological modeling has and will open up an
The last decade has shown tremendous abundance of new questions on the biological,
advances in animal tracking technology her- environmental, and physiological control of
alding the golden age of animal movement movement and its consequences for life-history
ecology. Recent construction of high-resolution patterns of organisms, populations, communi-
Global Positioning System and small designed ties, and ecosystems. The current approaches of

REFERENCES 17
simultaneous tracking of multiple animals (both Bauer, S., Lisovski, S., & Hahn, S. (2016). Timing is crucial
prey and predator) is opening new insights into for consequences of migratory connectivity. Oikos, 125,
605612. Available from https://doi.org/10.1111/
foodweb interactions and combining tracking oik.02706.
approaches with analytical tools will scale up Bazzi, G., Galimberti, A., Hays, Q. R., Bruni, I., Cecere,
our understanding in monitoring migration in a J. G., Gianfranceschi, L., . . . Rubolini., D. (2016).
dynamic, changing environment. Adcyap1 polymorphism covaries with breeding lati-
In summary, webelieve that a golden age of tude in a Nearctic migratory songbird, the Wilson’s
warbler (Cardellina pusilla). Ecology and Evolution, 6.
movement and migration tracking has indeed Available from https://doi.org/10.1002/ece3.2053.
arrived and the coming years will be a time of Bell, C. P. (1997). Leap-frog migration in the Fox Sparrow:
more discoveries. As tracking technology and Minimizing the cost of spring migration. Condor, 99,
analytical instrumentation continue to advance, 470477. Available from https://doi.org/10.2307/
integrating these approaches will open a whole 1369953.
Berthold, P. (1999). A comprehensive theory for the evolu-
new paradigm of research on migration. For tion, control, and adaptability of avian migration.
instance, real-time acquisition of data on the Ostrich, 70, 111. Available from https://doi.org/
movement and behavior of tagged birds cou- 10.1080/00306525.1999.9639744.
pled to data from stable isotope, molecular Bêty, J., Gauthier, G., & Giroux, J.-F. (2003). Body condi-
approaches, DNA barcoding, and remote sens- tion, migration and timing of reproduction in Snow
Geese: A test of the condition-dependent model of opti-
ing will provide transformational points mal clutch size. American Naturalist, 162, 110121.
whereby new insight could be achieved to Available from https://doi.org/10.1086/375680.
understand the intrinsic and extrinsic controls of Bindoff, A. D., Wotherspoon, S. J., Guinet, C., & Hindell,
animal migration, movement, and dispersal and M. A. (2017). Twilight-free geolocation from noisy light
finally to help us develop conservation data. Methods Ecology and Evolution, 19. Available
from https://doi.org/10.1111/2041-210X.12953.
approaches in entirely new ways. Bodewes, R., & Kuiken, T. (2018). Changing role of wild
birds in the epidemiology of avian influenza A viruses.
Advances in Virus Research, 100, 279307. Available
from https://doi.org/10.1016/bs.aivir.2017.10.007.
References Bortolotti, G. R., & Barlow, J. C. (1988). Some sources of
Åkesson, S. (2016). Flying with the winds: Differential variation in the elemental composition of Bald Eagle
migration strategies in relation to winds in moth and feathers. Canadian Journal of Zoology, 66, 19481951.
songbirds. Journal of Animal Ecology, 85, 14. Available Available from https://doi.org/10.1139/z88-285.
from https://doi.org/10.1111/1365-2656.12450. Bortolotti, G. R., Szuba, K. J., Naylor, B. J., & Bendell, J. F.
Ambrosini, R., Moller, A., & Saino, N. (2009). A quantita- (1988). Stability of mineral profiles of spruce grouse
tive measure of migratory connectivity. Journal of feathers. Journal of Wildlife Management, 52, 736743.
Theoretical Biology, 257, 203211. Available from Available from https://doi.org/10.2307/3800939.
https://doi.org/10.1016/j.jtbi.2008.11.019. Both, C., & Visser, M. E. (2001). Adjustment to climate
Asante, C. K., Jardine, T., Van Wilgenburg, S. L., & Hobson, change is constrained by arrival date in a long-distance
K. A. (2017). Tracing origins of waterfowl using the migrant bird. Nature, 411, 296298. Available from
Saskatchewan River Delta: Incorporating stable isotope https://doi.org/10.1038/35077063.
approaches in continent-wide waterfowl management Boulet, M., & Norris, D. R. (2006). The past and present of
and conservation. Condor, 119, 261274. Available from migratory connectivity. Ornithological Monographs, 61,
https://doi.org/10.1650/CONDOR-16-179.1. 113. Available from https://doi.org/10.2307/40166835.
Bairlein, F. (2001). Results of bird ringing in the study of Boulet, M., Gibbs, H. L., & Hobson, K. A. (2006).
migration routes. Ardea, 89, 719. Integrated analysis of genetic, stable isotope, and
Barron, D. G., Brawn, J. D., & Weatherhead, P. J. (2010). banding data reveal migratory connectivity and fly-
Meta-analysis of transmitter effects on avian behaviour ways in the northern Yellow Warbler (Dendroica pete-
and ecology. Methods in Ecology and Evolution, 1, chia; Aestiva group). Ornithological Monographs, 61,
180187. Available from https://doi.org/10.1111/ 2978. Available from https://doi.org/10.2307/
j.2041-210X.2010.00013.x. 40166837.

Boyle, W. A. (2017). Altitudinal bird migration in North Cohen, E. B., Hostetler, J. A., Hallworth, M. T., Rushing,
America. Auk, 134, 443465. Available from https:// C. S., Sillett, T. S., & Marra, P. P. (2017). Quantifying
doi.org/10.1642/AUK-16-228.1. the strength of migratory connectivity. Methods in
Boyle, W. A., Norris, D. R., & Guglielmo, C. G. (2010). Storms Ecology and Evolution, 112. Available from https://
drive altitudinal migration in a tropical bird. Proceedings doi.org/10.1111/2041-210X.12916.
of the Royal Society of London B, 277, 25112519. Available Conklin, J. R., Battley, P. F., Potter, M. A., & Ruthrauff,
from https://doi.org/10.1098/rspb.2010.0344. D. R. (2011). Geographic variation in morphology of
Braune, B., & Simon, M. (2003). Dioxins, Furans, and non- Alaska-breeding Bar-Tailed Godwits (Limosa lapponica)
ortho PCBs in Canadian Arctic seabirds. Environmental is not maintained on their nonbreeding grounds in
Science and Technology, 37, 30713077. Available from New Zealand. Auk, 128, 363373. Available from
https://doi.org/10.1021/es021082p. https://doi.org/10.1525/auk.2011.10231.
Brewer, A. D., Diamond, A. W., Woodsworth, E. J., Collins, Delong, R. L., Stewart, B. S., & Hill, R. D. (1992).
B. T., & Dunn, E. H. (2000). The atlas of Canadian bird Documenting migrations of northern elephant seals
banding, 192195. Volume 1: Doves, cuckoos and hum- using daylength. Marine Mammal Science, 8, 155159.
mingbirds through passerines. Ottawa, ON: Canadian Available from https://doi.org/10.1111/j.1748-7692.
Wildlife Service Special Publication. ,http://www. 1992.tb00375.x.
cws-scf.ec.gc.ca/publications/spec/atlas_e.cfm.. Dingle, H. (2014). Migration: The Biology of Life on the Move.
Chamberlain, C. P., Blum, J. D., Holmes, R. T., Feng, X., (2nd ed.). New York: Oxford University Press.
Sherry, T. W., & Graves, G. R. (1997). The use of Dingle, H., & Drake, V. A. (2007). What is migration?
stable isotope tracers for identifying populations of Bioscience, 57, 113121. Available from https://doi.
migratory birds. Oecologia, 109, 32141. Available from org/10.1641/B570206.
https://doi.org/10.1007/s004420050067. Donavan, T., Buzas, J., Jones, P., & Gibbs, H. L. (2006).
Chapman, B. B., Brönmark, C., Nilsson, J. Å., & Hansson, Tracking dispersal in birds: Assessing the potential of
L. A. (2011). The ecology and evolution of partial elemental markers. Auk, 123, 500511. Available from
migration. Oikos, 120, 17641775. Available from https://doi.org/10.1642/0004-8038(2006)123[500:TDIBAT]
https://doi.org/10.1111/j.1600-0706.2011.20131.x. 2.0.CO;2.
Chapman, J. W., Reynolds, D. R., & Smith, A. D. (2004). Dunn, P. O. (2004). Breeding dates and reproductive per-
Migratory and foraging movements in beneficial formance. Advances in Ecological Research, 35, 6987.
insects: A review of radar monitoring and tracking Available from https://doi.org/10.1016/S0065-2504(04)
methods. International Journal of Pest Management, 50, 35004-X.
225232. Available from https://doi.org/10.1080/ Dunn, P. O., & Winkler, D. W. (1999). Climate change has
09670870410001731961. affected the breeding date of tree swallows throughout
Charmantier, A., & Gienapp, P. (2014). Climate change and North America. Proceedings of the Royal Society of London
timing of avian breeding and migration: Evolutionary B, 266, 24872490. Available from https://doi.org/
versus plastic changes. Evolutionary Applications, 7, 1528. 10.1098/rspb.1999.0950.
Available from https://doi.org/10.1111/eva.12126. Egevang, C., Stenhouse, I. J., Phillips, R. A., Petersen, A.,
Chilson, P. B., Bridge, E., Frick, W. F., Chapman, J. W., & Fox, J. W., & Silk, J. R. D. (2010). Tracking of Arctic
Kelly, J. F. (2012). Radar aeroecology: Exploring the terns Sterna paradisaea reveals longest animal migra-
movements of aerial fauna through radio-wave remote tion. Proceedings of the National Academy of Sciences of
sensing. Biology Letters, 8, 698701. Available from the United States of America, 107, 20782081. Available
https://doi.org/10.1098/rsbl.2012.0384. from https://doi.org/10.1073/pnas.0909493107.
Cochran, W. W., Mouritsen, H., & Wikelski, M. (2004). Fallon, S. M., Fleisher, R. C., & Graves, G. R. (2006).
Migrating songbirds recalibrate their magnetic compass Malarial parasites as geographical markers in migratory
daily from twilight cues. Science, 304, 405408. Available birds? Biology Letters, 2, 213216. Available from
from https://doi.org/10.1126/science.1095844. https://doi.org/10.1098/rsbl.2005.0429.
Cohen, E. B., Hostetler, J. A., Royle, J. A., & Marra, P. P. Finch, T., Butler, S. J., Franco, A. M. A., & Cresswell, W.
(2014). Estimating migratory connectivity of birds when (2017). Low migratory connectivity is common in long-
re-encounter probabilities are heterogeneous. Ecology distance migrant birds. Journal of Animal Ecology, 86,
and Evolution, 4, 16591670. Available from https:// 662673. Available from https://doi.org/10.1111/1365-
doi.org/10.1002/ece3.1059. 2656.12635.

REFERENCES 19
Flockhart, D. T. T., Wassenaar, L. I., Martin, T. G., Heitmeyer, M. E., & Fredrickson, L. H. (1981). Do wetland
Hobson, K. A., Wunder, M. B., & Norris, D. R. (2013). conditions in the Mississippi Delta hardwoods influence
Tracking multi-generational colonization of the mallard recruitment? Transactions of the North American
breeding grounds by monarch butterflies in eastern Wildlife and Natural Resources Conference, 46, 4457.
North America. Proceedings of the Royal Society B, 280, Helm, B., & Gwinner, E. (2006). Migratory restlessness in
1087. Available from https://doi.org/10.1098/ an equatorial nonmigratory bird. PLoS Biology, 4, e110.
rspb.2013.1087, 2013. Available from https://doi.org/10.1371/journal.
Fry, B. (2006). Stable isotope ecology. New York: Springer. pbio.0040110.
Gauthreaux, S. A., & Belser, C. G. (2003). Radar ornithol- Hendrix, I., William, H., & Showers, W. B. (1992). Tracing
ogy and biological conservation. Auk, 120, 266277. black cutworm and armyworm (Lepidoptera:
Available from https://doi.org/10.1642/0004-8038 Noctuidae) northward migration using Pithecellobium
(2003)120[0266:ROABC]2.0.CO;2. and Calliandra pollen. Environmental Entomology, 21,
Gill, J. A., Norris, K., Potts, P. M., Gunnarsson, T. G., 10921096.
Atkinson, P. W., & Sutherland, W. J. (2001). The buffer Hewson, C. M., Thorup, K., Pearce-Higgins, J. W., &
effect and large-scale population regulation in migra- Atkinson, P. W. (2016). Population decline is linked to
tory birds. Science, 412, 436438. Available from migration route in the Common Cuckoo. Nature
https://doi.org/10.1038/35086568. Communications, 7, 12296. Available from https://doi.
Gómez, C., Bayly, N. J., Norris, D. R., Mackenzie, S. A., org/10.1038/ncomms1229.
Rosenberg, K. V., Taylor, P. D., . . . Cadena, C. D. Hill, R. (1994). Theory of geolocation by light levels. In B. J.
(2017). Fuel loads acquired at a stopover site influence LeBouef, & R. M. Laws (Eds.), Elephant seals: Population
the pace of intercontinental migration in a boreal song- ecology, behaviour, and physiology (pp. 227236).
bird. Scientific Reports, 7, 3405. Available from https:// Berkeley, CA: University of California Press.
doi.org/10.1038/s41598-017-03503-4. Hobson, K. A. (1999). Tracing origins and migration of
Greenberg, R., & Marra, P. P. (2005). Birds of two worlds: wildlife using stable isotopes: A review. Oecologia, 120,
The ecology and evolution of temperate-tropical migration 314326. Available from https://doi.org/10.1007/
systems. Baltimore, MD: Johns Hopkins University s004420050865.
Press. Hobson, K. A. (2003). Making migratory connections with
Hallworth, M. T., & Marra, P. P. (2015). Miniaturized GPS stable isotopes. In P. Berthold, E. Gwinner, & E.
tags identify non-breeding territories of a small breed- Sonnenschein (Eds.), Avian migration (pp. 379391).
ing migratory songbird. Scientific Reports, 5, 11069. Berlin Heidelberg, New York: Springer-Verlag.
Available from https://doi.org/10.1038/srep11069. Hobson, K. A. (2005). Stable isotopes and the determina-
Hario, M., & Uuksulainen, J. (1993). Mercury load accord- tion of avian migratory connectivity and seasonal inter-
ing to moulting area in primaries of the nominate race actions. Auk, 122, 10371048. Available from https://
of the Lesser Black-backed Gull. Larus f. fuscus. Ornis doi.org/10.1642/0004-8038(2005)122[1037:SIATDO]2.0.
Fennica, 70, 3239. CO;2.
Harrison, X. A., Blount, J. D., Inger, R., Norris, D. R., & Hobson, K. A., & Wassenaar, L. I. (1997). Linking breeding
Bearhop, S. (2011). Carry-over effects as drivers of fit- and wintering grounds of Neotropical migrant song-
ness differences in animals. Journal of Animal Ecology, birds using stable hydrogen isotopic analysis of feath-
80, 418. ers. Oecologia, 109, 142148.
Hays, G. C., Akesson, S., Godley, B. J., Luschi, P., & Hobson, K. A., Van Wilgenburg, S. L., Ferrand, Y.,
Santidrian, P. (2001). The implications of location accu- Gossman, F., & Bastat, C. (2013). A stable isotope (δ2H)
racy for the interpretation of satellite tracking data. approach to deriving origins of harvested Woodcock
Animal Behavior, 61, 10351040. (Scolopax rusticola) taken in France. European Journal of
Heckscher, C. M., Halley, M. R., & Stampul, P. M. (2015). Wildlife Research, 59, 881892. Available from https://
Intratropical migration of a Nearctic-Neotropical migra- doi.org/10.1007/s10344-013-0742-7.
tory songbird (Catharus fuscescens) in South America Holland, R. A., Thorup, K., Vonhof, M. J., Cochran, W. W.,
with implications for migration theory. Journal of & Wikelski, M. (2006). Bat orientation using Earth’s
Tropical Ecology, 31, 285289. Available from https:// magnetic field. Nature, 445, 702. Available from
doi.org/10.1017/S0266467415000024. https://doi.org/10.1038/444702a.

Jahn, A. E., Levey, D. J., Hostetler, J. A., & Mamani, A. M. Kirkeby, C., Wellenreuther, M., & Brydegaard, M. (2016).
(2010). Determinants of partial bird migration in the Observations of movement dynamics of flying insects
Amazon Basin. Journal of Animal Ecology, 79, 983992. using high resolution lidar. Scientific Reports, 6, 29083.
Available from https://doi.org/10.1111/j.1365-2656. Available from https://doi.org/10.1038/srep29083.
2010.01713.x. Levey, D., & Stiles, F. G. (1992). Evolutionary precursors of
Janssens, E., Dauwe, T., Bervoets, L., & Eens, M. (2002). long-distance migration: Resource availability and move-
Inter- and intraclutch variability in heavy metals in ment patterns in Neotropical landbirds. American
feathers of Great Tit nestlings (Parus major) along a pol- Naturalist, 140, 447476. Available from https://doi.org/
lution gradient. Archives of Environmental Contaminants 10.1086/285421.
and Toxicology, 43, 323329. Available from https:// Lajtha, K., & Michener, R. H. (Eds.), (1994). Stable isotopes
doi.org/10.1007/s00244-002-0138-2. in ecology and environmental science. Oxford: Blackwell
Jouventin, P., & Weimerskirch, H. (1990). Satellite tracking Scientific Publications.
of Wandering Albatrosses. Nature, 343, 746748. Liu, J., Xiao, H., Lei, F., Zhu, Q., Qin, K., Zhang, X.-W., . . .
Available from https://doi.org/10.1038/343746a0. Gao, G. F. (2005). Highly pathogenic H5N1 influenza
Kaimal, B., Johmson, R., & Hannigan, R. (2009). virus infection in migratory birds. Science, 309, 1206.
Distinguishing breeding populations of Mallards (Anas Available from https://doi.org/10.1126/science.1115273.
platyrhynchos) using trace elements. Journal of Marra, P. P., Hobson, K. A., & Holmes, R. T. (1998).
Geochemical Exploration, 102, 176180. Available from Linking winter and summer events in a migratory bird
https://doi.org/10.1016/j.gexplo.2009.02.001. by using stable-carbon isotopes. Science, 282,
Kaminski, R. M., & Gleusing, E. A. (1987). Density and 18841886. Available from https://doi.org/10.1126/
habitat related recruitment in mallards. Journal of science.282.5395.1884.
Wildlife Management, 51, 141148. Available from Marra, P. P., Norris, D. R., Haig, S. M., Webster, M. S., &
https://doi.org/10.2307/3801645. Royle, J. A. (2006). Migratory connectivity. In K. R.
Karasov, W. H., & Martinez del Rio, C. (2007). Isotope ecol- Crooks, & M. A. Sanjayan (Eds.), Connectivity conservation.
ogy. Physiological ecology (pp. 433478). Princeton, NJ: (pp. 157183). New York: Cambridge University Press.
Princeton University Press. Martin, T. M., Chades, I., Arcese, P., Marra, P. P.,
Kardynal, K. J., & Hobson, K. A. (2017). The pull of the Possingham, H. P., & Norris, D. R. (2007). Optimal
Central Flyway? Veeries breeding in western Canada conservation of migratory birds. PLoS One, 2. Available
migrate using an ancestral eastern route. Journal of Field from https://doi.org/10.1371/journal.pone.0000751, e571.
Ornithology, 88, 262273. Available from https://doi. Mayor, S. J., Guralnick, R. P., Tingley, M. W., Otegui, J.,
org/10.1111/jofo.12207. Withey, J. C., Elmendorf, S. C., . . . Schneider, D. C.
Kays, R., Crofoot, M. C., Jetz, W., & Wikelski, M. (2015). (2017). Increasing phenological asynchrony between
Terrestrial animal tracking as an eye on life and planet. spring green-up and arrival of migratory birds.
Science, 348. Available from https://doi.org/10.1126/ Scientific Reports, 7, 1902. Available from https://doi.
science.aaa247, aaa2478. org/10.1038/s41598-017-02045-z.
Kelly, J. F., Atudorei, V., Sharp, Z. D., & Finch, D. M. (2002). Mikkola, K. (1971). Pollen analysis as a means of studying
Insights into Wilson’s Warbler migration from analyses the migrations of Lepidoptera. Annales Entomologici
of hydrogen stable-isotope ratios. Oecologia, 130, 216221. Fennici, 37, 136139.
Available from https://doi.org/10.1007/s004420100789. Michener, R. M., & Lajtha, K. (Eds.), (2007). Stable isotopes
Kelly, J. F., Ruegg, K. C., & Smith, T. B. (2005). Combining in ecology and environmental science (2nd ed.). Oxford:
isotopic and genetic markers to identify breeding origins Blackwell Publishing.
of migrant birds. Ecological Applications, 15, 14871494. Mitchell, G. W., Guglielmo, C. G., & Hobson, K. A. (2015).
Available from https://doi.org/10.1890/04-1704. Measurement of whole-body CO2 production in birds
Kelsall, J. P. (1984). The use of chemical profiles from using real-time laser-derived measurements of hydro-
feathers to determine the origins of birds. In J. Ledger gen (δ2H) and oxygen (δ18O) isotope concentrations in
(Ed.), Proceedings of the fifth Pan-African Ornithological water vapor from breath. Physiological and Biochemical
Congress, Lilongwe, Malai, 1980 (pp. 501515). Zoology, 88, 599606.
Johannesburg: South African Ornithological Society. Moore, F. R., Smith, R. J., & Sandberg, R. (2005). Stopover
Ketterson, E. D., & Nolan, V., Jr. (1976). Geographic varia- ecology and intercontinental migrants: En route pro-
tion and its climatic correlates in the sex ratio of blems and consequences for reproductive performance.
eastern-wintering Dark-eyed Juncos (Junco hyemalis hye- In R. Greenberg, & P. Marra (Eds.), Birds of two worlds—
malis). Ecology, 57, 679693. Available from https:// The ecology and evolution of migration (pp. 251261).
doi.org/10.2307/1936182. Baltimore, MD: Johns Hopkins University Press.

REFERENCES 21
Moore, J. D., & Kremenetz, D. G. (2017). Migratory connec- Rakhimberdiev, E., Winkler, D. W., Bridge, E., Seavy,
tivity of American woodcock using band return data. N. E., Sheldon, D., Piersma, T., & Saveliev, A. (2016).
Journal of Wildlife Management, 81, 10631072. Available A hidden Markov model for reconstructing animal
from https://doi.org/10.1002/jwmg.21269. paths from solar geolocation loggers using templates
Morton, E. S. (1977). Intratropical migration in the Yellow- for light intensity. Movement Ecology, 3, 25. Available
Green Vireo and Piratic Flycatcher. Auk, 94, 97106. from https://doi.org/10.1186/s40462-015-0062-5.
Newton, I. (2006). Advances in the study of irruptive Ramenofsky, M., & Wingfield, J. C. (2007). Regulation of
migration. Ardea, 94, 433460. migration. BioScience, 57, 135143. Available from
Norris, D. R., & Marra, P. P. (2007). Seasonal interactions, https://doi.org/10.1641/B570208.
habitat quality and population dynamics in migratory Ramos, M. A. (1983). Seasonal movements of bird populations at
birds. Condor, 109, 535547. Available from https:// a neotropical study site in southern Veracruz, Mexico. Ph.D.
doi.org/10.1650/8350.1. dissertation. University of Minnesota, Minneapolis.
Norris, D. R., Marra, P. P., Kyser, T. K., Sherry, T. W., & Ramos, R., Sanz, V., Militão, T., Bried, J., Neves, V. C.,
Ratcliffe, L. M. (2004). Tropical winter habitat limits Biscoito, M., . . . González-Solı́s, J. (2015). Leapfrog
reproductive success on the temperate breeding migration and habitat preferences of a small oceanic
grounds in a migratory bird. Proceedings of the Royal seabird, Bulwer’s petrel (Bulweria bulwerii). Journal of
Society of London B, 271, 5964. Available from https:// Biogeography, 42, 16511664. Available from https://
doi.org/10.1098/rspb.2003.2569. doi.org/10.1111/jbi.12541.
Norris, D. R., Marra, P. P., Kyser, T. K., Royle, J. A., Ramos, M. A., & Warner, D. W. (1980). Analysis of
Bowen, G. J., & Ratcliffe, L. M. (2006). Migratory con- North American subspecies of migrant birds winter-
nectivity of a widely distributed Neotropical-Nearctic ing in Los Tuxtlas, southern Veracruz, Mexico. In A.
migratory songbird. Ornithological Monographs, 61, Keast, & E. S. Morton (Eds.), Migrant birds in the neo-
1428. Available from https://doi.org/10.2307/ tropics: Ecology, behavior, distribution, and conservation
40166836. (pp. 172180). Washington, DC: Smithsonian
Norris, D. R., Lank, D. B., Pither, J., Chipley, D., Ydenberg, Institution Press.
R. C., & Kyser, T. K. (2007). Trace elements identify Rappole, J. H., Derrickson, S. R., & Hubalek, Z. (2000).
wintering populations of a migratory shorebird. Migratory birds and the spread of West Nile virus in
Canadian Journal of Zoology, 85, 579583. Available from the western hemisphere. Emerging Infectious Diseases, 6,
https://doi.org/10.1139/Z07-024. 319328. Available from https://doi.org/10.3201/
Ochoas-Acuna, H., Sepulveda, M. S., & Gross, T. S. (2002). eid0604.000401.
Mercury in feathers from Chilean birds: Influence of Reed, K. D., Meece, J. K., Henkel, J. S., & Shukla, S. K.
location, feeding strategy, and taxonomic affiliation. (2003). Birds, migration and emerging zoonoses: West
Marine Pollution Bulletin, 44, 340349. Available from NileVirus, Lyme Disease, Influenza A and enteropatho-
https://doi.org/10.1016/S0025-326X(01)00280-6. gens. Clinical and Medical Research, 1, 512.
Owen, M., & Black, J. M. (1989). Factors affecting the sur- Ricklefs, R. E., Fallon, S. M., Latta, S. C., Swanson, B. L., &
vival of Barnacle Geese on migration from the breeding Bermingham, E. (2005). Migrants and their parasites: A
grounds. Journal of Animal Ecology, 58, 603617. bridge between two worlds. In R. Grrenberg, & P.
Available from https://doi.org/10.2307/4851. Marra (Eds.), Birds of two worlds (pp. 210221).
Parrish, J. R., Rogers, D. T., Jr., & Ward, F. P. (1983). Baltimore, MD: Johns Hopkins University Press.
Identification of natal locales of Peregrine Falcons (Falco Riley, J. R., Smith, A. D., Reynolds, D. R., Edwards, A. S.,
peregrinus) by trace element analysis of feathers. Auk, Osborne, I. H., Williams, J. L., . . . Poppy, G. M. (1996).
100, 560567. Tracking bees with harmonic radar. Nature, 379, 2730.
Perrins, C. M. (1970). Timing of birds breeding seasons. Available from https://doi.org/10.1038/379029b0.
Ibis, 112, 242255. Available from https://doi.org/ Robinson, S. K., Thompson, F. R., Donovan, T. M.,
10.1111/j.1474-919X.1970.tb00096.x. Whitehead, D. R., & Faaborg, J. (1995). Regional forest
Pulido, F. (2007). The genetics and evolution of avian fragmentation and nesting success of migratory birds.
migration. BioScience, 57, 165174. Available from Science, 267, 19871990. Available from https://doi.
https://doi.org/10.1641/B570211. org/10.1126/science.267.5206.1987.
Quinn, T. P. (2005). The behavior and ecology and Pacific Roff, D. A., & Fairbairn, D. J. (2007). The evolution and
Salmon and Trout. Seattle: American Fisheries Society, genetics of migration in insects. BioScience, 57,
Bethesda, MD and the University of Washington 155164. Available from https://doi.org/10.1641/
Press. B570210.

von Rönn, J. A., Harrod, C., Bensch, S., & Wolf, J. B. (2015). Strong, C., Zuckerberg, B., Betancourt, J. L., & Koenig,
Transcontinental migratory connectivity predicts para- W. D. (2015). Climatic dipoles drive two principal
site prevalence in breeding populations of the modes of North American boreal bird irruption.
European Barn Swallow. Journal of Evolutionary Biology, Proceedings of the National Academy of Sciences of the
28, 535546. Available from https://doi.org/10.1111/ United States of America, 112, E2795E2802. Available
jeb.12585. from https://doi.org/10.1073/pnas.1418414112.
Rubenstein, D. R., & Hobson, K. A. (2004). From birds to Stutchbury, B. J., Tarof, S. A., Done, T., Gow, E., Kramer,
butterflies: Animal movement and stable isotopes. P. M., Tautin, J., . . . Afanasyev, V. (2009). Tracking
Trends in Ecology and Evolution, 19, 256263. Available long-distance songbird migration by using geolocators.
from https://doi.org/10.1016/j.tree.2004.03.017. Science, 323, 896. Available from https://doi.org/
Ruegg, K. C., Anderson, E. C., Paxton, K. L., Apkenas, V., 10.1126/science.1166664.
Lao, S., Siegel, R. B., . . . Smith, T. B. (2014). Mapping Stutchbury, B. J., Siddiqui, R., Applegate, K., Hvenegaard,
migration in a songbird using high-resolution genetic G. T., Mammenga, P., Mickle, N., . . . Fraser, K. C.
markers. Molecular Ecology, 23, 57265739. Available (2016). Ecological causes and consequences of intratro-
from https://doi.org/10.1111/mec.12977. pical migration in temperate-breeding migratory birds.
Rundel, P. W., Ehleringer, J. R., & Nagy, K. A. (1988). American Naturalist, 188, S28S40. Available from
Stable isotopes in ecological research. Berlin: Springer-Verlag. https://doi.org/10.1086/687531.
Rushing, C. S., Hostetler, J. A., Sillett, T. S., Marra, P. P., Szep, T., Moller, A., Vallner, J., Kovacs, B., & Norman, D.
Rotenberg, J. A., & Ryder, T. B. (2017). Spatial and tem- (2003). Use of trace elements in feathers of sand martin
poral drivers of avian population dynamics across the Riparia riparia for identifying moulting areas. Journal of
annual cycle. Ecology, 98, 28372850. Available from Avian Biology, 34, 307320. Available from https://doi.
https://doi.org/10.1002/ecy.1967. org/10.1034/j.1600-048X.2003.03026.x.
Salomonsen, F. (1955). The evolutionary significance of Taylor, C. M., & Norris, D. R. (2010). Population dynamics
bird migration. Biologiske Meddelesler, 22, 162. in migratory networks. Theoretical Ecology, 3, 6573.
Sanpera, C., Ruiz, X., Moreno, R., Jover, L., & Waldron, S. Taylor, P. D., Crewe, T. L., Mackenzie, S. A., Lepage, D.,
(2007). Mercury and stable isotopes in feathers of Aubry, Y., et al. (2017). The Motus Wildlife Tracking
Audouin’s Gulls as indicators of feeding habits and System: A collaborative research network to enhance
migratory connectivity. Condor, 109, 268275. Available the understanding of wildlife movement. Avian
from https://doi.org/10.1650/0010-5422(2007)109[268: Conservation and Ecology, 12, 8. Available from https://
MASIIF]2.0.CO;2. doi.org/10.5751/ACE-00953-120108.
Shaffer, S. A., Tremblay, Y., Awkerman, J. A., Henry, Tuck, G. N., Polacheck, T., Croxall, J. P., Weimerskirch,
R. W., Teo, D. J., Anderson, D. A., . . . Costa, D. P. H., Prince, P. A., & Wotherspoon, S. (1999). The
(2005). Comparison of light- and SST-based geolocation potential of archival tags to provide long-term
with satellite telemetry in free-ranging albatrosses. movement and behaviour data for seabirds: First
Marine Biology, 147, 833843. results from Wandering Albatross Diomedea exulans of
Sibley, D. A. (2014). The Sibley guide to birds (2nd ed.). New South Georgia and the Crozet Islands. Emu, 99,
York: Alfred A. Knopf. 6068. Available from https://doi.org/10.1071/
Sillett, S., & Holmes, R. T. (2002). Variation in survivorship MU99008.
of a migratory songbird throughout its annual cycle. Urquhart, F. A., & Urquhart, N. R. (1978). Autumnal
Journal of Animal Ecology, 71, 296308. Available from migration routes of the eastern population of the mon-
https://doi.org/10.1046/j.1365-2656.2002.00599.x. arch butterfly (Danaus p. plexippus L.; Danaidae;
Smith, T. B., Clegg, S. M., Kimura, M., Ruegg, K. C., Mila, Lepidoptera) in North America to the overwintering
B., & Lovette, I. (2005). Molecular and genetic site in the Neovolcanic Plateau of Mexico. Canadian
approaches to linking breeding and wintering areas in Journal of Zoology, 56, 17591764. Available from
five Neotropical migrant passerines. In R. Greenberg, & https://doi.org/10.1139/z78-240.
P. Marra (Eds.), Birds of Two Worlds (pp. 222234). Visser, M. E., van Noordwijk, A. J., Tinbergen, J. M., &
Baltimore, MD: Johns Hopkins University Press. Lessells, C. M. (1998). Warmer springs lead to mis-
Stokely, J. M. (2005). The feasibility of utilizing the cellular timed reproduction in great tits (Parus Major).
infrastructure for urban wildlife telemetry. Ph.D. disserta- Proceedings of the Royal Society of London B, 265,
tion. Virginia Polytechnical Institute and Virginia State 18671870. Available from https://doi.org/10.1098/
University. rspb.1998.0514.

FURTHER READING 23
Webster, M. S., & Marra, P. P. (2005). The importance of under- breeding in a songbird. Oecologia, 181, 413422.
standing migratory connectivity. In R. Greenberg, & P. P. Available from https://doi.org/10.1007/s00442-015-
Marra (Eds.), Birds of two worlds: the ecology and evolution of 3527-8.
temperate-tropical migration systems (pp. 199209). Yohannes, E., Križanauskienė, A., Valcu, M., Bensch, S., &
Baltimore, MD: Johns Hopkins University Press. Kempenaers, B. (2009). Prevalence of malaria and
Webster, M. S., Marra, P. P., Haig, S. M., Bensch, S., & related haemosporidian parasites in two shorebird spe-
Holmes, R. T. (2002). Links between worlds: cies with different winter habitat distribution. Journal of
Unraveling migratory connectivity. Trends in Ecology Ornithology, 150, 287291. Available from https://doi.
and Evolution, 17, 7683. Available from https://doi. org/10.1007/s10336-008-0349-z.
org/10.1016/S0169-5347(01)02380-1. Yohannes, E., Mihai Valcu, R., Lee, W., & Kempenaers, B.
Wikelski, M., Kays, R. W., Jeremy Kasdin, N., Thorup, K., (2010). Resource use for reproduction depends on
Swenson, J. A., & Smith, G. W., Jr (2007). Going wild: spring arrival time and wintering area in an arctic
What a global small-animal tracking system could do breeding shorebird. Journal of Avian Biology, 41,
for experimental biologists. Journal of Experimental 580590. Available from https://doi.org/10.1111/
Biology, 210, 181186. Available from https://doi.org/ j.1600-048X.2010.04965.x.
10.1242/jeb.02629. Zaugg, S., Saporta, G., van Loon, E., Schmaljohann, H., &
Wikelski, M., Moskowitz, D., Adelman, J. S., Cochran, J., Liechti, F. (2008). Automatic identification of bird tar-
Wilcove, D. S., & May, M. L. (2006). Simple rules guide gets with radar via patterns produced by wing flap-
dragonfly migration. Biology Letters, 2, 325329. ping. Journal of the Royal Society Interface, 5, 10411053.
Available from https://doi.org/10.1098/rsbl.2006.0487. Available from https://doi.org/10.1098/rsif.2007.
Wikelski, M., Tarlow, E. M., Raim, A., Diehl, R. H., Larkin, 1349.
R. P., & Visser, G. H. (2003). Costs of migration in free-
flying songbirds. Nature, 423, 704. Available from
https://doi.org/10.1038/423704a. Further Reading
Winkler, D. W., Dunn, P. O., & McCulloch, C. E. (2002).
Predicting the effects of climate change on avian life- Murray, M. R., & Fuller, D. L. (2000). A critical review of
history traits. Proceedings of the National Academy of the effects of marking on the biology of vertebrates.
Sciences of the United States of America, 99, 1359513599. In L. Boitani, & T. K. Fuller (Eds.), Research techniques in
Available from https://doi.org/10.1073/pnas.212251999. animal ecology: Controversies and consequences. Methods
de Witt, C. (2002). An overview of brominated flame retar- and cases in conservation science (pp. 1564). New York:
dants in the environment. Chemosphere, 46, 583624. Columbia University Press.
Available from https://doi.org/10.1016/S0045-6535(01) Weisshaupt, N., Lehmann, V., Arizaga, J., & Maruri, M.
00225-9. (2017). Radar wind profilers and avian migration: A
Woodworth, B. K., Newman, A. E. M., Turbek, S. P., qualitative and quantitative assessment verified by
Dossman, B. C., Hobson, K. A., Wassenaar, L. I., . . . thermal imaging and moonwatching. Methods in Ecology
Norris, D. R. (2016). Differential migration and the link and Evolution, 8, 11331145. Available from https://
between winter latitude, timing of migration and doi.org/10.1111/2041-210X.12763.

C H A P T E R
2
Introduction to Conducting
Migration Studies
Leonard I. Wassenaar
International Atomic Energy Agency, Vienna, Austria
2.1 INTRODUCTION commercial, government, or university labora-

tories that measure samples on a fee-for-service
Since publication of the first edition of basis. This leads to a discussion between the
Tracking Animal Migration with Stable Isotopes a isotope analyst, who may have little knowledge
decade ago, the application of stable isotopes about the project, and an ecologist who may
(δ13C, δ2H, δ15N, δ18O, δ34S, 87Sr) in ecological have little knowledge about pertinent details
sciences has blossomed into numerous and of stable isotope measurements. The process
diverse studies of terrestrial and aquatic ani- usually begins with an enquiry from an ecolo-
mal migrations (Vander Zanden, Soto, Bowen, gist interested in using stable isotopes to inves-
& Hobson, 2016). Similarly, there have been tigate animal movement, but who has little
significant advances in stable isotope method- familiarity with what to do from an isotope
ologies along the way. This chapter briefly analytical perspective. What follows is a dia-
reviews the light stable isotope measurement logue that forms the framework of this chap-
terminology, methods and practices currently ter. What are stable isotopes? Which isotopes
applicable for use in migration ecology. could be used to answer research questions for
The analytical aspects of conducting migratory species of interest? What animal tis-
stable isotope measurements for migration sues are best used for isotope analysis? How
studies are particularly important to consider, to prepare the samples? How much sample
especially since few biologists and ecologists mass is needed? Are there caveats? How much
operate their own stable isotope instrumenta- will it cost? Why did duplicates give such dif-
tion laboratories. Most ecologists rely on ferent results?

26 2. INTRODUCTION TO CONDUCTING STABLE ISOTOPE MEASUREMENTS FOR ANIMAL MIGRATION STUDIES
These are relevant questions that should 2.1.1 What are Light Stable Isotopes?
be considered at the beginning of a project to
help ensure a successful outcome using Many of the individual elements of the peri-
stable isotopes. Addressing crucial analytical odic table have a range of stable isotopes, i.e.,
problems mid-way, after avoidable problems variants of one element having the same
emerge, or forging ahead without a good number of protons, but differing only in their
understanding of what stable isotope analy- number of neutrons (Criss, 1999; Hoefs, 2015;
ses can or cannot do, or measuring tissues Sharp, 2017). However, there are very few ele-
without understanding their dynamics, can mental stable isotopes that are easily measur-
lead to some painful discussions (Meier- able or of practical use in animal migration
Augenstein, Hobson, & Wassenaar, 2013; studies. The most useful elements are the so-
Wunder et al., 2009). Because of the high called “light isotopes” of carbon, hydrogen,
potential for misunderstandings, this chapter nitrogen, oxygen, and sulfur (CHNOS), impor-
attempts to bridge the gap by arming the tant elements which constitute the proteina-
ecologist with key information to ensure a ceous and building blocks of life: biosphere
basic understanding of the terminology and (plants, animals), hydrosphere (H2O) and
methods and issues around isotopic measure- atmosphere (N2, O2, and H2O). These five ele-
ments. Thereby, the ecologist gains confi- ments, in varying proportions, constitute
dence that the isotopic assays selected are almost 100% of the dry mass of animal and
appropriate, conducted correctly, and that plant tissues, ranging from B50% for carbon,
the results hopefully lead to meaningful and to around 6% for hydrogen in proteins
quantitative spatial information regarding (Table 2.1). Each of these elements have two or
migratory animal movement. more stable isotopes whose ratios vary widely
in nature. Other “heavier” elements (e.g., Hg,
TABLE 2.1 Approximate Elemental Abundances as (dry) Mass Fraction in wt. %, the Stable Isotope Ratios of
Interest, and Expected Stable Isotopic Ranges for Bulk Tissues (e.g., α- or β-Keratins Like Hair or Feathers) Commonly
Used in Migratory Research
Element Wt. % Isotope Ratios δ-Range Mass Required
LIGHT ISOTOPES
Carbona 3040 wt. % 13
C/12C 25 to 265m (PDB) 0.21.5 mg
Oxygen b
2740 wt. % 18 16
O/ O 110 to 130m (VSMOW) 0.20.5 mg
Nitrogen a
1219 wt. % 15
N/ N 14
22 to 125m (Air) 0.51.5 mg
Hydrogen b
68 wt. % 2 1
H/ H 2250 to 190m (VSMOW) 0.10.4 mg
Sulfur 520 wt. % 34 32
S/ S 220 to 130m (CDT) 12 mg
HEAVY ISOTOPES
Strontium ,100 2 x ppb 87
Sr/86Sr Absolute ratios 230 mgc
a
C and N isotopes are generally obtained simultaneously by CF-IRMS, but need sufficient sample obtain enough N.
b
H and O can be obtained simultaneously by pyrolytic methods.
c
Dependent on the Sr concentrations determined beforehand.
Keratins commonly contain amino acids such as glycine, alanine, and cysteine. Mass of sample typically required for each isotopic assay
also applies to other biological tissues like muscle, claw, and blood. The primary isotopic reference materials are PeeDee Belemnite (PDB),
Vienna Standard Mean Ocean Water (VSMOW), atmospheric N2 (AIR), and Canyon Diablo Triolite (CDT).

2.1 INTRODUCTION 27
Pb, Fe, and Sr) also have stable isotopes that blood, and claws) cannot have stable isotope
may be useful, but these elements typically ratios measured on bulk material directly,
occur at trace element (,ppmppb) mass frac- instead milligram-sized subsamples must be
tions and do not form the core of tissue protein sampled and quantitatively converted into
structure. The heavier isotopes are also diffi- ultrapure analyte gases like CO2, N2, or H2. In
cult to extract, easily contaminated, and costly other words, stable isotope assays are sample
to measure, or otherwise exhibit very little destructive. The second implication is mea-
useful isotope ratio differences in nature sured isotope ratios are obtained to 46
(Chapter 1: Animal Migration: A Context for decimal places, numbers which are difficult to
Using New Techniques and Approaches). One examine, and are not Système International
particularly useful “heavy isotope” is 87Sr, a (SI) quantities. Instead, by convention,
trace element that can be extracted from researchers receive their results tabulated as
tissues and has controlled geological distribu- positive or negative “δ-values” (delta values).
tions that may be utilized (Table 2.1; The δ-values are defined as the part-per-
Chapter 3: Isoscapes for Terrestrial Migration thousand (m, or per mille) deviation of the
Research, Chapter 4: Application of Isotopic ratios from a “zero-point” primary reference
Methods to Tracking Animal Movements). material (Fry, 2006; Hoefs, 2015). The idea of
The CHNOS elements have one highly “deviations in ratio differences” is even more
abundant “light” isotope (e.g., 12C; 98.894%) confusing to those accustomed to and who
and one or more “heavier” or rare isotopes prefer SI concentration units like μg/kg or
(e.g., 13C; 1.1056%) (Criss, 1999; Hoefs, 2015). mg/L. Because mass spectrometers measure
The ratios of the rare to common isotope vary very small isotope ratio differences between
minutely in natural materials due to a variety two analyte gases, an examination of the
of biogeophysical processes, but these tiny var- standard δ-equation reveals what the δ-value
iations in ratios can be used by scientists to really means:
great advantage (Brand & Coplen, 2012). It is
Ratiosample
exceedingly difficult to directly measure the δX in m 5 1 (2.1)
absolute isotope ratios, or the concentrations of Ratiostandard
one stable isotope in a tissue. To overcome
this, it was recognized over 70 years ago that it where X is the stable isotope of interest
was far easier and convenient to measure the (δ2HVSMOW, δ13CPDB, δ15NAIR, etc.) expressed
relative differences in rare-to-abundant isotope in m (Coplen, 2011). The right side of the equa-
ratios of a sample converted to a pure gas (e.g., tion is the isotopic ratio of the sample (gas) rel-
CO2, H2, SO2, and N2) compared to an identi- ative to a laboratory standard gas measured by
cal reference gas with known ratios, by using the mass spectrometer for the isotopes of inter-
isotope-ratio mass-spectrometers, or IRMS est (13C/12C, 18O/16O, 2H/1H). The ratio of the
(McKinney, McCrea, Epstein, Allen, & Urey, laboratory standard gas must be determined
1950). This is a first point of confusion for by calibration to an appropriate primary refer-
newcomers; stable isotopes are assays compar- ence material (usually defined to be “zero” m).
ing isotope ratios to isotope ratios (Fry, 2006). Because isotope-ratio differences lie in the
Generally, gas source IRMSs are used to 4th6th decimal place, the results from
measure the light isotopes in all environmental Eq. (2.1) are simply multiplied by 1000 to
materials. This has important implications for obtain numbers having 12 decimal places
the migration ecologist. The first implication is (e.g., δ2H 5 2150.1m). Because the primary ref-
that their biological samples (feather, muscle, erence materials and their ratios act as our

zero point, the reported δ-values may be nega- Fortunately, isotope fractionation of the light
tive or positive values, only because they are isotopes in nature is not only widespread, but
relative to the ratios of the primary reference highly diverse, and often characteristic in both
materials (Groning, 2004). Negative δ-values magnitude and its direction. Many examples
do not mean there are negative concentrations of isotope fractionation in environmental sys-
of the isotopes, and for simple 2- or 3-isotope tems are found in the recommended textbooks
systems one can easily convert negative listed below.
δ-values into positive mass fractions like ppm For the migratory ecologist, isotope fraction-
(Fry, 2006; Speakman, 1997). The primary ation usually involves using measured δ-value
reference material anchors were established differences between two or more related bulk
decades ago based on arbitrary natural materi- substrates because samples are often of neces-
als that were limited in quantity or are now sity “bulk” tissue materials with no clearly
exhausted, hence laboratories need to measure defined chemical stoichiometry that can be fol-
samples against appropriate well-calibrated sec- lowed (e.g., comparing feather vs food). For
ondary reference materials (Paul, Skrzypek, & example, in the global H-isotope system there
Forizs, 2007). are consistent and large δ-value “offsets”
Recently there was a proposal to replace the between water, plants, and bulk tissues of an
classical m (per mille) symbol with an SI com- organism. These isotope fractionations are here
pliant unit called the Urey (Ur) (Brand & more appropriately referred to as net isotopic
Coplen, 2012). Adoption of the Ur remains discriminations, since every hydrogen pool and
inconsistent in the scientific literature, hence in chemical form of hydrogen in all three sub-
this book we retain the conventional m symbol. strates cannot be fully known. Net isotope dis-
The conversion is 1m 5 1 mUr. crimination, while a needed oversimplification,
allows us to exploit stable isotopic measure-
ments in studying animal migration (see
Chapters 36). As with all oversimplifications,
2.1.2 Isotope Fractionation
the peril lies in the assumed details, and a
and Discrimination researcher may be required to defend the
Mass dependent isotope fractionation (Criss, assumptions with controlled experimentation
1999; Hoefs, 2015; Sharp, 2017) occurs when a (diet to tissue experiments), and maintain a
chemical, biological, or physical process drives critical eye toward what the measured isotopic
a change in the isotope ratios of the source data really represents.
material and/or reactant due to slight chemi-
cal differences arising from the subtle differ-
ences in isotopic masses. There are two
2.1.3 Isotope Mass Spectrometry
primary kinds of isotope fractionation pro-
Instrumentation
cesses. Unidirectional or irreversible isotope
fractionation is referred to as kinetic isotope The type of IRMS instrumentation currently
fractionation, whereas equilibrium fractionation is available to ecologists nowadays are predomi-
chemical reactions that are fully or partly nantly continuous-flow isotope-ratio mass-
reversible (Hoefs, 2015). In short, if there was spectrometers (CF-IRMS), which gained wide-
no isotope fractionation in nature, all compo- spread adoption since the 1990s (Fry, Brand,
nents of the hydrosphere, atmosphere, and Mersch, Tholke, & Garritt, 1992; Matthews &
biosphere would have identical isotopic ratios Hayes, 1978). Compared to their ultra-high
and stable isotopic assays would be pointless. precision dual-inlet counterparts (DI-IRMS)

2.1 INTRODUCTION 29
developed in the 1950s, there are key advan- preparative automation. Even δ34S, a histori-
tages to CF-IRMS-based assays. The pros are cally difficult assay, can be conducted using
low cost analyses and rapid throughput, which CF-IRMS to fit-for-purpose analytical precision
are achieved by linking preparative modules for migration studies (, 6 0.4m).
like Elemental Analyzers (EAs) or High- For organic tissues used in animal migration
Temperature Thermochemical (HTC) reactors studies, milligram subsamples are burned to
for 13C 1 15N, or 2H, and 2H 1 18O isotopes. convert samples to CO2, H2O, SO2, or N2 gas
The trade-off used to be lower analytical preci- using EAs. Schematic illustrations of some con-
sion, although CF-IRMS assays nowadays can temporary CF-IRMS systems for C, N, H, and
exceed dual-inlet precision due to extensive O isotope analyses are depicted in Fig. 2.1.
Elemental Analyzer for δ13C and δ13N
Auto sampler
O2 injection He flow in He Reference gas
injectors
Water trap
Quartz Magnesium CO2 N2
Quartz tube wool perchlorate
Combustion Reduction
1050ºC 650ºC
GC column Pneumatic
Chromium oxide 50ºC 3m PoropakQ
50ºC needle valve
Copper
Quartz chips Purge
N2 CO2
Silvered cobaltous TCD
Cobaltic oxide
Quartz Quartz NUPRO
wool wool He diluter isolation
(Optional) valve
Stand-by CF-IRMS
valve
Elemental Analyzer for δ2H or δ18O
Auto sampler
Oxygen setup
He flow in He
Hydrogen setup Reference gas
injectors
Ceramic outer Open split H2 CO

tube
Diluter
Glassy carbon (Optional)
inner tube H2 CO Pneumatic
Furnace needle valve
>1100– NiC TCD
Purge
1300ºC
Glassy carbon
Glassy Chips GC column
carbon 50ºC NUPRO
Quartz 1 m 5 A molecular
/ Cr wire isolation
wool sieve packed
column
valve
Quartz
wool Stand-by CF-IRMS
*90ºC for Hydrogen
valve
FIGURE 2.1 Typical Elemental Analyzer (EA, top right) preparative interface to a CF-IRMS system for δ13C, δ15N, or
δ34S (top left, S configuration not shown) and δ2H and δ18O (bottom left) in organic samples (courtesy of Elementar
GmbH). The Uni-Prep device (bottom right) and autosampler for automated isothermal online “comparative equilibra-
tion” and sample drying mounted to an EA-IRMS (courtesy Eurovector SpA.)

Much of the recent instrumental develop- the most recent and promising developments
ments have focused on automation and multi- and the pros and cons of using CSIA for 13C,
ple (13C, 15N, 34S) isotope assays, decreasing 15
N, and 2H analyses of selected amino acids
the sample size requirements, and increasing for use in migratory studies.
throughput while maintaining high precision Since 2010, lower-cost laser-based isotope
and accuracy. Some mass spectrometer com- analyzers brought exciting potential for easy to
panies promote multielement stable isotopic operate instruments compared to IRMSs. Laser
13
C 1 15N 1 34S measurements on single sam- isotope analyzers have already largely dis-
ples (Mambelli et al., 2016), although in prac- placed IRMS for water isotopes and some
tice, for the ecologist such assays are more greenhouse gases (Lis, Wassenaar, & Hendry,
difficult for samples with varying C:N:S 2008). Despite their potential, currently no
elemental ratios. Radiogenic 87Sr isotopes, automated sample preparative devices exist to
conversely, are conducted by trace-element convert biological bulk tissue samples to CO2
solid source IRMS. Numerous papers and or H2O gas for laser-based isotope measure-
books describe the routine preparative and ments. Perhaps someday the connection of
combustion-based conversions of environmen- automated EAs to lasers for δ2H or δ13C assays
tal samples for stable isotope analyses by may be achievable (Koehler & Wassenaar,
IRMS. Some summaries and historical perspec- 2012).
tives are found in Volumes 1 and 2 of the For further reading about stable isotope-
Handbook of Stable Isotope Techniques (de Groot, ratio mass spectrometry and broader applica-
2004). The cost (USD) of stable isotope analy- tions of stable isotopes in environmental chem-
ses in 2018 range from $720 for 13C 1 15N, to istry, the reader is referred to classic textbooks
$1550 for 18O/2H, $50100 for 34S, to published over the past decades (Hoefs, 2015),
$100200 or more for 87Sr. Costs, however, and online textbooks like Principles of
often vary widely and can be reduced through Stable Isotope Geochemistry (Sharp, 2017). For
collaborative partnerships, or by the ecologist newcomers to environmental isotope measure-
undertaking some of the costly sample prepa- ments in ecology the following introductory
ration steps. books are highly recommended: Stable Isotope
Recent promising developments in IRMS anal- Ecology (Fry, 2006), Stable Isotopes in Ecology
yses involve compound-specific or molecule- and Environmental Science (Lajtha & Michener,
specific isotope analysis (CSIA-IRMS), where tis- 2007), Isoscapes: Understanding Movement,
sue samples are processed for C, N, or H isotope Pattern, and Process on Earth through Isotope
analysis of selected targeted compounds using Mapping (West, Bowen, Dawson, & Tu, 2009),
either gas chromatography (GC) or high-pressure Stable Isotope Forensics: An Introduction to the
liquid chromatography separation interfaces to Forensic Application of Stable Isotope Analysis
an IRMS (Fogel, Griffin, & Newsome, 2016). (Meier-Augenstein, 2011), and Groundwater
CSIA assays promise distinctive advantages over Geochemistry and Isotopes (Clark, 2015).
traditional bulk tissue analyses, but suffer from
difficult or laborious sample preparation (i.e.,
derivatization), require highly specialized 2.2 SAMPLE COLLECTION
instrumentation, and accordingly low sample AND PREPARATIVE METHODS
throughput due to long processing times
required. Chapter 7, Amino Acid Isotope For animal migration studies, H (and possi-
Analysis: A New Frontier in Studies of Animal bly O) isotopes are typically of paramount
Migration and Foraging Ecology, summarizes interest. These two isotopes are considered

2.2 SAMPLE COLLECTION AND PREPARATIVE METHODS 31
global-spatial assays, because the patterns of H a high degree of confidence in making H and
and O isotopes in terrestrial ecosystems are O isotopic predictions into areas where no
controlled by global-scale hydrologic processes observational data exist (Bowen, Wassenaar, &
that are seasonally and spatially predictable by Hobson, 2005; Terzer et al., 2013). The oceans
latitude and elevation over multiyear time- are global O and H reservoirs that are mostly
frames, and over spatial scales ranging from isotopically homogenous and highly distinc-
kilometer to continental scales (Fig. 2.2; tive from terrestrial freshwater ecosystems,
Chapter 3: Isoscapes for Terrestrial Migration which allows for clear marine versus terrestrial
Research). These patterns are revealed in the distinctions to be made. The C, N, S, and Sr
50-year long-term Global Network for Isotopes isotopes, conversely, are categorized as local-
in Precipitation (www.iaea.org/water) and its spatial assays, because there is no a priori
spatial water isotope maps (Terzer, Wassenaar, reason their isotope ratios vary predictably in
Araguás-Araguás, & Aggarwal, 2013). A prom- time or continuously across landscapes to the
inent feature of isotopic spatial predictability is same degree as H and O isotopes do, although
FIGURE 2.2 Predicted growing-season hydrogen isotope (δ2H) patterns in precipitation across the globe (Terzer et al.,
2013). These distinctive H isotope patterns are mirrored in plants and upper-level trophic organisms and forms the foun-
dational “isoscape” for tracking animal movement. Migratory movements are often North2South, spanning continental
or regional isotopic gradients. Oxygen isotopes exhibit a similar pattern. Hydrogen and oxygen isoscape maps and data
can be found in the Global Network of Isotopes in Precipitation www.iaea.org/water, with public access to GNIP data
and maps at https://nucleus.iaea.org/wiser. Additional isoscape datasets for marine and aquatic systems are compiled at
http://wateriso.utah.edu/waterisotopes.

they could exhibit exploitable patterns at some large-scale isotopic patterns to make geospatial
relevant ecosystem scale (Chapter 3: Isoscapes interpretations in migratory studies.
for Terrestrial Migration Research). Intrasample isotopic heterogeneity is defined as
stable isotopic variance that occurs at the tiny
mg to μg (or mm to cm) scale within or along
the length of one sample used for isotopic
2.2.1 Tissue Sample Considerations
analyses (Hobson et al., 2017). Intrasample iso-
for Isotopic Assays topic heterogeneity is also greater than instru-
The first issue facing scientists using mental analytical uncertainty. For migrant
CHNOS stable isotopes in migration research animals, the problem may be further exacer-
involves the type of samples to be collected for bated in slowly growing tissues while the ani-
the species of interest for isotopic analysis. mal is moving across large spatial distances or
There are many possibilities: hair, claw, mus- feeding in different isotopic biomes, and can
cle, blood, wings, fins, thorax, tissue punch, lead to distinctive isotopic trends within a sin-
carapace, etc., but two clear distinctions can gle tissue (which may also be useful). One
be made from an isotopic perspective: fixed example of useful intrasample isotopic pat-
versus dynamic tissues (sometimes referred to terns is in fish otoliths which can be used to
as metabolically inactive vs metabolically active, infer movement (Chapter 6: Isotopic Tracking
respectively). A discussion of the pros and of Marine Animal Movement). The isotope
cons and potential of fixed versus active analyst’s uninformed answer to the problem of
tissues are fully discussed in Chapter 4, intrasample isotopic heterogeneity is often to
Application of Isotopic Methods to Tracking “pulverize the sample” to produce a homoge-
Animal Movements. Further, from an isotopic nous powder, a practice that could seriously
measurement aspect, the researcher needs to complicate matters, as demonstrated below.
be fully aware of issues of inter- and intra- Two contrasting examples of intrasample
sample isotopic heterogeneity because of the isotopic heterogeneity are illustrated in Fig. 2.3.
large potential for obtaining confounding One panel shows δ2H data along the length of a
isotopic results. This issue applies to both bald eagle flight feather. These data show dis-
bulk- and compound-specific isotope analy- tinctive longitudinal H isotopic patterns in the
ses. In short, what and where you sample for feather related to its southern migratory move-
isotope measurements on a tissue is highly ment during feather growth. As the feather
relevant and requires forethought. grew, it gained more positive δ2H values as the
Intersample isotopic heterogeneity is defined as eagle moved southward from Canada into
isotopic differences that occur among similar the United States, hence recorded a “net migra-
tissues (i.e., feathers) on the same animal due tory track” inside a single “fixed” feather. By
to inherent variance in isotope discriminations contrast, a single feather from a lesser scaup
occurring during biochemical synthesis of showed no significant intrasample tissue iso-
tissues. We expect some isotopic intersample topic differences because it was fully grown at
variance, e.g., among flight feathers grown by the molt site. The scaup feather represents
one individual at a location. Secondly, we an excellent candidate to assess molt origins
expect intersample isotopic variance for similar regardless of the position on the feather where
feathers from different birds at the same the subsample for isotope assays was taken. But,
location. Intersample isotopic heterogeneity of had the researcher been unaware the eagle
both types is always greater than instrumental feather was growing en route, and took one sub-
error for all isotopes, but is ideally less than sample for isotopic analysis to determine natal

Scaup feather Bald eagle feather FIGURE 2.3 Intrasample hydrogen (δ2H) iso-
topic heterogeneity in migratory bird feathers.
–128 (R) –114 (R) The feather on the left is from a lesser scaup
–117 –120 (Aythya affinis) that grew this feather at one loca-
–120 tion, and a bald eagle (Haliaeetus leucocephalus),
–121 –117 –112 right, that grew the feather during southward
migration. Subsamples for δ2H were taken from
–116 –114 the vane up and down each side, and from the
top and bottom of the rachis (R).
–111 –110
–124
–123
10 cm
–111
–111
26 cm
–96
–94
–95
–124 –124 –100
–90 –95
–85
–79
–128 (R) –82
–93 (R)
origin, highly misleading results could occur. 2.2.2 Expressing Sample Isotopic
Movement during tissue growth is one reason Uncertainty
why “duplicate” samples taken from different
parts of the same feather or tissue (often unwit- The sample heterogeneity issues described
tingly assumed by a researcher to be homoge- earlier will lead to varying degrees of sample
nous “repeats”) can give startlingly different isotopic variance, and is usually expressed in
isotopic results. In other words, what might be the standard deviation (SD) or standard error
presumed to be a single fixed tissue sample (SE) of n measurements of sample replication
representing only one location may in fact be or using population data. The uncertainty of
an interesting spatial tracking “recorder.” individual- or population-based measure-
Most investigators try to avoid issues of sam- ments must be properly propagated into the
ple heterogeneity by always making measure- assignment models to obtain realistic geospa-
ments on the same tissue (e.g., a focal tail or tial assignment and associated uncertainty
wing feather) and measuring the same part of estimates (Chapter 8: Design and Analysis
that tissue (e.g., just the vane section of the for Isotope-Based Studies of Migratory
distal section) (Hutchinson & Trueman, 2006; Animals, Chapter 9: Isoscape Computation
Mazerolle, Hobson, & Wassenaar, 2005; Coplen and Inference of Spatial Origins With Mixed
and Qi, 2013). Yet another consideration is Models Using the R package IsoriX).
whether or not endogenous reserves are mobi- Correct expression of isotope analytical
lized into tissue formation versus local diet only uncertainty is often overlooked, and unfortu-
(Fox, Hobson, & Kahlert, 2009). nately, there is very little consistency among

stable isotope laboratories in determining and 2.2.3 Other Sample Collection Issues
reporting measurement uncertainty (Wassenaar
et al., 2018). It is possible, e.g., to repeat a Museum or archeological tissue samples,
feather sample 10 times for δ2H and calculate a often attractive for historical information
SD that is lower than the uncertainty inherent regarding a migrating or extinct species, could
in the primary reference material VSMOW. have been chemically treated or preserved in
The uncertainty budget of an isotopic formalin in the past. The researcher must be
measurement contains several obtainable vigilant to potential isotope fractionation from
components: (1) the uncertainty of the pri- sample degradation or by contamination from
mary reference anchor, (2) the uncertainty of isotopic exchange with the chemicals used or
the secondary calibration standard(s) used, artifacts from applied preservatives (Hobson,
and (3) optionally, the uncertainty obtained Gibbs, & Gloutney, 1997). Testing for treatment
from replicates of the same sample (intrasam- effects may be required. One study demon-
ple variance). The point is this: reported strated that acetone preservation of archived
sample isotopic uncertainty (Uc) cannot be dragonflies had no effect on δ2H or δ18O values
less than the summed uncertainty inherent in of their chitin (Hobson, Soto, Paulson,
the reference and calibration materials used, Wassenaar, & Matthews, 2012).
and should be determined and conservatively Ethical considerations include whether a
reported using an error propagation method, tissue sample may be obtained nonlethally. For
like the squared root of the sum of the many species, several milligrams of hair, nail,
squares: fin clip, plucking a feather, or muscle biopsy
could be safely taken without affecting the
Uc 5 O½ðUVSMOW Þ2 1 ðUCBS Þ2 1 ðUREPEATS Þ2 . . . health of the individual (Hayden et al., 2015)
or tissues are easily regrown. For small migra-
To illustrate, a feather is measured five tory species like insects (moths, dragonflies)
times for δ2H. The uncertainty of the primary the entire body may be required to obtain suf-
reference material to which all results are ficient sample, hence the animal is euthanized.
reported (VSMOW) is 6 0.3m, the uncer- This may be a minor issue or a serious concern
tainty of the laboratory keratin calibration in the case of endangered species. Animal care
standard (i.e., CBS keratin) is 6 0.9m, and and national or international regulations must
the SD of five feather repeats is 6 0.2m (i.e., be observed when tissue sampling.
less than VSMOW, and indicating very low Finally, isotopic results obtained for fixed
intrasample variance which is an important or dynamic tissues will need to be related to
feature). The propagated uncertainty is on-the-ground known-origin equivalent sam-
correctly reported, to an appropriate number ples or predictively using water isotope spa-
of significant digits, by: tial patterns (e.g., isotopic “base maps” or
Uc 5O½ð0:3Þ2 1ð0:9Þ2 1ð0:2Þ5 61:0 m ðnot60:2 mÞ “isoscapes”) to facilitate quantification of
geospatial movement. The assumptions and
If the laboratory used for isotope analyses challenges in making robust geospatial
does not clearly reveal how their uncertainty connections and species-specific isoscape
is reported, it is worth asking how it was maps for migratory studies are fully dis-
done, and armed with information about cussed in Chapter 3, Isoscapes for Terrestrial
the calibration and control standards, can Migration Research and Chapter 8, Design
be determined with error propagation and Analysis for Isotope-Based Studies of
postprocessing. Migratory Animals.

In summary, it is incumbent upon the decompose, the sample cleaning procedure is
researcher to carefully consider the sample uncomplicated. If there is dirt or adherent
type and isotopic assay most likely to satisfy material, samples can be washed with distilled
the requirement of answering, wholly or in water, and air, oven, or freeze-dried. Where
part, the migration research question. Data there are natural oils (hair, feathers), samples
scrutiny requires knowledge of the biology of must be cleansed using a solvent. Solvent
the species and possibly experimentation with cleaning is strongly recommended to remove
the species or tissue under study, e.g., to better surface oils for C and H isotopes because oil,
establish water-diet-tissue isotope discrimina- grease, or waxes are markedly different (usu-
tion, or to verify tissue isotopic heterogeneity. ally depleted in the heavy isotope) compared
It is unwise to blindly extrapolate experimen- to bulk tissue, and so could impart uncon-
tal findings regarding intrasample isotopic trolled negative isotopic biases (Hobson et al.,
homogeneity from one to a different species, 2017). Note that natural oils on feathers or hair
as illustrated in Fig. 2.3. However, for many typically do not usually contain S or N, so
passerines, waterfowl, insects, and fish the solvent cleaning is not needed for these isoto-
water-dietary-tissue isotope discriminations pic assays. For tissue trace element isotopes
for δ2H and δ13C are remarkably uniform (e.g., 87Sr) sample cleaning and contamina-
(Clark, Hobson, & Wassenaar, 2006; Hobson, tion of tissues is a serious concern, since
1999; Soto, Hobson, & Wassenaar, 2016). results are easily compromised by dust
Notably, stable isotopic assays are never particles and by handling in the field or
deceptive; it may be our understanding of laboratory—extensive “clean-lab” procedures
the isotope biochemistry of the organism and are usually needed for using trace element
its tissue that is lacking (Meier-Augenstein isotopes (Bortolotti, 2010; Font, Nowell,
et al., 2013). Pearson, Ottley, & Willis, 2007).
For soft tissues like muscle, blood, or liver,
the approach to cleaning and preservation pro-
cedures is not as clear. In the field, these sam-
2.2.4 Sample Cleaning and Storing ples are usually stored cold, frozen, or
Once appropriate tissue samples have been somehow preserved to avoid decomposition.
collected there are questions concerning stor- In the laboratory, a longstanding debate
age and subsequent laboratory preparation for revolves around the fact that solvent or acid
isotope analysis. Field samples may be dirty, treatments alter the isotopic composition of the
matted, bloody, or greasy, so some degree of bulk living tissue to some degree by selectively
sample cleaning may be required to remove removing fatty acids, carbonate, or amino
extraneous contamination. While cleaning pro- acids, and thereby altering the original bulk
cedures for fixed tissues are often straightfor- isotopic composition (CNS) of the sample
ward, there is debate concerning what to do (Pinnegar & Polunin, 1999; Post et al., 2007;
with dynamic tissues (e.g., defatting, plasma Sotiropoulos, Tonn, & Wassenaar, 2004;
separation, and acid treatment). In short, any Yurkowski, Hussey, Semeniuk, Ferguson, &
sample treatment used should not alter the iso- Fisk, 2015). Opinions range from no cleaning,
topic integrity and signal of the sample, solvent cleaning, the use of C/N ratios to
whether it is a bulk tissue sample or targeted correct for fat content, or using empirical lipid
compound-specific analysis. correction models (Elliott, Davis, & Elliott,
For fixed tissues like feather, hair, nail, or 2014; Skinner, Martin, & Moore, 2016).
claws which are tough and do not easily Unfortunately, there is unlikely to ever be a

“one-size-fits-all” cleaning or lipid correction 2.2.5 Sample Weighing

approach that applies universally to all tissues.
Nevertheless, removal of lipids is generally All bulk tissue samples submitted for
agreed as an essential procedure to remove stable isotope measurements require careful
bias induced by 13C and 2H depleted lipids analytical microbalance weighing into special-
(Sessions, Burgoyne, Schimmelmann, & Hayes, ized cups or capsules prior to stable isotopic
1999; Soto, Wassenaar, & Hobson, 2013). By analysis. Many isotope laboratories perform
removing lipids, we are better assured that the weighing on a fee-for-service basis, and the cli-
H and C isotopic analysis is conducted on ent need not be overly concerned about the
bulk protein. technical details. Sample weighing, including
In some countries, samples shipped to labo- specialized capsules, cost $315 per sample.
ratories for isotopic analyses must be steril- Accurate and precise microbalance weighing
ized (e.g., Australia and New Zealand). of samples is critical as the target mass for
Testing has shown that autoclaving (high each isotope is instrument specific. This work
pressure 120 C and steaming for 15 minutes) cannot be automated and is painstakingly
has no measurable impact on the C, N, or H done by hand for each sample. However, given
isotopic composition of resilient keratins or the precautions above regarding intrasample
chitin (T. Horton, Personal communication). heterogeneity, clear subsampling instructions
However, depending on the resilience of the need to be given to the laboratory to avoid con-
fixed tissue, less extreme sterilization methods founding results. Cost savings and improve-
may be preferable (alcohol storage, irradia- ment in sample turnaround time may be
tion), the only criteria being that any steriliza- achieved if the researcher does the weighing
tion process will not alter the original isotopic and encapsulation. Only be aware microbalance
composition of the sample. It is not recom- weighing is tedious work and prone to external
mended to treat or wash samples with sodium interferences from drafty labs, static electricity,
hydroxide solutions, since NaOH decomposes and low relative humidity. Never begin any
organic samples. sample preparation work and microbalance
Both fixed and dynamic tissues for weighing without clear laboratory guidance.
stable isotopic analyses should be properly Target sample weights are usually deter-
stored before and after preparative procedures mined by the type of isotopic assay needed (C,
to prevent decomposition (and loss of isotope H, S) and the class of mass spectrometer that
information or fractionation). Storage may will be used (ranges from 100 μg to .5 mg).
include vials or envelopes at room tempera- Analytical microbalances therefore require a
ture, or samples stored frozen. If properly pre- readability of 6 0.001 mg or better. For each
pared and preserved, the isotopic integrity isotope, the mass of each sample depends on
with storage time will not be compromised. the sensitivity of the IRMS instrument and the
One known exception is formalin preservation mass fraction (dry weight %) of the element
which affects carbon and nitrogen isotope (i.e., C, H) in the sample. Never use a “gen-
ratios (Edwards, Turner, & Sharp, 2002; eral-purpose” laboratory balance for isotopic
Hobson et al., 1997; Barrow, Bjorndal, & Reich, weighing, and ensure readability and achiev-
2008). When short-term storage is required for able precision are compatible with the isotope
practical or field collection reasons, consider laboratory requirements. The types of sample
using 70% ethanolwater mixtures and plastic capsules also depend on the type of isotope
vials, then fully process the samples after analysis—usually they are ultra-high purity tin
returning from the field. (for C, N, and S) or silver capsules (for H and

O). Specialty sample capsules cost up to $2 dependencies are often a bit more forgiving
each and are available from scientific suppliers (i.e., target weight 1000 6 100 μg), but always
(see suppliers at www.isogeochem.com). depend on the instrument used and its opera-
It cannot be overstated—accurate and tional conditions. For δ2H, the δ-value mass
appropriately precise weighing is critical for dependency is highly sensitive for all types of
technical reasons. The main reason is the IRMS IRMS instruments and is often nonlinear, eas-
instrument-specific dependency of the mea- ily to 10m per 100 μg of sample in a positive or
sured δ-value on the mass of the element being negative direction. This means accuracy of tar-
analyzed. This effect arises from differential get weighing to 6 10 μg is usually needed for
gas pressures in the mass spectrometer ion H isotopes to reduce this significant source of
source or its tuning sensitivity, as shown in isotopic variance. In short, inaccurate or high
Fig. 2.4. The δ-value dependency on the mass weighing variance will translate to high isoto-
of sample used is often referred to as “ion pic variance in the sample results.
source linearity.” Linearity is not always quan- As noted, weighing pulverized or small
tified in laboratories on a per-run basis, nor is pieces of sample tissue to an accuracy of
the response always fully linear to allow for 6 10 μg is very difficult work; especially for
simple corrections. Usually, a laboratory tar- feathers, hair, or keratinaceous tissue powders
gets a specific mass that the analyst has prede- that are strongly affected by static electricity
termined gives the lowest ion source δ-value (i.e., powder explodes off the microspatula).
variance. Thus the person preparing samples The use of antistatic matting, cotton clothing,
must closely adhere to the sample target using a room air humidifier, and deploying
weight and uncertainty guidelines by the labo- antistatic guns and wrist straps are all useful
ratory. For C or N isotopes, these δ-value mass to reduce annoying static electricity effects.
–105 FIGURE 2.4 The dependence of instrumental

δ2H results on sample mass for two different
isotope-ratio mass spectrometers using homoge-
nized keratin powder. Weighing variance can
–110 result in δ2H changing by 10m per 100 μg of sam-
ple, well beyond acceptable uncertainty.
δD (o/oo) VSMOW
–115
–120
–125
–130
0 0.1 0.2 0.3 0.4 0.5 0.6

Keratin powder mass (mg)

To reiterate, inaccurate or imprecise weigh- species ranging from birds to mammals to

ing will cause serious isotopic variance for rep- insects and over many spatial scales (Bowen
licated results within and among laboratories, et al., 2005; Cormie, Schwarcz, & Gray, 1994;
and particularly for H isotopes. If the labora- Hobson & Wassenaar, 1997; Vander Zanden
tory has not established δ-mass dependencies, et al., 2016; Wassenaar & Hobson, 1998).
this is a good question to ask beforehand. If Using δ2H measurements requires obtaining
contracting out weighing to a laboratory, ask the H isotopic composition of the nonexchange-
for the sample weights recorded during the able H of tissue samples, which is carbon-
microbalance preparation and the elemental bound hydrogen that reflects the environmen-
composition (mass fraction) automatically tal H signal (water, diet) of the geographic
determined by the mass spectrometer (wt. %H, location of tissue formation. However, hydro-
C) along with the isotope δ-value data. The gen isotope analyses of organic samples suffer
mass fraction may not be routinely reported by from two seemingly minor, interrelated, but
laboratory unless explicitly requested. A cross critically important analytical concerns that do
plot of balance weight versus the % elemental not occur with the isotopes of CNS. These
(H or C) yield is a good proxy for assessing the concerns are the problem of (1) residual mois-
quality of weighing and its impact on isotopic ture in the sample plus, (2) uncontrolled H
variance. An example of target masses typically isotope exchange with exchangeable H continu-
used for stable isotope assays is summarized in ally interacting with ambient air moisture in
Table 2.1. A suggested weighing procedure is the laboratory. Trying to disentangle both
shown in Table 2.2. aspects to obtain the δ2H of nonexchangeable
One possible negative outcome of weighing H is not trivial, and has long been a subject of
errors or other sample handling mistakes is analytical method development starting in the
“isotopic outliers.” For example, samples from 1990s (Schimmelmann, Miller, & Leavitt, 1993;
a population of local birds are isotopically clus- Soto, Koehler, Wassenaar, & Hobson, 2017;
tered as expected, but there is an inexplicable Wassenaar & Hobson, 2000).
outlier. Either the outlier is correct, or the data Residual moisture are water molecules from
is faulty. The first thing the researcher should sample processing and from room air vapor
do is contact the isotope laboratory and double that quickly adhere to organic tissue samples
check if there was a weighing error or if sam- and are not fully removed due to insufficient
ples were somehow mistakenly switched. If all drying (Qi & Coplen, 2011; Soto et al., 2017).
is correct on the analytical side, repeating an Residual moisture H does not reflect the
outlier is warranted. If the outlier is correct, the location of tissue formation. The definition of
researcher can ponder the ecological signifi- residual moisture also varies considerably by
cance (e.g., recruitment of immigrants?). discipline, in the geotechnical literature it is
the gravimetric difference between the sample
after vacuum drying for 24 hours at 110 C.
Exchangeable H is the fraction of nonremova-
2.3 GLOBAL-SPATIAL ISOTOPES ble H atoms that form part of the organic
sample tissue plus any residual moisture. Both
2.3.1 Hydrogen Isotopes pools readily and differentially exchange H
Hydrogen (δ2H) isotopes are most fre- atoms with the 600022,000 ppm H2O in
quently used in studies of long-distance ambient laboratory air (Schimmelmann, 1991;
animal migration due to proven and successful Sessions & Hayes, 2005; Wassenaar &
track records for many diverse and iconic Hobson, 2000). Both residual moisture and

2.3 GLOBAL-SPATIAL ISOTOPES 39
TABLE 2.2 Example Procedure for Feather Stable Isotopic Analysis
Materials required: Analytical microbalance, tissue samples, clean culture tray(s), weighing utensils, methanol, Kimwipes,
tray template, tape, marker, silver or tin EA capsules.
1. Obtain a clean 96 position plastic culture tray (Elisa Plate) and print out an Excel sample template.
2. Ensure feathers are solvent cleaned (2:1 v/v chloroform/methanol 24-h soak and 2 3 rinse) to remove surface oils.
Air dry feathers in fume hood (24 h).
3. Clean weighing utensils using methanol and Kimwipes, allow to dry.
4. Cut off a small amount of feather material for analysis—cut samples from the same location on different feather
samples if possible (e.g., near tip). Feather pieces are best cut using small stainless steel surgical scissors.
5. Make sure the microbalance is clean and calibrated. Ensure the doors are closed when taring and weighing.
6. Tare a silver EA capsule,a handling only with tweezers, remove, and set on a clean metal surface. Use the smallest
available capsule that will contain the sample (e.g., 3.5 3 5.0 mm).
7. Using a spatula or tweezers, transfer a small amount of feather material into the capsule.
8. Reweigh, and continue adding or removing feather material until the target sample weight of 350 6 10 μg is obtained.b
With practice this will take ,35 min per sample. Ensure the microbalance is accurate and stay within prescribed
weight tolerance to avoid mass spectrometer linearity effects. Samples and references must be weighed to obtain
comparable elemental mass yield as the samples.
9. To seal the capsule, crimp the top of the capsule using a pair of straight edge tweezers and fold tightly (as if folding
down from the top of a paper bag). Then use the edge of the tweezers (use of two tweezers helps) to gently compact
the tin or silver capsule into a small cube or ball. There should be no stray edges, loose sides or sample material
poking out. Flattened samples (rather than cube/ball-shaped) or capsules with stray or loose edges can jam in an
autosampler, cross-contaminate samples, and ruin an analysis.
10. Record final sample weight and sample name in a spreadsheet. Place the sample capsule in the 96-position tray and
record the weight on the tray template. Clean all utensils with Kimwipes and methanol after each sample, air dry
briefly. Secure the lid of the sample tray with rubber bands or masking tape and label the tray when done. Ensure
samples cannot “jump” out of the cells when the Elisa lid is properly closed (some brands of trays allow this).
11. Record sample name and weights (in mg or μg) for each sample in the appropriate tray and its position (e.g., Tray 1,
pos A5). When completed, transfer this information to appropriate isotope laboratory sample submission form.
12. Use 3.5 3 5.0 mm silver or tin capsules designed for elemental isotope analysis. Suggested suppliers are Costech
(1-800-524-7219) and Elemental Microanalysis (1-800-659-9885).
a
Silver capsules must be used for δ2H and δ18O analyses, tin capsules for δ13C, δ 15N, and δ34S.
b
Consult isotope laboratory for target weights for each isotope.
Example illustrated for feathers for H assays, but also applies similarly to claw, hair, and can be adapted for the other isotopes (CNS).
exchangeable-H together form the fraction of Why does the net fraction of exchangeable H
exchangeable H (fex) of a sample. Unfortunately, matter? Because if left uncontrolled or ignored,
CF-IRMS instruments can only measure the identical samples and organic calibration stan-
total H of a sample and cannot distinguish dards will give incomparable δ2H or δ18O
between nonexchangeable H and fex H. The fex results at different laboratories or at different
immediately undergoes uncontrolled isotope times of the year depending on outside air
exchange whenever a sample is exposed to temperature and moisture sources. Residual
laboratory air. moisture heavily contributes to net fex of H.

An apt illustrative example of this confound- The preferred solution to remove moisture
ing effect is shown in a H-isotope intercompar- is by online heated vacuum drying (preferably
ison of keratins where the δ2H results .50 C), and using an approach where dried
(uncontrolled for fex) were unacceptably samples are never reexposed to air before their
incomparable among different laboratories isotope analysis. At 0.1 bar vacuum, the boil-
(Carter, Hill, Doyle, & Lock, 2009). ing point of water is reduced to only 50 C, a
It should be obvious that tissue samples temperature where many keratin and other
having 20% residual moisture content will organic tissues may be safely dried for
likely give a very different isotopic result for 12 hours without any detrimental isotopic
δ2H and δ18O and a higher fex than after being alteration. Residual moisture removal reduces
dried to ,1% residual moisture content. fex and hence what is left reflects the tissue
Ideally, tissue samples should have all the exchangeable H. Currently, the only commer-
residual moisture removed before H isotope cial drying solution that meets all of these
analysis, so that fex represents only the remain- criteria is the Uni-Prep device for online use
ing nonremovable tissue exchangeable-H. with HTC EAs and IRMS systems (Wassenaar,
Sample drying is therefore required, although Hobson, & Sisti, 2015).
drying practices range widely in both
approaches and efficacy. Drying methods
2.3.2 The “Comparative Equilibration”
include air drying, N2 flushing, desiccator or
vacuum desiccator drying (with various desic-
Approach to Organic δ 2H Analyses
cant chemicals), oven or vacuum-oven drying Another useful way to control net fex and to
(60110 C), and freeze-drying. Reported dry- determine the δ2H of nonexchangeable H is an
ing times range from hours to days, sometimes idea called comparative equilibration (Wassenaar
using the samples already sealed preweighed & Hobson, 2003). Comparative “equilibration”
in the capsules. While the drying methods leverages the fact that net H isotope exchange
used are rarely questioned, for H and O iso- between ambient lab air moisture and fex H in
topes there is an additional problem as tissue samples is fast and reaches equilibrium in
samples are reexposed to air again following ,96 hours at room temperature, and much
drying and during subsequent handling and faster at hotter temperatures (Bowen et al.,
weighing. Depending on the hygroscopic 2005; Soto et al., 2017; Wassenaar & Hobson,
properties of the tissue (and whether in pieces 2000). The “comparative” aspect means that
or powdered), samples will immediately begin each run of prepared unknown tissue samples
to reabsorb water moisture from the laboratory includes at least 23 matrix-equivalent organic
room air (Bowen, Chesson, Nielson, Cerling, & calibration standards whose nonexchangeable
Ehleringer, 2005). Moisture reabsorption con- H δ2H values are known (see below) and are
tinues until air-sample H2O equilibrium is prepared along with unknown samples.
reached during the handling, weighing, or During handling, as ambient laboratory mois-
when encapsulated in trays for hours or weeks, ture temporally changes its 2H content, both
or in the EA carousel. Drying and handling dif- the samples and references equally reequili-
ferences are one reason why fex values reported brate their total exchangeable H in an identical
for keratins are so wildly inconsistent in the fashion. These “comparatively equilibrated”
literature (ranging from 2% to 12%), likely standards and unknowns are then isolated
from incomplete or incomparable moisture from the atmosphere using a common zero-
removal methods and different laboratory blank EA autosampler and analyzed together
humidity conditions. by IRMS in a single analysis session.

The comparative equilibration method results for organic samples measured among
proved so successful it was commercialized laboratories worldwide, as depicted in Fig. 2.5
into an online sample preparation device called and Table 2.3 (Kelly, Bridge, Fudickar, &
the Uni-Prep (Wassenaar et al., 2015) (www. Wassenaar, 2009; Soto et al., 2017; Wassenaar &
eurovector.at). The Uni-Prep allows exhaustive Hobson, 2003).
online vacuum moisture removal over a wide In most laboratories H isotope measure-
range of isothermal temperatures (thereby ments of tissue samples are performed on H2
reducing fex), and subsequently facilitates con- derived from high-temperature EA pyrolysis
trolled equilibration with injected H2O vapor and CF-IRMS (Fig. 2.1). Pure H2 is used as the
under strictly isothermal conditions to “reset” sample analysis gas and the isotopic reference
the exchangeable H of all samples and stan- gas. A high-temperature EA and autosampler
dards to a uniform H isotopic composition. The is used to pyrolyze samples to a pulse of H2
advantage of the Uni-Prep approach is that gas (and N2 and CO gas). The pyrolysis col-
samples and standards are never left uncontrol- umn consists of a ceramic or Ni-carbide tube
lably exposed to ambient air moisture. Whether partially filled with glassy carbon chips and/
using the Uni-Prep or not, comparative equili- or chromium powder held at 11001350 C,
bration at room temperatures with air exposed followed by a molecular sieve GC column at
samples and standards can be used success- 80100 C (Gehre et al., 2015; Schimmelmann
fully to obtain comparable and robust δ2H et al., 2016; Soto et al., 2017). A GC column is
0 FIGURE 2.5 Laboratory intercom-

–20 Lab 1 parison of δ2H for 18 individual feathers
–40 Lab 2 using the “comparative equilibration”
–60 Lab 3 approach. No attempt was made to
homogenize the feathers nor were they
δD
–80
screened for intrasample heterogeneity.
–100
The average δ2H range per sample
–120
between laboratories was only 7m (with
–140 additional data from T. Jardine and R.
–160 Doucett).
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Sample #
TABLE 2.3 Laboratory Intercomparison of δ2H Results for Powdered Keratins Run Over Many Months Using the
“Comparative Equilibration” Approach
Lab 1 mean Lab 1 SD N Lab 2 mean Lab 2 SD n
Moose hair 2 163.5 2.1 84 2 164.7 2.4 54

Vole hair 2 106.2 2.3 85 2 105.1 2.6 21
Human hair (IAEA-085) 2 70 2.4 84 2 68.7 2.6 51
Keratin (Spectrum) 2 117.2 1.9 103 2 116.1 2.7 106
The results show excellent reproducibility and comparability of results among different laboratories
Courtesy T. Jardine and R. Doucette, unpublished data.

used to resolve sample H2 from N2 and CO. As a result, laboratories previously used an
All δ2H results are reported in units of per mil array of in-house, or inappropriate standards
(m) relative to the VSMOW-SLAP standard (like minerals, nonequivalent organic stan-
scale using new keratin reference materials dards), and generally inconsistent approaches
(Table 2.4). Recently, the use of Cr-based, or (Meier-Augenstein et al., 2013), leading to
C/Cr mixed bed reactors instead of pure C more confusion and unacceptably irreproduc-
chips has been shown to improve the δ2H ible results among laboratories.
results for nitrogenous organics like keratins Recently, a suite of four new keratinous
by reducing isotope fractionation that arises organic keratin standards was made formally
from formation of HCN (Coplen & Qi, 2016; available by Environment Canada and the
Gehre et al., 2015). The sample throughput for USGS. All are available through the USGS ref-
δ2H is fast, about 2.5 minutes per sample, and erence materials web portal (https://isotopes.
at low cost. Analytical uncertainty is often usgs.gov/lab/referencematerials.html). These
better than 6 1.0m. keratin standards (Table 2.4) span a wide
range of δ2H values, and are fit-for-purpose for
use with most proteinaceous tissues (keratin,
muscle), although this assertion should be
2.3.3 Organic Reference Materials
further tested by varied tissue type. These
for Nonexchangeable H new keratin standards are USGS42 and
The success of applying the “comparative USGS43 (Indian and Tibetan human hair),
equilibration” approach for δ2H assays funda- CBS (Caribou hoof), and KHS (Kudu horn).
mentally hinges on using matrix equivalent Their currently assigned δ2HVSMOW values
organic reference materials with similar are 272.9m, 244.4m, 2157.0m and 235.3m,
exchangeable H and hygroscopic properties as respectively (Coplen & Qi, 2016; Soto et al.,
the unknown samples, and having known δ2H 2017). Their assigned values and fex are tabu-
values carefully calibrated to the VSMOW- lated in Table 2.4.
SLAP scale. Hydrogen isotope calibration of It is important to note that these four kerati-
complex organics has long been problematic nous reference materials were released several
due to lack of appropriate reference materials. years ago, and were initially incompatible with
TABLE 2.4 Recommended Organic Keratin Standards δ2HVSMOW Values for Nonexchangeable H and for δ18OVSMOW
of Oxygen From 2017
Standard δ 2HVSMOW Wt. % H fex Ha δ 18OVSMOW Wt. % O Material Type Source
KHS 2 35.3 6 1.1m 6.6 6 0.1% 1.1 6 0.4 1 21.2 6 0.3 B22% Kudu Horn Powder Env. Canadab
USGS43 2 44.2 6 1.0m 6.3 6 0.1% 1.3 6 0.3 1 14.11 6 0.1 22% Human Hair Powder USGS
USGS42 2 72.2 6 0.9m 6.3 6 0.1% 1.0 6 0.3 1 8.56 6 0.1m 22% Human Hair Powder USGS
CBS 2 157.0 6 0.9m 6.6 6 0.1% 1.1 6 0.3 1 2.4 6 0.3m B22% Caribou Hoof Powder Env. Canadab
a
Determined by online vacuum drying at 105 C and dual-water equilibrations using the Uni-Prep device.
b
Prepared by L.I. Wassenaar and K.A. Hobson at Environment Canada, Saskatoon, Canada in 2010.
These keratin H and O standards can be purchased in 0.5 g quantities (approx. 1 year supply) from the USGS at https://isotopes.usgs.
gov/lab/referencematerials.html. Assigned δ2H values, fex, and dry mass fractions H are from Soto et al. (2017). Comparable δ2H and δ18O
values and dry mass fractions of H and O are from Qi et al. (2016) and Coplen and Qi (2013). Conversion equations for previous
(deprecated) δ2H values for all of the standards can be found in Soto et al. (2017).

each other. They underwent testing and revi- been fully described for wood reference mate-
sions, the currently accepted and unified δ2H rials (Qi, Coplen, & Jordan, 2016). The use of
results were published in 2017 (Soto et al., the Uni-Prep device will greatly facilitate and
2017). This means years of older publications improve the speed and reliability of dual-
using the initial USGS (USGS43/44) and equilibration experiments.
Environment Canada (CBS/KHS) assigned
δ2H values need their data corrected to main-
tain comparability with these unified stan-
dards going forward. The conversions for the
2.3.4 Oxygen Isotopes
old δ2H data are as follows: While oxygen isotopes of tissues should
For older (pre-2017) publications using dep- essentially give the same spatial information as
recated CBS and KHS δ2H values: H isotopes, there are some nonanalytical dis-
advantages to using δ18O. First, the global
Revised δ2 HVSMOW 5 10:774 1 0:852 δ2 Hold value range of δ18O values in animal tissues is rela-
(2.2) tively small, only around a 15m range. Along
For older (pre-2017) publications using dep- with an analytical uncertainty that is relatively
recated USGS 43 and USGS 44 δ2H values: high (B6 1.0m), oxygen isotopes have a lower
signal-to-noise ratio than hydrogen isotopes in
Revised δ2 HVSMOW 5 5:743 1 0:993 δ2 Hold value helping to resolve geospatial information.
(2.3) Second, the flow of oxygen in trophodynamics
is more complicated than for hydrogen. There
A good δ2H IRMS analysis template for are more sources of isotopically variable oxy-
comparative equilibration would be to use at gen (drinking water, air, food) and sinks
least 24 CBS and KHS as standards in each (exhaled CO2 and O2, urine, sweat, excrement),
run for conducting two-point normalization, which altogether cause more complicated net
and measure USGS42 and USG43 as unknown oxygen isotope fractionations between diet,
controls to ensure accurate results are obtained. drinking water, and tissue. Oxygen isotopes
Notably these keratinous standards need to be can be complicated to the point of being
prepared along with the unknown samples impossible for migratory animal tracking,
and undergo identical preparation procedures. partly depending on water use in the organism
Despite the excellent δ-range of these new ref- (Pietsch, Hobson, Wassenaar, & Tutken, 2011;
erence materials, in the future it would be use- Wolf, Bowen, & del Rio, 2011). On the other
ful to develop a fifth keratin standard with a hand, for some species, O and H isotopes are
δ2HVMSOW value in the 120m to 150m range reasonably well correlated (albeit noisily),
since many avian species and other organisms which suggests there may be some value in
exhibit such positive δ-values (De Ruyck, using oxygen as a second isotope dimension
Hobson, Koper, Larson, & Wassenaar, 2013). for migratory assignments. Excursions from
In the case where migrant tissue samples the expected linear H versus O relationship
are not closely matrix equivalent to these avail- may also be instructive to better separate dif-
able keratin standards, the researcher will be ferent migratory populations. The correlations
forced to conduct vapor equilibration experi- of δ18O in tissues compared to water isoscapes
ments with at least two isotopically distinctive for known origin samples are not nearly as
waters and run all samples twice to be able to good as hydrogen (Bowen et al., 2005; Hobson
calculate the δ2H of the nonexchangeable H. & Koehler, 2015; Hobson, deMent, Van
The dual-vapor equilibration approach has Wilgenburg, & Wassenaar, 2009; Nielson &

Bowen, 2010). Overall, there remains a paucity outside of the benzoic acid calibration range.
of δ18O data for migrant tissues, and the use of Accordingly, it is much preferable to use the
oxygen isotopes in conjunction with hydrogen recently developed keratinous materials
and other isotopes remains largely problematic USGS42, USGS43, CBS, and KHS (Table 5).
and unexplored. These new keratinous reference materials for
One clear benefit to oxygen isotopes is kera- δ18O are available for purchase from the USGS
tinous samples do not have any “exchange- (https://isotopes.usgs.gov/lab/referencema-
able-O” to be concerned about, however, the terials.html).
strong precautions concerning residual mois- The CF-IRMS analytical methods for H2 and
ture still apply. Exhaustive drying procedures CO are almost identical, and because H2 gas is
to remove adsorbed water will improve δ18O also produced in the same thermochemical
results. The drying procedures described reduction reaction as CO, some carbon-reactor
above for H still need to be followed, and are IRMS systems allow for dual δ18O and δ2H
preferably done online using a device like the assays on a single tissue sample (Hobson &
Uni-Prep so that dried samples never reabsorb Koehler, 2015). The only drawback is reduced
water from air during handling. sample throughput rate and a more compli-
Oxygen isotope analyses of organic tissues cated analytical setup involving multiple refer-
are performed using high-temperature pyroly- ence gases and IRMS peak jumping.
sis to CO and CF-IRMS (Fig. 2.1). Pure CO gas
is used as the sample analysis gas and isotopic
reference gas. A HTC analyzer and autosam-
2.4 LOCAL-SPATIAL ISOTOPES
pler is used to automatically pyrolyze submilli-
gram samples to a pulse of CO gas (and N2
2.4.1 Stable Carbon and Nitrogen
and H2 gas as also produced). The pyrolysis
column consists of a ceramic tube and glassy
Isotopes
carbon tube insert, filled to the hot zone with Stable-carbon and stable-nitrogen isotope
glassy carbon chips held at .12001350 C, assays can be considered local-spatial analyses,
followed by a molecular sieve GC column at but may be useful in delineating migratory
4060 C (Qi, Coplen, & Wassenaar, 2011). A populations or by providing additional infor-
1 m GC column is used to resolve sample CO mation indicating type of habitat (see
from the H2 and N2. All δ18O results are Chapter 1: Animal Migration: A Context for
reported in units of m relative to the VSMOW- Using New Techniques and Approaches,
SLAP standard scale using keratinous or other Chapter 4: Application of Isotopic Methods to
certified organic reference materials (i.e., IAEA Tracking Animal Movements). There may also
Benzoic Acid Standards) or by using primary be larger scale spatial patterns (Chapter 3:
reference waters VSMOW2 and SLAP2. The Isoscapes for Terrestrial Migration Research)
sample throughput rate for δ18O is about for δ13C and δ15N at regional or ecosystem
79 minutes per sample to allow for clean scales related to plant and ecozone types (C3
chromatographic separation of CO from inter- vs C4 dominated landscapes). Several studies
fering sample N2. that used δ2H were able to leverage additional
The IAEA benzoic acid standards have δ18O information from δ13C to improve the spatial
values of 123.1m relative to VSMOW (IAEA- resolution of migratory assignments in both
601) and 171.4m VSMOW (IAEA-604). It terrestrial and marine systems (Cherel &
should be noted that the δ18O of tissues gener- Hobson, 2007; Garcı́a-Pérez & Hobson, 2014;
ally fall between 1 10m and 120m, well Hobson, Wassenaar, & Taylor, 1999; Marra,

2.4 LOCAL-SPATIAL ISOTOPES 45
Hobson, & Holmes, 1998). The analysis of car- respectively. The hair C 1 N standards have
bon and nitrogen isotopes is routine in similar C: N ratios as proteinaceous tissues,
stable isotopes laboratories. The sample size and combined with the glutamic acids, fully
requirements range from 0.5 to 1.5 mg, span the isotopic range encountered for
depending on the C:N ratio of the sample. The migrant tissues in nature. These C 1 N refer-
cost is low, often around $10 for both isotopes. ence materials are available for purchase from
There are very few complications other than the USGS (https://isotopes.usgs.gov/lab/
the consideration about sample preparation referencematerials.html). Sample throughput is
(lipid removal is recommended for 13C), the between 9 and 14 minutes per sample, with an
general recommendation is to lipid extract analytical uncertainty around 6 0.2m.
samples for 13C but not for 15N, depending on
the tissue type.
The analyses of δ13C and δ15N are typically
coupled isotope measurements done on the
2.4.2 Sulfur Isotopes
same organic tissue sample. Samples are flash For migratory organism tissues (hair, mus-
combusted using an EA (Fig. 2.1). Purified cle, feather), sulfur is contained mainly in the
CO2 and N2 are used as the sample analysis form of amino acids (e.g., cysteine) often at
gas and the isotopic reference gases. A stan- low elemental concentrations (B, 45 wt. %).
dard EA and autosampler holding up to 100 Conventionally, a 34S analysis on organic sub-
samples are used to quantitatively combust strates requires a lengthy preparative process
samples to a pulse of CO2 and N2 gas (com- that involves quantitative highly oxidative
bustion H2O is scrubbed out with a trap). The 850 C or flash (Parr) bomb conversion of the
oxidation column consists of a quartz tube par- total S in a tissue sample to an appropriate sul-
tially filled to the hot zone with chromium fate salt. The sulfate salt is converted to puri-
oxide held at 1050 C, followed by reduction fied BaSO4 or to Ag2S. These purified matrices
column filled with copper wires (to reduce are in turn converted to SO2 or SF6 gas for
NOX to N2) held at 600800 C, followed by a analyses on a gas source DI-IRMS (Mayer &
packed GC column held at 3550 C. The GC Krouse, 2004).
often has a thermal conductivity detector to The highest precision δ34S analyses are still
quantify and resolve CO2 from N2. Isotope done using conventional DI-IRMS, but very
peak jumping is used on the IRMS to switch few isotope laboratories nowadays offer this
between nitrogen and carbon isotopes. All capability. However, CF-IRMS methods
δ13C results are reported in units of per mille evolved significantly over the past decades,
(m) relative to the PDB standard. For organic with the advantage of lower cost and high
samples and tissues, calibration is recom- sample throughput. Like C, tissue samples
mended by using three to four certified converted to BaSO4 or Ag2S can be combusted
organic reference materials (L-glutamic acids in an EA to produce SO2 gas. The SO2 is sepa-
or hair keratins). The glutamic acids standards rated from CO2 and N2 using a GC column
(USGS41a, USGS40) have δ13C values of (Mayer & Krouse, 2004). Analysis can be made
136.55m and 226.39m relative to VPDB, and using BaSO4 or Ag2S, which requires initial
δ15N values of 147.55m and 24.52m relative to preparative sample conversion steps. The pre-
AIR, respectively. Two powdered hair C 1 N cision for this method is about 6 0.4m, and
standards (USGS42, USGS43) have δ13C values requires about 15 minutes per sample
of 221.09m and 221.28m VPDB, and δ15N (,100 μg as S). More recent advances forgo the
values of 18.05 and 18.44m relative to AIR, initial sample conversion to an inorganic sulfur

salt, and use direct EA combustion of tissue or development will remain in the realm of a
keratin samples to SO2 without prior conver- few specialized laboratories.
sion to BaSO4 or Ag2S. This approach is prom- In short, the most uncomplicated and cost-
ising, but is complicated by variable or high C: effective method for 34S remains offline com-
N:H:S ratios commonly found in biological bustion of organic tissue samples to pure
samples. The high C:N:H:S ratios and larger BaSO4 salt and subsequent 34S analysis by EA-
sample mass required cogenerate large IRMS using inorganic 34S calibration standards.
amounts of CO2, N2, and H2O and subsequent The additional preparation costs and special-
separation of these undesired gases by GC or ized IRMS assays result in higher costs per
trapping is essential. However, more recent sample ( . $3060).
innovations enabled concurrent C 1 N 1 S
assays by using multiple chromatographic and
chemical gas trapping methods (Fry, 2007;
2.4.3 Isotopes of Trace
Mambelli et al., 2016). The prevailing consen-
sus seems to be direct combustion and analysis
Elements—87Sr/86Sr
is feasible when samples have .0.1 wt. % S Another type of local-spatial stable isotope
(still requiring 25 mg of keratin sample), and analysis used in animal migration studies is
the resulting δ34S precisions will be on the the stable isotope ratios of the trace element
order of 6 0.5m, and this is sufficiently precise strontium (87Sr/86Sr) (Chamberlain et al.,
for many migration studies (marine vs terres- 1997). One of the key advantages of the “heavy
trial assignment). Analytical vigilance to the isotopes” over light stable isotopes is that there
EA is required because the δ18O of the SO2 is little to no isotopic fractionation from geo-
produced changes as the analyzer oxidant logic sources through the food web and into
reagents deplete, requiring IRMS corrections to tissues (Blum, Taliaferro, & Holmes, 2001).
obtain correct δ34S values (Fry, Silva, Kendall, Hence the Sr isotopes among all species in
& Anderson, 2002). local food webs are expected to show fidelity
One further complication with direct to the 87Sr/86Sr ratios of the underlying bed-
tissue combustion is appropriate organic cal- rock or soil. Thus 87Sr/86Sr variations in and
ibration standards having a wide range of among landscapes and continents may be
δ34S values. These are nonexistent, and large distinctive, but these can be highly variable at
discrepancies occur when mixing inorganic small scales in areas of complex geology
standards (Ag2S) and organic samples hav- and are not a priori suitable for continuous
ing high C:N:S ratios by EA-IRMS. Recently, interpolation (Chapter 3: Isoscapes for
a set of hair standards for δ34S was devel- Terrestrial Migration Research). As with δ34S,
oped by the USGS. The USGS42 and USGS43 the 87Sr/86Sr differences between terrestrial
hair standards have δ34S (CDT) values of and marine environments are distinctive
17.84m and 110.46m, respectively, unfortu- (Kennedy, Chamberlain, Blum, Nislow, & Folt,
nately a very small δ-range. Further, given 2005). Hence, similarly to the other local-
the large sample sizes needed (25 mg), the spatial isotopes, Sr isotopes are best used in
USGS hair standards are costly and limited conjunction with δ2H or other light isotopes in
in quantity (0.5 g). The need to include many a multiisotope approach.
references within a CF-IRMS autorun means One important difference between the light
these USGS reference materials will be isotopes and Sr isotopes in fixed tissues like
quickly used up, so it is likely that calibra- feathers is that Sr is a trace element at exceed-
tion and organic S reference material ingly low concentrations (e.g., ,20 μg Sr/g of

2.4 LOCAL-SPATIAL ISOTOPES 47
feather) (Font et al., 2007). The only exception regarding spatial interpretations is critical to
is calcium-bearing tissues that contain much consider. To date, the application of 87Sr iso-
higher Sr concentrations (e.g., bones, teeth, and topes in movement or migration studies are
otoliths) (Blum et al., 2001). For noncalcium- largely limited to proof of concept testing
bearing tissues this means the potential for (Chau et al., 2017; Font et al., 2012), with no
extraneous contamination (e.g., entrained dust, widespread application yet available.
handling, and background) is critically high Finally, 87Sr/86Sr isotope analyses are very
and must be quantified and requires rigorous costly in comparison to the light isotopes
cleaning and QA/QC procedures (Font et al., ( . $200300). These mass spectrometry costs
2007; Vautour, Poirier, & Widory, 2015). There do not include initial prescreening for trace Sr
is no standardized agreement on how this concentrations, nor cost of clean-lab chemical
should be done. digestions. Only few laboratories worldwide
Further, trace Sr concentrations means that have Sr isotope capabilities on offer for com-
up to 25 mg of sample may be required. All mercial assays. This and the large sample size
samples require prescreening to determine the requirements suggest that Sr isotopes will
Sr content by inductively-coupled-plasma remain a specialized assay for projects where
mass spectrometry (ICP-MS) before prepara- the value of the anticipated outcomes exceeds
tive procedures for isolating Sr for isotopic the high analytical cost considerations.
assays are begun. Further, extensive wet chem-
ical or microwave digestions and selective ion
chromatography are required to isolate Sr for
isotopic analysis. Difficulties arise with organic
2.4.4 Biomineral C, O and Sr Isotopes
samples in being able to fully extract all avail- As noted in Chapter 6, Isotopic Tracking of
able Sr. Sr isotopic ratios are determined using Marine Animal Movement, using marine iso-
thermal ionization mass spectrometry and scapes for assessing fish migratory movement
using the NBS-987 standard. No organic or often utilize isotope and trace elements in
keratinous Sr isotope standards exist. accretionary carbonate biominerals (fish oto-
The problem of intrasample isotopic vari- lith) or keratin (whale baleen). Baleen keratin
ability has not been rigorously tested for Sr and marine chitins are easily measured, as
isotopes, although differences in Sr concentra- described earlier. Fish otoliths, however,
tions and isotope ratios between the rachis and require specialized sampling apparatus. To
vane of individual feathers have been reported obtain calcite or aragonite from otolith annual
(Font et al., 2007), although these differences rings, micromilling or laser ablation prepara-
were comparatively smaller than potential tive device are needed, which uses a comput-
geospatial differences. From the preceding sec- erized microscope to target and remove
tion this could be a major issue for slowly accretionary carbonate bands or targets that
growing tissues of birds or animals that are represent a season or a time of growth (akin to
moving among areas having variable 87Sr/86Sr, sampling tree rings). The carbonate is analyzed
especially given the large sample require- for δ13C, δ18O, or 87Sr isotope analyses by
ments. For example, when we consider feath- IRMS or ICP-MS. The stable isotope analyses
ers of small birds weigh 1020 mg, or hair are conducted on tiny ,1030 μg carbonate
strands 1 mg or less, sample tissue pooling samples, and are painstaking and costly
may be required. The example of the eagle (Campana, Fowler, & Jones, 1994; McCulloch,
feather heterogeneity and the implications for Cappo, Aumend, & Müller, 2005; Wurster,
confounding Sr isotope interpretations Patterson, & Cheatham, 1999).

2.5 CONCLUSIONS Carter, J. F., Hill, J. C., Doyle, S., & Lock, C. (2009). Results
of four inter-laboratory comparisons provided by the
Forensic Isotope Ratio Mass Spectrometry (FIRMS) net-
This chapter briefly reviews some practical work. Science & Justice, 49, 127137.
aspects of measuring stable isotope analyses Chamberlain, C. P., Blum, J. D., Holmes, R. T., Feng, X. H.,
for use in animal migration research. The Sherry, T. W., & Graves, G. R. (1997). The use of isotope
researcher is encouraged to be discriminating tracers for identifying populations of migratory birds.
Oecologia, 109, 132141.
and critical in the application of stable isotopes.
Chau, T. H., Tipple, B. J., Hu, L., Fernandez, D. P., Cerling,
Carefully consider the isotopes to be used, and T. E., Ehleringer, J. R., & Chesson, L. A. (2017).
all aspects of sampling and sample preparation, Reconstruction of travel history using coupled δ18O and
87
as well as isotopic measurements. Only when Sr/86Sr measurements of hair. Rapid Communications
utmost confidence in the stable isotope analyses in Mass Spectrometry, 31, 583589.
Cherel, Y., & Hobson, K. A. (2007). Geographical variation
are assured the researcher can proceed with the
in carbon stable isotope signatures of marine predators:
task of making spatial interpretations outlined A tool to investigate their foraging areas in the Southern
in the subsequent chapters. Ocean. Marine Ecology Progress Series, 329, 281287.
Clark, I. D. (2015). Groundwater geochemistry and isotopes.
Boca Raton, FL: CRC Press.
Clark, R. G., Hobson, K. A., & Wassenaar, L. I. (2006).
Acknowledgments Geographic variation in the isotopic (δD, δ13C, δ15N,
Tim Jardine and Richard Doucett contributed data to δ34S) composition of feathers and claws from lesser sca-
Figs. 2.4 and 2.5 and Table 2.3. up and northern pintail: Implications for studies of
migratory connectivity. Canadian Journal of Zoology-
Revue Canadienne De Zoologie, 84, 13951401.
Coplen, T. B. (2011). Guidelines and recommended terms
References for expression of stable-isotope-ratio and gas-ratio mea-
Barrow, L. M., Bjorndal, K. A., & Reich, K. J. (2008). Effects surement results. Rapid Communications in Mass
of preservation method on stable carbon and nitrogen Spectrometry: RCM, 25, 25382560.
isotope values. Physiological and Biochemical Zoology, 81, Coplen, T. B., & Qi, H. (2013). Recognizing the potential pit-
688693. falls of hydrogen isotopic analysis of keratins with steam
Blum, J. D., Taliaferro, E. H., & Holmes, R. T. (2001). equilibration to infer origins of wildlife, food, and peo-
Determining the sources of calcium for migratory song- ple. Rapid Communications in Mass Spectrometry, 27, 2569.
birds using stable strontium isotopes. Oecologia, 126, Coplen, T. B., & Qi, H. (2016). A revision in hydrogen iso-
569574. topic composition of USGS42 and USGS43 human-hair
Bortolotti, G. R. (2010). Flaws and pitfalls in the chemical stable isotopic reference materials for forensic science.
analysis of feathers: Bad newsgood news for avian Forensic Science International, 266, 222225.
chemoecology and toxicology. Ecological Applications, Cormie, A. B., Schwarcz, H. P., & Gray, J. (1994).
20, 17661774. Determination of the hydrogen isotopic composition of
Bowen, G. J., Chesson, L., Nielson, K., Cerling, T. E., & bone collagen and correction for hydrogen exchange.
Ehleringer, J. R. (2005). Treatment methods for the Geochimica et Cosmochimica Acta, 58, 365375.
determination of δ2H and δ18O of hair keratin by Criss, R. E. (1999). Principles of stable isotope distribution.
continuous-flow isotope-ratio mass spectrometry. Rapid New York: Oxford University Press.
Communications in Mass Spectrometry, 19, 23712378. De Ruyck, C., Hobson, K. A., Koper, N., Larson, K. W., &
Bowen, G. J., Wassenaar, L. I., & Hobson, K. A. (2005). Wassenaar, L. I. (2013). An appraisal of the use of
Global application of stable hydrogen and oxygen iso- hydrogen-isotope methods to delineate origins of
topes to wildlife forensics. Oecologia, 143, 337348. migratory saw-whet owls in North America. The
Brand, W. A., & Coplen, T. B. (2012). Stable isotope deltas: Condor, 115, 366374.
Tiny, yet robust signatures in nature. Isotopes in Edwards, M. S., Turner, T. F., & Sharp, Z. D. (2002). Short-
Environmental and Health Studies, 48, 393409. and long-term effects of fixation and preservation on
Campana, S. E., Fowler, A. J., & Jones, C. M. (1994). Otolith stable isotope values (δ13C, δ15N, δ34S) of fluid-
elemental fingerprinting for stock identification of preserved museum specimens. Copeia, 11061112.
Atlantic cod (Gadus morhua) using laser ablation Elliott, K. H., Davis, M., & Elliott, J. E. (2014). Equations
ICPMS. Canadian Journal of Fisheries and Aquatic Sciences, for lipid normalization of carbon stable isotope ratios in
51, 19421950. aquatic bird eggs. PLoS One, 9, e83597.

REFERENCES 49
Fogel, M. L., Griffin, P. L., & Newsome, S. D. (2016). Hobson, K. A. (1999). Tracing origins and migration of
Hydrogen isotopes in individual amino acids reflect wildlife using stable isotopes: A review. Oecologia, 120,
differentiated pools of hydrogen from food and water 314326.
in Escherichia coli. Proceedings of the National Academy of Hobson, K. A., & Koehler, G. (2015). On the use of
Sciences, 113, E4648E4653. stable oxygen isotope (δ18O) measurements for tracking
Font, L., Nowell, G. M., Pearson, D. G., Ottley, C. J., & avian movements in North America. Ecology and
Willis, S. G. (2007). Sr isotope analysis of bird feathers Evolution, 5, 799806.
by TIMS: A tool to trace bird migration paths and Hobson, K. A., & Wassenaar, L. I. (1997). Linking breeding
breeding sites. Journal of Analytical Atomic Spectrometry, and wintering grounds of neotropical migrant song-
22, 513522. birds using stable hydrogen isotopic analysis of feath-
Font, L., van der Peijl, G., van Wetten, I., Vroon, P., van ers. Oecologia, 109, 142148.
der Wagt, B., & Davies, G. (2012). Strontium and lead Hobson, K. A., Gibbs, H. L., & Gloutney, M. L. (1997).
isotope ratios in human hair: Investigating a potential Preservation of blood and tissue samples for stable-
tool for determining recent human geographical move- carbon and stable-nitrogen isotope analysis. Canadian
ments. Journal of Analytical Atomic Spectrometry, 27, Journal of Zoology-Revue Canadienne De Zoologie, 75,
719732. 17201723.
Fox, A. D., Hobson, K. A., & Kahlert, J. (2009). Isotopic evi- Hobson, K. A., Wassenaar, L. I., & Taylor, O. R. (1999).
dence for differential endogenous protein contributions Stable isotopes (δD and δ13C) are geographic indicators
to Greylag Goose Anser anser flight feathers. Journal of of natal origins of monarch butterflies in eastern North
Avian Biology, 40, 108112. America. Oecologia, 120, 397404.
Fry, B. (2006). Stable isotope ecology. New York: Springer. Hobson, K. A., deMent, S. H., Van Wilgenburg, S. L., &
Fry, B. (2007). Coupled N, C and S stable isotope measure- Wassenaar, L. I. (2009). Origins of American Kestrels
ments using a dual-column gas chromatography sys- wintering at two southern US sites: An investigation
tem. Rapid Communications in Mass Spectrometry, 21, using stable isotope (δD, δ18O) methods. Journal of
750756. Raptor Research, 43, 325337.
Fry, B., Brand, W., Mersch, F. J., Tholke, K., & Garritt, R. Hobson, K. A., Soto, D. X., Paulson, D. R., Wassenaar, L. I.,
(1992). Automated analysis system for coupled δ13C and & Matthews, J. H. (2012). A dragonfly (δ2H) isoscape
δ15N measurements. Analytical Chemistry, 64, 288291. for North America: A new tool for determining natal
Fry, B., Silva, S. R., Kendall, C., & Anderson, R. K. (2002). origins of migratory aquatic emergent insects. Methods
Oxygen isotope corrections for online delta S-34 analy- in Ecology and Evolution, 3, 766772.
sis. Rapid Communications in Mass Spectrometry, 16, Hobson, K. A., Plint, T., Serrano, E. G., Alvarez, X. M.,
854858. Ramirez, I., & Longstaffe, F. J. (2017). Within-wing iso-
Garcı́a-Pérez, B., & Hobson, K. A. (2014). A multi-isotope topic (δ2H, δ13C, δ15N) variation of monarch butter-
(δ2H, δ13C, δ15N) approach to establishing migratory flies: Implications for studies of migratory origins and
connectivity of Barn Swallow (Hirundo rustica). diet. Animal Migration, 4, 814.
Ecosphere, 5, art21. Hoefs, J. (2015). Stable isotope geochemistry (7th ed). Berlin;
Gehre, M., Renpenning, J., Gilevska, T., Qi, H., Coplen, New York: Springer.
T. B., Meijer, H. A., . . . Schimmelmann, A. (2015). On- Hutchinson, J., & Trueman, C. (2006). Stable isotope analy-
line hydrogen-isotope measurements of organic sam- ses of collagen in fish scales: Limitations set by scale
ples using elemental chromium-an extension for high architecture. Journal of Fish Biology, 69, 8741880.
temperature elemental-analyzer techniques. Analytical Kelly, J. F., Bridge, E. S., Fudickar, A. M., & Wassenaar,
Chemistry, 87, 51985205. L. I. (2009). A test of comparative equilibration for
Groning, M. (2004). International stable isotope reference determining non-exchangeable stable hydrogen
materials. In P. de Groot (Ed.), Handbook of stable isotope isotope values in complex organic materials. Rapid
analytical techniques (Vol. 1, pp. 874906). Amsterdam: Communications in Mass Spectrometry, 23, 23162320.
Elsevier. Kennedy, B. P., Chamberlain, C. P., Blum, J. D., Nislow,
de Groot, P. A. (2004). Handbook of stable isotope analytical K. H., & Folt, C. L. (2005). Comparing naturally occur-
techniques (1st ed.). Amsterdam: Elsevier. ring stable isotopes of nitrogen, carbon, and strontium
Hayden, B., Soto, D. X., Jardine, T. D., Graham, B. S., as markers for the rearing locations of Atlantic salmon
Cunjak, R. A., Romakkaniemi, A., & Linnansaari, T. (Salmo salar). Canadian Journal of Fisheries and Aquatic
(2015). Small tails tell tall talesintra-individual varia- Sciences, 62, 4857.
tion in the stable isotope values of fish fin. PLoS One, Koehler, G., & Wassenaar, L. I. (2012). Determination of the
10, e0145154. hydrogen isotopic compositions of organic materials

and hydrous minerals using thermal combustion laser Pietsch, S. J., Hobson, K. A., Wassenaar, L. I., & Tutken, T.
spectroscopy. Analytical Chemistry, 84, 36403645. (2011). Tracking cats: Problems with placing feline car-
Lajtha, K., & Michener, R. (2007). Stable isotopes in ecology and nivores on δ18O and δ2H isoscapes. PLoS One, 6, e24601.
environmental science. New York: Blackwell Scientific. Pinnegar, J. K., & Polunin, N. V. C. (1999). Differential frac-
Lis, G., Wassenaar, L. I., & Hendry, M. J. (2008). High- tionation of δ13C and δ15N among fish tissues:
precision laser spectroscopy D/H and 18O/16O mea- Implications for the study of trophic interactions.
surements of microliter natural water samples. Functional Ecology, 13, 225231.
Analytical Chemistry, 80, 287293. Post, D. M., Layman, C. A., Arrington, D. A., Takimoto, G.,
Mambelli, S., Brooks, P. D., Sutka, R., Hughes, S., Finstad, Quattrochi, J., & Montana, C. G. (2007). Getting to the
K. M., Nelson, J. P., & Dawson, T. E. (2016). High- fat of the matter: Models, methods and assumptions for
throughput method for simultaneous quantification of dealing with lipids in stable isotope analyses. Oecologia,
N, C and S stable isotopes and contents in organics and 152, 179189.
soils. Rapid Communications in Mass Spectrometry, 30, Qi, H., & Coplen, T. B. (2011). Investigation of preparation
17431753. techniques for δ2H analysis of keratin materials and a
Marra, P. P., Hobson, K. A., & Holmes, R. T. (1998). proposed analytical protocol. Rapid Communications in
Linking winter and summer events in a migratory bird Mass Spectrometry: RCM, 25, 22092222.
using stable carbon isotopes. Science, 292, 18841886. Qi, H., Coplen, T. B., & Wassenaar, L. I. (2011). Improved
Matthews, D. E., & Hayes, J. M. (1978). Isotope-ratio- online δ18O measurements of nitrogen- and sulfur-
monitoring gas chromatography-mass spectrometry. bearing organic materials and a proposed analytical
Analytical Chemistry, 50, 14651473. protocol. Rapid Communications in Mass Spectrometry:
Mayer, B., & Krouse, H. R. (2004). Procedures for sulfur RCM, 25, 20492058.
isotope abundance studies. In P. A. De Groot (Ed.), Qi, H., Coplen, T. B., & Jordan, J. A. (2016). Three whole-
Handbook of stable isotope analytical techniques (Vol. 1, wood isotopic reference materials, USGS54, USGS55,
pp. 538596). Amsterdam: Elsevier. and USGS56, for δ2H, δ18O, δ13C, and δ15N measure-
Mazerolle, D. F., Hobson, K. A., & Wassenaar, L. I. (2005). ments. Chemical Geology, 442, 4753.
Stable isotope and band-encounter analyses delineate Schimmelmann, A. (1991). Determination of the concentra-
migratory patterns and catchment areas of white- tion and stable isotopic composition of nonexchange-
throated sparrows at a migration monitoring station. able hydrogen in organic matter. Analytical Chemistry,
Oecologia, 144, 541549. 63, 24562459.
McCulloch, M., Cappo, M., Aumend, J., & Müller, W. Schimmelmann, A., Miller, R. F., & Leavitt, S. W. (1993).
(2005). Tracing the life history of individual barra- Hydrogen isotopic exchange and stable isotope ratios
mundi using laser ablation MC-ICP-MS Sr-isotopic and in cellulose, wood, chitin, and amino compounds.
Sr/Ba ratios in otoliths. Marine and Freshwater Research, In P. K. Swart, K. C. Lohmann, J. McKenzie, & S. Savin
56, 637644. (Eds.), Climate change in continental isotope records.
McKinney, C. R., McCrea, J. M., Epstein, S., Allen, H. A., & (pp. 367374). Washington, DC: American Geophysical
Urey, H. C. (1950). Improvements in mass spectro- Union.
meters for the measurement of small differences in iso- Schimmelmann, A., Qi, H., Coplen, T. B., Brand, W. A.,
tope abundance ratios. Review of Scientific Instruments, Fong, J., Meier-Augenstein, W., . . . Werner, R. A.
21, 724730. (2016). Organic reference materials for hydrogen, car-
Meier-Augenstein, W. (2011). Stable isotope forensics: An bon, and nitrogen stable isotope-ratio measurements:
introduction to the forensic application of stable isotope Caffeines, n-alkanes, fatty acid methyl esters, glycines,
analysis. West Sussex, UK: John Wiley & Sons. L-valines, polyethylenes, and oils. Analytical Chemistry,
Meier-Augenstein, W., Hobson, K. A., & Wassenaar, L. I. 88, 42944302.
(2013). Critique: Measuring hydrogen stable isotope Sessions, A. L., & Hayes, J. M. (2005). Calculation of hydro-
abundance of proteins to infer origins of wildlife, food gen isotopic fractionations in biogeochemical systems.
and people. Bioanalysis, 5, 751767. Geochimica Et Cosmochimica Acta, 69, 593597.
Nielson, K. E., & Bowen, G. J. (2010). Hydrogen and oxy- Sessions, A. L., Burgoyne, T. W., Schimmelmann, A., &
gen in brine shrimp chitin reflect environmental water Hayes, J. M. (1999). Fractionation of hydrogen isotopes
and dietary isotopic composition. Geochimica et in lipid biosynthesis. Organic Geochemistry, 30,
Cosmochimica Acta, 74, 18121822. 11931200.
Paul, D., Skrzypek, G., & Forizs, I. (2007). Normalization of Sharp, Z. (2017) Principles of stable isotope geochemistry. 2nd
measured stable isotopic compositions to isotope refer- Edition. http://digitalrepository.unm.edu/unm_oer/1/.
ence scalesA review. Rapid Communications in Mass Skinner, M. M., Martin, A. A., & Moore, B. C. (2016). Is
Spectrometry, 21, 30063014. lipid correction necessary in the stable isotope analysis

REFERENCES 51
of fish tissues? Rapid Communications in Mass environmental interest. Environmental Science and
Spectrometry, 30, 881889. Technology, 34, 23542360.
Sotiropoulos, M. A., Tonn, W. M., & Wassenaar, L. I. Wassenaar, L. I., & Hobson, K. A. (2003). Comparative
(2004). Effects of lipid extraction on stable carbon and equilibration and online technique for determination of
nitrogen isotope analyses of fish tissues: Potential con- non-exchangeable hydrogen of keratins for use in ani-
sequences for food web studies. Ecology of Freshwater mal migration studies. Isotopes in Environmental and
Fish, 13, 155160. Health Studies, 39, 211217.
Soto, D. X., Wassenaar, L. I., & Hobson, K. A. (2013). Wassenaar, L. I., Hobson, K. A., & Sisti, L. (2015). An
Stable hydrogen and oxygen isotopes in aquatic food online temperature-controlled vacuum-equilibration
webs are tracers of diet and provenance. Functional preparation system for the measurement of δ2H values
Ecology, 27, 535543. of non-exchangeable-H and of δ18O values in organic
Soto, D. X., Hobson, K. A., & Wassenaar, L. I. (2016). Using materials by isotope-ratio mass spectrometry. Rapid
hydrogen isotopes of freshwater fish tissue as a tracer Communications in Mass Spectrometry: RCM, 29, 397407.
of provenance. Ecology and Evolution, 6, 77767782. Wassenaar, L. I., Terzer-Wassmuth, S., Douence, C.,
Soto, D. X., Koehler, G., Wassenaar, L. I., & Hobson, K. A. Araguas-Araguas, L., Aggarwal, P. K., & Coplen, T. B.
(2017). Re-evaluation of the hydrogen stable isotopic (2018). Seeking excellence: An evaluation of 235
composition of keratin calibration standards for wildlife international laboratories conducting water isotope
and forensic science applications. Rapid Communications analyses by isotope-ratio and laser-absorption spec-
in Mass Spectrometry: RCM, 31, 11931203. trometer. Rapid Communications in Mass Spectrometry,
Speakman, J. (1997). Doubly labelled water: Theory and prac- 32, 393406.
tice. Springer Science & Business Media. West, J. B., Bowen, G. J., Dawson, T. E., & Tu, K. P. (2009).
Terzer, S., Wassenaar, L. I., Araguás-Araguás, L. J., & Isoscapes: Understanding movement, pattern, and process on
Aggarwal, P. K. (2013). Global isoscapes for δ18O and Earth through isotope mapping. New York: Springer.
δ2H in precipitation: Improved prediction using region- Wolf, N., Bowen, G. J., & del Rio, C. M. (2011). The influ-
alized climatic regression models. Hydrology and Earth ence of drinking water on the δD and δ18O values of
System Sciences, 17, 47134728. house sparrow plasma, blood and feathers. Journal of
Vander Zanden, H. B., Soto, D. X., Bowen, G. J., & Hobson, Experimental Biology, 214, 98103.
K. A. (2016). Expanding the isotopic toolbox: Wunder, M. B., Hobson, K. A., Kelly, J., Marra, P. P.,
Applications of hydrogen and oxygen stable isotope Wassenaar, L. I., Stricker, C. A., & Doucett, R. R. (2009).
ratios to food web studies. Frontiers in Ecology and Does a lack of design and repeatability compromise sci-
Evolution, 4, 20. entific criticism? A response to Smith et al. (2009). Auk,
Vautour, G., Poirier, A., & Widory, D. (2015). Tracking 126, 922926.
mobility using human hair: What can we learn from Wurster, C. M., Patterson, W. P., & Cheatham, M. M.
lead and strontium isotopes? Science & Justice, 55, (1999). Advances in micromilling techniques: A new
6371. apparatus for acquiring high-resolution oxygen and
Wassenaar, L. I., & Hobson, K. A. (1998). Natal origins of carbon stable isotope values and major/minor elemen-
migratory monarch butterflies at wintering colonies in tal ratios from accretionary carbonate. Computers &
Mexico: New isotopic evidence. Proceedings of the Geosciences, 25, 11591166.
National Academy of Sciences of the United States of Yurkowski, D. J., Hussey, N. E., Semeniuk, C., Ferguson,
America, 95, 1543615439. S. H., & Fisk, A. T. (2015). Effects of lipid extraction
Wassenaar, L. I., & Hobson, K. A. (2000). Improved and the utility of lipid normalization models on δ13C
method for determining the stable-hydrogen isotopic and δ15N values in Arctic marine mammal tissues. Polar
composition (δD) of complex organic materials of Biology, 38, 131143.

C H A P T E R
3
Isoscapes for Terrestrial Migration
Research
Gabriel J. Bowen1 and Jason B. West2
1
University of Utah, Salt Lake City, UT, United States, 2Texas A&M University, College Station,
TX, United States
3.1 INTRODUCTION fundamental isotopic theory; some studies have

developed isoscapes using statistical analysis of
Isotope tracking of migratory terrestrial ani- known-origin animal tissue directly (e.g.,
mals (e.g., birds, bats, and insects) relies on the Hobson, Wassenaar, & Taylor, 1999), and
assimilation and fixation of intrinsic although this approach offers some advantages
stable isotopic markers (i.e., CHNOS) from the for applied work, it reflects a different and less
environment into animal body tissue. The generalizable paradigm that is not discussed
power of the isotopic markers relates to the here. A discussion of opportunities and future
extent and pattern of spatial isotope ratio varia- directions in isoscape modeling is offered. The
tions in the environmental substrates from primary goal of this chapter is to familiarize the
which they are assimilated (primarily food, researcher with isoscape data products for
water, and air). This chapter introduces these potential use in migration applications. Also
patterns of variation for most of the commonly described are the principles and methodology
applied or applicable stable isotope systems, underlying the development of these products
and describes methods by which the spatial and relevant to their informed use.
landscapes of environmental isotopic variation Maps of some isotopic landscapes have been
(isoscapes) are modeled and predicted at scales available at least since the early 1980s when
that are relevant to the study of migratory workers affiliated with the International
behavior and ecology. Examples of isoscapes for Atomic Energy Agency (IAEA) and World
some isotopic systems are presented along with Meteorological Organization’s Global Network
discussion of challenges and cautionary notes for Isotopes in Precipitation (GNIP) compiled
related to the creation and interpretation of iso- and contoured mean annual precipitation H
scapes. We focus here on environmental systems and O isotope ratio data to produce a map with
and modeling approaches that are rooted in near-global coverage (Yurtsever & Gat, 1981).

54 3. ISOSCAPES FOR TERRESTRIAL MIGRATION RESEARCH
The GNIP dataset is a prime example of a spa- additional researchers to participate in contin-
tial isotope monitoring network, providing a ued development of isoscapes relevant to
dataset that has motivated the development of migration research.
improved isoscapes for H and O in water
(Birks, Gibson, Gourcy, Aggarwal, & Edwards,
2002; Bowen, Ehleringer, Chesson, Stange,
3.2 PROCESS
& Cerling, 2007; Bowen & Revenaugh, 2003;
Bowen, Wassenaar, & Hobson, 2005; Bowen &
Mapping isotopic variation across space is
Wilkinson, 2002; Meehan, Giermakowski, &
accomplished through the identification, sim-
Cryan, 2004; Terzer, Wassenaar, Araguás-
plification, and modeling of the processes that
Araguás, & Aggarwal, 2013) and, more
lead to isotopic variation at the landscape level.
recently, for plants (West, Sobek, & Ehleringer,
Although a wide range of physical and chemi-
2008).
cal processes can produce isotopic discrimina-
Isoscapes of ecosystem-scale carbon isotope
tion and contribute to observed isotope
ratios were found to be relevant to the study
distributions, a relatively small subset of pro-
of the global carbon cycle and have been in
cesses often dominates landscape-level variabil-
development since the early 1990s (Lloyd &
ity. We begin by highlighting and reviewing
Farquhar, 1994; Still, Berry, Collatz, & DeFries,
these processes as they relate to isoscape
2003; Suits et al., 2005). The development of
modeling for terrestrial migration studies.
isoscapes for other isotopic systems is an
active but somewhat less mature field, with
products representing plant and soil nitrogen
3.2.1 Rayleigh Distillation
isoscapes at different scales (Amundson et al.,
2003; Fox-Dobbs, Doak, Brody, & Palmer, 2010;
and Precipitation δ 2H and δ 18O
Rascher, Hellmann, Máguas, & Werner, 2012) Global spatial variability in the H and O iso-
and Sr isoscapes (Bataille & Bowen, 2012; topic composition of meteoric water (precipita-
Bataille et al., 2014; Bataille, Laffoon, & Bowen, tion) was first reported by Dansgaard (see also
2012; Beard & Johnson, 2000) published in the Craig, 1961; Dansgaard, 1954, 1964), making it
literature. one of the longest studied examples of
Research on the spatial patterning of vari- landscape-level isotopic variation. The pro-
ous isotopes in the environment and their found variation in δ2H and δ18O values mea-
modeling and depiction is an active field, and sured for precipitation samples collected
although some data products are well docu- worldwide (Rozanski, Araguas-Araguas, &
mented and publicly available through Gonfiantini, 1993) can largely be attributed to
websites such as http://www.waterisotopes. a single phenomenon: the progressive drying
org, the scope of these products remains lim- of air masses as they lose moisture in the form
ited. At the same time, an increasing number of precipitation. The phase change reaction
of software tools are becoming available that that leads to the formation of water droplets
help researchers explore and model spatial iso- (or ice crystals) in clouds proceeds with a
tope data. Given that much of the data, theory, temperature-dependent equilibrium isotope
and software enabling isoscapes modeling effect, αe, through which water molecules con-
are freely available, a secondary goal of this taining heavy isotopic species are preferen-
chapter is to introduce fundamental considera- tially incorporated in the liquid or solid phase
tions and approaches underlying isoscapes during condensation (note that αe differs for
modeling. Our hope is that this encourages condensation to liquid vs solid phase and for

3.2 PROCESS 55
2
H- and 18O-bearing water molecules). Having snowmelt and summertime basinal rain in
grown to a size at which they fall from the mountainous regions) or have different post-
convecting cloud mass, rain droplets or snow precipitation histories (e.g., thermal spring
crystals are effectively removed from the cloud water and surface water).
system, taking with them a disproportionate Reevaporation of water can occur at many
concentration of the “heavy” isotopic species points in the postprecipitation history of all
and leaving the cloud vapor incrementally continental waters, including from leaf sur-
depleted in 2H and 18O. As this process pro- faces that intercept falling rain, soils, rivers,
ceeds, the δ2H and δ18O values of the cloud lakes, or reservoirs. Isotope fractionation dur-
vapor become progressively lower according ing evaporation does not follow a simple equi-
to the Rayleigh equation (given in terms of librium model, but involves a dynamic balance
ratios): of the phase change reaction and bidirectional
diffusive transport between a boundary layer
R 5 R0 f ðα21Þ (3.1) adjacent to the liquid surface and the free
where R is the isotope ratio of cloud vapor at atmosphere. These processes were synthesized
any point in time, R0 is the initial isotope ratio as the “CraigGordon model” (Craig &
of the air mass, and f is the fraction of vapor Gordon, 1965):
remaining. For the residual vapor produced dδ hðδ 2 δa Þ 2 ðδ 1 1ÞðΔε 1 ε=αÞ
through the condensation process, values of α 5 εt
d In f 1 2 h 1 Δε (3.2)
are less than 1, giving a progressive decrease
in vapor isotope ratios (and those of newly 5 εe 1 Δεk
formed precipitation) as an air mass dries. where δ and δa are the isotopic compositions
of the evaporating water body and atmo-
sphere, f is the remaining fraction of liquid, h
3.2.2 Hydrological Mixing, Evaporation, is atmospheric humidity, and ε, α, and Δε are
and Surface Water δ 2H and δ 18O the isotope effect and fractionation factor for
equilibrium evaporation, and the kinetic iso-
The primary postprecipitation processes
tope effect of evaporation, respectively. Under
affecting the stable isotopic composition of
low humidity conditions, evaporation can be
continental waters are hydrological mixing of
approximated as distillation process following
waters having different isotopic compositions
Eq. (3.1). The net isotope effect of evaporation
and reevaporation from the land surface.
is to increase the δ2H and δ18O values of resid-
Large datasets of H and O isotope ratios of
ual surface water, with the magnitude of
river and tap water clearly show the impact of
change scaling with the extent of evaporation
both processes on spatial variation in environ-
and affected by h and other parameters that
mental water isotopic compositions (Bowen
determine the evaporative fractionation.
et al., 2007; Bowen, Kennedy, Liu, & Stalker,
2011; Kendall & Coplen, 2001; Landwehr,
Coplen, & Stewart, 2014). Mixing represents a
3.2.3 Plant Water and Organic
linear process, weighted according to the volu-
metric concentration of the different sources.
H and O Isotopes
Mixing is a particularly important process Spatial patterns in meteoric water isotope
where waters are present that have very differ- ratios are intimately connected to the base of
ent isotopic compositions because they fell at all terrestrial food webs through the linkage of
different times or places (e.g., mountain water and primary productivity (plants). To a

first approximation, the isotopic composition Leaf water isotopic composition is impor-
of the water in plants may be simplified to two tant for two reasons: (1) leaf water is a poten-
pools: unfractionated water that matches the tial source of animal body water (Murphy,
source water isotopic composition (e.g., soil Bowman, & Gagan, 2007) and (2) the isotopic
water) and 2H- and 18O-enriched leaf water. composition of plant organic molecules is
Leaf water isotope ratios increase in response dependent on the isotopic composition of the
to evaporation in a manner analogous to open water at sites of photosynthesis and subse-
water bodies, with liquid to gas phase changes quent isotope effects of plant metabolism
occurring inside the leaf, diffusion of vapor (Roden, Lin, & Ehleringer, 2000). These isotope
through stomatal openings and the leaf bound- fractionations can be relatively large. For
ary layer, and isotopic exchange of leaf water example, observed fractionation factors for the
with atmospheric vapor. These isotope effects formation of cellulose have been estimated as
have been described in several models of leaf ε 5 27m for oxygen isotopes and ε 5 158m for
water enrichment, derived principally from hydrogen isotopes (Luo & Sternberg, 1992;
the CraigGordon model already described. Sternberg & Deniro, 1983; Yakir, DeNiro, &
The general equation for steady state leaf Gat, 1990). To date only a small fraction of the
water isotope ratios has been written as: vast array of plant compounds that animals
consume has been characterized in detail, since
ei 2 es es 2 ea
Re 5 α αk Rs 1 αkb Rs the focus of research to date has largely been
ei ei on cellulose and leaf waxes and their use in

ea paleo reconstruction (Gamarra, Sachse, &
1 RA
ei Kahmen, 2016; Kahmen et al., 2013; Sachse
et al., 2012). In addition, it remains unclear to
(3.3) what extent leaf water enrichment and its
where Re is the isotope ratio of evaporated within-leaf heterogeneity is in fact reflected in
leaf water, Rs is the isotope ratio of the consumer tissue isotope ratio variation,
source water, RA is the isotope ratio of the although this question is the subject of ongoing
atmospheric water vapor, ei is internal leaf research (Lehmann, Gamarra, Kahmen,
vapor pressure, es is the leaf surface vapor pres- Siegwolf, & Saurer, 2017).
sure, and ea is atmospheric vapor pressure
(Flanagan, Comstock, & Ehleringer, 1991).
3.2.4 Gas Exchange, Photosynthetic
Comparisons between CraigGordon predic-
tions and measurements have led to the recog-
Pathway, and Plant C Isotopes
nition of additional sources of variation, The carbon isotope ratios of plant tissues
including spatial heterogeneity in leaves, non- are dependent on the δ13C of atmospheric CO2
steady state effects, and the effects of diffusion and the isotope fractionation events that occur
of heavy isotopologues in leaves. Model modifi- during carbon fixation, including diffusion,
cations incorporate these effects (Barbour, dissolution, and hydration of CO2, and the fix-
Roden, Farquhar, & Ehleringer, 2004; ation reactions themselves (Cernusak et al.,
Dongmann, Nurnberg, Forstel, & Wagener, 2009; Diefendorf, Mueller, Wing, Koch, &
1974; Farquhar & Cernusak, 2005; Helliker & Freeman, 2010; Farquhar, Ehleringer, &
Ehleringer, 2000; Yakir, Berry, Giles, & Osmond, Hubick, 1989; O’Leary, Madhavan, & Paneth,
1994). These models and their potential applica- 1992; Osmond et al., 1973; Seibt, Rajabi,
tions are reviewed in Cernusak et al. (2016). Griffiths, & Berry, 2008). There are three

3.2 PROCESS 57
primary pathways of carbon fixation in plants: Stomatal resistance has an important effect
C3, C4, and CAM, names that refer to the num- on the carbon isotope ratios of C3 plants. The
ber of carbon atoms in the first stable product number and aperture of stomata control the
of photosynthesis (C3 and C4), or the nocturnal CO2 diffusion resistance between the atmo-
buildup of malic acid (CAM). The C3 pathway sphere and the substomatal cavities, and thus
utilizes ribulose-1,5-bisphosphate carboxylase/ determine the relative openness of the isotopic
oxygenase (Rubisco) as a catalyst to form a system with respect to the source CO2. As sto-
three-carbon molecule from atmospheric CO2 mata close, the supply of CO2 to the substoma-
during the day. The C4 pathway is primarily tal cavity, as well as its diffusion back out of
found in grasses and utilizes phosphoenolpyr- the leaf, slows. This slowing of CO2 exchange
uvate carboxylase (PEP carboxylase) as a cata- with the atmosphere results in a relative
lyst to fix HCO32 (atmospheric CO2 hydrated 13
C-enrichment of the products of photosyn-
in a reaction catalyzed by carbonic anhydrase) thesis because of a progressive 13C-enrichment
initially to a four-carbon molecule that is then of the internal leaf CO2 pool, and a propor-
(via decarboxylation) the source of CO2 for C3 tional increase in the flux of 13CO2 into the
photosynthesis. In C4 plants, the C3 photosyn- reactions of photosynthesis. Because of this
thesis phase is spatially isolated from the effect of stomatal aperture, climate and indi-
atmosphere and receives high concentrations vidual plant water use efficiencies exert a
of CO2 from the PEP carboxylase system. dominant influence on the carbon isotopic sig-
CAM plants also utilize PEP carboxylase to nature of C3 plant tissues. With greater water
catalyze the first step, but rather than isolate limitation, stomata tend to close, increasing the
Rubisco spatially as occurs in C4 plants, these δ13C of the resulting sugars, and ultimately
plants do the initial step of fixing CO2 at night other organic materials in plants (Hemming
when the vapor pressure deficit is lower. et al., 2005).
Rubisco strongly discriminates against 13CO2
(approximately 30m; O’Leary, 1981) giving
C3 plants strongly 13C-depleted carbon isotope
3.2.5 Nitrogen Isotopes in Soils
ratios. PEP carboxylase, on the other hand,
shows much lower discrimination (approxi-
and Plants
mately 2m), which results in C4 plants having Plant δ15N values are determined by the iso-
higher δ13C values (CAM plants show a topic composition of the plant N source and
wide range of values; Ehleringer, Rundel, & the isotope fractionations associated with
Nagy, 1986). nutrient N uptake and metabolism. Because
The differences in isotopic discrimination there are multiple, competing reactions in
between these two enzymes are large enough soils, many of which have strong N isotopic
that C3 and C4 plants have nearly completely effects, a general theory describing soil δ15N
nonoverlapping carbon isotope distributions has not yet emerged (Evans, 2001; Robinson,
(Ehleringer et al., 1986). This was observed 2001). Primarily because of this, plant nitrogen
very early in the study of stable isotope isotopic composition has been interpreted as
ratios in plants (Craig, 1953; Wickman, 1952), an integrator of the effects of important pro-
and has been utilized extensively to under- cesses that can inform questions related to
stand plantanimal interactions, both in changes in nitrogen cycling and nitrogen
modern and paleoenvironments (e.g., Cerling, sources, but that needs to be interpreted cau-
Harris, & Passey, 2003; Van Der Merwe, tiously and in primarily site-specific ways
Throp, & Bell, 1988). (Evans, 2001; Robinson, 2001). Large scale

comparisons have shown correlations between igneous rocks and carbonate rocks, variation in
both the δ15N of plants and soils and climate the 87Rb content of different rocks, and varia-
variables such as mean annual temperature tion in the age of the rocks (Faure & Powell,
and precipitation (Amundson et al., 2003; 1972).
Austin & Sala, 1999; Handley et al., 1999; When considering variation in ecosystem Sr
Ometto et al., 2006). Although much of the isotope ratios that might be incorporated into a
global observed variability remains unex- migrating animal through diet or water, the iso-
plained by these simple models, they do sug- topic variability of rock Sr is often “averaged”
gest dominant controls on soil and plant δ15N by hydrological and biogeochemical processes
based on preferential losses of 15N from sys- (e.g., see review by Capo, Stewart, &
tems (cf. Houlton, Sigman, & Hedin, 2006). Chadwick, 1998). Rock Sr is released to ecosys-
Clearly N inputs are an important control on tems by chemical weathering, and in areas
plant and soil δ15N also, since N fixation where multiple rock sources are available, Sr
inputs are near 0m, but can vary from 23m to isotope ratios within ecosystems can be heavily
1m (West, HilleRisLambers, Lee, Hobbie, & biased toward those of Sr sourced from individ-
Reich, 2005) and nitrogen in wet and dry ual rock types. In many cases, ecosystem
87
deposition can vary considerably, depending Sr/86Sr ratios are largely controlled by car-
on the nitrogen source (Pardo et al., 2006; bonate minerals, which weather rapidly and
Widory, 2007). More recent work suggests the have total Sr concentrations B26 times higher
possibility that remotely sensed reflectance than most other rocks (Faure & Powell, 1972;
data might yield accurate information on vege- Jacobson & Blum, 2000). Sources of Sr entering
tation δ15N (Wang, Okin, Wang, Epstein, & groundwater, surface water, and soils can often
Macko, 2007). differ, particularly in areas where windblown
dust contributes a significant fraction of Sr to
soils (e.g., Kennedy, Chadwick, Vitousek,
Derry, & Hendricks, 1998; Quade, Chivas, &
3.2.6 Ecosystem Sr Isotopes McCulloch, 1995) or where bedrock geology is
Isotopic variation in the trace element stron- particularly heterogeneous, meaning that mul-
tium (Sr), unlike that in the other light isotope tiple, isotopic distinct pools of Sr may be avail-
systems discussed here, is driven by the pro- able to plants and animals living in close
duction of one of the common isotopes, 87Sr, proximity. Nonetheless, spatial variability in
which forms during the β2 decay of geogenic 87
Sr/86Sr, primarily due to major variation in
87
Rb. Both 87Sr and the more common nonra- rock type distributions over large spatial scales,
diogenic 86Sr are stable isotopes. Differences in has been shown to propagate into ecosystems
the relative abundance of these Sr isotopes are and thereby represent a useful tracer for migra-
due to the level of production of the radiogenic tory ecology and paleoecology (e.g., Blum,
isotope and are thus tracers of Sr source. Taliaferro, & Holmes, 2001; Chamberlain et al.,
Because the half-life of 87Rb is long (B48.8 bil- 1997; Hoppe, Koch, Carlson, & Webb, 1999).
lion years), production of 87Sr occurs over geo-
logical timescales, meaning that most 87Sr/86Sr
variation in ecosystems can be linked to Sr iso- 3.3 PATTERN
tope variation in surficial geological sources.
Variation in the 87Sr/86Sr ratio of rocks occurs The processes discussed in the previous sec-
for a number of different reasons, including tions produce isotope variation in the environ-
variation in the initial Sr isotope ratio of ment, but in order to understand the existence

3.3 PATTERN 59
of systematic, predictable, isotopic variability parameters can be illustrated through exam-
useful for terrestrial migration research, we ples from the systems introduced earlier.
must consider the organization of these pro- Spatial variation in the isotopic composition
cesses with respect to space. Indeed, the fact of elemental sources is important in all systems
that many isotopically discriminating pro- where these sources exhibit landscape-level iso-
cesses are spatially patterned underlies the isotopic heterogeneity, and in many cases may be
scapes concept and the correlative application the primary determinant of the isotopic compo-
of stable isotope tracking in migration sition of biological systems. Strontium isotopes,
research. The creation of isoscapes for migra- which are not measurably fractionated by bio-
tion research requires that these patterns be logical systems, provide an example system in
mapped in space (and time), usually through which spatial variation in elemental sources
the mathematical transformation of maps controls the isotopic composition of plants and
depicting correlates or environmental drivers animals. Work by Kennedy and colleagues
of isotopic variation. We will first review the (1998) on the well-characterized soil chronose-
origin of these patterns, considering how they quences of the Hawaiian islands nicely illus-
relate to geological, physiographic, climatologi- trates such a relationship. In this case, soils that
cal, and biological (collectively, “environmen- are initially charged with Sr from the bedrock
tal”) drivers, and then introduce a two- basalt exist in a dynamic balance where Sr lost
endmember classification system for spatial through weathering and leaching from soils is
patterning of isotopic variation. replenished by Sr inputs from oceanic aerosols.
Because the basalt and marine Sr sources have
very different 87Sr/86Sr values, it is primarily
land surface age, in turn determined by the age
3.3.1 Spatial Organization and Isotope
of the bedrock basalt flows, that is the primary
Patterning determinant of Sr isotope ratios in soils and
Stable isotope variation in environmental plants. The ages of basalt flows, both on the Big
components reflects the spatial patterning of Island of Hawaii (Fig. 3.1A) and throughout the
environmental factors in three primary ways. island chain, are spatially patterned, creating a
First, geographic location can determine the strong spatial patterning of Sr isotope ratios in
elemental sources available for incorporation Hawaiian island food webs.
in biological substrates, and thus impact the In isotopic systems in which appreciable
isotopic composition of ecologically relevant isotopic fractionation occurs in the environ-
materials. Second, where isotopes are fraction- ment and biological systems, spatial variation
ated by environmental processes, the magni- in the local environmental conditions under
tude of isotope fractionation at a given site which isotope-fractionating processes occur
will respond to local environmental conditions. can have a large impact on the isotopic compo-
Third, for systems in which large-scale geo- sition of plants and animals. The photosyn-
graphic transport is important, the spatial thetic assimilation of carbon by plants, e.g., is
organization of transport processes (e.g., atmo- a highly fractionating process which uses car-
spheric circulation, runoff, and groundwater bon from a relatively isotopically homoge-
flow) will determine both the isotopic source neous environmental substrate, atmospheric
and the integrated history of fractionating pro- CO2. As such, there is relatively little potential
cesses affecting material in the local environ- for spatial isotopic variation in elemental
ment. Each of these modes relating isotopic sources to drive variation in plant δ13C values
variation to spatially varying environmental (with exceptions such as near urban centers,

FIGURE 3.1 Spatial patterning of an environmental property underlying variation in ecosystem Sr isotope ratios on
the Big Island of Hawai’i. (A) The spatial distribution of approximate landscape surface ages (after Trusdell, Wolfe, &
Morris, 2006) on the Big Island of Hawai’i. (B) A strong relationship exists between surface age and plant 87Sr/86Sr, reflect-
ing the decreased availability of basalt-derived Sr and increased accumulation of Sr from sea-salt aerosols with greater
surface age (Kennedy et al., 1998).
Pataki, Bowling, & Ehleringer, 2003), but great spatial variation in environmental factors is
potential for variation in the magnitude of photo- found in cases where transport history is a pri-
synthetic fractionation to produce spatial δ13C mary determinant of isotopic composition. We
variation. For C3 ecosystems, photosynthetic dis- have seen, e.g., the isotopic composition of
crimination is largely controlled by environmen- meteoric precipitation is dependent, among
tal parameters such as soil water availability, other factors, on the condensation history of
temperature, and atmospheric moisture content, the air mass from which the precipitation
which influence the gas exchange physiology of forms. As a result, δ2H and δ18O values of pre-
leaves. As a result, strong relationships often cipitation cannot be predicted a priori without
exist between climate variables such as tempera- consideration of the trajectory taken by the
ture or precipitation and plant or ecosystem atmospheric moisture reaching the site at
δ13C, leading to spatial patterning of carbon iso- which the precipitation condenses. Spatial
tope ratios that mimics variations in climate. organization clearly exists at many levels
This is nicely illustrated by measurements of the within the climate system, however, including
δ13C of ecosystem respiration along a sampling atmospheric circulation patterns and the tem-
transect in western Oregon (Fig. 3.2; Bowling, perature gradients that lead to condensation
McDowell, Bond, Law, & Ehleringer, 2002), from air masses, leading to spatial patterning
which demonstrates a strong relationship in transport and condensation history of pre-
between ecosystem carbon isotope ratios and cipitating air masses. Waters sampled along a
precipitation amount along a strong, coast-to- gradient in mean annual precipitation amount
continent gradient in precipitation amount. from the northern California coast to eastern
The third, and most complex, type of rela- Nevada (Fig. 3.2; Ingraham & Taylor, 1991),
tionship linking spatial isotopic variation to e.g., illustrate a strong, spatially determined

3.3 PATTERN 61
FIGURE 3.2 Landscape-level stable isotope variation related to spatial patterning of climatic gradients. (A) Strong gra-
dients in mean annual precipitation amount (Daly, Neilson, & Phillips, 1994) characterize two western USA isotope sam-
pling transects (bold lines). (B) Carbon isotope ratios of ecosystem respiration (dots, estimated from Keeling plots; Bowling
et al., 2002) show a strong relationship to position along the Oregon transect, and are broadly, negatively correlated with
precipitation amount (line). This reflects the strong relationship between local fractionation due to plant gas exchange pro-
cesses and climatology along the spatial climate gradient. (C) Hydrogen isotope ratios of meteoric water samples (ground-
water, springs, and streams; Ingraham & Taylor, 1991) vary continuously along a transect in northern California and
Nevada are positively correlated with changes in precipitation amount along the coast to interior gradient. The gradients
in the water isotopic composition and precipitation amount can each be related to variation in the processes governing air
mass water balance and transport along the dominant westerly atmospheric circulation trajectory for this region.
relationship between precipitation amount and isotopic patterning can take on a range of
isotopic composition, which reflects the pro- forms, depending on the spatial distribution of
gressive drying of air masses leaving the variation in environmental parameters. Its
northern Pacific Ocean and traversing western form has important implications for the types
North America. of migration research applications to which
each isotope system is suited. In this context,
we distinguish two endmember classes: spa-
tially continuous and spatially discrete patterns.
3.3.2 Two Classes of Isotopic Patterns Depending on the environmental determinants
Spatial patterning of isotopic variation in of spatial variation in a given isotope system
many systems closely mimics that of the and the spatial scale in question, isoscapes can
underlying environmental determinants. This depict isotopic variation that is spatially

continuous, discrete, or intermediate to these Berry, & Clark, 1998; Ehleringer, Cerling, &
endmember classes. Helliker, 1997), and in many cases the distribu-
Spatially continuous isoscapes are character- tion of these plant types can be considered to
ized by smooth variation in isotope ratios be discrete, particularly over large spatial scales
across space, giving continuous isotopic gradi- (Powell, Yoo, & Still, 2012; Still et al., 2003). The
ents along map transects. These isoscapes issue of scale is highly relevant to the concept
occur in cases where the underlying environ- of discrete isoscapes, as both transport pro-
mental determinants vary continuously in cesses and statistical variance in populations
space and the process through which they tend to smooth the boundaries of discrete iso-
impact landscape-level isotope ratios is a con- scapes, particularly at small spatial scales.
tinuous function. Continuous isoscapes are The recognition and distinction of isoscape
common but not ubiquitous where climatologi- pattern classes is critical for terrestrial migra-
cal factors drive isotope ratio variations: for tion ecology applications in that they largely
example, the continuous (though not unidirec- determine the types of ecological questions to
tional) change in precipitation amount that is which different isotopic systems are applica-
related to ecosystem δ13C values in western ble. Spatially continuous isoscapes offer impor-
Oregon or the progressive rainout of moisture tant, recognized opportunities for research in
during westerly circulation driving precipita- migratory connectivity, particularly at large
tion δ2H variation across northern California spatial scales. Because isotope ratio variation
and Nevada (Fig. 3.2). within continuous isoscapes is largest along
Spatially discrete isoscapes represent the directional, environmental gradients, these iso-
contrasting pattern of spatial variation in which scapes are most amenable to addressing ques-
isotope ratios of environmental substrates are tions dealing with the position of individuals
relatively invariant over some areas but change along geographic gradients. The continuous
abruptly and discretely across boundaries isotope ratio variation across these isoscapes
between invariant zones. Such patterns of “pat- can confound attempts to apply them to ques-
chy” spatial isotopic variation can occur where tions of explicit assignment of individuals to
environmental drivers vary discretely across discrete locations, although statistical assign-
space or where continuously varying environment techniques have been successfully
mental factors influence isotopic variability applied to questions that can be posed as
according to discrete functions. Isotopic varia- assignment to regions (Wunder, Kester, Knopf,
tion of Sr in Hawaiian island ecosystems & Rye, 2005; Wunder and Norris, this volume).
(Fig. 3.1) represents an excellent example of the Spatially discrete isoscapes are well suited in
first scenario, where the aerial extent of individ- some cases to address assignment questions
ual lava flows define land surfaces of common where the questions and constraints can be
age, presumably characterized by similar eco- well defined and overlap with isotopically
system Sr isotope ratios, and the discrete defined spatial domains. However, isotope
boundaries of these flows segment the isotopic systems that are characterized by discrete spa-
landscape. The second scenario can be illus- tial patterning cannot be applied to assignment
trated by the case of natural, climate-induced at spatial scales below that of the patches com-
variation in the distribution of C3 and C4 plants. prising the isoscape. If the scale of the ques-
Although driven by continuously varying cli- tions of interest in a particular study is too
mate parameters, the impact of climate on C4 small, e.g., these isotope systems may provide
distribution (and thus ecosystem δ13C values) is little useful information. For either type of pat-
described by a threshold function (Collatz, tern, knowledge and consideration of the

3.4 MAPPING ISOSCAPES 63
environmental isoscape in question during the processes determining spatial isotopic variabil-
project planning stage is warranted to assess ity within the system. These may range in
the potential of the system to provide informa- complexity and precision from purely theoreti-
tion addressing the questions being posed. cal first-principals constructs, to highly derived
Published applications of isotopes to terres- empirically calibrated functions. In many
trial migration research have capitalized on cases, tradeoffs exist between the accuracy of a
both classes of patterns. With the advent of model and its ability to be generalized or
hydrogen isotope ratios as a tool for wildlife applied over large spatial scales. Appropriate
forensics (Chamberlain et al., 1997; Hobson & model selection is critical to obtaining an accu-
Wassenaar, 1997) much emphasis has been rate and well-constrained isoscape for the sys-
focused on the application of continuous iso- tem and spatiotemporal domain under
scapes of δ2H in water to problems of migratory consideration, and these factors together with
connectivity. However, several examples focus- the availability of data and intended uses of
ing on reconstructing the migration of modern the product should guide the choice of model.
or ancient animals and humans continue to The least-specific class of models that can be
exploit discrete spatial variability of isotope applied to any isotopic system are geostatisti-
ratios, in particular those of Sr (Buzon, Conlee, cal interpolation models. These models
Simonetti, & Bowen, 2012; Chamberlain et al., describe the variation in isotope ratios in terms
1997; Hoppe et al., 1999). The relative strengths of location alone. Estimates are calculated as a
and weaknesses of each class of isoscape function of observed values at nearby loca-
should be considered in future application, and tions, meaning that the only data required are
in particular we suggest that important oppor- observations of the isotope ratio at several
tunities exist for coupling isotope systems with points within the spatial domain of interest.
continuous and discrete isoscapes character- A range of interpolation models of varying
ized by variation at different spatial scales. complexity are available. Some of the simpler
examples, such as triangulation, nearest neigh-
bor, and inverse distance techniques, are
widely applied in data exploration but are
3.4 MAPPING ISOSCAPES
prone to artifacts associated with sample dis-
tributions and offer limited potential to quan-
Isoscape maps are created through a multi-
tify the uncertainty of predictions. Kriging, a
step process that is often iterative (Bowen,
class of procedures in which the weights
2010). Information on isotope fractionating pro-
assigned to data are determined by a model of
cesses and patterns, represented as models, is
the covariance structure of the observations,
combined with geospatial data to produce site-
offers a more robust approach to the interpola-
specific isotope ratio predictions over a spatio-
tion problem (Isaaks & Srivastava, 1990).
temporal domain of interest. Major steps in the
Kriging has been applied widely in prediction
process include model selection and calibra-
of isoscapes for a range of isotope systems, but
tion, data acquisition, calculation, optimization
recent literature has explored other model fra-
of model residuals, and estimation of error.
meworks such as generalized additive models
for the interpolation process as well (Wang,
Hou, Masson-Delmotte, & Jouzel, 2010).
3.4.1 Model Selection and Calibration Interpolation models offer a convenient way of
For many isotope systems, multiple models visualizing and extending observations from a
are available or conceivable that describe the well-developed sampling network across

space, but the degree of detail in the resulting A third class of models includes first-
isoscape is entirely limited by that represented principles and process-based parameteriza-
in the measured or observational dataset. As a tions. Because of the complexity of the
result, the use of interpolation models in isola- processes underlying isotopic variability, there
tion is best restricted to cases where the obser- are few cases where models can be fully devel-
vational documentation of spatial isotopic oped based on first principles alone. Where
variation is extensive. As discussed in the fol- such models could be developed, it is likely
lowing sections, however, interpolation models that they would necessarily oversimplify the
can be used to improve isoscape accuracy systems of interest, limiting their applicability.
when used in combination with other types of For example, first-principles models of the
models. radiogenic production of 87Sr have been
More specialized models can be constructed applied, using data on rock ages and Rb con-
using derived parameters as proxies for the tents, to produce bedrock 87/86Sr maps, but the
environmental factors underlying spatial isoto- relevance of the resulting isoscapes to ecologi-
pic variation. These models are particularly cal 87/86Sr applications is limited by other fac-
useful in cases where the complexity or data tors influencing Sr bioavailability but not yet
requirements of first-principles or process- represented in the models (Bataille & Bowen,
based models are prohibitive, but where a 2012; Bataille et al., 2012). In contrast, process-
simpler or more readily obtained suite of sur- based models strive to faithfully represent the
rogate parameters can be substituted in their physical and chemical processes underlying
place. Derived models have been used to cre- isotopic variation but adopt parameterizations
ate global isoscapes of precipitation isotope that may group or simplify the details of these
ratios by simplifying the extremely complex processes. For example, commonly used mod-
suite of environmental factors underlying spa- els of photosynthetic 13C fractionation by C3
tial isotopic variation in precipitation to a plants explicitly represent the biophysics of
derived model of the form: CO2 gas exchange by leaves but incorporate
species-specific or biome-specific parameters
δ 5 a L2 1 bðLÞ 1 cðAÞ 1 d; (3.4) to describe the biological regulation of gas
where δ is the isotopic composition of precipi- exchange by the stomata (Ball, Woodrow, &
tation, L is latitude, A is altitude, and a, b, c, Berry, 1987; Lloyd & Farquhar, 1994). These
and d are empirically fitted parameters (Bowen model parameters must again be calibrated
& Wilkinson, 2002). As this example shows, against field data, but given the more funda-
the parameters used in derived models may be mental nature of these variables this work can
only indirectly related to the physical pro- often be done through laboratory studies or
cesses driving isotopic variation, but may be experiments and extended to the spatiotempo-
useful surrogates due to correlation with the ral domain.
first-order parameters. In most cases, parame-
ter values will need to be calibrated relative to
observational data, but if the model is
3.4.2 Geospatial Data
stable and the parameters are skillfully chosen, An overwhelming amount of geospatial
some derived models will be capable of data is available to support scientific research
extrapolation to combinations of parameter in the form of digital data archives, reanalysis
values not represented in the observational and data synthesis projects, GIS database
dataset. and decision support tools, and real-time

Earth-observing satellite data. Many types of grid of regularly spaced cells, each containing
geospatial data are relevant to isoscapes a single value. Rasters are a limited data for-
modeling, including physiographic informa- mat, in that each grid can only contain values
tion (e.g., position data such as latitude and for a single parameter, but they are useful for
longitude, elevation, and land surface slope), isoscapes modeling in that they are continuous
climate data (e.g., temperature, precipitation (i.e., they provide values or “no data” indica-
amount, and atmospheric humidity), geologi- tors for the entire spatial extent of the raster)
cal data (e.g., bedrock geological maps and soil and many rasters can be formatted uniformly
types), hydrological data (e.g., stream routing, to allow them to be combined and analyzed
aquifer distribution and depth), biological data together. Many geospatial data are distributed
(e.g., species or biome distributions, leaf area in raster format, and others can be easily con-
index, and the normalized differential vegeta- verted (e.g., a polygon that represents the
tion index), and socioeconomic and demo- aerial extent of a biome can be converted to a
graphic data (e.g., distribution of crop raster grid of 1’s and 0’s representing the pres-
production, population density, and land use). ence or absence of the biome across a spatial
Major data distributors include the World domain). The geometry of rasters is described
Data Center System, the National Aeronautics in terms of their extent (what are the bound-
and Space Administration, the National Center aries of the data?), resolution (what is the size
for Atmospheric Research, Oak Ridge National of an individual grid cell?), and projection
Laboratory Distributed Active Archive Center, (what coordinate system is used to describe
the National Oceanic & Atmospheric the spatial relationship between cells?). To
Administration, the US Geological Survey, the allow accurate calculation of isoscapes, raster
Climatic Research Unit, the European Centre data must be processed so that each of these
for Medium-Range Weather Forecasts, and properties is uniform across all datasets.
synthesis projects such as The International Lastly, many isoscapes models require spa-
Satellite Land-Surface Climatology Project. tially distributed observations of isotope ratios
Depending on the data type and provider, dif- as input or for calibration purposes. A few out-
fering levels of spatial and temporal coverage standing spatial observation networks have pro-
may be available: many datasets are available duced valuable spatially resolved isotopic
for individual states or countries, and the tem- datasets, including several IAEA programs such
poral sampling interval of satellite-gathered as the Global Networks for Isotopes in
data may be widely different from ground- Precipitation and Rivers, the US National
based observational products. These issues Oceanic and Atmospheric Administration’s
must be taken into account and reconciled, Cooperative Air Sampling Network, and the
e.g., through creating mosaics of data from United States Geological Survey’s North
multiple sources or averaging data over a com- American Stream Quality Accounting Network.
mon time window, before the data can be used In some cases significant, spatially extensive
in isoscapes models. (although temporally limited) data collections
Because the end goal of isoscapes modeling have resulted from investigator-driven research
is to produce continuous surfaces of isotopic (e.g., Bowen et al., 2007; Good, Mallia, Lin, &
variation in space, the data input to these mod- Bowen, 2014; Longinelli & Selmo, 2003). In addi-
els must be amenable to representation in ras- tion to these two modes of new data generation,
ter format. A raster is a representation of data the types of large-scale datasets required for iso-
in two or more dimensions (e.g., latitude and scape modeling are increasingly being compiled
longitude for many geospatial rasters) as a by researchers involved in metaanalysis or other

synthesis research projects (e.g., Hobson, Van In addition to the actual process of model
Wilgenburg, Wassenaar, & Larson, 2012; execution, important decisions must be made
Jasechko et al., 2013; http://wateriso.utah.edu/ at this step about the extent and resolution of
waterisotopes/pages/spatial_db/SPATIAL_DB. the isoscape that will be generated: how large
html). To the degree that these datasets can be an area will be modeled and at how fine a spa-
themselves aggregated and shared, they may tial division? The question of extent may be
offer another valuable source of isotope data one that is answered largely by practical con-
supporting improvements in isoscape modeling. straints, e.g., the aerial coverage required to
encompass the likely range of a migrant popu-
lation or the extent of a requisite geographic
dataset. In making decisions about the spatial
3.4.3 Model Calculations extent of modeling, however, it is important to
Following model selection, calibration, and consider the extent of any calibration data that
data assembly, a provisional isoscape is created were used and critically assess whether they
by executing the model calculations on a spa- are sufficient to support application of the iso-
tial grid using spatial data input layers. In scapes model at the desired extent or whether
most cases, this procedure is simply that of unwarranted extrapolation will be involved.
iteratively executing a set of calculations that Decisions about the spatial resolution at
solve the model equations at each grid cell in which model calculations are applied are in
the spatial domain. Model calculations thus many cases more subjective but are equally
require a computational routine that iterates important. A number of factors must be con-
through the cells, reads data for that cell from sidered in the selection of an appropriate spa-
the required data rasters, executes the model tial resolution for a particular isoscape and
calculations, and outputs the result to a new ecological application. Biological factors, e.g.,
output raster. These functions can be accom- the size of an individual’s range on the breed-
plished through hand-coded routines consist- ing or wintering grounds, will be relevant to
ing of iterative loops, file input/output determining the maximum resolution that is
statements, and mathematical operators written appropriate for a particular application. In a
in any programming language that supports more general sense, the number, density, and
these functions (e.g., BASIC, FORTRAN, C, geographic specificity of calibration data will
C11, and Java). They can also be implemen- have a strong influence on the degree of spa-
ted directly in a wide range of open source and tial specificity that is appropriate for the
commercial scientific software and program- modeling work. In this regard, it is important
ming environments that offer GIS functionality for both modelers and isoscape users to recog-
(e.g., ArcGIS, GMT, GRASS, GrADS, Python, R, nize that higher spatial resolution is not neces-
and SURFER) in the form of prebuilt tools that sarily better. In many cases, attempts to
accomplish lower level data handling (e.g., predict isotope distributions at high spatial
reading a raster data file or iterating through resolutions may actually compromise the over-
the spatial grid) transparently. In many cases, all quality of the resulting data products: if the
this can increase the efficiency of the calcula- resolution at which the calibration and calcula-
tion and routine-building process. Some of tion work is conducted exceeds that of the
these GIS offer well-developed graphical user physical spatial processes determining isotopic
interfaces that allow isoscape development to variability this can introduce artifactual or
be conducted by researchers having little or no over-specified dependence of the predicted
computer programming experience. isotope ratios on model variables.

3.4.4 Optimization of Residuals improving isoscape accuracy, however,

through the use of a geostatistical interpolation
Models, as simplified descriptions of the model. The geostatistical model can be applied
processes they represent, are inevitably only to interpolate a grid of predicted residual
approximations of reality with associated values that can be added to the isoscape model
errors and uncertainty. This is true of isoscape predictions to “correct” them against the obser-
models, which are often limited in their com- vational data. An early analysis of precipitation
plexity by the availability of spatial input data isotope ratios by Bowen and Wilkinson (2002)
that would allow their application over the illustrates the use of spatially autocorrelated
spatial scales of interest. The mismatch residuals (Fig. 3.3). In this case, spatially pat-
between model predictions and observational terned residuals suggested that aspects of the
data, known as residuals, can often be partly climatology and atmospheric circulation not
attributed to uncertainties in the data or the represented by the derived isoscape model
model parameters, but may carry important used in the study had a significant impact on
information about the inadequacies of the precipitation isotope ratios. This information
model. Where spatial datasets documenting could be used to develop an improved model
the isotopic values of interest at points within parameterization that incorporated these envi-
the modeling domain are available, isoscape ronmental factors, or, as was done in the cited
modeling offers two powerful ways to take study, to develop a residual correction using
advantage of residuals in order to optimize the geostatistical interpolation. Although this pro-
end data product. cess can be conducted and/or conceived of as
First, examination of model residuals can an independent “step” in the analysis work-
often produce insight into the model through flow, residual correction can also be incorpo-
comparison with ancillary datasets, leading to rated as a component of the primary isoscape
iterative improvements in the modeling. model, allowing all model components to be
Residuals can be considered test cases, and any optimized and applied simultaneously (Bowen,
mismatch with observational data may high- Kennedy, Henne, & Zhang, 2012; Bowen &
light missing processes that could be incorpo- Revenaugh, 2003).
rated to improve future modeling. GIS software
facilitates the comparison of residual values
with the wide range of available spatial data by
3.4.5 Uncertainty of Isoscape
allowing users to intersect and extract data at
specified locations from multiple datasets: for
Predictions
example, residual values from a network of iso- As we have shown, isoscapes are commonly
tope observing sites could be referenced against the product of multilevel modeling efforts
data showing land use categories to produce a involving derived spatial data products,
comparison of these data across all sites. imperfect model parameterizations, and geos-
Second, in many cases residuals are nonran- tatistical analysis. Developing quantitative
domly distributed in space. This spatial measures of the uncertainty of isoscapes pre-
autocorrelation presumably reflects some dictions represents an important step in the
location-dependent process that was incom- robust application of these products to migra-
pletely represented in the model, and analysis tion research. Prediction error, defined as the
of residual spatial structure can itself offer difference between an isoscape prediction at a
insight into missing model processes. Spatial given site and the true value of the modeled
autocorrelation offers another avenue to variable, is itself spatially variable. As such,

FIGURE 3.3 Model residuals (δ18O, m) for isoscape predictions of long-term, mean annual precipitation isotopic com-
position (reprinted from Bowen, G.J., & Wilkinson, B. (2002). Spatial distribution of δ18O in meteoric precipitation.
Geology, 30(4), 315318). Residual values were calculated as observed values—model predictions at precipitation moni-
toring sites. Low residuals values over the northern hemisphere continental interiors can be attributed to the extensive
rainout of 18O as air masses traverse the continents. High residuals over the mid- and high-latitude oceans can be attrib-
uted to extensive evaporation from the warm surface waters of the gyres in these regions. The large magnitude positive
residuals in eastern Africa reflect a combination of factors including proximity to Indian Ocean water sources, reevapora-
tion of falling rain, and relatively low decrease of isotopic values with elevation in this region.
the most useful data products documenting of accurate models for the relationship between
uncertainty are maps of standard errors, confi- tissue isotopic composition and that of the envi-
dence intervals, or related statistics across the ronmental substrates modeled by isoscapes are
modeling domain. Such maps are now fairly important sources of uncertainty that are
routinely produced in studies where isoscapes beyond the scope of this chapter.
are developed based on statistical models (e.g., Prediction error can be partitioned into two
Bowen & Revenaugh, 2003; Terzer et al., 2013), components: data error and model error.
but present a greater challenge and are less Sources of error in the data products used to cal-
commonly provided for first-principle based ibrate, drive, or optimize spatial stable isotope
isoscapes (e.g., Bataille & Bowen, 2012). It is models can be related to the handling and analy-
important to note that prediction error as dis- sis of individual samples (e.g., evaporation of
cussed here represents only one component of water from improperly stored samples prior to
error affecting ecological interpretations devel- H and O isotope ratio analysis), lack of heteroge-
oped using isoscapes. Selection of appropriate neity within large datasets (e.g., using data
isoscapes for application to a particular ques- having uneven temporal coverage to estimate
tion (e.g., representing the appropriate period long-term average isotopic values), or the gener-
of temporal averaging; Bowen et al., 2005; ation of derived data products (e.g., errors in cli-
Vander Zanden et al., 2014) and development mate model reanalyses or interpolated climate

3.5 ISOSCAPES FOR TERRESTRIAL MIGRATION RESEARCH 69
rasters). Model errors relate to the parameteriza- absence of local data, however, the inability of
tion of natural processes used for modeling and the fixed altitude effect parameterization to
the inaccuracies in these parameterizations. For correctly predict isotope ratios in this region
example, maps of precipitation isotope ratios would not have been recognized. In such
were produced by Bowen and Wilkinson (2002) cases, the only clear routes to improving
using a fixed coefficient for variation with alti- isoscape models and the estimation of error
tude (parameter c in Eq. 3.4) at all sites. This are through the parallel paths of expanded
simplifying and inaccurate assumption has since spatial data collections and improved model
been shown to impart bias to predictions in parameterization.
some parts of the world, e.g., the East African
highlands (Bowen & Revenaugh, 2003; Otte
et al., 2017; Terzer et al., 2013). The distinction
3.5 ISOSCAPES FOR TERRESTRIAL
between data and model error is often blurred
MIGRATION RESEARCH
in isoscapes modeling because many of the data
products used are themselves model-derived.
In this final section, we review many of the
Several methods are available for quantifica-
currently available isoscapes data products as
tion of isoscape prediction uncertainty, all of
they relate the study of terrestrial animal
which provide accurate assessment of data
migration.
error but may incompletely represent model
error. Metrics of error can be derived from the
covariance matrix in kriging or through error
propagation (assuming estimates of parameter
3.5.1 Water H and O Isoscapes
and variable uncertainties are available) in Isoscapes of H isotopes in water, particu-
process-based and derived-parameter models. larly in precipitation, have been widely
Resampling statistics (e.g., cross-validation, applied in migration research based on the
jackknife, and bootstrap methods; Wu, 1986) premise that the dominant source of hydrogen
can also be applied to generate estimates of in body tissues is environmental water, either
uncertainty in any of these cases, and while consumed directly or routed through diet (e.g.,
these methods are computationally intensive Bearhop et al., 2005; Chamberlain et al., 1997;
they have the significant advantage of being Hobson & Wassenaar, 1997; Norris, Marra,
insensitive to assumptions about probability Montgomerie, Kyser, & Ratcliffe, 2004).
distribution functions and covariance functions Analogous application of O isoscapes has been
of model parameters (i.e., they are nonpara- investigated in at least one case (Hobson,
metric methods). Each of these methods will Bowen, Wassenaar, Ferrand, & Lormee, 2004),
quantify error associated with noisy data by but so far has shown less promise. The most
calculating its impact of the precision with commonly referenced products are maps of
which model parameters are known. Model long-term average growing season precipita-
error, however, will only be adequately repre- tion isotope ratios (Fig. 3.4; e.g., Bowen et al.,
sented where data are available to document 2005; Meehan et al., 2004; Terzer et al., 2013)
its impact on the precision of the model. In the produced by derived-parameter modeling of
case of East African precipitation isotope ratios monthly data with geostatistical residual cor-
given earlier, e.g., the subsequent error analy- rection. These products are freely available on
sis of Bowen and Revenaugh (2003) was able the Internet, and although uncertainty esti-
to identify high uncertainty for estimates in mates specific to the growing season products
this region based on the data from a monitor- have not been produced, confidence intervals
ing station in this region (Addis Abba). In the for related annual average precipitation

FIGURE 3.4 Global isoscape of long-term average growing season precipitation δ2H values. This map represents the
precipitation amount-weighted average of monthly isoscapes produced using a derived parameter model (modified ver-
sion of Eq. 3.4) and interpolated residual correction. Climate data (temperature and precipitation) was interpolated from
the Global Historical Climatological Network (Peterson & Vose, 1997). Source: Modified from Bowen, G. J., Wassenaar, L. I.,
& Hobson, K. A. (2005). Global application of stable hydrogen and oxygen isotopes to wildlife forensics. Oecologia, 143(3), 337348.
doi:10.1007/s00442-004-1813-y.
isoscapes (Bowen & Revenaugh, 2003; Terzer climatological data accumulated over many
et al., 2013) provide a general indication of the decades of monitoring and measurement.
potential error in the growing season maps. Although the use of large, long-term average
The nature and extent of isotopic variability datasets increases the accuracy of long-term
across these isoscapes has been reviewed by average precipitation isoscapes (Bowen &
Bowen et al. (2005). In general, they provide Revenaugh, 2003), it is not clear how closely
the greatest power to constrain the location of these predictions reflect the environmental iso-
migration endpoints and pathways within the topic signal taken up by migrant individuals
northern hemisphere continental interiors, and that inhabit a location for a period of weeks or
in some cases may also be useful for differenti- months during a particular dry or wet year
ating habitats along altitudinal gradients. (Vander Zanden et al., 2014, 2015). Similarly,
Despite the temporally dynamic nature of H the rationale for restricting the data used to the
and O isotope ratios in precipitation (Bowen, “growing season” (usually defined as all months
2008; Good et al., 2014), most precipitation iso- with mean temperature . 0 C) has been that
scapes currently reflect long-term average (cli- water assimilated by plants and entering the
matological) values. This reflects the sparseness food chain will be primarily derived from
of water isotope data for individual regions growing-season precipitation, but other seasonal
and times. In order to achieve high levels of or annual isoscapes may be more relevant in
accuracy in continental to global scale analyses, some cases (e.g., Bowen et al., 2005). IsoMAP
it has generally been necessary to work with (http://isomap.org), an online water isoscape

modeling toolkit, offers a potential resource for yet available, but partial cross-validation
researchers who want to develop and explore results for the river water product of Bowen
time-specific isoscapes that may be more appro- et al. (2011), e.g., suggest root-mean-squared
priate to a specific migratory study system. The error values on par with those of precipitation
IsoMAP system allows users to identify and isoscapes for the same area (Bowen et al., 2011).
extract precipitation isotope data for particular
regions and time periods and develop new iso-
scapes using these data (Bowen, West, et al.,
2012). Although in most cases the accuracy of
3.5.2 Vegetation C Isoscapes
the resulting products will be lower than that Carbon isotope ratios have been primarily
achieved with climatological data, reduction of applied to studies of animal diet, largely
bias through the use of more temporally specific because of the distinct isotopic signal associ-
data may outweigh this in some applications. ated with the relative proportion of C3 versus
For some migratory animals H or O iso- C4 plants in the diet (Cerling, Harris,
scapes of precipitation will not be the appropri- Ambrose, Leakey, & Solounias, 1997;
ate choice for isotope tracking work because MacFadden & Cerling, 1994; Peters & Vogel,
individuals obtain these elements from drink- 2005; Sponheimer & Lee-Thorp, 1999). They
ing water or food webs based in hydrological have not been as widely used to study migra-
pools with properties that modulate natural tion (aquatic organisms being an important
patterns of precipitation isotope variability exception to this), perhaps due to disagree-
(e.g., river systems or irrigation water). In these ment in the literature over the utility of δ13C in
cases isoscapes representing other water yielding geographic information (Chamberlain,
sources could help to improve and expand the Bensch, Feng, Akesson, & Andersson, 2000;
use of water isotopes in migration research. Hobson et al., 2003; Wassenaar & Hobson,
Isoscapes of surface (river) water (Bowen et al., 2001). Vegetation δ13C isoscapes have signifi-
2011; Fekete, Gibson, Aggarwal, & Vorosmarty, cant potential to provide important insights by
2006) and tap water (Bowen et al., 2007) have allowing one to compare, e.g., spatially distrib-
now been produced for the contiguous United uted data with continuous grid predictions, or
States, and may be relevant to tracing some by providing an interpretation platform for
migrants, e.g., with aquatic habits or that might bird feathers produced in different but
be known to feed primarily in irrigated agricul- unknown locations. Significant spatial informa-
tural habitats (e.g., Hobson, Doward, Kardynal, tion may therefore be found from model pre-
& McNeil, 2018). These products have been dictions that incorporate the distribution of C3
developed using precipitation isoscapes to rep- and C4 plants and climatic drivers. By combin-
resent the isotope flux to the land surface and ing models of C3 versus C4 plant distributions,
modeling the modification of this signal using biophysical models of plant carbon isotope
hydrological (surface water) or geostatistical fractionation, and biome distributions derived
(tap water) models. The first-order patterns of from satellites to allow biome-specific plant
these isoscapes are similar to those of precipita- physiology to be incorporated, global plant
tion, but the surface and tap water isoscapes δ13C isoscapes have been produced (see
demonstrate significant differences relative to Fig. 3.5; Lloyd & Farquhar, 1994; Scholze,
precipitation products particularly in moun- Kaplan, Knorr, & Heimann, 2003; Suits et al.,
tainous regions and along the course of large, 2005). These plant δ13C isoscapes provide a
mountain-fed rivers. Comprehensive assess- useful framework for understanding observed
ments of the accuracy of these products are not spatial information in bird tissue δ13C (Pain

NP
(A)
60
30
EQ
–30
–60
SP
180 120 W 60 W 0 60 E 120 E 180
–28.0 –27.7 –27.3 –27.0 –26.7–26.3–26.0 –25.7 –25.3–25.0–23.5 –22.0 –20.5 –19.0–17.5 –16.0–14.5 –13.0
NP
(B)
60
30
EQ
–30
–60
SP
180 120 W 60 W 0 60 E 120 E 180
0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95
FIGURE 3.5 Global mean (A) and standard deviation (B) annual plant δ13C (m). These plant carbon isoscapes are hybrid
products derived from global distribution maps of C3/C4 vegetation (derived from satellite products and physiological model-
ing) and modeled physiological responses of C3 plants to atmospheric conditions for the years 198393 (continuous fields
from ECMWF) and constrained by the Normalized Difference Vegetation Index for those years. Source: Reproduced with per-
mission from Suits, N. S., Denning, A. S., Berry, J. A., Still, C. J., Kaduk, J., Miller, J. B., & Baker, I. T. (2005). Simulation of carbon
isotope discrimination of the terrestrial biosphere. Global Biogeochemical Cycles, 19, GB1017. doi:10.1029/2003GB002141.

et al., 2004), assuming one understands the 3.5.3 Vegetation N Isoscapes
relationships between birds and their food
source. It is of course necessary also to have Although mapping spatial variation in plant
some confidence in the relationship between δ15N has received less attention than plant
the isoscape prediction and the actual food δ13C, there is at least one published set of
source (e.g., if the model predicts leaf δ13C, global plant and soil δ15N maps (Amundson
and the bird eats primarily seeds, what is the et al., 2003). This modeling effort was based on
relationship between seed and leaf δ13C?). prior arguments that plant δ15N is related to
Several recent papers have explored the poten- the residence time of N in an ecosystem or N
tial for C isoscapes to be combined with others cycle “openness,” as well as empirical observa-
and report utility in identifying origins using tions consistent with this expectation that plant
this approach (Fox et al., 2017; Hobson & δ15N is negatively correlated with precipitation
Kardynal, 2016; Werner, Hobson, Van (e.g., Austin & Vitousek, 1998; Handley et al.,
Wilgenburg, & Fischer, 2016). If these variables 1999). Temperature is also positively correlated
can be understood, plant δ13C isoscapes offer with plant δ15N values and this relationship is
the potential for sophisticated interpretations. part of the Amundson et al. (2003) model
Greater exploration of these interfaces is (see Fig. 3.6). It has also been observed that
clearly warranted. animal δ15N is negatively correlated with
–180º –150º –120º –90º –60º –30º 0º 30º 60º 90º 120º 150º 180º
60º 60º
30º 30º
0º 0º
–30º –30º
–60º –60º
–180º –150º –120º –90º –60º –30º 0º 30º 60º 90º 120º 150º 180º
–8.5 –4.3 –3.6 –3.1 –2.7 –2.4 –2.1 –1.6 –1.1 –0.6 0 0.5 1.1 1.7 2.2 2.8 3.3 3.9 4.7
Plant δ15N (‰)
FIGURE 3.6 Global plant δ15N (m). This plant nitrogen isoscape was produced by executing a regression model in GIS
previously fit to observed plant δ15N and mean annual precipitation and temperature (Amundson et al., 2003). The model
was driven with observed climate parameters (continuous fields of MAP and MAT for the climate normal period 196190
from the Climate Research Unit; New, Lister, Hulme, & Makin, 2002). The model output was further masked using continu-
ous vegetation fields (DeFries, Townshend, & Hansen, 1999) eliminating areas with greater than 80% nonvegetated ground.

precipitation, a pattern that could be caused likely that a greater understanding of all sig-
by both changes in dietary δ15N or animal nificant sources of H and O inputs to animal
metabolism. Recent work suggests that the pat- metabolism would yield better predictive abil-
tern is the result of variation in dietary δ15N ity, making plant H and O isoscapes poten-
and not variation in animal metabolism indi- tially useful for understanding animal
cating that animals (in this case kangaroos) movement. This is especially true for animals
faithfully record dietary δ15N and that dietary (i.e., herbivores) that may obtain a significant
δ15N is itself linked to climate, likely through amount of their body water from plant water.
its effect on N cycle openness (Murphy & Plant δ2H and δ18O isoscapes generally are
Bowman, 2006). Support for retention of this derived using a combination of approaches.
geographic signal has also been found for war- Plant processes that discriminate against 2H or
18
blers (Chamberlain et al., 2000). Smaller scale O are modeled explicitly using biophysical
efforts have also found useful relationships models of the fractionation. These models are
with elevation (Voigt, Helbig-Bonitz, Kramer- themselves driven by parameters such as plant
Schadt, & Kalko, 2014), within-habitat hetero- source water isotopic composition and climate
geneity in resource availability (Smiley, that are either simulated within general circu-
Cotton, Badgley, & Cerling, 2016), and spatial lation models, or derived from surface water
variation in human impacts (Hall, Hale, Baker, or precipitation isoscapes. In addition, the
Bowling, & Ehleringer, 2015). Ongoing work resulting plant isoscapes (Fig. 3.7) may be then
promises to refine our understanding of eco- weighted using various approaches, including
system controls on foliar δ15N values and how maps of plant biome distributions or produc-
this variation is incorporated into animal tivity, or simulations of the same (West et al.,
tissues. 2008). Depending on the degree of understand-
ing or available data, the plant δ2H and δ18O
isoscapes may be general, such as a global
average leaf water isoscape, or quite specific,
3.5.4 Vegetation H and O Isoscapes such as a series of isoscapes depicting the
Since hydrogen and oxygen are found in changing spatial variation of leaf water for a
plant organic compounds and water, both single growing season. The degree of detail
plant water and organic isoscapes have been may be dictated by the availability of data or
produced for application to understanding bio- model understanding, or it may be dictated by
sphereatmosphere interactions, and others the specificity or generality of the question
such as to commerce or forensics (Cerling being asked. As with all isoscapes discussed,
et al., 2016; Ciais et al., 1997; Cuntz, Ciais, the approach is flexible and not a priori linked
Hoffmann, & Knorr, 2003; Farquhar et al., to any particular temporal or spatial scale.
1993; Kahmen et al., 2013). Although of course
it is recognized that spatial variation in leaf
water likely influences spatial patterns of ani-
mal H and O isotope ratios (Jeffrey, Denys,
3.5.5 Ecosystem Sr Isoscapes
Stoetzel, & Lee-Thorp, 2015), these plant iso- Large-scale Sr isoscapes have now been
scapes have not yet been explicitly applied to developed for bedrock and certain other sub-
improving understanding of migration, relying strates (e.g., river water and bioavailable Sr)
instead on the generally strong relationships across parts of North America and The
between animal isotope ratios (primarily H) Caribbean (Bataille & Bowen, 2012; Bataille
and drinking water isotope ratios. It seems et al., 2012, 2014; Beard & Johnson, 2000)

–180º –150º –120º –90º –60º –30º 0º 30º 60º 90º 120º 150º 180º
60º 60º
30º 30º
0º 0º
–30º –30º
–60º –60º
–180º –150º –120º –90º –60º –30º 0º 30º 60º 90º 120º 150º 180º
–17.8 –12.9 –10.5 –8.3 –6.1 –3.7 –1.3 1.1 3.3 5.5 7.7 10.1 12.5 14.9 17.1 19.3 21.3 23.6
Leaf water δ18O (‰)
FIGURE 3.7 Global annual average leaf water δ18O. This plant oxygen isoscape was produced by executing a physio-
logical model of leaf water enrichment (modified from Flanagan, L. B., Comstock, J. P., & Ehleringer, J. R. (1991).
Comparison of modeled and observed environmental influences on the stable oxygen and hydrogen isotope composition
of leaf water in Phaseolus vulgaris L. Plant Physiology, 96(2), 588596) using gridded annual average precipitation δ18O
(see Fig. 3.3) and monthly climate parameters (continuous fields of temperature and relative humidity for the climate nor-
mal period 196190 from the Climate Research Unit; New et al., 2002) to drive the model (West et al., 2008). Monthly
grids were averaged and the model output was further masked (as in Fig. 3.5) using continuous vegetation fields (DeFries
et al., 1999) eliminating areas with greater than 80 % nonvegetated ground.
(Fig. 3.8). As discussed earlier, local bedrock Sr origin of migratory organisms and identifying
isotope ratios will in many cases be only areas of high potential for Sr-based migration
loosely related to ecosystem 87Sr/86Sr relevant research. For example, areas of high 87Sr/86Sr
to migration research applications (Naiman, associated with very old (Precambrian) bed-
Quade, & Patchett, 2000). This result is borne rock in northern Minnesota and mountainous
out in several of these studies, which compare areas of the Rocky and Appalachian
modeled values with those measured from dif- Mountains likely represent areas where Sr iso-
ferent biological substrates and generally show tope ratios offer power to constrain the loca-
robust but noisy relationships which are tion and habitat of migrants. Relatively low
87
improved by incorporating processes beyond Sr/86Sr mapped in areas of recent low-silica
geological 87Sr production in the models volcanism in the northwestern United States
(Bataille & Bowen, 2012; Bataille et al., 2012). also offer potential for strong discrimination of
At their current level of development, such patterns of animal movement, and similar con-
isoscape data products are at minimum useful trasts have proven fruitful in discriminating
for guiding preliminary interpretations of the regions of production for Pacific salmon

FIGURE 3.8 Predicted 87Sr/86Sr for bedrock of the contiguous United States (modified from Bataille, C. P., & Bowen,
G. J. (2012). Mapping 87Sr/86Sr variations in bedrock and water for large scale provenance studies. Chemical Geology,
304305, 3952. doi:10.1016/j.chemgeo.2012.01.028). Values were modeled using bedrock-specific estimated initial
87
Sr/86Sr and 87Rb and Sr content and calculating 87Sr production following rock formation using the known decay rate of
87
Rb and rock ages represented. Geological data were obtained from United States Geological Survey state-level digital
geological maps.
(Brennan & Schindler, 2017). In regions where discussed here (H, C, N, O, and Sr), as well as
atmospheric deposition has been identified as the sulfur isotope system and some heavy ele-
an important mechanism of Sr addition to the ment isotope systems, is characterized by
landscape, Sr isoscapes can be improved by some level of known spatial variability and
incorporating information on aerosol 87Sr/86Sr thus presents opportunities for application to
in the modeling, rather than referring to migration research. The greatest potential for
bedrock values in isolation (Bataille et al., precise tracking of migratory connectivity
2012). using isotopes may exist where combinations
of discrete and continuous variation occur at
nested spatial scales. Isoscapes are produced
3.6 SUMMARY AND LOOK using models and data of varying complexity
FORWARD and specificity. All of these models and data
are imperfect and accurate estimates of predic-
Isoscape modeling represents an attempt to tion uncertainty and continued improvements
reproduce the natural fingerprint of in data and models are needed to advance the
stable isotopes on the landscape by synthesiz- scope and quality of isoscapes data products
ing geospatial data and models representing available to migration researchers. Moreover,
the processes underlying spatial isotopic varia- data products having greater specificity with
tion. The organization of these processes in the respect to substrate (e.g., lake vs river vs pre-
natural and anthropogenically modified envi- cipitation water isoscapes; leaf water isoscapes
ronment produces patterns of spatial variation for different plant types) and spatiotemporal
that differ among isotope systems and sub- domain are needed to reduce the reliance of
strates. Each of the stable isotope systems migration researchers on generalized products

3.6 SUMMARY AND LOOK FORWARD 77
that may be of limited (and poorly known) rel- important role that structured networks can
evance to their system of study. play in serving this need, we see a parallel
In the first edition of this book, we need and opportunity at this time to facilitate
highlighted two areas of emphasis that we the aggregation and distribution of data gener-
considered key to advancing the development ated by individual investigators and projects.
and application of isoscapes for migration Two databases, one hosted by the IAEA
research. Since that time, significant progress (https://nucleus.iaea.org/wiser) and a second
has been made in each area, and an opportu- managed by the group developing the IsoMAP
nity exists to revisit and update our guidance. platform (http://waterisotopes.org), exist as
First, we called for renewed and expanded platforms for the archival and redistribution of
efforts to collect spatiotemporally distributed investigator-generated water isotope data, but
isotope data, particularly through coordinated, at this point no community-wide initiative
multiinvestigator networks, that would exists to develop resources and standards for
increase the volume of data supporting iso- isotope data, be it for water or more broadly.
scape predictions. Some of the opportunities Several initiatives have recently emerged in
we envisioned in this area have come to fru- this area (e.g., Pauli et al., 2017; http://isome-
ition: in the United States, the National mo.com/index.html). A huge number of useful
Ecological Observatory Network has come isotope data exist in the literature and are
online and is providing standardized isotopic being generated each day that would be of
sampling nation-wide, and programs such as great value in isoscape development.
the IAEA’s water networks have expanded Development of a robust, community-driven
with new emphasis on river water isotope information management framework for
ratios and new growth in precipitation moni- aggregation and dissemination of these data
toring. These developments have been accom- would be of significant value but will require
panied, however, by other changes that we sustained and coordinated effort for many
had not envisioned. In particular, the past 10 years to come.
years have witnessed a significant expansion Second, in first edition of this book we
of investigator-driven collection of spatiotem- called for new, cyberinfrastructure-enabled
porally distributed isotope data. This trend tools for the development and distribution of
appears to stem from multiple factors: recogni- customized, fit-for-purpose isoscapes. This
tion of the value of isoscapes as a data analysis charge stemmed from the idea that, as migra-
framework has spurred new projects focused tion research applications expanded and
on gathering or aggregating data over large became more specific, there would be a greater
areas (Cotton, Cerling, Hoppe, Mosier, & Still, need for application-specific isoscapes, such as
2016; Hobson et al., 2012; Jasechko et al., 2013), those representing specific regions or periods
new analytical instrumentation has reduced of time. Since that time, the IsoMAP cyber-GIS
cost and technology barriers for some isotope gateway (http://isomap.org) has been devel-
analyses (e.g., water H and O) (Lis, Wassenaar, oped, largely to serve this purpose, with an
& Hendry, 2007), and social media has emphasis on precipitation water isotope ratios.
emerged as a platform for soliciting and coor- As already noted, IsoMAP allows development
dinating networks of volunteers to help of regional, time-specific isoscapes, and pro-
acquire samples (Good et al., 2014). vides these services using a browser-based
Data volume and availability continue to user interface backed by remote database and
remain a critical limitation to isoscape devel- computational resources and a data manage-
opment, and although we still recognize the ment system supporting robust and accessible

data analysis workflows (Bowen, West, et al., animals, and as a result improving our power
2012). The resulting data products may be of to use this tool to reconstruct and understand
greater relevance to specific migration applica- their movements in time and space.
tions than climatological versions of these pro-
ducts, and are now being adopted in some
migration research applications. Acknowledgment
However, work using such products has Parts of the research described in this chapter have
raised questions about the degree to which been supported by U.S. National Science Foundation
grants DBI-0743543 to GJB and JBW and EF-1241286 and
they actually provide improved representa-
DBI-1565128 to GJB. We also acknowledge support from
tions of the isotopic patterns existing within the Rosen Center for Advanced Computing (Purdue Univ.)
consumer tissues, and thus improve assign- and Center for High Performance Computing (Univ.
ments of migratory individuals (Vander of Utah).
Zanden et al., 2014, 2015). This has highlighted
a long-standing question in isotope-based
migration research: what is the “real” origin of
References
the spatial environmental signature assimi- Amundson, R., Austin, A. T., Schuur, E. A. G., Yoo, K.,
lated by migratory animals? For more than Matzek, V., Kendall, C., . . . Baisden, W. T. (2003).
Global patterns of the isotopic composition of soil and
two decades, now, our migration research has plant nitrogen. Global Biogeochemical Cycles, 17(1), 31.
referenced patterns of environmental isotope Austin, A. T., & Sala, O. (1999). Foliar d15N is negatively
variation in well-sampled substrates such as correlated with rainfall along the IGBP transect in
precipitation, and the success of these efforts Australia. Australian Journal of Plant Physiology, 26,
makes clear that these isoscapes represent key 293295.
Austin, A. T., & Vitousek, P. M. (1998). Nutrient dynamics
aspects of the spatial pattern fixed in consumer on a precipitation gradient in Hawai’i. Oecologia, 113(4),
tissues. As we seek greater specificity and 519529.
more robust assignments of migratory organ- Ball, J. T., Woodrow, I. E., & Berry, J. A. (1987). A model
isms, however, it is clear that we must con- for predicting stomatal conductance and its contribu-
tinue to improve our understanding of how tion to the control of photosynthesis under different
environmental conditions. In J. Biggins (Ed.), Progress
these animals sample isotope variability from in photosynthesis research (pp. 221224). Dordrecht:
the environment. Birds are not rain gauges. Martinus Nijhoff.
Isoscapes exist, or are being developed, for Barbour, M. M., Roden, J. S., Farquhar, G. D., & Ehleringer,
numerous other environmental substrates that J. R. (2004). Expressing leaf water and cellulose oxygen
may provide more direct representation of the isotope ratios as enrichment above source water reveals
evidence of a Peclet effect. Oecologia, 138(3), 426435.
resources available to study organisms. The Available from https://doi.org/10.1007/s00442-003-
opportunity exists now to advance the devel- 1449-3.
opment of these data products and integrate Bataille, C. P., & Bowen, G. J. (2012). Mapping
87
them into the analysis of spatial isotope pat- Sr/86Sr variations in bedrock and water for large scale
terns in consumer tissues, framing and provenance studies. Chemical Geology, 304305, 3952.
Available from https://doi.org/10.1016/j.chemgeo.
answering refined questions about how migra- 2012.01.028.
tory animals integrate isotope signatures from Bataille, C. P., Brennan, S. R., Hartmann, J., Moosdorf, N.,
their local environment and how to best repre- Wooller, M. J., & Bowen, G. J. (2014). A geostatistical
sent the underlying isoscape patterns. The framework for predicting variations in strontium con-
resulting process of isoscape selection and centrations and isotope ratios in Alaskan rivers.
Chemical Geology, 389(0), 115. Available from https://
optimization offers significant promise for doi.org/10.1016/j.chemgeo.2014.08.030.
improving the representation of spatial isotope Bataille, C. P., Laffoon, J., & Bowen, G. J. (2012). Mapping
patterns of real relevance to migratory multiple source effects on the strontium isotopic

REFERENCES 79
signatures of ecosystems: A case study from the Bowen, G. J., West, J. B., Zhao, L., Takahashi, G., Miller,
circum-Caribbean region. Ecosphere, 3(12), 118. C. C., & Zhang, T. (2012). Cyberinfrastructure for iso-
Available from https://doi.org/10.1890/ES12-00155.1. tope analysis and modeling. Eos, 93, 185187.
Beard, B. L., & Johnson, C. M. (2000). Strontium isotope Bowen, G. J., Kennedy, C. D., Henne, P. D., & Zhang, T.
composition of skeletal material can determine the birth (2012). Footprint of recycled water subsidies downwind
place and geographic mobility of humans and animals. of Lake Michigan. Ecosphere, 3(6), 53. Available from
Journal of Forensic Science, 45(5), 10491061. https://doi.org/10.1890/ES12-00062.1.
Bearhop, S., Fiedler, W., Furness, R. W., Votier, S. C., Bowling, D. R., McDowell, N. G., Bond, B. J., Law, B. E., &
Waldron, S., Newton, J., . . . Farnsworth, K. (2005). Ehleringer, J. R. (2002). 13C content of ecosystem respi-
Assortative mating as a mechanism for rapid evolution ration is linked to precipitation and vapor pressure def-
of a migratory divide. Science, 310, 502504. Available icit. Oecologia, 131, 113124.
from https://doi.org/10.1126/science.1115661. Brennan, S. R., & Schindler, D. E. (2017). Linking otolith
Birks, S. J., Gibson, J. J., Gourcy, L., Aggarwal, P. K., & microchemistry and dendritic isoscapes to map hetero-
Edwards, T. W. D. (2002). Maps and animations offer geneous production of fish across river basins.
new opportunities for studying the global water cycle. Ecological Applications, 27(2), 363377. Available from
Eos, Transactions, American Geophysical Union, Electronic https://doi.org/10.1002/eap.1474.
Supplement, 83. Available from http://www.agu.org/ Buzon, M. R., Conlee, C. A., Simonetti, A., & Bowen, G. J.
eos_elec/020082e.html. (2012). The consequences of Wari contact in the Nasca
Blum, J. D., Taliaferro, E. H., & Holmes, R. T. (2001). region during the Middle Horizon: Archaeological,
Determining the sources of calcium for migratory song- skeletal, and isotopic evidence. Journal of Archaeological
birds using stable strontium isotopes. Oecologia, 126(4), Science, 39, 26272636. Available from https://doi.org/
569574. 10.1016/j.jas.2012.04.003.
Bowen, G. J. (2008). Spatial analysis of the intra-annual var- Capo, R. C., Stewart, B. W., & Chadwick, O. A. (1998).
iation of precipitation isotope ratios and its climatologi- Strontium isotopes as tracers of ecosystem processes;
cal corollaries. Journal of Geophysical Research, 113, theory and methods. Geoderma, 82(13), 197225.
D05113. Available from https://doi.org/10.1029/ Cerling, T. E., Harris, J. M., Ambrose, S. H., Leakey, M. G.,
2007JD009295. & Solounias, N. (1997). Dietary and environmental
Bowen, G. J. (2010). Isoscapes: Spatial pattern in isotopic reconstruction with stable isotope analyses of herbivore
biogeochemistry. Annual Review of Earth & Planetary tooth enamel from the Miocene locality of Fort Ternan,
Sciences, 38, 161187. Available from https://doi.org/ Kenya. Journal of Human Evolution, 33, 635650.
10.1146/annurev-earth-040809-152429. Cerling, T. E., Harris, J. M., & Passey, B. H. (2003). Diets of
Bowen, G. J., & Revenaugh, J. (2003). Interpolating the iso- East African Bovidae based on stable isotope analysis.
topic composition of modern meteoric precipitation. Journal of Mammology, 84(2), 456470.
Water Resources Research, 39, 1299. Available from Cerling, T. E., Barnette, J. E., Bowen, G. J., Chesson, L. A.,
https://doi.org/10.1029/2003WR002086. Ehleringer, J. R., Remien, C. H., . . . West, J. B. (2016).
Bowen, G. J., & Wilkinson, B. (2002). Spatial distribution of Forensic stable isotope biogeochemistry. Annual Review
δ18O in meteoric precipitation. Geology, 30(4), 315318. of Earth and Planetary Sciences, 44, 175206.
Bowen, G. J., Wassenaar, L. I., & Hobson, K. A. (2005). Cernusak, L. A., Tcherkez, G., Keitel, C., Cornwell, W. K.,
Global application of stable hydrogen and oxygen iso- Santiago, L. S., Knohl, A., . . . Wright, I. J. (2009).
topes to wildlife forensics. Oecologia, 143(3), 337348. Viewpoint: Why are non-photosynthetic tissues gener-
Available from https://doi.org/10.1007/s00442-004- ally C-13 enriched compared with leaves in C-3 plants?
1813-y. Review and synthesis of current hypotheses. Functional
Bowen, G. J., Ehleringer, J. R., Chesson, L. A., Stange, E., & Plant Biology, 36(3), 199213. Available from https://
Cerling, T. E. (2007). Stable isotope ratios of tap water doi.org/10.1071/FP08216.
in the contiguous USA. Water Resources Research, 43, Cernusak, L. A., Barbour, M. M., Arndt, S. K., Cheesman,
W03419. Available from https://doi.org/10.1029/ A. W., English, N. B., Feild, T. S., . . . Farquhar, G. D.
2006wr005186. (2016). Stable isotopes in leaf water of terrestrial plants.
Bowen, G. J., Kennedy, C. D., Liu, Z., & Stalker, J. (2011). Plant Cell and Environment, 39(5), 10871102. Available
Water balance model for mean annual hydrogen and from https://doi.org/10.1111/pce.12703.
oxygen isotope distributions in surface waters of the Chamberlain, C. P., Blum, J. D., Holmes, R. T., Feng, X. H.,
contiguous USA. Journal of Geophysical Research, 116, Sherry, T. W., & Graves, G. R. (1997). The use of isotope
G04011. Available from https://doi.org/10.1029/ tracers for identifying populations of migratory birds.
2010JG001581. Oecologia, 109(1), 132141.

Chamberlain, C. P., Bensch, S., Feng, X., Akesson, S., & Dongmann, G., Nurnberg, H. W., Forstel, H., & Wagener,
Andersson, T. (2000). Stable isotopes examined across a K. (1974). On the enrichment of H218O in the leaves of
migratory divide in Scandinavian willow warblers transpiring plants. Radiation and Environmental
(Phylloscopus trochilus trochilus and Phylloscopus trochilus Biophysics, 11, 4152.
acredula) reflect their African winter quarters. Ehleringer, J. R., Rundel, P. W., & Nagy, K. A. (1986).
Proceedings of the Royal Society of London Series B- Stable isotopes in physiological ecology and food web
Biological Sciences, 267(1438), 4348. research. Trends in Ecology & Evolution, 1, 4245.
Ciais, P., Denning, A. S., Tans, P. P., Berry, J. A., Randall, Ehleringer, J. R., Cerling, T. E., & Helliker, B. R. (1997). C-4
D. A., Collatz, J. J. G., & Heimann, M. (1997). A three photosynthesis, atmospheric CO2 and climate.
dimensional synthesis study of 18O in atmospheric CO2 Oecologia, 112(3), 285299.
Part 1: Surface fluxes. Journal of Geophysical Research, Evans, R. D. (2001). Physiological mechanisms influencing
102, 58575872. plant nitrogen isotope composition. Trends in Plant
Collatz, G. J., Berry, J. A., & Clark, J. S. (1998). Effects of cli- Science, 6(3), 121126.
mate and atmospheric CO2 partial pressure on the Farquhar, G. D., & Cernusak, L. A. (2005). On the isotopic
global distribution of C-4 grasses: Present, past, and composition of leaf water in the non-steady state.
future. Oecologia, 114(4), 441454. Functional Plant Biology, 32(4), 293303.
Cotton, J. M., Cerling, T. E., Hoppe, K. A., Mosier, T. M., & Farquhar, G. D., Ehleringer, J. R., & Hubick, K. T. (1989).
Still, C. J. (2016). Climate, CO2, and the history of North Carbon isotope discrimination and photosynthesis.
American grasses since the Last Glacial Maximum. Annual Review of Plant Physiology and Plant Molecular
Science Advances, 2(3), e1501346. Biology, 40, 503537. Available from https://doi.org/
Craig, H. (1953). The geochemistry of the stable carbon iso- 10.1146/annurev.pp.40.060189.002443.
topes. Geochimica et Cosmochimica Acta, 3, 5392. Farquhar, G. D., Lloyd, J., Taylor, J. A., Flanagan, L. B.,
Craig, H. (1961). Isotopic variations in meteoric waters. Syvertsen, J. P., Hubick, K. T., . . . Ehleringer, J. R. (1993).
Science, 133, 17021703. Vegetation effects on the isotope composition of oxygen
Craig, H., & Gordon, L. I. (1965). Deuterium and oxygen- in atmospheric CO2. Nature, 363(6428), 439443.
18 variations in the ocean and the marine atmosphere. Faure, G., & Powell, J. L. (1972). Strontium isotope geology.
In E. Tongiorgi (Ed.), Proceedings of a conference on New York: Springer-Verlag.
stable isotopes in oceanographic studies and paleotempera- Fekete, B. M., Gibson, J. J., Aggarwal, P., & Vorosmarty,
tures (Vol. 9-130). Spoleto, Italy. C. J. (2006). Application of isotope tracers in continental
Cuntz, M., Ciais, P., Hoffmann, G., & Knorr, W. (2003). A scale hydrological modeling. Journal of Hydrology, 330,
comprehensive global three-dimensional model of δ18O 444456. Available from https://doi.org/10.1016/j.
in atmospheric CO2: 1. Validation of surface processes. jhydrol.2006.04.029.
Journal of Geophysical Research, 108(D17), 4527. Flanagan, L. B., Comstock, J. P., & Ehleringer, J. R. (1991).
Daly, C., Neilson, R. P., & Phillips, D. L. (1994). A Comparison of modeled and observed environmental
statistical-topographic model for mapping climatologi- influences on the stable oxygen and hydrogen isotope
cal precipitation over mountainous terrain. Journal of composition of leaf water in Phaseolus vulgaris L. Plant
Applied Meteorology, 33(2), 140158. Physiology, 96(2), 588596.
Dansgaard, W. (1954). The O18-abundance in fresh water. Fox, A. D., Hobson, K. A., de Jong, A., Kardynal, K. J.,
Geochimica et Cosmochimica Acta, 6, 241260. Koehler, G., & Heinicke, T. (2017). Flyway population
Dansgaard, W. (1964). Stable isotopes in precipitation. delineation in Taiga Bean Geese Anser fabalis fabalis revealed
Tellus, 16, 436468. by multi-element feather stable isotope analysis. Ibis, 159(1),
DeFries, R. S., Townshend, J. R. G., & Hansen, M. C. 6675. Available from https://doi.org/10.1111/ibi.12417.
(1999). Continuous fields of vegetation characteristics at Fox-Dobbs, K., Doak, D. F., Brody, A. K., & Palmer, T. M.
the global scale at 1km resolution. Journal of Geophysical (2010). Termites create spatial structure and govern eco-
Research, 104(16), 1116. system function by affecting N2 fixation in an East
Diefendorf, A. F., Mueller, K. E., Wing, S. L., Koch, P. L., & African savanna. Ecology, 91, 12961307. Available
Freeman, K. H. (2010). Global patterns in leaf C-13 dis- from https://doi.org/10.1890/09-0653.1.
crimination and implications for studies of past and Gamarra, B., Sachse, D., & Kahmen, A. (2016). Effects of leaf
future climate. Proceedings of the National Academy of water evaporative 2 H-enrichment and biosynthetic frac-
Sciences of the United States of America, 107(13), tionation on leaf wax n-alkane delta2 H values in C3 and
57385743. Available from https://doi.org/10.1073/ C4 grasses. Plant Cell and Environment, 39(11), 23902403.
pnas.0910513107. Available from https://doi.org/10.1111/pce.12789.

REFERENCES 81
Good, S. P., Mallia, D. V., Lin, J. C., & Bowen, G. J. (2014). Hobson, K. A., Van Wilgenburg, S. L., Wassenaar, L. I., &
Stable isotope analysis of precipitation samples Larson, K. (2012). Linking Hydrogen (δ2H) Isotopes in
obtained via crowdsourcing reveals the spatiotemporal feathers and precipitation: Sources of variance and con-
evolution of superstorm sandy. PLoS One, 9(3), e91117. sequences for assignment to isoscapes. PLoS One, 7(4),
Available from https://doi.org/10.1371/journal. e35137. Available from https://doi.org/10.1371/jour-
pone.0091117. nal.pone.0035137.
Hall, S. J., Hale, R. L., Baker, M. A., Bowling, D. R., & Hobson, K. A., Doward, K., Kardynal, K. J., & McNeil, J. N.
Ehleringer, J. R. (2015). Riparian plant isotopes reflect (2018). Inferring origins of migrating insects using iso-
anthropogenic nitrogen perturbations: Robust patterns scapes: A case study using the true armyworm, Mythimna
across land use gradients. Ecosphere, 6(10). Available unipuncta, in North America. Ecological Entomology.
from https://doi.org/10.1890/ES15-00319.1. Available from https://doi.org/10.1111/een.12505.
Handley, L. L., Austin, A. T., Robinson, D., Scrimgeour, Hoppe, K. A., Koch, P. L., Carlson, R. W., & Webb, S. D.
C. M., Raven, J. A., Heaton, T. H. E., . . . Stewart, G. R. (1999). Tracking mammoths and mastodons:
(1999). The N-15 natural abundance (delta N-15) of eco- Reconstruction of migratory behavior using strontium
system samples reflects measures of water availability. isotope ratios. Geology, 27(5), 439442.
Australian Journal of Plant Physiology, 26(2), 185199. Houlton, B. Z., Sigman, D. M., & Hedin, L. O. (2006).
Helliker, B. R., & Ehleringer, J. R. (2000). Establishing a Isotopic evidence for large gaseous nitrogen losses
grassland signature in veins: O-18 in the leaf water of from tropical rainforests. Proceedings of the National
C-3 and C-4 grasses. Proceedings of the National Academy Academy of Sciences of the United States of America, 103
of Sciences of the United States of America, 97(14), (23), 87458750.
78947898. Ingraham, N. L., & Taylor, B. E. (1991). Light stable isotope
Hemming, D., Yakir, D., Ambus, P., Aurela, M., Besson, systematics of large-scale hydrologic regimes in California
C. K., Black, K., . . . Vesala, T. (2005). Pan-European and Nevada. Water Resources Research, 27(1), 7790.
d13C values of air and organic matter from forest eco- Isaaks, E. H., & Srivastava, R. M. (1990). An introduction to
systems. Global Change Biology, 11, 10651093. applied geostatistics. New York: Oxford University Press.
Hobson, K. A., & Kardynal, K. J. (2016). An isotope (delta Jacobson, A. D., & Blum, J. D. (2000). Ca/Sr and 87Sr/86Sr
S-34) filter and geolocator results constrain a dual geochemistry of disseminated calcite in Himalayan sili-
feather isoscape (delta H-2, delta C-13) to identify the cate rocks from Nanga Parbat: Influence on river-water
wintering grounds of North American Barn Swallows. chemistry. Geology, 28(5), 463466.
Auk, 133(1), 8698. Available from https://doi.org/ Jasechko, S., Sharp, Z. D., Gibson, J. J., Birks, S. J., Yi, Y., &
10.1642/AUK-15-149.1. Fawcett, P. J. (2013). Terrestrial water fluxes dominated
Hobson, K. A., & Wassenaar, L. I. (1997). Linking breeding by transpiration. Nature, 496, 347350.
and wintering grounds of neotropical migrant song- Jeffrey, A., Denys, C., Stoetzel, E., & Lee-Thorp, J. A.
birds using stable hydrogen isotopic analysis of feath- (2015). Influences on the stable oxygen and carbon iso-
ers. Oecologia, 109, 142148. topes in gerbillid rodent teeth in semi-arid and arid
Hobson, K. A., Wassenaar, L. I., & Taylor, O. R. (1999). environments: Implications for past climate and envi-
Stable isotopes (δD and δ13C) are geographic indicators ronmental reconstruction. Earth and Planetary Science
of natal origins of monarch butterflies in eastern North Letters, 428, 8496. Available from https://doi.org/
America. Oecologia, 120, 397404. 10.1016/j.epsl.2015.07.012.
Hobson, K. A., Wassenaar, L. I., Mila, B., Lovette, I., Kahmen, A., Hoffmann, B., Schefuss, E., Arndt, S. K.,
Dingle, C., & Smith, T. B. (2003). Stable isotopes as indi- Cernusak, L. A., West, J. B., & Sachse, D. (2013). Leaf
cators of altitudinal distributions and movements in an water deuterium enrichment shapes leaf wax n-alkane
Ecuadorean hummingbird community. Oecologia, 136 delta D values of angiosperm plants II: Observational
(2), 302308. Available from https://doi.org/10.1007/ evidence and global implications. Geochimica et
s00442-003-1271-y. Cosmochimica Acta, 111, 5063. Available from https://
Hobson, K. A., Bowen, G. J., Wassenaar, L. I., Ferrand, Y., doi.org/10.1016/j.gca.2012.09.004.
& Lormee, H. (2004). Using stable hydrogen and oxy- Kendall, C., & Coplen, T. B. (2001). Distribution of oxygen-
gen isotope measurements of feathers to infer geo- 18 and deuterium in river waters across the United
graphical origins of migrating European birds. States. Hydrological Processes, 15(7), 13631393.
Oecologia, 141, 477488. Available from https://doi. Kennedy, M. J., Chadwick, O. A., Vitousek, P. M., Derry,
org/10.1007/s00442-004-1671-7. L. A., & Hendricks, D. M. (1998). Changing sources of

base cations during ecosystem development, Hawaiian latitude, and feather color in a migratory songbird.
Islands. Geology, 26(11), 10151018. Science, 306, 22492250.
Landwehr, J. M., Coplen, T. B., & Stewart, D. W. (2014). O’Leary, M. H. (1981). Carbon isotope fractionation in
Spatial, seasonal, and source variability in the plants. Phytochemistry, 20, 553567.
stable oxygen and hydrogen isotopic composition of O’Leary, M. H., Madhavan, S., & Paneth, P. (1992). Physical
tap waters throughout the USA. Hydrological Processes, and chemical basis of carbon isotope fractionation in
28(21), 53825422. plants. Plant Cell and Environment, 15(9), 10991104.
Lehmann, M. M., Gamarra, B., Kahmen, A., Siegwolf, Ometto, J. P. H. B., Ehleringer, J. R., Domingues, T. F.,
R. T. W., & Saurer, M. (2017). Oxygen isotope fractiona- Berry, J. A., Ishida, F. Y., Mazzi, E., . . . Martinelli, L. A.
tions across individual leaf carbohydrates in grass and (2006). The stable carbon and nitrogen isotopic compo-
tree species. Plant Cell and Environment, 40(8), sition of vegetation in tropical forests of the Amazon
16581670. Available from https://doi.org/10.1111/ Basin, Brazil. Biogeochemistry, 79(1-2), 251274.
pce.12974. Osmond, C. B., Allaway, W. G., Sutton, B. G., Troughton,
Lis, G., Wassenaar, L. I., & Hendry, M. J. (2007). High- J. H., Queiroz, O., Luttge, U., & Winter, K. (1973).
precision laser spectroscopy D/H and 18O/16O mea- Carbon isotope discrimination in photosynthesis of
surements of microliter natural water samples. CAM plants. Nature, 246, 4142.
Analytical Chemistry. Available from https://doi.org/ Otte, I., Detsch, F., Gütlein, A., Scholl, M., Kiese, R.,
10.1021/ac701716q. Appelhans, T., & Nauss, T. (2017). Seasonality of
Lloyd, J., & Farquhar, G. D. (1994). 13C discrimination dur- stable isotope composition of atmospheric water input
ing CO2 assimilation by the terrestrial biosphere. at the southern slopes of Mt. Kilimanjaro, Tanzania.
Oecologia, 99(34), 201215. Hydrological Processes, 31(22), 39323947. Available
Longinelli, A., & Selmo, E. (2003). Isotopic composition of from https://doi.org/10.1002/hyp.11311.
precipitation in Italy: A first overall map. Journal of Pain, D., Green, R., Gieβing, B., Kozulin, A., Poluda, A.,
Hydrology, 270(12), 7588. Ottosson, U., . . . Hilton, G. (2004). Using stable isotopes
Luo, Y.-H., & Sternberg, L. (1992). Spatial D/H heterogene- to investigate migratory connectivity of the globally
ity of leaf water. Plant Physiology, 99(1), 348350. threatened aquatic warbler Acrocephalus paludicola.
MacFadden, B. J., & Cerling, T. E. (1994). Fossil horses, car- Oecologia, 138(2), 168174.
bon isotopes and global change. Trends in Ecology and Pardo, L. H., Templer, P. H., Goodale, C. L., Duke, S.,
Evolution, 9, 481485. Groffman, P. M., Adams, M. B., . . . Wessel, W. (2006).
Meehan, T. D., Giermakowski, J. T., & Cryan, P. M. (2004). Regional assessment of N saturation using foliar and
GIS-based model of stable hydrogen isotope ratios in root delta N-15. Biogeochemistry, 80(2), 143171.
North American growing-season precipitation for use Pataki, D. E., Bowling, D. R., & Ehleringer, J. R. (2003).
in animal movement studies. Isotopes in Environmental Seasonal cycle of carbon dioxide and its isotopic com-
and Health Studies, 40(4), 291300. position in an urban atmosphere: Anthropogenic and
Murphy, B. P., & Bowman, D. M. J. S. (2006). Kangaroo biogenic effects. Journal of Geophysical Research-
metabolism does not cause the relationship between Atmospheres, 108(D23).
bone collagen delta N-15 and water availability. Pauli, J. N., Newsome, S. D., Cook, J. A., Harrod, C.,
Functional Ecology, 20(6), 10621069. Steffan, S. A., Baker, C. J., . . . Cerling, T. E. (2017).
Murphy, B. P., Bowman, D. M. J. S., & Gagan, M. K. (2007). Opinion: Why we need a centralized repository for iso-
The interactive effect of temperature and humidity on topic data. Proceedings of the National Academy of
the oxygen isotope composition of kangaroos. Sciences, 114(12), 29973001.
Functional Ecology, 21, 757766. Available from https:// Peters, C. R., & Vogel, J. C. (2005). Africa’s wild C-4 plant
doi.org/10.1111/j.1365-2435.2007.01284.x. foods and possible early hominid diets. Journal of
Naiman, Z., Quade, J., & Patchett, P. J. (2000). Isotopic evi- Human Evolution, 48(3), 219236.
dence for eolian recycling of pedogenic carbonate and Peterson, T. C., & Vose, R. S. (1997). An overview of the
variations in carbonate dust sources throughout the Global Historical Climatology Network temperature
southwest United States. Geochimica et Cosmochimica data base. Bulletin of the American Meteorological Society,
Acta, 64(18), 30993109. 78, 28372849.
New, M., Lister, D., Hulme, M., & Makin, I. (2002). A high- Powell, R. L., Yoo, E. H., & Still, C. J. (2012). Vegetation
resolution data set of surface climate over global land and soil carbon-13 isoscapes for South America:
areas. Climate Research, 21(1), 125. Integrating remote sensing and ecosystem isotope mea-
Norris, D. R., Marra, P. P., Montgomerie, R., Kyser, T. K., surements. Ecosphere, 3(11). Available from https://doi.
& Ratcliffe, L. M. (2004). Reproductive effort, molting org/10.1890/Es12-00162.1.

REFERENCES 83
Quade, J., Chivas, A. R., & McCulloch, M. T. (1995). Global Biogeochemical Cycles, 19, GB1017. Available from
Strontium and carbon isotope tracers and the origins of https://doi.org/10.1029/2003GB002141.
soil carbonate in South Australia and Victoria. Terzer, S., Wassenaar, L., Araguás-Araguás, L., &
Palaeogeography Palaeoclimatology Palaeoecology, 113, Aggarwal, P. (2013). Global isoscapes for δ18O and δ2H
103117. in precipitation: Improved prediction using regionalized
Rascher, K. G., Hellmann, C., Máguas, C., & Werner, C. climatic regression models. Hydrology and Earth System
(2012). Community scale 15N isoscapes: Tracing the Sciences, 17(11), 47134728.
spatial impact of an exotic N2-fixing invader. Ecology Trusdell, F. A., Wolfe, E. W., & Morris, J. (2006). Digital
Letters, 15(5), 484491. database of the geological map of the Island of Hawai’i (144,
Robinson, D. (2001). d15N as an integrator of the nitrogen version 1.0). Retrieved from ,http://pubs.usgs.gov/
cycle. Trends in Ecology and Evolution, 16(3), 153162. ds/2005/144/..
Roden, J. S., Lin, G. G., & Ehleringer, J. R. (2000). A mecha- Van Der Merwe, N. J., Throp, J. A. L., & Bell, R. H. V.
nistic model for interpretation of hydrogen and oxygen (1988). Carbon isotopes as indicators of elephant diets
isotope ratios in tree-ring cellulose. Geochimica et and African environments. African Journal of Ecology, 26
Cosmochimica Acta, 64(1), 2135. (2), 163172.
Rozanski, K., Araguas-Araguas, L., & Gonfiantini, R. Vander Zanden, H. B., Wunder, M. B., Hobson, K. A., Van
(1993). Isotopic patterns in modern global precipitation. Wilgenburg, S. L., Wassenaar, L. I., Welker, J. M., &
In P. K. Swart, K. C. Lohmann, J. McKenzie, & S. Savin Bowen, G. J. (2014). Contrasting assignment of migra-
(Eds.), Climate change in continental isotopic records tory organisms to geographic origins using long-term
(pp. 136). Washington, DC: American Geophysical versus year-specific precipitation isotope maps. Methods
Union. in Ecology and Evolution, 5(9), 891900.
Sachse, D., Billault, I., Bowen, G. J., Chikaraishi, Y., Vander Zanden, H. B., Wunder, M. B., Hobson, K. A., Van
Dawson, T. E., Feakins, S. J., . . . Kahmen, A. (2012). Wilgenburg, S. L., Wassenaar, L. I., Welker, J. M., &
Molecular paleohydrology: Interpreting the hydrogen- Bowen, G. J. (2015). Space-time tradeoffs in the devel-
isotopic composition of lipid biomarkers from photo- opment of precipitation-based isotope models for deter-
synthesizing organisms. Annual Review of Earth and mining migratory origin. Journal of Avian Biology, 46,
Planetary Sciences, 40, 221249. 658667. Available from https://doi.org/10.1111/
Scholze, M., Kaplan, J. O., Knorr, W., & Heimann, M. jav.00656.
(2003). Climate and interannual variability of the Voigt, C. C., Helbig-Bonitz, M., Kramer-Schadt, S., &
atmosphere-biosphere 13 CO. Geophysical Research Kalko, E. K. V. (2014). The third dimension of bat
Letters, 30(2), 1097. migration: Evidence for elevational movements of
Seibt, U., Rajabi, A., Griffiths, H., & Berry, J. A. (2008). Miniopterus natalensis along the slopes of Mount
Carbon isotopes and water use efficiency: Sense and Kilimanjaro. Oecologia, 174(3), 751764. Available from
sensitivity. Oecologia, 155(3), 441454. Available from https://doi.org/10.1007/s00442-013-2819-0.
https://doi.org/10.1007/s00442-007-0932-7. Wang, L. X., Okin, G. S., Wang, J., Epstein, H., & Macko,
Smiley, T. M., Cotton, J. M., Badgley, C., & Cerling, T. E. S. A. (2007). Predicting leaf and canopy N-15 composi-
(2016). Small-mammal isotope ecology tracks climate tions from reflectance spectra. Geophysical Research
and vegetation gradients across western North Letters, 34(2).
America. Oikos, 125(8), 11001109. Available from Wang, Y., Hou, S., Masson-Delmotte, V., & Jouzel, J.
https://doi.org/10.1111/oik.02722. (2010). A generalized additive model for the spatial dis-
Sponheimer, M., & Lee-Thorp, J. A. (1999). Isotopic evi- tribution of stable isotopic composition in Antarctic sur-
dence for the diet of an early hominid, Australopithecus face snow. Chemical Geology, 271(3), 133141. Available
africanus. Science, 283(5400), 368370. from https://doi.org/10.1016/j.chemgeo.2010.01.004.
Sternberg, L., & Deniro, M. J. (1983). Isotopic composition Wassenaar, L. I., & Hobson, K. A. (2001). A stable-isotope
of cellulose from C3, C4, and CAM plants growing approach to delineate geographical catchment areas of
near one another. Science, 220(4600), 947949. Available avian migration monitoring stations in North America.
from https://doi.org/10.1126/science.220.4600.947. Environmental Science & Technology, 35(9), 18451850.
Still, C. J., Berry, J. A., Collatz, G. J., & DeFries, R. S. (2003). Werner, S. J., Hobson, K. A., Van Wilgenburg, S. L., &
Global distribution of C3 and C4 vegetation: Carbon Fischer, J. W. (2016). Multi-isotopic (delta H-2, delta C-
cycle implications. Global Biogeochemical Cycles, 17, 1006. 13, delta N-15) tracing of molt origin for Red-Winged
Suits, N. S., Denning, A. S., Berry, J. A., Still, C. J., Kaduk, Blackbirds associated with agro-ecosystems. PLoS One,
J., Miller, J. B., & Baker, I. T. (2005). Simulation of car- 11(11). Available from https://doi.org/10.1371/journal.
bon isotope discrimination of the terrestrial biosphere. pone.0165996.

West, J. B., HilleRisLambers, J., Lee, T. D., Hobbie, S. E., & Wunder, M. B., Kester, C. L., Knopf, F. L., & Rye, R. O.
Reich, P. B. (2005). Legume species identity and soil (2005). A test of geographic assignment using isotope
nitrogen supply determine symbiotic nitrogen-fixation tracers in feathers of known origin. Oecologia, 144,
responses to elevated atmospheric [CO2]. New 607617. Available from https://doi.org/10.1007/
Phytologist, 167(2), 523530. s00442-005-0071-y.
West, J. B., Sobek, A., & Ehleringer, J. R. (2008). A simpli- Yakir, D., DeNiro, M. J., & Gat, J. R. (1990). Natural deute-
fied GIS approach to modeling global leaf water iso- rium and oxygen-18 enrichment in leaf water of cotton
scapes. PLoS One, 3(6), e2447. Available from https:// plants grown under wet and dry conditions: Evidence
doi.org/10.1371/journal.pone.0002447. for water compartmentation and its dynamics. Plant
Wickman, F. E. (1952). Variations in the relative abundance Cell and Environment, 13(1), 4956.
of the carbon isotopes in plants. Geochimica et Yakir, D., Berry, J. A., Giles, L., & Osmond, C. B. (1994).
Cosmochimica Acta, 2, 243254. Isotopic heterogeneity of water in transpiring leaves:
Widory, D. (2007). Nitrogen isotopes: Tracers of Identification of the component that controls the 18O of
origin and processes affecting PM10 in the atmo- atmospheric O2 and CO2. Plant, Cell and Environment,
sphere of Paris. Atmospheric Environment, 41(11), 17, 7380.
23822390. Yurtsever, Y., & Gat, J. R. (1981). Atmospheric waters. In J. R.
Wu, C. F. J. (1986). Jackknife, bootstrap and other resam- Gat, & R. Gonfiantini (Eds.), Stable isotope hydrology:
pling methods in regression analysis. The Annals of Deuterium and oxygen-18 in the water cycle (pp. 103142).
Statistics, 14(4), 12611295. Vienna: International Atomic Energy Agency.

C H A P T E R
4
Application of Isotopic Methods to
Tracking Animal Movements
Keith A. Hobson
University of Western Ontario, London, ON, Canada
4.1 INTRODUCTION 20 years did investigations start to focus on

using stable isotopes to trace animal origins
The application of stable isotope methods in and movements, and this has now evolved
ecology has undergone a fascinating evolution into a diverse field of investigation with
over past decades. Early applications involving forensic as well as ecological implications.
carbon and nitrogen isotope analyses focused Stable isotope methods are now considered
on plant physiology and photosynthetic mainstream in ecology, and are well estab-
pathways and the tracing of nitrogenous com- lished in the diverse toolbox available to study
pounds in terrestrial and aquatic systems. It animal movement.
was not until the early 1990s that isotopic lin- Fundamentally, the application of stable
kages were made between primary productiv- isotopes as a tracer of animal origin relies on
ity and animal tissues through several trophic three basic isotopic principles:
levels. A key development was the observation
1. Tissues in consumer (all animals including
that stable isotope measurements of animal tis-
humans) stable isotope (CHNOS) values
sues provide information on source of nutri-
reflect the dietary and drinking water they
ents and relative trophic position, and this led
are in “equilibrium” with. If diet and water
to vast improvements in the way animal ecol-
used by a migratory organism differs
ogy, physiology, and nutrition is studied.
isotopically and spatially, then isotope
From the simplistic concept of “you are what
values in the consumer may potentially
you eat plus a few parts per mil” refinements
provide information on previous locations
were made involving sophisticated isotope
or biomes.
mixing models, insights into how nutritional
2. The period over which this spatial isotopic
quality affects trophic isotope discrimination
information is retained depends on the
and the evolution of compound-specific versus
tissue type. For metabolically active tissues,
bulk tissue isotope methods. Only in the past

86 4. APPLICATION OF ISOTOPIC METHODS TO TRACKING ANIMAL MOVEMENTS
this represents a moving window of dietary that diet-tissue isotope discrimination factors
information. For metabolically inactive are influenced by quality of the diet, and so
tissues, spatial information will be “locked are likely not to be a static variable for wild
in” indefinitely, but reflects the short period animal populations. Because our ability to
of growth of that tissue. place an organism onto an isoscape map is
3. Mechanisms related to dietary transfer of sensitive to the true isotope discrimination fac-
isotopic signals to consumer tissues tors, researchers should evaluate their esti-
including isotopic discrimination, exercise, mates based on an honest assessment of how
and metabolic routing are known and well they know such factors. Obtaining this
accounted for. information requires dietary experiments with
the organism of interest, or, at least a sensitiv-
In practice, it is unlikely that all three princi-
ity analysis to determine the effect of varying
ples are satisfied or fully known with sufficient
discrimination values on the outcome of GIS
confidence. However, depending on the organ-
models or other methods used to “place” an
ism, much of the uncertainty can be con-
organism onto a map.
strained and, as we shall see, inferences can
still be made with respect to previous prove-
nance of individuals based on isotopic mea-
surements of their tissues. A careful blend of 4.2.1 Tissue and Isotopic Turnover: The
knowledge of the life history of the organism, Moving Window
knowledge of likely dietary isotopic landscapes It is well known that stable isotope values
or “isoscapes” experienced by that organism, in consumer tissues reflect an integration of
and the physiological parameters that can feeding events over various time periods.
influence isotopic inferences makes up the art Tieszen, Boutton, Tesdahl, and Slade (1983)
of using stable isotopes to accurately track were the first to conduct “diet switch” experi-
migratory organisms. The devil, of course, lies ments, whereby captive animals were allowed
in the details and assumptions made. to achieve equilibrium under one dietary
regime and then being switched to an isotopi-
cally different diet (Fig. 4.1). Tissue isotopes
4.2 TOWARD ISOTOPIC tracked the diet switch and uptake of the new
ASSIGNMENT OF ORIGINS isotopic dietary signal. This approach has been
applied successfully for many species, and
The “you are what you eat plus a few parts most find an exponential uptake curve to
per mil” maxim is formalized in the equation describe the pattern of isotopic dietary change
in tissues.
δCt 5 δd 1 Δdt
DðtÞ 5 a 1 b expð 2ctÞ
where δCt is the measured stable isotope value
of a tissue in the consumer, δd is the where D(t) is the stable isotope value of the tis-
stable isotope value of the diet, and Δdt is the sue at time t, a is the asymptotic tissue value, b
diet-tissue isotope discrimination factor. We is the absolute change in tissue isotope value
know that the isotope discrimination factor is between initial and asymptotic conditions, and
an oversimplification, and it does not necessar- c is a rate constant defining tissue turnover.
ily take into account metabolic routing of spe- When researchers wish to consider effects of
cific macronutrients such as proteins, lipids, growth (k) as well as metabolic turnover (m),
and carbohydrates. Research has also shown the overall rate constant c can be expressed as

4.2 TOWARD ISOTOPIC ASSIGNMENT OF ORIGINS 87
–18 FIGURE 4.1 Conceptual depiction of the way in
which different tissues will respond to an isotopic diet
–20 switch. We expect close coupling between the diet iso-
δ C of diet
tope trajectory and fast turnover tissues like liver and
–22 blood plasma. Much slower response is expected for
slow turnover tissues like bone collagen.
13
–24
–18 –26
δ13C of tissues
–20 Muscle
–22
Collagen
–24 Liver
–26
0 50 100 150 200 250
Time in days
(k 1 m). This approach works to provide esti- wind tunnels is clearly one effective way to
mates of elemental (C,N) turnover rates in explore turnover rates in migratory birds and
various tissues of birds, fish, and mammals insects. These results are encouraging and sug-
(e.g., Bosley, Witting, Chambers, & Wainright, gest published tissue isotopic turnover rates
2002; Dalerum & Angerbjörn, 2005; Hesslein, may be appropriate for isotopic studies on
Hallard, & Ramlal, 1993). One disadvantage of migratory organisms. The other good news is
these studies is they were based on sedentary, that elemental turnover rates across animals
nonexercised individuals in controlled labora- appear to follow expectations based on body
tory settings. Can the elemental turnover rates size (Thomas & Crowther, 2015). It is possible,
obtained from such sedentary experimental then, to estimate turnover rates for various tis-
studies be applied to migrating individuals sues based on the body mass of the organism
where, in the example of birds, undergo hours of interest even though that species has not
of sustained flight? Might we not expect more been tested experimentally (Martı́nez del Rio
rapid elemental turnover in tissues of exercis- & Carleton, 2012).
ing versus sedentary organisms? This is still Others have investigated a different
not fully clear, but Hobson and Yohannes approach to analyzing and interpreting nutri-
(2007) used Rosy Starlings (Sturnus roseus) fly- ent uptake curves based on isotopic dietary
ing in a wind tunnel to provide a first approxi- switch experiments (Cerling, Bowen,
mation of this effect for the cellular fraction of Ehleringer, & Sponheimer, 2007). Instead of fit-
blood. They performed a C3 to C4 diet switch ting exponential equations to estimate turn-
on birds that flew for several hours per day, over rates, they used a decay-lapse function
and then contrasted the isotopic turnover rates technique to estimate relative contributions.
with those of an unexercised control group. This approach involves the determination of
Interestingly, no difference in isotope turnover the so-called “reaction progress variables” that
was found for carbon isotopes between the are derived from a process that involves line-
two groups. This suggests that blood produc- arizing the decay curves. Interestingly, this
tion was unaffected isotopically by exercise, at approach suggests curves represent more than
least to the level measurable in this experi- one source pool, with each having a different
ment. More studies are needed and the use of elemental turnover rate. Perhaps the best

interpretation is that essential amino acids are migratory organism, we need to decide on a
transferred quickly from diet to tissues, convention that best quantifies the time period
whereas nonessential amino acids are manu- represented by the isotopic measurement.
factured from dietary components and so rep- Most authors have considered that a tissue
resent a lag time prior to incorporation into realistically represents about 3 half-lives, or
consumer tissues (but see Chapter 7: Amino the time required for 87.5% of the original sig-
Acid Isotope Analysis: A New Frontier in nal to be replaced by a new signal. Put another
Studies of Animal Migration and Foraging way, we should at least be able to detect 12.5%
Ecology). This way of looking at elemental of the original material remaining by our isoto-
turnover in animals shows potential for under- pic measurement. While this is a rule-of-
standing how elements from diet and body thumb, the ability to resolve between an origi-
stores are routed to consumer tissues, and how nal tissue signal and the asymptotic signal
these differ temporally in terms of dietary inte- reached at a new location also depends upon
gration. However, this important development the magnitude of the isotopic difference
by no means negates the conventional between these two signals. The greater the
approach, and the net elemental turnover mea- isotopic difference between initial and final
sured by fitting the exponential function pro- dietary conditions (i.e., the greater the value b),
vides a phenomenological estimate of the time and the smaller the variance associated with
period a given isotopic measurement of an each equilibrium condition, the more confident
organism represents (see also Carleton, Kelly, we are of detecting isotopic information from a
Anderson-Sprecher, & Martı́nez del Rio, 2008). previous location (Fig. 4.2).
Once we have assessed a realistic estimate Phillips and Eldridge (2006) explored the
of the half-life of an element for a tissue of a utility of using more than one tissue to detect
Biome 2
Tissue 1
Tissue 2
δX (per mil)
Biome 1
t′1 t′2 t1 t2
Time (days)
FIGURE 4.2 When an organism moves from one isotopic biome to another, our ability to detect the original biome sig-
nal will depend on the tissue we choose and the magnitude of isotopic separation between the two biomes. Here, tissue 1
has a faster elemental turnover rate than tissue 2. The bands about each biome indicate the isotopic resolution or measure-
ment error corresponding to organisms in that biome. The scenario of reducing the isotopic distance is demonstrated with
the arrows. Here, we can see that the time over which we can detect the original biome signal is reduced (primed
notation).

the time an individual organism has spent in a (k 1 m) and estimates of these errors are
new environment. Such information is less typically poorly known for most species. Finally,
useful for estimating where an organism came all models assume a direct dietary source of
from, but provides insights into the value of nutrients to tissues and do not consider situa-
the migratory stopover environment. This tions where organisms may be metabolizing
approach is based on the contrast between tis- stored nutrients for maintenance.
sues with different turnover rates, typically by The development of stable isotope methods
comparing a fast turnover tissue like liver, to tracking migratory wildlife has provided
plasma, or breath CO2 and a slower turnover a rich literature to illustrate the breadth of
tissue like muscle or the cellular fraction of applications using the light isotopes of C, N,
blood. The model assumes that the researcher H, O, and S (reviewed in Boecklen, Yarnes,
knows the initial and asymptotic tissue isotope Cook, & James, 2011; Hobson, 1999, 2005, 2008;
values, the measured isotope value of tissues Layman et al., 2012; Rubenstein & Hobson,
at some time after arrival, and the necessary 2004). With the increase in research in this
rate constants associated with the tissues used. field and interest in forensic tracing of plant
The model does a good job of estimating the and animal products (many studies with simi-
time since diet shift and the magnitude of the lar aims), it is no longer practical to provide a
isotopic difference between initial and asymp- detailed documentation of this field. Rather,
totic conditions, except in circumstances where this chapter briefly summarizes key concepts
the elapsed time was a fraction of a half-life and important findings, with due homage paid
of the slower turnover tissue, or when the to the original works.
diet shifts were small (i.e., less than 10 times In general, the isotope applications can be
the measurement error). Klaassen, Piersma, categorized into (1) inferences of animal ori-
Korthals, Dekinga, and Dietz (2010) and gins based on biome markers using isotopes of
Heady and Moore (2013) provide additional C, N, and S and (2) those using modeled
developments of models addressing time since continental-scale δ2H, δ13C, and δ15N isoscapes.
arrival or diet switch. The use of δ2H measurements in particular has
As with all models, solutions are based on brought tremendous opportunities, but also
assumptions. In the case of discerning previ- challenges as we attempt to fill in the informa-
ous geographical provenance of a migratory tion gaps associated with this element, hence
organism, we typically make the assumption hydrogen isotopes are discussed separately.
that the organism was in equilibrium with its
previous diet upon arrival. This will depend
on the tissue, and so there would be a much
higher likelihood of equilibrium conditions
4.2.2 Direct Isotope Tracers
reached for shorter turnover tissues. Use of Isotopic discrimination relevant to food
intermediate- or long-turnover tissues may be webs and animal movement in general is a
less useful for those migrants that move phenomenon of the light stable isotopes (e.g.,
quickly among different stopover locations. C, N, H, O, and S). For heavier elements, often
There are some circumstances where we are little or no isotope discrimination occurs
interested in knowing if the arrival signature is resulting in a more direct link to surficial geol-
different from the local food web signature, and ogy, hydrology, or underlying isoscapes. There
less concerned that it can be associated with a are however many exceptions and abiotic and
particular location. There will also be error biotic processes are involved in fractionation
associated with estimates of the rate constants of several metals (e.g., Coutaud, Meheut,

Viers, Rols, & Pokrovsky, 2014; Croal, Johnson, isotope as well as methodological improve-
Beard, & Newman, 2004) and so each isotope ments making it possible to measure relatively
must be considered on a case-by-case basis. small tissue samples (Chapter 2: Introduction
Unfortunately, isotopic analyses of the heavier to Conducting Stable Isotope Measurements
elements are costly and difficult because they for Animal Migration Studies).
are more involved than isotope ratio mass Natural variations in strontium isotope
spectrometry and require clean labs and the values of bedrock are determined by bedrock
use of more sophisticated instrumentation type and age varies regionally. Sr isotope sig-
such as Inductively coupled plasma mass spec- nals in tissues are relatively unvarying tempo-
trometry (ICPMS). Nonetheless, a few ele- rally, leading to the possibility of developing
ments such as Sr, Pb, and Hg show useful permanent isoscapes useful for the study of
isotopic structure that can vary spatially and animal movements. Sr has no discrimination
thus be applied to tracing animal movements. occurring from soils to higher trophic levels
Sulfur is one of the light elements for which (Flockhart, Kyser, Chipley, Miller, & Norris,
we typically expect little isotopic discrimina- 2015). Beard and Johnson (2000) modeled an
tion, but there are few S isotope studies on tis- expected 87Sr/86Sr isoscape for the United
sue types. Sulfur in consumer tissues is located States based on patterns of surficial geology
in sulfur-bearing amino acids (e.g., cysteine and applied that model to assigning human
and methionine) and so δ34S measurements are skeletal remains. We expect a general increase
closely linked to dietary protein pathways. in the 87Sr/86Sr ratio from west to east in
Unlike the other light isotopes, we expect little North America due to increasing age of bed-
S isotopic discrimination between diets and rock and Sellick, Kyser, Wunder, Chipley, and
consumer tissues because there is little oppor- Norris (2009) showed that adding δ87Sr mea-
tunity for the essential amino acids to be isoto- surements to δ2H in feathers of tree swallows
pically modified in consumers (Arneson & (Tachycineta bicolor) across a longitudinal gradi-
MacAvoy, 2005; Richards, Fuller, Sponheimer, ent in North America improved the precision
Robinson, & Ayliffe, 2003). As a result, δ34S of assignments to origin. Barnett-Johnson,
measurements are useful direct tracers in food Pearson, Ramos, Grimes, and MacFarlane
web and migration studies (Hebert & (2008) successfully characterized expected
Wassenaar, 2005; Krouse, Stewart, & Grinenko, δ87Sr values in watersheds supporting specific
1991). Proximity to coastlines influences terres- populations of Atlantic salmon. Examination
trial food web δ34S values due to the fallout of of δ87Sr values in otoliths allowed these
sulfur compounds derived from sea spray authors to identify the natal stream origin of
(Jamieson & Wadleigh, 2000; Zazzo, Monahan, fish taken at sea. That work underlined the
Moloney, Green, & Schmidt, 2011). In terres- amount of work required to first understand
trial systems, δ34S values in consumer tissues underlying δ87Sr patterns in specific regions of
are influenced by underlying geology with interest and the scale of movements of interest.
marine-derived sediments and volcanic rocks By combining lithology-specific parameters in
having relatively enriched values compared to addition to bedrock age in predictive models,
other substrates. Animals linked to sources of progress has been made in refining the δ87Sr
sulfur in turn derived from anaerobic sedi- model isoscape for the continental United
ments in marshes can also show enrichment. States by Bataille and Bowen (2012) and for the
We expect the use of δ34S measurements to Caribbean by Bataille et al. (2012).
increase substantially in the coming years due Like Sr, the stable isotopes of trace elements
to the lack of discrimination involving this of lead (204Pb, 206Pb, 207Pb, and 208Pb) reflect

bedrock source from which they are derived. Photochemical transformations of Hg com-
Anthropogenic sources of Pb can be traced pounds can result in isotopic changes that are
due to derivation from different ore sources independent of mass. The even-mass
and there has been considerable application of isotopes of Hg show mass dependent fraction-
Pb isotope tracing because this element is an ation whereas the odd isotopes do not. Point
important environmental pollutant. Most nota- et al. (2011) investigated Hg isotope concentra-
bly, Pb isotopes have proven useful for tracing tions in the eggs of seabirds occupying areas
distributions and origins of leaded gasoline, with different ice cover in coastal Alaska and
ash from coal burning, and incineration profound considerable isotopic variation that
ducts in the environment (reviewed by could be linked to the extent of photochemical
Komarek, Ettler, Chrastny, & Mihaljevic, 2008). breakdown of MeHg into less toxic forms.
Measurements of concentrations and Thus the Hg isotopic composition of seabird
stable isotopes of Pb in animals have proven tissues carried spatial information related to
useful in evaluating sources of exposure. For latitudinal patterns of ice cover in spring.
example, sources of incidental ingestion of Pb While the focus of this book is almost
by raptors or scavengers consuming hunter- entirely on the abundant light isotopes, it is
killed prey have been linked directly to Pb worth recalling heavier isotopes such as those
ammunition used (Finkelstein et al., 2010). An of Sr, Pb, and Hg and assays of the (nonisoto-
interesting application on sources of elephant pic) concentrations of trace elements have been
ivory in Africa involved the measurement of used considerably to infer origins and move-
provenance through the tracing of isotopes in ments of biota (Chapter 2: Introduction to
ivory of several elements including Pb (Vogel, Conducting Stable Isotope Measurements for
Eglington, & Auret, 1990) but few studies have Animal Migration Studies). The most useful
used Pb isotopes to infer animal movements or heavy isotope situations involve improved
provenance beyond stock identification. assignment of individuals to a relatively small
Stewart, Outridge, and Stern (2003) examined number of possible populations that can be
the 208Pb/207Pb and 206Pb/207Pb ratios in teeth well described in terms of their trace element
of Walrus (Odobenus rosmarus) in the Canadian or stable isotope composition. We will later
Arctic and were able to infer life history move- revisit the concept of statistically combining
ments. Walrus are particularly amenable to isotopic and nonisotopic data to improve
this approach because they feed almost exclu- assignments of animals to their origin.
sively on sedentary bivalves and so are closely
linked to the geochemical environment and
localized food webs. 4.2.3 Tracing Involving Isotopic
While the lack of isotopic discrimination Discrimination Factors or Calibrations
between biota and underlying inorganic sub-
strates for heavier trace elements allow for 4.2.3.1 Nitrogen Isotopes
direct creation of isoscapes, recent investiga- Despite extensive use of N stable isotopes in
tions into the behavior of the stable isotopes of delineating animal diets and trophic relation-
Hg (198Hg, 199Hg, 200Hg, 201Hg, and 202Hg) ships, the use of δ15N measurements in tracing
highlight their complexity in nature. Mercury origins of animals is rare. This is due to the
is deposited in polar regions through atmo- fact that δ15N values in plant and animal tis-
spheric transport and a variety of biogeochem- sues vary tremendously, regionally and at
ical processes result in methylation of Hg into small spatial scales due to numerous natural
a bioavailable form (MeHg), which is toxic. and anthropogenic influences on the N cycle

(Vitousek et al., 1997) which range from land- Thus inferring origins using bulk tissue N iso-
use practices, fertilizer use, sewage disposal, tope analyses requires knowledge of diet and
and the release of nitrogenous compounds into this can be a challenge for omnivores that may
the atmosphere (Pardo & Naddlehoffer, 2010). move across regions with changing baseline
Such N isotope variation is impossible to δ15N values. Tissue δ15N measurements repre-
model in terms of predictable, continent-wide, sent a means of tracing protein pathways
isoscapes. However, in more natural settings, derived from diet because this element is
foliar δ15N has been modeled globally by largely absent in lipids and carbohydrates.
Craine et al. (2009) and influences of climatic This means that linking animals back to iso-
variables, N and P availability, N fixation pro- scapes using tissue δ15N values is theoretically
cesses, and types of mycorrhizal fungi have more feasible for carnivores and frugivores
been identified as controlling factors. Thus than for omnivores. For essential amino acids,
while general large-scale phenomena affecting nitrogen will largely be incorporated with little
terrestrial food web δ15N values are under- isotopic discrimination into the protein pool of
stood, high isotopic variance at more local to the consumer. Nonessential amino acids typi-
regional scales is to be expected. Nonetheless, cally involve more opportunities for isotopic
some researchers have attempted to use the discrimination during protein synthesis and
foliar δ15N isoscape provided by Craine et al. so the net discrimination we see for δ15N mea-
(2009) to produce tissue-specific isoscapes to surements in consumers will reflect the degree
assist in assignment of birds to molt origins in to which the diet meets the amino acid
Africa (Hobson, Møller, & Van Wilgenburg, requirement of the consumer (Robbins,
2012; Hobson, Van Wilgenburg, Faaborg, et al., Felicetti, & Sponheimer, 2005).
2014; Hobson, Van Wilgenburg, Wassenaar, & In general, poorer quality diets result in
Larson, 2012; Hobson, Van Wilgenburg, greater overall diet-tissue discrimination for
15
Wassenaar, Powell, et al., 2012; Veen et al., N than high-quality diets. An important
2014). More typically, however, δ15N values in derived variable in experiments designed to
animal tissues have been used to infer the type establish tissue-specific δ15N values in migra-
of biome supporting animals during tissue tory animals is the elemental C:N ratio of the
growth. Generally, marine sources of N are diet, as this ratio provides a useful indicator of
typically more enriched in 15N than terrestrial diet quality and the N isotope discrimination
sources. In terrestrial systems, untilled soils factor to apply in natural situations. N isotopic
are less enriched in 15N than those exposed by discrimination will also depend on the means
agriculture. Following this principle, Hobson of voiding nitrogenous waste. Here, a major
(1999) demonstrated that feathers grown in difference occurs between aquatic inverte-
boreal biomes are more depleted in 15N than brates that void nitrogen via ammonia com-
those from agricultural landscapes. In general, pared to terrestrial vertebrates (Post, 2002).
hotter, dryer regions have food webs with There is also evidence that ungulates adapted
higher δ15N values compared with those in to arid conditions conserve water by recycling
cooler or wetter areas. urea that ultimately influences whole body
Another fundamental issue complicating the tissue δ15N values (Ambrose & DeNiro, 1986;
application of δ15N measurements to infer ani- Sealy, van der Merwe, Lee Thorp, & Lanham,
mal origin is that this isotope is influenced by 1987). Hobson, Alisauskas, and Clark (1993)
trophic position, with bulk tissue δ15N values also determined that birds that fast and
increasing by about 2.5m5m with each tro- undergo significant protein catabolism during
phic level (Layman et al., 2012; Post, 2002). incubation, like geese breeding at high

latitudes, also experience an increase in body physiological and ecological factors that can
δ15N values. influence tissue δ15N independent of location.
Knowledge of these sorts of physiological These limitations may be partially addressed
processes is necessary when using tissue δ15N through compound-specific amino acid δ15N
values of migratory organisms to infer origins. analyses which can inherently account for tro-
The prevailing consensus is that researchers phic position and baseline δ15N. In general, N
should strive to use the most parsimonious isotopes are best considered as providing addi-
nitrogen isotope discrimination value associ- tional locational information due to biome
ated with their specific organism of interest. A characteristics and known land use practices
review of N isotopic discrimination across sev- in terrestrial systems.
eral taxa by Vanderklift and Ponsard (2003)
identified mode of excretion and environment 4.2.3.2 Carbon Isotopes
(marine, freshwater aquatic, and terrestrial) as Unlike nitrogen and sulfur, carbon is present
important factors (see also Boecklen et al., 2011; in all dietary macromolecules (protein, fat, and
Post, 2002). In their analysis, Caut, Angulo, and carbohydrates), hence δ13C measurements of
Courchamp (2009) provide a summary of δ15N consumer tissues reflect these various sources.
discrimination factors that will prove useful The diverse sources of carbon contribute to vari-
and suggest that these factors are also depen- able diet-tissue δ13C discrimination factors com-
dent upon dietary or baseline δ15N values, but pared to other light elements. However, in
more research is needed to confirm mechan- many cases, lipids in diets are transferred
isms underlying this suggestion. directly with little isotopic modification to
A significant advance in the use of lipids in the consumer. Carbohydrates are
compound-specific isotope analyses (CSIAs) has often burned directly for energy production,
been the identification of amino acids whose producing CO2 as the only carbon byproduct,
δ15N values largely reflect dietary source (i.e., and hence δ13C values in breath CO2 are used
show little change in δ15N with trophic level) to trace origins of carbohydrates in diet.
compared to those that show strong trophic dis- Unfortunately, we have little idea of the appro-
crimination, the so-called trophic amino acids. priate carbon isotopic discrimination factors
For example, glutamate is a trophic amino acid that currently apply between dietary substrates
compared to phenylalanine. Measuring the δ15N and breath CO2 (McCue & Welch, 2016;
difference between these amino acids within the Podlesak, McWilliams, & Hatch, 2005). Carbon
consumer can thus control any changes in base- isotope values of protein theoretically originate
line δ15N and trophic position. This phenome- from all three dietary macromolecules, but is
non clearly helps resolve two of the primary more likely associated with dietary proteins
limitations to using bulk tissue δ15N values to especially for carnivores. In general, we expect
infer origins of migratory animals, the ambigu- lower diet-tissue isotopic discrimination fac-
ity related to an animal coming from a region of tors for δ13C than for δ15N. However, when
unknown baseline δ15N and unknown trophic using bulk tissues, researchers need to be
position. It is possible then, to imagine a source aware of the significant intertissue differences
amino acid (say phenylalanine) δ15N isoscape to in C isotope discrimination factors linking
assist with assignments. animals with isoscapes.
While it is possible to generate tissue δ15N For the most part, δ13C values in animal
isoscapes for the purposes of spatial assign- tissues are used to infer diet and biome infor-
ment, the use of this isotope is limited and mation. The strong correspondence between
should be used with caution due to the many δ13C values in primary production and

photosynthetic pathway (C3, C4, and CAM) pro- 4.2.3.3 Hydrogen and Oxygen
vides strong linkages to climate in terrestrial sys- As emphasized below and in Chapter 3,
tems. In marine systems, algal growth rates Isoscapes for Terrestrial Migration Research,
influence food web δ13C values and are strongly hydrogen isotopes are particularly powerful for
linked to nutrient availability and sea-surface tracking migratory wildlife. However, this ele-
temperature (Chapter 6: Isotopic Tracking of ment presents challenges in terms of fully
Marine Animal Movement). Spatial information understanding how δ2H measurements of con-
associated with food web δ13C values related to sumer tissue relate to hydrogen sources that, in
plant or primary productivity distributions form most terrestrial systems, are driven by the global
the basis for tracing animal origin and move- water cycle. Like carbon, hydrogen occurs in all
ment. A particularly useful development in gen- three dietary macromolecules and so recognition
erating expected plant or animal tissue δ13C of metabolic routing is important. As outlined
isoscapes has been the modeling of expected rel- in Chapter 2, Introduction to Conducting
ative distributions of C3 versus C4 plant Stable Isotope Measurements for Animal
ecozones (Still & Powell, 2010). These can be Migration Studies, measurement of δ2H in the
converted into spatially explicit δ13C predicted nonexchangeable fraction of hydrogen is chal-
surfaces. Hobson, Van Wilgenburg, Wassenaar, lenging but within the consumer, a portion of
and Larson (2012) applied a 1m C isotope dis- the hydrogen in any tissue exchanges with body
crimination factor between predicted plant tissue water, a component which is presumably more
δ13C and herbivorous insects and another 1m C labile than dietary-derived hydrogen. Drinking
isotope discrimination value linking insects to water as well as diet thus constitutes a source of
feathers of insectivorous birds to produce a H in animals and this makes it difficult to
feather δ13C isoscape for Africa. Several sources derive linkages between precipitation δ2H and
of error are associated with continental patterns tissue δ2H that are universally applicable. Using
and like 15N, can be heavily influenced by a controlled laboratory study, Hobson et al.
anthropogenic factors such as agriculture (e.g., (1999) maintained quail (Coturnix japonica) on a
the extensive planting of C4 crops such as corn, single diet but exposed groups to drinking
sorghum, and millet), irrigation, and land use. water of vastly different δ2H values. They found
Future δ13C isoscapes will involve year-specific that H from drinking water accounted for about
agricultural crop layers for cases where migrant 20% of the total H in various tissues.
animals use crops such as those expected for Interestingly, this was the case for lipids with
migrant pest insects (Hobson et al., 2018). no exchangeable H bonds indicating that body
Recent advances in the use of compound- water can exchange with H in precursor mole-
specific δ13C analyses of plant and animal tis- cules involved in lipid synthesis. Because such
sues opened new avenues for tracing animal details do not exist for most animal systems,
movements. For example, plants, fungi, and overall H isotope discrimination factors are esti-
bacteria have characteristic amino acid δ13C mated using phenomenological approaches, as
patterns because of unique pathways of amino discussed below. Evidence suggests that most of
acid synthesis (Larson & Hobson, 2009). Thus the H isotope discrimination between environ-
CSIA using δ13C analyses can be applied to mental waters and animal tissues occurs at the
infer patterns of dietary shifts of tissues syn- incorporation of H from water into plant tissues
thesized at different times and geographical with little subsequent trophic changes.
regions (Chapter 7: Amino Acid Isotope The use of δ18O measurements to track wild-
Analysis: A New Frontier in Studies of Animal life remains rare because of technological con-
Migration and Foraging Ecology). straints of routinely measuring oxygen isotopes

4.3 MOVEMENTS INFERRED WITHOUT ISOSCAPES 95
in animal tissues. That situation has now chan- terrestrial food webs with marine organisms
ged due to online pyrolytic techniques and the having more positive δ13C, δ15N, δ34S, δ2H, and
development of keratin O isotope standards δ18O values compared to terrestrial counter-
(Chapter 2: Introduction to Conducting parts (Hobson, 1999). As many migratory
Stable Isotope Measurements for Animal organisms use terrestrial and marine biomes
Migration Studies). Pekarsky et al. (2015) suc- throughout their annual cycles, these marine
cessfully used feather δ18O measurements to versus terrestrial isotopic differences become
infer breeding origins of Eurasian Cranes (Grus useful (e.g., Atkinson et al., 2005). Plant phy-
grus) wintering in Israel. Moreover, that study siologists have also pioneered the use of
clearly indicated how modeled feather δ18O stable isotope measurements to discern C3-,
isoscapes could be derived using satellite data. C4-, and CAM-based photosynthetic pathways
The tight coupling between δ18O and δ2H using δ13C and δ2H measurements with clear
values in meteoric water provides opportunity implications for reconstructing animal diets
to infer additional information related to cli- and origins (Wolf & Martinez del Rio, 2000).
mate, but that correlation between H and O Investigations into the effect of water-use effi-
isotopes breaks down in food webs involving ciency mechanisms in C3 plants that generally
animals, due to varying degrees based on fac- leads to an enrichment of plant tissue 13C have
tors affecting each isotope differently (Hobson also alluded to winter origins of migratory
& Koehler, 2015). Thus, in many cases, no addi- birds (Marra, Hobson, & Holmes, 1998;
tional information is gained from δ18O over Bearhop, Furness, Hilton, Votier, & Waldron,
δ2H measurements on the same tissue. As well, 2003) and to the existence of carry over effects
δ2H values typically span a much wider range that can influence subsequent life history
than δ18O measurements in terrestrial food stages (Norris, Marra, Kyser, Sherry, &
webs and so can potentially provide greater Ratcliffe, 2004). While most of these applica-
resolution with respect to source discrimination tions involve indirect spatial association
and an overall better signal-to-noise ratio. through knowledge of animal life history and
Oxygen occurs in proteins but not in lipids or the biomes they use, these studies paved the
carbohydrates. However, sources of oxygen way for more spatially explicit approaches to
include drinking water and air for metabolism tracing animal movements and so a brief
and thus, like H, it is difficult to predict O history is warranted.
isotopic discrimination factors associated with One of the earliest applications of
each contribution and “working values” stable isotope methods to investigate animal
will need to be derived largely from future spatial movement was by Killingly (1980) who
examination of wild and captive animals. inferred the temperature of water during cal-
cite formation of barnacles attached to the skin
of California Gray Whales (Eschrichtius robus-
4.3 MOVEMENTS INFERRED tus) using δ18O measurements and by Killingly
WITHOUT ISOSCAPES and Lutcavage (1983) who examined δ18O and
δ13C measurements in barnacles on loggerhead
The strong link between stable isotope turtles (Caretta caretta). That work has since
values in animal and plant tissues and biogeo- been followed by studies on the inorganic frac-
chemical processes have provided the basis for tion of otoliths in freshwater and marine fish
linking organisms to regions or biomes. Some to infer movements (e.g., Kennedy, Folt, Blum,
of the earliest isotopic investigations revealed & Chamberlain, 1997; Meyer-Rochow, Cook, &
distinct differences between marine and Hendy, 1992). Other examples of using marine

isoscapes to infer spatial movements of marine Migratory movements of fish with an anad-
mammals are by isotopic analyses of the romous life stage present an isotopic opportu-
baleen plates of the western Arctic population nity and, have the added advantage of
of bowhead whales (Balaena mysticetus) migrat- carrying an isotopic record in their otoliths
ing annually between the Beaufort and Bering and scales (Nelson, Northcote, & Hendy, 1989;
seas (Schell, Saupe, & Haubenstock, 1989) and Trueman & Moore, 2007) and soft tissues
southern right whales (Eubalaena australis) that (Hobson et al., 2007). Kennedy et al. (1997)
annually cross the Southern ocean conver- and Harrington, Kennedy, Chamberlain,
gence, a zone of dramatic changes in food web Blum, and Folt (1998) demonstrated how
δ13C and δ15N (Best & Schell, 1996). Trueman stable isotopes of several elements can be used
et al. (Chapter 6: Isotopic Tracking of Marine on the organic and inorganic fractions of oto-
Animal Movement) provide more information liths to identify natal streams of Atlantic
on the derivation and use of marine isoscapes salmon (Salmo salar) intercepted as adults at
to infer movements of fish and marine sea. Essentially, the suite of δ13C, δ15N, and
mammals. δ87Sr measurements form unique combina-
Within terrestrial and freshwater habitats, tions of values reflecting the geological sub-
there clearly is substantial isotopic variability strate and land-use practices surrounding
that can be used to examine movements of drainage basins of key salmon-producing
fish. Hesslein, Capel, Fox, and Hallard (1991) streams (Barnett-Johnson et al., 2008, reviewed
used δ13C, δ15N, and δ34S measurements of by Walther & Limburg, 2012).
muscle in broad whitefish (Coregonus nasus) Recently, there has been interest in using
and lake whitefish (Coregonus clupeaformis) in freshwater fish tissue δ2H values because
two freshwater regions of the Mackenzie Delta stream and river H inputs to lake systems may
in Northern Canada to infer their movements. have very different water δ2H values (Doucett,
The movement of animals between marine, Marks, Blinn, Caron, & Hungate, 2007). Thus
estuarine, and terrestrial or freshwater habitats freshwater systems have great potential for a
holds great potential for inferring their past multiisotope approach to trace migrations and
habitat use and potential migratory origins. movements of aquatic species. They also have
Tietje and Teer (1988) were among the first to the advantage of being reasonably tightly con-
use stable isotope methods to investigate how strained spatially and it should be possible to
wintering Northern Shoveler (Anas clypeata) literally create multiisotopic basemaps of the
ducks use coastal and inland freshwater wet- major aquatic space used by migrant fish
lands and were able to demonstrate sedentary (Soto, Hobson, & Wassenaar, 2016).
behavior among late wintering individuals. In terrestrial systems, some of the earliest
Other studies have primarily used δ13C mea- applications of isotopic measurements used for
surements to infer movement of piscivorous tracking origins of animals were conducted in
birds between marine and freshwater habitats Africa. Two simultaneous yet independent
(Bearhop et al., 1999; Mizutani, Fukuda, studies used stable isotope measurements of
Kabaya, & Wada, 1990). Hobson, Blight, and elephant (Loxodonta africana) ivory and bone
Arcese (2015) used a multiisotope approach to collagen to infer origins of that material as a
investigate use of terrestrial, marine, and fresh- forensic tool to counter the illegal ivory trade
water resources by coastal wintering gulls near (Van der Merwe et al., 1990; Vogel et al., 1990).
an urban center and the use of agricultural Elephants feeding primarily on grasses sample
land by coastal shorebirds (Hobson, Slater, a C4 food web and so have more positive δ13C
Lank, Milner, & Gardiner, 2013). values compared to those feeding in

4.3 MOVEMENTS INFERRED WITHOUT ISOSCAPES 97
woodlands on C3 browse. Elephants feeding in Unlike most Nearctic migratory birds,
more arid areas may also have higher δ15N Palearctic species typically replace flight feath-
values than those in more mesic habitats ers on their wintering ground in Africa, so it
(Heaton, 1987). Combined with assays of Pb was possible to infer aspects of the wintering
and Sr isotopes, these studies showed strong habitats using isotopic analyses of these feath-
segregation among several African elephant ers. Feathers grown in mesic habitats of sub-
populations and underlined the forensic utility Saharan West Africa are expected to have
of stable isotopes to infer origins of several lower δ13C values than those grown in Central,
taxa. Unfortunately, some of the early enthusi- East, or South Africa and that was consistent
asm was later tempered by the observation of with analyses performed by Bensch,
strong year-to-year variations in food web Bengtsson, and Åkesson (2006) on two subspe-
δ15N values within the Amboseli National cies of flycatcher breeding in Sweden.
Park presumably due to climatic variation Following the example of Møller and Hobson
(Koch et al., 1995). This illustrates the need to (2004) who examined African grown feathers
know the natural range of variation in of barn swallows breeding in Europe,
stable isotope patterns spatially and tempo- Schmaltz, Loonstra, Wymenga, Hobson, and
rally when evaluating the accuracy of the tech- Piersma (2017) used blood and feather δ13C,
nique when inferring animal origins. δ15N, and δ2H values to infer likely origins of
Fortunately, many terrestrial systems are Ruffs (Philomachus pugnax) arriving in Europe
less erratic isotopically and show consistent during spring migration. In addition to a mul-
isotopic patterns over decadal and longer time tivariate analysis involving three isotopes, the
frames. This is especially the case with the use bivariate plot they produced of δ13C versus
of δ13C measurements to track the use by ani- δ15N established quadrats inferring (1) birds
mals of C3, C4, and CAM food webs. An inter- wintering in Europe based on high δ15N
esting application of δ13C measurements to (manure-based) agriculture and low δ13C (C3)
investigate mechanisms affecting the phenol- plants; (2) birds wintering in sub-Saharan
ogy of animal migration was that of Africa and associated with livestock (high
Flemming, Nunez, and Sternberg (1993) who δ15N) and C4 plants (high δ13C); (3) birds from
showed that the nectarivorous bat Leptonycteris sub-Saharan Africa associated with C4 irri-
curasoae switched from C3 flowering plants gated agriculture (high δ13C and low δ15N)
during the winter to CAM flowering columnar and birds molting in unknown areas but asso-
cacti as it migrated north in the spring. The bat ciated with low δ13C and low δ15N (mesic or
tissue isotopic data revealed how the species irrigated C3 biomes possibly associated with
has adapted to the phenology of CAM “nectar rice agriculture).
corridors” during northward migration. The analysis of the multiisotope structure of
Other studies have exploited the strong C4 breeding populations of animals can also be
signal of agricultural corn to infer the origins used as a means of detecting immigrants into
of migratory herbivorous birds (Henaux, those populations even when the isotopic
Powell, Vrtiska, & Hobson, 2012) and their rel- structure of the landscape is unknown. For
ative dependence on C3- and C4-based food example, Hobson, Wassenaar, and Bayne
webs (Wassenaar & Hobson, 2000). However, (2004) used this approach to investigate mini-
δ13C isoscapes in North America or other areas mum estimates of dispersal into breeding
with intense agricultural production can have populations of American Redstarts and
a mix of C3 and C4 plants and so will be com- Ovenbirds (Seiurus aurocapillus), using δ2H
plicated to model. measurements of feathers. The challenge is to

derive statistically defensible criteria for distin- Hobson, Wassenaar, and Taylor (1999) first
guishing among local and immigrant indivi- created an isotopic basemap corresponding to
duals and the associated risk of making an butterflies produced throughout their breeding
error in assignment. range during the summer of 1996. An isotopic
The power of isotopic inference in assigning basemap is composed of isotope measure-
tissues of animals and plants to regions or ments made on individuals from known
biomes depends to a large degree on the life locations that spans the entire breeding range.
history of the organism in question and the This feat was accomplished through the aid of
confidence one can associate with isotopic pat- the nonprofit MonarchWatch organization
terns in nature. A more powerful approach is (monarchwatch.org), who were able to solicit
the use of tissue-specific isoscapes that provide volunteers and educators from 86 locations
predictive surfaces (isoscapes) of expected across the monarch breeding range to success-
tissue isotope values. fully raise 412 butterflies on milkweed grown
locally. Only milkweed watered by rainfall
was used. From that sample, butterflies from
4.4 USING ISOSCAPES 33 sites were selected for δ2H and δ13C analy-
ses of wing tissue performed to produce a
The first comprehensive application of δ2H year-specific basemap depicting isotopic pat-
measurements in the study of migratory ani- terns for C and H isotopes.
mals was an investigation of the isotopic struc- In addition to the wild-reared group of
ture of populations of Monarch butterflies monarchs, the relationship between δ2H and
(Danaus plexippus) wintering in Mexico. The δ13C of milkweed tissue and chitin in wings
eastern population of the Monarch Butterfly in and the water used to raise milkweed was
North America overwinter in roost sites in the investigated under controlled laboratory con-
high-altitude Oyamel Fir (Abies religiosa) for- ditions using three batches of known δ2H
ests of central Michoacan and Mexico states. In water. Those captive studies showed extremely
spring, only gravid females migrate north, tight (r2 5 0.99) relationships, demonstrating
reaching Texas, USA, where they lay eggs on that insect wing chitin δ2H is derived exclu-
milkweed (Asclepias species) plants. The new sively from water available to plants with
generation emerging travels further north to most of the isotopic discrimination occurring
repeat the process at higher latitudes. Finally, between water and plants (see also Ostrom,
in one of the most spectacular migrations of Colunga-Garcia, & Gage, 1997). However, in
any animal, in late summer monarchs 46 their investigation of wild monarchs,
generations removed from those ancestors that Wassenaar and Hobson (1998) applied the
migrated northward from Mexico the previous derived basemap based on the “outdoor” sam-
spring then return to the same roost sites that ple to portray origins of monarchs who were
they have never seen before. produced during 1996 and later collected from
Tracking origins where the overwintering all known winter roost sites in Mexico that
butterflies were produced has emerged as a winter (i.e., 199697). The authors reasoned
central question in the conservation of that the calibration based on the outdoor mon-
Monarchs due to their declining numbers and archs from known locations would better
the real possibility their migratory phenomenon encompass natural isotopic variation. That
may go extinct. Only by identifying the key approach resulted in the insight that the winter
production zones, we can target conservation roost sites were panmictic, made up of butter-
efforts effectively on the breeding grounds. flies from all over the breeding range, and

4.4 USING ISOSCAPES 99
most importantly, revealed that half the mon- study was conducted to see if tissues grown
arch population was produced largely in along an altitudinal gradient reflected such
Kansas, Nebraska, Iowa, Missouri, Wisconsin, patterns. Hobson et al. (2003) examined δ2H
Illinois, Michigan, Indiana, and Ohio corre- and δ13C values of feathers of hummingbirds
sponding to the corn, soybean, and dairy pro- inhabiting the Ecuadorean Andes and found
ducing region of the Midwest USA. Thus good agreement between actual and predicted
while conservation of this species was previ- feather δ2H based entirely on a global model
ously focused almost entirely on the precari- (Chapter 3: Isoscapes for Terrestrial Migration
ous winter roosts in Mexico, these isotope Research). As expected, feather δ13C values
studies pointed to the possibility that prime increased with altitude. Thus, for any given
monarch breeding habitat was concentrated in species, it may be possible to estimate approxi-
areas of intense agricultural production in the mate elevations at which feathers, and other
United States where milkweed was controlled tissues were grown. By examining different
and where genetically modified corn was tissues with different windows of isotopic
being used that produced BtK, a bacterium integration, the possibility exists to infer previ-
that targets Lepidoptera (Losey, Rayor, & ous altitudinal movements.
Carter, 1999).
Since these landmark studies, stable isotope
measurements have been used to infer (1) pat-
terns of monarch spring recolonization in east-
4.4.1 Assignment to Bins
ern North America (Flockhart et al., 2013; The creation of predictive, tissue-specific,
Miller, Wassenaar, Hobson, & Norris, 2012), (2) isoscapes allows for a variety of approaches to
the origins of wintering individuals in western inferring past origins and movements of ani-
North America (Yang, Ostrovsky, Rogers, & mals to “place them on the map.” This topic is
Welker, 2016), (3) the effects of natal origin on elaborated in Chapter 6, Isotopic Tracking of
parasite loads (Altizer, Hobson, Davis, De Marine Animal Movement. To date, research-
Roode, & Wassenaar, 2015), (4) the role of wing ers have relied on assignment of individuals or
coloration in flight distance (Hanley, Miller, populations to arbitrary spatial bins or regions
Flockhart, & Norris, 2013), and (5) general of interest, or have adopted a more spatially
conservation concerns related to areas of high explicit assignment approach. Both approaches
productivity (Flockhart, Brower, et al., 2017; rely on the ability to link individual tissues
Flockhart, Dabydeen, et al., 2017). δ-values to an expected value at a given scale
An interesting aspect of the behavior of δ2H (ranging from thousands of km2 in the case of
is that deuterium in precipitation tends to rain bins to hundreds of m2 in the case of pixels).
out more at lower elevations than at higher The success of this approach clearly rests on
elevations. This is a well-known phenomenon how well modeled isoscapes represent reality.
that results in an altitudinal “depletion” in 2H Dichotomous assignment to bins can be useful,
from 21m to 24m per 100 m rise in elevation, especially if those bins represent management
depending on the gradient and temperature units or regions relevant to a question. For
(Clark & Fritz, 1997). Similarly, the demands example, Norris et al. (2006) wanted to know
of plant adaptations to harsher growing condi- the degree of migratory connectivity between
tions at higher altitudes tend to result in plants breeding and wintering grounds of “populations”
with higher δ13C values at higher elevations. of the Neotropical migrant, American redstart
As there are several species that perform alti- (Setophaga ruticilla). Those authors looked
tudinal migrations, especially in the tropics, a at 12 wintering populations and assigned

individuals to one of the five breeding origin Potential problems with the “assignment to
bins using likelihood assignment. Each winter- bins” approach involve how well researchers
ing population could then be categorized as can estimate the expected mean and variance
representing various proportions of breeding in isotope values associated with regions.
ground origins and thereby reveal patterns of Derivation of expected isotope values associ-
connectivity. In this case, western breeding ated with bins is usually accomplished by
birds clearly wintered in the western portion sampling isoscape points within bins and the
of their nonbreeding range and eastern birds degree to which such values are useful will
largely in the Caribbean. Flockhart, Brower, depend on the nature and robustness of the
et al. (2017) used this approach to assign underlying isoscape and the sample coverage.
monarch butterflies from the winter roost sites Problems also arise with assignments close to
in Mexico to one of the six breeding regions in sharp bin boundaries.
the United States and Canada. That approach
was taken because there was considerable
interest in the role of the American Midwest
4.4.2 Spatially Explicit Assignments
agricultural region to monarch production Several applications using δ2H measure-
at continental scales. Ashley, Hobson, Van ments have involved migratory birds in North
Wilgenburg, North, and Petrie (2010) used a America that had a strong conservation moti-
hybrid approach by assigning harvested vation (Figs. 4.3 and 4.4). These studies have
American Black Ducks (Anas rubripes) to relied on derived calibration algorithms to link
origins in a spatially explicit way but within individuals and populations to tissue isoscapes
three distinct flyway bins (Eastern, Central, (reviewed in Hobson, 2008; Table 4.1).
and Western). Unfortunately, there are exceptionally few
FIGURE 4.3 Migratory connectivity determined using stable isotope analyses of feathers based on the results of (A)
Rubenstein et al. (2002) for the Black-throated Blue Warbler (Setophaga caerulescens) and (B) the leapfrog migration pattern
of the Wilson’s Warbler (Wilsonia pusilla) discovered by Kelly, Atudorei, Sharp, and Finch (2002).

FIGURE 4.4 Examples of spatially explicit assignment to molt origins of Golden-winged Warblers (Vermivora chrysop-
tera) sampled at two sites on their wintering grounds and showing strong differences in migratory connectivity.
Assignment was based on δ2H measurements of feathers and subsequent assignment to a feather δ2H isoscape for North
America using probabilistic assignment techniques discussed in Chapter 8, Design and Analysis for Isotope-Based Studies of
Migratory Animals and Chapter 9, Isoscape Computation and Inference of Spatial Origins With Mixed Models Using the R
package IsoriX. Figure legends depict the number of individuals assigned to the same pixel (see Hobson et al., 2016).
studies that have attempted to derive calibra- (resident, short distance, and Neotropical) influ-
tion algorithms on any continent. In North enced the calibration. Lott and Smith (2006)
America, Hobson, Van Wilgenburg, Wassenaar, provided a feather δ2H isoscape and
and Larson (2012) examined songbird feathers calibration algorithm for North American rapto-
from an extensive (Monitoring Avian rial birds and Cryan, Bogan, Rye, Landis, and
Productivity and Survivorship: MAPS) collec- Kester (2004), Cryan, Stricker, and Wunder
tion housed by the Center for Tropical (2014), and Popa-Lisseanu, Sörgel, Luckner,
research at the University of California at Los Wassenaar, and Ibáñez (2012) provided the first
Angeles (UCLA). That study was based on 544 δ2H isoscapes for bat hair. Calibration algo-
feathers from 40 species representing 140 rithms for insects have been recently summa-
known locations. In addition to choosing feath- rized by Hobson, Doward, Kardynal, and
ers from several guilds and across a large lati- McNeil (2018).
tudinal gradient, the researchers specified that It is important that we increase our under-
feathers be taken, where possible, from indivi- standing of the isotopic variance to be expected
duals that were philopatric to sampling site as for precipitation-based isoscapes across taxa
indicated by band returns. That study showed and across geography despite the difficulty in
significant within-population variance but the acquiring and sampling known-origin tissues
model accounted for B80% of the variance and across large spatial gradients. Almost nothing
suggested that foraging guild (ground, canopy, is known about appropriate calibrations for
and shrub/aerial) and migratory strategy Africa where millions of Palearctic-Afrotropical

TABLE 4.1 Relationship Between Stable Hydrogen Isotope Ratios of Environmental Waters (δ2Hp) and the δ2H
Values of Animal Tissues Assumed to Have Been Produced at Known Sites (See Hobson, 2008 for Additional
Examples)
Speciesa Calibrationb Source
BIRDS
Waterfowl δ2H 5 227.8 1 0.95δ2Hp Clark, Hobson, and Wassenaar (2006)
δ H 5 257 1 0.84δ Hp
2 2
Hebert and Wassenaar (2005)
Songbirds (NGF) δ H 5 217.6 1 0.95δ Hp

2 2
Hobson, Møller, et al. (2012)
Songbirds (NO) δ H 5 227.1 1 0.95δ Hp
2 2
Songbirds (SDGF) δ H 5 223.0 1 0.95δ Hp
2 2
Songbirds (SDO) δ H 5 236.9 1 0.95δ Hp
2 2
Songbirds (RGF) δ H 5 227.9 1 0.95δ Hp
2 2
Songbirds (RO) δ H 5 211.2 1 0.95δ Hp
2 2
Raptors δ H 5 25.6 1 0.91δ Hp

2 2
Lott and Smith (2006)
C
BATS
δ2H 5 225 1 0.8δ2Hp Cryan et al. (2004)
δ H 5 242.6 1 0.73δ Hp
2 2
Cryan et al. (2014)
δ H 5 216.8 1 1.07δ Hp
2 2
Popa-Lisseanu et al. (2012)
INSECTS
Monarch butterfly wild δ2H 5 279 1 0.62δ2Hp Hobson et al. (1999)

Monarch butterfly lab δ H 5 253 1 0.5δ Hp
2 2
Hobson et al. (1999)
Dragonfly spp δ2H 5 242.5 1 0.91δ2Hp Hobson, Van Wilgenburg,
Wassenaar, and Larson (2012)
Beetle spp δ2H 5 33.2 1 1.6δ2Hp Gröcke, Schimmelmann, Elias, and Miller (2006)
δ H 5 34.7 1 1.4δ Hp
2 2
Gröcke et al. (2006)
Hoverfly δ H 5 225.2 1 1.04δ Hp
2 2
Quin et al. (2011)
True army worm δ H 5 295 1 0.42δ Hp
2 2
Hobson et al. (2018)
a
NGF, Neotropical migrant ground forager; NO, Neotropical migrant other; RGF, resident ground forager; RO, resident other; SDGF, short-distance
migrant ground forager; SDO, short-distance migrant other.
b
Note that δ2Hp can refer to (1) amount-weighted mean growing-season precipitation, (2) amount-weighted mean annual precipitation, or (3) laboratory waters.
c
See Chapter 5, Tracking of Movements of Terrestrial Mammals Using Stable Isotopes, for a more complete list.
migrant birds breeding in Europe molt their southern part of the African continent and the
feathers in winter. The hydrogen isotope base- extremely enriched region in the northeast,
map for Africa shows dramatic changes season- centered on Sudan and Ethiopia. In an investi-
ally. An interesting and potentially very useful gation into potential wintering sites of the
feature is the more depleted values in the endangered aquatic warbler, Pain et al. (2004)

did not find δ2H measurements to be particu- and the underlying variance associated with the
larly useful and instead advocated the use of precipitation surface. This is because we know
δ15N and δ13C measurements to better define that such variance changes across isoscapes.
wintering areas in Africa. On the other hand, Fortunately, variance surfaces for amount
Yohannes, Hobson, Pearson, Wassenaar, and weighted, mean annual or growing season δ2H
Biebach (2005), Yohannes, Hobson, and Pearson and δ18O precipitation are now available
(2007) investigated δ2H together with δ15N and (Terzer, Wassenaar, Araguás-Araguás, &
δ13C measurements in feathers of several Aggarwal, 2013) and analytical refinements for
migrant passerines moving through East Africa assignment purposes are expected (Chapter 8:
where some of them stop to molt en route Design and Analysis for Isotope-Based Studies
to more southern wintering areas. A recent of Migratory Animals). Regardless, we are
analysis of feathers from museum collections quickly approaching the limits of refinement in
revealed a generally poor correlation with our approach to using stable isotopes to assign
expected precipitation δ2H (Gutiérrez-Expósito, individuals and populations to origins using a
Ramı́rez, Afán, Forero, & Hobson, 2015). single (primarily δ2H) isotope. It should also be
Most approaches to date propagate known recognized that conclusions derived using the
error associated with derived calibrations link- isotope approach have been inherently conser-
ing mean annual or growing season precipita- vative, examining continent wide patterns
tion with animal tissues using the average of of migratory connectivity (Hobson, Van
the (tissue) residuals of the calibration. An Wilgenburg, Faaborg, et al., 2014), origins of
important component of deriving calibration harvested populations, derivation of migratory
relationships linking animal tissues with divides, and so on. More spatially sensitive
underlying isoscapes is that a measure of site- questions regarding animal origins will require
specific isotopic variance in animal tissues can greater sophistication and almost certainly will
be estimated from the residuals of such regres- require the addition of additional isotopic and
sions. Hobson, Van Wilgenburg, Wassenaar, nonisotopic information, and this will be an
and Larson (2012), Hobson, Van Wilgenburg, area of great future interest.
Faaborg, et al. (2014) documented the standard
deviations of the residuals found for their cali-
bration work linking feather δ2H values with
4.4.3 Incorporating Multiple Sources
expected mean growing season precipitation
δ2H (δ2Hp). The overall model indicated a
of Information
residual SD of B13m. Specific migratory and As reviewed in Chapter 1, Animal
foraging guild residual values varied from Migration: A Context for Using New
B18m in short-distance migrant ground fora- Techniques and Approaches, the tools for
gers to B10m for nonground foraging investigating animal origins and movements
Neotropical migrants. Importantly, these esti- has never been so diverse or powerful. It helps
mates of variance can be propagated in assign- to consider isotopic applications as part of a
ment of individuals to origin (Chapter 8: much larger (forensic) initiative. Chapter 8,
Design and Analysis for Isotope-Based Studies Design and Analysis for Isotope-Based Studies
of Migratory Animals). That approach provides of Migratory Animals, outlines in detail
a realistic estimate of error, but a better aspects of sample design and analysis. For
approach will be to use spatially explicit variance example, several researchers have used multi-
associated with both the isotope values of tis- ple isotopes (δ2H, δ34S, δ13C, δ15N, and δ87Sr)
sues grown at a site (i.e., within population) in a multivariate approach to examine origins

of migratory populations (Caccamise, Reed, inappropriately (Hobson, Van Wilgenburg,

Casteli, Wainright, & Nichols, 2000; Hebert & Faaborg, et al., 2014; Kery & Schaub, 2012). A
Wassenaar, 2005; Yohannes et al., 2007), but recent area of great promise is combining spa-
adding more isotopes is not necessarily a tial structure associated with genetic markers
guarantee of better spatial resolution. As dis- (i.e., genescapes) and stable isotope markers
cussed, the formal incorporation of more than (Boulet, Gibbs, & Hobson, 2006; Clegg, Kelly,
one isotope into migratory assignments has Kimura, & Smith, 2003; Kelly, Ruegg, & Smith,
rarely been used. To date, this has been 2005; Paxton, Yau, Moore, & Irwin, 2013).
accomplished primarily via the use of multi- Chabot, Hobson, Van Wilgenburg, & McQuat,
variate normal distributions whereby more 2012 and Rundell et al. (2013) approached this
than one (assumed largely orthogonal) iso- formally by integrating genetic admixture coef-
scape has been applied. In addition to the ficients as prior information to increase the
familiar precipitation isoscapes involving pri- power of isotopic assignment using Bayesian-
marily δ2H measurements, other isoscapes based models. The efficacy of this approach
include those based on modeled plant C3 ver- will depend on the species in question and
sus C4 δ13C distributions (Still & Powell, how well the genescape has been described
2010) and on soil δ15N predictions (Craine for a species. Conveniently, material for
et al., 2009). Hobson, Van Wilgenburg, stable isotope and genetic analyses is often
Wassenaar, and Larson (2012) used that available from the same sample (e.g., feather tip
approach to create expected δ2H, δ13C, and vs base). Ruegg et al. (2017) presented a formal
δ15N feather isoscapes and proceeded to means of combining stable isotope, genetics
query surfaces simultaneously to identify dis- and habitat suitability models for assignment
tinct isotopic clusters to assign European purposes. Rushing, Ryder, Saracco, and
breeding birds to potential origins in Africa. Marra (2014) demonstrated for Wood Thrush
That approach initially involved binning indi- (Hylocichla mustelina) how combining morpho-
viduals to clusters (see also Lopez-Calderon metric and isotope data in assignment models
et al., 2017) but later, spatially explicit assign- can significantly improve assignment accuracy.
ments were used (Hobson, Van Wilgenburg, Other means of constraining origins using
Wesolowski, et al., 2014; Veen et al., 2014). stable isotope approaches involve the obvious
The same multiisotope approach was used to step of delimiting potential origins by known
assign migratory barn swallows (Hirundo breeding and wintering distributions before
rustica) to molt origins in South America performing assignments. In addition, some
(Hobson & Kardynal, 2016). stable isotopes provide important threshold
The use of prior information in Bayesian information to limit samples to only those
assignment models underscores one of the key regions where the precipitation isoscape is
advantages in this approach. Priors used in valid (namely, terrestrial systems). Here, the
assigning individuals include expected popu- use of δ15N, δ13C, and δ34S measurements can
lation density distributions on the breeding potentially reveal marine inputs to diets which
grounds (Royle & Rubenstein, 2004) and the would deter the use of δ2H or δ18O to assign
use of movement recovery vectors to constrain individuals to terrestrial precipitation iso-
potential origins or destinations of migrants scapes (Lott, Meehan, & Heath, 2003; Table 4.1,
(Gunnarsson et al., 2012; Van Wilgenburg & and Fig. 4.5, see also Hobson & Kardynal,
Hobson, 2011). However, such applications are 2016). Other authors have used this threshold
not necessarily straightforward, and the approach to separate origins of birds also
Bayesian prior can skew migratory origins using agricultural versus more natural biomes

4.5 CHALLENGES 105
140 standardized analytical approaches. Beyond
issues related to measurement error, there is
100 Coastal bird eaters the ongoing issue of the appropriate precipita-
tion to tissue calibration relationship to use to
create spatially explicit isoscapes. We now
60
δDf−p (per mil)
realize that such calibration algorithms are

sensitive to factors related to both animal
20 physiology, diet, and ecological guild in addi-
Inland bird eaters tion to the abiotic factors driving the underly-
ing elemental isoscapes (Hobson, Van
–20
Wilgenburg, Wassenaar, & Larson, 2012).
Hydrogen is an element that will exchange
–60 with weak OH or NH bonds and this can
Coastal
Inland generalists generalists take place with drinking water and overall
–100 body water (Hobson et al., 1999). Body δ2H
–15 –10 –5 0 5 10 15 20 values can increase from heat stress
δ34S (per mil) (McKechnie, Wolf, & Martinez del Rio, 2004)
FIGURE 4.5 The relationship between the difference and presumably as a function of work or high
between feather δ2H and predicted precipitation δ2H and metabolism that results in increased body
feather δ34S for nine species of raptors breeding in North evapotranspiration. Powell and Hobson (2006)
America (from Lott et al., 2003). This figure illustrates the found that Wood Thrush (H. mustelina) grow-
way in which birds having access to marine protein can be
ing feathers in Georgia had higher feather δ2H
distinguished by their high δ2H and δ34S values.
values than expected from the feather δ2H iso-
scape, and speculated that heat stress during
(Schmaltz et al., 2017; Yerkes, Hobson, molt may have been a factor. While we assume
Wassenaar, Macleod, & Coluccy, 2008). that trophic 2H discrimination effects are
minor and that most of the precipitation to
tissue discrimination occurs between precipita-
4.5 CHALLENGES tion and plants, this requires further investiga-
tion (Birchall, O’Connel, Heaton, & Hedges,
The development of isotopic techniques to 2005). Notably, it is possible that feathers of
track animal movement represents a huge birds grown in the nest differ from those of
breakthrough in the way we approach a suite the adults feeding them due to differences in
of practical and theoretical issues associated metabolism, drinking water, thermal regime,
with the ecology and conservation of migra- feather growth rate, and diet (Haché, Hobson,
tory species. However, despite excellent Bayne, Van Wilgenburg, & Villard, 2014;
advances, there are significant challenges that Langin et al., 2007; Studds et al., 2012).
must be addressed. Most of these apply to the Another concern for users of the hydrogen
use of δ2H measurements and related iso- isotope base maps for the various continents is
scapes. As noted in Chapter 2, Introduction the variability in the IAEA GNIP dataset for
to Conducting Stable Isotope Measurements any given year where organisms are sampled
for Animal Migration Studies, the adoption of (Chapter 3: Isoscapes for Terrestrial Migration
robust measurement protocols for δ2H (and to Research, Vander Zanden et al., 2015). As
some extent δ18O) in animal tissues are demonstrated with the monarch study,
paramount and practitioners must adopt the only foolproof way to avoid impact of

long-term variance is to create a base map for on the extent of evapotranspiration from water-
the year of interest, but this is beyond the bodies. However, waterfowl feather δ2H values
scope of most researchers for most organisms. follow closely the expected growing-season
Second, if a study site is close to one of the average value for at least the temperate region
IAEA GNIP sampling stations, then it may be of North America (Clark et al., 2006), but the
possible for the researcher to potentially derive potential for regional departures in shallow
a year-specific tissue value for the site of inter- wetland systems requires careful consideration
est but this would only provide local informa- (Coulton, Clark, Hobson, Wassenaar, & Hebert,
tion most appropriate for discerning local from 2009).
immigrant individuals (Hobson, 2005). More The early years of development for the
realistically, if researchers can obtain animal assignment of animals to origins using
tissues from known individuals grown in the stable isotopes revealed problem areas. For
year of interest that could act as a reasonable birds, much attention focused on applications
proxy for local integrated isotope values (Van involving shorebirds (Farmer, Cade, & Torres-
Wilgenburg, Hobson, Brewster, & Welker, Dowdall, 2008; Rocque, Ben-David, Barry, &
2012). Regardless, the effect of sampling year Winker, 2006) and raptors (Smith et al., 2009).
on the accuracy of assignment will vary spa- Apart from issues involving fundamental mis-
tially and by the degree to which climate prox- understanding of isotopic applications to trac-
ies such as ENSO operate in any given system ing bird origins (Larson & Hobson, 2009)
or by the degree to which food webs are sup- shorebirds in general can be problematic
ported by episodic (e.g., monsoons) versus because the origin and period of molt in adults
multiple precipitation events (Ehleringer, is often poorly known on and off the breeding
Phillips, Schuster, & Sandquist, 1991; grounds. Moreover, this group consists of
Rozanski, Araguas-Araguas, & Gonfiantini, birds that use a vast array of habitats during
1993; Villacis, Vimeux & Taupin, 2008). Our their breeding, migration, and wintering life
general poor understanding of which rainfall stages. These habitats range from terrestrial
matters in food web H flow persists. The good upland, freshwater, brackish, and marine and
relationship obtained between feather δ2H and so represent a formidable array of isotopic
mean annual growing season δ2H in North endpoints. Several species also move between
America (Hobson, Møller, et al., 2012; Hobson, temperate regions across hemispheres and this
Van Wilgenburg, Wassenaar, & Larson, 2012; can lead to nondiagnostic feather δ2H values.
Hobson, Van Wilgenburg, Wassenaar, Powell, In the case of raptors, apart from early issues
et al., 2012; Hobson & Wassenaar, 1997) was which dealt with either methodology or a
for forest birds distributed through the central lack of understanding (Wunder et al., 2009),
region of the continent. Closed-canopy forest evidence suggests that assigning adult raptors
with shallow root systems may well integrate to feather isoscapes remains particularly
food web δ2H available to birds and other ani- challenging.
mals over such long periods. However, is this For birds, molt patterns are reasonably
the case for more pulsed ecosystems like grass- well known for most species. However,
lands or deserts? In other riparian systems, stable isotope measurements themselves have
snowmelt may have the greatest influence on provided important qualifiers. The molt of flight
local food web δ2H. In other systems where feathers of northern populations of the logger-
animals may be influenced by aquatic emer- head shrike (Lanius ludovicianus) are essentially
gent insects, tissues grown later in the season bimodal with inner primaries, secondaries,
may differ from those grown earlier depending and tail feathers usually being molted on the

4.6 SUMMARY 107
breeding grounds but other feathers being analyses to delineate C3 versus C4 or CAM
grown on the wintering grounds following a food webs or the use of stable hydrogen
suspension in molt during migration (Chabot, isotope analyses to further separate C4 and
Hobson, Craig, & Lougheed, 2017; Perez & CAM pathways provide distinct advantages.
Hobson, 2007). Other birds undergo prealternate Terrestrial organisms that also spend part of
molt of some body feathers on the wintering their lives in marine or estuarine situations
grounds prior to migration allowing us to inves- lend themselves to isotopic tracking using sev-
tigate aspects of winter origins or habitat use eral elements. Altitudinal migrants constrained
(Mazerolle & Hobson, 2005; Mehl, Alisauskas, by latitude and longitude also represent a use-
Hobson, & Kellett, 2004). Unfortunately, infor- ful application of δ2H and δ18O measurements
mation on the reliability or extent of prealternate providing the movement represents at least
molt or on the extent of delayed molt in migrat- several hundred meters. We will have more
ing birds is often not available (Hobson, Brua, trouble in cases where underlying isotopic gra-
Hohman, & Wassenaar, 2000). Another alterna- dients are less distinct or where alternative
tive is to use claws that are continuously grow- origins overlap isotopically.
ing. Birds captured soon after their arrival on The application of deuterium measurements
the breeding grounds should have claws that in animal tissues to place them on continental
have retained information from the wintering isoscapes undoubtedly has provided the single
grounds (Bearhop, Hilton, Votier, & Waldron, greatest impact in the field of isotopic tracking.
2004; Mazerolle & Hobson, 2005). While we Again, the success of the H isotope approach
need more controlled studies to establish will depend very much on which part of the
growth rates of claws for a variety of species, isoscape we are dealing with. Distinguishing
contrasting stable isotope values of claws between Arctic and prairie origins of migratory
against a metabolically active tissue like blood birds in North America or between those from
can in fact provide insight into periods where Scandinavia or Spain in Europe will be rela-
these tissues “agree” isotopically. tively straightforward. We are faced with more
of a challenge in distinguishing between birds
or other animals originating across latitudinal
4.6 SUMMARY bands on both continents or from regions that
are more spatially restricted (e.g., Szymanski,
This chapter suggested that situations Afton, & Hobson, 2006; Coulton et al., 2009).
where the three principles needed for success- So, while this chapter has shown that the iso-
ful isotopic tracking of migratory animals are tope approach has provided an exciting new
met will be rare. The degree to which research- tool to researchers and conservationists, it is
ers are successful in applying isotopic methods not a “silver bullet” to be applied without full
will depend on the organism of interest, its recognition of the inherent limitations. How
geographical range, and ecophysiology. then might the field proceed from here? We
Applications also fundamentally depend on now realize that the application of a single pre-
how well we know the nature and behavior of cipitation to tissue δ2H calibration algorithm
the appropriate isoscapes. The most elegant across diverse species or particular geographic
applications will usually be situations where regions will inherently involve error, hence a
alternative isoscapes are different and species sensitivity analysis should be part of the
experience simple isotopically dichotomous approach in future studies (Hobson, Van
situations during their travels. Here, the long- Wilgenburg, Wassenaar, Moore, & Farrington,
standing success in using stable carbon isotope 2007). In addition, we should realize the risks

in placing all of our eggs in one isotopic basket PLoS One, 10(11), e0141371. Available from https://doi.
(i.e., explore complimentary tools). org/10.1371/journal.pone.0141371.
Ambrose, S. H., & DeNiro, M. J. (1986). The isotopic ecol-
This chapter has dealt primarily with avian ogy of East African mammals. Oecologia, 69, 395406.
applications, the next chapter focuses on terres- Arneson, L. S., & MacAvoy, S. E. (2005). Carbon, nitrogen,
trial mammals followed by a chapter on track- and sulfur diet-tissue discrimination in mouse tissues.
ing marine animals. Chapter 7, Amino Acid Canadian Journal of Zoology, 83, 989.
Isotope Analysis: A New Frontier in Studies of Ashley, P., Hobson, K. A., Van Wilgenburg, S. L., North,
N., & Petrie, S. (2010). Linking Canadian harvested
Animal Migration and Foraging Ecology, juvenile American Black Ducks to their natal areas
provides an overview of the intriguing develop- using stable isotope (δD, δ13C, and δ15N) methods.
ments now taking place using compound- Avian Conservation and Ecology, 5(2), 7. URL: ,http://
specific methods and Chapter 8, Design and www.ace-eco.org/vol5/iss2/art7/..
Analysis for Isotope-Based Studies of Migratory Atkinson, P. W., Baker, A. J., Bevan, R. M., Clark, N. A.,
Cole, K. B., Gonzalez, P. M., . . . Robinson, R. A. (2005).
Animals, deals with the important and dynamic Unravelling the migration and moult strategies of long-
area of study design and statistical assignment distance migrant using stable isotopes: Red Knot
using stable isotope and other data. Chapter 9, Calidris canutus, movements in the Americas. Ibis, 147,
Isoscape Computation and Inference of Spatial 738749.
Origins With Mixed Models Using the R pack- Barnett-Johnson, R., Pearson, T. E., Ramos, F. C., Grimes,
C. B., & MacFarlane, R. B. (2008). Tracking natal origins
age IsoriX, presents one of the several software of salmon using isotopes, otoliths, and landscape geol-
packages for assignment currently available. By ogy. Limnology and Oceanography, 53(4), 16331642.
this point, the reader should be encouraged by Bataille, C. P., et al. (2012). Mapping multiple source effects
the breadth of past isotopic applications to on the strontium isotopic signatures of ecosystems from
tracking migratory animals and realize the tre- the circum-Caribbean region. Ecosphere, 3(12), art118.
Bataille, C. P., & Bowen, G. J. (2012). Mapping 87Sr/86Sr
mendous scope for future developments. The variations in bedrock and water for large scale prove-
need for caution and consideration of the nance studies. Chemical Geology, 304305, 3952.
numerous assumptions involved are sobering Beard, B. L., & Johnson, C. M. (2000). Strontium isotope
(this chapter and Chapter 8: Design and composition of skeletal material can determine the birth
Analysis for Isotope-Based Studies of Migratory place and geographic mobility of humans and animals.
Journal of Forensic Sciences, 45, 10491061.
Animals). Nonetheless, more and more we are Bearhop, S., Furness, R. W., Hilton, G. M., Votier, S. C., &
coming to terms with the nature of isotopic var- Waldron, S. (2003). A forensic approach to understand-
iance in the natural world and Chapter 10, ing diet and habitat use from stable isotope analysis of
Outlook for Using Stable Isotopes in Animal (avian) claw material. Functional Ecology, 17, 270275.
Migration Studies, provides a summary of Bearhop, S., Hilton, G. M., Votier, S. C., & Waldron, S.
(2004). Stable isotope ratios indicate that body condition
thoughts about the path ahead. in migrating passerines is influenced by winter habitat.
Proceedings of the Royal Society of London (Series B), 271,
215218.
Acknowledgment Bensch, S., Bengtsson, G., & Åkesson, S. (2006). Patterns of
Len Wassenaar provided valuable edits on an earlier stable isotope signatures in willow warbler Phylloscopus
version of this chapter. trochilus feathers collected in Africa. Journal of Avian
Biology, 37, 323330.
Bearhop, S., Thompson, D. R., Waldron, S., Russell, I. C.,
Alexander, G., & Furness, R. W. (1999). Stable isotopes
References indicate the extent of freshwater feeding by cormorants
Altizer, S., Hobson, K. A., Davis, A. K., De Roode, J. C., & Phalacrocorax carbo from inland fisheries in England.
Wassenaar, L. I. (2015). Do healthy monarchs migrate Journal of Applied Ecology, 36, 7584.
farther? Tracking natal origins of parasitized vs. unin- Best, P. B., & Schell, D. M. (1996). Stable isotopes in south-
fected monarch butterflies overwintering in Mexico. ern right whale (Eubalaena australis) baleen as indicators

REFERENCES 109
of seasonal movements, feeding and growth. Marine δ34S) composition of feathers and claws from lesser sca-
Biology, 124, 483494. up and northern pintail: implications for studies of
Birchall, J., O’Connel, T. C., Heaton, T. H. E., & Hedges, migratory connectivity. Canadian Journal of Zoology, 84,
R. E. M. (2005). Hydrogen isotope ratios in animal body 13951401.
protein reflect trophic level. Journal of Animal Ecology, Clegg, S. M., Kelly, J. F., Kimura, M., & Smith, T. B. (2003).
74, 877881. Combining genetic markers and stable isotopes to
Boecklen, W. J., Yarnes, C. T., Cook, B. a, & James, A. C. reveal population connectivity and migration pattern in
(2011). On the use of stable isotopes in trophic ecology. a Neotropical migrant, Wilson’s Warbler (Wilsonia pusil-
Annual Review of Ecology, Evolution, and Systematics, 42, la). Molecular Ecology, 12, 819830.
411440. Coulton, D. W., Clark, R. G., Hobson, K. A., Wassenaar,
Bosley, K. L., Witting, D. A., Chambers, R. C., & Wainright, L. I., & Hebert, C. (2009). Temporal sources of deute-
S. C. (2002). Estimating turnover rates of carbon and rium (δD) in waterfowl feathers across a prairie-to-
nitrogen in recently metamorphosed winter flounder boreal gradient. Condor, 111, 255265.
Psuedopleuronectes americana with stable isotopes. Marine Coutaud, A., Meheut, M., Viers, J., Rols, J. L., &
Ecology Progress Series, 236, 233240. Pokrovsky, O. S. (2014). Zn isotope fractionation during
Boulet, M., Gibbs, H. L., & Hobson, K. A. (2006). interaction with phototrophic biofilm. Chemical Geology,
Integrated analysis of genetic, stable isotope, and band- 390, 4660.
ing data reveal migratory connectivity and flyways in Craine, J. M., et al. (2009). Global patterns of foliar nitrogen
the northern Yellow Warbler (Dendroica petechia; Aestiva isotopes and their relationships with climate, mycor-
group). Ornithological Monographs, 61, 2978. rhizal fungi, foliar nutrient concentrations, and nitrogen
Caccamise, D. F., Reed, L. M., Casteli, P. M., Wainright, availability. New Phytologist, 183, 980992.
S. C., & Nichols, T. C. (2000). Distinguishing migratory Croal, L. R., Johnson, C. M., Beard, B. L., & Newman, D. K.
and resident Canada geese using stable isotope analy- (2004). Iron isotope fractionation by Fe (II)-oxidizing
sis. Journal of Wildlife Management, 64, 10841091. photoautotrophic bacteria. Geochimica et Cosmochimica
Carleton, S. A., Kelly, L., Anderson-Sprecher, R., & Acta, 68, 12271242.
Martı́nez del Rio, C. (2008). Should we use one-, or Cryan, P. M., Bogan, M. A., Rye, R. O., Landis, G. P., &
multi-compartment models to describe 13C incorpo- Kester, C. L. (2004). Stable hydrogen isotope analysis of
ration into animal tissues? Rapid Communications in bat hair as evidence for seasonal molt and long-distance
Mass Spectrometry, 22, 30083014. migration. Journal of Mammalogy, 85, 9951001.
Caut, S., Angulo, E., & Courchamp, F. (2009). Variation in Cryan, P. M., Stricker, C. A., & Wunder, M. B. (2014).
discrimination factors (Δ15N and Δ13C): The effect of Continental scale, seasonal movements of a heterother-
diet isotopic values and applications for diet recon- mic migratory tree bat. Ecological Applications, 24,
struction. Journal of Applied Ecology, 46, 443453. 602616.
Cerling, T. E., Bowen, G., Ehleringer, J. R., & Sponheimer, Dalerum, F., & Angerbjörn, A. (2005). Resolving temporal
M. (2007). The reaction progress variable and isotope variation in vertebrate diets using naturally occurring
turnover in biological systems. In T. E. Dawson, & stable isotopes. Oecologia, 144, 647658.
R. T. W. Siegwolf (Eds.), Stable isotopes as indicators of Doucett, R. R., Marks, J. C., Blinn, D. W., Caron, M., &
ecological change (pp. 163171). London: Academic Hungate, B. A. (2007). Measuring terrestrial subsidies
Press. to aquatic foodwebs using stable-isotopes of hydrogen.
Chabot, A., Hobson, K. A., Craig, S., & Lougheed, S. Ecology, 88, 15871592.
(2017). Molt in Loggerhead Shrike (Lanius ludovicianus) Ehleringer, J. R., Phillips, S. L., Schuster, W. S. F., &
is influenced by sex, latitude and migration. Ibis, 160, Sandquist, D. R. (1991). Differential utilization of sum-
301312. mer rains by desert plants. Oecologia, 88, 430434.
Chabot, A. M., Hobson, K. A., Van Wilgenburg, S. L., & Farmer, A., Cade, B. S., & Torres-Dowdall, J. (2008).
McQuat, G. J. (2012). Assigning breeding ground ori- Fundamental limits to the accuracy of deuterium iso-
gins to migrant birds: Advances using a novel Bayesian topes for identifying the spatial origin of migratory ani-
approach combining genetic and stable isotope mar- mals. Oecologia, 158, 183192.
kers. PLoS One, 7(8), e43627. Available from https:// Finkelstein, M. E., George, D., Scherbinski, S., Gwiazda, R.,
doi.org/10.1371/journal.pone.0043627. Johnson, M., Burnett, J., . . . Smith, D. R. (2010). Feather
Clark, I. D., & Fritz, P. (1997). Environmental isotopes in lead concentrations and (207)Pb/(206) Pb ratios reveal
hydrogeology. New York: Lewis Publishers. lead exposure history of California Condors
Clark, R. G., Hobson, K. A., & Wassenaar, L. I. (2006). (Gymnogyps californianus). Environmental Science &
Geographic variation in the isotopic (δD, δ13C, δ15N, Technology, 44, 26392647.

Flemming, T. H., Nunez, R. A., & Sternberg, L. S. L. (1993). distance in wild-caught migratory monarch butterflies.
Seasonal changes in the diets of migrant and non- Behavioral Ecology, 24, 11081113.
migrant nectarivorous bats as revealed by carbon Harrington, R. R., Kennedy, B. P., Chamberlain, C. P.,
stable isotope analysis. Oecologia, 94, 7275. Blum, J. D., & Folt, C. L. (1998). 15N enrichment in agri-
Flockhart, D. T., Brower, L. P., Ramirez, M. I., Hobson, cultural catchments: Field patterns and applications to
K. A., Wassenaar, L. I., Altizer, S., & Norris, D. R. tracking Atlantic salmon (Salmo salar). Chemical Geology,
(2017). Regional climate on the breeding grounds pre- 147, 281294.
dicts variation in the natal origin of monarch butterflies Heady, W. N., & Moore, J. W. (2013). Tissue turnover and
overwintering in Mexico over 38 years. Global Change stable isotope clocks to quantify resource shifts in anad-
Biology. Available from https://doi.org/10.1111/ romous brown trout. Oecologia, 172, 2134.
gcb.13589. Heaton, T. H. E. (1987). The 15N/14N ratios of plants in
Flockhart, D. T., Dabydeen, A., Satterfield, D., Hobson, South Africa and Namibia: Relationship to climate and
K. A., Wassenaar, L. I., & Norris, D. R. (2017). Patterns coastal/saline environments. Oecologia, 74, 236246.
of parasitism in monarch butterflies during the breed- Hebert, C., & Wassenaar, L. I. (2005). Feather
ing season in eastern North America. Ecological stable isotopes in western North American waterfowl:
Entomology. Available from https://doi.org/10.1111/ Spatial patterns, underlying factors, and management
een.12460. implications. Wildlife Society Bulletin, 33, 92102.
Flockhart, D. T., Kyser, T. K., Chipley, D., Miller, N. G., & Henaux, V., Powell, L., Vrtiska, M., & Hobson, K. A.
Norris, D. R. (2015). Experimental evidence shows no (2012). Establishing winter origins of migrating Lesser
fractionation of strontium isotopes (Sr/Sr) among soil, Snow Geese using stable isotopes. Avian Conservation
plants, and herbivores: Implications for tracking wild- and Ecology, 7. Available from http://dx.doi.org/
life and forensic science. Isotopes in Environmental and 10.5751/ACE-00515-070105.
Health Studies, 51, 372381. Hesslein, R. H., Capel, M. J., Fox, D. E., & Hallard, K. A.
Flockhart, D. T., Wassenaar, L. I., Martin, T. G., Hobson, (1991). Stable isotopes of sulphur, carbon, and nitrogen
K. A., Wunder, M. B., & Norris, D. R. (2013). Tracking as indicators of trophic level and fish migration in the
multi-generational colonization of the breeding grounds lower Mackenzie River basin, Canada. Canadian Journal
by monarch butterflies in eastern North America. of Fisheries and Aquatic Science, 48, 22582265.
Proceedings of the Royal Society of London, Series B: Hesslein, R. H., Hallard, K. A., & Ramlal, P. (1993).
Biological Sciences, 280, 1087. Replacement of sulfur, carbon, and nitrogen in tissue of
Gröcke, D. R., Schimmelmann, A., Elias, S., & Miller, R. F. growing broad whitefish (Coregonus nasus) in response
(2006). Stable hydrogen-isotope ratios in beetle chitin: to a change in diet traced by δ34S, δ13C, and δ15N.
preliminary European data and re-interpretation of Canadian. Journal of Fisheries and Aquatic Science, 50,
North American data. Quaternary Science Reviews, 25, 20712076.
18501864. Hobson, K. A. (1999). Stable-carbon and nitrogen isotope
Gunnarsson, G., Latorre-Margalef, N., Hobson, K. A., Van ratios of songbird feathers grown in two terrestrial
Wilgenburg, S. L., Elmberg, J., Olsen, B., & biomes: Implications for evaluating trophic relation-
Waldenström, J. (2012). Disease dynamics and bird ships and breeding origins. Condor, 101, 799805.
migration Linking Mallards Anas platyrhynchos and Hobson, K. A. (2005). Stable isotopes and the determina-
Influenza A virus in time and space. PLoS One, 7(4), tion of avian migratory connectivity and seasonal
e35679. Available from https://doi.org/10.1371/jour- interactions. Auk, 122, 10371048.
nal.pone.0035679. Hobson, K. A. (2008). Applying isotopic methods to tracking
Gutiérrez-Expósito, C., Ramı́rez, F., Afán, I., Forero, M. G., animal movements. In K. A. Hobson, & L. I. Wassenaar
& Hobson, K. A. (2015). A deuterium feather isoscape (Eds.), Tracking animal migration using stable isotopes (1st
for sub-Saharan Africa. PLoS One, 10(9). Available from ed., pp. 4578). London: Academic Press.
https://doi.org/10.1371/journal.pone.0135938. Hobson, K. A., Alisauskas, R. T., & Clark, R. G. (1993).
Haché, S., Hobson, K. A., Bayne, E. M., Van Wilgenburg, Stable-nitrogen isotope enrichment in avian tissues due
S. L., & Villard, M. A. (2014). Tracking natal dispersal to fasting and nutritional stress: implications for isoto-
in a coastal population of a migratory songbird using pic analyses of diet. Condor, 95(1993), 388394.
feather stable isotope (δ2H, δ34S) tracers. PLoS One, 9(4). Hobson, K. A., Blight, L. K., & Arcese, P. (2015). Human-
Available from https://doi.org/10.1371/journal. induced long-term shifts in gull diet from marine to ter-
pone.0094437. restrial sources in North America’s coastal Pacific:
Hanley, D., Miller, N. G., Flockhart, D. T. T., & Norris, More evidence from more isotopes (δ2H, δ34S).
D. R. (2013). Forewing pigmentation predicts migration Environmental Science and Technology, 49, 1083410840.

REFERENCES 111
Hobson, K. A., Brua, R. B., Hohman, W. L., & Wassenaar, Afrotropical migrant birds to origins. Ecosphere, 3(5), 44.
L. I. (2000). Low frequency of “double molt” of remiges Available from https://doi.org/10.1890/ES12-00018.1.
in ruddy ducks revealed by stable isotopes: Hobson, K. A., Van Wilgenburg, S. L., Wesolowski, T.,
Implications for tracking migratory waterfowl. Auk, Maziarz, M., Biljsma, R. G., Grendelmeier, A., & Mallord,
117, 129135. J. W. (2014). A multi-isotope (δ2H, δ13C, δ15N) approach to
Hobson, K. A., Doward, K., Kardynal, K., & McNeil, J. establishing migratory connectivity in Palearctic-
(2018). Inferring origins of migrating insects using iso- Afrotropical migrants: An example using Wood Warblers
scapes: A case study using the true armyworm, Mythimna Phylloscopus sibilatrix. Acta Ornithologica, 49, 5769.
unipuncta, in North America. Ecological Entomology. Hobson, K. A., & Wassenaar, L. I. (1997). Linking breeding
Available from https://doi.org/10.1111/een.12505. and wintering grounds of Neotropical migrant song-
Hobson, K. A., & Kardynal, K. (2016). An isotope (δ34S) fil- birds using stable hydrogen isotopic analysis of feath-
ter and geolocator results constrain a dual feather iso- ers. Oecologia, 109, 142148.
scape (δ2H, δ13C) to identify the wintering grounds of Hobson, K. A., Wassenaar, L. I., & Bayne, E. (2004). Using
North American Barn Swallows. Auk, 133, 8698. isotopic variance to detect long-distance dispersal and
Hobson, K. A., & Koehler, G. (2015). On the use of stable- philopatry in birds: An example with ovenbirds and
oxygen isotope (δ18O) measurements for tracking avian American redstarts. Condor, 106, 732743.
movements in North America. Ecology and Evolution. Hobson, K. A., Wassenaar, L. I., Milá, B., Lovette, I.,
Available from https://doi.org/10.1002/ece3.1383. Dingle, C., & Smith, T. B. (2003). Stable isotopes as indi-
Hobson, K. A., Møller, A., & Van Wilgenburg, S. L. (2012). cators of altitudinal distributions and movements in an
A multi-isotope (δ13C, δ15N, δ2H) approach to connect- Ecuadorean hummingbird community. Oecologia, 136,
ing European breeding and African wintering popula- 302308.
tions of barn swallows (Hirundo rustica). Animal Hobson, K. A., Wassenaar, L. I., & Taylor, O. R. (1999).
Migration. Available from https://doi.org/10.2478/ Stable isotopes (δD and δ13C) are geographic indicators
ami-2012-0002. of natal origins of Monarch butterflies in eastern North
Hobson, K. A., Slater, G. L., Lank, D. B., Milner, R. L., & America. Oecologia, 120, 397404.
Gardiner, R. (2013). Agricultural lands subsidize winter Hobson, K. A., & Yohannes, E. (2007). Establishing elemen-
diet of Pacific Dunlin (Calidris alpine pacifica) at two tal turnover in exercising birds using a wind tunnel:
major estuaries. Condor, 115, 515524. Implications for stable isotope tracking of migrants.
Hobson, K. A., Van Wilgenburg, S. L., Faaborg, J., Toms, Canadian Journal of Zoology, 85, 703708.
J. D., Rengifo, C., Llanes Sosa, A., et al. (2014). Jamieson, R. E., & Wadleigh, M. A. (2000). Tracing sources
Connecting breeding and wintering grounds of of precipitation sulfate in eastern Canada using
Neotropical migrant songbirds using stable hydrogen stable isotopes and trace metals. Journal of Geophysical
isotopes: A call for an isotopic atlas of migratory con- Research, 105, 549556.
nectivity. Journal of Field Ornithology, 85, 237257. Kelly, J. F., Atudorei, V., Sharp, Z. D., & Finch, D. M.
Hobson, K. A., Van Wilgenburg, S. L., Roth, A. M., (2002). Insights into Wilson’s Warbler migration from
Bennett, R. E., Bayly, N. J., Chavarria-Duriaux, L., . . . analyses of hydrogen stable-isotope ratios. Oecologia,
Ritterson, J. D. (2016). Golden-winged Warbler migra- 130, 216221.
tory connectivity derived from stable isotopes. Studies Kelly, J. F., Ruegg, K. C., & Smith, T. B. (2005). Combining
in Avian Biology, 49, 193203. isotopic and genetic markers to identify breeding origins
Hobson, K. A., Van Wilgenburg, S. L., Wassenaar, L. I., & of migrant birds. Ecological Applications, 15, 14871494.
Larson, K. (2012). Linking hydrogen (δ2H) isotopes in Kennedy, B. P., Folt, C. L., Blum, J. D., & Chamberlain,
feathers and precipitation: Sources of variance and con- C. P. (1997). Natural isotope markers in salmon. Nature,
sequences for assignment to global isoscapes. PLoS One, 387, 766.
7(4), e35137. Available from https://doi.org/10.1371/ Kery, M., & Schaub, M. (2012). Bayesian population analysis
journal.pone.0035137. using WinBUGS: A hierarchical perspective. London:
Hobson, K. A., Van Wilgenburg, S., Wassenaar, L. I., Academic Press.
Moore, L. I., & Farrington, J. (2007). Estimating origins Killingly, J. S. (1980). Migrations of California gray whales
of three species of Neotropical migrants at a Gulf coast tracked by oxyen-18 variations in their epizoic barna-
stopover site: Combining stable isotope and GIS tools. cles. Science, 207, 759760.
Condor, 109, 256267. Killingly, J. S., & Lutcavage, M. (1983). Loggerhead turtle
Hobson, K. A., Van Wilgenburg, S. L., Wassenaar, L. I., movements reconstructed from 18O and 13C profiles
Powell, R. L., Still, C. J., & Craine, J. M. (2012). A multi- from commensal barnacle shells. Estuarine and Coastal
isotope (δ13C, δ15N, δ2H) feather isoscape to assign Shelf Science, 16, 345349.

Klaassen, M., Piersma, T., Korthals, H., Dekinga, A., & Mazerolle, D., & Hobson, K. A. (2005). Estimating origins
Dietz, M. W. (2010). Single-point isotope measurements of short-distance migrant songbirds in North America:
in blood cells and plasma to estimate the time since Contrasting inferences from hydrogen isotope measure-
diet switches. Functional Ecology, 24, 796804. ments of feathers, claws, and blood. Condor, 107,
Koch, P. L., Heisinger, J., Moss, C., Carlson, R. W., Fogel, 280288.
M. L., & Behrensmeyer, A. K. (1995). Isotopic tracking McCue, M.D. &Welch, K.C., Jr. (2016). 13C Breath-testing in
of change in diet and habitat use in African elephants. animals: Theory, applications and future directions.
Science, 267, 13401343. Journal of Comparative Physiology B, 186, 265-285.
Komarek, M., Ettler, V., Chrastny, V., & Mihaljevic, M. McKechnie, A. E., Wolf, B. O., & Martinez del Rio, C.
(2008). Lead isotopes in environmental sciences: A (2004). Deuterium stable isotope ratios as tracers of
review. Environment International, 34, 562577. water resource use: An experimental test with rock
Krouse, H. R., Stewart, J. W. B., & Grinenko, V. A. (1991). dove. Oecologia, 140, 191200.
Pedosphere and biosphere. In H. R. Krouse, & V. A. Mehl, K. R., Alisauskas, R. T., Hobson, K. A., & Kellett,
Grinenko (Eds.), Stable isotopes: Natural and anthropo- D. K. (2004). To winter east or west? Heterogeneity in
genic sulphur in the environment (pp. 267306). Toronto, winter site philopatry in a central Arctic population of
ON: John Wiley and Sons. King Eiders. Condor, 106, 241251.
Langin, K. M., Reudink, M. W., Marra, P. R., Norris, D. R., Meyer-Rochow, V. B., Cook, I., & Hendy, H. (1992). How
Kyser, T. K., & Ratcliffe, L. M. (2007). Hydrogen isoto- to obtain clues from the otoliths of an adult fish about
pic variation in migratory bird tissues of known origin: the aquatic environment it has been in as a larvae.
Implications for geographic assignment. Oecologia, 152, Comparative Biochemistry and Physiology, 103A,
449457. 333335.
Larson, K., & Hobson, K. A. (2009). Assignment to breed- Miller, N. G., Wassenaar, L. I., Hobson, K. A., & Norris,
ing and wintering grounds using stable isotopes: A D. R. (2012). Migratory connectivity of the monarch
comment on lessons learned by Rocque et al. Journal of butterfly (Danaus plexippus): Patterns of spring re-
Ornithology, 150, 709712. colonization in eastern North America. PLoS One, 7(3),
Layman, C. A., Araujo, M. S., Boucek, R., Hammerschlag- e31891. Available from https://doi.org/10.1371/jour-
Peyer, C. M., Harrison, E., Jud, Z. R., . . . Bearhop, S. nal.pone.0031891.
(2012). Applying stable isotopes to examine food-web Mizutani, H., Fukuda, M., Kabaya, Y., & Wada, E. (1990).
structure: An overview of analytical tools. Biological Carbon isotope ratio of feathers reveals feeding behav-
Reviews, 87, 545562. ior of cormorants. Auk, 107, 400403.
Lopez-Calderon, C., Hobson, K. A., Marzal, A., Balbontin, Møller, A. P., & Hobson, K. A. (2004). Heterogeneity in
J., Reviriego, M., Magallanes, S., . . . Moller, A. P. (2017). stable isotope profiles predicts coexistence of two popu-
Wintering areas predict age-related breeding phenology lations of barn swallows Hirundo rustica differing in
in a migratory passerine bird. Journal of Avian Biology, morphology and reproductive performance. Proceedings
48, 001009. of the Royal Society of London, 271, 13551362.
Losey, J. E., Rayor, L. S., & Carter, M. E. (1999). Transgenic Nelson, C. S., Northcote, T. G., & Hendy, C. H. (1989).
pollen harms monarch larvae. Nature, 399, 214. Potential use of oxygen and carbon isotopic composi-
Lott, C. A., Meehan, T. D., & Heath, J. A. (2003). tion of otoliths to identify migratory and non-migratory
Estimating the latitudinal origins of migratory birds stocks of the New Zealand common smelt: A pilot
using hydrogen and sulfur stable isotopes in feathers: study. New Zealand Journal of Marine and Freshwater
Influence of marine prey base. Oecologia, 134, Research, 23, 337344.
505510. Norris, D. R., Marra, P. P., Bowen, G. J., Ratcliffe, L. M.,
Lott, C. A., & Smith, J. P. (2006). A GIS approach to Royle, J. A., & Kyser, K. T. (2006). Migratory connectiv-
estimating the origins of migratory raptors in North ity of a widely distributed songbird, The American
America using hydrogen stable isotope ratios in feathers. Redstart (Setophaga ruticilla). Ornithological Monographs,
Auk, 118, 1623. 61, 1428.
Marra, P. P., Hobson, K. A., & Holmes, R. T. (1998). Norris, D. R., Marra, P. P., Kyser, T. K., Sherry, T. W., &
Linking winter and summer events in a migratory bird Ratcliffe, L. M. (2004). Tropical winter habitat limits
using stable carbon isotopes. Science, 282, 18841886. reproductive success on the temperate breeding
Martı́nez del Rio, C., & Carleton, S. A. (2012). How fast grounds in a migratory bird. Proceedings of the Royal
and how faithful: The dynamics of isotopic incorpo- Society of London Series B, 271, 5964.
ration into animal tissues. Journal of Mammalogy, 93, Ostrom, P. H., Colunga-Garcia, M., & Gage, S. H. (1997).
353359. Establishing pathways of energy flow for insect

REFERENCES 113
predators using stable isotope ratios: Field and labora- Quin, A., Menozzi, P., Coulon, M., Hamilton, A. J.,
tory evidence. Oecologia, 109, 108113. Sarthou, J. P., Tsafack, N., . . . Ponsard, S. (2011). Can
Pain, D. J., Green, R. E., Giessing, B., Kozulin, A., Poluda, deuterium stable isotope values be used to assign the
A., Ottosson, U., . . . Hilton, G. M. (2004). Using geographic origin of an auxiliary hoverfly in south-
stable isotopes to investigate migratory connectivity of western France? Rapid Communications in Mass
the globally threatened aquatic warbler Acrocephalus Spectrometry, 25, 27932798.
paludicola. Oecologia, 138, 168174. Richards, M. P., Fuller, B. T., Sponheimer, M., Robinson,
Pardo, L. H., & Nadelhoffer, K. J. (2010). Using nitrogen T., & Ayliffe, L. (2003). Sulphur isotopes in palaeodie-
isotope ratios to assess terrestrial ecosystems at regional tary studies: A review and results from a controlled
and global scales. In J. B. West, G. J. Bowen, T. E. feeding experiment. International Journal of
Dawson, & K. P. Tu (Eds.), Isoscapes: Understanding Osteoarchaeology, 13, 3745.
movements, pattern and process on Earth through isotope Robbins, L. C. T., Felicetti, A., & Sponheimer, M. (2005).
mapping (pp. 221250). New York, NY: Springer. The effect of dietary protein quality on nitrogen isotope
Paxton, K. L., Yau, M., Moore, F. R., & Irwin, D. E. (2013). discrimination in mammals and birds. Oecologia, 144,
Differential migratory timing of western populations of 534540.
Wilson’s warbler (Cardellina pusilla) revealed by mito- Rocque, D. A., Ben-David, M., Barry, R. P., & Winker, K.
chondrial DNA and stable isotopes. Auk, 130, 689698. (2006). Assigning birds to wintering and breeding
Pekarsky, S., Angert, A., Haese, B., Werner, M., Hobson, grounds using stable isotopes: Lessons from two
K. A., & Nathan, R. (2015). Enriching the isotopic tool- feather generations among three intercontinental
box for migratory connectivity analysis: A new migrants. Journal of Ornithology, 147, 395404.
approach for migratory species breeding in remote or Royle, J. A., & Rubenstein, D. R. (2004). The role of species
unexplored areas. Diversity and Distributions. Available abundance in determining breeding origins of migra-
from https://doi.org/10.1111/ddi.12306. tory birds with stable isotopes. Ecological Applications,
Perez, G., & Hobson, K. A. (2007). Feather deuterium mea- 14, 17801788.
surements reveal origins of migratory western Rozanski, K., Araguas-Araguas, L., & Gonfiantini, R.
Loggerhead Shrikes (Lanius ludovicianus excubitorides) (1993). Isotopic patterns in modern global precipitation.
wintering in Mexico. Diversity and Distributions, 13, In P. K. Swart, K. C. Lohmann, J. MacKenzie, & S.
166171. Savin (Eds.), Climate change in continental isotopic records
Phillips, D. L., & Eldridge, P. M. (2006). Estimating the tim- (pp. 137). American Geological Union, Geophysical
ing of diet shifts using stable isotopes. Oecologia, 147, Monograph Series.
195203. Rubenstein, D. R., Chamberlain, C. P., Holmes, R. T.,
Podlesak, D. W., McWilliams, S. R., & Hatch, K. A. (2005). Ayres, M. P., Waldbauer, J. R., Graves, G. R., & Tuross,
Stable isotopes in breath, blood, feces and feathers can N. C. (2002). Linking breeding and wintering ranges of
indicate intra-individual changes in the diet of migra- a migratory songbird using stable isotopes. Science, 295,
tory songbirds. Oecologia, 142, 501510. 10621065.
Point, D., Sonke, J. E., Day, R. D., Roseneau, D. G., Rubenstein, D. R., & Hobson, K. A. (2004). From birds to
Hobson, K. A., Vander Pol, S. S., . . . Becker, P. R. butterflies: Animal movement patterns and stable isotopes.
(2011). Methylmercury degradation influenced by sea- Trends in Ecology and Evolution, 19, 256263.
ice cover in northern high-latitude marine ecosystems. Ruegg, K. C., Anderson, E. C., Harrigan, R. J., Paxton,
Nature Geoscience, 4, 188194. K. L., Kelly, J. F., Moore, F., & Smith, T. B. (2017).
Popa-Lisseanu, A. G., Sörgel, K., Luckner, A., Wassenaar, Genetic assignment with isotopes and habitat suitability
L. I., Ibáñez, C., et al. (2012). A triple-isotope approach (GAIAH), a migratory bird case study. Methods in
to predict the breeding origins of European bats. PLoS Ecology and Evolution. Available from https://doi.org/
One, 7(1), e30388. Available from https://doi.org/ 10.1111/2041-210X.12800.
10.1371/journal.pone.00303. Rundell, C. W., Wunder, M. B., Alvarado, A. H., Ruegg,
Post, D. M. (2002). Using stable isotopes to estimate trophic K. C., Harrigan, R., Schuh, A., . . . Novembre, J. (2013).
position: Models, methods, and assumptions. Ecology, Novel statistical methods for integrating genetic and
83, 703718. stable isotope data to infer individual-level migratory
Powell, L. A., & Hobson, K. A. (2006). Enriched feather connectivity. Molecular Ecology, 22, 41634176.
hydrogen isotope values for Wood Thrushes sampled Rushing, C. S., Ryder, T. B., Saracco, J. F., & Marra, P. P.
in Georgia, USA, during the breeding season: (2014). Assessing migratory connectivity for a long-
Implications for quantifying dispersal. Canadian Journal distance migratory bird using multiple intrinsic mar-
of Zoology, 84, 13311338. kers. Ecological Applications, 24, 445456.

Schell, D. M., Saupe, S. M., & Haubenstock, N. (1989). Thomas, S. M., & Crowther, T. W. (2015). Predicting rates
Bowhead whale (Balaena mysticetus) growth and feeding of isotopic turnover across the animal kingdom: A syn-
as estimated by δ13C techniques. Marine Biology, 103, thesis of existing data. Journal of Animal Ecology, 84,
433443. 861870.
Schmaltz, L., Loonstra, A. H. J., Wymenga, E., Hobson, Tieszen, L. L., Boutton, T. W., Tesdahl, K. G., & Slade,
K. A., & Piersma, T. (2017). Quantifying the non- N. A. (1983). Fractionation and turnover of
breeding provenance of staging ruffs, Philomachus pug- stable carbon isotopes in animal tissues: Implications
nax, using stable isotopes analysis of different tissues. for δ13C analysis of diet. Oecologia, 57, 3237.
Journal of Ornithology. Available from https://doi.org/ Tietje, W. D., & Teer, J. G. (1988). Winter body condition of
10.1007/s10336-017-1488-x. Northern Shovelers on freshwater and saline habitats.
Sealy, J. C., van der Merwe, N. J., Lee Thorp, J. A., & In D. J. Batt, R. H. Chaebreck, L. H. Fredrickson, &
Lanham, J. L. (1987). Nitrogen isotopic ecology in D. G. Raveling (Eds.), Waterfowl in winter (pp. 353377).
southern Africa: Implications for environmental and Minneapolis: University of Minnesota Press.
dietary tracing. Geochimica et Cosmochimica Acta, 51, Trueman, C. N., & Moore, A. (2007). Use of the
27072717. stable isotope composition of fish scales for monitoring
Sellick, M. J., Kyser, T. K., Wunder, M. B., Chipley, D., & aquatic ecosystems. In T. E. Dawson, & R. T. W.
Norris, D. R. (2009). Geographic variation of strontium Siegwolf (Eds.), Stable isotopes as indicators of ecological
and hydrogen isotopes in avian tissue: Implications for change (pp. 145161). London: Academic Press.
tracking migration and dispersal. PLoS One, 4, e4735. Vanderklift, M. A., & Ponsard, S. (2003). Sources of varia-
Smith, A. D., Lott, C. A., Smith, J. P., Donohue, K. C., tion in consumer-diet δ15N enrichment: A meta analy-
Wittenberg, S., Smith, K. G., & Goodrich, L. (2009). sis. Oecologia, 136, 169182.
Deuterium measurements of raptor feathers: Does a Van der Merwe, N. J., Lee Thorp, J. A., Thackeray, J. F.,
lack of reproducibility compromise geographic assign- Hall-Martin, A., Kruger, F. J., Coertzees, H., . . .
ment? Auk, 126, 4146. Lindeque, M. (1990). Source-area determination of ele-
Soto, D., Hobson, K. A., & Wassenaar, L. I. (2016). Using phant ivory by isotopic analysis. Nature, 346, 744746.
stable-hydrogen isotopes of freshwater fish tissue as a Veen, T., Hjernquist, M. B., Van Wilgenburg, S. L.,
tracer of provenance. Ecology and Evolution. Available Hobson, K. A., Folmer, E., Font, L., & Klaassen, M.
from https://doi.org/10.1002/ece32519. (2014). Identifying the African wintering grounds of
Stewart, R. E. A., Outridge, P. M., & Stern, R. A. (2003). hybrid flycatchers using a multi-isotope (δ2H, δ13C,
Walrus life-history movements reconstructed from lead δ15N) assignment approach. PLoS One, 9(5), e98075.
isotopes in annual layers of teeth. Marine Mammal Available from https://doi.org/10.1371/journal.
Science, 19, 806818. pone.0098075.
Still, C. J., & Powell, R. (2010). Continental-scale distribu- Villacis, M., Vimeux, F., & Taupin, J. D. (2008). Analysis of
tions of vegetation stable carbon isotope ratios. In J. B. the climate controls on the isotopic composition of pre-
West, G. J. Bowen, T. E. Dawson, & K. P. Tu (Eds.), cipitation (δ18O) at Nuevo Rocafuerte, 74.5 W, 0.9 S,
Isoscapes: Understanding movements, pattern and process on 250 m, Ecuador. Comptes Rendu Geoscience, 340, 19.
Earth through isotope mapping (pp. 179195). New York: Vitousek, P. M., Aber, J., Howarth, R. W., Tilman, G. D.,
Springer. Matson, P. A., Schindler, D. W., . . . Tilman, D. G.
Studds, C. E., McFarland, K. P., Aubry, Y., Rimmer, C. C., (1997). Human alteration of the global nitrogen cycle:
Hobson, K. A., Marra, P. P., & Wassenaar, L. I. (2012). auses and consequences. Ecological Applications, 7,
Stable-hydrogen isotope measures of natal dispersal 737750.
reflect observed population declines in a threatened Vogel, J. C., Eglington, B., & Auret, J. M. (1990). Isotopic
migratory songbird. Diversity and Distributions, 18, fingerprints in elephant bone and ivory. Nature, 346,
919930. 747749.
Szymanski, M., Afton, A., & Hobson, K. A. (2006). Use of Walther, B. D., & Limburg, K. E. (2012). The use of otolith
stable isotope methodology to determine natal origin of chemistry to characterize diadromous migrations.
hatch-year mallards shot during fall in Minnesota. Journal of Fish Biology, 81, 796825.
Journal of Wildlife Management, 71, 13171324. Wassenaar, L. I., & Hobson, K. A. (1998). Natal origins of
Terzer, S., Wassenaar, L. I., Araguás-Araguás, L. J., & migratory Monarch butterflies at wintering colonies in
Aggarwal, P. K. (2013). Global isoscapes for δ18O and Mexico: New isotopic evidence. Proceedings of the
δ2H in precipitation: Improved prediction using region- National Academy of Sciences, 95, 1543615439.
alized climatic regression models. Hydrology and Earth Wassenaar, L. I., & Hobson, K. A. (2000). Stable-carbon
System Sciences, 17, 4713. and hydrogen isotope ratios reveal breeding origins of

FURTHER READING 115
red-winged blackbirds. Ecological Applications, 10, Yohannes, E., Hobson, K. A., & Pearson, D. J. (2007).
911916. Feather stable-isotope profiles reveal stopover habitat
Van Wilgenburg, S., & Hobson, K. A. (2011). Combining selection and site fidelity in nine migratory species
feather stable isotope (δD) and band recovery data to moving through sub-Saharan Africa. Journal of Avian
improve probabilistic assignment of migratory birds to Biology, 38, 347355.
origin. Ecological Applications, 21, 13401351. Vander Zanden, H. B., Wunder, M. B., Hobson, K. A., Van
Van Wilgenburg, S. L., Hobson, K. A., Brewster, K. R., & Wilgenburg, S. L., Wassenaar, L. I., Welker, J. M., &
Welker, J. M. (2012). Addressing uncertainty in asses- Bowen, G. J. (2015). Space-time tradeoffs in the devel-
sing dispersal in threatened migratory species using opment of precipitation isotope models for determining
stable hydrogen isotope analysis of feathers. Endangered migratory origin. Journal of Avian Biology, 46, 658667.
Species Research, 16, 1729. Zazzo, A., Monahan, F. J., Moloney, A. P., Green, S., &
Wunder, M., Hobson, K. A., Kelly, J., Marra, P., Schmidt, O. (2011). Sulphur isotopes in animal hair
Wassenaar, L. I., Stricker, C., & Doucette, R. (2009). track distance to sea. Rapid Communication in Mass
Does a lack of design and repeatability compromise sci- Spectrometry, 25, 23712378.
entific criticism? A response to Smith et al. Auk, 126,
922926.
Wolf, B. O., & Martinez del Rio, C. (2000). Use of saguaro Further Reading
fruit by white-winged doves: Isotopic evidence of a
tight ecological association. Oecologia, 124, 536543. Cormie, A. B., Schwarcz, H. P., & Gray, J. (1994).
Yang, L. H., Ostrovsky, D., Rogers, M. C., & Welker, J. M. Relationship between the hydrogen and oxygen iso-
(2016). Intra-population variation in the natal origins and topes of deer bone and their use in the estimation of
wing morphology of overwintering western monarch relative humidity. Geochimica et Cosmochimica Acta, 60,
butterflies Danaus plexippus. Ecography, 39, 9981007. 41614166.
Yerkes, T., Hobson, K. A., Wassenaar, L. I., Macleod, R., & Haramis, G. M., Jorde, D. G., Macko, S. A., & Walker, J. L.
Coluccy, J. M. (2008). Stable Isotopes (δD, δ13C, δ15N) (2001). Stable isotope analysis of canvasback winter diet
reveal associations among geographic location and con- in Upper Chesapeake Bay. Auk, 118, 10081017.
dition of Alaskan Northern Pintails. Journal of Wildlife Hebert, C., & Wassenaar, L. I. (2001). Stable nitrogen iso-
Management, 72, 715725. topes in waterfowl feathers reflect agricultural land use
Yohannes, E., Hobson, K. A., Pearson, D., Wassenaar, L. I., in western Canada. Environmental Science and
& Biebach, H. (2005). Stable isotope analyses of feathers Technology, 35, 34823487.
help identify autumn stopover sites of three long- Hobson, K. A. (1987). Use of stable-carbon isotope analysis
distance migrants in northeastern Africa. Journal of to estimate marine and terrestrial protein content in
Avian Biology, 36, 235241. gull diets. Canadian Journal of Zoology, 65, 12101213.

C H A P T E R
5
Tracking of Movements of Terrestrial
Mammals Using Stable Isotopes
Christian C. Voigt1,2 and Linn S. Lehnert1,2
1
Leibniz Institute for Zoo and Wildlife Research, Berlin, Germany, 2Freie Universität Berlin,
Berlin, Germany
5.1 INTRODUCTION estimated based on a single encounter. For

example, the origin of a bat killed by a wind
Isotopic tracking of movements has been turbine can be assessed based on the
used in the study of mammal biology since the stable hydrogen isotope ratios of its fur
1990s. Yet only recently has this field gained (Baerwald, Patterson, & Barclay, 2014; Lehnert
momentum due to more laboratories offering et al., 2014; Voigt, Lindecke, Schönborn, Kramer-
their isotopic services, the standardized analysis Schadt, & Lehmann, 2016; Voigt, Popa-Lisseanu,
of hydrogen stable isotope ratios of organic Niermann, & Kramer-Schadt, 2012). Third, since
materials, and the importance of establishing the isotopic tracer is inactive in biologically inert
animal movements in conservation biology. The material, such as keratin, isotopic tracking may
huge benefit of isotope tracking terrestrial mam- help to unravel the movement biology of ani-
mals over alternative methods such as tagging mals in historic times based on museum collec-
with transmitters or loggers is threefold: First, tions (Ossa, Kramer-Schadt, Peel, Scharf, &
this approach depends only on the use of minute Voigt, 2012). This offers the unique possibility to
tissue samples (about 0.5 mg, depending on study movements of extinct mammals and to
number and type of elements used) for measur- compare the migration patterns of historic speci-
ing endogenous markers, making this technique mens with those of extant conspecifics.
applicable to any mammal, independent of size. Previous reviews in this field focused on
This facilitates the collection of samples in a min- general aspects related to tracking migratory
imally invasive way, which is an important animals (West, Bowen, Cerling, & Ehleringer,
aspect for animal welfare considerations. 2006; Wunder, 2012; Chapter 4: Application
Second, this approach does not require the of Isotopic Methods to Tracking Animal
recapture of the same individual. Consequently, Movements) or on taxon-specific applications
the origin of a small-sized animal can be (Crawford, McDonald, & Bearhop, 2008). Here,

118 5. TRACKING OF MOVEMENTS OF TERRESTRIAL MAMMALS USING STABLE ISOTOPES
we review published studies aimed at using mention other studies focusing on dispersal,
stable isotopes to track terrestrial mammals, local movements and diurnal movements as
ranging from small-sized bats to elephants, well (Table 5.1). Our chapter does not cover
covering three commonly used isotopic elements forensic studies for tracking the origin of animal
(H, C, and N) and including local, latitudinal, products such as ivory from source countries; a
and elevational movements. The focus of this topic that has been addressed before in other
chapter will be on migratory movement, namely publications for wildlife (e.g., Peterson & Fry,
seasonal bidirectional movements, but we 1987; Ziegler, Merker, Streit, Boner, & Jacob, 2016)
TABLE 5.1 Past Studies in Which Stable Isotopes Were Used to Track the Origin or Movements of Terrestrial
Mammals (Sorted Following Orders)
Section Order Climate Zone Country Species Isotopes Source
Elevational migration Artiodactyla Temperate Europe; northern Ovis aries, C; N Männel,

part Auerswald,
of the European Bos taurus, and Schnyder
Alps Capra hircus (2007)
Latitudinal migration Artiodactyla Temperate Europe; Ireland O. aries S Zazzo et al.

(2011)
Latitudinal migration Carnivora Temperate; USA; midwestern Puma concolor H; C Henaux et al.
subtropical (2011)
Latitudinal migration Carnivora Temperate North America Lynx rufus, O Pietsch and
Tütken (2016)
P. concolor
Latitudinal migration Carnivora Subtropic USA; California Gulo gulo C; N Moriarty et al.
(2009)
Latitudinal migration Chiroptera Tropics; Mexico Leptonycteris curasoae C Fleming et al.
subtropics (1993)
Glossophaga soricina
Latitudinal migration Chiroptera Subtropics; USA; Mexico Tadarida brasiliensis C; N Wurster,
arid McFarlane, and
Bird (2007)
Latitudinal migration Chiroptera Tropics Costa Rica Carollia castanea, C Voigt et al.
(2012)
Carollia sowelli,
Carollia perspicillata
Latitudinal migration Chiroptera Temperate Canada Myotis lucifugus, C; N Segers and
Broders (2015)
Myotis septentrionalis
Latitudinal migration Chiroptera Tropics Madagascar Pteropus rufus, C; N Reuter et al.
(2016)
Eidolon dupreanum,
Rousettus
madagascariensis
(Continued)

5.1 INTRODUCTION 119
Latitudinal migration Chiroptera Temperate Europe Nyctalus noctula, H; C; N Voigt et al.

(2016)
Pipistrellus pipistrellus,
Pipistrellus nathusii
Elevational migration Chiroptera Tropics Honduras Artibeus toltecus, H Erzberger,
Popa-Lisseanu,
Micronycteris microtis,
Lehmann,
Artibeus jamaicensis, and Voigt
(2011)
Sturnira ludovici,
Sturnira lilium,
Myotis keaysi,
Molossus ater
Elevational migration Chiroptera Tropics Tanzania Miniopterus natalensis H; C; N Voigt et al.
(2014)
Latitudinal migration Chiroptera Temperate USA Lasiurus cinereus H Cryan et al.
(2004)
Latitudinal migration Chiroptera USA Lasiurus borealis, H Britzke et al.
(2009)
Myotis sodalist,
M. septentrionalis,
M. lucifugus
Latitudinal migration Chiroptera Temperate USA; Indiana M. sodalist H Britzke et al.
(2012)
Latitudinal migration Chiroptera Temperate USA Perimyotis subflavus H Fraser et al.
(2012)
Latitudinal migration Chiroptera Tropics Africa Eidolon helvum, H; C; N Ossa et al.

(2012)
Rousettus aegyptiacus,
Lissonycteris angolensis,
Epomophorus wahlbergi,
Hypsignathus
monstrosus,
Epomops franqueti,
Epomophorus crypturus
Latitudinal migration Chiroptera Temperate USA; Michigan M. lucifugus H Sullivan et al.

(2012)
Latitudinal migration Chiroptera Temperate USA L. cinereus cinereus H Cryan et al.
(2014)
(Continued)


Latitudinal migration Chiroptera Temperate Europe N. noctula H Voigt et al.

(2014)
Latitudinal migration Chiroptera Temperate USA Lasionycteris H Fraser et al.

noctivagans (2017)
Latitudinal migration Chiroptera Temperate Europe N. noctula H
Latitudinal migration Chiroptera Temperate Europe P. nathusii, H Voigt, Popa-

Lisseanu, et al.
P. pipistrellus, (2012)
N. noctula,
Nyctalus leisleri
Latitudinal migration Chiroptera Temperate Canada L. cinereus, H; C; N Baerwald et al.
(2014)
L. noctivagans
Latitudinal migration Chiroptera Temperate Europe N. noctula H Lehnert et al.

(2014)
Local movements Chiroptera Tropics Central America Desmodus rotundus C Voigt and
Kelm (2006)
Local movements Chiroptera Tropics Central America C. castanea, C Voigt, Voigt-
Heucke et al.
C. sowelli (2012)
C. perspicillata
Local movements Chiroptera Tropics Central America Phyllostomus discolor, C Voigt (2010)
A. jamaicensis
Artibeus lituratus,
Artibeus watsoni,
Ectophylla alba
Local movements Chiroptera Tropics Central America More than 10 species Rex et al.
(2011)
Local movements Chiroptera Tropics Madagascar 16 bat species N,C Dammhahn

and Goodman
(2014)
Local movements Chiroptera Tropics Madagascar P. rufus, N,C Reuter et al.
(2016)
Eidolon helvum
R. madagascariensis
Latitudinal migration Proboscidea Tropics Africa; Kenya Loxodonta africana H; C; N Cerling et al.
(2009)

5.2 STABLE ISOTOPES AND MOVEMENTS OF TERRESTRIAL MAMMALS 121
and humans (Meier-Augenstein, 2011). Yet it is on a nectar corridor formed by columnar cacti
important to note that the approaches from tropical Mexico to the Sonoran and
described in this chapter and elsewhere in this Chihuahuan desert in the United States. It is
book could also be applied in wildlife forensic important in the context of this chapter to
studies. acknowledge that the use of δ13C values in
A large proportion of isotopic papers pub- this specific case did not reveal the spatial
lished on mammal migration have focused on pattern of migratory movement or the origin
bats and using tissue δ2H measurements as the of animals based on probability of origin
isotopic tracer. The likely reason for this approaches, yet it highlighted the relevance of
taxonomic bias is that bats are elusive and dif- columnar cacti as a food source for migratory
ficult to track with alternative methods, such nectar-feeding bats. The suitability of δ13C
as relatively heavy GPS units and loggers, and values for use in tracing spatial origins
that migratory bats cover long distances, cross- depends on whether δ13C values vary in base-
ing several precipitation-based δ2H isoclines line food web samples, such as plant matter,
during their annual latitudinal journeys. across latitudinal, longitudinal, or elevational
Lastly, the wind energy-wildlife conflict has gradients. In contrast to hydrogen isotopes, we
stimulated several papers on bats in this field. are missing large-scale continental maps
Nonetheless, we aimed at being comprehen- describing spatial variation in plant δ13C and
sive in referencing and discussing also studies δ15N values (but see Chapter 2: Introduction
on non-Chiropteran species. to Conducting Stable Isotope Measurements
for Animal Migration Studies, Chapter 3:
Isoscapes for Terrestrial Migration Research).
Therefore studies on mammals mostly used
5.2 STABLE ISOTOPES AND δ13C and δ15N values of consumer tissue or
MOVEMENTS OF TERRESTRIAL keratin in conjunction with δ2H values to
MAMMALS confirm movement and migration patterns or
to highlight the variable isotopic background
5.2.1 Carbon and Nitrogen of ecosystems that animals cross along their
One of the first stable isotope papers on the seasonal journey. For example, when covering
ecology of a migratory mammal established long distances, mobile animals such as migrat-
δ13C values to assess the diet of nectar-feeding ing bats may incorporate isotopes from a
bat species in North America (Fleming, variety of food sources. Consequently, depend-
Nuñez, & Sternberg, 1993). During those early ing on the tissue chosen, variation in δ13C and
days, stable isotope measurements depended δ15N values of individuals captured at a
on relatively large sample mass, which forced specific stopover site can be high when the
researchers to collect muscular and skin catchment area of this stopover site is large
tissue from sacrificed animals and from (Segers & Broders, 2015). Intraindividual
museum specimens. In the larger migratory approaches have also been used, e.g., when
Leptonycteris curasoae, δ13C values indicated a comparing the variation of δ13C and δ15N
strong dependency on nectar from columnar values across tissues within individuals, if
cacti with a CAM photosynthetic pathway, higher variances could best be explained by
whereas the smaller, nonmigratory Glossophaga individuals encountering more habitats with
soricina exhibited δ13C values indicative of a variable isotopic compositions (Voigt et al.,
strong dependency on nectar from C3 plants. 2016). In some rare cases, baseline data on the
These findings confirmed that L. curasoae relied geographic variation of δ13C and δ15N values

in food webs leading to bats is available, so feeding habits and movement of bats (Reuter,
that δ13C and δ15N values in consumer tissues Wills, Lee, Cordes, & Sewall, 2016; Rex,
can be used to test movement patterns inferred Michener, Kunz, & Voigt, 2011; Voigt, Voigt-
from δ2H values in fur keratin. For example, Heucke, & Kretzschmar, 2012). Voigt and
bats performing elevational movements along colleagues used breath δ13C values of fruit-
the slopes of Mount Kilimanjaro in Tanzania eating bats to assess the strata in which they
were detected using isotope sampling (Voigt foraged, since fruit-eating bats oxidize dietary
et al., 2016). Again, the general idea behind carbohydrates quickly (Voigt, 2010). At the
this approach lies in the assumption that same study site, Voigt, Voigt-Heucke, et al.
tissues with different isotopic retention times (2012) analyzed the movement of fruit-eating
integrate over specific retrospective periods bats between rainforest and cleared areas
and thus shed light over the isotopic variation (with expected higher δ13C values). In other
of consumed food sources during the studies, δ13C values of tail hair were used as a
specific period of tissue formation (Chapter 4: proxy to assess movements of African ele-
Application of Isotopic Methods to Tracking phants between protected and heavily used,
Animal Movements). overgrazed communal areas, since in the
In general, continental patterns of δ13C and latter habitat elephants switch from grass to a
δ N values are often too small in relation to
15
browse-dominated diet (Cerling, Wittemyer,
the isotopic variation experienced by mammals Ehleringer, Remien, & Douglas-Hamilton,
at local scales. For example, in most cases dif- 2009). In an earlier study, authors also used
ferences in δ13C values between plants with a δ13C and δ15N values in conjunction with GPS
C4/CAM and C3 photosynthetic pathway are data to explain feeding behavior of GPS
larger than most continental gradients of δ13C tagged elephants in relation to their local
values for C3 plants (Chapter 3: Isoscapes for movements and use of crops (Cerling et al.,
Terrestrial Migration Research). Yet, the strong 2006).
contrast in δ13C values between C3 and
C4/CAM plants can be used to infer local
movements of mammals. For example, δ13C
values in tissues or breath of vampire bats
5.2.2 Hydrogen
(Desmodus rotundus) suggested a diet based on Initial mammalian studies using δ2H values
C4 foods that were provided by livestock feed- as a tool for forensic investigations focused on
ing on grasses even though vampires were the source of hydrogen in biologically inert
captured in an C3 rainforest environment body products, such as hair keratin. Sharp,
(Voigt, Grasse, Rex, Hetz, & Speakman, 2008; Atudorei, Panarello, Fernández, and Douthitt
Voigt & Kelm, 2006). This indicated a commut- (2003) demonstrated that about 31% of hydro-
ing behavior of vampire bats between roosts in gen in human hair originates from drinking
the rainforests and feeding areas on pastures. water. They also pointed out the problem of
Further, C3 plants may vary in δ13C values exchangeable hydrogen in hair keratin (9% of
when they are exposed to a gradient of water total hydrogen in human hair, Sharp et al.,
stress or additionally as a result of soil respira- 2003). The fact that hydrogen in organic mole-
tion in tropical forests, a pattern referred to as cules might exchange with ambient humidity
the “canopy effect” (Buchmann, Brooks, & when attached to oxygen or nitrogen, is an
Ehleringer, 2002). Accordingly, fruits and issue which hampered the adoption of
nectar may vary in δ13C values in relation to this approach in initial years (Chapter 2:
height. This gradient has been used to infer Introduction to Conducting Stable Isotope

5.3 ISOTOPIC RETENTION IN TISSUES: THE RETROSPECTIVE MOVING TIME WINDOW 123
Measurements for Animal Migration Studies, catchment areas of bats observed in hibernac-
Meier-Augenstein, Hobson, & Wassenaar, ula (Britzke, Loeb, Romanek, Hobson, &
2013). Importantly, spatial information derived Vonhof, 2012; Sullivan, Bump, Kruger, &
from isotopic tracking is built on transfer func- Peterson, 2012; Voigt et al., 2014) and that for
tions (i.e., isotopic difference between precipi- bats killed at wind turbines (Baerwald et al.,
tation and tissue) which require the collection 2014; Lehnert et al., 2014; Voigt, Popa-Lisseanu,
of baseline data across a large geographical et al., 2012). An additional study addressed the
range (e.g., Ehleringer et al., 2008; Popa- migration behavior of the African straw-
Lisseanu et al., 2012). Such transfer functions colored fruit bat, Eidolon helvum and other
are beneficial for developing and using tissue- African flying foxes (Ossa et al., 2012).
specific isoscapes for assignment purposes
(Chapter 4: Application of Isotopic Methods to
Tracking Animal Movements).
In the first study in which δ2H was used to
5.3 ISOTOPIC RETENTION IN
assess the spatial behavior of a free-ranging
TISSUES: THE RETROSPECTIVE
mammal species, Cryan, Bogan, Rye, Landis,
MOVING TIME WINDOW
and Kester (2004) used keratin δ2H values of
bat hair to assess the timing of molt and the
5.3.1 Fur and Molting Patterns
migration behavior of Lasiurus cinereus, the Isotopic tracking studies in mammals use
North American hoary bat. First, they mostly biologically inert material such as fur.
confirmed that molting occurs before migration Fur consists of various keratinous amino acids
in this species; an important point when bats dominated by cysteine, serine, and glutamic
captured during migration should be assigned acids (for humans: Meier-Augenstein et al.,
to their summer origin (i.e., molting area) 2013). The relative contribution of amino acid
based on keratin δ2H values. Secondly, they composition might vary across taxa, particu-
found low variation of keratin δ2H values dur- larly between mammalian fur keratin and avian
ing the molting period, which suggested that feather keratin. In isotopic tracking studies that
δ2H values in keratin material are a reliable focus on migratory animals, it is frequently
predictor of δ2H values of long-term average assumed that individuals molt before they
precipitation, one of the basic assumptions migrate. This is often the case, yet poorly docu-
underlying this approach. Follow-up studies mented for mammals. Fraser, Longstaffe, and
focusing on the same and additional North Fenton (2013) reviewed the literature on molt-
American species established information ing patterns in bats, particularly in context to
about the general migration behavior of these stable isotope studies. They highlighted that
species for North America, yet studies varied molting occurs in bats mostly during a single
in calibration standards (Britzke, Loeb, event in summer or fall, yet with
Hobson, Romanek, & Vonhof, 2009; Cryan, notable exceptions. Importantly, the phenology
Stricker, & Wunder, 2014; Fraser, Brooks, & of molting may vary across species and within
Longstaffe, 2017; Fraser, McGuire, Eger, species (age and sex). Also, molting usually
Longstaffe, & Fenton, 2012). In parallel, isotopic occurs gradually, i.e., body parts shade gradu-
transfer functions were established for ally their fur with areas covered by old fur adja-
European bats (Popa-Lisseanu et al., 2012), cent to areas with regrown fur, a pattern called
enabling European researchers to assign the asynchronous fur growth. Fraser et al. (2013)
summer origin of migratory bats as well. On concluded that fur samples should be taken
both continents, most studies focused on the dorsally from adult males to be most likely

representative of the area of summer residency material and therefore, these body products
of a bat. Most information on molting in bats carry a constant isotopic fingerprint that can be
has been obtained from studies of the temperate used for dietary and spatial tracking of mam-
zone, thus we are data deficient for the clear mals. For example, long tail hair and whiskers
majority of bats in subtropical and tropical areas can be cut into smaller segments and each of
where the species density of Chiroptera is high- these increments record the isotopic composi-
est. In a detailed intraindividual study on molt- tion of the ingesta at the time of growth
ing patterns in free-ranging Nathusius bats, (Schwertl, Auerswald, & Schnyder, 2003). For
Voigt, Lindecke, Schönborng, Kramer-Schadt, this approach, it is essential to quantify the
and Lehmann (2016) emphasized that the exact growth rates for the specific species under
onset of molting was difficult to define. Further, study (Table 5.2). On the one hand, growth
individuals varied largely with respect to fur rates vary across taxa and within species, so
renewal, with females requiring about 5055 that it is important to establish the growth rate
days and males 2550 days to complete molt. of a selected study organism before the study.
Molting ended around mid-July for this On the other hand, it seems questionable
European migratory bat species, suggesting that whether growth rates measured in captivity
δ2H values in fur keratin may integrate over a reflect those observed in free-ranging animals.
period from early June to mid-July. The high Therefore these growth rate values must be
mobility of bats and the high variability in the treated with caution and uncertainties should
timing of molt within and across species make it be included in statistical models. Thus far, we
difficult to define exact periods for collecting lack studies that used the potential of
reference data for the derivation of transfer hydrogen-based isotopic tracking using whisker
functions (see 5.4.1). In many cases, researchers or tail hair increments; possibly, because large-
avoided any potential bias by using keratin δ2H sized mammals travel too slowly and thus do
values of nonmigratory species, i.e., those not cross ecosystem isoclines to make this
species that remain in an area presumably with approach fruitful. However, Henaux, Powell,
similar isotopic composition throughout a year. Hobson, Nielsen, and LaRue (2011) developed
Other non-bat mammals have also distinct an isotopic spatial track for cougars on a rela-
molting patterns, particularly in the temperate tively small scale based on stable hydrogen iso-
zone where ambient temperatures vary tope ratios in increments of claw material. Such
largely between summer and winter (Ling, an approach might be feasible if growth rates
1970). Species-specific molting patterns may and local spatial scales are well established for
vary according to the geographic site, taxonomy the study species and area, respectively.
(e.g., subspecies level), sex, and age. In sum-
mary, the timing of molting is highly relevant
for isotopic tracking of migratory movements
based on stable hydrogen isotope ratios in fur
5.3.3 Tissue Turnover Rates
keratin, and therefore the researcher needs to Within mammals, organs differ in their iso-
pay attention to this aspect of a species’ biology. topic retention, i.e., the rate at which
stable isotopes are replaced as part of growth,
maintenance, and degeneration of tissues
(Martinez del Rio & Carleton, 2012). The isoto-
5.3.2 Other Keratinous Body Products
pic times are usually quantified using diet-
In general, mammalian tail hair, whiskers, switch experiments (e.g., Ayliffe et al., 2004;
and claws consist of biologically inert keratin Podlesak et al., 2008). Either material is

5.3 ISOTOPIC RETENTION IN TISSUES: THE RETROSPECTIVE MOVING TIME WINDOW 125
TABLE 5.2 Growth Rates of Tail Hair and Whiskers in Selected Mammal Species
Common Growth Rate
Tissue Species Name (mm /day) Source
Tail hair Equus caballus Horse 0.630.88 West et al. (2004)
Tail hair E. caballus Horse 1.28 Sharp et al. (2003)

Tail hair E. caballus Horse 0.83 Dunnett (2005)
Tail hair Loxodonta africana African 0.55 6 0.11 (M) Cerling et al. (2006)
Elephant
0.81 6 0.13 (F)
Tail hair L. africana African 0.731.04 Cerling et al. (2009)
Elephant
Tail hair L. africana African 0.711.03 Wittemyer, Cerling, and
Elephant Douglas-Hamilton (2009)
Tail hair L. africana African 0.170.37 Codron et al. (2013)
Elephant
Tail hair Bos taurus Cow 0.691.06 Schwertl et al. (2003)
Tail hair B. taurus Cow 0.510.63 Fisher, Wilson, Leach, and Scholz
(1985)
Tail hair E. caballus Horse 0.390.87 Ayliffe et al. (2004)
Whiskers Canis lupus Wolf 0.370.49 McLaren, Crawshaw, and
Patterson (2015)
Whiskers Panthera leo, Panthera Lion, Leopard 0.65 (0.050.84) Mutirwara, Radloff, and
pardus Codron (2017)
Whiskers Mustela erminea Stoat 0.60 Spurr (2002)
Whiskers Meles meles European 0.43 (0.230.83) Robertson, McDonald, Delahay,
badger Kelly, and Bearhop (2013)
Note that we have not listed those of pet or laboratory animals. Single numbers refer to mean growth rates (with addition of 6 1 standard
devation, if possible) and two numbers to the range of observed growth rates. Abbreviations: F, females; M, males.
collected with minimal invasiveness, such as control; a condition that is usually met only in
blood (separated into hematocrit and plasma) captivity. However, it is important to keep in
skin or muscle (using biopsy punches) or ani- mind that the captive diet might not necessar-
mals are euthanized during the diet switch. ily reflect the diet of wild animals and this
Currently, studies are heavily biased toward flaw might affect the isotopic retention times.
small mammals, since isotopic retention in For example, Voigt, Matt, Michener, and Kunz
each organ scales allometrically with body (2003) found low isotopic retention times for
mass, i.e., for a given organ tissue isotopic carbon stable isotopes in nectar-feeding bats
retention time increases with the body mass of fed a low nitrogen diet, whereas Mirón,
animals (Thomas & Crowther, 2015). Diet- Herrera, Ramı́rez, and Hobson (2006) showed
switch experiments usually require a setting in faster isotopic retention times, when the same
which the diet of animals is under complete species was fed a nitrogen-enriched diet.

After having gained a good understanding of appropriate. On the one hand, isotopic transfer
organ-specific isotopic retention times in each functions corresponding to single-species may
study animal, it is possible to use suffer from small sample sizes. Thus transfer
stable isotopes of specific organs as tracers of functions may not catch the full variation
diet and origin (Voigt et al., 2014). Thus far, observed over the range of expected tissue δ2H
the spatial aspect of this approach has been values. On the other hand, multispecies trans-
impaired by the inability to measure consis- fer functions may overestimate the variance
tently (and among laboratories) δ2H values of because of species-specific differences in δ2H
nonexchangeable H in materials other than values caused by food-web or guild-specific
keratin. Development of calibration standards deviations (Voigt, Lehmann, & Greif, 2015).
for metabolically active tissues like plasma,
hematocrit, muscle tissues, among others are
needed in the future (see Chapter 2:
5.4.2 Causes and Consequences of
Introduction to Conducting Stable Isotope
Measurements for Animal Migration Studies)
Variation
but issues related to exchange with body water There are numerous sources of variation in
have not been resolved. using mammal tissue δ2H values for animal
tracking. Laboratory analytical measurement
issues have been covered in Chapter 2,
Introduction to Conducting Stable Isotope
5.4 APPLICATION OF
Measurements for Animal Migration Studies,
STABLE ISOTOPES TO THE STUDY
and it should be mandatory now for all analy-
OF MIGRATORY MOVEMENTS ses to appropriately control for the effect of
exchangeable hydrogen on δ2H values (Meier-
5.4.1 Transfer Functions Augenstein et al., 2013). Second, food items
Several transfer functions exist linking may vary in isotopic composition and thus, tis-
mammalian keratin δ2H values with expected sues of consumers feeding on food items with
long-term precipitation δ2H values (Table 5.3). contrasting isotopic composition may reflect
For example, one based on human hair these differences even though they might
(Ehleringer et al., 2008), several for North have been exposed to the same δ2H values of
American bats (Cryan et al., 2004; modified in precipitation and ground water (Birchall,
Baerwald et al., 2014), for European bats O’Connel, Heaton, & Hedges, 2005; Britzke
(Popa-Lisseanu et al., 2012, modified in Voigt et al., 2009; Voigt, Schneeberger, & Luckner,
et al., 2014; Table 5.3). So far, studies have yet 2013). Food related differences in δ2H values
to establish robust transfer functions for carni- across species may rather originate from con-
vores, probably because of specific aspects in trasting δ2H values in food items in addition to
the biology of this taxon, possibly the low those factors acting on variance in precipita-
dependency on surface water for drinking tion δ2H (Chapter 3: Isoscapes for Terrestrial
(Pietsch, Hobson, Wassenaar, & Tütken, 2011). Migration Research, Voigt et al., 2015). For
Two points are important in the context of iso- example, syntopic bats showed contrasting
topic transfer functions of free-ranging mam- δ2H values in fur keratin when feeding on ter-
mals: Large variances in regression models restrial or limnic food items (Voigt et al., 2015),
hamper the accuracy of the predictive model. most likely because baseline δ2H values of
Also, we generally lack species-specific trans- plant matter differed between both habitats.
fer functions, which are presumed to be most Such differences in food webspecific isotopic

5.4 APPLICATION OF STABLE ISOTOPES TO THE STUDY OF MIGRATORY MOVEMENTS 127
TABLE 5.3 List of Published H Isotope Transfer Functions for a Variety of Mammal Species
Taxon Continent Species Transfer Function Source
Chiroptera Europe Multispecies δ Hf 5 1.07*δ Hp16.84

2 2 scaled to A
Popa-Lisseanu et al. (2012)
Chiroptera Europe Multispecies δ Hf 5 1.37*δ Hp 1 5.52
2 2 scaled to A
Voigt et al. (2014)
Chiroptera Europe Multispecies δ Hf 5 0.92*δ Hp30.72
2 2 scaled to A
-
δ Hf 5 0.9*δ Hp9.38
2 2 scaled to B
δ2Hf 5 0.77*δ2Hp 1 3.44 scaled to USGS
Chiroptera Africa Multispecies δ2Hf 5 1.52*δ2Hp 1 54.09 scaled to A

Ossa et al. (2012)
Chiroptera North Lasiurus cinereus δ Hf 5 0.79*δ Hp 2 24.81

2 2 scaled to C
Cryan et al. (2004)
America
Chiroptera North L. cinereus δ2Hf 5 0.73*δ2Hp 2 42.61scaled to C

Cryan et al. (2014)
America
Chiroptera North Myotis δ2Hf 5 0.79*δ2Hp 2 4.73 (M)scaled to B and C
Britzke et al. (2009)
America septentrionalis
δ Hf 5 1.25*δ Hp 1 18.48 (F)
2 2 scaled to B and C
δ Hf 5 0.98*δ Hp 1 5.48 (M 1 F)
2 2 scaled to B
and C
Chiroptera North Myotis lucifugus δ2Hf 5 0.49*δ2Hp 2 30.90 (M)scaled to B and C

America
δ Hf 5 0.33*δ Hp40.41 (F)
δ Hf 5 0.52*δ Hp30.82 (M 1 F)
2 2 scaled to B and
C
δ2Hf 5 2.69*δ2Hp 1 96.93scaled to B and C

Sullivan et al. (2012)
Chiroptera North Myotis sodalis δ2Hf 5 0.90*δ2Hp 2 0.59 (M)scaled to B and C
America
δ2Hf 5 0.71*δ2Hp8.17 (F)scaled to B and C
δ2Hf 5 0.83*δ2Hp2.97 (M 1 F)scaled to B and
C
Chiroptera North Perimyotis δ2Hf 5 20.04*δ2Hp1.79*δ2Hp 2 45.61 Fraser et al. (2012)

America subflavius (M)scaled to C
δ2Hf 5 20.03*δ2Hp1.61*δ2Hp 2 40.38
(F)scaled to C
Chiroptera North Lasiurus borealis δ2Hf 5 28.2*δ2Hp58.80 (M)scaled to B and C
America
δ Hf 5 1.35*δ Hp3.60 (F)
δ Hf 5 0.48*δ Hp26.10 (M 1 F)
2 2 scaled to B and
C
δ2Hf 5 1.48*δ2Hp 1 13.95 (M)scaled to C Pylant, Nelson, and Keller

(2014)
δ2Hf 5 1.75*δ2Hp 1 18.02 (F)scaled to C
Pylant et al. (2014)
δ Hf 5 1.67*δ Hp 1 16.84 (M 1 F)
2 2 scaled to C
Pylant et al. (2014)
(Continued)


Taxon Continent Species Transfer Function Source
Ungulates North Odocoileus δ HBoneCollagen 5 0.86*δ Hriver.

2 2
Pietsch et al. (2011)
water 1 20.28
scaled to B
America virginianus
Lagomorpha North Sylvilagus δ2Hf 5 0.80*δ2Hp25.46scaled to B

Pietsch et al. (2011)
America floridanus
Abbreviations: F, females; M, males. Information on whether data are scaled to keratin standards from the Berlin laboratory (A), the
Saskatoon laboratory (B), or other standards (C).
compositions suggest that it is necessary to been captured before the second molt are
establish and use taxon- or guild-specific isoto- included in regression models. A large
pic transfer functions where possible. For variation in the transfer function used to pre-
example, it might be necessary to use δ2H dict the summer origin of mammals might
values from nonmigratory bat species from a lead to imprecise geographical assignments of
terrestrial food web, i.e., not depending on animals.
aquatic insects, as a proxy for a migratory spe-
cies that also belongs to terrestrial food webs.
Third, interspecific differences in δ2H values 5.4.3 Case Study
may also stem from species-specific molting Here, we use a case study to describe the
patterns that may cause fur keratin to vary specific steps needed to engage in an isotopic
according to the mean δ2H value driving the tracking study of a migratory bat. Specifically,
food and drinking water used by the species at we elaborate on practical considerations,
the time of tissue synthesis (Britzke et al., whereas details of the modeling are described
2009). Naturally, the protocol for collecting fur in Chapter 9, Isoscape Computation and
should be consistent within a study. Lastly, it Inference of Spatial Origins With Mixed
is unknown how lactation changes the isotopic Models Using the R package IsoriX. The spe-
composition of juveniles, since suckling young cific example deals with the geographical
assimilate hydrogen from water and nutrients origin of bats killed by wind turbines in
in maternal milk. Since δ2H values of body Germany. Over the past decades wind energy
water might not necessarily be correlated with has become a prospering industry worldwide
δ2H values of fur keratin in mammals (Voigt (Arnett et al., 2016). Wind energy poses risks to
et al., 2013), assimilation of hydrogen from volant animals, with large numbers of birds
milk by juveniles might lead to altered δ2H and bats dying at wind turbines each year
values in fur keratin compared to correspond- (Hayes, 2013; Voigt et al., 2015). Two of the
ing values in adult conspecifics. Indeed, main challenges in solving this dilemma are,
lipid-rich foods, which are depleted in 2H, first, to establish measures to mitigate the
associated with nursing may drive dietary δ2H negative impacts on wildlife and, second, to
values more negative (Soto, Koehler, quantify the consequences of increased mortal-
Wassenaar, & Hobson, 2017). Since juveniles ity on affected source populations. In case of
carry the isotopic signature of the lactation bats, it is impossible to tell whether a bat killed
period in their fur keratin, it is likely that this at a wind turbine originates from a local or dis-
might cause additional variation in transfer tant population based on morphological or
functions if δ2H values of juveniles that have genetic traits. Therefore we applied isoscape

5.4 APPLICATION OF STABLE ISOTOPES TO THE STUDY OF MIGRATORY MOVEMENTS 129
origin models based on δ2H values of fur We started by fitting a δ2H precipitation
keratin collected from individual bats killed at isoscape for the relevant geographical area that
wind turbines, specifically from the common covers the distribution range of N. noctula in
noctule bat (Nyctalus noctula). Beforehand, we Europe (and partly beyond). We used mea-
established a relationship between δ2H values sured δ2Hp values of Europe made available
in fur keratin of nonmigratory bats and that of via the Global Network of Isotopes in
mean annual precipitation water. This relation- Precipitation (GNIP) to fit a pair of geostatisti-
ship enabled us to predict δ2H values in precip- cal models, approximating the relationship
itation water of the most likely places of origin. between topographic features of a location and
Collection of samples: In most studies, fur its amount weighted mean annual δ2Hp signa-
samples are collected from the dorsal part of ture. Subsequently, we built a δ2Hp isoscape
bats, mostly from the interscapular area. The for Europe using the geospatial model to pre-
amount of fur collected can be kept small, since dict the spatial distribution of δ2Hp values in
the stable isotope analysis requires only a our area of interest (Chapter 3: Isoscapes for
minute sample mass (e.g., 0.5 mg). However, in Terrestrial Migration Research). It is important
case of multielemental approaches, larger fur to note that the density of weather stations
samples are required. It is important that car- contributing with δ2Hp values varies spatially,
casses must be relatively fresh, because decom- which may lead to a limitation in predictive
position of fur caused by exposure to rain and power in sparsely covered regions, such as
high temperature might alter the isotopic com- northern regions. However, potential temporal
position of keratin. Yet, this process can take patterns in data availability should also be
considerable time and probably depends on explored. IsoriX provides tools to quantify
local ambient conditions such as warm ambient the prediction variance across the area of inter-
temperature and rainfall that will hasten the est (Fig. 9.4).
process of decomposition. Collected fur sam- The isotopic transfer function is required to
ples should be dried and stored in a small relate the measured δ2H values in fur keratin
envelope or plastic vial. To prevent any further to δ2H values in precipitation. We obtained
decomposition, it is advisable to store samples data for this function by combining datasets of
in a freezer or a dry place. nonmigratory, insectivorous bat species from
Preparation and analysis of samples: The labo- terrestrial food webs (Popa-Lisseanu et al.,
ratory based processes involved in the prepa- 2012) with those of noctule bats sampled
ration and isotopic analysis of samples are during their nonmigratory period across
mentioned in Chapter 2, Introduction to Europe (Voigt et al., 2014). By using a mixed-
Conducting Stable Isotope Measurements for species transfer function we likely increased
Animal Migration Studies. the variance within the dataset (Fig. 5.1),
Reconstructing of the origin of animals based on which might potentially lead to less accurate
stable hydrogen isotope ratios: The newly devel- assignments compared with a species-specific
oped R-package IsoriX was used for statistical dataset. However, to date we lack the data
analysis and Chapter 9, Isoscape Computation basis for a single species transfer function in
and Inference of Spatial Origins With Mixed Europe that covers the full isotopic range of
Models Using the R package IsoriX, provides possible places of origin. While acknowledging
details on all the different statistical substeps limitations in the spatial resolution of assign-
related to the workflow in IsoriX. Here we ments, we are convinced that providing
introduce and discuss the process and out- assignments based on a large multispecies
come of this analysis. dataset lead to more conservative and thus

FIGURE 5.1 The graph shows our H isotope transfer function (IZW Berlin: blue line and black dots) relating measured
δ2H values in fur keratin of nonmigratory bats to δ2H values in precipitation water across Europe. In order to demonstrate
the effect different laboratory standards used in different laboratories may have on the resulting data. We additionally
rescaled our dataset to two other common standards and added the resulting transfer functions to the plot (Saskatoon
Laboratory: green line and green crosses; and USGS 42: red line and red triangles). δ2H values measured at the stable isotope
laboratory IZW Berlin can be rescaled to USGS 42 using the following equation: 0.8388*δ2HIZW.Berlin 1 29.22. The respective
formula for rescaling δ2H values measured at the Saskatoon stable isotope laboratory can be found in Soto et al. (2017). As
can be seen the results differ, which is why one should exercise due care when analyzing a mixed dataset measured at dif-
ferent laboratories or with different standards.
also more reliable assignments than using a Geographical assignments: In the case study,
small but single species dataset which we aimed to assign the likely origin for 14 noc-
might be based on smaller sample sizes. Since tule bats killed at wind turbines. We focused on
we know that measured δ2H values in bats individuals found during the autumn migration
may vary significantly between age classes period after the molting season, below wind
(Baerwald et al., 2014; Britzke et al., 2009), we turbines. Results of our assignments indicate
considered adult individuals only both for the that most individuals found dead below wind
transfer function and the assignments. Results turbines originated from local populations,
for the transfer function are shown in Fig. 5.1. while one individual came from a distant popu-
For illustration, we plotted our transfer func- lation in Eastern Europe according to its δ2H
tion dataset scaled to three different standards values (Fig. 5.2). It is obvious from the derived
used in different laboratories (IZW Laboratory, probability maps that areas with similar δ2H
Saskatoon Laboratory, USGS 42). The keratin values in precipitation water yield a similar
standards of the IZW laboratory originate from probability of origin for animals. Since so-called
Swedish and Spanish sheep (wool) and isoclines, i.e., areas of similar isotopic composi-
Tanzanian goat (pelage) and have been estab- tion, follow latitude reasonably closely in the
lished using the dual-inlet method as outlined northern hemisphere, isoscape origin models
in Wassenaar and Hobson (2000, 2003). poorly resolve east-western movements of

5.5 FUTURE DIRECTIONS 131
FIGURE 5.2 Predicted geographical provenance of 14 Nyctalus noctula killed at wind turbines in eastern Germany.
Geographical areas marked green indicate areas of likely origin while gray areas indicate unlikely origin. Based on visual
inspection of 14 individual assignments generated in IsoriX (Chapter 9: Isoscape Computation and Inference of Spatial
Origins With Mixed Models Using the R package IsoriX) we decided to pool all 13 individuals for which the locality of
death could not be excluded as likely origin into the category “regional bats” and provide a group assignment as a final
plot (left graph). The individual assignment of the only migratory bat is shown in the right graph.
animals. However, it is possible to refine the non-keratinous matrices (Meier-Augenstein

isotopic approach by including a priori infor- et al., 2013; Soto et al., 2017). Further technical
mation. In a previous study, we combined the advances, possibly in compound specific anal-
isoscape origin model with assumed heading ysis of stable hydrogen isotope ratios, may
directions and travel distances in migratory stimulate more research (Chapter 7: Amino
noctule bats that were obtained from banding Acid Isotope Analysis: A New Frontier in
data (Voigt et al., 2014). This example shows Studies of Animal Migration and Foraging
that it is possible to specify the area of likely Ecology). We recommend international efforts
origin when using such priors. Other priors be undertaken to provide more reference mate-
could be a further set of stable isotopes, such as rial from migratory species during their molt-
δ34S measurements used as a proxy for distance ing period to serve as a global databank. By
to the coast (Zazzo, Monahan, Moloney, doing this, raw isotopic data would be avail-
Green, & Schmidt, 2011). able combined with spatial information (longi-
tude and latitude), which would facilitate
future efforts to refine isoscape origin models.
5.5 FUTURE DIRECTIONS This effort should include non-keratinous
mammalian materials as well. Establishing
We envision a promising future for the small-scale isoscapes in the range of a few
application of stable isotopes in the study square kilometers could help in elucidating
of movements in terrestrial mammals. spatial movements of less mobile species.
Recently, analytical developments have Lastly, isotopic tracking can be combined with
become widely available which will enable us other techniques such as GPS collaring or mul-
to analyze hydrogen stable isotope ratios in tisensor data loggers.

5.6 SUMMARY Kramer-Schadt and Alexandre Courtiol for advice regard-

ing statistical questions and the participants of the two
International Summer Schools for Stable Isotopes in
Isotopic tracking of movements has devel- Animal Ecology held in 2014 and 2016 at the IZW for dis-
oped as an important tool for mammal ecology. cussion. We thank Alexandre Courtiol, Len Wassenaar,
To date, the focus of research has been on highly and Keith Hobson for comments on an earlier draft of the
mobile smaller mammals, such as bats, and manuscript.
most studies have been based on the analysis of
δ2H values in keratinous materials. Almost all
previous isotopic tracking studies in mammals References
are based on inert body products, such as fur,
whiskers, tail hair, and claw material, yet the Arnett, E., Baerwald, E. F., Mathews, F., Rodrigues, L.,
Rodriguez-Duran, A., Rydell, J., . . . Voigt, C. C. (2016).
full potential of these body products has not yet Impacts of wind energy development on bats: A global per-
been explored. For example, δ2H values in whis- spective. Bats in the anthropocene: Conseration of bats in a
ker material have not yet been used in isoscape changing world (pp. 295323). Cham: Springer.
origin models. Overall, research in isotopic Ayliffe, L. K., Cerling, T. E., Robinson, T., West, A. G.,
tracking has yielded important insights into the Sponheimer, M., Passey, B. H., . . . Ehleringer, J. R.
(2004). Turnover of carbon isotopes in tail hair and
large-scale movements of mammals, particularly breath CO2 of horses fed an isotopically varied diet.
in research related to conservation questions. Oecologia, 139(1), 1122.
Yet, stable isotope tracking of mammals has not Baerwald, E. F., Patterson, W. P., & Barclay, R. M. R.
reached the same momentum and widespread (2014). Origins and migratory patterns of bats killed by
use as for birds and insects. Currently, we wind turbines in southern Alberta: Evidence from
stable isotopes. Ecosphere, 5(9), 117.
foresee four important developments that will Birchall, J., O’Connel, T. C., Heaton, T. H. E., & Hedges,
foster new research in the area of isotopic track- R. E. M. (2005). Hydrogen isotope ratios in animal body
ing of animals in general and that of mammals protein reflect trophic level. Journal of Animal Ecology,
in particular: New instruments for analyzing 74, 877881.
stable hydrogen isotope ratios will facilitate Britzke, E. R., Loeb, S. C., Hobson, K. A., Romanek, C. S.,
& Vonhof, M. J. (2009). Using hydrogen isotopes to
research on non-keratinous matrices, the avail- assign origins of bats in the eastern United States.
ability of international organic isotopic stan- Journal of Mammalogy, 90(3), 743751.
dards will make measurements of hydrogen Britzke, E. R., Loeb, S. C., Romanek, C. S., Hobson, K. A.,
stable isotope ratios comparable across studies, & Vonhof, M. J. (2012). Variation in catchment areas of
the addition of more reference data for local Indiana bat (Myotis sodalis) hibernacula inferred from
stable hydrogen (δ2H) isotope analysis. Canadian Journal
food webs and for consumer tissues will facili- of Zoology, 90(10), 12431250.
tate the interpretation of data in future and the Buchmann, N., Brooks, J. R., & Ehleringer, J. R. (2002).
development of new software tools such as Predicting daytime carbon isotope ratios of atmospheric
IsoriX and IsoMap will enable any researcher to CO2 within forest canopies. Functional Ecology, 16(1),
develop assignment models with ease. In sum- 4957.
Cerling, T. E., Wittemyer, G., Ehleringer, J. R., Remien,
mary, we envision a multitude of possibilities to C. H., & Douglas-Hamilton, I. (2009). History of
engage in isotope tracking studies of mammals animals using isotope records (HAIR): A 6-year
and therefore encourage researchers to follow in dietary history of one family of African elephants.
this direction. Proceedings of the National Academy of Sciences, 106(20),
80938100.
Cerling, T. E., Wittemyer, G., Rasmussen, H. B., Vollrath,
F., Cerling, C. E., Robinson, T. J., & Douglas-Hamilton,
Acknowledgments I. (2006). Stable isotopes in elephant hair document
We would like to thank the technical staff of the migration patterns and diet changes. Proceedings of the
Stable Isotope Laboratory at the Leibniz Institute for Zoo National Academy of Sciences of the United States of
and Wildlife Research for help with the analysis, Stephanie America, 103(2), 371373.

REFERENCES 133
Codron, J., Kirkman, K., Duffy, K. J., Sponheimer, M., Lee- Fraser, E. E., McGuire, L. P., Eger, J. L., Longstaffe, F. J., &
Thorp, J. A., Ganswindt, A., Clauss, M., & Codron, D. Fenton, M. B. (2012). Evidence of latitudinal migration
(2013). Stable isotope turnover and variability in tail hairs in tri-colored bats, Perimyotis subflavus. PLoS One, 7(2),
of captive and free-ranging African elephants (Loxodonta e31419.
africana) reveal dietary niche differences within popula- Hayes, M. A. (2013). Bats killed in large numbers at United
tions. Canadian Journal of Zoology, 91(3), 124134. States wind energy facilities. BioScience, 63(12),
Crawford, K., Mcdonald, R. A., & Bearhop, S. (2008). 975979.
Applications of stable isotope techniques to the ecology Henaux, V., Powell, L. A., Hobson, K. A., Nielsen, C. K., &
of mammals. Mammal Review, 38(1), 87107. LaRue, M. A. (2011). Tracking large carnivore dispersal
Cryan, P. M., Bogan, M. A., Rye, R. O., Landis, G. P., & using isotopic clues in claws: An application to cougars
Kester, C. L. (2004). Stable hydrogen isotope analysis across the Great Plains. Methods in Ecology and
of bat hair as evidence for seasonal molt and long-distance Evolution, 2(5), 489499.
migration. Journal of Mammalogy, 85(5), 9951001. Lehnert, L. S., Kramer-Schadt, S., Schönborn, S., Lindecke,
Cryan, P. M., Stricker, C. A., & Wunder, M. B. (2014). O., Niermann, I., & Voigt, C. C. (2014). Wind farm facil-
Continental-scale, seasonal movements of a heterother- ities in Germany kill noctule bats from near and far.
mic migratory tree bat. Ecological Applications, 24(4), PLoS One, 9(8), e103106.
602616. Ling, J. K. (1970). Pelage and molting in wild mammals
Dammhahn, M., & Goodman, S. M. (2014). Trophic niche with special reference to aquatic forms. The Quarterly
differentiation and microhabitat utilization revealed by Review of Biology, 45(1), 1654.
stable isotope analyses in a dry-forest bat assemblage at Männel, T. T., Auerswald, K., & Schnyder, H. (2007).
Ankarana, northern Madagascar. Journal of Tropical Altitudinal gradients of grassland carbon and nitrogen
Ecology, 30(2), 97109. isotope composition are recorded in the hair of grazers.
Dunnett, M. (2005). The diagnostic potential of equine hair: Global Ecology and Biogeography, 16, 583592.
A comparative review of hair analysis for assessing Martinez del Rio, C. M., & Carleton, S. A. (2012). How fast
nutritional status, environmental poisoning, and drug and how faithful: The dynamics of isotopic incorpo-
use and abuse. In J. Pagan, & R. J. Georg (Eds.), ration into animal tissues. Journal of Mammalogy, 93(2),
Advances in equine nutrition III (pp. 85106). 353359.
Kentucky: Kentucky Equine Research. McLaren, A. A. D., Crawshaw, G. J., & Patterson, B. R. (2015).
Ehleringer, J. R., Bowen, G. J., Chesson, L. A., West, A. G., Carbon and nitrogen discrimination factors of wolves
Podlesak, D. W., & Cerling, T. E. (2008). Hydrogen and and accuracy of diet inferences using stable isotope analy-
oxygen isotope ratios in human hair are related to sis. Wildlife Society Bulletin, 39, 788796.
geography. Proceedings of the National Academy of Meier-Augenstein, W. (2011). Stable isotope forensics: An
Sciences, 105(8), 27882793. introduction to the forensic application of stable isotope anal-
Erzberger, A., Popa-Lisseanu, A. G., Lehmann, G. U., & ysis (Vol. 3). Chichester: John Wiley & Sons Ltd.
Voigt, C. C. (2011). Potential and limits in detecting altitu- Meier-Augenstein, W., Hobson, K. A., & Wassenaar, L. I.
dinal movements of bats using stable hydrogen isotope (2013). Critique: Measuring hydrogen stable isotope
ratios of fur keratin. Acta Chiropterologica, 13(2), 431438. abundance of proteins to infer origins of wildlife, food
Fisher, D. D., Wilson, L. L., Leach, R. M., & Scholz, R. W. and people. Bioanalysis, 5(7), 751767.
(1985). Switch hair as an indicator of magnesium and Mirón, L. L. M., Herrera, L. G. M., Ramı́rez, N., & Hobson,
copper status of beef cows. American Journal of K. A. (2006). Effect of diet quality on carbon and nitro-
Veterinary Research, 46(11), 22352240. gen turnover and isotopic discrimination in blood of a
Fleming, T. H., Nuñez, R. A., & Sternberg, L. D. S. L. New World nectarivorous bat. Journal of Experimental
(1993). Seasonal changes in the diets of migrant and Biology, 209(3), 541548.
non-migrant nectarivorous bats as revealed by carbon Moriarty, K. M., Zielinski, W. J., Gonzales, A. G., Dawson,
stable isotope analysis. Oecologia, 94(1), 7275. T. E., Boatner, K. M., Wilson, C. A., . . . Schwartz, M. K.
Fraser, E. E., Brooks, D., & Longstaffe, F. J. (2017). (2009). Wolverine confirmation in California after
Stable isotope investigation of the migratory behavior of nearly a century: Native or long-distance immigrant?.
silver-haired bats (Lasionycteris noctivagans) in eastern Northwest Science, 83(2), 154162.
North America. Journal of Mammalogy, 98(5), 12251235. Mutirwara, R., Radloff, F. G., & Codron, D. (2017). Growth
Fraser, E. E., Longstaffe, F. J., & Fenton, M. B. (2013). Moulting rate and stable carbon and nitrogen isotope trophic dis-
matters: The importance of understanding moulting cycles crimination factors of lion and leopard whiskers. Rapid
in bats when using fur for endogenous marker analysis. Communications in Mass Spectrometry. Available from
Canadian Journal of Zoology, 91(8), 533544. https://doi.org/10.1002/rcm.8003.

Ossa, G., Kramer-Schadt, S., Peel, A. J., Scharf, A. K., & Segers, J. L., & Broders, H. G. (2015). Carbon (δ13C) and
Voigt, C. C. (2012). The movement ecology of the nitrogen (δ15N) stable isotope signatures in bat fur indi-
straw-colored fruit bat, Eidolon helvum, in sub-Saharan cate swarming sites have catchment areas for bats from
Africa assessed by stable isotope ratios. PLoS One, 7(9), different summering areas. PLoS One, 10(4), e0125755.
e45729. Sharp, Z. D., Atudorei, V., Panarello, H. O., Fernández, J., &
Peterson, B. J., & Fry, B. (1987). Stable isotopes in ecosys- Douthitt, C. (2003). Hydrogen isotope systematics of
tem studies. Annual Review of Ecology and Systematics, 18 hair: Archeological and forensic applications. Journal of
(1), 293320. Archaeological Science, 30(12), 17091716.
Pietsch, S. J., Hobson, K. A., Wassenaar, L. I., & Tütken, T. Soto, D. X., Koehler, G., Wassenaar, L. I., & Hobson, K. A.
(2011). Tracking cats: Problems with placing feline car- (2017). Re-evaluation of the hydrogen stable isotopic
nivores on δ18O, δD isoscapes. PLoS One, 6(9), e24601. composition of keratin calibration standards for wildlife
Pietsch, S. J., & Tütken, T. (2016). Oxygen isotope composi- and forensic science applications. Rapid Communications
tion of North American bobcat (Lynx rufus) and puma in Mass Spectrometry. Available from https://doi.org/
(Puma concolor) bone phosphate: Implications for 10.1002/rcm.7893.
provenance and climate reconstruction. Isotopes in Spurr, E. B. (2002). Rhodamine B as a systemic hair marker
Environmental and Health Studies, 52(12), 164184. for assessment of bait acceptance by stoats (Mustela
Podlesak, D. W., Torregrossa, A. M., Ehleringer, J. R., erminea). New Zealand Journal of Zoology, 29(3), 187194.
Dearing, M. D., Passey, B. H., & Cerling, T. E. (2008). Sullivan, A. R., Bump, J. K., Kruger, L. A., & Peterson, R. O.
Turnover of oxygen and hydrogen isotopes in the body (2012). Bat-cave catchment areas: Using stable isotopes
water, CO2, hair, and enamel of a small mammal. (δD) to determine the probable origins of hibernating
Geochimica et Cosmochimica Acta, 72(1), 1935. bats. Ecological Applications, 22(5), 14281434.
Popa-Lisseanu, A. G., Sörgel, K., Luckner, A., Wassenaar, Thomas, S. M., & Crowther, T. W. (2015). Predicting rates
L. I., Ibáñez, C., Kramer-Schadt, S., . . . Mysłajek, R. W. of isotopic turnover across the animal kingdom: A syn-
(2012). A triple-isotope approach to predict the breed- thesis of existing data. Journal of Animal Ecology, 84(3),
ing origins of European bats. PLoS One, 7(1), e30388. 861870.
Pylant, C. L., Nelson, D. M., & Keller, S. R. (2014). Voigt, C. C. (2010). Insights into strata use of forest animals
Stable hydrogen isotopes record the summering using the ‘canopy effect’. Biotropica, 42(6), 634637.
grounds of eastern red bats (Lasiurus borealis). PeerJ, 2, Voigt, C. C., Grasse, P., Rex, K., Hetz, S. K., & Speakman,
e629. Available from https://doi.org/10.7717/ J. R. (2008). Bat breath reveals metabolic substrate use
peerj.629. in free-ranging vampires. Journal of Comparative
Reuter, K. E., Wills, A. R., Lee, R. W., Cordes, E. E., & Physiology B, 178(1), 916.
Sewall, B. J. (2016). Using stable isotopes to infer the Voigt, C. C., Helbig-Bonitz, M., Kramer-Schadt, S., &
impacts of habitat change on the diets and vertical Kalko, E. K. (2014). The third dimension of bat migra-
stratification of frugivorous bats in Madagascar. PLoS tion: Evidence for elevational movements of
One, 11(4), e0153192. Miniopterus natalensis along the slopes of Mount
Rex, K., Michener, R., Kunz, T. H., & Voigt, C. C. (2011). Kilimanjaro. Oecologia, 174(3), 751764.
Vertical stratification of Neotropical leaf-nosed bats Voigt, C. C., & Kelm, D. H. (2006). Host preference of the com-
(Chiroptera: Phyllostomidae) revealed by stable carbon mon vampire bat (Desmodus rotundus; Chiroptera) assessed
isotopes. Journal of Tropical Ecology, 27(3), 211222. by stable isotopes. Journal of Mammalogy, 87(1), 16.
Robertson, A., McDonald, R. A., Delahay, R. J., Kelly, S. D., Voigt, C. C., Lehmann, D., & Greif, S. (2015). Stable isotope
& Bearhop, S. (2013). Whisker growth in wild Eurasian ratios of hydrogen separate mammals of aquatic and
badgers Meles meles: Implications for stable isotope and terrestrial food webs. Methods in Ecology and Evolution,
bait marking studies. European Journal of Wildlife 6(11), 13321340.
Research, 59(3), 341350. Voigt, C. C., Lehnert, L. S., Popa-Lisseanu, A. G.,
Rohrig. (2014). Windenergie Report Deutschland (2014) Ciechanowski, M., Estók, P., Gloza-Rausch, F., . . .
Fraunhofer-Institut für Windenergie und Teige, T. (2014). The trans-boundary importance of arti-
Energiesystemtechnik IWES Kassel. ISBN 978-3- ficial bat hibernacula in managed European forests.
8396-0854-8. Biodiversity and Conservation, 23(3), 617631.
Schwertl, M., Auerswald, K., & Schnyder, H. (2003). Voigt, C. C., Lindecke, O., Schönborn, S., Kramer-Schadt,
Reconstruction of the isotopic history of animal diets S., & Lehmann, D. (2016). Habitat use of migratory bats
by hair segmental analysis. Rapid Communications in killed during autumn at wind turbines. Ecological
Mass Spectrometry, 17(12), 13121318. Applications, 26(3), 771783.

FURTHER READING 135
Voigt, C. C., Matt, F., Michener, R., & Kunz, T. H. (2003). Wittemyer, G., Cerling, T. E., & Douglas-Hamilton, I. (2009).
Low turnover rates of carbon isotopes in tissues of two Establishing chronologies from isotopic profiles in seri-
nectar-feeding bat species. Journal of Experimental ally collected animal tissues: An example using tail hairs
Biology, 206(8), 14191427. from African elephants. Chemical Geology, 267(1), 311.
Voigt, C. C., Popa-Lisseanu, A. G., Niermann, I., & Wunder, M. (2012). Determining geographic patterns of
Kramer-Schadt, S. (2012). The catchment area of wind migration and dispersal using stable isotopes in kera-
farms for European bats: A plea for international regu- tins. Journal of Mammalogy, 93(2), 360367.
lations. Biological Conservation, 153, 8086. Wurster, C. M., McFarlane, D. A., & Bird, M. I. (2007).
Voigt, C. C., Schneeberger, K., & Luckner, A. (2013). Spatial and temporal expression of vegetation and
Ecological and dietary correlates of stable hydrogen iso- atmospheric variability from stable carbon and nitrogen
tope ratios in fur and body water of syntopic tropical isotope analysis of bat guano in the southern United
bats. Ecology, 94(2), 346355. States. Geochimica et Cosmochimica Acta, 71(13),
Voigt, C. C., Voigt-Heucke, S. L., & Kretzschmar, A. S. 33023310.
(2012). Isotopic evidence for seed transfer from succes- Zazzo, A., Monahan, F. J., Moloney, A. P., Green, S., &
sional areas into forests by short-tailed fruit bats Schmidt, O. (2011). Sulphur isotopes in animal hair
(Carollia spp.; Phyllostomidae). Journal of Tropical track distance to sea. Rapid Communications in Mass
Ecology, 28(2), 181186. Spectrometry, 25, 23712378.
Wassenaar, L. I., & Hobson, K. A. (2000). Improved Ziegler, S., Merker, S., Streit, B., Boner, M., & Jacob, D. E.
method for determining the stable-hydrogen isotopic (2016). Towards understanding isotope variability in
composition (δD) of complex organic materials of envi- elephant ivory to establish isotopic profiling and
ronmental interest. Environmental Science & Technology, source-area determination. Biological Conservation, 197,
34(11), 23542360. 154163.
Wassenaar, L. I., & Hobson, K. A. (2003). Comparative
equilibration and online technique for determination of
non-exchangeable hydrogen of keratins for use in ani- Further Reading
mal migration studies. Isotopes in Environmental and
Health Studies, 39(3), 211217. Coplen, T. B., & Qi, H. (2012). USGS42 and USGS43:
West, A. G., Ayliffe, L. K., Cerling, T. E., Robinson, T. F., Human-hair stable hydrogen and oxygen isotopic refer-
Karren, B., Dearing, M. D., & Ehleringer, J. R. (2004). ence materials and analytical methods for forensic sci-
Short-term diet changes revealed using stable carbon ence and implications for published measurement
isotopes in horse tail-hair. Functional Ecology, 18(4), results. Forensic Science International, 214(1), 135141.
616624. Fraser, K. C., McKinnon, E. A., & Diamond, A. W. (2010).
West, J. B., Bowen, G. J., Cerling, T. E., & Ehleringer, J. R. Migration, diet, or molt? Interpreting stable-hydrogen
(2006). Stable isotopes as one of nature’s ecological isotope values in Neotropical bats. Biotropica, 42(4),
recorders. Trends in Ecology & Evolution, 21(7), 408414. 512517.

C H A P T E R
6
Isotopic Tracking of Marine Animal
Movement
Clive N. Trueman and Katie St John Glew
University of Southampton, Southampton, United Kingdom
6.1 INTRODUCTION biomass of between 1 and 10 billion tonnes

(Irigoien et al., 2014).
Seventy percent of the Earth’s surface is Quantifying and describing migration of
covered by water, and 99.8% of the marine animals is especially important, as fish-
habitable volume of the planet, excluding the ing is the only remaining large-scale hunting
aerial realm is marine (Bauer & Hoye, 2014; of wild animals. Global total capture fishery
Dawson, 2012). Stop and read that again: production was 93 million tonnes in 2014, with
99.8% of the habitable volume of the planet 81.5 million tonnes coming from marine fisher-
provides an awful lot of physical space for ani- ies (FAO, 2016). Fisheries and aquaculture pro-
mals to move around in! Migrations connect vide around 17% of the world’s protein for
habitat, direct fluxes of nutrients and facilitate human use and may provide more than 70%
gene flow (Bauer & Hoye, 2014). Migration is a of human protein intake in some coastal and
common behavior across most habitats, but island countries (FAO, 2016). Fisheries and
organized directed movements are especially aquaculture also support the livelihoods of
common in animals living in water (Rodhouse, 10%12% of the world’s human population
2004; Secor, 2015a). The Arctic tern (Sterna (FAO, 2016). Industrial fishing has the capacity
paradisaea) conducts the longest migrations of to significantly deplete local, regional, and
any animal on the planet, migrating between global fish populations, with consequences for
Greenland and Antarctica, a distance of both humans and ecosystems. Knowledge of
77,000 km each year, with lifetime movements the spatial structure of marine animal commu-
potentially exceeding 2 million km (Egevang nities is urgently needed for effective
et al., 2010). However, the largest migration by ecosystem-based spatial management of the
biomass on earth is the daily movement of world’s seas and oceans. Marine environments
mesopelagic fishes between deep-waters and are relatively inaccessible and difficult to
the sea surface with an uncertain but large observe for humans, consequently life histories

138 6. ISOTOPIC TRACKING OF MARINE ANIMAL MOVEMENT
of marine animals other than a few commer- are also highly complex, offering a diverse
cially or culturally relevant species are relat- suite of behavioral opportunities and potential
vely poorly known. For these reasons, indirect migration between habitats. Finally, vertical
evidence about space use, movements and water column migrations are ubiquitous features
migrations of marine animals is particularly of aquatic ecosystems on daily, seasonal, and
valuable. Fortunately, marine systems offer a ontogenetic scales.
good range of isotopic gradients across which Many air breathing, terrestrially reprodu-
animals move, providing profitable targets cing marine animals such as seabirds, pinni-
for isotope-based reconstructions of marine peds, and sea turtles are restricted to oceanic
animal migrations. islands or isolated beaches for breeding, and
then disperse over wide regions (Boyd, 2004;
Phillips, Bearhop, McGill, & Dawson, 2009).
Behavior and foraging locations during the
6.1.1 Migration in Aquatic
nonbreeding periods are often poorly under-
Environments stood, but if tissue growth occurs outside of
Drivers for directed migration include sea- the breeding area, tissue isotopes can provide
sonal variations in the distribution of food valuable evidence of behavior during cryptic
resources, reproduction and changes in habi- life stages (Reich, Bjorndal, & Bolten, 2007).
tat requirements. In three-dimensional fluid Nomadic movements are also extremely com-
environments like the sea, manipulating prey mon in marine systems as animals track
is difficult, and predators cannot easily pro- hydrographic features such as frontal systems
cess food items larger than their gape width, that move in time and space.
leading to food webs that are structured by In this chapter we will draw on examples
size (what you can eat) rather than species using isotopes to study migration as defined in
(Jennings, Barnes, Sweeting, & Polunin, Chapter 1, Animal Migration: A Context for
2008). As many marine organisms increase in Using New Techniques and Approaches; the
size over several orders of magnitude, adults methods and approaches described can, of
and juveniles have distinctly different diets course, also be used to investigate space use
and predators, and consequently, distinctive during foraging.
habitat requirements. Ontogenetic migrations
and repeated migrations associated with
reproduction and spawning are, therefore,
6.1.2 Using Stable Isotopes to Infer
common in marine organisms (Rodhouse,
2004; Secor, 2015a). Temperate marine envir-
Migration in Marine Settings
onments are also highly seasonal: pelagic pri- In the simplest sense, migration or move-
mary producers are all small, ephemeral ment may be inferred when the isotopic com-
organisms, and plankton biomass is limited position of an animal’s tissues is inconsistent
by light and water temperature in winter, with isotopic compositions expected from diet
and by grazing pressure in summer, result- in area in which it was sampled (see
ing in strong pulses of production (i.e., Chapter 3: Isoscapes for Terrestrial Migration
blooms). Many marine predators conduct Research and Chapter 4: Application of
seasonal latitudinal feeding migrations fol- Isotopic Methods to Tracking Animal
lowing algal bloom conditions, and the asso- Movements for more detailed discussions).
ciated food webs, into high-latitude regions. Ideally, the area where the tissue could have
Coastal and marginal marine environments been grown can be inferred from spatially

6.2 PART 1: MECHANISM, SPATIAL STRUCTURE, AND ISOSCAPES 139
explicit reference samples and/or derived iso- literature is strongly biased toward studies of
scapes (see Chapter 3: Isoscapes for Terrestrial marine mammals. Of necessity we omit many
Migration Research). When considering marine excellent and important studies of marine
environments, spatial scale is a major consider- migration using stable isotopes, and therefore,
ation. The open oceans are vast and remote encourage the reader to voyage more exten-
spaces that are logistically challenging to visit, sively through the literature.
and collection of spatially explicit reference
data needed to develop isoscapes is exception-
ally challenging. Coastal and shelf seas are 6.2 PART 1: MECHANISM, SPATIAL
generally accessible by boat, and physical sam- STRUCTURE, AND ISOSCAPES
ples can be used to create spatial reference
datasets or isoscapes for direct mapping or the 6.2.1 Isotopic Variability in Marine
development of calibration algorithms for Systems
modeling (see Chapter 3: Isoscapes for
Terrestrial Migration Research). Boat-based Excellent reviews of the mechanisms under-
sampling at spatially relevant scales is less pinning spatial variation in stable isotopes of
practical in the open oceans. Fortunately the marine and estuarine systems are available
mechanisms responsible for large-scale spatial (Bouillon, Connolly, & Gillikin, 2011; McMahon,
variability in stable oxygen, carbon, and nitro- Hamady, & Thorrold, 2013; Michener & Lajtha,
gen isotope compositions of seawater and phy- 2008; Ramos & Gonzalez-Solis, 2012; Ryabenko,
toplankton communities are relatively well 2013), and readers are encouraged to consult
understood, and are related to oceanographic these sources. Isoscapes have been produced
variables that are measured and modeled at across the marine environment using a variety
high spatio-temporal resolution (e.g., of animal tissues, over numerous spatial and
LeGrande & Schmidt, 2006; Yool, Popova, & temporal scales (Fig. 6.1, Table 6.1), and for
Anderson, 2013). Models can be used to pre- many different ecological applications.
dict the isotopic composition and variability in However, despite the spatial coverage outlined
time and space across ocean basins where in Fig. 6.1, isotopic variability in most oceanic
physical samples are unavailable. In this chap- regions is typically represented using interpola-
ter we discuss using sample-based and model- tion between data points that are irregularly dis-
based predictions of isotopic variability to tributed in space and time (see Chapter 3:
study marine animal migration. Isoscapes for Terrestrial Migration Research and
Part 1 of the chapter discusses biogeochemi- Chapter 8: Design and Analysis for Isotope-
cal mechanisms giving rise to spatial patterns Based Studies of Migratory Animals for a thor-
of stable isotopes in marine systems, empha- ough discussion of spatial isoscape modeling).
sizing differences inherent in marine compared
to terrestrial environments and outlining some
6.2.2 Oxygen, Hydrogen, and Strontium
of the complicating factors that are particularly
important to consider for studying marine sys-
Isotopes
tems. In Part 2, we draw on the extensive and The majority of stable isotope-based investi-
growing literature to provide examples of the gations of terrestrial animal movements have
use of isotopes to reconstruct migrations in drawn on spatial variations in the isotopic
marine animals. We chose studies representing composition of oxygen and hydrogen, driven
a range of aquatic organisms, tissue types, and by predictable Rayleigh distillation processes
scales of movement, while recognizing that the and underpinned by a relatively extensive

δ13C isoscape
δ15N isoscape
δ13C, δ15N isoscapes
18
δ O isoscape
FIGURE 6.1 Areas of the global ocean for which isoscape models have been published (citations and details in
Table 6.1).
network of precipitation isotope sampling sta- oxygen, hydrogen, and most dissolved ions,
tions (Bowen, 2010, Chapter 3: Isoscapes for limiting the signal available to track animal
Terrestrial Migration Research). Strontium iso- movements at small scales. The isotopic com-
tope ratios (87Sr/86Sr) have also been used position of oxygen in sea surface waters are
effectively based on the relationship between relatively well constrained (LeGrande &
tissue strontium compositions and the under- Schmidt, 2006) and largely vary as a function
lying geology (Bataille & Bowen, 2012; of salinity. At high latitudes, discharge of iso-
Brennan et al., 2015; Stuben, Berner, topically light river water and glacial meltwa-
Chandrasekharam, & Karmakar, 2003). These ter to the sea results in lower sea surface water
applications connect the isotopic composition δ18O values, and many northern polar latitudes
of animal tissues to that of local environmental show seawater δ18O values lower than 22m
water or diet (see Chapter 3: Isoscapes for (Fig. 6.2A). Higher δ18O values are observed in
Terrestrial Migration Research, Chapter 4: seawater in highly evaporative regions where
16
Application of Isotopic Methods to Tracking O is preferentially removed, such as the sub-
Animal Movements, and Chapter 8: Design tropical gyres and in semi-enclosed basins
and Analysis for Isotope-Based Studies of (seawater δ18O values in the North Atlantic
Migratory Animals). The large volume of the Subtropical gyre and Mediterranean Sea range
oceans effectively eliminates small-scale or dis- between 10.5m and 22m, Fig. 6.2A). Apart
crete spatial variations in isotopic ratios of from these extremes, however, the range in

TABLE 6.1 Summary of Marine Isoscape Models Known to the Authors in the Available Scientific Literature
Even
Spatial
Bulk/ Coverage
Continuous/ CSIA- Baseline Model/in situ of in situ
Location Isotopes Author Discrete? AA Organism Samples Method Variance? Time Scale Data?
GLOBAL
Global δ15N Somes et al. Continuous Bulk POM Model Climate model 1 marine No Simulate over NA
(2010, 2013) ecosystem model different time
scales
Global δ13C, δ13C Hofmann Continuous Bulk DIC, POC Model 3D model of No Simulate over NA
-DIC et al. (2000) phytoplankton and different time
zooplankton community scales
dynamics
Global δ13C, δ13C Tagliabue Continuous Bulk DIC, POC Model Physical model 1 PISCES No Simulate over NA
-DIC and Bopp ocean-ecosystem model different time
(2008) (Aumont and Bopp, 2006) scales
Global δ13C -DIC Schmittner Continuous Bulk DIC Model Physical 1 marine No 19902005 NA
et al. (2013) ecosystem model
(Schmittner et al., 2008)
Global δ13C Magozzi Continuous Bulk Plankton Model Coupled physics- Variance 200110, NA
et al. (2017) and Discrete biogeochemistry model reported annually
(NEMO-MEDUSA), used within averaged
to estimate carbon discrete
isotopic fractionation provinces
Global δ18O LeGrande Continuous Bulk Seawater In situ Nearest neighbor No 50 years No
and Schmidt (Schmidt et al., interpolation of
(2006) 1999) isotopically distinct
regions
Global and δ18O Trueman Continuous Bulk Seawater and fish In situ Fish otolith isotope values No 30 1 years No
Atlantic et al. (2012) otoliths seawater estimated from measured
Ocean samples 1 seawater samples
temperature (Schmidt et al., 1999), and
data SST data
ATLANTIC
Atlantic δ15N, δ13C, McMahon Continuous Bulk Seawater and In situ Ocean Data View: Data Misfits Data compiled No
Ocean δ13C-DIC, et al.(2013) zooplankton (published Interpolating Variation calculated across numerous
δ18O δ2H data) Analysis years
(Continued)
Even
Spatial
Bulk/ Coverage
Gulf of δ N, δ C
15 13
Vander Continuous Bulk Loggerhead turtles In situ GIS—ordinary kriging Variance 2011/12 Yes
Mexico Zanden (scute issue) isoscape
et al. (2015)
Gulf of δ15N, δ13C Radabaugh Continuous Bulk 3 fish species In situ Ordinary kriging No 2009/10 Yes
Mexico et al. (2013) (muscle tissue),
benthic algae and
POM
Gulf of δ15N, δ13C Radabaugh Continuous Bulk 7 fish species In situ 1 Multiple regression Variation 2009/10 Yes
Mexico and Peebles (muscle tissue), environmental models between fish plots from
(2014) benthic algae and data isotope data and mean
POM environmental variables. isotopic
Ordinary kriging of data value of
each
species
Bermuda δ15N Fourqurean Continuous Bulk Seagrass In situ Ordinary kriging No 200608 No
Platform et al. (2015)
North Sea, δ15N Jennings Continuous Bulk Queen Scallops In situ 1 Linear model used to Model 2001 No
Irish Sea and Warr environmental predict N15 values at variance
and (2003) data environmental data points assessed
English
Channel
North Sea, δ13C Barnes and Continuous Bulk Queen Scallops In situ 1 Linear model used to Model 2001 No
Irish Sea Jennings environmental predict C13 values at variance
and (2009) data environmental data points assessed
English
Channel
North Sea δ15N, δ13C MacKenzie Continuous Bulk Lion’s mane In situ 1 Ordinary kriging, linear Variance 2011 No
et al. (2014) jellyfish environmental model used to predict isoscape
data isotope values at
environmental data
points. Ordinary kriging
of modeled data
North sea δ15N, δ13C Trueman Continuous Bulk Lion’s mane In situ Ordinary kriging Variance 2015 Yes
et al. (2017) jellyfish isoscape
Baltic Sea δ18O, δ13C Torniainen Continuous Bulk Seawater 1 Atlantic In situ Otolith isotope values Simple 201011 Yes
-DIC et al. (2017) salmon estimated from predictive sensitivity
otoliths 1 salinity model using seawater analysis
data O18 and salinity data.
Ocean Data View: Data
Interpolating Variation
Analysis
NW δ15N, δ13C Ceriani et al. Continuous Bulk Loggerhead turtles In situ ARC GIS—Bayesian Cross 200913 No
Atlantic (2014, 2017) (skin tissue) kriging validation
statistics
PACIFIC
Bering, δ15N, δ13C Schell et al. Discrete Bulk Zooplankton In situ Grouped samples into No 198594 Yes
Chukchi (1998) regions separated by
and different water masses—
Beaufort average value in each area
Seas
Bering Sea δ15N Jones et al. Continuous Bulk Krill, pollock, thick- In situ Arc GIS—inverse distance No 2008/09 Yes
(2014) billed murre and weighting interpolations.
black-legged Cross shelf pattern
kittiwake muscle observed and modeled
using all species
Equatorial δ15N Graham Continuous Bulk Bigeye and In situ Isotope values normalized No Unknown No
Pacific et al. (2010) yellowfin tuna against average value for
muscle samples the species in that region,
differences plotted across
space
Eastern δ15N Olson et al. Continuous Bulk Copepods and In situ GAM models to predict No Copepods 5 2003, Yes
Tropical (2010) yellowfin tuna copepod N15 values tuna 200305 (copepods)
pacific across space. Tuna isotope
values—ordinary kriging
Pacific atoll δ18O Detjen et al. Continuous Bulk Barnacles attached In Arc GIS—Barnacle oxygen No 2011 Yes
(2015) to green sea situ 1 model isotopes predicted from (Seawater
turtles 1 seawater water isotope values and data)
isotope values SST. Contour maps
(Continued)
Even
Spatial
Bulk/ Coverage
(LeGrande &
Schmidt, 2006)
California δ15N- Vokhshoori Continuous CSIA- Mussels—but In Model isotope values and No 200910 Yes
Coast CSIA-AA and AA measuring primary situ 1 model latitude. Arc GIS—simple
McCarthy productivity interpolation
(2014)
Southern δ15N, δ13C Kurle and Continuous Bulk POM In situ Arc GIS—Inverse distance No 201213 Yes
Californian McWhorter weighting. Mapped (seasonal)
Bight (2017) distance from mean
isotopic values
Temperate δ15N, δ13C Mackey Continuous Bulk Red Algae In situ 1 Linear model used to No 2013 (summer No
Australasia et al. (2015) environmental predict C13 values at and winter)
data environmental data
points. Ocean Data
View—Data Interpolating
Variation Analysis
South West δ15N, δ13C Pethybridge Continuous Bulk Albacore Tuna In situ 1 GAM model created No 2009/10 Yes
Pacific et al. (2015) environmental using SST, Chl, and lat/
data long to predict isotope
values. Contour plots
created
South East δ15N Young et al. Continuous Bulk Yellowfin tuna In GAM model of yellowfin No Integrated over No
Pacific (2015) situ 1 model tuna and other predatory many years
fish isotope values,
adjusted with modeled
baseline (Somes et al.,
2013)
Eastern δ15N Zupcic- Continuous Bulk Surface sediment In situ Contour interpolation plot No Integrated over No
Tropical Moore et al. samples many years
pacific (2017)—data
from Tesdal
et al. (2013)
Western δ15N Houssard Continuous Bulk POM and yellowfin In situ GAM models of POM and No 19922015 No
and Central et al. (2017) and bigeye tuna tuna, using a two
Pacific dimension thin plate
regression spline fitted to
the samples by location
SOUTHERN OCEAN
Southern δ15N, δ13C Jaeger et al. Continuous Bulk Albatross blood In situ Interpolation of blood No 2008 No
Ocean (2010) plasma plasma isotopes
Southern δ13C Quillfeldt Continuous Bulk Phytoplankton and In situ Arc GIS—nearest No Integrated over No
Ocean et al. (2010) POM neighbor interpolation many years
South δ13C Ceia et al. Continuous Bulk Albatross blood In situ ARC GIS—natural No MayOct 2009 No
Georgia (2015) and plasma neighbor interpolation
(A) δ18Ow (B) δ18Ooto

4
0 2
–2 0
–4 –2
–6 –4
(C) δ13CPOM (D) δ15NPOM

10
–20 8
–22 6
–24 4
–26 2
–28 0
–30 −2
FIGURE 6.2 Estimates of global-scale spatial variation in (A) ocean water δ O values, (B) otolith δ O values, (C)
18 18
δ13C values in particulate organic matter at the surface, and (D) δ15N values in particulate organic matter at the sur-
face. Data for (A) are drawn from LeGrande, A. N. & Schmidt, G. A. (2006). Global gridded data set of the oxygen
isotopic composition in seawater. Geophysical Research Letters, 33, L12604 . Values in (B) are estimated by applying the
otolith oxygen fractionation equation from Høie, H. (2004). Temperature-dependent fractionation of stable oxygen iso-
topes in otoliths of juvenile cod (Gadus morhua L.). ICES Journal of Marine Science, 61, 243251 to the estimated water
isotope values shown in (A) and applying mean climatological sea-surface temperature estimates derived from the
NEMO-MEDUSA global ocean model (Yool et al., 2013). Values in C are derived from Magozzi, S., Yool, A., Vander
Zanden, H., Wunder, M., & Trueman, C. (2017). Using ocean models to predict spatial and temporal variation in
marine carbon isotopes. Ecosphere, 8 and are model estimates of biomass-weighted annual average δ13C values in par-
ticulate organic matter at the ocean surface. Values in D are derived from Schmittner, A. & Somes, C. J. (2016).
Complementary constraints from carbon (13C) and nitrogen (15N) isotopes on the glacial ocean’s soft-tissue biological
pump. Paleoceanography, 31, 669693 and are model estimates of annual average δ15N values in particulate organic
matter at the ocean surface.
surface seawater oxygen isotope compositions 1.5*103 years). Consequently, marine waters
is relatively small (Fig. 6.2A). Localized isoto- essentially have homogenous 87Sr/86Sr ratios.
pic gradients may be sufficient to identify ani- In marginal marine environments, such as
mal movement between distinctive water estuaries and some shallow shelf seas, mixing
masses, but the limited range of δ18O and δ2H between marine and fresh water creates salin-
values in seawater largely prevents direct pre- ity (and isotopic) gradients that produce pro-
diction of spatial origin based on tissue δ18O nounced chemical patterns. There is a wealth
and δ2H values. Similarly, the residence time of trace-element-based tracers suitable for
of strontium in seawater is approximately tracking movements of animals across salinity
4*106 years (Henderson, 1984), much longer gradients (i.e., elemental ratios of Sr, Ba, and
than the mean mixing time of the oceans (ca. other ions with low residence times in marine

waters) which have been described and 6.2.3 Processes Driving Spatial Patterns
reviewed in detail (e.g., Bouillon et al., 2011; in δ13C Values of Primary Production
Sturrock, Trueman, Darnaude, & Hunter, 2012;
Trueman, MacKenzie, & Palmer, 2012). The carbon isotopic composition of marine
While δ18O values of seawater are relatively phytoplankton is largely controlled by the car-
homogenous, stable isotopes of oxygen are bon isotopic composition of dissolved inor-
fractionated during incorporation into biomin- ganic carbon (DIC) in ambient water, modified
erals (i.e., aragonite and calcite in shells and by isotopic fractionation of carbon isotopes
otoliths), and the magnitude of isotope frac- during photosynthesis. The extent of fraction-
tionation is directly related to water tempera- ation of 13C/12C ratios during photosynthesis
ture. Lifetime or seasonal movements across in phytoplankton is driven by the concentra-
water temperature gradients can, therefore, be tion and isotopic composition of dissolved
reconstructed from oxygen isotopic patterns inorganic carbon in seawater, phytoplankton
obtained from cross-sections of incrementally growth rates, and cell sizes (Laws, Popp,
grown biominerals. The isotopic composition Bidigare, Kennicutt, & Macko, 1995; Rau,
of oxygen in a biomineral like fish otolith ara- Riebesell, & Wolf-Gladrow, 1996). Most of the
gonite is related to the δ18O of seawater and variables influencing δ13C values in phyto-
water temperature through a linear relation- plankton are indirectly related to sea-surface
ship of the form temperature (SST), and spatial variations in
δ13C values of oceanic phytoplankton broadly
δ18 Ootolith 2 δ18 Owater 5 B 2 AT (6.1) follow temperature gradients, with lower δ13C
values at high latitudes (Fig. 6.2C). Nutrient
where A and B are empirically determined
availability also influences phytoplankton δ13C
coefficients (Høie, 2004) and T is the water
values by influencing the growth rate and tax-
temperature. Biomineral δ18O values show
onomic composition of phytoplankton commu-
more spatial variation than seawater δ18O
nities. Regions where phytoplankton growth is
values, largely reflecting differences in water
limited by nutrient availability such as the sub-
temperature, but also influenced by water
tropical oceanic gyres have relatively low δ13C
salinity (Fig. 6.2B). Experimentally determined
phytoplankton values for a given water tem-
values of A and B appear to vary significantly
perature, while upwelling regions supporting
among species (Godiksen, Svenning,
high plankton growth rates have higher δ13C
Dempson, Storm-Suke, & Power, 2010; Høie,
values (Magozzi, Yool, Vander Zanden,
2004; Kalish, 1991; Stephenson, 2001), so that
Wunder, & Trueman, 2017). Spatio-temporal
species-specific determination of A and B coef-
variability in δ13C values of phytoplankton has
ficients are required. If water temperature and
been modeled (Hofmann et al., 2000; Magozzi
water isotopic compositions are known, it is
et al., 2017; Schmittner & Somes, 2016;
possible to estimate differences in otolith δ18O
Tagliabue & Bopp, 2008). Isotope-enabled
values between distinct water masses, and
models based on differing underlying general
therefore, track potential movements across
circulation-biogeochemical models and differ-
temperature and/or salinity gradients.
ing implementations of the equations governing
Biomineral-hosted isotope methods are
carbon isotope fractionation produce relatively
restricted to animals with biomineralized tis-
similar estimates of the spatial distribution of
sues, and are largely limited to reconstructing
δ13C values. The similarity between model pre-
movements across large thermal and/or salin-
dictions highlights the fact that the first-order
ity gradients.
controls on spatial distributions of δ13C values

appear relatively simple and are closely corre- sediments, where microbes use nitrate (NO32)
lated with SST (Magozzi et al., 2017). instead of dissolved O2 as the electron acceptor
during respiration, converting it to gaseous
N2O and N2 which can degas to the atmo-
6.2.4 Nitrogen Isotopes sphere. Denitrification strongly discriminates
Nitrogen in marine environments occurs in against 15N, with kinetic isotope effects
four main forms, e.g., nitrate (NO32), nitrite between 20m30m (Somes et al., 2010) leading
(NO22), ammonia (NH4), and N2 gas, and the to a pronounced enrichment of 15N in the
15
N/14N ratios are fractionated during conver- remaining dissolved nitrate pool. Water col-
sion of one form to another. Consequently, the umn denitrification primarily occurs in ocean
interrelated processes controlling spatial and regions with low-dissolved oxygen concentra-
temporal variations in the N isotopic composition such as the eastern tropical North Pacific,
tion of marine primary production are far the eastern tropical South Pacific, the northern
more complex than those described above for Arabian Sea, and the Gulf of Mexico “Dead
carbon. The main processes of the marine Zone” where atmospheric gas exchange rates
nitrogen cycle and their influences on δ15N are too low to meet aquatic oxygen demand
values of primary production are well (Fig. 6.2D).
described (Ryabenko, 2013). Briefly diazo- Organisms such as phytoplankton may
trophs are organisms such as the cyanobacte- assimilate nitrogenous nutrients (ammonium,
rium Trichodesmium that fix atmospheric N2 nitrate, or nitrite) directly from the water col-
gas dissolved in water. Nitrogen fixation has a umn. Dissolved nitrate is rapidly assimilated
relatively small kinetic isotope effect (22m to by phytoplankton, hence, concentrations of
12m) (Ryabenko, 2013); therefore, newly fixed nitrate in the ocean surface are generally very
nitrogen has δ15N values close to atmospheric low. Assimilation of nitrogen is accompanied
nitrogen (by definition 0m). Nitrogen fixation by a positive kinetic isotope effect which is
is energetically costly compared to assimilation larger for ammonia assimilation (114m to
of ammonia, and therefore, fixation is likely to 127m) than for nitrate (5m to 10m) (Ryabenko,
be inhibited where concentrations of fixed 2013). Microbial nitrification is the oxidization
nitrogen are relatively high due to preferential of reduced ammonia (NH41) to nitrite (NO22)
use of alternative N sources as a primary and nitrate (NO32). The kinetic isotope effect
nutrient. The extent of nitrogen fixation is, associated with nitrification is complex, as
therefore, highest in warm, well-lit ocean multiple pathways can be utilized under dif-
regions with relatively low amounts of fixed fering oxygenation conditions or microbial
nitrogen and potentially areas in receipt of strains. Experimental determination of net iso-
atmospheric iron nutrient deposition (Somes topic fractionation effects associated with cul-
et al., 2010). The subtropical Atlantic Ocean tures of ammonia-oxidizing bacteria range
appears to be a region of relatively intense from 14.2m to 38.2m (Casciotti, McIlvin, &
nitrogen fixation, and therefore, has relatively Buchwald, 2010). By contrast, experimental
low δ15N values in seawater particulate determination of the isotopic effect associated
organic matter (POM) (Fig. 6.2D). with nitrite oxidation to nitrate shows a nega-
Addition of nitrogen (oxidized forms) to the tive kinetic isotopic effect (212.8m). The com-
oceanic N pool is balanced by loss of nitrogen bined kinetic isotopic effect of nitrification
through anaerobic microbial denitrification. from ammonia to nitrate may vary depending
Denitrification largely occurs in sub- to anoxic on the external conditions and microbial com-
conditions in the water column or in the munity, but a global average isotope effect

associated with nitrate assimilation is fre- Niño-Torres, Gallo-Reynoso, Galván-Magaña,
quently estimated to be 5m (Somes et al., 2010), Escobar-Briones, & Macko, 2006).
similar to the average enrichment of oceanic As little S fractionation occurs between die-
nitrogen δ15N values compared to atmospheric tary trophic levels (0m to 23m from marine
N2. The processes above altogether combine to phytoplankton to predators), S isotopic gradi-
produce systematic variations in spatial distri- ents are also observed in coastal consumers
butions of δ15N in POM values. Atlantic foraging throughout and between salinity gra-
waters, and particularly, sub-tropical Atlantic dients, and foraging on different primary pro-
waters are characterized by relatively low δ15N ducers at the base of the food web. Thus δ34S
POM values, reflecting high rates of nitrogen ratios are often used to investigate coastal and
fixation, balanced by relatively high δ15N POM estuarine resource use and connectivity
values in the eastern Pacific and northern between marine and freshwater ecosystems
Arabian Sea associated with denitrification (Lott, Meehan, & Heath, 2003; MacAvoy,
(Fig. 6.2D). Macko, & Garman, 1998; Weber et al., 2002).
6.2.5 Sulfur Isotopes 6.2.6 Multiple Isotopic Variation in

Shelf Seas, Coasts, and Estuaries
In open-ocean environments, δ34S values are
relatively uniform at approximately 121m The biochemical mechanisms outlined ear-
with little spatial variability due to extensive lier explain broad-scale distribution of various
mixing (Peterson & Fry, 1987). Marine sulfur stable isotope (CNS) compositions observed in
isotopic composition is predominantly con- POM in open marine systems. In shallow shelf
trolled by the large S isotopic fractionations seas and coastal environments, additional vari-
that occurs during bacterial sulfate reduction ables may complicate the spatial distribution
where lighter 32S isotopes are preferentially of stable isotope compositions in water and
reduced to insoluble sulfides (H2S, pyrite), POM, potentially providing isotopic gradients
resulting in distinctive isotopic differences at smaller spatial scales than typically seen in
between pelagic and benthic environments, the open ocean. Isotopic gradients in δ13C
with low δ34S sulfide values in marine sedi- values of POM between coastal and offshore
ments (210m) (Bottrell & Newton, 2006). regions are frequently observed, but explana-
Strong S isotopic differences also occur tions of the mechanisms underpinning such
between marine and terrestrial ecosystems, isotopic gradients are often unclear. Miller and
with freshwater δ34S values usually negative, Page (2012) suggest that increased nutrient
but ranging between 230m and 160m, largely supply in coastal regions can potentially
dependent on bedrock lithology, atmospheric increase phytoplankton growth rates, there-
deposition extent, and anthropogenic inputs fore, limiting preferential loss of 13C. In coastal
(Weber, Hutcheon, McKeegan, & Ingram, systems, a greater range of primary producers
2002). These isotopic differences may result in may contribute to supplying carbon and nitro-
clear δ34S gradients between estuarine and gen to aquatic food chains. In general terrestri-
coastal environments, with increased δ34S ally derived POM is depleted in 13C compared
ratios observed with distance from shore (Fry, to marine phytoplankton (Bouillon et al., 2011;
2002). δ34S values, therefore, vary among pri- Michener & Lajtha, 2008), and other potential
mary producers using different sources of S contributions to relatively 13C- enriched POM
(Connolly, Guest, Melville, & Oakes, 2004; in coastal regions include macrophytes and

benthic microalgae. Benthic microalgal produc- to wider marine food webs (France, 1995;
tion is also typically enriched in 13C compared Miller & Page, 2012).
to pelagic phytoplankton by around 7m In shallow marine environments, mixing of
(France, 1995), attributed to limitation of diffu- the water column and resuspension of POM
sion of CO2 away from benthic boundary can have a profound effect on isotopic gradi-
layers, such that regions with strong bottom ents. For example, in the North Sea, a shallow
currents may show lower isotopic distinction shelf sea between the United Kingdom and
between pelagic and benthic microalgae (and northeast European mainland, δ13C values in
by extension limited coastal-offshore gradients consumer tissues vary spatially by more than
in δ13C values) (Bouillon et al., 2011; France, 5m, and δ15N values by more than 8m (Barnes
1995). Similarly, macrophytes typically exhibit & Jennings, 2009; Jennings & Warr, 2003;
relatively positive δ13C values compared to in MacKenzie, Longmore, Preece, Lucas, &
situ phytoplankton, but macrophytes show Trueman, 2014) (Fig. 6.3). Depth and water
large isotopic variability, exceeding 6m in a temperature are primary environmental corre-
single leaf of the eelgrass Zostera marina and lates for both δ13C and δ15N values, while dis-
8m in a single stipe of the kelp Laminaria longir- tance to shore has weak predictive power
uris (Stephenson, Tan, & Mann, 1984). (Barnes & Jennings, 2009; Jennings & Warr,
Macrophytes may provide substantial nutri- 2003; MacKenzie et al., 2014). Spatial patterns
ents to grazers and secondary consumers in δ13C and δ15N values in pelagic food webs
within the immediate food web (Koenigs, of the North Sea are best explained by a com-
Miller, & Page, 2015), but they are generally bination of river-borne nutrient input, hydrol-
thought to play a limited role in carbon flow ogy, and winter resuspension of sedimentary
(A) (B)
62
62
δ13C δ15N
60
60
–14
16
–15
58
58
–16 14
56
56
−17 12
54
54
–18
10
–19
52
52
8
50
50
–5 0 5 10 –5 0 5 10
FIGURE 6.3 Isoscapes of (A) δ C and (B) δ N values across the North Sea interpolated from jellyfish tissues. Source:
13 15
After MacKenzie, K., Longmore, C., Preece, C., Lucas, C., & Trueman, C. (2014). Testing the long-term stability of marine isoscapes
in shelf seas using jellyfish tissues. Biogeochemistry, 114; Trueman, C. N., MacKenzie, K. M., & St John Glew, K. (2017).
Stable isotope-based location in a shelf sea setting: Accuracy and precision are comparable to light-based location methods. Methods in
Ecology and Evolution, 8, 232240.

nutrients (Trueman, MacKenzie, & St John pattern is largely driven by
Glew, 2017). Anthropogenic inputs to marine predictable physical or geological processes
systems further complicate local patterns of and incorporation of isotopes into tissues
isotopic distribution, while providing addi- occurs primarily through diet and drinking
tional potential isotopic signals for spatial water with significant, but limited, trophic
analysis (Morris et al., 2009). Sulfur isotopes alteration (see Chapter 2: Introduction to
are increasingly used in coastal marine sys- Conducting Stable Isotope Measurements for
tems because the anoxic environments com- Animal Migration Studies, Chapter 3:
monly found in organic-rich coastal and Isoscapes for Terrestrial Migration Research,
shallow marine sediments yield sulfur which Chapter 4: Application of Isotopic Methods to
is highly depleted in 34S compared to the rela- Tracking Animal Movements, and Chapter 5:
tively homogenous δ34S values of open marine Tracking of Movements of Terrestrial
environments (Peterson & Fry, 1987). Thus, Mammals Using Stable Isotopes ). The isotopic
while coastal and marine environments fre- composition of carbon and nitrogen in animal
quently show distinct multi-isotopic composi- tissues is influenced by both spatial-temporal
tions, predicting the scale and even direction variability at the base of the food web and the
of stable isotope gradients between terrestrial nature and number of isotopic fractionation
and marine environments in those areas is per- steps between the test animal and primary
haps more challenging than could be inferred production (but see Chapter 7: Amino Acid
from much of the isotopic literature. Isotope Analysis: A New Frontier in Studies of
Animal Migration and Foraging Ecology on
compound-specific isotope methods for further
details). Critically, any studies tracking animal
6.2.7 Construction and Use of Marine
location using C and N isotopes are effectively
Isoscapes tracking the geographic area where carbon and
The statistical methods associated with con- nitrogen were originally assimilated into pri-
struction of predictive isoscapes and probabi- mary production. This may coincide with the
listic assignment to determine organism physical location of the animal, but only where
origins are described in detail in Chapter 3, all prey species throughout the food chain
Isoscapes for Terrestrial Migration Research remain in the same isotopically characterized
and Chapter 8, Design and Analysis for region. The isotopic composition of animal tis-
Isotope-Based Studies of Migratory Animals, sue represents a weighted average resulting
and are essentially identical for marine and from all the trophic interactions occurring
terrestrial systems. However, there are unique between primary production and assimilation
considerations for the use of stable isotopes to into the target animal, and may therefore rep-
track migration in marine systems. In marine resent a wide temporal and spatial range, com-
systems, most isotope-based migration plicating the interpretation of spatial origin.
research uses δ13C and δ15N measurements, Interpreting an animal’s geographic origin
and their spatial patterns are largely a conse- from C, N, and S isotope compositions thus
quence of biochemical and physiological pro- becomes more challenging with increasing tro-
cesses in primary production, transferred phic distance between the test animal and the
through food webs via trophic interactions. reference isoscape. Increasing trophic distance
This contrasts with the bulk of terrestrial introduces more potential isotopic fraction-
migration isotope work using oxygen, hydro- ation steps, and for spatial offset through
gen, and strontium as tracers, where spatial migratory prey (and thus more uncertainty)

between the test and reference organisms. consumers. Marine isoscapes have been
Longer tissue integration times associated with generated across different spatial scales rang-
low metabolic rates also increase the potential ing from geographically isolated coastal
for foraging over isotopically variable diets areas and shelf seas (Barnes & Jennings, 2009;
during isotopic turnover, and thus, developing Jennings & Warr, 2003; Radabaugh & Peebles,
tissue isotopic compositions that reflect a 2014; Trueman et al., 2017; Vander Zanden
weighted average of multiple locations, rather et al., 2015), broad coastal regions (Fourqurean,
than a single location. Isotopic averaging is not Manuel, Coates, Kenworthy, & Boyer, 2015;
necessarily detrimental since averaging over Vokhshoori & McCarthy, 2014), to ocean
time reduces the requirement for temporally basins (Jaeger, Lecomte, Weimerskirch,
or seasonally explicit isoscapes (see Fig. 6.4). Richard, & Cherel, 2010; McMahon et al., 2013;
Quillfeldt, Masello, McGill, Adams, & Furness,
2010) and across global oceans (LeGrande &
Schmidt, 2006; Magozzi et al., 2017; Somes
6.2.8 Construction of Marine
et al., 2010) (Fig. 6.1, Table 6.1). The spatial
Isoscapes—Reference Organisms scale of an elemental isoscape is frequently
Most applications of stable isotopes to study limited by sampling constraints, resulting in
migration of marine animals are based on spa- numerous local- and regional-scale isoscapes.
tial variations in carbon, nitrogen, and sulfur As marine isoscapes are commonly con-
isotopic compositions in primary production, structed from carbon and nitrogen isotope
passed through the food chain to higher compositions in tissues of different isotopic
–22 –24 –26 –28 FIGURE 6.4 Simulation of the effect of temporal
averaging through trophic interactions on baseline
variability in δ13C values. (A) Simulations of oceanic
δ13CPOM values across the North Atlantic for
January taken from Magozzi, S., Yool, A., Vander
Zanden, H., Wunder, M., & Trueman, C. (2017).
–21
Using ocean models to predict spatial and temporal

variation in marine carbon isotopes. Ecosphere, 8 and
(B) monthly median simulated δ13C values averaged
T.L. 4
over the region displayed in (A). δ13C values in sim-
ulated POM are plotted in black and show seasonal
–22
variation around 2m. Each successive line suggests

an additional trophic level with temporal attenua-
T.L. 3
tion increasing by 2 months and a trophic fraction-
δ13CPOM
ation of 1m. At trophic level 4 (6 months integration

–23
time) seasonal variability in δ13C values is below

0.5m.
T.L. 2
–24
POM
–25
February April June August October December

baseline reference organisms (e.g., phytoplank- Vinette, 1998). However, copepods have
ton and various zooplankton), combining relatively complex trophic ecology, potentially
isoscapes can be problematic, introducing conducting daily and seasonal vertical water
additional sources of variance and uncertainty column migrations, overwintering and storing
relating to the biology, ecology, and physiol- lipids to maintain vertical position. Crustacean
ogy of the differing reference organisms and zooplankton often contain high lipid contents
the spatial and temporal scale over which they and have carbonate exoskeletons, so that con-
integrate carbon and nitrogen into their sistent chemical preparative treatments may be
tissues. needed to obtain appropriate stable isotope
Phytoplankton and POM can be sampled to data (see Chapter 2: Introduction to
construct marine isoscapes (Table 6.1), but Conducting Stable Isotope Measurements for
their isotopic composition may vary over short Animal Migration Studies). Sessile inverte-
distances and timescales as cell generation brates (i.e., mussels and scallops) are obvious
times are short, and the isotopic composition targets for reference organisms in benthic sys-
of cell tissues reflects localized environmental tems, but the distribution of benthic inverte-
conditions. Consumers integrate spatio- brates is generally tightly coupled to the
temporal variations in primary production and nature of the aquatic substrate and water
POM, and marine C and N isoscapes are fre- depth, so it may be difficult to obtain spatially
quently built from low trophic-level organisms consistent sampling. Feeding method should
(Fourqurean et al., 2015; Kurle & McWhorter, be considered since filter feeders sampling rel-
2017; Mackey, Hyndes, Carvalho, & Eyre, atively fresh, sinking phytoplankton, and POM
2015; McMahon et al., 2013; Trueman et al., may develop contrasting isotopic compositions
2017). The choice of a reference organism is compared to detrital feeders in the same
not straightforward. An ideal reference organ- location.
ism will be spatially widespread or have Phytoplankton-based baseline reference iso-
reduced mobility, will be evenly distributed tope samples are not suitable for assessing
across the study area, expressing no preference movement in animals specializing on isotopi-
for patchy habitats, and have an isotopically cally distinct diets or habitats (e.g., turtles
generalist diet close in trophic level to the spe- grazing on seagrass beds). Here baseline refer-
cies of interest (Barnes & Jennings, 2009; ences must be constructed from specifies-
Jennings & Warr, 2003; Lorrain et al., 2015; specific data or known dietary items wherever
MacKenzie et al., 2014; Trueman et al., 2017). possible (Vander Zanden et al., 2015) If the ref-
Inevitably, the choice of reference organisms erence organisms used to construct isoscapes
used to construct an isoscape will reflect a are not identical to the test animal, isotopic cal-
trade-off between the expressed and desired ibration (or offsets) between the reference and
traits, weighted according to the details of the test species must be known or experimentally
study areas and principal aims of the study. derived. This calibration is essentially equiva-
In pelagic systems, crustacean zooplankton lent to the terrestrial rainfall-tissue offset
such as copepods and euphausiids present described for oxygen and hydrogen isotope
attractive targets for construction of marine assignment work (see Chapter 3: Isoscapes for
isoscapes as they are widely distributed, do Terrestrial Migration Research and Chapter 8:
not move over large lateral distances during Design and Analysis for Isotope-Based Studies
the timescale of tissue assimilation, and are of Migratory Animals). If matched sets of ref-
key components of pelagic marine food webs erence and test animals are not available in the
(McMahon et al., 2013; Schell, Barnett, & study area, estimates of isotopic calibrations

can be drawn from estimated trophic enrich- Schmittner et al., 2013; Somes et al., 2010;
ment factors and trophic separation between Somes, Oschlies, & Schmittner, 2013; Tagliabue
test and reference organisms, and any tissue- & Bopp, 2008). Models are an inexpensive prac-
specific isotopic offset, but the uncertainties tical alternative to extensive sample collection,
associated with these estimates need to be but lack local-scale detail. Model estimates are
propagated and will reduce assignment accu- often based on critical assumptions, and hence,
racy and precision (Trueman et al., 2017) offer a simplified representation of isotopic
Aquatic organism tissue type also requires space (Magozzi et al., 2017). The development
consideration (Trueman et al., 2012). of model-based isoscapes is limited by our
Metabolically active tissues, such as blood liver knowledge of the isotopic physiology of pri-
and skin, represent recent diet consumption mary producers (especially for nitrogen).
and snap shots in isotopic time, whereas less Assessing the accuracy and precision of model-
metabolically active tissues, such as bone colla- generated isoscapes is difficult, usually due to a
gen, reflect diet over longer time periods lack of field sample and calibration data.
(Hobson, 2005; Ramos & Gonzalez-Solis, 2012; Nonetheless, model isoscapes effectively cap-
Wolf, Carleton, & Del rio, 2009). Tissue choice ture large-scale spatial variability in carbon and
will, therefore, result in the integration of nitrogen isotopes in particulate organic matter,
ingested diet items across different timescales and help to inform research where differences
(Tucker, Macdonald, & Seminoff, 2014). If between predicted and measured data arise.
studying short-term processes, tissues with Oceanic regions likely to be isotopically distinc-
higher metabolic rates will be needed, whereas tive can be predicted a priori by quizzing ocean
tissues with longer integration times may aver- models, which may help to design isotope-
age short-term variability and provide a based studies of migration (Fig. 6.5). In our
clearer picture of directed movement between view, generalized isoscape models are cur-
two isotopically contrasting areas. Some of the rently unsuitable for direct geolocation of
caveats with tissue type isotope analyses are migrant marine animals, but provide valuable
described in Chapter 2, Introduction to tools to explore mechanisms underpinning of
Conducting Stable Isotope Measurements for spatio-temporal variability in stable isotopes of
Animal Migration Studies. primary production and provide a platform for
in silico (i.e., computational) simulation. Model
isoscapes have been used effectively to inter-
pret ocean-basin scale compilations of
6.2.9 Construction of Marine
stable isotope data in squid, tuna, and sharks
Isoscapes—Global Models (Bird et al., 2018; Navarro, Coll, Somes, &
Coupled ocean physics-biogeochemistry Olson, 2013; Young et al., 2014, 2015).
models can be used to generate isoscape mod-
els either by tracking isotopes of carbon and
nitrogen throughout the model or by using out- 6.2.10 Temporal Variability in Baseline
put variables to predict isotopic compositions
in surface production. Various model combina-
Isoscapes
tions have predicted spatio-temporal variations Marine nutrient availability and photosyn-
in carbon and nitrogen isotope ratios in POM thesis rates fluctuate over seasonal, yearly, and
across the global ocean, based on different longer time scales in turn influencing CNS iso-
assumptions and combinations of variables tope ratios (Hannides, Popp, Landry, &
(Hofmann et al., 2000; Magozzi et al., 2017; Graham, 2009; Kurle & McWhorter, 2017;

(A) 7
1
(B) (C) (D)
10
10
4
8
3
6
2
δ15NPOM
δ NPOM
δ18Ooto
1
4
4
15
0
2
−1
0
−2
−2
−2
−3
−3 −2 −1 0 1 2 3 4 −30 −28 −26 −24 −22 −20 −18 −30 −28 −26 −24 −22 −20 −18
18
δ Ooto δ13CPOM δ13CPOM
FIGURE 6.5 (A) Potential isotopically distinct regions of the global ocean estimated from simulated δ13C, δ15N, and
δ18Ooto isoscape models shown in Fig. 6.2. Distinct regions classified by k-means clustering with y-defined clusters was
set. (BD) Scatterplots showing cluster separation in two-dimensional isotope space. δ13C and δ15N provide the most
distinct clusters (C), while δ13C and δ18Ooto values co-vary strongly though the shared influence of sea-surface tempera-
ture (D).
Quillfeldt et al., 2015), with temporal variabil- Peebles, 2013), although significant seasonal
ity outweighing spatial variability in some and yearly fluctuations have also been
highly dynamic regions (Quillfeldt et al., observed in predators (Kurle, Sinclair,
2015). Temporal and spatial variability in car- Edwards, & Gudmundson, 2011; Quillfeldt
bon and nitrogen isotope compositions is par- et al., 2015). In some marine environments, C
ticularly prevalent in primary producers, and N isotope ratios remain relatively
whereas variations are dampened in higher stable (Ceia et al., 2015; Jones, Hoover,
trophic-level organisms due to slower tissue Heppell, & Kuletz, 2014), e.g., North Sea broad
turnover rates (Ceia et al., 2014, 2015; Graham, scale isotopic patterns have remained rela-
Koch, Newsome, McMahon, & Aurioles, 2010; tively consistent over decadal to centennial
Montoya, 2007; Radabaugh, Hollander, & time periods (Barnes & Jennings, 2009; Barrett

et al., 2011; Jennings & Warr, 2003; MacKenzie behavior and biology of aquatic animals
et al., 2014; Trueman et al., 2017), likely due to combine to provide important differences.
the dominant influence of static environmental Marine animals are commonly underwater,
variables on isotopic composition within this and hence, difficult to observe directly, but in
area (MacKenzie et al., 2014). the case of commercial fishes in particular,
In marine basins with complex circulation large volumes of observational data are often
and upwelling patterns, the spatial distribution collected, typically relating to abundances and
of C and N isotopic values may be dynamic in distributions of populations in time and space.
time, complicating the use of spatial reference Individual-level data are less common, most
samples or static isoscapes (Kurle & frequently based on classic point mark-
McWhorter, 2017; Kurle et al., 2011). The tem- recapture studies. Coupling isotopic data with
poral resolution of marine carbon and nitrogen electronic tagging is becoming more common
isoscapes depends in part on the position of practice in terrestrial isotope migration work
the sample organism in the food chain, the tis- but is incredibly challenging for fishes.
sue taken, and the isotopic turnover time. Recovery of tags is problematic in fishes,
Isoscapes have been developed capturing depending on re-capture of tagged individuals
short-term seasonal variability from tissues of and return of the tag. Isotopic information can
organisms with rapid isotopic assimilation be linked with tag-derived known movement
rates (Ceia et al., 2015; MacKenzie et al., 2014; data only if the organism is returned
Radabaugh et al., 2013; Trueman et al., 2017), (Darnaude et al., 2014). Pop-up satellite tags
and approximately annual-scale averages are increasingly deployed on large fishes, and
using tissues with slow integration time while isotopic samples may be taken at the
(Jennings & Warr, 2003; Vander Zanden et al., point of deployment, it is extremely difficult to
2015). Longer-term averages have been pro- obtain tissue samples relating to the time win-
posed from integration of data over multiple dow recorded by the tag. Coupled tag-based
years (LeGrande & Schmidt, 2006; McMahon geolocation and isotopic data can, however, be
et al., 2013; Quillfeldt et al., 2010). obtained for central-place surfacing foragers
Understanding the time frame represented by such as pinnipeds and turtles during the
an isoscape, temporal variability in the local breeding season. Many marine animals possess
baseline and the timescale of isotopic integra- incrementally grown tissues, including oto-
tion within the food chain is essential for effec- liths, vertebrae, teeth, baleen, eye lenses, gla-
tive use of isoscapes (and isotopic data in dius, and vibrissae, and sequential sampling of
general) in ecological applications (Graham incremental tissues represents a major tool in
et al., 2010; Quillfeldt et al., 2015; Ramos & marine migration research, in part compensat-
Gonzalez-Solis, 2012). ing for the difficulty of repeatedly sampling
individuals.
6.3 PART 2: EXAMPLES OF

ISOTOPES AND ISOSCAPES FOR 6.3.1 Oxygen Isotope Records of
MARINE MIGRATION RESEARCH
Migration From Biomineralized Tissues
The general principles of tracking animal Spatial variations in water δ18O and temper-
migration using isotope-based methods are ature provide predictable gradients in the δ18O
similar between terrestrial and marine systems, values of biomineralized tissues, particularly
but aspects of the marine environment and the in calcite or aragonite (both are forms of

6.3 PART 2: EXAMPLES OF ISOTOPES AND ISOSCAPES FOR MARINE MIGRATION RESEARCH 157
CaCO3) of invertebrate shells and otoliths. The 2011; Trueman et al., 2012). Atlantic bluefin
chemical composition of otolith aragonite tuna (Thunnus thynnus) are highly migratory,
(termed “microchemistry” in the otolith litera- commercially valuable, and under intense fish-
ture) has long been used to infer the geo- ing pressure. The relatively high rates of evap-
graphic origin and stock structure of fishes oration and oxygen isotopic enrichment of
(Campana, 1999). Otolith trace-element-based Mediterranean seawater have been used to
approaches are well-suited to testing whether quantify the extent of trans-oceanic tuna
populations are likely to have been drawn migration and stock exchange between the
from distinct source areas, but the mechanisms main spawning areas in the Gulf of Mexico
underpinning trace element variability in and the Mediterranean Sea (Rooker et al., 2008,
marine waters, and the physiological influences 2014). Values of δ18O in otoliths from yearling
affecting fractionation of trace elements tuna recovered from two spawning regions
between water and otolith aragonite are not differed with relatively high δ18O values in
fully understood (Sturrock et al., 2014, 2015). Mediterranean-origin fish. Otolith cores reflect
Consequently, trace-element approaches are the early years of life, and the isotopic compo-
restricted to determining the “most likely” sitions of core regions of otoliths extracted
match for a fish or population of fishes between from adult fish caught in Canadian waters
a discrete suite of reference populations. One suggested exclusive origins from the western
exception is movement across salinity gradi- Gulf of Mexico spawning areas, while natal
ents, where large and predictable gradients in origins of maturing tuna entering the
trace element compositions are reflected within Mediterranean Sea indicated high levels of
otolith aragonite, providing a powerful tracer spawning site fidelity in the eastern stock
for movement between freshwater and marine (Rooker et al., 2014; Schloesser, Gillanders,
conditions (Walther & Limburg, 2012). Neilson, Secor, & Rooker, 2010). Tagging work
In open marine environments, oxygen iso- identified cross-oceanic migrations of bluefin
topes offer information on migration where tuna, particularly during preadult phases of
movement occurs across temperature gradients life (Block et al., 2005), indicating potential for
or between isotopically distinctive water bod- mixing between stocks. As a result, isotope-
ies. In two pioneering examples of the inferen- based approaches for estimating stock mixing
tial use of isotopes to assess migration of and trans-oceanic migrations in Atlantic tuna
marine animals, the isotopic compositions of have been adopted in stock assessment tools
oxygen in barnacles attached to gray whales by the International Commission for the
(Eschrichtius robustus) (Killingley, 1980) and Conservation of Atlantic Tuna, one of the few
loggerhead turtles (Caretta caretta) (Killingley examples where stable isotope-based methods
& Lutcavage, 1983) were used to infer water are contributing directly to marine resource
temperature, and thus, the latitudinal migra- management and conservation (Secor, 2015b).
tion. Detjen, Sterling, and Gómez (2015) took a Atlantic salmon (Salmo salar) are another
similar approach, identifying likely residence highly migratory fish with commercial and
areas for green sea turtles (Chelonia mydas) cultural significance. Populations of wild
using barnacle δ18O isoscapes, simulated from Atlantic salmon experienced dramatic declines
sea-surface temperature and surface water since the 1980s, linked to increased marine
δ18O values. mortality (Friedland, Shank, Todd, McGinnity,
Values of δ18O in otoliths provide valuable & Nye, 2014). Hanson, Wurster, EIMF, and
direct tracers of movement of teleost fishes Todd (2010) attempted to use otolith δ18O
across isotopic gradients (e.g., Bouillon et al., values to infer the marine feeding regions of

salmon returning to Scottish rivers. Values of brackish water nursery grounds before migrat-
δ18O determined at high spatial precision ing to open marine waters, where they
clearly indicated abrupt transitions from fresh- remained for 3 years then returning to mar-
water to marine waters, but uncertainty ginal brackish waters, presumably to spawn.
around the form of the δ18O fractionation Biomineral δ18O values are powerful tracers
equation (Eq. 6.1) limited the practical value of of movement where animals migrate across
the method (see below for alternative well-known temperature and salinity gradi-
approaches to identifying feeding areas of ents. Biominerals may show growth incre-
Atlantic salmon). ments providing records of movements
Fish otolith δ18O values (δ18Ooto) are rela- through whole life history. Otoliths are impor-
tively sensitive to ambient δ18Owater values, tant in fisheries research and are commonly
which are often poorly constrained particularly collected and stored in archives allowing ret-
in enclosed coastal shelf waters. Uncertainty in rospective analyses of the nature of migratory
reconstructed water temperatures can and behaviors across decadal timescales. The
should be propagated in any migration assign- chemical stability of inorganic biominerals
ment approach. In a study of plaice extends the potential for isotope-based migra-
(Pleuronetes platessa) where otoliths were recov- tion research into historic and prehistoric
ered from individual fish that had data logger timescales. The requirement for precise esti-
tags, precise models of salinity were needed to mates of water δ18O and uncertainty sur-
achieve estimates of likely location consistent rounding terms in Eq. (6.1) currently limits
between otolith δ18O and tidal pressure geo- the precision of geolocation using biomineral
location methods (Darnaude et al., 2014). δ18O values.
Biomineral δ18O values vary with temperature
and water mass origin, and are therefore also
well-suited for reconstructing vertical migra-
6.3.2 Carbon and Nitrogen Isotopes in
tions. Several studies have reconstructed life-
time movements in deep-water fishes and
Migration
cephalopod molluscs using otolith δ18O values Three main approaches have been used to
(Chang, Liu, Liao, & Shiao, 2015; Shephard, link marine animal tissue isotopic composi-
Trueman, Rickaby, & Rogan, 2007; Shiao, Liu tions to their geographic origin, requiring dif-
& Sui, 2016; Shiao, Sui, Chang, & Chang, 2017) fering levels of accompanying isotopic data.
revealing common patterns of shallow pelagic All the following approaches draw on the
larval and juvenile life stages followed by assumptions and limitations outlined earlier
ontogenetic depth migrations of varying associated with uncertain diet-tissue isotopic
complexity. fractionation, isotopic incorporation rates, and
Biomineralized tissues may be preserved in temporal variability in the isotopic composi-
sediments offering insights into movements of tions of primary production.
historic, prehistoric and fossil fishes. In its simplest form, the CNS isotopic com-
Ontogenetic migrations have been identified in positions of tissues of a migratory marine ani-
a late Cretaceous fossil fish from the Fox Hills mal are compared to those values expected at
Formation of South Dakota (Carpenter, the capture location, between individuals or
Erickson, & Holland, 2003). Sequential analy- across incrementally grown tissues within
ses of oxygen and carbon stable isotopes across individuals. Isotopic differences are then inter-
the growth axis of four well-preserved otoliths preted with reference to assumed drivers of
suggested that juvenile fish spent a year in aquatic spatial isotopic variability. In this case,

little background information is needed, but regions, and thus likely foraging areas, were
the precision and potentially accuracy of estimated from Bayesian mixing model analy-
inferred movement patterns is limited. For ses (mixSIR) extended to consider known tem-
example, Humboldt (jumbo) squid (Dosidicus poral variations in shark distributions and
gigas) from the eastern Pacific Ocean undergo isotopic turnover rates (Moore & Semmens,
extensive daily vertical migrations and proba- 2008). The mixing model results suggested that
bly also latitudinal migrations. Squid have a juvenile white sharks primarily foraged within
chitinous tissue, the gladius, which grows coastal regions before initiating migrations and
incrementally throughout its life. Sequential C foraging within the oligotrophic central Pacific
and N isotope sampling along the gladius of at sexual maturity. By contrast, salmon sharks
33 Humboldt squid collected from 6 locations appeared to forage in oceanic subtropical gyre
in the eastern Pacific (Ruiz-Cooley, Villa, & regions as juveniles before moving to forage in
Gould, 2010) revealed complex ontogenetic relative coastal regions after age 6.
variations in both δ13C and δ15N values, imply- In a study of humpback whales (Megaptera
ing differences in ontogenetic behaviors within novaeangliae) from the Pacific Ocean,
and between sites, but lacked spatially explicit Witteveen, Worthy, and Roth (2009) used a
reference isotope data, which limited the inter- classification-tree approach to link summer
pretation of potential migratory movement feeding areas to δ13C and δ15N values of skin
patterns. samples taken from whales in discrete feeding
Tissue CNS isotopic data can be compared areas. The classification-tree model was then
qualitatively or quantitatively to isotopic data used to infer summer feeding locations for ani-
from other individuals of the target species mals sampled in winter breeding aggregations
from discrete potential origin sites or from on the assumption that feeding is limited dur-
prey to estimate the likely origin. Pelagic ing winter breeding. Tissue isotopic composi-
sharks are difficult to observe, and direct tions can also be matched to locations by
satellite-based tagging is very expensive limit- comparison to CN isotopic analyses of tissues
ing the number of individuals that can be of individuals or populations whose locations
studied. The vertebral centra of sharks and are known from alternative direct tagging
rays grows incrementally, providing a host for approaches. This approach is particularly well-
construction of continuous isotopic life histo- developed in studies of seabirds and pinni-
ries. Carlisle et al. (2012, 2014) inferred likely peds where individuals are tagged at breeding
foraging areas for white sharks (Carcharodon colonies, and combinations of tagging and iso-
carcharias) and salmon sharks (Lamna ditropis) topic data used to infer nonbreeding behavior
in the eastern Pacific based on data for isotopic (Jaeger et al., 2010). For example, Phillips et al.
compositions of prey from different potential (2009) used feather C and N isotope data to
broad foraging locations. The isotopic compo- investigate nonbreeding foraging behaviors in
sition of resources from differing locations 15 species of seabirds breeding at Bird Island
were estimated from literature data and sum- in South Georgia. Tracking information and
marized as means and variances, under the isotopic data were available for 7 species,
assumption that averaged literature-derived allowing isotopic compositions associated with
prey data represented the isotopic composi- foraging in either Antarctic, sub-Antarctic,
tions of the prey fields encountered by the sub- sub-tropical, or continental shelf habitats to be
ject sharks. The likely proportional defined. These broad divisions were (qualita-
contributions of prey from isotopically discrete tively) used to infer likely habitat preferences

for 8 other species for which no tracking data 6.3.3 Assignment to Marine Isoscapes
were available. Isotope-based geo-location will
always be less precise than location deter- In terrestrial systems, isotope-based studies
mined from direct tagging approaches; how- of connectivity and migration have benefitted
ever, isotope-based methods can be more from the isoscape approach, with a corre-
easily applied to large numbers of individuals, spondingly large literature (discussed in
providing information about population-level Chapter 3: Isoscapes for Terrestrial Migration
frequencies of movement behaviors. Research and West, Bowen, Dawson, & Tu,
In a study of green turtles in the eastern 2010). In contrast, we know of only three
Mediterranean, the CNS isotopic composition examples in the literature where marine ani-
of a large proportion of 230 measured indivi- mals have been assigned to likely geographic
duals nesting in northern Cyprus were incon- origins based on probabilistic comparisons
sistent with foraging locations identified from between isotopic compositions of animal tis-
a smaller suite of satellite tagged individuals. sues and continuous-surface isoscapes. In part,
Assuming annual fidelity in feeding grounds, this reflects difficulties or lack of effort associ-
isotope data were used to target individuals ated with constructing sample-based isoscapes
for subsequent satellite tagging, revealing the in marine systems, and the uncertainties asso-
unrecognized importance of a single foraging ciated with tissue turnover times and trophic
site in coastal northern Egypt (Bradshaw et al., fractionation. With careful consideration of
2017). underlying assumptions and associated uncer-
None of the approaches described earlier tainties, however, isotope-based assignment
produce spatially explicit predictions of forag- using isoscapes may prove extremely effective
ing location for populations or individuals. If in marine systems. The first published exam-
populations are thought to conserve migration ple (Vander Zanden et al., 2015) investigated
routes, or at least migrate to consistent forag- foraging behavior of loggerhead sea turtles in
ing regions, predictive linkages between SST the Gulf of Mexico region. Conceptually this
and isotopic compositions can be used to approach is analogous to terrestrial avian
derive maps of likely foraging area. Remote research, as female sea turtles are accessible
sensing data provide estimates of global sea- while nesting, but nonbreeding movements are
surface temperature at high spatial resolution. poorly understood. Vander Zanden et al.
Regions with high temporal covariance (2015) derived a species-specific isoscape for
between SST (driver) and δ13C and δ18O values loggerhead sea turtles by ordinary kriging of
in the organism of interest (response) represent the isotopic compositions of carbon and nitro-
likely foraging areas. MacKenzie et al. (2011) gen in scutes from satellite-tracked turtles. The
applied this approach to carbon isotope com- performance of the marine isoscape-based
positions measured in over 30 years of assignment was then evaluated by assigning
archived Atlantic salmon scales, demonstrating 19 further turtles of known origin. Critically,
that Atlantic salmon returning to two regions the isoscape approach provided an estimate of
in the United Kingdom foraged in different likely origin with 79% accuracy with a spatial
regions of the north Atlantic. These analyses precision of 25% of the assessed area (see
highlighted the Norwegian Sea and Iceland Chapter 3: Isoscapes for Terrestrial Migration
Basin or west Greenland as likely foraging Research and Chapter 8: Design and Analysis
areas, consistent with the current understand- for Isotope-Based Studies of Migratory
ing of salmon migrations based on tag- Animals for a more detailed discussion of
recapture and genetic data. assignment accuracy and precision threshold

methods). In a similar approach in the North from the herring fishery (Trueman et al., 2017).
Sea, Trueman et al. (2017) evaluated the per- A small number of adult herring were caught
formance of isoscape-based assignments where in late August in the southern North Sea, and
the baseline reference organisms differed from these individuals were assigned to the princi-
species to be assigned. This is an important ple foraging areas in the north central North
distinction, as it allowed for a baseline refer- Sea, indicating that some herring migrated
ence isoscape to be used to study movement rapidly to the spawning grounds. Juvenile her-
for a wide range of taxa, rather than requiring ring caught in the southern North Sea were
species-specific isoscape surfaces. Uncertainty assigned to local foraging areas, consistent
associated with calibrating isoscapes between with assumptions of restricted movements in
taxa with different trophic levels and tissue juvenile herring. Continuous assignment
types must be estimated and propagated into approaches were also used to infer origin in
assignment models (Trueman et al., 2017). In Atlantic salmon in the Baltic Sea drawing on
the North Sea example, marine isoscapes were isoscapes created from δ13CDIC and δ18O values
created by ordinary kriging of δ13C and δ15N measured in seawater and δ13C and δ18O mea-
values from lion’s mane jellyfish (Cyanea capil- surements in otoliths (Torniainen et al., 2017).
lata) sampled regularly across the North Sea These studies demonstrate that isoscape-based
(Fig. 6.3), and the resulting isoscape models assignments can be successful in marine set-
were used to assign scallops (Aequipecten opercu- tings, recovering patterns of movement, forag-
laris) and herring (Clupea harengus) of known ing, and migratory connectivity: effort placed
origin. Scallops were assigned to areas repre- in the construction of sample-based marine
senting 30% of the total study area, and 75% of isoscapes is, therefore, likely to be rewarded.
sampled scallops were accurately assigned with
a mean linear error on the order of 102 km.
Migratory fish in theory present a more
6.3.4 Incremental Tissues and
challenging problem, as isotopic turnover rates
in tissues available for isotopic analyses may
Movement
be longer than residency in a single region. A rich picture of individual-level movement
Turnover rates are related to metabolism, feed- can be recovered from sequential isotopic anal-
ing rates, and somatic growth (Thomas & yses of incrementally grown tissues. This
Crowther, 2015) so smaller animals with high- approach is particularly well-developed for
er metabolic rates will have faster tissue turn- marine mammal research. The keratinous whis-
over rates. Herring are abundant pelagic kers (vibrissae) of pinnipeds and the baleen
forage fish in north-east Atlantic waters, and plates of mysticete whales are chemically sta-
conduct ontogenetic and seasonal migrations. ble, easily handled tissues offering a time series
Jellyfish-based isoscapes were used to estimate of isotopic compositions. The application of
foraging areas for 300 herring captured at the stable isotope analyses to marine mammal ecol-
end of the summer feeding season across ogy was reviewed previously (Newsome,
the North Sea. Combined likely foraging Clementz, & Koch, 2010) and the literature has
areas for 300 herring assessed from isoscape grown considerably in succeeding years. To
assignments were similar to capture locations infer movement patterns from isotopic records
indicating minimal movement between of incrementally grown tissues, a clear under-
spring-summer feeding and capture in standing of growth rates and growth dynamics
August and September, and the assignments is needed. Experimental and inferential meth-
closely matched the distribution of catches ods have been proposed to identify annual

growth increments in vibrissae and baleen, and Isotopic data derived from incrementally
to infer changes in growth rates through ontog- grown tissues provide repeated measurements
eny (Beltran et al., 2015), however, experimen- on individual animals and are particularly
tal validation of growth dynamics is currently well-suited for identifying individual speciali-
only available for a few species. zations within generalist populations
Migration and foraging in baleen whales (Kernaléguen et al., 2012; Vander Zanden,
was one of the first applications of isotopes to Bjorndal, Reich, & Bolten, 2010). In an elegant
migratory ecology (Best & Schell, 1996; Schell, study of Antarctic fur seals (Arctocephalus
Saupe, & Haubenstock, 1989) and gazella), Kernaléguen et al. (2012) used high-
stable isotope methods have frequently been resolution sampling along vibrissae of 35 indi-
used to infer movement in baleen whales, vidual seals to reveal species-sexual and
partly due to the ease of handling of baleen. individual-scale spatial specializations in for-
Commonly, carbon and nitrogen isotopic pro- aging behavior. In the Southern Ocean, strong
files along baleen plates are compared to those isotopic gradients associated with the
for potential prey recovered from proposed Subtropical and Polar Fronts (Fig. 6.2) effec-
feeding grounds (e.g., Caraveo-Patiño, tively distinguished individuals conducting
Hobson, & Soto, 2007). Cyclical variations in local and distant-water foraging, and within
δ13C and δ15N values are often assumed to distant foraging animals, between seals forag-
show annual periodicity, and consequently, ing north or south of the sub-Antarctic islands
have been used to reconstruct baleen growth on which they breed (Kernaléguen et al., 2012).
rates. It should be noted, however, that the Where isotopic gradients are less pronounced,
physiological mechanism underpinning cycli- interpreting isotopic variations in incremen-
cal variations in δ13C and δ15N values in sea- tally grown tissues remains challenging, and
sonally fasting marine mammals is unclear in the following section, we outline how simu-
(Aguilar, Gimenez, Gomez-Campos, Cardona, lation modeling may help to infer movement
& Borrell, 2014; Best & Schell, 1996; Caraveo- patterns from sequential isotope data.
Patiño et al., 2007; Cardona, Vales, Aguilar,
Crespo, & Zenteno, 2017). Seasonal variations
between relatively high and low δ13C and δ15N 6.4 COMPUTER MODELING AND
values can only reflect movement between iso- SIMULATION
topically distinct regions if whales forage con-
tinuously throughout the year. δ15N values Sample-based isoscapes are relatively sim-
could in theory become (a) enriched during ple to construct for coastal and shelf sea envir-
winter fasting if protein demands are met onments, but the oceanic regions remain
through tissue catabolism or (b) depleted dur- challenging. Literature-based marine isotopic
ing winter fasting if reduced levels of urea data are inconsistent in space and time, and
production limit the loss of nitrogen waste while compilations of literature data provide
metabolites that are relatively depleted in 15N information on basin-scale spatial variation in
(Aguilar et al., 2014). Similarly, δ13C values in stable isotope compositions (McMahon et al.,
baleen or vibrissae could be relatively depleted 2013), they lack the spatial or temporal resolu-
during winter fasting if energy demands are tion required for most assignment applica-
largely met through respiration of stored lipids tions. Mechanistic isotope models offer an
providing 13C-depleted respiratory carbon for alternative approach for exploring spatial and
assimilation into nonessential amino acids temporal isotopic variability in open-ocean set-
(Cardona et al., 2017). tings. As described earlier, spatial variations in

6.4 COMPUTER MODELING AND SIMULATION 163
δ18O values of otolith and barnacle biomineral (Navarro et al., 2013; Young et al., 2015). In a
have been estimated directly from observed or recent global-scale study of shark foraging,
modeled SST and salinity and used directly to Bird et al. (2018) compiled records of the δ13C
infer origin and migration (Killingley, 1980; values of shark muscle from more than 5000
Killingley & Lutcavage, 1983). Global mecha- individual sharks from 113 species and identi-
nistic models of spatial variation in δ15NPOM fied species as oceanic, coastal, or deep water
values have been used to remove spatial vari- feeding groups. Shark samples were recovered
ance from bulk isotope data and infer trophic from large latitudinal gradients, and Bird et al.
level from predator δ15N values (Navarro (2018) compared latitudinal δ13C gradients
et al., 2013; Young et al., 2015), but have not observed in shark muscle to those predicted
yet been used extensively to assist in the inter- from an isotope-enabled global biogeochemical
pretation of movement and migration in model (Magozzi et al., 2017). δ13C records from
marine animals. Meta-analyses of compiled lit- shelf-foraging sharks closely tracked predicted
erature isotope data are attractive tools for latitudinal gradients in δ13C values of local
inferring large-scale ecological process (Pauli phytoplankton, implying that shelf sharks in
et al., 2017), and global-scale isoscape models general derive most of their carbon from local
provide important context for interpreting food webs and do not perform large latitudinal
such opportunistically assembled isotope data migrations (Fig. 6.6A). In oceanic sharks,
(A) (B)
–5
–5
–10
–10
–15
–15
δ13C
δ13C
–20
–20
–25
–25
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Distance to Equator (degrees)

FIGURE 6.6 Relationship between tissue δ13C values and modeled δ13CPOM values in 5000 individual sharks from
shelf (A) and oceanic (B) habitats. Latitudinal gradients of δ13C values in shelf dwelling sharks closely correspond to gra-
dients expected from local phytoplankton (black circles and pink regression envelopes) when adjusted for trophic level effects
(gray regression envelope) indicating that shelf sharks largely derive nutrients from the local area. Oceanic sharks show
restricted range in tissue δ13C values implying global reliance on nutrients from mid-latitudes. Source: Data from Bird, C.
S., et al., 2018. A global perspective on the trophic geography of sharks. Nature Ecology and Evolution, https://doi.org/10.1038/
s41559-017-0432-z. Online Publication January 18.

latitudinal δ13C gradients are much smaller isotopic composition of the tissue under study
than predicted from local phytoplankton to a combination of environmental, behavioral,
(Fig. 6.6B), implying assimilation of nutrients and physiological variables. Here we explore
from a restricted latitudinal range between how such simulation models could help
around 30 and 50 deg of latitude (Bird et al., to interpret migration patterns from
2018). stable isotope records of whale baleen. In an
In the final section of this chapter, we out- early, influential, and carefully argued paper
line some directions where modeling may (Best & Schell, 1996) inferred movements of
inform future isotope-based studies of move- southern right whales (Eubalaena australis)
ment in oceanic marine ecosystems. Despite stranded on the South African coast based pri-
attractive qualities as robust environmental marily on carbon isotope profiles along baleen
recorders, tissue isotope compositions are plates. Oscillations in δ13C values along the
influenced by a wide range of potential drivers baleen plate were interpreted with respect to
(Boecklen, Yarnes, Cook, & James, 2011), and measured latitudinal isotopic gradients in phy-
interpreting the cause(s) contributing to mea- toplankton (Fig. 6.7A). Briefly, 13C-enriched
sured variation in isotopic compositions can be excursions were interpreted as reflecting forag-
difficult. Some of the most challenging vari- ing around the subtropical convergence at
ables to consider in the context of migratory around 45 S. The timing of isotopic peaks in
organisms include temporal variability in base- δ13C values, however, did not converge with
line isotopes, tissue turnover rates at all stages latitudinal distributions expected based on
of the food chain between primary production whaling data. Peak δ13C values occurred in the
and the subject animal, temporal variability in Austral summer (January to April), whereas
feeding rate (e.g., Fig. 6.4) and physiological northernmost sightings of whales occur in late
effects on isotopic turnover and routing. fall-winter (May to July). Furthermore, the
Assessing these possible factors through logi- baleen isotope record suggested feeding in
cal argument is extremely challenging but they northern waters (i.e., relatively positive δ13C
can be investigated through computer simula- values) in the austral summer, despite whaling
tion modeling. and sighting data indicating a northerly distri-
Isotope-enabled global biogeochemical mod- bution of whales in the austral winter and the
els provide estimates of spatial and temporal presence of whales as far north as 20 S in June
variability in stable isotope baselines. Outputs and July. This inconsistency was interpreted as
from these models can be used as inputs to reflecting limited feeding and subsistence
food web or migration models. In the context while in northern waters from protein assimi-
of migration research, agent-based movement lated in southern waters (Best & Schell, 1996).
models can be used to generate potential Here we apply a simulation approach to incor-
movement tracks. These are then coupled to porate temporal variance in isotopic baselines
temporally specific isoscape models to gener- and explore the likely isotopic expression asso-
ate a record of the isotopic composition of diet ciated with differing migration and feeding
available at each time step. Finally, a physio- scenarios. We used a simple agent based
logical model can be applied to consider model for whale movements which sets the
effects of tissue turnover rates and/or molecu- likelihood, direction, and extent of movement
lar (isotopic) routing. By creating simulation on a daily basis as a probabilistic function of
models, all variables can be adjusted, offering behavioral state, SST, water depth, and phyto-
insight into the sensitivity of the ultimate plankton concentration (as a proxy for

(A) (B)
(C) (D)
0
−20
25
−40
50
(E) (F)
−20.0
−29.0 −27.5 −26.0 −24.5 −23.0 −21.5 −20.0 −18.5

−21.0
Simulated δ C
−22.0
13
Measured δ C
13
−23.0
−24.0
−25.0
−26.0
0 90 180 300 420 540 660 780 900 1020 1140 1260 1380 1500 1620 1740 0 90 180 300 420 540 660 780 900 1020 1140 1260 1380 1500 1620 1740
Day number
FIGURE 6.7 Example output of simulation modeling of a δ13C record from the baleen of a Southern right whale (Best
& Schell, 1996). Here a simple individual (agent)-based movement model was coupled to the carbon isotope-enabled
NEMO-MEDUSA ocean biogeochemistry-ecosystem model (Magozzi et al., 2017) and used to simulate baleen isotope
records expected under two migration scenarios: undirected foraging seeking only areas of higher production (A, C, E) or
directed latitudinal migrations matching observation records (B, D, F). A and B display core locations occupied by 50 sim-
ulated whales over a 5-year simulation. (C) and (D) show the distribution of latitudes visited by the simulated whales and
(E) and (F) show the simulated baleen δ13C records (gray) with the measured record (red). Directed migration consistent
with sightings and whaling records captures most of the features of the measured profile, suggesting that physiological
process associated with fasting is not required to explain the observed record.
zooplankton food availability). At each loca- 6.5 CONCLUSION

tion in the simulations, we extracted phyto-
plankton δ13C values from the underlying Stable isotopes (CNS) have a long and var-
spatio-temporal δ13C isoscape (Magozzi et al., ied history of application to marine migration
2017) resulting in a time series of the δ13C ecology. Isotopic approaches are particularly
values of diet potentially assimilated for each effective where animals possess incremental
day of the simulation. A more detailed tissues, and the isotopic composition of otolith
description of the agent-based model and asso- aragonite, whale baleen, and pinniped vibris-
ciated R code is available at https://github. sae in particular has yielded important insights
com/clivetrueman/Animal.migration.isotopes. into movement and foraging ecology.
We compared simulated baleen isotope data In marine systems, most spatial information
to δ13C profiles recovered from a 14.8 m long is provided by stable carbon and nitrogen iso-
adult male stranding around Port Elizabeth on topes, and the interpretation of isotopic data is
October 27, 1989 (Best & Schell, 1996). We confounded by spatial and dietary influences.
tested two migratory hypotheses to assess Assigning origin based on dietary isotopes is
whether the baleen isotope record could be effectively tracing the spatial origin of primary
explained by migration. We simulate a whale production averaged through the food web,
resident in the southern Indian Ocean, with a which is fundamentally different from tracing
starting location in January, at 40 S, north of the spatial location of the animal itself. In both
the subtropical convergence, foraging through- marine and terrestrial systems, isotopic
out the year, continually seeking areas of higher approaches drawing spatial inferences from
primary production. Second, we constrained carbon and nitrogen isotopes will always be
simulated movements to match latitudinal less reliable when applied to high trophic level
distributions of whales from fishery observa- animals in tissues with long isotopic assimila-
tions. Simulated distributions, monthly lati- tion rates and in predators targeting migratory
tudes, and associated simulated baleen δ13C prey.
profiles for 50 simulated whales are shown in The use of stable isotopes in marine spatial
Fig. 6.7. Foraging with no seasonally directed ecology has also been limited by a lack of cred-
movement yields highly variable δ13C records ible, precise, and accurate reference isoscapes
due to movement across the subtropical conver- covering ocean-basin scales. In terrestrial sys-
gence and could not reproduce the measured tems, the development of global oxygen and
baleen profile (Fig. 6.7C). The simple directed deuterium isoscapes encouraged many move-
migratory model produced a relatively good ment ecologists to adopt isotopic approaches
match to the measured profile. While inconsis- to assign origin to animals. The construction of
tencies remained, it was not clear that extensive mechanistic models predicting isotopic compo-
periods of fasting were needed to explain the sitions of phytoplankton across the global
observed baleen profile (Fig. 6.7D). Coupling ocean, validated against increasingly dense
isoscape and agent (individual)-based models field measurements, promises to do the same
allows us to design in silico computational for marine systems. Stable isotopes of carbon
experiments to better interpret measured isoto- at the base of the marine food chain vary
pic data. Simulation modeling approaches may largely with latitude, and are particularly
be especially helpful for considering combined effective tracers where consumers such as
temporal, spatial, physiological, and food-web baleen whales migrate into high latitudes in
influences on tissue isotopic compositions. summer, exploiting seasonal production.

REFERENCES 167
Stable isotopes of nitrogen show more com- Interpreting the expansion of sea fishing in medieval
plex spatial distributions, with strong gradi- Europe using stable isotope analysis of archaeological
cod bones. Journal of Archaeological Science, 38,
ents associated with the biogeochemistry of 15161524.
nitrogen fixation. At the scale of ocean basins, Bataille, C. P., & Bowen, G. J. (2012). Mapping 87Sr/86Sr
δ15N values may be most effective at identify- variations in bedrock and water for large scale prove-
ing movement between oligotrophic and nance studies. Chemical Geology, 304305, 3952.
nutrient-rich waters. Computer modeling Bauer, S., & Hoye, B. J. (2014). Migratory animals couple
biodiversity and ecosystem functioning worldwide.
offers methods to test sensitivity of isotopic Science, 344, 1242552.
tracers to changes in movement behavior in Beltran, R. S., Connolly sadou, M., Condit, R., Peterson, S.,
the context of variations in temporal and spa- Reichmuth, C., & Costa, D. P. (2015). Fine-scale whisker
tial baselines and food web dynamics. We growth measurements can reveal temporal foraging patterns
hope to see more studies combining isotopic from stable isotope signatures,. Marine Ecology Progress
Series (523, pp. 243253).
data with direct tagging, genetic and modeling Best, P. D., & Schell, D. M. S. (1996). Stable isotopes in
methods. southern right whale (Eubalaena australis) baleen as
Managing marine fisheries arguably pre- indicators of seasonal movements, feeding and growth.
sents the greatest need for information con- Marine Biology, 124, 483494.
cerning individual-scale movements and Bird, C. S., et al. (2018). A global perspective on the trophic
geography of sharks. Nature Ecology and Evolution.
migrations. Isotope methods could provide Available from https://doi.org/10.1038/s41559-017-
much needed insights, but have not been 0432-z, Online Publication January 18.
widely adopted in fishery management. This Block, B. A., Teo, S. L. H., Walli, A., Boustany, A.,
may reflect a lack of familiarity with the Stokesbury, M. J. W., Farwell, C. J., . . . Williams, T. D.
approaches, but also the relative lack of preci- (2005). Electronic tagging and population structure of
Atlantic bluefin tuna. Nature, 434, 11211127.
sion associated with isotopic assignment in Boecklen, W. J., Yarnes, C. T., Cook, B. A., & James, A. C.
marine systems. We hope that increased atten- (2011). On the Use of Stable Isotopes in Trophic
tion, methodological development, and assem- Ecology. Annual Review of Ecology, Evolution, and
bly of spatial reference data in marine systems Systematics, 42, 411440.
will allow isotopic approaches to contribute Bottrell, S. H., & Newton, R. J. (2006). Reconstruction of
changes in global sulfur cycling from marine sulfate
meaningfully to fishery management and isotopes. Earth-Science Reviews, 75, 5983.
traceability of marine products within global Bouillon, S., Connolly, R. M., & Gillikin, D. P., 2011. Use of
trade networks. stable isotopes to understand food webs and ecosystem func-
tioning in estuaries (pp. 143173). Waltham: Academic
Press.
Bowen, G. J. (2010). Isoscapes: Spatial pattern in isotopic
References biogeochemistry. Annual Review of Earth and Planetary
Aguilar, A., Gimenez, J., Gomez-Campos, E., Cardona, L., Sciences, 38, 161187.
& Borrell, A. (2014). delta15N value does not reflect fast- Boyd, I. L. (2004). Migration of marine mammals. In D.
ing in mysticetes. PLoS One, 9, e92288. Werner (Ed.), Biological resources and migration. Berliln:
Aumont, O., & Bopp, L. (2006). Globalizing results from Springer-Verlag.
ocean in situ iron fertilization studies. Global Bradshaw, P. J., Broderick, A. C., Carreras, C., Inger, R.,
Biogeochemical Cycles, 20, GB2017. Available from Fuller, W., Snape, R., . . . Godley, B. J. (2017).
https://doi.org/10.1029/2005GB002591. Satellite tracking and stable isotope analysis high-
Barnes, C., & Jennings, J. T. B. (2009). Environmental corre- light differential recruitment among foraging areas
lates of large-scale spatial variation in the δ13C of in green turtles. Marine Ecology Progress Series, 582,
marine animals. Estuarine, Coastal and Shelf Science, 81, 201214.
368374. Brennan, S. R., Zimmerman, C. E., Fernandez, D. P.,
Barrett, J. H., Orton, D., Johnstone, C., Harland, J., Van Cerling, T. E., McPhee, M. V., & Wooller, M. J. (2015).
neer, W., Ervynck, A., . . . Enghoff, I. B. (2011). Strontium isotopes delineate fine-scale natal origins

and migration histories of Pacific salmon. Science Chang, N. N., Liu, E. Y., Liao, Y. C., & Shiao, J. C. (2015).
Advances, 1, e1400124. Vertical habitat shift of viviparous and oviparous deep-
Campana, S. E. (1999). Chemistry and composition of fish sea cusk eels revealed by otolith microstructure and
otoliths. Pathways, mechanisms and applications. stable-isotope composition. Journal of Fish Biology, 86,
Marine Ecology Progress Series, 188, 263297. 845853.
Caraveo-Patiño, J., Hobson, K. A., & Soto, L. A. (2007). Connolly, R. M., Guest, M. A., Melville, A. J., & Oakes,
Feeding ecology of gray whales inferred from stable- J. M. (2004). Sulfur stable isotopes separate producers
carbon and nitrogen isotopic analysis of baleen plates. in marine food-web analysis. Oecologia, 138, 161167.
Hydrobiologia, 586, 1725. Darnaude, A. M., Sturrock, A., Trueman, C. N., Mouillot,
Cardona, L., Vales, D., Aguilar, A., Crespo, E., & Zenteno, D., EIMF., Campana, S. E., & Hunter, E. (2014).
L. (2017). Temporal variability in stable isotope ratios of Listening in on the past: What can otolith δ18O values
C and N in the vibrissa of captive and wild adult South really tell us about the environmental history of fishes?
American sea lions Otaria byronia: More than just diet PLoS One, 9, e108539.
shifts. Marine Mammal Science, 33, 975990. Dawson, M. N. (2012). Species richness, habitable volume,
Carlisle, A. B., Goldman, K. J., Litvin, S. Y., Madigan, D. J., and species densities in freshwater, the sea, and on
Bigman, J. S., Swithenbank, A. M., . . . Block, B. A. land. Frontiers of Biogeography, 4, 105116.
(2014). Stable isotope analysis of vertebrae reveals onto- Detjen, M., Sterling, E., & Gómez, A. (2015). Stable isotopes
genetic changes in habitat in an endothermic pelagic in barnacles as a tool to understand green sea turtle
shark. Proceedings of the Royal Society B: Biological (Chelonia mydas) regional movement patterns.
Sciences, 282, 20141446. Biogeosciences, 12, 70817086.
Carlisle, A. B., Kim, S. L., Semmens, B. X., Madigan, D. J., Egevang, C., Stenhouse, I. J., Phillips, R. A., Petersen, A.,
Jorgensen, S. J., Perle, C. R., . . . Block, B. A. (2012). Fox, J. W., & Silk, J. R. D. (2010). Tracking of Arctic
Using stable isotope analysis to understand the migra- terns Sterna paradisaea reveals longest animal migration.
tion and trophic ecology of northeastern Pacific white Proceedings of the National Academy of Science of the
sharks (Carcharodon carcharias). PLoS One, 7, e30492. United States of America, 107, 20782081.
Carpenter, S. J., Erickson, J. M., & Holland, F. D., jr FAO, 2016. The state of world fisheries and aquaculture 2016.
(2003). Migration of a Late Cretaceous fish. Nature, Contributing to food security and nutrition for all, Rome.
423, 7074. Fourqurean, J. W., Manuel, S., Coates, K., Kenworthy, W.,
Casciotti, K. L., McIlvin, M., & Buchwald, C. (2010). & Boyer, J. N. (2015). Water quality, isoscapes and stoi-
Oxygen isotopic exchange and fractionation during bac- chioscapes of seagrasses indicate general P limitation
terial ammonia oxidation. Limnology and Oceanography, and unique N cycling in shallow water benthos of
55, 573762. Bermuda. Biogeosciences, 12, 62356249.
Ceia, F. R., Paiva, V. H., Fidalgo, V., Morais, L., Baeta, A., France, R. L. (1995). Stable isotopic survey of the role of
Crisóstomo, P., . . . Ramos, J. A. (2014). Annual and sea- macrophytes in the carbon flow of aquatic foodwebs.
sonal consistency in the feeding ecology of an opportu- Vegetation, 124, 6772.
nistic species, the yellow-legged gull Larus michahellis. Friedland, K., Shank, B. V., Todd, C. D., McGinnity, P., &
Marine Ecology Progress Series, 497, 273284. Nye, J. A. (2014). Differential response of continental
Ceia, F. R., Ramos, J. A., Phillips, R. A., Cherel, Y., Jones, stock complexes of Atlantic salmon (Salmo salar) to the
D. C., Vieira, R. P., & Xavier, J. C. (2015). Analysis of Atlantic Multidecadal Oscillation. Journal of Marine
stable isotope ratios in blood of tracked wandering Systems, 133, 7787.
albatrosses fails to distinguish a δ13C gradient within Fry, B. (2002). Conservative mixing of stable isotopes
their winter foraging areas in the southwest Atlantic across estuarine salinity gradients: A conceptual frame-
Ocean. Rapid Communications in Mass Spectrometry, 29, work for monitoring watershed influences on down-
23282336. stream fisheries production. Estuaries and Coasts, 25,
Ceriani, S. A., Roth, J. D., Sasso, C. R., Mcclellan, C. M., 264271.
James, M. C., Haas, H. L., . . . Bagley, D. A. (2014). Godiksen, J. A., Svenning, M., Dempson, J. B., Storm-Suke,
Modeling and mapping isotopic patterns in the A., & Power, M. (2010). Development of a species-
Northwest Atlantic derived from loggerhead sea tur- specific fractionation equation for Arctic charr
tles. Ecosphere, 5, 124. (Salvelinus alpinus (L.)): An experimental approach.
Ceriani, S. A., Weishampel, J. F., Ehrhart, L. M., Mansfield, Hydrobiologia, 650, 6777.
K. L., & Wunder, M. B. (2017). Foraging and recruit- Graham, B. S., Koch, P. L., Newsome, S. D., McMahon,
ment hotspot dynamics for the largest Atlantic logger- K. W., & Aurioles, D. (2010). Using isoscapes to trace
head turtle rookery. Scientific Reports, 7, 16894. the movements and foraging behavior of top predators

REFERENCES 169
in oceanic ecosystems. In J. B. West, G. J. Bowen, T. E. Jones, N. M., Hoover, B. A., Heppell, S. A., & Kuletz, K. J.
Dawson, & K. P. Tu (Eds.), Isoscapes: Understanding (2014). A cross-shelf gradient in δ 15 N stable isotope
movement, pattern, and process on Earth through isotope values of krill and pollock indicates seabird foraging
mapping. Dordrecht: Springer. patterns in the Bering Sea. Deep Sea Research Part II:
Hannides, C. C. S., Popp, B. N., Landry, M. R., & Graham, Topical Studies in Oceanography, 109, 241250.
B. S. (2009). Quantification of zooplankton trophic posi- Kalish, J. M. (1991). 13C and 18O isotopic disequilibria in
tion in the North Pacific Subtropical Gyre using fish otoliths: Metabolic and kinetic effects. Marine
stable nitrogen isotopes. Limnology and Oceanography, Ecology Progress Series, 75, 191203.
54, 5061. Kernaléguen, L., Cazelles, B., Arnould, J. P., Richard, P.,
Hanson, N. N., Wurster, C. M., EIMF., & Todd, C. D. Guinet, C., & Cherel, Y. (2012). Long-term species, sex-
(2010). Comparison of secondary ion mass spectrometry ual and individual variations in foraging strategies of
and micromilling/continuous flow isotope ratio mass fur seals revealed by stable isotopes in whiskers. PLoS
spectrometry techniques used to acquire intra-otolith One, 7, e32916.
delta18O values of wild Atlantic salmon (Salmo salar). Killingley, J. S. (1980). Migrations of California gray whales
Rapid Communications in Mass Spectrometry, 24, tracked by oxygen-18 variations in their epizoic barna-
24912498. cles. Science, 207, 759760.
Henderson, P. (1984). Inorganic geochemistry. Oxford: Killingley, J. S., & Lutcavage, M. (1983). Loggerhead turtle
Pergamon Press. movements reconstructed from 18O and 13C profiles
Hobson, K. A. (2005). Using stable isotopes to trace long- from commensal barnacle shells. Estuarine, Coastal and
distance dispersal in birds and other taxa. Diversity and Shelf Science, 16, 345349.
Distributions, 11, 157164. Koenigs, C., Miller, R. J., & Page, H. M. (2015). Top preda-
Hofmann, M., Wolf-Gladrow, D. A., Takahashi, T., tors rely on carbon derived from giant kelp Macrocystis
Sutherland, S. C., Six, K. D., & Maier-Reimer, E. (2000). pyrifera. Marine Ecology Progress Series, 537, 18.
Stable carbon isotope distribution of particulate organic Kurle, C. M., & McWhorter, J. K. (2017). Spatial and tem-
matter in the ocean: A model study. Marine Chemistry, poral variability within marine isoscapes: Implications
72, 131150. for interpreting stable isotope data from marine sys-
Høie, H. (2004). Temperature-dependent fractionation of tems. Marine Ecology Progress Series, 568, 3145.
stable oxygen isotopes in otoliths of juvenile cod Kurle, C. M., Sinclair, E. H., Edwards, A. E., &
(Gadus morhua L.). ICES Journal of Marine Science, 61, Gudmundson, C. J. (2011). Temporal and spatial varia-
243251. tion in the δ15N and δ13C values of fish and squid from
Houssard, P., Lorrain, A., Tremblay-Boyer, L., Allain, V., Alaskan waters. Marine Biology, 158, 23892404.
Graham, B. S., Menkes, C. E., . . . Leroy, B. (2017). Laws, E. A., Popp, B. N., Bidigare, R. R., Kennicutt, M. C.,
Trophic position increases with thermocline depth in & Macko, S. A. (1995). Dependence of phytoplankton
yellowfin and bigeye tuna across the Western and carbon isotopic composition on growth rate and CO2
Central Pacific Ocean. Progress in Oceanography, 154, (aq)—Theoretical considerations and experimental
4963. results. Geochimica et Cosmochimica Acta, 59, 11311138.
Irigoien, X., Klevjer, T. A., Røstad, A., Martinez, U., Boyra, LeGrande, A. N., & Schmidt, G. A. (2006). Global gridded
G., Acuña, J. L., . . . Kaartvedt, S. (2014). Large mesope- data set of the oxygen isotopic composition in seawater.
lagic fishes biomass and trophic efficiency in the open Geophysical Research Letters, 33, L12604.
ocean. Nature Communications, 5, 3271. Lorrain, A., Graham, B. S., Popp, B. N., Allain, V., Olson,
Jaeger, A., Lecomte, V. J., Weimerskirch, H., Richard, P., & R. J., Hunt, B. P., & Menkes, C. E. (2015). Nitrogen iso-
Cherel, Y. (2010). Seabird satellite tracking validates the topic baselines and implications for estimating foraging
use of latitudinal isoscapes to depict predators’ forag- habitat and trophic position of yellowfin tuna in the
ing areas in the Southern Ocean. Rapid Communications Indian and Pacific Oceans. Deep Sea Research Part II:
in Mass Spectrometry, 24, 34563460. Topical Studies in Oceanography, 113, 188198.
Jennings, S., Barnes, C., Sweeting, C. J., & Polunin, N. V. Lott, C. A., Meehan, T. D., & Heath, J. A. (2003).
(2008). Application of nitrogen stable isotope analysis Estimating the latitudinal origins of migratory birds
in size-based marine food web and macroecological using hydrogen and sulfur stable isotopes in feathers:
research. Rapid Communications in Mass Spectrometry, 22, Influence of marine prey base. Oecologia, 134, 505510.
16731680. MacAvoy, S. E., Macko, S. A., & Garman, G. C. (1998).
Jennings, S., & Warr, K. (2003). Environmental correlates of Tracing marine biomass into tidal freshwater ecosys-
large-scale spatial variation in the delta 15N of marine tems using stable sulfur isotopes. Naturwissenschaften,
animals. Marine Biology, 142, 11311140. 85, 544546.

MacKenzie, K., Longmore, C., Preece, C., Lucas, C., & eastern Pacific Ocean. Progress in Oceanography, 86,
Trueman, C. (2014). Testing the long-term stability of 124138.
marine isoscapes in shelf seas using jellyfish tissues. Pauli, J. N., Newsome, S. D., Cook, J. A., Harrod, C.,
Biogeochemistry, 114. Steffan, S. A., Baker, C. J., . . . Hayden, B. (2017).
MacKenzie, K. M., Palmer, M. R., Moore, A., Ibbotson, Opinion: Why we need a centralized repository for iso-
A. T., Beaumont, W. R., Poulter, D. J., & Trueman, topic data. Proceedings of National Academy of Sciences of
C. N. (2011). Locations of marine animals revealed by the United States of America, 114, 29973001.
carbon isotopes. Scientific Reports, 1, 21. Peterson, B. J., & Fry, B. (1987). Stable isotopes in ecosys-
Mackey, A. P., Hyndes, G. A., Carvalho, M. C., & Eyre, tem studies. Annual Review of Ecology, Evolution, and
B. D. (2015). Physical and biogeochemical correlates of Systematics, 18, 293320.
spatio-temporal variation in the δ 13C of marine macro- Pethybridge, H. R., Young, J. W., Kuhnert, P. M., & Farley,
algae. Estuarine, Coastal and Shelf Science, 157, 718. J. H. (2015). Using stable isotopes of albacore tuna and
Magozzi, S., Yool, A., Vander Zanden, H., Wunder, M., & predictive models to characterize bioregions and exam-
Trueman, C. (2017). Using ocean models to predict spa- ine ecological change in the SW Pacific Ocean. Progress
tial and temporal variation in marine carbon isotopes. in Oceanography, 134, 293303.
Ecosphere, 8. Phillips, R. A., Bearhop, S., McGill, R. A., & Dawson, D. A.
McMahon, K. W., Hamady, L. L., & Thorrold, S. R. (2013). (2009). Stable isotopes reveal individual variation in
A review of ecogeochemistry approaches to estimating migration strategies and habitat preferences in a suite
movements of marine animals. Limnology and of seabirds during the nonbreeding period. Oecologia,
Oceanography, 58, 697714. 160, 795806.
Michener, R., & Lajtha, K. (2008). Stable isotopes in ecology and Quillfeldt, P., Ekschmitt, K., Brickle, P., McGill, R. A.,
environmental science. New York: John Wiley and Sons. Wolters, V., Dehnhard, N., & Masello, J. F. (2015).
Miller, R. J., & Page, H. M. (2012). Kelp as a trophic Variability of higher trophic level stable isotope data
resource for marine suspension feeders: A review of in space and timea case study in a marine ecosys-
isotope-based evidence. Marine Biology, 159, 13911402. tem. Rapid Communications in Mass Spectrometry, 29,
Montoya, J. P. (2007). Natural abundance of 15N in marine 18.
planktonic ecosystems. In K. Lajtha, & R. H. Michener Quillfeldt, P., Masello, J. F., McGill, R. A. R., Adams, M., &
(Eds.), Stable isotopes in ecology and environmental science. Furness, R. W. (2010). Moving polewards in winter: A
New York: Wiley. recent change in the migratory strategy of a pelagic sea-
Moore, J. W., & Semmens, B. X. (2008). Incorporating bird? Frontiers in Zoology, 7.
uncertainty and prior information into stable isotope Radabaugh, K. R., Hollander, D. J., & Peebles, E. B. (2013).
mixing models. Ecology Letters, 11, 470480. Seasonal δ 13C and δ 15N isoscapes of fish populations
Morris, E. P., Peralta, G., Benavente, J., Freitas, R., along a continental shelf trophic gradient. Continental
Rodrigues, A. M., Quintino, V., . . . Pérez-Lloréns, J. L. Shelf Research, 68, 112122.
(2009). Caulerpa prolifera stable isotope ratios reveal Radabaugh, K. R., & Peebles, E. B. (2014). Multiple regres-
anthropogenic nutrients within a tidal lagoon. Marine sion models of δ 13C and δ 15N for fish populations in
Ecology Progress Series, 390, 117128. the eastern Gulf of Mexico. Continental Shelf Research,
Navarro, J., Coll, M., Somes, C. J., & Olson, R. J. (2013). 84, 158168.
Trophic niche of squids: Insights from isotopic data in Ramos, R., & Gonzalez-Solis, J. (2012). Trace me if you can:
marine systems worldwide. Deep Sea Research Part II: The use of intrinsic biogeochemical markers in marine
Topical Studies in Oceanography, 95, 93102. top predators. Frontiers in Ecology and the Environment,
Newsome, S. D., Clementz, M. T., & Koch, P. L. (2010). 10, 258266.
Using stable isotope biogeochemistry to study marine Rau, G. H., Riebesell, U., & Wolf-Gladrow, D. (1996). A
mammal ecology. Marine Mammal Science. model of photosynthetic 13C fractionation by marine
Niño-Torres, C. A., Gallo-Reynoso, J. P., Galván-Magaña, phytoplankton based on diffusive molecular CO2
F., Escobar-Briones, E., & Macko, S. A. (2006). Isotopic uptake. Marine Ecology Progress Series, 133, 275285.
analysis of δ13C, δ15N, and δ34S “a feeding tale” in teeth Reich, K. J., Bjorndal, K. A., & Bolten, A. B. (2007). The
of the longbeaked common dolphin, Delphinus capensis. ‘lost years’ of green turtles: Using stable isotopes to
Marine Mammal Science, 22, 831846. study cryptic lifestages. Biological Letters, 3,
Olson, R. J., Popp, B. N., Graham, B. S., Lopez-Ibarra, 712714.
G. A., Galvan-Magana, F., Lennert-Cody, C. E., . . . Fry, Rodhouse, P. G. K. (2004). Migration of fishery resources
B. (2010). Food-web inferences of stable isotope spatial in the world’s oceans. In W. Dietrich (Ed.), Biological
patterns in copepods and yellowfin tuna in the pelagic resources and migration. Berlin: Springer-Verlag.

REFERENCES 171
Rooker, J. R., Arrizabalaga, H., Fraile, I., Secor, D. H., Shephard, S., Trueman, C., Rickaby, R., & Rogan, E. (2007).
Dettman, D. L., Abid, N., . . . Santos, M. N. (2014). Juvenile life history of NE Atlantic orange roughy from
Crossing the line: Migratory and homing behaviors of otolith stable isotopes. Deep Sea Research Part I:
Atlantic bluefin tuna. Marine Ecology Progress Series, Oceanographic Research Papers, 54, 12211230.
504, 265276. Shiao, J. C., Liu, E. Y., & Sui, T. D. (2016). Up-and-down
Rooker, J. R., Secor, D. H., De metrio, G., Schloesser, R. W., shift in residence depth of slickheads
Block, B. A., & Neilson, J. D. (2008). Natal homing and (Alepocephalidae) revealed by otolith stable oxygen
connectivity in Atlantic bluefin tuna populations. isotopic composition. Journal of Fish Biology, 88,
Science, 322, 742744. 12651272.
Ruiz-Cooley, R. I., Villa, E. C., & Gould, W. R. (2010). Shiao, J.-C., Sui, T.-D., Chang, N.-N., & Chang, C.-W.
Ontogenetic variation of δ13C and δ15N recorded in the (2017). Remarkable vertical shift in residence depth
gladius of the jumbo squid Dosidicus gigas: Geographic links pelagic larval and demersal adult jellynose fish.
differences. Marine Ecology Progress Series, 399, 187198. Deep Sea Research Part I: Oceanographic Research Papers,
Ryabenko, E. (2013). Stable isotope methods for the study 121, 160168.
of the nitrogen cycle. In E. Zambianchi (Ed.), Topics in Somes, C. J., Oschlies, A., & Schmittner, A. (2013). Isotopic
oceanography. Rijeka: InTech. constraints on the pre-industrial oceanic nitrogen bud-
Schell, D. M., Barnett, B. A., & Vinette, K. A. (1998). get. Biogeosciences, 10, 5889.
Carbon and nitrogen isotope ratios in zooplankton of Somes, C. J., Schmittner, A., Galbraith, E. D., Lehmann,
the Bering, Chukchi and Beaufort seas. Marine Ecology M. F., Altabet, M. A., Montoya, J. P., . . . Eby, M.
Progress Series, 162, 1123. (2010). Simulating the global distribution of nitrogen
Schell, D. M., Saupe, S. M., & Haubenstock, N. (1989). isotopes in the ocean. Global Biogeochemical Cycles, 24,
Bowhead whale (Balaena mysticetus) growth and feeding GB4019.
as estimated by d13C techniques. Marine Biology, 103, Stephenson, P. (2001). Analysis of stable isotope ratios to
433443. investigate stock structure of red emperor and Rankin
Schloesser, R. W., Gillanders, B., Neilson, J. D., Secor, cod in northern Western Australia. Journal of Fish
D. H., & Rooker, J. R. (2010). Natal origin of Atlantic Biology, 58, 126144.
bluefin tuna (Thunnus thynnus) from Canadian waters Stephenson, R. L., Tan, F. C., & Mann, K. H. (1984).
based on otolith δ13C and δ18O. Canadian Journal of Stable carbon isotope variability in marine macrophytes
Fisheries and Aquatic Sciences, 67, 563569. and its implications for food web studies. Marine
Schmidt, G.A., Bigg, G.R. and Rohling, E.J. 1999. Global Biology, 1981.
Seawater Oxygen-18 Database, http://data.giss.nasa. Stuben, D., Berner, Z., Chandrasekharam, D., & Karmakar,
gov/o18data/, NASA Goddard Institute of Space J. (2003). Arsenic enrichment in groundwater of West
Science, New York. Bengal, India: Geochemical evidence for mobilization of As
Schmittner, A., Oschlies, A., Matthews, H. D., & Galbraith, under reducing conditions, . Applied Geochemistry (18,
E. D. (2008). Future changes in climate, ocean circulation, pp. 14171434). .
ecosystems, and biogeochemical cycling simulted for a Sturrock, A. M., Hunter, E., Milton, J. A., Johnson, R. C.,
business-as-usual CO2 emission scenario until year Waring, C. P., Trueman, C. N., & Leder, E. (2015).
4000AD. Global Biogeochemical Cycles, 22, GB1013. Quantifying physiological influences on otolith micro-
Available from https://doi.org/10.1029/2007/GB002953. chemistry. Methods in Ecology and Evolution, 6, 806816.
Schmittner, A., Gruber, N., Mix, A., Key, R., Tagliabue, A., Sturrock, A. M., Trueman, C. N., Darnaude, A. M., &
& Westberry, T. (2013). Biology and airsea gas Hunter, E. (2012). Can otolith elemental chemistry ret-
exchange controls on the distribution of carbon isotope rospectively track migrations in fully marine fishes?
ratios (δ13C) in the ocean. Biogeosciences, 10, 57935816. Journal of Fish Biology, 81, 766795.
Schmittner, A., & Somes, C. J. (2016). Complementary con- Sturrock, A. M., Trueman, C. N., Milton, J. A., Waring,
straints from carbon (13C) and nitrogen (15N) isotopes C. P., Cooper, M. J., & Hunter, E. (2014). Physiological
on the glacial ocean’s soft-tissue biological pump. influences can outweigh environmental signals in oto-
Paleoceanography, 31, 669693. lith microchemistry research. Marine Ecology Progress
Secor, D. H. (2015a). Migration ecology of marine fishes. Series, 500, 245264.
Baltimore: Johns Hopkins University Press. Tagliabue, A., & Bopp, L. (2008). Towards understanding
Secor, D. H. (2015b). Synopsis of regional mixing levels for global variability in ocean carbon-13. Global
Atlantic bluefin tuna estimated from otolith Biogeochemical Cycles, 22, 13.
stable isotope analysis, 20072014. Collective Volume of Tesdal, J.-E., Galbraith, E., & Kienast, M. (2013).
Scientific Papers ICCAT, 71, 16831689. Nitrogen isotopes in bulk marine sediment: Linking

seafloor observations with subseafloor records. Walther, B. D., & Limburg, K. E. (2012). The use of otolith
Biogeosciences, 10, 101. chemistry to characterize diadromous migrations.
Thomas, S. M., & Crowther, T. W. (2015). Predicting rates Journal of Fish Biology, 81, 796825.
of isotopic turnover across the animal kingdom: A syn- Weber, P. K., Hutcheon, I. D., McKeegan, K. D., & Ingram,
thesis of existing data. Journal of Animal Ecology, 84, B. L. (2002). Otolith sulfur isotope method to recon-
861870. struct salmon (Oncorhynchus tshawytscha) life history.
Torniainen, J., Lensu, A., Vuorinen, P. J., Sonninen, E., Canadian Journal of Fisheries and Aquatic Sciences, 59,
Keinänen, M., Jones, R. I., . . . Kiljunen, M. (2017). 587591.
Oxygen and carbon isoscapes for the Baltic Sea: Testing West, J. B., Bowen, G. J., Dawson, T. E., & Tu, K. P.
their applicability in fish migration studies. Ecology and (2010). Isoscapes. Understanding movement, pattern, and
Evolution, 7, 22552267. process on Earth through isotope mapping. New York:
Trueman, C. N., MacKenzie, K. M., & Palmer, M. R. (2012). Springer.
Identifying migrations in marine fishes through stable- Witteveen, B. H., Worthy, G. A. J., & Roth, J. D. (2009).
isotope analysis. Journal of Fish Biology, 81, 826847. Tracing migratory movements of breeding North
Trueman, C. N., MacKenzie, K. M., & St John Glew, K. Pacific humpback whales using stable isotope analysis.
(2017). Stable isotope-based location in a shelf sea set- Marine Ecology Progress Series, 393, 173183.
ting: Accuracy and precision are comparable to light- Wolf, N., Carleton, S. A., & Del rio, C. M. (2009). Ten years
based location methods. Methods in Ecology and of experimental animal isotopic ecology. Functional
Evolution, 8, 232240. Ecology, 23, 1726.
Tucker, A. D., Macdonald, B. D., & Seminoff, J. A. (2014). Yool, A., Popova, E. E., & Anderson, T. R. (2013). MEDUSA-
Foraging site fidelity and stable isotope values of log- 2.0: An intermediate complexity biogeochemical model
gerhead turtles tracked in the Gulf of Mexico and of the marine carbon cycle for climate change and ocean
Northwest Caribbean. Marine Ecology Progress Series, acidification studies. Geoscientific Model Development, 6,
502, 267279. 17671811.
Vander Zanden, H. B., Bjorndal, K. A., Reich, K. J., & Young, J., Olson, R., Ménard, F., Kuhnert, P., Duffy, L.,
Bolten, A. B. (2010). Individual specialists in a general- Allain, V., . . . Graham, B. (2015). Setting the stage for a
ist population: Results from a long-term stable isotope global-scale trophic analysis of marine top predators: A
series. Biological Letters, 6, 711714. multi-workshop review. Reviews in Fish Biology and
Vander Zanden, H. B., Tucker, A. D., Hart, K. M., Fisheries, 25, 261272.
Lamont, M. M., Fujisaki, I., Addison, D. S., . . . Bowen, Young, J. W., Olson, R. J., Ménard, F., Kuhnert, P. M.,
G. J. (2015). Determining origin in a migratory marine Duffy, L. M., Allain, V., . . . Choy, C. A. (2014). Setting
vertebrate: A novel method to integrate stable isotopes the stage for a global-scale trophic analysis of marine
and satellite tracking. Ecological Applications, 25, top predators: A multi-workshop review. Reviews in
320335. Fish Biology and Fisheries, 25, 261272.
Vokhshoori, N. L., & McCarthy, M. D. (2014). Compound- Zupcic-Moore, J. R., Ruiz-Cooley, R. I., Paliza, O., Koch,
specific δ15N amino acid measurements in littoral mus- P. L., & Mccarthy, M. D. (2017). Using stable isotopes to
sels in the California upwelling ecosystem: A new investigate foraging variation and habitat use of sperm
approach to generating baseline δ15N isoscapes for whales from northern Peru. Marine Ecology Progress
coastal ecosystems. PloS One, 9, e98087. Series, 579, 201212.

C H A P T E R
7
Amino Acid Isotope Analysis: A New
Frontier in Studies of Animal Migration
and Foraging Ecology
Kelton W. McMahon1 and Seth D. Newsome2
1
University of Rhode Island, Narragansett, RI, United States, 2University of New Mexico,
Albuquerque, NM, United States
7.1 INTRODUCTION Ecogeochemistry is the application of geo-

chemical techniques to fundamental questions
Whether it is diel, meter-scale migrations of in ecology (McMahon, Hamady, & Thorrold,
zooplankton toward the ocean surface or the 2013a). This field offers a powerful suite of
annual 70,000 km global migrations of Arctic tools to explore the intricate connections
Terns (Sterna paradisaea), animal migration between movement and foraging dynamics.
represents one of the most biologically signifi- Physical, chemical, and biological processes in
cant redistributions of biomass on Earth nature create geospatial structure in the distri-
(Dingle, 2014). Accordingly, there is a need to bution of stable isotopes; mapping these distri-
develop powerful, precise, and integrative butions illustrates the invisible “isoscapes” in
tools to describe and quantify these spectacu- which organisms operate (Bowen, 2010). As
lar migratory movements (Hobson, 1999; animals interact with isoscapes, they incorpo-
Hussey et al., 2015; Kays, Crofoot, Jetz, & rate the local isotopic signals of the environ-
Wikelski, 2015; Chapter 1: Animal Migration: ment into their tissues (Graham, Koch,
A Context for Using New Techniques and Newsome, McMahon, & Aurioles, 2010;
Approaches). Given that the acquisition and Hobson, 1999). Herein lies one of the biggest
allocation of dietary resources is a fundamen- challenges in interpreting isotope data in the
tal requirement for all animals and an integral context of movement ecology: the isotope com-
part of animal movement (Fryxell & Sinclair, position of a consumer reflects information
1988), movement ecology is intimately linked about both the baseline isoscape and the sub-
with foraging ecology in space and time. sequent biochemical modifications of organic

174 7. AMINO ACID ISOTOPE ANALYSIS: A NEW FRONTIER IN STUDIES OF ANIMAL MIGRATION AND FORAGING ECOLOGY
matter as it flows through food webs three major themes: (1) using CSIA-AA to
(Chapter 4: Application of Isotopic Methods to refine isoscapes, (2) disentangling movement
Tracking Animal Movements). and foraging ecology, and (3) examining
The trophic modification of organic matter movement and resource utilization across dif-
and its isotopic composition results in a series ferent ecosystems. We provide a brief over-
of complex ecogeochemical “filters” between view of the analytical process needed to
the baseline isoscape and the corresponding generate and interpret CSIA-AA data and an
consumer tissue isotope δ-value. To use outlook toward future research needed to
stable isotopes to track animal movement expand and refine the use of CSIA-AA for
among isoscapes, we need to understand how studying animal movement and foraging
these filters modify the isotope value of consu- ecology.
mers relative to their baseline isoscapes.
Conventional isotopic approaches to examin-
ing movement ecology have focused on the
7.2 PRIMER ON AMINO ACID
isotope analysis of “bulk” tissues, i.e., the
BIOCHEMISTRY AND ISOTOPE
weighted average of all components within in
DISCRIMINATION
a specific tissue. While quite successful
(Hobson & Wassenaar, 2008), it can be difficult
To track animal movement through space
to determine whether variation in bulk tissue
and time with stable isotopes, we must be able
isotope values is due to differences in (1) diet
to link the isotope δ-value(s) of a consumer tis-
(e.g., trophic level), (2) tissue types analyzed,
sue to an underlying baseline isoscape.
(3) physiology or metabolism, (4) isotopic com-
However, the isotope values of the baseline iso-
position at the base of the food web, or (5)
scape (e.g., primary producers) are potentially
some combination of these factors (Post, 2002).
modified through consumer metabolism in a
This can be particularly problematic when
variety of ways depending on the number and
studying mobile organisms in which resource
isotopic effect of enzymatic reactions, as well as
and habitat use change across space and time.
the flux of elements through these metabolic
Whereas bulk tissue isotope analysis
pathways (reviewed in Hayes, 2001; McMahon
averages all macromolecules in a sample,
& McCarthy, 2016; Ohkouchi et al., 2017). The
compound-specific isotope analysis of amino
differential isotope fractionation of individual
acids (CSIA-AA) takes a molecular approach
AAs during metabolism provides a mechanism
based on well-established biochemical path-
to disentangle signals of the baseline isoscape
ways. The power of the molecular approach
from subsequent metabolic modifications of
lies in the differential isotopic (e.g., δ2H, δ13C,
consumer isotope values. Therefore in this sec-
δ15N) discrimination of individual AAs during
tion we present a brief primer on the basics of
the transfer of organic matter between diet and
how AA metabolism influences nitrogen, car-
consumer (i.e., trophic transfer) or biochemical
bon, and hydrogen isotopic discrimination.
processing within the consumer. In this chap-
ter, we explore the potential of CSIA-AA to
study movement and foraging dynamics. We
start with a brief primer on the roles of metab-
7.2.1 Nitrogen Metabolism
olism and physiology on AA stable isotopic Transamination (transferring an amine group)
discrimination. We highlight the strengths and and deamination (removing an amine group)
limitations of the CSIA-AA approach to move- are the two dominant enzymatic processes that
ment ecology by describing case studies under control the flow of nitrogen, and thus nitrogen

7.2 PRIMER ON AMINO ACID BIOCHEMISTRY AND ISOTOPE DISCRIMINATION 175
isotope fractionation, in proteinaceous AAs thought to directly reflect the δ15N of the
(Braun, Vikari, Windisch, & Auerswald, 2014). baseline isoscape without the confounding issue
The diversity of transaminases and deami- of trophic isotope discrimination. However,
nases, each with different N isotope effects, alternative metabolic pathways can impart N
and the variable degree of transamination and isotope fractionation via transamination or
deamination among AAs, result in a wide deamination for many source AAs. 15N-labeling
range of nitrogen isotopic discrimination pat- experiments have shown that the central pool
terns among individual AAs (McMahon & of dietary nitrogen can be incorporated into
McCarthy, 2016). For δ15N analysis, protein Phe, albeit at low levels relative to most other
AAs are commonly divided into two groups, AAs (Hoskin, Gavet, Milne, & Lobley, 2001).
the minimally discriminating “source” AAs Identifying under what dietary and physiologi-
and the heavily discriminating “trophic” AAs cal conditions source AA 15N discrimination
(Popp et al., 2007), based on their nitrogen becomes significant is a current area of research
metabolism and corresponding isotopic dis- in CSIA-AA.
crimination. While this division is often con- Finally, it should be noted that several other
fused with the more familiar essential versus AAs that were originally termed “source”
nonessential AA groupings for carbon, these AAs, namely Glycine (Gly), Serine (Ser), and
groupings not only represent different AAs, Threonine (Thr) (Popp et al., 2007), were done
but are also based on fundamentally different so based largely on the empirical results
biochemical mechanisms related to nitrogen reported in McClelland and Montoya (2002).
cycling (source and trophic AAs) versus syn- Subsequently, across a much broader range of
thesis of carbon sidechains of AAs (essential consumers, the variability in mean ( 6 SD)
and nonessential AAs). Below we explore the Δ15NCD values for these AAs has been shown
mechanisms of differential 15N discrimination to be extremely large: Gly (3.9 6 4.9m,), Ser
among AAs, which allow ecologists to disen- (2.9 6 4.6m), and Thr (25.8 6 3.2m) (McMahon
tangle the relative influence of baseline versus & McCarthy, 2016). As a result, we suggest
trophic isotope variability on consumer δ15N caution be used when interpreting δ15N values
values. of these particular AAs in the context of ani-
mal movement and foraging ecology without
7.2.1.1 Minimally Fractionating “Source” further mechanistic studies of their isotopic
Amino Acids discrimination patterns.
The source AAs (e.g., phenylalanine: Phe;
methionine: Met; lysine: Lys) show relatively 7.2.1.2 Heavily Fractionating “Trophic”
little N isotope discrimination between diet and Amino Acids
consumer, likely because their dominant meta- The trophic AAs (glutamic acid: Glu; aspar-
bolic pathways typically do not form or break tic acid: Asp; alanine: Ala; isoleucine: Ile; leu-
CN bonds during metabolism (O’Connell, cine: Leu; proline: Pro; valine: Val) typically
2017). A metaanalysis of published controlled undergo significant N isotopic discrimination
feeding studies (70 species, 317 individuals, and during metabolism associated with extensive
88 distinct consumerdiet combinations) found transamination and deamination (O’Connell,
minimal N isotopic discrimination of Phe 2017). The canonical trophic AA Glu often has
(Δ15NCD 5 20.1 6 1.6m), Met (0.4 6 0.4m), and the highest mean Δ15NCD of all AAs
Lys (0.8 6 1.5m) between consumer (C) and diet (6.4 6 2.5m; McMahon & McCarthy, 2016). The
(D) (McMahon & McCarthy, 2016). The δ15N other trophic AAs typically exhibit discrimina-
values of these “source” AAs are therefore tion patterns that closely resemble Glu because

they exchange nitrogen with the central nitro- (Wu, 2009). Typically, δ13C data are reported
gen pool in an organism via transamination for seven essential AAs (Ile, Leu, Lys, Met,
chains linked to Glu (McCarthy, Lehman, & Phe, Thr, and Val), which contribute B25%
Kudela, 2013; O’Connell 2017). 40% of the AA budget of animal tissues com-
Together, the δ15N values of a consumer’s monly analyzed by ecologists (Wolf,
trophic and source AAs provide ecologists Newsome, Peters, & Fogel, 2015). Conversely,
with a potential tool to calculate consumer tro- nonessential AAs are those that organisms can
phic position (TPCSIA-AA) that is internally de novo synthesize from a common carbon
indexed to the δ15N of the base of the food pool (Wu, 2009). δ13C data are routinely
web. However, several controlled feeding reported for eight nonessential AAs (Gly, Ser,
studies have noted that the degree of N iso- Ala, Glu, Asp, Pro, arginine: Arg, and tyrosine:
tope discrimination of trophic AAs between Tyr), which contribute B60%75% to the total
diet and consumer is not constant AAs budget of animal tissues (Wolf et al.,
(Chikaraishi, Steffan, Takano, & Ohkouchi, 2015). Some of these nonessential AAs are con-
2015; McMahon, Polito, Abel, McCarthy, & sidered conditionally essential (e.g., Pro, Arg,
Thorrold, 2015; McMahon, Thorrold, Elsdon, & and Tyr), meaning their de novo synthesis can
McCarthy, 2015; Yamaguchi et al., 2017). As be limited under certain physiological condi-
with trophic patterns observed in bulk tissues tions. In contrast to the general classification
(Vanderklift & Ponsard, 2003), diet quality and scheme used for δ15N (source vs trophic) that
consumer mode of nitrogen excretion (e.g., is empirically derived, these essential and non-
ammonia vs urea) are potentially key variables essential designations for δ13C are based on
influencing the degree of trophic AA isotopic well-established biochemical pathways.
discrimination (Germain, Koch, Harvey, &
McCarthy, 2013; McMahon, Thorrold, et al., 7.2.2.1 Essential Amino Acids
2015; O’Connell, 2017). This variability in Animals have lost the enzymatic pathways
nitrogen isotopic discrimination among trophic required to synthesize sufficient quantities of
AAs can significantly impact the accuracy of essential AAs and hence must acquire the
trophic position estimates (e.g., Dale, intact carbon skeletons of essential AAs
Wallsgrove, Popp, & Holland, 2011; Lorrain directly from their dietary proteins (Howland
et al., 2009). Therefore we need to increase the et al., 2003; McMahon, Fogel, Elsdon, &
number of controlled feeding studies examin- Thorrold, 2010; O’Brien, Fogel, & Boggs, 2002)
ing AA isotope fractionation to better under- or from their prokaryotic gut microbiome
stand and account for the underlying (Ayayee et al. 2015; Newsome, Fogel, Kelly, &
mechanisms controlling variability in trophic Martinez del Rio, 2011). As a result, essential
AA 15N discrimination. AAs typically show minimal 13C isotopic dis-
crimination between diet (or gut microbe
essential AAs) and consumer tissue (Howland
et al., 2003; McMahon et al., 2010). Consumer
7.2.2 Carbon Metabolism essential AA δ13C values therefore represent
For carbon, AAs have conventionally been the isotopic signature of the producers of those
classified into two categories, essential and AAs at the base of the food web, without the
nonessential AAs, which relate to an organ- confounding variable of significant trophic
ism’s ability for de novo synthesis of AA side discrimination.
chains. Animals cannot synthesize essential On the other hand, plants, algae (protists),
AAs at a rate adequate for normal growth bacteria, and fungi can synthesize essential

7.2 PRIMER ON AMINO ACID BIOCHEMISTRY AND ISOTOPE DISCRIMINATION 177
AAs de novo from a variety of inorganic and and (3) route them directly from nonessential
organic carbon sources. These organisms use a AAs produced by the gut microbiome from
variety of pathways and associated isotope dietary carbohydrate (Ayayee et al., 2015;
effects to synthesize common essential AAs Newsome et al., 2011) and lipid precursors
(Hayes, 2001). This metabolic diversity in syn- (Newsome et al. 2014). Nonessential AAs can
thesis pathways imprints on the relative δ13C be further grouped into two general types: (1)
values of essential AAs synthesized by differ- those synthesized from intermediaries in gly-
ent groups of producers. Multivariate analyses colysis (Gly/Ser/Ala) and (2) those synthe-
using carbon isotope data from a suite of sized from intermediaries in the tricarboxylic
essential AAs results in unique essential AA acid (TCA) cycle (Glu/Asp/Pro/Arg), both of
δ13C “fingerprints” among plants, algae, bacte- which have distinct sources of carbon and syn-
ria, and fungi (Larsen, Taylor, Leigh, & thesis pathways. However, routing of intact
O’Brien, 2009; Larsen et al., 2013; Scott et al. nonessential AAs directly from dietary protein
2006). Assuming the contribution of essential into consumer tissues (just like essential AAs)
AAs synthesized de novo by gut microbiota is is energetically most favorable (Schwarcz,
minimal (but see Ayayee, Jones, & Sabree, 1991), as the use of premanufactured carbon
2015 and Newsome et al. 2011 for counterargu- skeletons reduces the metabolic costs of syn-
ments), then direct assimilation (routing) of thesis (Wu et al., 2014). Isotopic discrimination
dietary essential AAs results in little C isotopic of C between nonessential AAs and bulk diet
alteration of these molecules as they are is highly variable and is likely dependent on
passed from diet to consumer (Howland et al., the degree of de novo synthesis, the nonpro-
2003; McMahon et al., 2010). By extension, the tein substrates utilized during synthesis (e.g.,
essential AA δ13C fingerprints of consumers at carbohydrates vs lipids), and the degree of
any trophic level could be used to evaluate the direct routing of nonessential AAs sourced
relative importance of different sources of pri- from dietary protein (Howland et al., 2003;
mary production at the base of the food chain McMahon et al., 2010; McMahon, Polito, et al.,
(Arthur, Kelez, Larsen, Choy, & Popp, 2014; 2015; Newsome et al., 2011). Empirical studies
McMahon, McCarthy, Sherwood, Larsen, & have demonstrated that the relative degree of
Guilderson, 2015; McMahon, Thorrold, de novo synthesis versus direct routing of non-
Houghton & Berumen, 2016). As discussed in essential AAs is primarily driven by dietary
Section 7.4.2, when these distinct primary protein content (Newsome et al., 2011;
producers occupy spatial gradients or discrete Newsome, Wolf, Peters, & Fogel, 2014; O’Brien
habitats, the AA C isotope fingerprinting et al., 2002). Understanding the patterns in
approach can be used to identify foraging direction and magnitude of nonessential AA
across space and time. carbon isotope fractionation, as well as the
underlying drivers, is an area of active
7.2.2.2 Nonessential Amino Acids research in ecogeochemistry.
In contrast to essential AAs, animals can
acquire nonessential AAs in several ways: (1)
synthesize them de novo from a bulk carbon
pool of protein and nonprotein dietary macro-
7.2.3 Hydrogen Metabolism
molecules (Newsome et al., 2011; O’Brien Only one study has reported δ2H values of
et al., 2002; Wu et al., 2014), (2) route them individual AAs (Fogel, Griffin, & Newsome,
directly or indirectly (via the gut microbiome) 2016). However, given the B200m δ-range in
from dietary protein (McMahon et al., 2010), global δ2H isoscapes (Bowen, 2010; McMahon

et al., 2013a), AA δ2H analysis, just like hydro- used to trace animal movement. But for her-
gen isotope analysis of bulk tissues, may be bivorous or frugivorous species that con-
the most promising system for studying terres- sume relatively low protein diets, AA δ2H
trial animal movement and migration patterns. may be a more faithful recorder of local envi-
While food is the only source of carbon and ronmental water, and thus may be a strong proxy
nitrogen available for proteinaceous tissue syn- for tracking latitudinal or altitudinal movements.
thesis in animals, hydrogen from both food
and water is used to construct tissues
(Hobson, 1999; Wolf, Newsome, Fogel, & 7.3 ACCOUNTING FOR
Martinez del Rio, 2013). The experiment CONSUMER PHYSIOLOGY
reported in Fogel et al. (2016) focused on bac-
teria (Escherichia coli) grown in water of vary- Animals that migrate use a diverse array of
ing δ2H values and on medium that did or life history strategies to maximize fitness when
did not contain protein. Interestingly, the engaging in long-distance movements and
essential and nonessential classification extended residence in different ecosystems:
scheme appears to work well for AA δ2H, aquatic versus terrestrial, freshwater versus
likely because most of the hydrogen in AAs is marine, and tropics versus high latitudes
bonded to carbon and does not exchange with (Johnson & Gaines, 1990). These migration
body water. This experiment showed that strategies are associated with major shifts in
most ( . 80%) of the hydrogen in E. coli cells organismal physiology (Gwinner, 2012). For
was routed directly from a protein-rich sub- example, animal migration is often accompa-
strate (tryptone). The only exception to this nied by major shifts in energy output associ-
pattern was in the glycolytic nonessential AA ated with the physical act of migrating,
alanine, which had B40%50% of its hydro- alterations to metabolic rate associated with
gen derived from environmental water (i.e., environmental temperature shifts and/or
cellular water) even when exogenous protein physical activity, and changes in body condi-
was available for cellular synthesis. tion associated with fasting and/or shifts to
By contrast, a relatively high proportion of endogenous energy reserves (e.g., body fat or
hydrogen (B35%75%) in both nonessential muscle). These changes are often concurrent
and essential AAs was sourced from environ- with additional changes in physiology associ-
mental water when E. coli were grown on a ated with the drivers of migration, such as
glucose medium containing no protein. Amino shifts in hormones and body condition associ-
acid δ2H data from controlled feeding experi- ated with reproduction.
ments on other organisms have yet to be There is a wealth of bulk tissue isotope data
reported, however, the biochemical pathways on how major physiological stressors associ-
(glycolysis and TCA cycle) used by E. coli to ated with migration, reproduction, and disease
synthesize nonessential AAs are like those impact consumer C isotopic discrimination
used by heterotrophic eukaryotes (Kanehisa, (e.g., Gannes, Martinez del Rio, & Koch, 1998),
Furumichi, Tanabe, Sato, & Morishima, 2017). but comparatively little information for indi-
Overall, these patterns suggest that the major- vidual AAs. For example, McMahon, Polito,
ity (B70%80%) of hydrogen in the proteina- et al. (2015) found that the Δ13CCD of nones-
ceous tissues of omnivorous and carnivorous sential AAs in captive penguins varied sig-
animals is obtained from diet, which may nificantly depending on the duration of fasting
obscure the relationships between tissue δ2H prior to feather molt. The strong 13C-depletion
and that of precipitation that are commonly of AAs associated with the TCA cycle

7.3 ACCOUNTING FOR CONSUMER PHYSIOLOGY 179
(Glu, Asp, and Pro) appeared to be related to respectively (e.g., Newsome et al., 2010;
increased reliance on endogenous lipid stores, Newsome, Wolf, Bradley, & Fogel, 2017).
which are catabolized and readily converted to Conversion of one reservoir into another
TCA precursors during nutritional stress from should be traceable at the molecular level,
molting and migration. Similar patterns have especially for compounds (e.g., nonessential
been observed in captive mice fed lipid-rich AAs) that consumers can synthesize de novo
diets (Newsome et al., 2014). These variable while minimally fractionating essential AAs
AA isotopic discrimination patterns associated provide concurrent information about the
with changes in physiology, disease, and sources of primary production at the base of
resource utilization make it challenging to the food web supporting those consumers
accurately interpret migration and residence (e.g., McMahon, Polito, et al., 2015). Note that
patterns of mobile consumers. There is a criti- isotopic effects of such changes in movement,
cal need for more controlled feeding experi- foraging, and physiology will only be observ-
ments in the lab and comparative field studies able in tissues with isotopic incorporation
examining the impacts of these major changes rates that can record shifts on the same time
in physiology on the isotopic discrimination of scales as the changes in resource allocation
individual AAs, which can create additional and use.
mismatches between consumer isotope values Another topic linked to exploring migra-
and the underlying isoscape. tion and physiology with CSIA-AA that
With proper calibration, these AA isotopic represents both a current complication and a
discrimination patterns that currently pose potential opportunity is the role that the gut
challenges to tracking animal migration could microbiome plays in consumer protein metab-
instead provide powerful opportunities to olism. Gut microflora likely play an important
study organismal physiology of migrating ani- role in the protein metabolism of many ani-
mals. Animals that roam over vast distances mals, particularly herbivorous and omnivo-
are inherently difficult to capture and sample, rous hosts consuming diets deficient in
often making it impossible to reliably capture dietary protein quantity or quality. In such
the same individual multiple times to provide cases, the gut microbiome often supplements
a longitudinal ecological or ecophysiological the essential AA budget of its host organism
record. CSIA-AA could enable tracing of car- by converting nonprotein dietary macromole-
bon as it is shuttled among tissue types, help- cules, such as carbohydrates or even lipids,
ing to illuminate temporal variation in the into the carbon skeletons needed for de novo
mobilization of different reservoirs of endoge- synthesis of essential AAs. Only a few studies
nous versus exogenous resources. Empirical focused on a small suite of essential AAs (Lys
data collected from a wide variety of organ- and Thr) have examined this topic in humans
isms including bacteria, fungi, plants, and ani- (Metges, 2000), rats (Torrallardona, Harris,
mals show that endogenous lipids and protein, Coates, & Fuller, 1996), fish (Newsome et al.,
which are the two largest endogenous reser- 2011), and insects (Ayayee et al., 2015) in a
voirs that animals use during times of nutri- laboratory setting. This topic merits continued
tional stress, have very different δ13C and δ2H study as the gut microbiome adds an impor-
values (DeNiro & Epstein, 1977; Estep & tant filter between consumers and the
Hoering, 1980; Hayes, 2001). Bulk adipose baseline isoscape, as well as a potential
tissue, or lipids extracted from proteinaceous opportunity to explore patterns of resource
tissues, have δ13C and δ2H values 6m8m and utilization, metabolism, and physiology in
B100m150m lower than tissue proteins, migratory species.

7.4 CASE STUDIES IN MOVEMENT baseline isoscape signal as it moves up the

AND FORAGING ECOLOGY USING food chain imparts additional isotopic discrim-
CSIA-AA ination and uncertainty that can obscure the
links between consumer isotope values and
Perhaps the most striking advantage of the baseline isoscape. CSIA-AA has the poten-
CSIA-AA is the ability to isolate information tial to expand and refine the use of isotopes
about both the baseline isoscape and food web for studying animal movement by constraining
structure from a single sample. In this section, (1) the development of baseline isoscapes (this
we explore case studies under three important section) and (2) the offsets between baseline
themes to the development and application of isoscapes and consumer stable isotope values
CSIA-AA in movement and foraging ecology: (Section 7.4.2).
(1) using CSIA-AA to refine isoscapes, (2) Generating isoscapes using minimally iso-
disentangling movement and trophic dynamics tope fractionating AAs of primary or second-
in space and time, and (3) examining move- ary consumers provides a temporal integration
ment and resource utilization across different of the baseline signal without the confounding
ecosystem types. issue of significant trophic discrimination. For
example, Vokhshoori and McCarthy (2014)
used δ15N values of source AAs from sessile,
filter-feeding mussel tissue (Mytilus california-
7.4.1 Developing CSIA-Based Isoscapes nus) to generate CSIA-AA isoscapes across 10
Isoscapes form a valuable framework for latitude in the dynamic coastal zone of the
tracking the movement of animals in space California Current System. They found strong
and time (Graham et al., 2010; Hobson, 1999; latitudinal gradients in mussel Phe δ15N
McMahon, Hamady, & Thorrold, 2013b; values, reflecting the mixing of northern 15N-
Wunder & Norris, 2008). Primary producers depleted water brought south by surface flow
integrate the physical, chemical, and biological of the California Current and southern 15N-
cycling of elements in the environment and enriched water brought north via the
form a link to upper trophic level consumers. California Undercurrent from the denitrifica-
The tissues (e.g., leaves) of most primary pro- tion zone of the Eastern Tropical Pacific. In
ducers in terrestrial environments integrate systems that are spatially and temporally con-
environmental information over long time strained, the empirical generation of AA-based
periods and thus may be suitable for the con- isoscapes may provide a robust geospatial
struction of isoscapes at the primary producer framework for examining animal movement.
level. In contrast, the typically short life spans The primary reason why CSIA-AA has not
and fast isotopic incorporation rates of primary been routinely used to construct isoscapes is
producers in marine environments (e.g., phyto- analytical. The requirement of significant spe-
plankton) relative to upper trophic level consu- cialized isotope mass spectrometer instrumenta-
mers can create a mismatch between the tion, time and expense of analyses, and analyst
highly dynamic baseline isoscape and con- expertise required for proper interpretation of
sumer isotope values. To circumvent this issue, CSIA-AA data mean it is not feasible to analyze
marine isoscapes are often empirically con- hundreds or thousands of samples to generate
structed using data from primary consumers large-scale isoscapes from AA isotope data. In
with longer integration times that more closely all likelihood, the true value of CSIA-AA to the
match those of the target organism. The bio- development of isoscapes will lie in the ability
chemical and physiological processing of that to help constrain the underlying drivers of

7.4 CASE STUDIES IN MOVEMENT AND FORAGING ECOLOGY USING CSIA-AA 181
observed geospatial structure in isoscapes that often (1) use source AAs to examine local bio-
are constructed using bulk tissue data. For geochemical cycling without the confounding
example, MacKenzie, Longmore, Preece, Lucas, issue of consumer trophic discrimination or (2)
and Trueman (2014) used lion’s mane jellyfish compare source and trophic AAs to calculate
(Cyanea capillata) to construct a robust pelagic consumer trophic level while controlling for
isoscape for the North Sea. However, jellyfish variation in baseline isotope values.
are opportunistic feeders, preying on a wide The logical extension of these CSIA-AA
range of pelagic organisms. To address this approaches is to use the isotope values of min-
potential trophic variability issue, the authors imally isotope fractionating AAs in consumers
compared their isoscapes to similar isoscapes as proxies for baseline isotope values to track
generated from benthic, sessile scallops in the movement across known geochemical gradi-
North Sea from Jennings and Warr (2003). They ents, while controlling for potential filters
concluded that the remarkably similar gradients between baseline and consumer imposed by
in isotope value between these two consumers, trophic discrimination. One of the first studies
despite their differences in diet (trophic level) to use CSIA-AA to infer animal movement
and habitat, suggested that the observed North was by Popp et al. (2007), who found strong
Sea isoscape was a function of regional hydrog- latitudinal gradients in the source AA δ15N
raphy and biogeochemical cycling rather than values of Pacific yellowfin tuna (Thunnus alba-
trophic dynamics. A further test of this hypoth- cares) in the eastern tropical Pacific that mir-
esis could apply CSIA-AA of select samples rored patterns in the local baseline isotope
along the major isotopic gradients. Thus bulk values of particulate organic matter. They con-
SIA facilitates the sample sizes necessary to cluded that these tuna were not undertaking
generate robust isoscapes in space and time, significant ocean basin scale migrations, at
while CSIA-AA provides a tool to further vali- least on the monthly time scale of tissue turn-
date the structure of the observed isoscapes over (Bradley, Madigan, Block, & Popp, 2014).
through insights into the underlying mechan- Dale et al. (2011) took this approach one step
isms generating such geospatial patterns. further, using CSIA-AA to identify ontogenetic
shifts in nursery habitat use of brown stingray
(Dasyatis lata) across an isoscape in coastal
Hawai’i. They were able to resolve shifts in
7.4.2 Disentangling Movement and
both movement (migration from nursery habi-
Foraging Ecology tats within the bay to adult habitats offshore)
Movement and foraging ecology are funda- and foraging (diet shifts with concurrent
mentally linked. Not only is foraging ecology a increases in trophic level) of this benthic pred-
major driver of animal migrations (Fryxell & ator across a complex tropical isoscape.
Sinclair, 1988), but food is often a primary vec- A number of other studies have employed
tor for transferring the biogeochemical signals CSIA-AA to examine movement associated
of baseline δ13C, δ15N, and δ2H isoscapes into with foraging (e.g., Lorrain et al., 2009; Ruiz-
consumer tissues (Graham et al., 2010; Wunder Cooley, Koch, Fiedler, & McCarthy, 2014),
& Norris, 2008). Perhaps the most common breeding (e.g., Seminoff et al., 2012; Vander
applications of CSIA-AA in ecogeochemistry Zanden et al., 2013), and range expansion (e.g.,
have been to tease apart the relative influence Ruiz-Cooley, Ballance, & McCarthy, 2013)
of baseline and trophic isotope variability on across a variety of spatial and temporal scales.
consumer isotope values (see Fig. 1 of Several of these studies report substantial
McMahon & McCarthy, 2016). These studies underestimates of top consumer trophic level

based on CSIA-AA in both field and laboratory To date, most studies using CSIA-AA to
settings (e.g., Dale et al., 2011; Germain et al., examine migration have focused on large-scale
2013; Lorrain et al., 2009; Ruiz-Cooley et al., (.100s km) movements of organisms relative
2013, 2014), which likely reflects challenges to known geochemical gradients. The use of
with our current understanding of the under- isotopes to geolocate organisms to specific
lying mechanisms of trophic AA isotopic dis- habitats distributed across smaller scales
crimination (O’Connell, 2017). Characterizing (,10 km) is a potentially powerful but chal-
the sources and magnitudes of these AA tro- lenging goal. McMahon, Berumen, and
phic discrimination factors is an area of active Thorrold (2012) used an essential AA δ13C fin-
research (e.g., Chikaraishi et al., 2015; Germain gerprinting approach to assess ontogenetic
et al., 2013; McMahon, Polito, et al., 2015; migration patterns of Ehrenberg snapper
McMahon, Thorrold, et al., 2015; Steffan et al., (Lutjanus ehrenbergii) among discrete juvenile
2013), as are new approaches to incorporating nursery habitats on the scale of 10s km. To do
this variability into equations that estimate tro- so, they characterized spatially separated habi-
phic level (e.g., McMahon, Polito, et al., 2015). tat signatures using resident fishes (reflecting
As our understanding of AA isotopic discrimi- local food web structure; McMahon, Berumen,
nation improves, so will our ability to calculate Mateo, Elsdon, & Thorrold, 2011) and then
more accurate and precise trophic levels using compared those signatures to the core of adult
CSIA-AA. fish otoliths (representing the period of time
δ2H analysis of individual AAs could pro- when those fish were juveniles; McMahon,
vide better spatial resolution than bulk tissue Fogel, Johnson, Houghton, & Thorrold, 2011).
approaches to assigning origins in animals that In doing so, they were able to identify impor-
undertake continental-scale migrations. tant nursery habitats and migration corridors
Specifically, there are two potentially fruitful for an ecologically and economically important
approaches that could be used to construct coral reef fish in the Red Sea. In situations like
compound-specific hydrogen isoscapes. First, this with spatially and isotopically distinct
nonessential AAs like Ala appear to be synthe- habitats, CSIA-AA can provide a powerful tool
sized from a high proportion (40%50%) of for identifying habitat use and residence pat-
water-derived hydrogen. Thus δ2H analysis of terns in mobile consumers.
Ala may provide a more direct link between As with all indirect tools for tracking animal
the hydrogen isotopic composition of local pre- movement, we need to validate the patterns of
cipitation and that of tissues, of which only movement determined by CSIA-AA with
20%30% is sourced from body water. A known locations. For instance, Seminoff et al.
second approach would be to focus on the δ2H (2012) used satellite telemetry and source AA
of essential AAs, which are overwhelmingly δ15N analysis to show that patterns in bulk tis-
derived from food ($90%), and thus may sue isotope values of Pacific leatherback sea
more faithfully record regional or even land- turtles (Dermochelys coriacea) reflected diver-
scape level variation in the δ2H of resources gent turtle migratory strategies to distinct for-
than bulk tissues that are synthesized from a aging groups rather than spatial variation in
combination of hydrogen from water and trophic dynamics. Similarly, Polito et al. (2017)
food. However, AA δ2H compound-specific used archival geolocation tags to show that the
approaches need to be tested with feeding strong segregation of essential AA δ13C values
experiments in the laboratory followed by in Antarctic penguin tail feathers was indica-
analysis of tissues from known origin indivi- tive of distinct migration strategies from the
duals (e.g., Langin et al., 2007). Antarctic Peninsula east into the ice-covered

7.4 CASE STUDIES IN MOVEMENT AND FORAGING ECOLOGY USING CSIA-AA 183
Weddell Sea versus west into the ice-free Scotia trophic AA Glu. Assessing trophic dynamics
Sea. There is an obvious need to expand upon of consumers that move and forage between
these efforts across a variety of stable isotopes, these distinct ecosystem types (e.g., with dif-
ecosystems, and spatiotemporal scales. While ferent primary producer β values) will require
no isotopic approach will ever provide the geo- new trophic position models that take this var-
location resolution of satellite telemetry, iso- iability into account. For instance, to examine
topes have the advantage of recording dietary the trophic positions of prehistoric human
inputs over a variety of timescales depending populations in Rapa Nui, South Pacific,
on tissue type. Thus coupling isotopes with Jarman et al. (2017) used an amino acid δ13C
telemetry could potentially discriminate fingerprinting approach to identify the relative
between movement through and actual use of contribution of marine and terrestrial primary
resources in particular habitats. production supporting humans, and then used
an amino acid nitrogen isotope mass balance
model with a β mixed value that multiplies the
estimated fraction of marine and terrestrial
7.4.3 Constraining Movement Across
foods with their corresponding known marine
Ecosystem or terrestrial β values. This analytically inten-
Most studies utilizing bulk and CSIA-AA sive, but innovative approach, likely reflects
isotope methods to characterize animal move- the next generation in trophic dynamics mod-
ment have focused on movement within a sin- els using CSIA-AA. But the success of this
gle ecosystem type (e.g., an ocean basin or a approach necessitates a better understanding
terrestrial grassland). Far fewer studies have of the parameters (e.g., B) that go into the
used CSIA-AA to rigorously examine move- model. Furthermore, there have been very few
ment across ecosystems types (e.g., between controlled feeding studies to examine trophic
marine and terrestrial systems, or from C3 for- discrimination of individual AAs among terres-
ests to C4 grasslands). Some early work in this trial consumers, particularly vertebrates and
area assessed use of marine versus terrestrial upper trophic level consumers (but see
resources by coastal consumers (including Nakashita et al., 2011; Steffan et al., 2015).
humans) (e.g., Hare, Fogel, Stafford, Mitchell, Uric-acid producing insects, for instance,
& Hoering, 1991; Honch, McCullagh, & appear to have very different isotopic discrimi-
Hedges, 2012; Naito, Honch, Chikaraishi, nation patterns of Glu and Phe than many
Ohkouchi, & Yoneda, 2010). However, inter- urea and uric-acid producing aquatic consu-
pretation of source or essential AA isotope mers (see discussion in McMahon &
data can be challenging when adjacent ecosys- McCarthy, 2016). In agreement with other
tems have different primary producer sources recent comments (O’Connell & Collins, 2017),
(Jarman et al., 2017). In comparison to marine we caution that more data are needed on pri-
producers, very little is known about δ15N mary producer β values and AA isotopic dis-
discrimination patterns among source AAs in crimination factors between diet and consumer
terrestrial plants (e.g., Chikaraishi et al., 2009). from a wide range of natural marine and ter-
Preliminary data suggest that C3 and C4 terres- restrial ecosystems before we can confidently
trial plants have distinct β values (difference use AA-based equations to quantify trophic
between trophic and source AAs in primary level, especially for consumers that forage
producers), and these differences appear to be across multiple primary producer types (e.g.,
larger in magnitude than observed trophic dis- marine algae, terrestrial C3 plants, and terres-
crimination of the most heavily fractionating trial C4 plants). Similar datasets are needed to

further refine the capabilities and limitations carboxyl and amino terminus of sequential
of using essential AA δ13C fingerprints to AAs in a polypeptide chain. AA purification
address questions in both aquatic and terres- with a cation exchange resin (e.g., Dowex) is
trial ecology. often necessary when dealing with heteroge-
neous materials that contain a mixture of pro-
tein, lipids, and carbohydrates (e.g., blood
7.5 CSIA-AA METHODOLOGY plasma, plants, algae, and biofilms) or when
working with biomineral tissues that have
Successful application of CSIA-AA to ques- interfering compounds (e.g., otoliths)
tions in movement and foraging ecology (Amelung & Zhang, 2001). However, signifi-
requires significant analytical capital and cant isotopic fractionation can occur with some
biochemical expertise to properly analyze, types of column resins or procedures (e.g.,
interpret, and evaluate molecular isotope data. Hare et al., 1991), so consistent analytical pro-
Most notably, the isotope analysis of individ- cedures with careful evaluation of isotopic
ual compounds requires significantly more fractionation should always be employed.
time for sample preparation and analysis com- The vast majority of papers reporting AA
pared to conventional analysis of bulk tissues isotope data use GC-C-IRMS, which requires
via elemental analyzer-isotope ratio mass spec- the derivatization of individual AAs to reduce
trometry (EA-IRMS). A telling proxy for the AA polarity and increase AA volatility prior to
intense analytical nature of CSIA-AA is the separation and isotopic analysis (Silfer, Engel,
average sample size in published datasets, Macko, & Jumeau, 1991). Derivatization neutra-
often in the dozens rather than the hundreds lizes polar carboxyl (COOH), amino (NH2),
or thousands. While CSIA-AA is a potentially and hydroxyl (OH) groups in AAs by repla-
powerful methodology in animal ecology and cing active hydrogen atoms with nonpolar moi-
ecophysiology, it is currently best suited for a eties. The three most common derivatization
well-defined question that can be answered reagents used in ecological and geochemical
with a relatively small dataset. studies are: trifluoroacyl-isopropyl ester (TFA/
Contemporary analytical methods for deter- AA/iPr, e.g., Silfer et al., 1991), pivaloyl-
mining δ13C, δ15N, or δ2H values of individual isopropyl ester (Pv/AA/iPr, e.g., Chikaraishi,
AAs consists of two phases: a wet chemistry Kashiyama, Ogawa, Kitazato, & Ohkouchi,
phase to purify AAs and an analytical phase to 2007), and methoxycarbonyl AA ester (MOC/
separate the AAs and measure their AA, e.g., Walsh, He, & Yarnes, 2014). As is the
stable isotope values. In the wet chemistry case with all wet chemistry, great care should
phase, samples are typically lipid extracted, be taken to choose the appropriate derivatiza-
acid hydrolyzed, purified, and then derivation protocol for the application at hand. Each
tized (if analyzing via gas chromatograph- of these procedures has trade-offs, e.g., in the
combustion (GC-C-)-IRMS, see below). Lipid amount of time required to derivatize, the sta-
extraction with polar solvents (e.g., chloro- bility of the resulting derivatives, the safety
form:methanol or petroleum ether) is neces- risks of exposure (both to humans and instru-
sary when working with samples that contain ments), and the ability to maximize the number
.2% lipids, as excess lipids can interfere with of AAs measured (Ohkouchi et al., 2017). A
the derivatization process and degrade chro- smaller number of studies have successfully
matography. Hydrolysis with strong (B6 N) used high-pressure liquid chromatography-
hydrochloric acid liberates individual AAs by (HPLC)-C-IRMS, which does not require deriv-
breaking the peptide bonds that bind the atization, to analyze the stable isotope values of

7.6 SUMMARY AND FUTURE WORK 185
individual AAs (Broek & McCarthy, 2014; isotope fractionation of the reaction (for sam-
McCullagh, Juchelka, & Hedges, 2006). ple equations see Fogel et al., 2016; O’Brien
However, GC-C-IRMS continues to be the pre- et al., 2002). Corrections accounting for the
ferred analytical approach for CSIA-AA. proportion of extrinsic versus intrinsic carbon
While the wet chemistry phase is time and hydrogen vary depending on individual
intensive, the major bottleneck in sample AA structure. No extrinsic nitrogen is added
throughput is instrument analytical time. during derivatization so post hoc corrections
Methods vary, but separation of AAs via GC of δ15N data are similar to those used for
typically uses a 5060 m nonpolar GC column correcting bulk tissue isotope data produced
and a temperature ramping procedure lasting by EA-IRMS.
4075 minutes for an individual sample
injection. Samples are typically analyzed in
triplicate and bracketed by an AA reference 7.6 SUMMARY AND FUTURE
material that is derivatized along with the tar- WORK
get samples. Reference materials are mixtures
of commercially available AAs (e.g., Sigma CSIA-AA has great potential to enhance the
Aldrich) with known isotope values indepen- study of animal migration, habitat use, and
dently measured via EA-IRMS prior to deriva- foraging ecology. However, like all indirect
tization. Under ideal analytical conditions tracer-based proxies, there are a number of
when instrumentation is performing well, this assumptions and inherent limitations that
run sequence will yield AA isotope data for must be considered when interpreting AA iso-
B34 unique samples (run in triplicate with tope data. The success of any isotope-based
standards) in a B24-hour time period. But application to track resource utilization, be it
note that GC-C-IRMSs are complex instru- diet or habitat use, is dictated by our under-
ments that require constant attention and standing of two fundamental principles of
expert knowledge to operate. Generating and how the isotopic composition of consumer diet
maintaining good chromatography on any GC and tissues are related: (1) isotopic discrimina-
system often transcends the boundary between tion factors that control the offset between con-
science and art and takes time to learn. sumer and baseline isoscapes and (2) tissue
High-quality peak separation in the chroma- incorporation rates that influence the time
tography is critical for generating high-quality scale of integration.
AA isotope data. Peak coelution prevents accu- First, we must consider the accuracy of the
rate integration and isotopic measurement. AA discrimination factors used to link the iso-
Ensuring that the trace returns as close as topic composition of the consumer to the
possible to baseline between peaks requires underlying baseline isoscape. The analysis of
the purification steps discussed earlier, proper source AAs (δ15N) and essential AAs (δ13C and
selection of column type, modifications to the δ2H) offers a major advantage on this front.
temperature ramps during analysis, and However, even these AAs with generally mini-
adjustments to peak and baseline integration mal isotopic discrimination exhibit some
on the chromatograph postanalysis. Once degree of change in isotope value during tro-
optimal chromatography is achieved, deriva- phic transfer, which when propagated through
tized AA isotope data must be corrected post several trophic levels, can impart a significant
hoc for the extrinsic carbon or hydrogen shift in consumer AA isotope values relative to
added to the carboxyl and amino terminus of the baseline isoscape. There is an obvious
each AA as well as the associated kinetic need for more data on the mechanisms that

influence carbon, nitrogen, and hydrogen iso- accurate consumer trophic positions using
topic discrimination in individual AAs. This CSIA-AA will require choosing trophic and
requires controlled laboratory experiments sources AAs with comparable turnover rates.
cross-validated with natural experiments However, with proper calibration, variability
across an increasingly wide range of consumer in individual AA incorporation rates may also
taxa, consumerresource relationships (e.g., provide a framework to create an isotopic
diet composition and quality, and gut micro- clock, as has been done previously by compar-
biota contribution), and consumer physiologi- ing bulk isotope values of multiple tissues
cal conditions (e.g., fasting and disease). with different incorporation rates, to determine
Recent developments in position-specific the length of time an individual has spent on a
stable isotope analysis may open new doors to new resource (e.g., habitat or food) (Phillips &
mechanistically study biochemical pathways Eldridge, 2006).
along the synthesis and transformation of com- Finally, the accuracy of geolocation assign-
pounds (Gauchotte-Lindsay & Turnbull, 2016). ments using CSIA-AA or any isotope approach
Our improved understanding of isotopic dis- is only as good as the accuracy of the underly-
crimination factors will also necessitate better ing isoscapes. Most current regional or conti-
universal practices for accounting for uncer- nental scale isoscapes used to track large-scale
tainty in AA isotopic measurements and migrations are hampered by the undersampling
associated parameters (e.g., β and Δ in trophic and uneven distribution of data across space
level calculations) used in CSIA-AA applica- and time (e.g., Jennings & Warr, 2003;
tions (e.g., Ohkouchi et al., 2017). McMahon et al., 2013b; Schell, Barnett, &
Second, the accuracy of an isotope approach Vinette, 1998), as well as the paucity of calibra-
to tracking animal movement and habitat use tion samples of known origin (e.g., Langin
is dependent upon our understanding of the et al., 2007). With advances in our understand-
integration time of the sampled tissue relative ing of isotope systematics (e.g., Hayes, 2001;
to the organism’s residence time in a habitat. Macko, Estep, Engel, & Hare, 1986), biogeo-
The rate of isotope incorporation into the AAs chemical and ecosystem models (e.g., Magozzi,
of a consumer can vary significantly as a func- Yool, Vander Zanden, Wunder, & Trueman,
tion of the degree of transamination/deamina- 2017), and statistical approaches (e.g., Wunder
tion (nitrogen) or synthesis (carbon and & Norris, 2008), significant opportunities exist
hydrogen), the kinetic isotope effects associ- to expand and refine the development of spa-
ated with AA transport, and the biochemical tially and temporally robust isoscapes through
reactions controlling metabolic breakdown the development of more mechanistic and com-
(Chikaraishi et al., 2007; McCarthy et al., 2013). prehensive isoscape models (Trueman,
While isotope turnover rates of individual MacKenzie, & Palmer, 2012). When coupled
AAs is still relatively unknown, several studies with the improved understanding of individual
have found that half-lives of individual AAs in AA isotopic discrimination and turnover rates,
Pacific bluefin tuna (Thunnus orientalis) the CSIA-AA approach holds great promise for
(Bradley et al., 2014) and Pacific white shrimp improving our understanding of animal move-
(Litopenaeus vannamei) (Downs et al., 2014) var- ment, habitat use, and foraging ecology.
ied from weeks (e.g., Pro) to .1 year (e.g.,
Ser). Understanding AA isotope turnover rates
is critical to constraining the temporal and spa- Acknowledgment
tial integrations of the local baseline isotope This chapter is dedicated to Dr. Marilyn L. Fogel. Through
signals into consumer tissues. Calculation of her pioneering work in compound-specific stable isotope

REFERENCES 187
analysis and her years of dedication to education and men- contents, bulk and amino acid stable isotopes. Marine
toring, her fingerprints can be seen throughout this chapter Ecology Progress Series, 433, 221236.
and the works cited within. DeNiro, M. J., & Epstein, S. (1977). Mechanism of carbon
isotope fractionation associated with lipid synthesis.
Science, 197(4300), 261263.
Dingle, H. (2014). Migration: The biology of life on the move.
References USA: Oxford University Press.
Amelung, W., & Zhang, X. (2001). Determination of amino Downs, E. E., Popp, B. N., & Holl, C. M. (2014). Nitrogen
acid enantiomers in soils. Soil Biology and Biochemistry, isotope fractionation and amino acid turnover rates in
33(4), 553562. the Pacific white shrimp Litopenaeus vannamei. Marine
Arthur, K. E., Kelez, S., Larsen, T., Choy, C. A., & Popp, Ecology Progress Series, 516, 239250.
B. N. (2014). Tracing the biosynthetic source of essential Estep, M. F., & Hoering, T. C. (1980). Biogeochemistry of
amino acids in marine turtles using δ13C fingerprints. the stable hydrogen isotopes. Geochimica et
Ecology, 95(5), 12851293. Cosmochimica Acta, 44(8), 11971206.
Ayayee, P. A., Jones, S. C., & Sabree, Z. L. (2015). Can 13C Fogel, M. L., Griffin, P. L., & Newsome, S. D. (2016).
stable isotope analysis uncover essential amino acid Hydrogen isotopes in individual amino acids reflect
provisioning by termite-associated gut microbes? PeerJ, differentiated pools of hydrogen from food and water
3, e1218. in Escherichia coli. Proceedings of the National Academy of
Bowen, G. J. (2010). Isoscapes: Spatial pattern in isotopic Sciences, 113(32), E4648E4653.
biogeochemistry. Annual Review of Earth and Planetary Fryxell, J. M., & Sinclair, A. R. E. (1988). Causes and conse-
Sciences, 38, 161187. quences of migration by large herbivores. Trends in
Bradley, C. J., Madigan, D. J., Block, B. A., & Popp, B. N. Ecology & Evolution, 3(9), 237241.
(2014). Amino acid isotope incorporation and enrich- Gannes, L. Z., Martinez del Rio, C., & Koch, P. (1998).
ment factors in Pacific bluefin tuna, Thunnus orientalis. Natural abundance variations in stable isotopes and
PLoS One, 9(1), e85818. their potential uses in animal physiological ecology.
Braun, A., Vikari, A., Windisch, W., & Auerswald, K. Comparative Biochemistry and Physiology Part A:
(2014). Transamination governs nitrogen isotope hetero- Molecular & Integrative Physiology, 119(3), 725737.
geneity of amino acids in rats. Journal of Agricultural and Gauchotte-Lindsay, C., & Turnbull, S. M. (2016). On-line
Food Chemistry, 62(32), 80088013. high-precision carbon position-specific stable isotope
Broek, T. A., & McCarthy, M. D. (2014). A new approach analysis: A review. TrAC Trends in Analytical Chemistry,
to δ15N compound-specific amino acid trophic position 76, 115125.
measurements: Preparative high pressure liquid chro- Germain, L. R., Koch, P. L., Harvey, J., & McCarthy, M. D.
matography technique for purifying underivatized (2013). Nitrogen isotope fractionation in amino acids
amino acids for stable isotope analysis. Limnology and from harbor seals: Implications for compound-specific
Oceanography: Methods, 12(12), 840852. trophic position calculations. Marine Ecology Progress
Chikaraishi, Y., Kashiyama, Y., Ogawa, N. O., Kitazato, H., Series, 482, 265277.
& Ohkouchi, N. (2007). Metabolic control of nitrogen Graham, B. S., Koch, P. L., Newsome, S. D., McMahon,
isotope composition of amino acids in macroalgae and K. W., & Aurioles, D. (2010). Using isoscapes to trace the
gastropods: Implications for aquatic food web studies. movements and foraging behavior of top predators in oceanic
Marine Ecology Progress Series, 342, 8590. ecosystems. Isoscapes (pp. 299318). Netherlands: Springer.
Chikaraishi, Y., Ogawa, N. O., Kashiyama, Y., Takano, Y., Gwinner, E. (Ed.), (2012). Bird migration: Physiology and eco-
Suga, H., Tomitani, A., . . . Ohkouchi, N. (2009). physiology. Berlin: Springer Science & Business Media.
Determination of aquatic food-web structure based on Hare, P. E., Fogel, M. L., Stafford, T. W., Mitchell, A. D., &
compound-specific nitrogen isotopic composition of Hoering, T. C. (1991). The isotopic composition of car-
amino acids. Limnology and Oceanography: Methods, 7 bon and nitrogen in individual amino acids isolated
(11), 740750. from modern and fossil proteins. Journal of
Chikaraishi, Y., Steffan, S. A., Takano, Y., & Ohkouchi, N. Archaeological Science, 18(3), 277292.
(2015). Diet quality influences isotopic discrimination Hayes, J. M. (2001). Fractionation of carbon and hydrogen
among amino acids in an aquatic vertebrate. Ecology isotopes in biosynthetic processes. Reviews in
and Evolution, 5(10), 20482059. Mineralogy and Geochemistry, 43(1), 225277.
Dale, J. J., Wallsgrove, N. J., Popp, B. N., & Holland, K. N. Hobson, K. A. (1999). Tracing origins and migration of
(2011). Nursery habitat use and foraging ecology of the wildlife using stable isotopes: A review. Oecologia, 120
brown stingray Dasyatis lata determined from stomach (3), 314326.

Hobson, K. A., & Wassenaar, L. I. (2008). Tracking animal using amino acid stable isotope fingerprinting. PLoS
migration with stable isotopes (Vol. 2). Academic Press. One, 8(9), e73441.
Honch, N. V., McCullagh, J. S., & Hedges, R. E. (2012). Lorrain, A., Graham, B., Ménard, F., Popp, B., Bouillon, S.,
Variation of bone collagen amino acid δ13C values in Van Breugel, P., & Cherel, Y. (2009). Nitrogen and car-
archaeological humans and fauna with different dietary bon isotope values of individual amino acids: A tool to
regimes: Developing frameworks of dietary discrimina- study foraging ecology of penguins in the Southern
tion. American Journal of Physical Anthropology, 148(4), Ocean. Marine Ecology Progress Series, 391, 293306.
495511. MacKenzie, K. M., Longmore, C., Preece, C., Lucas, C. H.,
Hoskin, S. O., Gavet, S., Milne, E., & Lobley, G. E. (2001). & Trueman, C. N. (2014). Testing the long-term stability
Does glutamine act as a substrate for transamination of marine isoscapes in shelf seas using jellyfish tissues.
reactions in the liver of fed and fasted sheep? British Biogeochemistry, 121(2), 441454.
Journal of Nutrition, 85(5), 591597. Macko, S. A., Estep, M. L. F., Engel, M. H., & Hare, P. E.
Howland, M. R., Corr, L. T., Young, S. M., Jones, V., Jim, (1986). Kinetic fractionation of stable nitrogen isotopes
S., Van Der Merwe, N. J., . . . Evershed, R. P. (2003). during amino acid transamination. Geochimica et
Expression of the dietary isotope signal in the com- Cosmochimica Acta, 50(10), 21432146.
pound-specific δ13C values of pig bone lipids and Magozzi, S., Yool, A., Vander Zanden, H. B., Wunder,
amino acids. International Journal of Osteoarchaeology, 13 M. B., & Trueman, C. N. (2017). Using ocean models to
(12), 5465. predict spatial and temporal variation in marine carbon
Hussey, N. E., Kessel, S. T., Aarestrup, K., Cooke, S. J., isotopes. Ecosphere, 8(5), e01763.
Cowley, P. D., Fisk, A. T., . . . Flemming, J. E. M. (2015). McCarthy, M. D., Lehman, J., & Kudela, R. (2013).
Aquatic animal telemetry: A panoramic window into Compound-specific amino acid δ15N patterns in marine
the underwater world. Science, 348(6240), 1255642. algae: Tracer potential for cyanobacterial vs. eukaryotic
Jarman, C. L., Larsen, T., Hunt, T., Lipo, C., Solsvik, R., organic nitrogen sources in the ocean. Geochimica et
Wallsgrove, N., . . . Popp, B. N. (2017). Diet of the prehis- Cosmochimica Acta, 103, 104120.
toric population of Rapa Nui (Easter Island, Chile) shows McClelland, J. W., & Montoya, J. P. (2002). Trophic rela-
environmental adaptation and resilience. American tionships and the nitrogen isotopic composition of
Journal of Physical Anthropology, 164(2), 343361. amino acids in plankton. Ecology, 83(8), 21732180.
Jennings, S., & Warr, K. J. (2003). Environmental correlates McCullagh, J. S., Juchelka, D., & Hedges, R. E. (2006).
of large-scale spatial variation in the δ15N of marine Analysis of amino acid 13C abundance from human and
animals. Marine Biology, 142(6), 11311140. faunal bone collagen using liquid chromatography/iso-
Johnson, M. L., & Gaines, M. S. (1990). Evolution of dis- tope ratio mass spectrometry. Rapid Communications in
persal: Theoretical models and empirical tests using Mass Spectrometry, 20(18), 27612768.
birds and mammals. Annual Review of Ecology and McMahon, K. W., Berumen, M. L., Mateo, I., Elsdon, T. S., &
Systematics, 21(1), 449480. Thorrold, S. R. (2011). Carbon isotopes in otolith amino
Kanehisa, M., Furumichi, M., Tanabe, M., Sato, Y., & acids identify residency of juvenile snapper (Family:
Morishima, K. (2017). KEGG: New perspectives on gen- Lutjanidae) in coastal nurseries. Coral Reefs, 30, 11351145.
omes, pathways, diseases and drugs. Nucleic Acids McMahon, K. W., Berumen, M. L., & Thorrold, S. R. (2012).
Research, 45(D1), D353D361. Linking habitat mosaics and connectivity in a coral reef
Kays, R., Crofoot, M. C., Jetz, W., & Wikelski, M. (2015). seascape. Proceedings of the National Academy of Sciences,
Terrestrial animal tracking as an eye on life and planet. 109(38), 1537215376.
Science, 348(6240), aaa2478. McMahon, K. W., Fogel, M. L., Elsdon, T. S., & Thorrold,
Langin, K. M., Reudink, M. W., Marra, P. P., Norris, D. R., S. R. (2010). Carbon isotope fractionation of amino acids
Kyser, T. K., & Ratcliffe, L. M. (2007). Hydrogen isoto- in fish muscle reflects biosynthesis and isotopic routing
pic variation in migratory bird tissues of known origin: from dietary protein. Journal of Animal Ecology, 79(5),
Implications for geographic assignment. Oecologia, 152 11321141.
(3), 449457. McMahon, K. W., Fogel, M. L., Johnson, B. J., Houghton,
Larsen, T., Taylor, D. L., Leigh, M. B., & O’Brien, D. M. L. A., & Thorrold, S. R. (2011). A new method to recon-
(2009). Stable isotope fingerprinting: A novel method struct fish diet and movement patterns from δ13C
for identifying plant, fungal, or bacterial origins of values in otolith amino acids. Canadian Journal of
amino acids. Ecology, 90(12), 35263535. Fisheries and Aquatic Sciences, 68, 13301340.
Larsen, T., Ventura, M., Andersen, N., O’Brien, D. M., McMahon, K. W., Hamady, L. L., & Thorrold, S. R. (2013a).
Piatkowski, U., & McCarthy, M. D. (2013). Tracing car- Ocean ecogeochemistry: A review. Oceanography and
bon sources through aquatic and terrestrial food webs Marine Biology: An Annual Review, 51, 327374.

REFERENCES 189
McMahon, K. W., Hamady, L. L., & Thorrold, S. R. (2013b). Newsome, S. D., Wolf, N., Peters, J., & Fogel, M. L. (2014).
A review of ecogeochemistry approaches to estimating Amino acid δ13C analysis shows flexibility in the rout-
movements of marine animals. Limnology and ing of dietary protein and lipids to the tissue of an
Oceanography, 58(2), 697714. omnivore. Integrative and Comparative Biology, 54(5),
McMahon, K. W., & McCarthy, M. D. (2016). Embracing 890902.
variability in amino acid δ15N fractionation: O’Brien, D. M., Fogel, M. L., & Boggs, C. L. (2002).
Mechanisms, implications, and applications for trophic Renewable and nonrenewable resources: Amino acid
ecology. Ecosphere, 7, e01511. turnover and allocation to reproduction in Lepidoptera.
McMahon, K. W., McCarthy, M. D., Sherwood, O. A., Proceedings of the National Academy of Sciences, 99(7),
Larsen, T., & Guilderson, T. P. (2015). Millennial-scale 44134418.
plankton regime shifts in the subtropical North Pacific O’Connell, T. C. (2017). ‘Trophic’ and ‘source’ amino acids
Ocean. Science, 350(6267), 15301533. in trophic estimation: A likely metabolic explanation.
McMahon, K. W., Polito, M. J., Abel, S., McCarthy, M. D., Oecologia, 184(2), 317326.
& Thorrold, S. R. (2015). Carbon and nitrogen isotope O’Connell, T. C., & Collins, M. J. (2017). Comment on
fractionation of amino acids in an avian marine preda- “Ecological niche of Neanderthals from Spy Cave
tor, the gentoo penguin (Pygoscelis papua). Ecology and revealed by nitrogen isotopes of individual amino acids
Evolution, 5(6), 12781290. in collagen”. Journal of Human Evolution. Available from
McMahon, K. W., Thorrold, S. R., Elsdon, T. S., & https://doi.org/10.1016/j.jhevol.2017.05.006.
McCarthy, M. D. (2015). Trophic discrimination of Ohkouchi, N., Chikaraishi, Y., Close, H. G., Fry, B., Larsen,
nitrogen stable isotopes in amino acids varies with diet T., Madigan, D. J., . . . Ogawa, N. O. (2017). Advances
quality in a marine fish. Limnology and Oceanography, 60 in the application of amino acid nitrogen isotopic anal-
(3), 10761087. ysis in ecological and biogeochemical studies. Organic
McMahon, K. W., Thorrold, S. R., Houghton, L. A., & Geochemistry, 113, 150174.
Berumen, M. L. (2016). Tracing carbon flow through Phillips, D. L., & Eldridge, P. M. (2006). Estimating the tim-
coral reef food webs using a compound-specific ing of diet shifts using stable isotopes. Oecologia, 147(2),
stable isotope approach. Oecologia, 180(3), 809821. 195203.
Metges, C. C. (2000). Contribution of microbial amino acids Polito, M. J., Hinke, J. T., Hart, T., Santos, M., Houghton,
to amino acid homeostasis of the host. The Journal of L. A., & Thorrold, S. R. (2017). Stable isotope analyses
Nutrition, 130(7), S1857S1864. of feather amino acids identify penguin migration strat-
Naito, Y. I., Honch, N. V., Chikaraishi, Y., Ohkouchi, N., & egies at ocean basin scales. Biology Letters, 13(8),
Yoneda, M. (2010). Quantitative evaluation of marine 20170241.
protein contribution in ancient diets based on nitrogen Popp, B. N., Graham, B. S., Olson, R. J., Hannides, C. C.,
isotope ratios of individual amino acids in bone colla- Lott, M. J., López-Ibarra, G. A., . . . Fry, B. (2007).
gen: An investigation at the Kitakogane Jomon site. Insight into the trophic ecology of yellowfin tuna,
American Journal of Physical Anthropology, 143(1), 3140. Thunnus albacares, from compound-specific nitrogen iso-
Nakashita, R., Suzuki, Y., Akamatsu, F., Naito, Y. I., Sato- tope analysis of proteinaceous amino acids. Terrestrial
Hashimoto, M., & Tsubota, T. (2011). Ecological Ecology, 1, 173190.
application of compound-specific stable nitrogen isotope Post, D. M. (2002). Using stable isotopes to estimate trophic
analysis of amino acids: A case study of captive and position: Models, methods, and assumptions. Ecology,
wild bears. Researches in Organic Geochemistry, 27, 7379. 83(3), 703718.
Newsome, S. D., Bentall, G. B., Tinker, M. T., Oftedal, Ruiz-Cooley, R. I., Ballance, L. T., & McCarthy, M. D.
O. T., Ralls, K., Estes, J. A., & Fogel, M. L. (2010). (2013). Range expansion of the jumbo squid in the NE
Variation in δ13C and δ15N dietvibrissae trophic dis- Pacific: δ15N decrypts multiple origins, migration and
crimination factors in a wild population of California habitat use. PLoS One, 8(3), e59651.
sea otters. Ecological Applications, 20(6), 17441752. Ruiz-Cooley, R. I., Koch, P. L., Fiedler, P. C., & McCarthy,
Newsome, S. D., Fogel, M. L., Kelly, L., & Martinez del M. D. (2014). Carbon and nitrogen isotopes from top
Rio, C. (2011). Contributions of direct incorporation predator amino acids reveal rapidly shifting ocean bio-
from diet and microbial amino acids to protein synthe- chemistry in the outer California current. PLoS One, 9
sis in Nile tilapia. Functional Ecology, 25(5), 10511062. (10), e110355.
Newsome, S. D., Wolf, N., Bradley, C. J., & Fogel, M. L. Schell, D. M., Barnett, B. A., & Vinette, K. A. (1998).
(2017). Assimilation and isotopic discrimination of Carbon and nitrogen isotope ratios in zooplankton of
hydrogen in tilapia: Implications for studying animal the Bering, Chukchi and Beaufort seas. Marine Ecology
diet with δ2H. Ecosphere, 8(1), e01616. Progress Series, 162, 1123.

Schwarcz, H. P. (1991). Some theoretical aspects of isotope Vander Zanden, H. B., Arthur, K. E., Bolten, A. B., Popp,
paleodiet studies. Journal of Archaeological Science, 18(3), B. N., Lagueux, C. J., Harrison, E., . . . Bjorndal, K. A.
261275. (2013). Trophic ecology of a green turtle breeding popu-
Scott, J. H., O’Brien, D. M., Emerson, D., Sun, H., lation. Marine Ecology Progress Series, 476, 237249.
McDonald, G. D., Salgado, A., & Fogel, M. L. (2006). Vokhshoori, N. L., & McCarthy, M. D. (2014). Compound-
An examination of the carbon isotope effects associated specific δ15N amino acid measurements in littoral mus-
with amino acid biosynthesis. Astrobiology, 6(6), sels in the California upwelling ecosystem: A new
867880. approach to generating baseline δ15N isoscapes for
Seminoff, J. A., Benson, S. R., Arthur, K. E., Eguchi, T., coastal ecosystems. PLoS One, 9(6), e98087.
Dutton, P. H., Tapilatu, R. F., & Popp, B. N. (2012). Walsh, R. G., He, S., & Yarnes, C. T. (2014). Compound-
Stable isotope tracking of endangered sea turtles: specific δ13C and δ15N analysis of amino acids: A rapid,
Validation with satellite telemetry and δ15N analysis of chloroformate-based method for ecological studies. Rapid
amino acids. PLoS One, 7(5), e37403. Communications in Mass Spectrometry, 28(1), 96108.
Silfer, J. A., Engel, M. H., Macko, S. A., & Jumeau, E. J. Wolf, N., Newsome, S. D., Fogel, M. L., & Martinez del
(1991). Stable carbon isotope analysis of amino acid Rio, C. (2013). The relationship between drinking water
enantiomers by conventional isotope ratio mass spec- and the hydrogen and oxygen stable isotope values of
trometry and combined gas chromatography/isotope tissues in Japanese Quail (Cortunix japonica). The Auk,
ratio mass spectrometry. Analytical Chemistry, 63(4), 130(2), 323330.
370374. Wolf, N., Newsome, S. D., Peters, J., & Fogel, M. L. (2015).
Steffan, S. A., Chikaraishi, Y., Currie, C. R., Horn, H., Variability in the routing of dietary proteins and lipids
Gaines-Day, H. R., Pauli, J. N., . . . Ohkouchi, N. to consumer tissues influences tissue-specific isotopic
(2015). Microbes are trophic analogs of animals. discrimination. Rapid Communications in Mass
Proceedings of the National Academy of Sciences, 112(49), Spectrometry, 29(15), 14481456.
1511915124. Wu, G. (2009). Amino acids: Metabolism, functions, and
Steffan, S. A., Chikaraishi, Y., Horton, D. R., Ohkouchi, N., nutrition. Amino Acids, 37(1), 117.
Singleton, M. E., Miliczky, E., . . . Jones, V. P. (2013). Wu, G., Bazer, F. W., Dai, Z., Li, D., Wang, J., & Wu, Z.
Trophic hierarchies illuminated via amino acid isotopic (2014). Amino acid nutrition in animals: Protein synthe-
analysis. PLoS One, 8(9), e76152. sis and beyond. Annual Review of Animal Biosciences, 2
Torrallardona, D., Harris, C. I., Coates, M. E., & Fuller, (1), 387417.
M. F. (1996). Microbial amino acid synthesis and utili- Wunder, M. B., & Norris, D. R. (2008). Analysis and design
zation in rats: Incorporation of 15N from 15NH4Cl into for isotope-based studies of migratory animals.
lysine in the tissues of germ-free and conventional rats. Terrestrial Ecology, 2, 107128.
British Journal of Nutrition, 76(5), 689700. Yamaguchi, Y. T., Chikaraishi, Y., Takano, Y., Ogawa,
Trueman, C. N., MacKenzie, K. M., & Palmer, M. R. (2012). N. O., Imachi, H., Yokoyama, Y., & Ohkouchi, N.
Identifying migrations in marine fishes through stable- (2017). Fractionation of nitrogen isotopes during amino
isotope analysis. Journal of Fish Biology, 81(2), 826847. acid metabolism in heterotrophic and chemolithoauto-
Vanderklift, M. A., & Ponsard, S. (2003). Sources of trophic microbes across Eukarya, Bacteria, and
variation in consumer-diet δ15N enrichment: A meta- Archaea: Effects of nitrogen sources and metabolic
analysis. Oecologia, 136(2), 169182. pathways. Organic Geochemistry, 111, 101112.

C H A P T E R
8
Design and Analysis for Isotope-Based
Studies of Migratory Animals
Michael B. Wunder1 and D. Ryan Norris2
1
University of Colorado Denver, Denver, CO, United States, 2University of Guelph,
Guelph, ON, United States
8.1 INTRODUCTION provenance, and this consideration is lacking

in the literature. Here we provide a step-by-
The use of stable isotopes to track migration step guide for researchers interested in design-
and other types of movements is widely ing isotope-provenance studies. Furthermore,
adopted around the world. Isotopes are partic- the practice of isotopic geographic assignments
ularly attractive for studies on movement has advanced since the first edition of this
because individuals only have to be captured book. Thus we updated the approaches and
once to estimate their origin from a previous perspectives, removed details on approaches
period of the annual cycle, and isotopes are that are (or should be) obsolete, and discuss
more cost-effective than external tracking issues we think are important for conducting
devices. However, like any technique, there rigorous research in this field.
are challenges associated with properly execut- We provide better and more accessible
ing an isotope-based study to estimate geo- information that can be used to design and
graphic origin. execute a study using stable isotopes to track
In the first edition of this book, we pub- movements of animals. We hope that this
lished a chapter that primarily focused on information will be useful for everyone, from
statistical and quantitative problems of geo- students and professors to wildlife managers.
graphic assignments for individuals of unknown We believe the information in this chapter is
origin (Wunder & Norris, 2008). Looking back, broadly transferable to anyone interested in
we did not provide a systematic treatment of the provenance of animals, including those
challenges associated with the “design” of in the fields archeology, anthropology, social
isotope-based studies for studies of migratory and physical geography, and forensic science.

192 8. DESIGN AND ANALYSIS FOR ISOTOPE-BASED STUDIES OF MIGRATORY ANIMALS
8.2 PLANNING YOUR STUDY downloadable, global H and O data and pro-
vides isoscape maps and GIS-friendly grids
Is this a section you should skip? We hope (www.iaea.org/water) on a monthly, annual,
not. Checking out some simple considerations or seasonal basis (Terzer et al., 2013). A grow-
prior to an isotope study to track migratory ing number academic-run sampling sites also
animals is tremendously helpful and will offer isoscape capabilities based GNIP data
avoid later frustration or failure. Below, we and other aggregated water and other isotope
describe key considerations that increase the data (www.isomap.org).
chances of successfully executing an isotope- Several scientists developed process-based
based provenance project. isoscapes for carbon (Magozzi, Yool, Vander
(1) Identify the likely region of assignment and Zanden, Wunder, & Trueman, 2017; Powell,
evaluate known isotopic gradients from this area. Yoo, & Still, 2012) and strontium isotopes
By likely region of assignment, we mean the (Bataille, Laffoon, & Bowen, 2012). There are
entire geographic area from where you think also a growing number of studies describing
that the migratory organism might originate. isotopic patterns and variations in tissues of
This may be the entire breeding or nonbreed- known-origin in species and various taxo-
ing range of a species or a smaller area within nomic groups (Chapter 4: Application
these ranges. Regardless, it is important to of Isotopic Methods to Tracking Animal
know what kind of stable isotopic variation Movements). These resources are helpful, even
you will be confronted with when assigning if your study subject is not that species but is
individuals of unknown origin. You will also closely related or shares similar ecology.
need to have some idea of the spatial resolu- Even with resources for evaluating expected
tion you feel comfortable with when making target isotopic variation, predicting what is an
isotopic assignments. Of course, your level of appropriate geographic variation for your
“comfort” depends on the research question(s). study is tricky, especially with little experience
For example, if the question requires assigning or prior data. It can only be evaluated in the
individuals with a high degree of confidence context of variation generated from non-
to an area of a 100200 km2, you will likely geographic processes. In other words, we
need fine scale isotopic gradients and almost consider that geographic isotopic variation is
certainly multiple isotopes (or potentially other the signal we want to detect, and biological
approaches, see below). On the other hand, if and measurement variation is the noise from
assigning individuals to broad continental which that signal must be recovered. One
regions adequately answers your research thing that helps tremendously is to have an
question, then coarser-scale gradients and a idea of the degree of isotope variation
single isotope may be sufficient. expected from both analytical and sampling
There are several key resources available to errors, especially for variances within an indi-
investigate geographic variation in isotopes vidual (i.e. repeated samples) and for var-
(termed isoscapes, see Chapter 3: Isoscapes for iances among tissues between known-origin
Terrestrial Migration Research for more individuals from the same location (Chapter 1:
details) and they continue to grow in number Animal Migration: A Context for Using New
and degree of refinement. For hydrogen and Techniques and Approaches, Chapter 2:
oxygen isotopes, the International Atomic Introduction to Conducting Stable Isotope
Energy Agency’s Global Network for Isotopes Measurements for Animal Migration Studies).
in Precipitation (GNIP) provides up-to-date, We discuss this in more detail below. Overall,

8.2 PLANNING YOUR STUDY 193
gather all the available information on sources analytical error. Clearly, differences of this size
of isotopic variation and evaluate whether an are in no way “significant”.
isotope-based study would be feasible or not; The two other major sources of isotopic
some ecosystems and problems are well suited variance occur within tissues of the same type
to the approach, but others are not. We don’t from a single individual and within tissues of
think this process needs to be a formal, quanti- the same type from different known-origin
tative analysis; a thorough qualitative assess- individuals at the same location (Chapter 4:
ment of all available sources of information Application of Isotopic Methods to Tracking
will be extremely valuable. Animal Movements). While we know much
(2) Evaluate sources of isotope variance: analyti- less about what factors influence these
cal error, variance among multiple samples of the sources of variance than we do about what
same tissue type from a single individual and vari- influences analytical error, there are a grow-
ance among single samples from multiple known- ing number of studies on the physiological
origin individuals at the same location. and ecological factors governing isotopic
When assigning individuals of unknown ori- variation within and among individuals.
gin, there are “nuisance” sources of isotope var- Between individuals, these may include age,
iation (although they could provide useful data sex, metabolic rate, water use efficiency, sub-
in other contexts); this is simply “noise” that tle differences in foraging ecology, and even
diminishes the geographic signal we want to reproductive state. The factors driving varia-
recover. The larger the noise, the more it tion in isotope values within tissues from a
detracts from your ability to assign individuals single individual is a more complex topic that
to areas with a high degree of confidence (we we will not address here but that has largely
also strongly advocate carrying forward these to do with metabolic routing and growth rates
sources of uncertainty when performing geo- (Chapter 4: Application of Isotopic Methods
graphic assignments, see Section 8.3). One of to Tracking Animal Movements).
these sources is analytical error, which is inher- The point is that these sources of isotopic
ent variance in measurement associated with variation should be considered collectively and
the analysis of tissue samples using an Isotope in conjunction with predicted spatial variation
Ratio Mass Spectrometer (see Chapter 2: when deciding on whether and which isotopes
Introduction to Conducting Stable Isotope will effectively help answer the research ques-
Measurements for Animal Migration Studies). tion. For example, you might want to use
In lighter isotopes, such as deuterium, the vari- hydrogen isotopes to assign individuals to one
ance associated with repeated measurements of of two areas that differ by B20m. However, if
a homogenized sample is typically ,2m, we had poor analytical error in deuterium
whereas the variance in heavier isotopes tends ( 6 4m) and knew that known-origin tissues
to be smaller (,1m). Nevertheless, the magni- from individuals sampled at the same location
tude of analytical error must be considered rela- vary between 5m and 15m, a H isotope model
tive to the variation observed over the isoscape alone may be insufficient to assign animals to
and the desired spatial resolution for assign- one of these two places with a high degree of
ment. Analytical error almost always comprises confidence.
the smallest component of variation, often by a (3) Determine, at least preliminarily, which
wide margin, however we frequently see stable isotopes to use. Once sources of variation
authors reporting “statistically significant” dif- and geographic variation in isotopes have
ferences in mean isotope values across locations been identified and evaluated, the next step is
that are as or smaller in magnitude than the to plan which isotopes you may want to

analyze to answer your research question(s). values of an existing environmental isoscape

The driving factors will be sources of isotopic need to be calibrated to values expected for
variation, as discussed above. Hence, there are tissues in the species of interest. This calibra-
significant advantages of using multiple iso- tion approach also requires known-origin
topes, particularly if they vary over space tissues, but not from across the full geo-
orthogonally (see Chapter 3: Isoscapes for graphic extent of the potential range (see
Terrestrial Migration Research). This spatial Section 8.3). There are an increasing number
complementarity between isotopes can increase of calibration estimates for a wide range of
confidence in assignments and should, there- species (Chapter 4: Application of Isotopic
fore, also be a prominent consideration in a Methods to Tracking Animal Movements),
decision on which isotopes to analyze. Such and before you start it will be important to
spatial complementarity could outweigh high collect these estimates and evaluate whether
within- and between-individual variance or the information can be applied to your organ-
analytical error associated with an isotope. ism. Calibration estimates typically come in
Another practical aspect to consider when the form of a line equation describing the
choosing which and how many isotopes to relationship between the isotope value in the
analyze is the analytical cost (see Chapter 2: animal tissue (y) and the value derived from
Introduction to Conducting Stable Isotope the isoscape (x), adjusted by an intercept (β 0)
Measurements for Animal Migration Studies). and a slope (β). If you must adopt a calibra-
(4) Determine whether there is any calibration tion estimate derived from a different species,
information available for your species or a similar we recommend using the same family which
species. There are two general approaches to shares a similar feeding ecology and geo-
build a migrant isoscape: (1) model the spatial graphic range as your study species (and the
patterns directly using known-origin tissue same tissue type).
from the species you are studying and/or (2) The slope estimate is of particular impor-
model the relationship between an existing tance in calibration equations because it will
environmental isoscape and a more spatially influence how you interpret and evaluate
limited sample of known origin tissues; i.e., environmental-based isotopes. Ideally, β 5 1,
calibrate the environmental isoscape values to which means that isotopic values in animal tis-
reflect expected values in the tissue and use sue are scaled one-to-one according to a fixed
the calibrated isoscape for inference (see offset, β 0. In contrast, if β is between 0 and 1,
Section 8.3). Although the number of pub- then the range of isotope values observed in
lished known-origin isotope studies are animal tissues over space are compressed rela-
increasing, the probability you will find data tive to the range of isotope values of the
for the species you are studying is still low. environmental-based isoscape. In other words,
Developing known-origin isoscapes adds sig- you will see a smaller range in animal tissues
nificant effort and cost to your study but this isotope values over the predefined geographic
remains the ideal approach, and one we space than the environmental-based isoscape
strongly advocate if resources are available indicates. Conversely, if β is .1, then range of
(Chapter 4: Application of Isotopic Methods to isotope values observed in animal tissue over
Tracking Animal Movements). space is increased relative to the range of
Where it’s not possible to develop an iso- isotope values in the environmental-based
scape directly from known-origin tissues, isoscape. What causes β to be lesser than or
(e.g., because of a comparatively limited greater than 1 is unknown but, in the case of
extent of spatial sampling frames), isotope H isotopes, could be related to geographic

8.3 SAMPLING CONSIDERATIONS 195
variation in water use efficiency. Nevertheless, 8.3 SAMPLING CONSIDERATIONS
the implications of variations in β are largely
overlooked when planning and executing Upon deciding which isotopes will be feasi-
a study. ble and informative; a next step might be to
(5) Collect information on the tissue you wish to estimate the cost. The laboratory costs for
use in the organism you will be studying. The pri- stable isotope analysis are usually incurred on a
mary way isotopes are utilized to track move- per-sample basis, so it helps to have a sense for
ment is that (ideally) a tissue is grown in one how many samples need to be analyzed. The
location at known time and the animal can be field costs for sampling animal tissues can vary
sampled at another location in a subsequent widely, so we might also want to know which
time to estimate where that tissue was grown. tissues to sample, when to sample them, and
Most effective applications will therefore from which geographic locations. Reviewing
understand the growth rates and molting sche- the literature for guidance on these design con-
dules for the focus species. Information on siderations can be difficult; there are now hun-
growth rates of related species may be useful dreds of published studies using stable isotopes
but we caution about casting too wide of a net to study some aspect of animal migration, yet
for obtaining estimates because growth rates most give little to no justification for decisions
and molting schedules may vary due to a wide that were made about study design.
range of ecological and physiological reasons This section presents general guiding princi-
(see Chapter 1: Animal Migration: A Context ples for designing sampling efforts in studies
for Using New Techniques and Approaches). that use stable isotopes for a range of questions
(6) Consider adding alternative approaches. in animal migration ecology. The broader dis-
There is no guarantee that isotopes will be the cussion on sampling here is divided into three
best or only method to answer your research parts: (1) basic considerations related to the
question. External (and expensive) tracking type and amount of tissue and the timing of
devices, such as satellite and GPS tags or light- collection, (2) considerations for sampling ani-
logging geolocators, provide direct information mals of known origin as related to the con-
about provenance and explicit movement struction of models for assigning geographic
routes for species large enough to carry them. origins, and (3) considerations for sampling
Tagged animals can be also sampled for animals of unknown origin as related to com-
isotopes, which would provide known-origin mon animal migration research questions. For
tissues to calibrate an isoscape for use with each scenario in (2) and (3), we consider the
previously uncaptured animals. Trace element question: Suppose you can collect and analyze
analysis estimates the concentrations of several 100 samples; how best to allocate resources to
low-concentration elements simultaneously those samples? The answer to this question
and can be helpful for differentiating between varies by situation and is offered only as an
potential sites of origin. However, the high informal guideline meant to be interpreted
degree of site specificity calls into question only in relation to the answers for the same
whether trace element profiles form question in different scenarios.
predictable spatial gradients and it’s unclear
whether interpolation would be meaningful. If
not, then these markers may only be effective
when all potential sites are sampled (see
8.3.1 Tissue Amount, Type, and Timing
Chapter 1: Animal Migration: A Context for In all cases, one needs to know specific
Using New Techniques and Approaches). mass requirements for the instruments in the

isotope lab where samples will be analyzed. the previous winter. In some cases, molt sche-
For a range of reasons that are not important dules might prevent the use of stable isotopes
to our discussion, costs per sample and target for inferring origins of interest. Other useful
mass for analysis of different material types tissues include those that are inert once formed,
will vary among labs. Before starting any sam- but that experience continuous “shedding” or
pling program, therefore, we advise discussing growth (whiskers, otoliths, claws, fingernails,
these details with your lab (see Chapter 2: etc.). The trick with this type of material is to
Introduction to Conducting Stable Isotope calibrate growth rates so that distances along
Measurements for Animal Migration Studies). the material can be used to target specific time
Understanding the physiology of tissue periods of interest. It’s worth noting that the
dynamics is important. For example, migration same assumption about dietary equilibrium
studies commonly sample metabolically inert applies to these tissues, however.
tissues such as keratin (feathers, hair), chitin
(wings), or calcium carbonate (otoliths) because
such tissues remain isotopically unchanged 8.3.2 Sampling Considerations for
after formation (see Chapter 4: Application of Creating Assignment Models Using
Isotopic Methods to Tracking Animal Known-Origin Material
Movements). If metabolic pools are not in equi-
librium with the local diet prior to the onset of Before any individuals of unknown origin
tissue development, however, the resulting tis- can be assigned geographic origins, we need
sue will reflect the chemistry of dietary to assemble a model for assigning geographic
resources from both the location of growth and origins. The most effective models are con-
the location immediately prior to growth. In structed from tissues of known origin, sam-
some cases, such mixtures can generate isotopic pled from across the full gradient of parameter
profiles that vary widely from those that are space for each major source of variance. Those
created under conditions satisfying the single parameter spaces are not the same for all
geography dietary equilibrium assumption (see isotope-based assignment questions. Spatial
Wunder, Jehl, & Stricker, 2012). assignment models can be generated directly
Understanding of the natural history of the from tissues of the species under study, or by
target species helps determine which tissues to calibrating an extant isoscape for a different
sample and when to sample them. For example, material (e.g., water, soils, plants, and dis-
Woodworth et al. (2016) used tail feathers to solved inorganic carbon) using a small sample
estimate wintering locations of breeding savan- of known origin tissues. In both cases, it is pos-
nah sparrows (Passerculus sandwichensis). They sible to modify sampling designs to minimize
noted the feather condition as a way to estimate the need for extrapolation across geographic or
the probability that it was grown recently (i.e., isotopic space.
during the previous winter) rather than during
the previous summer. In general, it makes sense 8.3.2.1 Models That Directly Link Tissue
to sample feather tracts that are known to be Chemistry and Geography
replaced during discrete time periods that are The most direct way to fit an assignment
relevant for the study question. For example, model for the species under study is to collect
birds with complex molts that are sampled a sample of animal tissues of known origin
during the breeding season might carry flight from across the geographic range of interest.
feathers grown during the previous breeding The geographic area targeted for making
season and also contour feathers grown during assignments (e.g., a wintering range) can be

subdivided at various resolutions, depending locations across the geographic extent of the
on the application. The spatial resolution of grid. Here the aim is to characterize variance
the model structure will determine how best to over geography, so sampling should be bal-
allocate fixed sampling efforts. anced across the geography as a first step. If
For example, studies that aim to differentiate there are additional covariates in the interpola-
among a few to several predefined candidate tion model (e.g., elevation, temperature, and
origins (e.g., assignments to biogeographic or distance to coastline), then sampling can be
political regions) are best served by sampling comparatively saturated in geographic regions
many individuals from each of the few candi- that support the highest variances of those
date origins. These so-called nominal designs covariates.
depend on characterizing the isotope distribu- So back to our question: Suppose you can
tions within each (named) region. In these collect and analyze only 100 samples; how best
cases, the aim for sampling is to obtain a rep- to do it? The answer depends on whether the
resentative sample from each region. In our assignment region is partitioned into several
experience with empirical datasets, isotope named units, or whether it is to be partitioned
values for tissues of known origin in animal on a regular grid at some specified geographic
populations are usually normally distributed. resolution. For cases where the assignment
Thus tests based on the assumptions for a nor- region is predivided into n regions, we suggest
mal distribution will depend on the sample that the same number of samples should be
being large enough to reliably estimate the collected from each of the n regions. So, for a
parameters of that distribution. There are rules study that wants to differentiate among 10
of thumb for the size of such samples (e.g., regions, we might take 10 samples from each
NB2050), which would suggest target sam- region. This may not be a large enough sample
ple sizes from each of the named regions of to reliably estimate the model parameters,
interest. However, we note that isotopic gradi- especially if the samples need to be dispersed
ents can exist within regions that are defined across a known isotope gradient within any of
by other information (e.g., political boundaries the regions. On the other hand, if the aim is to
and animal population density contours). In make assignments to a regular grid of dimen-
such cases, assumptions of normality and the sion r*c, we suggest that the 100 samples be
suitability of requisite sample sizes will not distributed sparsely across the geographic
hold unless the spatial extent of the sampling extent of interest. This can be done uniformly
frame considers the confounding gradient. if the isoscape will depend only on space.
A recent solution to the problems suggested Alternately, if spatial covariates are to be used
above is to make assignments to origin by in the construction of the isoscape, we suggest
gridding a spatially interpolated model surface spatially clustering sample locations where
(isoscape). In these applications, known-origin variances of spatial covariates are highest. In
tissues are modeled over space, with or with- this way, spatial variances within the isoscape
out additional covariates, using a regular grid are better estimated from the sample. In all the
of cells (a raster). As a research community, cases, the resultant estimated distributions of
we have often referred to this approach as isotopes (for each named region or grid cell)
using a “continuous” surface, but note that a can be weighted by the spatially collocated
more appropriate title might be hyperdimen- distributions of other covariates, such as rela-
sional nominal approach. These gridded mod- tive abundance, to correct for unequal propor-
els are best fit from sampling frames that tional sampling across covariates among
produce one or a few individuals from many regions or cells.

8.3.2.2 Models That Indirectly Link Tissue we need minimally sample from the locations
Chemistry and Geography with the minimum and maximum isotope
Spatial origin assignment models can also values in the baseline isoscape.
be created by calibrating an existing isoscape Now again to our question: Suppose you
model with known-origin tissue samples from can afford only 100 samples? In this case, we
the species under study. Thus this approach is want to distribute the samples evenly across
like creating isoscapes directly from tissues of the range of isoscape values for the region of
interest in that both require tissues from interest. For example, we might first restrict
animals with known geographic histories. The (mask) the baseline isoscape for water to the
main difference is that geographic patterns for wintering range of our study species. Suppose
isotope values in the calibration approach are we then query the H isotope isoscape and find
provided from models using samples of some the minimum predicted value for the winter-
other material, which reduces the need for tis- ing range is 2125m and the maximum value is
sues samples from as many locations. The cali- 249m, giving a range of 80m. If we had 100
bration from values for the unrelated material samples, we might identify 20 geographic loca-
to those for the material of interest is usually tions where the isoscape predicted values
done with some type of regression model. The B4m apart and sample 23 animals from each
x variable is the isotope value of the unrelated location. Or, at the minimum, we might
material (rain, soil, etc.) and the y variable is instead find locations where the isoscape pre-
the isotope value in the tissue of interest dicted 249m and 2125m and sample one ani-
(feather, fur, otolith, etc.). This implies that the mal from each of those two sites and save the
calibration tissue must be of known origin, so remaining funds for page charges!
the isotope value (y) can be paired with the
isoscape value (x) at that known location. 8.3.2.3 Creating Assignment Models
That is, geographic colocation of sampled Without Sampling Known-Origin Material
tissue and isoscape prediction allow the cali- We strongly advise against taking this
bration. Because the isoscape provides pre- approach but recognize that often it is the only
dicted values for every cell on the grid, and option. This model building approach does not
because the calibration is a regression of require sampling any tissues of known origin.
isotope values (and not of geographic values), It rests entirely on the assumption that surro-
tissue sampling designs should span the range gate species are representative of the species of
of isotopes in the isoscape, rather than the interest, and that previously published work
range of geographic locations within the target was rigorous and transparent. There is no for-
assignment region. To minimize the sampling mal way to test these assumptions without
effort while still preventing the need for sampling tissues of known origin, so if tissues
extrapolation, we can identify, and sample of known origin are unavailable, conclusions
locations associated with the endpoints of the necessarily rest on untestable assumptions. If
isotopic range covered by the isoscape for the possible, we recommend making the extra
unrelated material. That is, the calibration effort to use known-origin material as
curve for translating values from the isoscape described by any of the previously mentioned
for unrelated material to values for the tissue modeling approaches instead.
of interest should cover the full extent of Now our question: Suppose you can collect
isotope values in the isoscape. To ensure this, and analyze 100 samples? Save the funds for

collecting samples of known origin to train geographic locations, sample tissues from 5
your model or hire a statistician to tell you that individuals.
your conclusions will be otherwise intractable!
8.3.3.2 Migratory Catchment and Dispersal
Questions
8.3.3 Sampling Considerations for Migratory catchment and dispersal studies
Typical Research Questions About sample a single aggregation (breeding site,
Animals of Unknown Origin stopover site, and wintering locale) to under-
stand the extent to which distinct breeding
8.3.3.1 Migratory Connectivity Questions populations use the site. Such studies typically
Migratory connectivity research focuses on concern the geographic “catchment” region for
the geographic structure of seasonal connectiv- birds using the site. Did young birds disperse
ity across the full geographic range of a spe- into the breeding area for disparate natal
cies. The general isotope sampling plan for regions? Are birds from multiple breeding
such studies is to collect tissues that were areas passing through a migration corridor?
developed during the season opposite from What is the geographic extent of summer ori-
when the sampling occurs. For example, to gins for bats killed by wind turbines? For
connect breeding birds to nonbreeding locales, power considerations in such studies, it makes
sample from birds during the breeding season, sense to sample deeply at the location of inter-
but take feathers that were grown during the est as a priority, and widely across time as
previous nonbreeding period; to connect non- funds allow. That is, collect sample tissues
breeding birds to breeding areas, sample dur- from as many individuals of the species of
ing the nonbreeding period, but take feathers interest as funding allows. This will increase
that were grown during the previous breeding the chances of sampling passage individuals
season. For power considerations, prioritize that originated from low-density areas.
number of geographic locations over number Now to our 100-sample question: Because
of individuals at each location. It will be more these studies ask about the geographic origins
informative to have fewer samples from more of individuals at one or a few sites, we would
locations than to have more samples from want to collect all 100 samples from that site.
fewer locations, as connectivity studies are If the question concerned migratory stopover,
interested in the geographic variation of con- we might want to distribute the samples
nections across seasons. To increase the robust- evenly across the migratory period. If the
ness of the study, sample more deeply within question concerned dispersal or some other
each geographic location as funds allow. demographic-related population level parame-
Our question: suppose you can collect and ter, we might want to consider balancing our
analyze 100 samples? To the extent possible, samples across age and sex classes with the set
we want to represent the distribution across of animals using the site.
(not within) geographies for both seasons of
interest. For example, we might identify 10 8.3.3.3 Migratory Carry-Over Questions
distinct accessible breeding locations that Migratory carry over studies are interested
span the geographic extent of the breeding in questions of how conditions and experi-
range and 10 distinct and accessible non- ences during one season influence those in a
breeding locations. From each of these 20 subsequent period. Frequently, these studies

are interested in how habitat use and migra- spatially interpolated model can provide predic-
tory timing are related to demographic rates, tions for a series of gridded points across a
and so sampling is most often conducted at a geography of interest. For example, δ2H values
single or a few locations. Among-individual in precipitation have been empirically modeled
variation in demographic rates is often small for terrestrial regions (Bowen, Wassenaar, &
and difficult to estimate, so such studies Hobson, 2005; Terzer et al, 2013). The genera-
would do well to sample as many individuals tion of these models produces isoscapes that
as possible from the single site of interest. can be used to infer regions of probable tissue
And if you can collect only 100 samples? As growth for individuals of unknown origin
with the catchment studies, we want to collect (Chapter 3: Isoscapes for Terrestrial Migration
all our samples from the same site, but can Research).
consider stratifying over time, age, sex, etc. as The key assumption, however, is that geo-
the study plan demands. For example, these graphic patterns in environmental
types of studies sometimes concern histories stable isotopes are faithfully maintained or
and fates of mated pairs, so we might want to translated through food webs to the animal of
ensure we sample each member from as many interest. In other words, the isotopic discrimina-
pairs as possible. tion between the stable isotopes in dietary
sources and the tissue in the species of interest
is both predictable and constant in time and
space (Chapter 4: Application of Isotopic
8.4 DATA ANALYSIS AND
Methods to Tracking Animal Movements,
MODELING CONSIDERATIONS
Chapter 5: Tracking of Movements of
Terrestrial Mammals Using Stable Isotopes).
8.4.1 Calibration and the Basic
After an animal incorporates isotopic values
Assignment Problem from the environment at some location, meta-
An essential premise in using stable iso- bolically slow or inert tissues sampled later,
topes to track the movement of migratory ani- when the animal is in a different area, can be
mals is that detectable and predictable patterns used to infer previous geographic locations
exist in the spatial distributions of stable (Wunder, 2010).
isotopes in the environment. For example, Just as all isotope-ratio mass spectrometry
there are relatively strong, well-known, and data depend on run-time calibrations with uni-
globally predictable geographic gradients in versally accepted reference materials (measure-
the hydrogen (δ2H) and oxygen (δ18O) isotope ment standards), so too do models for
compositions of meteoric water (Dansgaard, assigning individuals to locations depend on
1964). Geographic gradients in other isotope sys- assign-time calibrations using geographic stan-
tems are less well understood, but modeling dards (tissues of known geographic origin).
progress continues (Chapter 3: Isoscapes for The accuracy of geographic model calibrations
Terrestrial Migration Research). is improved by using geographic standards
When geographic patterns are detected for with attributes like the samples for assignment
stable isotopes they are in fact inferred from a to geographic provenance. This means that
finite number of sampling points across the geographic assignment models perform best
landscape. Although patterns in δ2H of feath- when calibrated using geographic standards
ers have been modeled as linear functions of from the same species as the unknown sam-
latitude and longitude (Kelly, Atudorei, Sharp, ples, and that bracket the range of isotope
& Finch, 2002; Rubenstein et al., 2002), a values and all other covariates in both time

8.4 DATA ANALYSIS AND MODELING CONSIDERATIONS 201
and space from which the samples are drawn. “δ” notion (Chapter 2: Introduction to
Ideally, a calibration dataset would include Conducting Stable Isotope Measurements for
samples obtained from across all potential Animal Migration Studies). The full range of
areas in which the migratory species of interest possible values is never covered by any natu-
could have originated (see Section 8.3). ral dataset, and the range of interest is almost
Alternatively, isoscapes can be derived always narrow enough that a careful linear
directly from tissues sampled from the species calibration is an excellent approximation
of interest over the period of interest (Hobson, (Gröning, 2004). The important point to recog-
Wassenaar, & Taylor, 1999). In such cases, the nize is that the δ-value of a sample is not an
calibration is incorporated directly into the exact value but is also derived from a linear
spatial interpolation. Either way, an isoscape is regression of results from primary reference
a static surface that is generally then treated as materials (standards). IRMS measurements of
a deterministic process to describe the spatial bulk unknown samples (of variable mass and
pattern in the isotope of interest. As we dis- isotopic heterogeneity) cannot be 100% accu-
cuss below, calibrated models represent only rate and are affected by inherent variability in
the first step in the modeling process for the preparative procedures. This point should
assigning provenance. not be overlooked by those wanting to use
We recognize that in most cases, developing stable isotopes in ecology. It is not reasonable,
species-, tissue-specific or even compound- e.g., to claim that a difference of 10m between
specific isoscapes may be unrealistic due to two populations is meaningful if the analytical
both logistical difficulties (e.g., inaccessible error associated with the isotope measure-
areas) and financial and analytical constraints. ments is 6 2m. This is because typical hypoth-
There is, therefore, a critical need to test esis tests for “significant” differences between
general hypotheses that attempt to explain means assume that the δ-values are exactly
observed differences in isotope discrimination known. However, the δ-values are each only
factors among and within species, among dif- known to within 6 2m, which means a 4m dif-
ferent tissues, and across space. Testing should ference for two different measurements of the
include a combination of in situ studies of same exact sample is to be expected. More
wild animals and controlled lab studies, ide- importantly, this suggests the utility of using
ally in a back-and-forth way that promotes stochastic or probabilistic approaches to ana-
understanding of mechanistic factors in natu- lyze conclusions about stable isotope data.
ral settings. Only by advancing understanding
in this way will we be able to derive some
general rules about variation in isotope dis- 8.4.3 Basic Statistical Assumptions
crimination and be able to identify optimal
sampling schemes for generating more robust 8.4.3.1 Statistical Independence
assignment models. The assumption of statistical independence
requires samples be drawn at random from
the at-large target population of interest.
Linear models that relate stable isotope values
8.4.2 Characteristics of Isotope Data for
to latitude, e.g., clustered sampling designs
Use in Studying Migratory Animals (e.g., many individuals from a few locations)
Measurements of stable isotope values using illustrate a classic violation of this assumption.
IRMS are reported relative to a known interna- For example, suppose that feathers from 140
tionally accepted reference and use the typical birds were sampled for δ2H. Suppose that

those 140 birds were taken from only 10 sites is IRMS measurement error derived from run-
that vary in latitude. In this case, there are not time calibrations, and it is a simple matter to
really 138 degrees of freedom for fitting a lin- measure the calibration residuals to quantify
ear model (140 minus one for the slope and analytical error. Although the variance is usu-
one for the intercept); there are only 8 degrees ally not the same for each analysis run, the
of freedom. In this example, site is the more asymptotic standard deviation of these residuals
appropriate sample unit because we are trying is generally all that is reported. As more is
to relate stable isotope values to geographic learned about systematic deviations from this
variables, and each of the 14 birds sampled assumption of process homogeneity, relevant
from a single site share the same value for the covariates can be added to adjust models.
response variable (latitude). The samples for
each site are better treated as pseudo-replicates
than as independent samples. Statistically, this
8.4.4 Population Versus Individual-Level
problem can be overcome by including a ran-
dom effect term (site) in the linear model or
Geographic Assignments
considering a repeated measures modeling Are we interested in the distribution of geo-
framework. graphic origins estimated from δ-values for
individuals in a sample population of migra-
8.4.3.2 Identically Distributed (Process tory animals, or are we interested in the geo-
Homogeneity) graphic origin associated with the average
Implicit in stable isotope studies that seek δ-value for a population of migratory animals?
provenance of migratory animals is the These subtly different questions have tradi-
assumption that variation in isotope values tionally been treated in different ways.
among individuals at a location is governed by However, we argue that population-level ques-
the same variance generating processes. For tions about the origin of migratory animals are
example, we assume that all individuals of the best treated by compiling individual-level
target species respond in the same way to assignments rather than summarizing δ-values
environmental stresses, foraging at roughly the from a given sample of individuals (e.g., using
same position in relatively similar food webs, the average δ-value). This is because we are
and developing tissues at roughly the same not really interested in δ-values themselves,
rate. This strongest assumption, that a site is but rather in a transformation of those values
isotopically homogeneous, will almost never to some geographic location(s). Because there
be met. is not a perfect 1:1 transformation from δ-value
Isotopic variation among individuals stems to geographic coordinates, it is important to
from differences in what they consume, when transform the δ-value for each individual data
they consume it, and under what conditions point into a geographic value before determin-
they develop tissues, all of which potentially ing the population-level characteristics of the
contain useful information. The second source distribution of geographies. Otherwise, we for-
of stable isotopic variability comes from isoto- feit gains from quantifying systematic and
pic heterogeneity within an individual animal. informative variance.
Organic tissues within an animal may turnover The goal of many studies of migratory ani-
at different rates. As such, this too can provide mals is to describe geographic structure in the
useful information about changes in geogra- sampled population. This implies an interest
phy, environmental conditions, diet, or life in the characteristics of the distribution of
history tradeoffs. The third source of variance assignments, rather than in the geographic

8.5 ASSIGNMENT MODEL TYPES 203
assignment for a distribution characteristic isotopes, but also work fine with any combination
(e.g., the mean). Assigning individuals to geo- of continuous or discrete predictor variables.
graphic origins one at a time and exploring Hebert and Wassenaar (2005a, 2005b)
properties of the resultant geographic distribu- employed this approach to assign mallards
tion is more flexible and natural than using the (Anas platyrhynchos) and northern pintails
average of δ-values to determine geographic (Anas acuta) to one of four predefined geo-
histories for a sample population for two graphic regions in North America using the
important reasons. First, describing a popula- values of δ2H, δ34S, δ13C, and δ15N in feathers.
tion by calculating the arithmetic average Their application used univariate thresholds
(mean) and standard deviation for the isotopic for decision branching. They tested the
data implicitly assumes normality, but many robustness of their model by assigning
interesting geographic structures are mixtures known-origin data not used to generate the
of distributions. Second, δ-values in the same classification tree. They did not, however,
tissue from different individuals from the explore the robustness of the approach to
same population (location) are never identical. how the potential geographic origins were
Moreover, repeated samples from the same predefined. As with all assignment methods
individual are not even expected to be identi- discussed here, the efficacy of the approach
cal. Because of this, the transformation from will depend strongly on how the potential tar-
δ-values to geographic location is not done get geographic regions are defined.
with complete certainty. By assigning indivi- The branching thresholds in classification
duals to geographic locations first, we can trees can be fixed points along univariate gra-
propagate the variance associated with the dients as in Hebert and Wassenaar (2005a,
transformation and provide a more compre- 2005b) or they can be linear combinations of
hensive estimate for the geographic structure the predictor variables, as in discriminant
of the sample population. function analysis. Rule sets derived from clas-
sification trees provide no specific informa-
tion about how close the call is for a given
split. In other words, it is very difficult to
8.5 ASSIGNMENT MODEL TYPES quantify the relative strength favoring one
branch versus another. There are a host of
8.5.1 Classification Trees software programs that employ various algo-
A classification tree is a derived hierarchy of rithms to ensure optimization of the tradeoff
decision rules for assigning novel data to one of between number of branching splits and pre-
two or more classes. Each decision rule provides dictive accuracy, but optimizations are just
a fork, or “branching event” in the flow toward that—optimal solutions given the information
assigning a data point to one of the predefined at hand. If the likelihood for one region is
classes (regions). Classification trees do not nearly identical to that for another, there will
require distributional assumptions and can com- still always be an optimal solution for differ-
bine discrete and continuous covariates. They entiating between them, but this is not neces-
rely on clustering algorithms or similar recursive sarily saying we have a great deal of
computations. Thus, classification trees will be confidence in doing so. As with any statistical
especially useful for future exploratory work modeling approach, this is just another way
and pilot studies, especially as more isotopes of saying that the performance of classification
and trace elements are used because they accom- trees is limited by the quality and quantity
modate not only multimodal distributions of of data.

8.5.2 Likelihood-Based Methods probability of B given A, or the probability of

some model parameters or hypotheses, given
One form of likelihood-based methods is data. P(A|B) is the likelihood function, or the
discriminant analysis which can be used to probability of the observed data given some
classify a sample into one of two or more clas- model parameters or hypotheses. P(B) is
ses (regions). It has been used by several referred to as the prior or marginal probability
isotope-based studies to estimate the prove- for B, or the probability of some model param-
nance of animals (Caccamise, Reed, Castellie, eter without knowledge from data, and P(A) is
Wainwright, & Nichols, 2000; Farmer et al., a marginal probability that serves as a normal-
2004; Kelly, Ruegg, & Smith, 2005; Rocque, izing constant. P(A) integrates all possible out-
Ben-David, Barry, & Winker, 2006; Szymanski, comes for A, regardless of B. Because A is
Afton & Hobson, 2007; Wassenaar & Hobson, conditioned on B, we sum or integrate over P
2000). Parameters for region-specific likelihood (A|B)*P(B). That is, we integrate over all possi-
functions (always some form of the normal ble outcomes where A occurs.
distribution) are then estimated from isotope The use of probabilities using Bayes’ Rule
data collected in each predefined region. The makes the interpretation of results more
likelihood functions for each region are then straightforward than with likelihoods alone
evaluated for the isotope value measured for because we talk directly about the probability
an individual of unknown origin, and the indi- of model parameters or hypotheses, rather
vidual is then simply assigned to the region than indirectly about parameters by describing
with the highest valued likelihood (or proba- the probability of observing the same data in a
bility). One advantage of likelihood-based hypothetical replicate sampling event. In fact,
methods is that they consider not only the discriminant analysis can be thought of as a
mean isotope value of each region, but also special case of Bayes where the likelihood is
some measure of the variability. normal, and the prior is uniform over the
Likelihoods can easily be converted to prob- candidate regions. Discriminant analysis is,
abilities using Bayes’ Rule, which is a formal therefore, naturally extended into a more
way to invert conditional probabilities. Bayes’ Bayesian approach by giving structure to the
Rule is simply an algebraic manipulation of an prior probability distribution.
equality that springs from the definition of As an example, Royle and Rubenstein
conditional probability. The probability of two (2004) assumed normal distributions for the
events both happening, P(A,B) is the same as stable isotope values in feathers collected from
the probability of one event happening P(A), each of three broad regions and argue that
joined with the probability of the second event relative abundance is a good proxy for the
happening, given that the first event occurred, prior probability that an individual of
P(B|A). The vertical line means “given,” or unknown origin came from any of those three
“conditioned on.” In other words, P(A,B) 5 P regions. In other words, they point out that in
(A)*P(B|A). the absence of any isotopic information, the
Because it doesn’t matter which event probability of an individual originating from
occurs first, we can also write P(A,B) 5 P(B)*P any given region is proportional to the relative
(A|B). Recognizing that these expressions are abundance of animals from that region. Norris
equal, we can write P(B)*P(A|B) 5 P(A)*P(B| et al. (2006) also used relative abundance for
A), rearrange and get P(B|A) 5 P(A|B)*P(B)/ the prior probability distribution, but they dif-
P(A), which is Bayes’ Rule. In this expression, fered from Royle and Rubenstein (2004) by
P(B|A) is often referred to as the posterior using precipitation-based isoscape values to

REFERENCES 205
define the sampling probability distributions distance from a true data station where δ2H in
for each region. In cases where relative popu- precipitation was directly measured. Thus the
lation abundance estimates are not available, overall confidence of assignments to these
the number of individuals sampled from each regions was much lower.
potential area of origin can be used as a proxy This stochastic modeling extension was use-
for the prior probability (Wunder, Kester, ful for exploring the relative sensitivity of con-
Knopf, & Rye, 2005). clusions to the assumptions that isoscapes are
Assuming normal distributions for the like- perfectly predicted and that δ2H in feathers is
lihood function is by no means a requirement. perfectly measured. It represents a prelimi-
Given the relatively sparse nature of the data nary, if not also rudimentary, approach toward
in these studies (as with many migratory stud- studying the state of our understanding of the
ies), the assumption of normality was used to relationship between δ2H in feathers and geog-
marginalize the influence of outliers in the raphy. This method can be extended to include
data and to simplify computations. other factors known to influence the certainty
We used simulations to explore the reliabil- of our modeling efforts. For example, we
ity of asymptotic results from Bayesian assign- might consider the effects of reasonable sto-
ments given stochastic fluctuations associated chastic fluctuations in the functional relation-
with IRMS measurements and with the process ship between δ2H in precipitation and that in
of spatially interpolating sparse data on δ2H in feathers (variable isotopic discrimination).
precipitation (Wunder & Norris, 2008). There,
we reevaluated a study by Norris et al. (2006)
that calculated the mean and standard devia-
tion of expected δ2H values for five predefined
8.6 CONCLUSION
breeding regions. The δ2H values were
Stable isotopes are useful for studying geo-
extracted from spatially interpolated precipita-
graphic histories of previously unmarked
tion maps (Bowen et al., 2005) and standard
migratory animals. In this chapter, we have
deviation values for the interpolation in the
provided a fundamental overview of design
raster ranged from 5m to 13.6m. We considered
and analysis considerations for such studies.
not only variation in expected values from the
In doing so, we highlight assumptions associ-
interpolated precipitation-based δ2H isoscape
ated with various models, show how sources
but also the uncertainty associated with mea-
of variance can be more directly incorporated,
suring δ2H in feathers and the uncertainty
and emphasize the importance of understand-
associated with the spatial interpolation itself.
ing the mechanisms that generate isotopic vari-
The latter portion of uncertainty captures the
ation. Our hope is that future research will
variation associated with how far a geographic
adopt and further refine these approaches.
point is away from an actual data sampling
station; the further away a point is the higher
variability there will be in interpolating the
δ2H value. Regions with high variability in any References
of the distributions of predicted δ2H likely Bataille, C. P., Laffoon, J. E., & Bowen, G. J. (2012).
reflected higher amounts of topographic relief Mapping multiple source effects on the strontium isoto-
and temperature variation over its range. pic signatures of ecosystems from the circum-
Caribbean region. Ecosphere, 3(12), art118.
Regions with high spatial interpolation uncer- Bowen, G. J., Wassenaar, L. I., & Hobson, K. A. (2005).
tainty were associated with similar physio- Global application of stable hydrogen and oxygen iso-
graphic features, but also reflected the relative topes to wildlife forensics. Oecologia, 143, 337348.

Caccamise, D. F., Reed, L. M., Castellie, P. M., Wainwright, Rocque, D. A., Ben-David, M., Barry, R. P., & Winker, K.
S., & Nichols, T. C. (2000). Distinguishing migratory (2006). Assigning birds to wintering and breeding
and resident Canada Geese using stable isotope analy- grounds using stable isotopes: Lessons from two
sis. Journal of Wildlife Management, 64, 10841091. feather generations among three intercontinental
Dansgaard, W. (1964). Stable isotopes in precipitation. migrants. Journal of Ornithology, 147, 395404.
Tellus, 16, 436468. Royle, J. A., & Rubenstein, D. R. (2004). The role of species
Farmer, A., Abril, M., Fernandez, M., Torres, J., Kester, C., abundance in determining breeding origins of migra-
& Bern, C. (2004). Using stable isotopes to associate tory birds with stable isotopes. Ecological Applications,
migratory shorebirds with their wintering locations in 14, 17801788.
Argentina. Ornitologia Neotropical, 15, 377384. Rubenstein, D. R., Chamberlain, C. P., Holmes, R. T.,
Gröning, M. (2004). International stable isotope reference Ayres, M. P., Waldbauer, J. R., Graves, G. R., & Tuross,
materials. In Pd Groot (Ed.), Handbook of stable isotope N. C. (2002). Linking breeding and wintering ranges of
analytical techniques (Vol. 1, pp. 874906). Amsterdam: a migratory songbird using stable isotopes. Science, 295,
Elsevier. 10621065.
Hebert, C. E., & Wassenaar, L. I. (2005a). Feather Szymanski, M. L., Afton, A. D., & Hobson, K. A. (2007).
stable isotopes in western North American waterfowl: Use of stable isotope methodology to determine natal
Spatial patterns, underlying factors, and management origins of mallards at a fine scale within the upper
applications. Wildlife Society Bulletin, 33, 92102. Midwest. Journal of Wildlife Management, 71, 13171324.
Hebert, C. E., & Wassenaar, L. I. (2005b). Stable isotopes Terzer, S., et al. (2013). Global isoscapes for δ18O and δ2H
provide evidence for poor pintail production on the in precipitation: Improved prediction using regional-
Canadian prairies. Journal of Wildlife Management, 69, ized climatic regression models. Hydrology and Earth
101109. System Sciences, 17, 47134728.
Hobson, K. A., Wassenaar, L. I., & Taylor, O. R. (1999). Wassenaar, L. I., & Hobson, K. A. (2000). Stable carbon
Stable isotopes (δD and δ13C) are geographic indicators and hydrogen isotope ratios reveal breeding origins of
of natal origins of monarch butterflies in eastern North red-winged blackbirds. Ecological Applications, 10,
America. Oecologia, 120, 397404. 911916.
Kelly, J. F., Atudorei, V., Sharp, Z. D., & Finch, D. M. Woodworth, B. K., Newman, A. E. M., Turbek, S. P.,
(2002). Insights into Wilson’s warbler migration from Dossman, B. C., Hobson, K. A., Wassenaar, L. I., . . .
analyses of hydrogen stable-isotope ratios. Oecologia, Norris, D. R. (2016). Differential migration and the link
130, 216221. between winter latitude, timing of migration, and
Kelly, J. F., Ruegg, K. C., & Smith, T. B. (2005). Combining breeding in a songbird. Oecologia, 181, 413422.
isotopic and genetic markers to identify breeding ori- Wunder, M. B. (2010). Using isoscapes to model probabil-
gins of migrant birds. Ecological Applications, 15, ity surfaces for determining geographic origins. In G.
14871494. Bowen, J. West, K. Tu, & T. Dawson (Eds.), Isoscapes:
Magozzi, S., Yool, A., Vander Zanden, H. B., Wunder, Understanding movement, pattern, and process on Earth
M. B., & Trueman, C. N. (2017). Using ocean models to through isotope mapping. New York: Springer-Verlag.
predict spatial and temporal variation in marine carbon Wunder, M. B., Jehl, J. R., & Stricker, C. (2012). The early
isotopes. Ecosphere, 8(5), e01763. Available from bird gets the shrimp: Confronting assumptions of
https://doi.org/10.1002/ecs2.1763. isotopic equilibrium and homogeneity in a wild bird
Norris, D. R., Marra, P. P., Bowen, G. J., Ratcliffe, L. M., population. Journal of Animal Ecology, 81, 12231232.
Royle, J. A., & Kyser, T. K. (2006). Migratory connectivity Wunder, M. B., Kester, C. L., Knopf, F. L., & Rye, R. O.
of a widely distributed songbird, the American redstart (2005). A test of geographic assignment using isotope
(Setophaga ruticilla). Ornithological Monographs, 61, 1428. tracers in feathers of known origin. Oecologia, 144,
Powell, R. L., Yoo, E.-H., & Still, C. J. (2012). Vegetation 607617.
and soil carbon-13 isoscapes for South America: Wunder, M. B., & Norris, D. R. (2008). Improved estimates
Integrating remote sensing and ecosystem isotope mea- of certainty in stable isotope-based methods for track-
surements. Ecosphere, 3(11), 109. Available from http:// ing migratory animals. Ecological Applications, 18,
dx.doi.org/10.1890/ ES12-00162.1. 549559.

C H A P T E R
9
Isoscape Computation and Inference of
Spatial Origins With Mixed Models
Using the R package IsoriX
Alexandre Courtiol1, François Rousset2, Marie-Sophie Rohwäder1,
David X. Soto3, Linn S. Lehnert1, Christian C. Voigt1,4,
Keith A. Hobson5,6, Leonard I. Wassenaar7 and
Stephanie Kramer-Schadt1
1
Leibniz Institute for Zoo and Wildlife Research, Berlin, Germany, 2Université de Montpellier,
Montpellier, France, 3KU Leuven, Leuven, Belgium, 4Freie Universität Berlin, Berlin, Germany,
5
University of Western Ontario, London, ON, Canada, 6Environment and Climate Change Canada,
Saskatoon, SK, Canada, 7International Atomic Energy Agency, Vienna, Austria
9.1 INTRODUCTION principle, it is in fact far more difficult in prac-

tice. Currently, there are no accepted or stan-
One of the major challenges facing research- dardized predictive geospatial modeling
ers using stable isotopes in evaluating migra- approaches that can guarantee consistent out-
tory origins of animals is how to use the comes, and scientists are faced with a confus-
stable isotope data obtained from a migrating ing array of methods and approaches using
species and place it “on the map.” Maps different assumptions and analytical techni-
depicting migratory origins of individuals, ques (Chapter 8: Design and Analysis for
migratory catchment areas of populations, and Isotope-Based Studies of Migratory Animals).
seasonal movements are powerful tools to help Therefore when tackling a migratory problem,
managers visualize and implement effective the scientist faces the difficult task of imple-
conservation efforts of many species, often menting and comparing them. This limitation
over large regions. While placing organisms calls for the development of software offering
on the map may seem straightforward in users a simple interface to such complex

DOI: https://doi.org/10.1016/B978-0-12-814723-8.00009-X 207 © 2019 Elsevier Inc. All rights reserved.
208 9. ISOSCAPE COMPUTATION AND INFERENCE OF SPATIAL ORIGINS WITH MIXED MODELS USING THE R PACKAGE ISORIX
methods. Ideally, such software would also but contains default user settings to restrain
offer developers means to compare, probe, and the burden of choices for inexperienced users.
test different methods. IsoriX is freely distributed as an R package
For these reasons, we developed IsoriX—the (Courtiol, Rousset, Rohwaeder, & Kramer-
first R-based package designed to model iso- Schadt, 2018).
scapes (spatial landscapes of environmental iso- IsoriX can be used to perform spatially
topic variation; Chapter 3: Isoscapes for explicit predictions from isotopic measurements
Terrestrial Migration Research) and to perform of samples collected within an area of interest,
geographic assignments (i.e., to infer the loca- so long as the measured δ-values are known to
tions of origin of samples from their vary spatially. The obtained predictions can
stable isotope ratios; Chapter 4: Application of then be used in performing successful geo-
Isotopic Methods to Tracking Animal graphic assignments if the measured δ-values
Movements). R is a cost-free statistical environ- show a strong spatial structure. Stable isotope
ment well known by scientists around the δ-values such as hydrogen stable isotopes (here-
world. The R package “IsoriX” implements the after, H isotopes) measured in precipitation
isoscape assignment methods outlined by water typically fulfill these conditions at a con-
Courtiol and Rousset (2017), which shares sim- tinental scale (Chapter 3: Isoscapes for
ilarities with IsoMAP and concepts outlined in Terrestrial Migration Research). IsoriX can be
Chapter 3, Isoscapes for Terrestrial Migration used for any application relying on geographic
Research, and Chapter 8, Design and Analysis assignments done this way, such as tracing
for Isotope-Based Studies of Migratory samples of (migrating) animals back to their
Animals. Because IsoriX is open source, scien- locations of origin. IsoriX therefore represents a
tists can work together to improve the soft- tool of interest for wildlife migration and con-
ware and its features into the future. servation, wildlife trade monitoring, and in for-
The purpose of this chapter is to demon- ensics. Plants, aquatic organisms, or other
strate the use of IsoriX for mapping origins, samples (e.g., water, wine) can also be assigned
using a case study of migratory bats. We walk- to the most likely place of origin using IsoriX
through the procedures and highlight the providing adequate spatial isotopic measure-
assumptions and caveats that researchers need ments and patterns exist.
to be aware of. It is our hope that this worked In this chapter, we demonstrate IsoriX in an
out example will guide and stimulate others to illustrative case study to determine the origin
use and improve the IsoriX package. of migrating bats using H isotopes. We use the
H isotope ratios of precipitation water to fit an
isoscape and illustrate geographic assignments
9.2 WHAT IS ISORIX? to determine, using H isotopes from fur sam-
ples, whether some bats (common noctule
IsoriX is a software package aiming at turn- bats, Nyctalus noctula) found dead underneath
ing the complex statistical problems of iso- wind turbines located in Germany originated
scape construction and geographic locally or had migrated from elsewhere (see
assignments into straightforward workflows, Chapter 5: Tracking of Movements of
starting with the preparation of datasets and Terrestrial Mammals Using Stable Isotopes).
finishing with the production of high quality The purpose of this chapter is to guide newco-
maps meeting publication standards (Fig. 9.1). mers through the entire IsoriX assignment pro-
Because the optimal workflow will differ cess for learning purposes. While we
depending on the application, IsoriX is flexible demonstrate a H isoscape assignment in this

9.2 WHAT IS ISORIX? 209
FIGURE 9.1 Example of a H isoscape predicted with IsoriX. This isoscape represents the mean hydrogen isotopic com-
position of precipitation water predicted across Europe. Source: Data obtained from GNIP at www.iaea.org/water.
chapter, the IsoriX approach is equally applica- mapped. IsoriX provides maps representing
ble to other stable isotopes (i.e., 18O, 13C) and variance terms around the predicted isoscape.
chemical tracers. IsoriX can then be used to conduct geographi-
An isotope-based assignment workflow in cal assignments, whether the samples to be
IsoriX is summarized as follows: The user pro- assigned have isotope values following the
vides an isotopic dataset containing δ-values same geographical variation as the source sam-
from source samples (e.g., δ2H values in precipi- ples, or not. In the latter case, IsoriX is used to
tation; called source data or baseline data), an fit a new calibration model based on known-
optional dataset containing δ-values of known- origin calibration samples. The fitted calibra-
origin calibration samples (e.g., sedentary organ- tion model defines how the isotope values of
isms; also called reference samples), and a the samples are related to those of the environ-
dataset of isotope δ-values from the assignment ment where they were collected (e.g., isotopic
samples (e.g., the organisms of unknown ori- values of precipitation). The calibration model
gin). In the first step, a pair of geostatistical is used by IsoriX to perform the geographic
models is fitted to the source samples. These assignment of the assignment samples.
fitted models are used by IsoriX to predict the Otherwise, one can directly proceed to the
isotope isoscape, i.e., to predict the isotope assignment step in IsoriX without the need to
source values across the area that needs to be perform the calibration procedure.

Performing these three main steps (fitting Archive Network (CRAN). For more casual
the isoscape, calibration, and geographic material, several introductory books are avail-
assignment) within a single software package able (e.g., Getting Started with R by Beckerman,
presents two major benefits. First, it makes it Childs, & Petchey, 2017), as well as video
possible to tailor each step according to the tutorials on the net.
data at hand and the research question in There are several reasons for the wide-
mind. Second, IsoriX offers the possibility to spread use of R. R is cost-free and open source
propagate the uncertainty stemming from each (R Core Team, 2017) and can run on computer
step into the final assignment task, thereby systems using Linux, Mac, or Windows, and
strongly increasing the robustness of geo- on laptops to supercomputers. R is a program-
graphic assignments (see also Chapter 8: ming language that is relatively easy to learn,
Design and Analysis for Isotope-Based Studies which makes it possible for researchers to cre-
of Migratory Animals). ate new functionalities (Chambers, 2016).
The rest of this chapter is organized into These new functionalities are distributed to
four main parts (Sections 9.39.5 and other R users in the form of archive folders
Appendix). The first part (Section 9.3) briefly called extensions or packages, which are easy to
introduces R for new users unfamiliar with install or update. Therefore the number of
this statistical environment, and provides packages continues to grow and CRAN—the
information about the internal structure of main online repository for hosting such
IsoriX, such as the names of the main functions packages—reached over 12,000 packages in
and their purpose. The second part January 2018. IsoriX is one of these packages.
(Section 9.4) illustrates a simple example of the A growing number of ecologists use and
different steps needed to complete an assign- rely on R packages. Most use packages of
ment procedure and discusses the assumptions generic tools to organize, analyze, and visual-
being made at each step. The third part ize their data. Examples of well-known R
(Section 9.5) discusses the future of IsoriX. We packages are “sp” (Bivand, Pebesma, &
also provide a summary of the statistical Gómez-Rubio, 2013) or “raster” (Hijmans,
framework used in IsoriX as an Appendix to 2016) used to create or edit spatial objects,
this chapter. “lme4” (Bates, Mächler, Bolker, & Walker,
2015) used to fit linear mixed-effects models
(LMMs; a type of linear regression), and “lat-
tice” (Sarkar, 2008) or “ggplot2” (Wickham,
9.3 THE ISORIX PACKAGE 2009) used to produce high-quality figures.
Packages developed for specific research meth-
9.3.1 The R Environment odologies also offer interfaces between users
The IsoriX software package must be run with little experience in programming, or with
within the software environment called “R,” little knowledge in the methods they wish to
which is the most used statistical software apply, and the complex procedures needed for
environment for data analysis (Rexer such methods.
Analytics, 2015). A full introduction to R is In the context of isoscapes, R offers several
beyond the scope of this chapter, however, packages to fit dietary mixing models for infer-
with millions of R users, there is no shortage ence on the diet of organisms by comparison
of online material to learn how to use and proof the isotopic composition of tissues to their
gram R. A thorough official documentation is potential food sources. These include
freely available on the Comprehensive R “IsotopeR” (Ferguson & Hopkins, 2016),

9.3 THE ISORIX PACKAGE 211
“MixSIAR” (Stock & Semmens, 2013), and use IsoriX; this makes the workflow fully
“simmr” (Parnell, 2016). Keep in mind these transparent, reproducible, easy to communi-
mixing models are not the same as the statistical cate, debug, and revise. Once an IsoriX script
framework of mixed models used in IsoriX. is written, users will discover that recycling
Methods used to perform isotope mixing anal- their script for other purposes is straightfor-
ysis conservatively “mix” probability distribu- ward. Adapting the same workflow to another
tions. They are distinctive from mixed models dataset often requires editing only a single line
which “mix” fixed effects and random effects. of code. Fifth, as in other R packages, the
source code behind all the functions of IsoriX
is open source. Thus anyone can read our com-
puter code. This transparency implies that
9.3.2 IsoriX and R experts can review procedures and provide
The first version of IsoriX was made avail- feedback for improvements. R offers the possi-
able on CRAN on November 2016, and since bility for advanced users to experiment with
then has been regularly updated to improve existing IsoriX functions to modify or extend
functionalities and documentation. functionality. They can do so for their own pri-
Because IsoriX is an R package, there are vate purpose, but this also opens the possibil-
several benefits compared to alternative imple- ity that IsoriX becomes a more collective
mentations of assignment workflows (e.g., the project (see Section 9.5). IsoriX is distributed
IsoMAP toolkit). First, many ecologists have under a license GNU General Public License
learned R for other reasons, and are familiar v3.0, which means that developers can also
with the R working environment. Second, copy code and use it in their own alternative R
because R is an interactive language, it enables packages or other software, by providing refer-
users to analyze their data in detail and to ence to IsoriX and provided that their work is
visualize intermediate results at each step of distributed under the same license terms.
the workflow. Third, it allows users to inte-
grate IsoriX into wider workflows; e.g., users
may link the inputs or outputs of their calibra-
9.3.3 Dependencies
tion or geographic assignment to additional
statistical analyses or Geographic Information IsoriX does not depend on any external pro-
System (GIS) systems to study the ecological grams or server-based computations because
determinants of intersample variation. For all processing in IsoriX is done on the com-
large isoscape projects, users can connect puter on which IsoriX is installed. This auton-
IsoriX to relational database management sys- omy makes IsoriX easy to install and to deploy
tems such as MariaDB, MySQL, Oracle, on any computer compatible with R. Like most
PostgreSQL, or SQLite to handle data more other R packages, IsoriX does rely internally
efficiently. Several R packages make these con- on functions provided by other R packages,
nections straightforward via the unified which themselves rely on other R packages
DataBase Interface for R implemented in the (Fig. 9.2). We will now introduce and discuss
package “DBI” (R Special Interest Group on three key packages on which IsoriX depends,
Databases (R-SIG-DB), Wickham, & Müller, because it facilitates the understanding of the
2017). Fourth, R users must write a script capacities of IsoriX.
defining exactly their workflow to use IsoriX. Many functions in IsoriX rely on the R pack-
The need to explicitly define the workflow by age “raster” (Hijmans, 2016). This package
a few lines of code is one of the key reasons to enables users to manipulate gridded spatial

numDeriv
nloptr
viridisLite
IsoriX
proxy
spaMM latticeExtra
sp RColorBrewer
raster
MASS rasterVis
Rcpp nlme
RcppEigen lattice
Depends
Imports Matrix
hexbin
LinkingTo
zoo
FIGURE 9.2 IsoriX dependencies on other R packages. This figure illustrates the direct and indirect dependencies of
IsoriX v0.8 to other R packages. It was drawn using the R package: “miniCRAN” (de Vries, 2017). The different colors rep-
resent different type of dependencies (full package dependencies, called Depends in R jargon, are show in blue; dependen-
cies to specific R functions from other packages, called Imports, are shown in red; dependencies to specific routines coded
in the C language, called LinkingTo, are shown in gray). For clarity, dependencies to so-called “base” packages and
optional dependencies are not shown.
data known as rasters. We use rasters in IsoriX means that IsoriX outputs can be used to pro-
to handle certain environmental predictors duce maps directly in R, or that maps pro-
that may be used to determine and fit an iso- duced can be easily exported into standard
scape (e.g., elevation or temperature data), to GIS file formats such as geotiffs, using the R
store the isoscape predictions and the variance packages “maptools” (Bivand & Lewin-Koh,
information associated with it, and to store the 2017) or “rgdal” (Bivand, Keitt, & Rowlingson,
outcomes of the geographic assignments. We 2017), and then loaded into a standalone GIS
chose to rely on the R package “raster” software such as QuantumGIS (free and open
because: (1) it is popular and unlikely to disap- source; www.qgis.org) or ESRI ArcGIS (com-
pear; (2) it is efficient to manipulate small ras- mercial license).
ters and rasters too large for the random Because not every migration ecologist is
access memory of a computer; (3) it is widely familiar with GIS systems, we coded our own
used and known by most ecologists using R as plotting methods to produce high-quality
a GIS. In practice, the dependence on “raster” maps directly in R. These plotting methods

9.3 THE ISORIX PACKAGE 213
rely on the package “rasterVis” (Perpiñán & isofit, isoscape, calibfit, and isofind.
Hijmans, 2016), allowing the display of rasters Respectively, these functions fit a pair of geos-
using the “lattice” package (Sarkar, 2008)—a tatistical models to the source data, use the fit-
popular derivative of the powerful visual envi- ted models to predict the isoscape and the
ronment “grid” available in R. (Another more complementary rasters storing variance infor-
recent popular derivative of “grid” is mation, fit the calibration model, and perform
“ggplot2” (Wickham, 2009).) Again, basing our geographic assignments of samples to their
graphic system on the R package “lattice” locations of origin based on (1) their isotopic
offers the possibility for an experienced user to compositions, (2) the predicted isoscape and
modify the maps generated by IsoriX, even complementary rasters, and (3) the fitted iso-
though our package already offers flexibility in tope calibration model. IsoriX also provides
the graphic production. additional functions such as functions to
Finally, IsoriX depends on the R package download large datasets of environmental data
“spaMM” (Rousset & Ferdy, 2014), which we or predictors (e.g., getelev), functions to pre-
use to estimate all of the parameters required pare datasets so they can be used by the core
to fit an isoscape, to describe the relationship functions (e.g., prepsources, prepraster), plot-
between the isotopic composition of the cali- ting methods (all accessible via the so-called
bration samples and that of the environment, “generic” function plot), display methods
and to perform the geographic assignments. (accessible via the generic functions print and
The package “spaMM” is designed to fit mixed summary), and other useful functions.
models of variable complexity using IsoriX comes with preloaded data. This
likelihood-based methods. This complexity includes datasets for testing IsoriX functions
goes from simple linear models to hierarchical (e.g., GNIPDataDE, CalibDataBat, and
generalized LMMs. We chose the R package AssignDataBat), as well as data to enhance the
“spaMM” because our fitting approach plots produced by IsoriX such as polygons that
requires more complex model structures than can be used to mask large bodies of water
those typically covered by, e.g., “lme4” (Bates (OceanMask), or polygons used to display bor-
et al., 2015). Besides, “spaMM” is developed ders between countries (CountryBorders). All
by one of us (F.R.). Therefore functionalities the datasets from IsoriX start with a capital-
developed in response to needs arising from ized noun, while functions are in lowercase.
IsoriX, but which may also be useful for other The documentation gives the source of pre-
mixed model applications are typically inte- loaded data and is important to read before
grated in “spaMM” and not in IsoriX, which using these data. The function documentation
keeps the source code of IsoriX focused around provides instructions on use, mostly from a
its main purpose. syntactic point of view and indicates what
inputs a given function needs, optional argu-
ments it may have, and what the default set-
tings of the function rely on. It is important to
9.3.4 Internal Structure become familiar with this information because
An R package is essentially a collection of the same function can exhibit different beha-
functions, data, and documentation. In IsoriX, viors depending on how the user sets the func-
these functions are divided into two broad cat- tion arguments.
egories: core functions and additional func- What the documentation does not explain
tions. The core functions are those doing the in detail is how the functions operate behind
hard work in IsoriX. There are only four: the curtain; which statistical assumptions they

are based on; the functions or settings that isoscape, e.g., an isotopic assignment of
should be used, depending on the data and unknown feathers using an isoscape derived
the research question; and what the workflow from feather samples of known origin (calibra-
should look like. This is the aim of the next tion samples). We chose, however, to limit this
section, and further details are found in other illustration to a situation we find is common:
publications (e.g., Courtiol & Rousset, 2017) the isoscape is derived from H isotope mea-
and online tutorials. surements of precipitation (usually based on
the Global Network of Isotopes in
Precipitation (GNIP) δ2H, Chapter 3: Isoscapes
9.4 THE ISORIX WORKFLOW for Terrestrial Migration Research) and the
researcher wants to infer the origin of the
In this section we illustrate how to imple- unknown tissues. As noted in the introduction,
ment a simple but complete isotope assign- the goal of the workflow is to identify whether
ment workflow in IsoriX (Fig. 9.3). IsoriX can bats found dead underneath wind turbines in
handle different types of assignment work- Germany were local bats or had migrated from
flows, e.g., the situation in which both the elsewhere, and if so, from where (see
source samples used to fit the isoscape and Chapter 5: Tracking of Movements of
assignment samples are from the same Terrestrial Mammals Using Stable Isotopes).
FIGURE 9.3 An example of a simple assignment workflow in IsoriX. This diagram represents the different stages of
the workflow described in this chapter. Note that IsoriX is flexible and can be used for different workflows. Function
names are displayed in blue.

9.4 THE ISORIX WORKFLOW 215
The workflow is divided into three main steps install the IsoriX package and dependencies by
that must be done sequentially: modeling of means of the R function install.packages:
the isoscape, fit of the calibration model, and
install.packages("IsoriX",
the geographic assignment of the unknown
dependencies 5 TRUE)
samples.
By illustrating and solving this bat assign- Second, each time after starting R, one must
ment problem with H isotopes, we detail the load the packages used for the IsoriX work-
main functionalities of IsoriX, but also discuss flow using the function library. For the dem-
the assumptions made in our approach. It is onstration workflow presented here, we need
important to discuss assumptions explicitly, to load four packages (which you may have to
because each step leading to an assignment install too if they are missing):
procedure (which may have important man-
library(IsoriX)
agement implications) is associated with
library(lattice)
assumptions about all different sources of
library(grid)
variation. There are indeed many sources of
library(raster)
variation that can impact the precision and
robustness of geographic assignments (see also
Chapter 2: Introduction to Conducting
Migration Studies, Chapter 5: Tracking of
9.4.2 Modeling the H Isoscape
Movements of Terrestrial Mammals Using The first step of our example workflow will
Stable Isotopes, Chapter 8: Design and be to fit the H isoscape. Recall that an isoscape
Analysis for Isotope-Based Studies of is a map representing the distribution of a
Migratory Animals). This includes isotope ana- stable isotope values over a given geographic
lytical and preparative error, inter- and intra- area (Chapter 3: Isoscapes for Terrestrial
sample isotopic variation, errors associated Migration Research). Note that in the context
with environmental predictors, the structure of of IsoriX, the predicted isoscape is associated
the statistical models, and the estimation of the with different measurements of the isotopic
parameters. Ideally, a geographic assignment variation around the predicted isoscape.
should account for and propagate all known
sources of isotopic variation. However, this
goal may never be achieved, since although 9.4.2.1 Preparing the Raw Source H
some sources of variation will be ultimately Isotope Data
considered during the geographic assignment, To model a H isoscape, we require mea-
others will be neglected for practical reasons, surements of source samples, i.e., the H isoto-
or may be negligible compared to others. pic composition of the environment at known
locations having explicit geographic coordi-
nates. Here, we use hydrogen isotope ratios
(δ2 H) in precipitation obtained from the GNIP
9.4.1 Preparing the R Environment for
(http://www-naweb.iaea.org/napc/ih/IHS_
the IsoriX Workflow resources_gnip.html), which is a rich source
Before demonstrating the IsoriX workflow, of global isotope data for building water-
the user needs to set up an R session in the fol- based H and O isoscapes. GNIP data are free
lowing way: First, after installing R (as to download after the registration on the
detailed on CRAN’s website), one should IAEA website.

Regardless of which source data are being source value: spatial variation is described by
used, the source data must be imported into R the first four columns of the table and the tem-
and stored as a data frame (the most common poral variation is described by columns year
type of object used to represent tables in R). and month. Note, however, that the require-
Tabular data stored as text files such as ment to provide a unique value for a given
comma-separated values files can be imported year-month combination at a given location
in R using the R function read.table. Reading implies that any variation in the isotope value
tables created in LibreOffice or Microsoft Excel occurring within a given month (for a given
is also possible by means of a dedicated pack- year and at a given sampling location) is
age such as “readODS” (Schutten, Chan, & neglected in our analyses, which is
Leeper, 2016), “readxl” (Wickham & Bryan, acceptable because organisms buffer fine-scale
2017), or “gdata” (Warnes et al., 2017). variation in their tissues through dietary aver-
The data frame created should present a aging (Vander Zanden et al., 2014, 2015).
structure similar to that shown in Table 9.1.
This implies that, as with the example dataset 9.4.2.2 Selecting and Aggregating the
used here (GNIPData), the data frame must have Source Isotope Data
at least six columns with the following infor- The next step requires selecting and aggre-
mation: sampling location (source_ID), latitude gating the isotope data that will be used to fit
(lat) and longitude (long) in decimal degrees, the required isoscape. This is done using the
year and the month of the sample, and the iso- function prepsources (#1 in Fig. 9.3). The role
tope delta value (source_value). The data frame of this function is to transform raw isotope
can also contain other relevant covariates data into a data frame (or data frames) that
needed to fit the data, such as elevation. In can be directly used for the fitting procedure
this data frame, for each sampling location, performed by the core function isofit.
each row must represent a unique year-month Aggregating the isotope data is necessary to
combination (e.g., for one location, there can achieve acceptable computing time for model
be no more than 24 rows for 2 years of obser- fitting in IsoriX.
vations). If several records are available for a As for other IsoriX functions, we detail
given year-month combination (e.g., multiple some key arguments influencing the behavior
records for January 1990), the user must thus of the function prepsources, but users should
summarize this information as a single number refer to the R documentation of the package
(e.g., by taking the mean) before using the data for more detail. Documentation is accessed
frame in IsoriX. from within R, by using the function help.
By preparing the data in Table 9.1, we retain Here, type help(prepsources) or its shortcut?
important sources of isotopic variation in prepsources to access documentation.
TABLE 9.1 GNIPData: a Data Frame Containing the H Isotope Composition of Environmental Water Sources
source_ID lat long elev year month source_value
NY ALESUND 78.91 11.93 7 1990 1 2 72.8
NY ALESUND 78.91 11.93 7 1990 2 2 67.6

NY ALESUND 78.91 11.93 7 1990 3 2 107.0
This table shows the first three rows of a larger data frame.

The function prepsources selects the relevant numerical vectors setting the months and years
isotope data by applying spatial and temporal to be included in the analysis. This argument
restrictions on the raw data. Spatial restriction can be used for eliminating data associated
is provided by four arguments (long_min, with, e.g., an outlier drought year. Note, how-
long_max, lat_min, and lat_max), which allow ever, that these arguments should be used
the user to define the appropriate maximum with caution because restricting months and
(rectangular) spatial area over which the iso- years under consideration decreases the
scape needs to be fitted. It is a “maximum” amount of source data being used, which can
area in the sense that the actual area depends result in a larger assignment uncertainty. Even
on the most extreme spatial locations of the for animals that are known to migrate during
source samples. a specific season, better accuracy is often
It is important to define the geographic area obtained by selecting yearly data instead of the
adequately. A practical difficulty is that the seemingly relevant ones (Vander Zanden et al.,
candidate area may not always be known or 2014). This is because the predicted isoscape
clearly identified. On the one hand, predefin- will be more precise but also because animals
ing a very small geographic area may result in integrate the isotopic composition of their
an uncertain isoscape due to the reduction in drinking water and diet during tissue growth
the amount of source isotope data and (Soto, Wassenaar, & Hobson, 2013). For these
increases the risk that the true location of reasons, we recommend sticking to the default
unknown origin falls outside the selected area. setting of IsoriX which considers source iso-
On the other hand, opting for a large geo- tope data from all months and years.
graphic area increases computational time, and The function prepsources aggregates the
may lead to a poor isoscape because the rela- selected data. The function allows the applica-
tionship between environmental predictors tion of different aggregation schemes, which
needed to fit the isoscape often differ in differ- influence the type of isoscape predictions gen-
ent geographic areas (Terzer, Wassenaar, erated. Commonly, aggregation operates on a
Araguás-Araguás, & Aggarwal, 2013). Also, if specific temporal scale and aims at summariz-
the calibration step is conducted using IsoriX, ing repeated measurements into a single
the user needs to include those locations in the δ-value mean and associated variance. We use
selected geographic area. Finding an adequate here the simplest aggregation scheme. This
and appropriate definition for the geographic scheme reduces all observations into one mean
area requires some practice and knowledge of isotope δ-value and one variance at each
available data, but again, because the work- source sampling location. In this example, the
flow in IsoriX is a simple text script, it is easy mean represents the average of the H isotope
to modify the spatial coordinates defining a values recorded at a given location, and the
geographic area and rerun the whole workflow variance represents the monthly measurement
to assess the influence of having different spa- variation (if available) during the entire
tial restrictions. IsoriX allows the user to quan- recording period. Because the recording period
tify the uncertainty associated with the may cover several years, the variance captures
predicted isoscape, which can provide further monthly fluctuations and fluctuations across
guidance about appropriate spatial restrictions. years. This aggregation scheme makes the
The function prepsources allows the user to assumption particularly explicit that a given
filter the isotope source data based on the tem- mean and variance are sufficient descriptors of
poral information via the arguments month and the temporal distribution of the isotopic com-
year. These arguments are provided as position at that location. Further modeling

steps will thus have to approximate such a dis- This call means that the function prepsources
tribution by the normal distribution. Note, takes our raw data (GNIPData), selects all
however, that other approaches to fit an iso- precipitation sampling locations within an
scape from nonaggregated data also make this area spanning from 30 degrees East to 60
assumption. degrees West and from 30 degrees North to 70
The more sophisticated aggregation degrees North, and aggregates them according
schemes proposed by the function allow one to to the default (simple) temporal aggregation
further disentangle within-year and between- scheme. The resulting data frame is stored in a
year variations, and to apply defined weight- new object which is here named GNIPDataEUagg
ing during the aggregation procedure. The (Table 9.2).
alternative aggregation schemes offer the pos- Users can generate the aggregated data
sibility to predict an annual isoscape that is, frame for the source isotope data by other
e.g., weighted by the monthly precipitation- means than using our function. No matter how
amounts. To obtain such predictions, monthly the table was built, the structure must be identi-
isotope measurements are weighted by the cal to the one illustrated in Table 9.2. In particu-
amount of precipitation water fallen during lar, each row must correspond to a unique
that month. Weighting can be done in two dif- location, and the columns of such a data frame
ferent ways in IsoriX; by directly weighting the must provide the information required to fit the
observations during the aggregation process models, i.e., it must include a name or code for
before the models are being fitted or by fitting the sampling location (source_ID), the mean iso-
a different isoscape for each month, and then tope δ-value (mean_source_value), variance in
weighting and aggregating the predicted iso- δ-values (var_source_Value), the number of
scapes to obtain a single one (IsoriX provides observations that have been aggregated
specific functions for this; see online tutorials (n_source_value), as well as the other covariates
for more details). that will be used in the models (i.e., latitude,
To select the source H isotope data over an longitude, and elevation). The name and type
area corresponding roughly to Europe and of the R objects behind each column must be
perform the simple temporal aggregation identical to those produced by the function pre-
scheme, we need to write the following line of psources, which is why we recommend nonex-
code in our R script: perienced users to use this IsoriX function. If
there is no repetition for a given location, the
GNIPDataEUagg ,- prepsources
variance will be missing (coded as “NA” in R).
(data 5 GNIPData, long_min 5 -30,
If not all variances are missing, IsoriX will run
long_max 5 60, lat_min 5 30, lat_max 5 70)
despite such missing information.
TABLE 9.2 GNIPDataEUagg: a Data Frame Containing the Aggregated Source H Isotope Data
source_ID mean_source_value var_vource_value n_source_value lat long elev
ADANA 2 23.86 273.27 336 36.98 35.30 73

AGUALVA 2 16.22 48.86 12 38.77 2 27.19 260
AL BAQAA 2 26.89 128.80 10 32.09 35.77 700
This table shows the first three rows of a data frame containing a total of 327 source locations.

9.4.2.3 Fitting the Isoscape elsewhere (Courtiol & Rousset, 2017). Briefly,
Once the aggregated dataset is completed, the function uses the package “spaMM”
an isoscape must be fitted, which is done by (Rousset & Ferdy, 2014) to fit both models
fitting two interrelated statistical models to the using the powerful statistical framework of
source isotope data (#2 in Fig. 9.3). These mod- mixed-effects models. In particular, “spaMM”
els are called the mean model and the residual models the spatial structure by means of the
dispersion model and we refer to the fitted mod- joint estimation of the fixed effects of environ-
els as the mean fit and the residual dispersion fit, mental covariates and of spatially independent
accordingly. and spatially autocorrelated random effects. In
The function isofit starts by fitting the simple terms, this means that while the model
residual dispersion model. This first model fits can account for the effect of environmental
describes how the variance in source isotope covariates on the variation in source isotope
values differs over the geographic area. values (or their variance), the model also cap-
Remember that under the aggregation scheme tures spatial structure not accounted for by
used, this variance represents the variation in these predictors. To model this spatial struc-
isotope values across months (spanning over ture, isofit uses the Matérn autocorrelation
several years). The residual dispersion model function. This autocorrelation function cap-
allows characterization of the extent to which tures the fact that measurements taken close
the isotopic composition of the precipitation together in space tend, on average, to be more
water varies in each location. Alternatives to similar to each other than measurements from
IsoriX often make the assumption that isotopic more distant areas. To what extent such mea-
variation is constant over the geographic area. surements are “more similar” is the question
However, this variation varies spatially, and addressed by “spaMM” during the autocorre-
considering this variation of the variation leads lation fitting procedure. The same type of
to a better fit of the source H isotope data modeling framework for the geographic auto-
(Courtiol & Rousset, 2017). Under the hood, the correlation is used in IsoMAP (Bowen, Liu,
residual dispersion model is a Gamma Vander Zanden, Zhao, & Takahashi, 2014;
Generalized Linear Mixed-effects Model Bowen et al., 2016), but the difference in IsoriX
(Gamma GLMM, see Appendix for details). is that autocorrelation is considered for the
The function isofit then fits the mean mean model and also for the residual disper-
model. This model describes how the mean sion model. IsoMAP assumes no residual vari-
source values vary geographically. The statisti- ation among source samples coming from the
cal definition of the mean model is a LMM. The same geographical location, although it may
only statistical twist is that the residual variance allow large isotopic variation among nearby
of this model is not estimated during the fit of samples — the so-called nugget effect; Cressie,
the mean model, but during the fit of the resid- 1993). In IsoriX we do not constrain the spatial
ual dispersion model. Therefore the residual patterns to be similar for both the mean and
variance of the mean model is defined by the the residual dispersion model.
predicted values of the residual dispersion fit. Despite the complexity concealed behind
The full statistical explanation of what iso- the isofit function, using this function to fit
fit does is beyond the scope of this chapter our isoscape is straightforward in practice. The
(see Appendix for a summary). Details and function takes as main arguments the name of
comparisons to other methods can be found the aggregated dataset, such as the one we

created earlier, and named lists of booleans The new R object created by isofit (here
(i.e., of TRUE or FALSE) for each model, indi- called EuropeFit) is a list of class ISOFIT. The
cating which environmental covariates to benefits of defining specific classes for R
include as fixed effects and which random objects is to conceal the complexity: one can
effects to consider during the fit. Specifically, use traditional functions such as plot, print,
the argument mean_model_fix specifies fixed and summary as R will automatically select the
effects to be considered for fitting the mean appropriate version of these functions for
model. The default is to consider no covariate, each class of object. In particular, a simple
but here we consider elevation and the abso- call to plot(EuropeFit) will display diagnos-
lute value for latitude, since those covariates tic plots for the fits. For advanced users, it is
offer good fit of H isotope values in Europe useful to know that the object EuropeFit con-
(Courtiol & Rousset, 2017). Similarly, disp_mo- tains two lists of importance: mean_fit and
del_fix specifies that fixed effects are consid- disp_fit. These lists are themselves of class
ered in the residual dispersion model and the HLfit (a class defined in the package
default, which we will use here, is to consider “spaMM”), which means that they corre-
no covariate. In the current implementation of spond to objects created by “spaMM” and
IsoriX, one could also consider elevation, abso- can be used as such. Those familiar with
lute latitude, squared latitude, longitude, and using “spaMM” or other statistical fitting
squared longitude in each model. In the future, function in R (e.g., lm, lmer, and gam) will be
other covariates may be added such as temper- familiar with how to extract details of fitted
ature and precipitation amount. For random models, such as the estimates, the likelihood
effects, the equivalent arguments are mean_mo- of the fits, or Akaike Information Criteria
del_rand and disp_model_rand. The default set- (AIC). The generic function AIC is tradition-
ting for random effects is to include for each ally used to compute the latter to perform
model one random effect capturing variation model selection. The AIC method imple-
between the sampling locations of the sources mented in “spaMM,” called, e.g., as AIC
that is independent of geographic position (EuropeFit$mean_fit) or AIC(EuropeFit$dis-
(e.g., the variation caused by different sample p_fit), provides the conditional AIC (cAIC;
processing) and one random effect with Vaida & Blanchard, 2005), which is more
Matérn autocorrelation. These random effects appropriate than the traditional AIC to select
are named uncorr and spatial, respectively. the most appropriate model for spatial pre-
We highly recommend estimating the Matérn diction (see Courtiol & Rousset, 2017) among
random effects for most applications, but not fits of different models to the same data by
estimating the random effects is useful to isofit (the smaller the cAIC value, the better
speed up computation time tremendously and the fit).
thus to produce a quick draft of a workflow Because fitting models takes seconds to
before fine-tuning it. This could be done by hours—depending on the data, structure of the
setting both mean_model_rand and disp_model_- model and the capability of the computer—it
rand to list(uncorr 5 FALSE, spatial 5 FALSE). is useful to save the fitted object to be able to
We can fit the geostatistical models by the reload it in the future without redoing the
following line of code: computation again. This is done using the R
functions save and load. Note, however, that
EuropeFit ,- isofit(data 5 GNIPDataEUagg,
we recommend refitting the models after each
mean_model_fix 5 list(elev 5 TRUE,
update of the package “IsoriX,” “spaMM,” or
lat_abs 5 TRUE))
of R itself.

9.4.2.4 Preparing the Structural Raster Multiresolution Terrain Elevation Data 2010
The model fits are used to generate predic- (GMTED2010) at 30 arc-seconds spatial resolu-
tions about the response values of the mean tion for the whole world (https://topotools.cr.
model (mean source isotope values) and of the usgs.gov/gmted_viewer/; higher resolutions
residual dispersion model (residual isotope can be downloaded from the website). Note that
dispersion variances). Technically speaking, the high resolution of this raster means this file
the fitted models could be used to predict the is large ( . 650 MB) even though we compressed
expected δ-values for any (real) value of the cov- it on our server. Importantly, we do not need to
ariates considered in the model through the download this file multiple times. It is sufficient
fixed and random effects. However, for such to store it at a known location on the user hard
predictions to be reliable, they are ideally per- drive.
formed within the range of observed values for Once the elevation raster has been down-
such covariates. In other words, extrapolation loaded, we have to import the file (here called
should be limited. Moreover, for predictions to gmted2010_30mn.tif) in R by means of the pack-
be usable in practice, predictions should be age “raster” (Hijmans, 2016). Assuming the file
drawn for a finite (not infinite) number of loca- is in our current working directory, and we
tions, which we call prediction locations. have loaded the package “raster” (see
IsoriX uses a structural raster object to define Section 9.4.1), the structural raster can be
both the geographic area that will be covered by imported and turned into an R raster object,
predictions and its resolution. Recall that a ras- which we can name ElevWorld:
ter is a georeferenced matrix providing data ElevWorld ,- raster("gmted2010_30mn.tif")
across on a regular spatial grid. The structural
raster is used to provide information about the Then, we use the IsoriX function prepraster
covariates that are needed to generate predic- to clip the raster to the appropriate spatial extent
tions, e.g., elevation, precipitation (a RasterBrick and to resize it to the appropriate resolution for
may be used as an alternative to the RasterLayer our bat study (#3 in Fig. 9.3). The clipping can
if the raster must contain multiple layers to store be controlled manually, but we included a use-
the covariates; RasterBrick and RasterLayer ful argument isofit handing over the fitted
being two classes from the package “raster”). models to the function prepraster so that the
To generate predictions for the isoscape, we range of latitude and longitude is automatically
need information about the geographic coordi- selected based on the extent of the aggregated
nates (i.e., latitude and longitude) and the sam- source isotope data (here: GNIPDataEUagg). By
ple site elevation. We thus need a raster with a default the clipping will select an area 5 % wider
single layer storing the elevation data as our than that covered by the source isotope data
structural raster. We call this the elevation raster. (this can also be adjusted if necessary using the
Because high-quality data rasters may be hard to argument margin_pct).
find, we included the function getelev in IsoriX Choosing how to resize the resolution is
which directly downloads a reliable elevation more subtle. The higher the resolution of the
raster from our server and stores it on the user raster, the better it is for the geographic assign-
hard drive (assuming an internet connection). To ment, but the more it consumes computer
do this, we call getelev() after setting the work- resources. Our advice is to start by using a
ing directory in R to the location where the file coarsely aggregated structural raster and imple-
will be saved (or one can use the path argument ment a first draft of the workflow to obtain
from the function getelev). The raster down- maps quickly and to later fine-tune the settings
loaded by our function is the Global for the display of these maps. Once the work-
flow is established, we recommend using a

structural raster with a higher resolution for the ElevEuropeFine ,- prepraster

final product. The reduction of the resolution of (raster 5 ElevWorld, isofit 5 EuropeFit,
the structural raster is controlled in prepraster aggregation_factor 5 4)
by two arguments: aggregation_factor and
A simple call to plot(ElevEuropeFine) makes
aggregation_fn. The first argument defines the
it possible to visualize the effect of the
number of adjacent raster cells (horizontally
aggregation.
and vertically) that are aggregated to form a
new cell of lower resolution. The second defines
the function used for such an aggregation, with 9.4.2.5 Predicting the H Isoscape
the default function being the mean. Once the models are fitted and the struc-
The aggregation factor and the aggregation tural raster is prepared, we can finally proceed
function can both impact the geographic to predict the H isoscape using the function
assignments at the end of the workflow. isoscape (#4 in Fig. 9.3). Here, the only argu-
Importantly, even the original elevation rasters ments to provide are the structural raster and
provided by GMTED2010 have themselves the fitted models:
undergone an aggregation procedure which
EuropeIsoscape ,- isoscape
influences the assignment (it is possible to
(raster 5 ElevEuropeFine,
download from their website elevation rasters
isofit 5 EuropeFit)
aggregated with other functions than the
mean). Imagine, with an extreme example, that The function isoscape performs a prediction
we coarsely average the elevation of a moun- based on the fitted models, using the data con-
tain range (e.g., the Alps) into one single raster tained in the structural raster (here, the latitude,
cell. The aggregation function is the function longitude, and elevation). One important
applied to all the cells that will be combined assumption made during the fitting and the pre-
into one. If we took the mean value (default diction procedure is that the covariates are with-
choice), we would end up with a raster cell out error. This implies that latitude, longitude,
that will have a value close to the average ele- and elevation are assumed to be exact. While it
vation of the mountain chain. But if instead we is fine to consider coordinates as exact (at least
used the function maximum, the cell would for predictions), errors in the estimation of eleva-
have a value close to the maximal elevation, tions are neglected. For the elevation raster used
i.e., the highest summit of the Alps. Whichever here, the official documentation of GMTED2010
function is most adequate for the geographic states that such errors are expected to be low on
assignment strongly depends on where the average (root mean square error varies between
samples we aim to assign originate from. If the 25 and 42 m, when the elevation raster is used at
samples correspond to migratory organisms its native resolution of 30 arc-seconds and not
living all over mountain slopes, then the mean aggregated further).
is the better choice, but if they live on moun- The output of the function isoscape is an
tain tops, then the maximum will be best. The object of class ISOSCAPE, which is a list containing
higher the resolution, the less the aggregation all the information required to map the isoscape
function will make a difference. predictions and the complementary rasters. The
For the bat study, we choose an aggregation object EuropeIsoscape is composed of a set of ras-
factor of 4 and the default aggregation function ters (a RasterBrick) stored in the element of the
(i.e., the mean) to generate our structural ras- list called isoscape, and by the spatial points cor-
ter, which we call ElevEuropeFine. This can be responding to the locations of the sources stored
done with the following code: in an element called sp_points. The spatial

points are included in the object so that the plot- sources, country borders, or to mask large
ting can be based on the single object returned water bodies. The simplest use of the plotting
by the function isoscape. function is to just type plot(EuropeIsoscape),
The element isoscape (i.e., here but the plot obtained would be quite different
EuropeIsoscape$isoscape) stores several rasters from Fig. 9.1. To improve the figure, we
of class RasterLayer combined into a single instead used the following call:
object of class RasterBrick. Three of these rasters
plot(x 5 EuropeIsoscape, sources 5 list
are of particular interest: mean, mean_predVar, and
(draw 5 FALSE), mask 5 list
mean_residVar. The raster called mean contains
(fill 5 "lightgrey"), palette 5 list
point predictions for the mean isotope source
(range 5 c(-130, 10), step 5 10,
values. This is the predicted H isoscape. The ras-
n_labels 5 15, fn 5 viridisLite::viridis))
ter mean_predVar contains the prediction variance
associated with each point prediction This example only shows a sample of the plot
(Fig. 9.4A). This variance directly quantifies the settings that can be changed. Here, the code
isotope uncertainty of the predicted isoscape. shows that we decided not to draw the locations
This uncertainty typically decreases in areas that of the sources, we used the default mask which
are dense in source data and increases where hides large water bodies, but we changed the
data is sparse. The raster mean_residVar contains filling color to light gray, that we used the color
the residual variance associated with the isotope scale provided by the palette viridis from the R
source values (Fig. 9.4B). This variance is an esti- package “viridisLite” (automatically installed
mation of the observed temporal variability in during the installation of IsoriX), and we chose
measurements at each prediction location. This to constrain the beginning and ending of the
residual variance exists because there is natural δ-scale between 2130 and 110 per mil, to
variation in the isotopic composition of precipi- change color every 10 per mil, and to draw 15
tation at a given location (see Section 9.4.2.2). labels matching the color changes.
Each of these rasters can be accessed using their Users familiar with "grid" and "lattice" R
name. For example, the raster storing the iso- graphic environments can further refine
scape can be retrieved by typing EuropeIsoscape these plots. As one illustration, we created
$isoscape$mean. the panel in Fig. 9.4 by loading, as mentioned
in Section 9.4.1, the packages “grid”
9.4.2.6 Plotting the Predicted H Isoscape (included in any installation of R) and “lat-
and Associated Rasters tice” (installed with IsoriX) and running the
To map a given raster from the output of following code:
the function isoscape, we implemented a plot-
PlotMeanPredVar ,- plot(x 5 EuropeIsoscape,
ting method accessible by calling the generic
which 5 "mean_predVar", plot 5 FALSE)
function plot. This method should be provided
PlotMeanResidVar ,- plot
with the entire object created by the function
(x 5 EuropeIsoscape,
isoscape and with the name of the raster to be
which 5 "mean_residVar", plot 5 FALSE)
plotted. If no raster name is provided, the
print(PlotMeanPredVar, split 5 c(1, 1, 1, 2),
default is to plot the raster mean, i.e., the pre-
more 5 TRUE)
dicted H isoscape. Other arguments can be
defined to fine-tune the outcome of the predic- print(PlotMeanResidVar, split 5 c(1, 2, 1, 2))
tive isoscape map produced. One can, e.g.,
grid.text(c("A", "B"), x 5 0, y 5 c(0.99,
define the color scale or choose how to display
0.5), hjust 5 -1, gp 5 gpar(cex 5 1.5))
(or not display) the locations of the sample

FIGURE 9.4 Maps showing the prediction variance (A) and the residual dispersion variance (B) of the predicted H iso-
scape. The prediction variance quantifies the uncertainty of the predictions shown in Fig. 9.1. The residual dispersion vari-
ance quantifies here the between month variation in H isotopic composition. Red triangles indicate the sampling locations
of the GNIP sources.

Plotting rasters other than the one corre- 9.4.3.1 Preparing the Data for the
sponding to the traditional predicted isoscape Calibration
(i.e., the raster mean) greatly benefits assignment Obtaining good data for the calibration fit is
workflows. In particular, a map of the mean critical for getting accurate and precise geo-
prediction variance mean_predVar (Fig. 9.4A) graphic assignments (Chapter 5: Tracking of
directly depicts the uncertainty of the predicted Movements of Terrestrial Mammals Using
isoscape. This allows users to diagnose poorly Stable Isotopes). Ideally, we would prefer
fitted models or evaluate inadequate datasets. It species-specific isotope data from many cali-
also informs where additional data sources bration samples covering the entire potential
should be sampled in order to increase the reli- assignment area and thus spanning the whole
ability of the isoscape predictions, and thus the range of each covariate considered in the geos-
geographic assignment. The mean residual vari- tatistical models and consequently inhabiting
ance mean_residVar (Fig. 9.4B) shows where iso- isotopically different locations (Chapter 8:
tope values vary the most/least over time (any Design and Analysis for Isotope-Based Studies
temporal scale defined by the aggregation proof Migratory Animals). This would reduce
cess, see Section 9.4.2.2). risks associated with extrapolation when using
the fitted calibration model during the geo-
graphic assignment. We also prefer many
9.4.3 Fitting an Isotope Calibration
repeated measurements per location to reliably
Model estimate the intersample variance in isotopic
After predicting the isoscape, the second composition, which captures the effect of tem-
step of the workflow is to characterize the rela- poral variation in source isotope values at each
tionship between the isotopic composition of site. We also need the timing of the sampling
the environment (i.e., the source H isotope events for these calibration samples to match
data) and the isotopic composition of the sam- the overall sampling period of the source iso-
ples we want to assign (our bat samples). For tope data. We would prefer to use samples
this, we need H isotope measurements of bats comparable to those that need to be assigned
of known origin (i.e., the calibration samples). such as organisms from the same species and
We also need the H isotopic composition of the same physiological state, since the relation-
the environment at the locations where the bat ship between the isotopic composition of an
calibration samples are being collected. For the organism and that of its environment is known
latter, the common practice is to directly infer to be influenced by such factors (e.g.,
source H isotope data from the predicted iso- Gorokhova, 2017; Soto, Hobson, & Wassenaar,
scape, which is what IsoriX does by default. 2016; Voigt, Lehmann, & Greif, 2015). In prac-
With calibration data at hand, we can proceed tice, it is exceedingly rare (and costly) to
to fit a statistical model, which will be used meet all these requirements, but one should
during the geographic assignment to convert attempt to get as close as possible to the ideal
the isotope values of the assignment samples situation described.
into their source isotope value equivalent. We We use here a dataset called CalibDataBat,
call this statistical model the calibration model which contains stable H isotope ratios of the
and the fitted model the calibration fit. In other nonexchangeable hydrogen in fur keratin of
publications, the latter is also referred to as the nonmigratory bats sampled across Europe
isotope calibration function, isotope calibration (Table 9.3). The calibration dataset must con-
curve, or isotope transfer function. tain at least a column site_ID of class factor

TABLE 9.3 CalibDataBat: a Data Frame Containing small compared to natural variation, thus the
the Calibration Data variance between technical replicates can usu-
site_ID long lat elev sample_ID sample_value ally be neglected (Soto, Koehler, Wassenaar, &
Hobson, 2017, but see Chapter 2), Introduction
site_41 18.87 46.20 90 1 2 71.8 to Conducting Stable Isotope Measurements
site_41 18.87 46.20 90 2 2 103.4 for Animal Migration Studies.
site_41 18.87 46.20 90 3 2 68.9
9.4.3.2 Fitting the Calibration Model
This table shows the first three rows of a data frame containing a
total of 335 calibration samples. We designed the function calibfit to fit the
calibration model (#5 in Fig. 9.3). This function
takes two arguments: the geostatistical models
TABLE 9.4 AssignDataBat: a Data Frame Containing fitted by isofit and the calibration dataset.
the Assignment Data With our bat examples this gives:
sample_ID lat long sample_value CalibBats ,- calibfit(data 5 CalibDataBat,
isofit 5 EuropeFit)
Nnoc_1 51.72 13.68 2 86.6
Nnoc_2 51.72 13.68 2 94.0 The object CalibBats created here is a list of
class CALIBFIT, with its own printing and plot-
Nnoc_3 51.72 13.68 2 88.7
ting methods. A simple call to plot(CalibBats)
This table shows the first three rows of a data frame containing a thus produces a plot allowing one to check the
total of 14 assignment samples. fit obtained (Fig. 9.5).
The calibration model that is fitted is a par-
ticular LMM (see Appendix). The particularity
describing the sampling site, the columns con- lies in the calibration model controlling for the
taining all covariates required for the geostatis- uncertainty of the isoscape predictions. The
tical models to predict the isoscape values at function calibfit not only considers the pre-
these locations (here long, lat, and elev), and a
diction variance associated with each predic-
column sample_value containing the measured
tion from the mean model (which varies
H isotope values of the calibration samples.
spatially) but also the prediction covariances
An important constraint to keep in mind is between the different predictions of the isotope
that only measurements on multiple calibra- values across the isoscape. Moreover, the vari-
tion samples per location capture the variation ation between calibration samples is captured
between different samples exposed to the by the residual variance of the calibration fit.
same isotope environment. Therefore IsoriX A strong assumption made while fitting the
needs multiple calibration samples for several
calibration model is that the relationship
(not necessarily all) sampling locations. That
between the isotope values of the calibration
means that some levels of site_ID need to be
samples and those of the environment are line-
repeated on several rows. If laboratory mea- arly related.
surements are replicated, we recommend to
consider the mean measurement value for
each sampling unit (here for each bat) instead
9.4.4 Assigning Samples to Location of
of considering multiple rows for a given cali-
bration sample in order to avoid confounding
Origin
biological replicates and technical ones. In gen- The third and final step of the workflow is
eral, measurement errors for H isotopes are to perform the geographic assignment. For

FIGURE 9.5 Calibration fit. Points represent the bat
–40 H isotope calibration data. The continuous line represents
the regression line and the dashed lines provide the 95 %
Isotopic value in the calibration sample
confidence interval of this regression line. The uncer-

tainty associated with the source isotope values (x-axis)
–60
is not represented but it is accounted for during the fit-
ting procedure.
–80
–100
–120
–140
–90 –80 –70 –60 –50 –40 –30
Predicted isotopic value in the environment
this, we use the predicted isoscape and com- information is only considered for plotting.
plementary rasters constructed during the first Here we are going to assign a single group of
step, the calibration fit obtained during the bats whose data are given in Table 9.4.
second step, and the isotope data of the assign-
ment samples. 9.4.4.2 Running the Geographic
Assignments
9.4.4.1 Preparing the Data for the Running the geographic assignment is the
Assignment task of the function isofind (#6 in Fig. 9.3).
At least two pieces of information are This function needs the isoscape object (con-
required from the bat samples that need to be taining all the rasters produced by the function
assigned: a label and its H isotope value. isoscape), the calibration fit and the data frame
Irrespective of the number of samples to with the data to assign. The call to isofind
assign, this information should be provided as should therefore resemble the following:
a data frame with the column names sample_ID
AssignedBats2 ,- isofind
and sample_value. The assignment function
(data5AssignDataBat2,
isofind can be used to perform two types of
isoscape 5 EuropeIsoscape,
geographic assignments: an individual assign-
calibfit 5 CalibFit)
ment and a group assignment for all samples
in the data frame. (If one wishes to assign dif- The output of the function, here stored
ferent groups of samples separately, one can under the name AssignedBats, is a list of class
prepare several data frames and run the ISOFIND containing three elements: the element
assignment function on each of the groups sample storing all rasters related to the assign-
independently.) The dataset for the geographic ment of each single sample, group storing the
assignment can also contain the latitude and rasters related to the group assignment, and
longitude of the sampling sites but this sp_points storing the spatial points of the

source locations as well as the ones for the cali- against an assignment at the corresponding
bration samples and the assignment samples location, while a large difference where the
(if available). As for the output of the function prediction variance is very large may not. The
isoscape, storing all this information into a sin- variance of the test also depends on (2) the
gle R object produced by isofind allows one to residual variation among samples at the
plot detailed assignment maps easily. unknown location of origin. In the current
First, isofind runs the assignment test for workflow, we estimate this variance from the
each sample at each candidate location or cell residual variance of the calibration fit, which
from our structural raster. Specifically, for we assume is constant. Other approaches are
each candidate location we test whether a possible. If assignment samples are of the
given assignment sample comes from this loca- same kind as the source sample (and thus
tion, and therefore whether the isoscape value when no calibration model exists), the variance
at its unknown location of origin is consistent is instead given by the residual dispersion var-
with the predicted isoscape value at the candi- iance of the isoscape. The distribution of the
date location. We use the difference between assignment test also depends on (3) the predic-
these two isoscape values as test statistics. To tion variance of the calibration fit (which
compute the test statistics, isofind predicts the decreases with the size of the calibration data-
isoscape value of the assignment sample at its set) given the predicted isoscape value at the
unknown location of origin from the isotope candidate location. Finally, the distribution of
values of the sample by inverting the calibra- the assignment test depends on (4) a covari-
tion relationship (see Appendix). The pre- ance term between the prediction error of the
dicted isoscape value of the candidate location calibration fit at the candidate location, and
is directly the point prediction of the isotope the prediction error of the isoscape at this loca-
value at each prediction location. The test station. This last term relates the uncertainty of
tistics for all locations can be retrieved for each the isoscape to one of the calibration fit. This
sample from the object produced by isofind. covariance is neglected in other approaches. In
For example, the raster stored under the name IsoriX, we can consider it because its computa-
AssignedBats$sample$stat$Nnoc_1 contains such tion rests on the returned fit objects of both the
information for the first bat. calibration fit and the mean fit, which there-
Under the null hypothesis (the location of fore need to be jointly available in computer
origin 5 the candidate location), the variance memory.
of the test statistic is the variance of the differ- For each candidate location of origin, the
ence between the two isoscape values that P-value of the assignment test is also stored in
defines it. The variance of the test statistics is the object produced by isofind (e.g.,
also stored as a raster (e.g., AssignedBats$sam- AssignedBats$sample$pv$Nnoc_1). The P-value of
ple$stat_var$Nnoc_1). the test is the probability that an assignment
By relating the isoscape values to the sample originating from the candidate location
unknown isotope value that the mean fit esti- has a larger absolute value than the observed
mates (see Appendix), one finds that the vari- test statistic. Even if a location presents a perfect
ance of the test statistic depends on (1) the match (P-value close to one), it does not mean
prediction variance of the mean fit and is that the location is the correct place of origin. It
therefore not constant over the entire spatial only means that the candidate location has an
range (see Appendix). Therefore a small differ- isotopic composition like the one of its true ori-
ence in isoscape values at a location where the gin, irrespective of where this location may be.
prediction variance is very small may argue This limitation is of course not specific to IsoriX.

The P-value raster associated with each assign- between the two plotting methods are the
ment sample can therefore be used to directly same. One main difference, however, is the
extract the result of the test of an assignment argument which is used to indicate which ras-
hypothesis. For example, to test whether the first ter to plot (mean, mean_predVar, . . .) is now
bat to assign (Nnoc_1) originates from its sam- replaced by an argument who. If who is set to
ple location we can simply type: group, the function will show a single map
representing the geographic assignment for the
extract(AssignedBats$sample$pv[[1]], cbind
group. Otherwise, to obtain a single assign-
(AssignDataBat$long[1], AssignDataBat$lat
ment map per sample, one can provide to who
[1]))
a vector of the samples to be plotted (as indi-
Note that we use here the function extract ces or using their ID). Specifically, we obtained
from the package “raster.” The outcome of this the geographic assignments displayed in
call gives a nonsignificant P-value (P 5 .99), so Fig. 9.6 as follows:
we cannot rule out a local origin for the first
plot(AssignedBats, who 5 1:14, sources 5 list
bat. Repeating this procedure for all unknown
(draw 5 FALSE), calibs 5 list(draw 5 FALSE))
bat assignment samples shows that we can
reject a local origin only for bat Nnoc_15. The probability maps obtained confirm the
Once the function isofind has assigned all examination of the P-value rasters, with the
samples, it also performs a group assignment. sampling location (eastern Germany) not being
The P-value of the geographic assignment test rejected as a source of origin for all but one of
for the entire group (here stored as the raster the bats (Nnoc_15). This is true even though
AssignedBats$group$pv) is computed by combin- the highest P-values for these potentially sed-
ing, at each location, the P-value of the geo- entary individuals are outside of Germany
graphic assignment for all assignment samples (e.g., Nnoc_2). The plotting method can use
using Fisher’s combination of probabilities the information from the assignment test to
method (Fisher, 1925). The null hypothesis of define confidence regions with any given level
this group assignment is that all assignment L, that can be visualized on an assignment
samples originated from a location with identi- map (see Section 9.5), and that should contain
cal distribution of source isotope values as the the true location of origin with the given prob-
one being tested. As a consequence, the test ability L. By default, a 95 %-confidence region
considers that all assignment samples come is displayed in gray as shown on Fig. 9.6.
from the same location of origin (or at least Another default setting is to apply a mask to
from a location presenting the same isotopic hide the outcome of the geographic assign-
composition), and if this assumption is not ful- ment over the large water bodies. One can use
filled, the test may exclude all candidate loca- a second mask to hide areas using the argu-
tions. Therefore, we recommend always first ment mask2. For instance, this can be used to
exploring individual assignments even for hide areas outside the IUCN known distribu-
users only interested in group assignments. tion map for the respective species. If more
than two masks are necessary, one can either
9.4.4.3 Plotting the Geographic combine several masks into one and provide it
Assignments to this argument mask2, or if the masks need to
Geographic assignments can be plotted by be visually distinct, one can use the function
calling the generic plot function which invokes layer form the R package “latticeExtra”
a plotting method like the one described for (Sarkar & Andrews, 2016; see IsoriX online
isoscapes (see Section 9.4.2.6). Most arguments documentation for examples).

FIGURE 9.6 Probability maps of the geographic assignment for each of the 14 bats found dead at the bottom of wind
turbines in Germany. Probabilities depict the P-values of the assignment test for each assignment sample.
To make a bat group assignment, we again 9.5 THE FUTURE OF ISORIX

perform the geographic assignment, but this
time without that individual that is likely to In future versions of the IsoriX package, we
originate from outside Germany (Nnoc_15). plan to overcome limitations of the current
We display the assignment for all the remain- version. Future versions of IsoriX should be
ing bats as a group (Fig. 9.7). This can be done more comprehensive, faster, and safer.
by typing: We plan to add new functionalities to exist-
AssignDataBat2 ,- subset(AssignDataBat,
ing functions and create new functions as
sample_ID ! 5 "Nnoc_15")
needed to make IsoriX more comprehensive.
AssignedBats2 ,- isofind
For example, we plan to allow the residual
(data 5 AssignDataBat2,
variance of the calibration fit to vary geo-
isoscape 5 EuropeIsoscape,
graphically as dictated by the isoscape. We
calibfit 5 CalibBats)
also plan to allow geographic assignments
plot(AssignedBats, who 5 "group")
using isoscapes and/or calibration fitted

9.5 THE FUTURE OF ISORIX 231
FIGURE 9.7 Probability maps of the geographic assignment for the bats including all bats shown in Fig. 9.6 except bat
Nnoc_15 Red triangles indicate sampling locations of the sources. Blue crosses indicate the sampling locations of the calibra-
tion samples. White diamonds indicate the sampling locations of the assignment samples.
outside IsoriX. Yet, relying on published iso- hence we expect speed improvements. Further,
tope calibration or transfer functions without we will implement several functions, including
having access to the raw data imposes severe isoscape, so that they can benefit from parallel
restrictions (e.g., it prohibits the prediction computing since most computers now possess
variances and covariances of the underlying several central processing units and R can han-
predicted isoscape to be used while fitting the dle this.
calibration model). This is why we will Because we are progressively increasing the
include some published key calibration data- number of checks that functions perform on
sets within IsoriX (if authors granted access to the user inputs, detecting common input mis-
the raw data), so that users can refit the pub- takes and returning meaningful error messages
lished calibration model using their isoscapes or warning messages particularly benefit users
fitted in IsoriX. taking their first steps in the R environment
The package “spaMM” on which IsoriX for isoscape assignments. We will keep
relies for heavy computation keeps making improving and extending online documenta-
good progress in terms of computation speed, tion by providing tutorials reflecting different

possible applications of IsoriX. In particular, of the international summer school on stable isotopes in
IsoriX was conceived with migratory terrestrial animal ecology at IZW for their feedback on IsoriX. We
thank the stable isotope laboratory of the IZW for analyz-
animals in mind, but it could be used to ing samples. We thank Carlos Alberto and Myles Menz for
study other organisms (e.g., plants or marine stimulating new applications for IsoriX. We also thank
organisms) and each application rests on sub- Deepayan Sarkar for guidance on programming with the R
tle differences in the kind of data, plots and package “lattice.” We thank Stefan Terzer for providing
model structures needed. We therefore would the core GNIP H data used in this chapter. Finally, we
thank Sinah Drenske for helping with the conversion of
like to provide guidance for these other appli- our LaTeX manuscript into a *.docx file.
cations as well.
As IsoriX improves, we encourage users to
regularly check for the release of updates of
IsoriX and of the R packages it depends on, References
and to install these updates. This can be done Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015).
automatically within R using the function Fitting linear mixed-effects models using lme4. Journal
update.packages. All changes made on IsoriX of Statistical Software, 67(1), 148. Available from
are made in a transparent fashion. Each time https://doi.org/10.18637/jss.v067.i01.
Beckerman, A., Childs, D., & Petchey, O. (2017). Getting started
the package is updated, information about sig- with R. Oxford University Press. Available from https://
nificant changes can be obtained by the com- doi.org/10.1093/acprof:oso/9780198787839.001.0001.
mand news(package 5 “IsoriX”). All changes Bivand, R., Keitt, T., & Rowlingson, B. (2017). rgdal: Bindings
can be tracked in detail using the system git for the “Geospatial” data abstraction library. https://cran.r-
on GitHub, and since CRAN archives all ver- project.org/package 5 rgdal.
Bivand, R., & Lewin-Koh, N. (2017). maptools: Tools for read-
sions of each package, it is possible to use a ing and handling spatial objects. https://cran.r-project.
file comparison tool to compare source files org/package 5 maptools.
from different release versions. Bivand, R. S., Pebesma, E., & Gómez-Rubio, V. (2013).
Improving an R package requires time and Applied spatial data analysis with R. New York: Springer.
energy, but since this package has been pro- Available from https://doi.org/10.1007/978-1-4614-
7618-4.
grammed within the R environment anyone Bowen, G. J., Liu, Z., Vander Zanden, H. B., Zhao, L., &
can contribute to the IsoriX project: not only Takahashi, G. (2014). Geographic assignment with
by writing new code or editing existing one, stable isotopes in IsoMAP. Methods in Ecology and
but also by contributing to the documenta- Evolution, 5(3), 201206. Available from https://doi.
tion, reporting potential bugs, or requesting org/10.1111/2041-210x.12147.
Bowen, G. J., Putman, A., Oerter, E., Vander Zanden, H. B.,
additional features. We believe that anyone Woo, J., Wunder, M., & Zhao, L. (2016). Exploring physi-
can improve IsoriX, irrespective of their cal and biological connections using IsoMAP. Geological
levels of expertise, and we encourage users Society of America. Available from https://doi.org/
to do so. To coordinate collective effort, we 10.1130/abs/2016am-286274.
are hosting the developmental version of Chambers, J. M. (2016). Extending R. Boca Raton, FL:
Chapman & Hall.
future IsoriX releases on the development Courtiol, A., & Rousset, F. (2017). Modelling isoscapes
platform GitHub (https://github.com/cour- using mixed models. https://doi.org/10.1101/
tiol/IsoriX). 207662.
Courtiol, A., Rousset, F., Rohwaeder, M.-S., & Kramer-
Schadt, S. (2018). IsoriX: Isoscape computation and inference
Acknowledgments of spatial origins using mixed models. https://cran.r-project.
org/package 5 IsoriX.
We are thankful to our beta testers (Vojtěch Brlı́k, Edgars Cressie, N. A. C. (1993). Statistics for spatial data. New York:
Lediņš, Marco Thoma, and Peter de Vries) and to students Wiley.

REFERENCES 233
Davison, A. C. (2003). Statistical models. Cambridge: Soto, D. X., Wassenaar, L. I., & Hobson, K. A. (2013).
Cambridge University Press. Available from https:// Stable hydrogen and oxygen isotopes in aquatic food
doi.org/10.1017/CBO9780511815850. webs are tracers of diet and provenance. Functional
Diggle, P. J., Ribeiro, P. J., & Christensen, O. F. (2003). An Ecology, 27(2), 535543. Available from https://doi.
introduction to model-based geostatistics. In J. Møller org/10.1111/1365-2435.12054.
(Ed.), Spatial statistics and computational methods Stock, B. C., & Semmens, B. X. (2013). MixSIAR GUI user
(pp. 4386). New York: Springer Verlag. manual. https://doi.org/10.5281/zenodo.56159.
Ferguson, J., & Hopkins, J. (2016). IsotopeR: Stable isotope mix- Terzer, S., Wassenaar, L. I., Araguás-Araguás, L. J., &
ing model. https://cran.r-project.org/package 5 IsotopeR. Aggarwal, P. K. (2013). Global isoscapes for delta18O
Fisher, R. A. (1925). Statistical methods for research work- and delta2H in precipitation: Improved prediction
ers. Oliver and Boyd, Edinburgh. using regionalized climatic regression models.
Gorokhova, E. (2017). Individual growth as a non-dietary Hydrology and Earth System Sciences Discussions, 10(6),
determinant of the isotopic niche metrics. Methods in 73517393. Available from https://doi.org/10.5194/
Ecology and Evolution. Available from https://doi.org/ hessd-10-7351-2013.
10.1111/2041-210X.12887. Vaida, F., & Blanchard, S. (2005). Conditional Akaike infor-
Hijmans, R. J. (2016). raster: Geographic data analysis and mation for mixed-effects models. Biometrika, 92(2),
modeling. https://cran.r-project.org/package 5 raster. 351370. Available from https://doi.org/10.1093/bio-
Parnell, A. (2016). simmr: A stable isotope mixing model. met/92.2.351.
https://cran.r-project.org/package 5 simmr. Vander Zanden, H. B., Wunder, M. B., Hobson, K. A.,
Perpiñán, O., & Hijmans, R. (2016). rasterVis. http://oscar- Van Wilgenburg, S. L., Wassenaar, L. I., Welker, J. M.,
perpinan.github.io/rastervis/. & Bowen, G. J. (2014). Contrasting assignment of
R Core Team. (2017). R: A language and environment for sta- migratory organisms to geographic origins using long-
tistical computing. Vienna. https://www.r-project.org/. term versus year-specific precipitation isotope maps.
R Special Interest Group on Databases (R-SIG-DB), Methods in Ecology and Evolution, 5(9), 891900.
Wickham, H., & Müller, K. (2017). DBI: R database inter- Available from https://doi.org/10.1111/2041-
face. https://cran.r-project.org/package 5 DBI. 210X.12229.
Rexer Analytics. (2015). Data Science Survey Rexer analytics. Vander Zanden, H. B., Wunder, M. B., Hobson, K. A., Van
http://www.rexeranalytics.com/data-science-survey.html. Wilgenburg, S. L., Wassenaar, L. I., Welker, J. M., &
Rousset, F., & Ferdy, J.-B. (2014). Testing environmental and Bowen, G. J. (2015). Space-time tradeoffs in the devel-
genetic effects in the presence of spatial autocorrelation. opment of precipitation-based isoscape models for
Ecography, 37(8), 781790. Available from http://dx.doi. determining migratory origin. Journal of Avian Biology,
org/10.1111/ecog.00566. 46(6), 658667. Available from https://doi.org/
Sarkar, D. (2008). Lattice: Multivariate data visualization with 10.1111/jav.00656.
R. New York: Springer. Available from http://lmdvr.r- Venables, W. N., & Ripley, B. D. (2002). Modern applied sta-
forge.r-project.org. tistics with S. New York: Springer. Available from
Sarkar, D., & Andrews, F. (2016). latticeExtra: Extra graphi- https://doi.org/10.1007/978-0-387-21706-2.
cal utilities based on lattice. https://cran.r-project.org/ Voigt, C. C., Lehmann, D., & Greif, S. (2015). Stable isotope
package 5 latticeExtra. ratios of hydrogen separate mammals of aquatic and
Schutten, G.-J., Chan, C., & Leeper, T. J. (2016). readODS: terrestrial food webs. Methods in Ecology and Evolution,
Read and write ODS files. https://cran.r-project.org/ 6(11), 13321340. Available from https://doi.org/
package 5 readODS. 10.1111/2041-210x.12414.
Soto, D. X., Hobson, K. A., & Wassenaar, L. I. (2016). de Vries, A. (2017). miniCRAN: Create a mini version of
Using hydrogen isotopes of freshwater fish tissue as a CRAN containing only selected packages. https://cran.r-
tracer of provenance. Ecology and Evolution, 6(21), project.org/package 5 miniCRAN.
77767782. Available from https://doi.org/10.1002/ Warnes, G. R., Bolker, B., Gorjanc, G., Grothendieck, G.,
ece3.2519. Korosec, A., Lumley, T., et al. (2017). gdata: Various R
Soto, D. X., Koehler, G., Wassenaar, L. I., & Hobson, K. A. programming tools for data manipulation. https://cran.r-
(2017). Re-evaluation of the hydrogen stable isotopic project.org/package 5 gdata.
composition of keratin calibration standards for wildlife Wickham, H. (2009). ggplot2: Elegant graphics for data analy-
and forensic science applications. Rapid Communications sis. Springer-Verlag, New York. http://ggplot2.org.
in Mass Spectrometry, 31(14), 11931203. Available from Wickham, H., & Bryan, J. (2017). readxl: Read excel files.
https://doi.org/10.1002/rcm.7893. https://cran.r-project.org/package 5 readxl.

APPENDIX: STATISTICAL function. This function has two parameters:

FRAMEWORK the spatial scale parameter ρ and the smooth-
ness parameter ν. The geographical distance is
This appendix provides a brief statistical measured as an orthodromic distance to
summary of the IsoriX approach. We refer the account for the curvature of the Earth. The ran-
reader to our technical publication for detailed dom effect used to model the differences
explanations, justifications, and implementa- between sources that are uncorrelated is
tion details concerning the modeling of iso- defined as:
scapes (Courtiol & Rousset, 2017) and to RandomUncorrg BNð0; ΩÞ; (9.4)
online material and future publication(s) for
more details about the calibration and assign- with Ω being a diagonal covariance matrix
ment procedure. In this technical appendix, we with constant variance. The residual dispersion
assume basic knowledge about Generalized error Errorgi is defined as:
Linear Mixed-effects Models (GLMM). For
Errorgi BNð0; φg Þ; (9.5)
those not familiar with GL(M)M, we recom-
mend reading chapter 7 of Venables & Ripley Because the expected values φg for the vari-
(2002), Diggle, Ribeiro & Christensen (2003), ance of the residual error cannot be directly
and Courtiol & Rousset (2017). observed, we model it using the residual disper-
sion model, which is a Gamma GLMM fitted to
the observed variances in isotope measure-
A.1 The Mean Model and the Residual ments at each location g. The mean value for
Dispersion Model this model is

In IsoriX, a given source measurement i at a D
φg 5 exp FixD
g 1 RandomSpatialg
location g, which we denote sourcegi is mod- (9.6)
eled as the sum of a fixed effect, two random 1 RandomUncorrD g ;
effects and an error structure. We write the
mean model as the following LMM: and the variance of the Gamma is 2φ2g = ng 2 1
with ng the number of months sampled at the
Sourcegi 5 Fixg 1 RandomSpatialg location g under the simple aggregation
(9.1)
1 RandomUncorrg 1 Errorgi : scheme defined in Section 9.4.2.2 (see Courtiol
& Rousset, 2017). The symbols D are not math-
The fixed part of the model is defined ematical exponents but labels to coefficients
(under the settings considered here) as: related to the definition of the residual error.
Fixg 5 β 0 1 β lat 3 jlatjg 1 β elev 3 elevg : (9.2) Here,
g 5 β0 ;
D
The random effect used to model the spatial FixD (9.7)
autocorrelation is defined as:
RandomSpatialD
g BNð0; Σ Þ;
D
(9.8)
RandomSpatialg BNð0; ΣÞ; (9.3)
with again ΣD being a variance-covariance
with Σ being a variance-covariance matrix matrix for which the covariance between
wherein the covariance between source loca- source location depends on their geographical
tion is the product of a variance, and of a cor- distance by means of the Matérn correlation
relation depending on their geographical function. This Matérn function is also defined
distance according to the Matérn correlation by ρD and ν D . Also,

APPENDIX: STATISTICAL FRAMEWORK 235
RandomUncorrD
g BNð0; Ω Þ
D
(9.9) Eq. (9.10). Note that δg 2 δ^ g is a random effect
with covariance matrix given by the mean fit,
with ΩD being a diagonal covariance matrix and that such a calibration model is fitted by
with constant variance. "spaMM".
We estimate all parameters by likelihood max-
imization, using restricted likelihood (REML) for
estimation of random-effect parameters. A.3 The Assignment Test
Let us call zi is the isotope measurement of
the assignment sample i. We can construct a
A.2 The Calibration Model test that an assignment sample comes from
We assume that the isotope measurement some candidate position g by considering the
calibgi of a calibration sample i collected at a distribution of the test statistic
location g is a linear function of the true b
unknown value δg of the isotope composition tig 5 δbi 2 δbg ; (9.14)
of the environment at this location: ^
where δ^ i is the predicted source value at the
calibgi 5 β C0 1 β Cδ δg 1 ECgi ; (9.10) location of origin which is deduced from the
calibration fit:
with the residual term ECgi representing the resid- c
b
δbi 5 zi 2 β^ 0 =β^ δ ;
c
ual variation among samples at the location g. (9.15)
We assume this residual term to be normally
distributed with mean zero and a given vari- and where δ^ g is again the predicted source
ance. In the current workflow, we consider this value at the candidate location.
variance constant but this assumption will be To characterize the distribution of the test
relaxed in future version of the package. Again, statistic, we first replace zi by its expression
the symbols C are not mathematical exponents (following Eq. 9.10) in the previous expression
but labels to coefficients related to the definition so to clarify its relationship with the true isoto-
of the calibration model. However, we do not pic composition at the origin (δi ):
know δg and therefore fit the following calibra- c
b
δbi 5 β C0 1 β Cδ δi 1 ECi 2 β^ 0 =β^ δ :
C
tion to the known isoscape value, i.e., the predic- (9.16)
tion of the mean fit model δ^ g :
Then, we replace in this equation the relation-
calibgi 5 β C0 1 β Cδ δ^ g 1 εCgi : (9.11) ship between the true unknown δi value and the
isoscape value predicted by the mean model at
The error terms of the two calibration mod-
the unknown location of origin, δ^ i 5 δi 1 EA i where
els are then related by:
EA
i is the prediction error of our mean model at
εCgi 5 β Cδ ðδg 2 δ^ g Þ 1 ECgi : (9.12) the unknown location of origin, yielding:

Thus the calibration model to be fitted can β C0 1 β Cδ ðδ^ i 2 EA
i Þ 1 E C
i 2 ^C
β 0
be rewritten as ti 5
β^ δ^ i
C
δ
calibgi 5 β C0 1 β Cδ δ^ g 1 β Cδ ðδg 2 δ^ g Þ 1 ECgi ; (9.13) C C
ðβ 0 2 β^ 0 Þ β δ
C
ECi EA
i β δ
C
5 21 1 2 :
where the difference δg 2 δ^ g represents the pre- β^ δ 1 δ^ i
C
β^ δ
C
β^ δ
C
β^ δ
C
diction error of the mean source value at the

sampling location g by the mean fit, and where (9.17)
the residual error term is the one defined in

The variance of this statistic can be approxi- All these terms can be computed from the
mated from Eq. (9.17) by the so-called delta fitted models. The first three variance compo-
method (see e.g., section 2.2.2 in Davison, 2003). nents are directly obtained: VarðEA i Þ is the pre-
Accordingly, we write all estimates as parame- diction variance of the fit of the mean model,
ter values minus some small perturbations, eC0 VarðECi Þ is the residual dispersion variance of
and eCδ (opposite to the estimation errors of the the calibration fit, VarðeC0 1 eCδ δ^ i Þ is the variance
parameters), here considered of the same order of error in prediction of isotope values by the
of magnitude as ECi and EAi , and we evaluate the calibration fit, deduced from covariance matrix
Taylor expansion up to second order in both of estimation errors of its coefficients. The last
perturbation terms, and take the expectation of term on the right represents the covariation
the second-order terms. This yields: between the same uncertainty in prediction,
and the uncertainty EA i in δi given the source
VarðECi Þ VarðeC0 1 eCδ δ^ i Þ data. It can be shown that the covariance term
Varðti Þ VarðEA
i Þ1 1
β^ δ β^ δ
C2 C2
is a function of the covariances of predictions
of the mean model at the calibration positions
2CovðeC0 1 eCδ δ^ i ; EA
i Þ
1 and predictions of the mean model at the can-
β^ δ
C
didate locations of origin of the assignment
(9.18) sample.

C H A P T E R
10
Outlook for Using Stable Isotopes in
Animal Migration Studies
Keith A. Hobson1, Leonard I. Wassenaar2, Gabriel J. Bowen3,
Alexandre Courtiol4, Clive N. Trueman5, Christian C. Voigt4,
Jason B. West6, Kelton W. McMahon7 and Seth D. Newsome8
1
Environment and Climate Change Canada, Saskatoon, SK, Canada, 2International Atomic Energy
Agency, Vienna, Austria, 3University of Utah, Salt Lake City, UT, United States, 4Leibniz Institute
for Zoo and Wildlife Research, Berlin, Germany, 5University of Southampton, Southampton, United
Kingdom, 6Texas A&M University, College Station, TX, United States, 7University of Rhode Island,
Narragansett, RI, United States, 8University of New Mexico, Albuquerque, NM, United States
A quantifiable understanding of animal populations are regulated, and how environ-

movement supports a host of research topics, mental changes affect life-history tactics.
ranging from the evolution of life history strate- As noted in the previous chapters, the past
gies to the ecology and physiology of actual decade has seen tremendous progress in the
migration and dispersal. The lack of basic infor- advancement of the use of naturally occurring
mation on migratory connectivity for most of stable isotope measurements and associated
the world’s migratory species remains the most spatial modeling approaches. However,
crucial gap in our knowledge, despite this gap knowledge gaps remain, and improvements
is primarily technological and not conceptual are needed through new research. It is impos-
(Wikelski et al., 2007). Quantitative knowledge sible to fully predict how the use of
of how individuals and populations are spa- stable isotopes in migration research will prog-
tially connected between all periods of their ress over the next decade as technological, ana-
annual cycle is the key. Without knowledge of lytical, and scientific innovations proceed,
where individuals spend their time during the often in unpredictably novel ways.
year, it is impossible to deduce the factors that In this summary chapter, we put forward
influence behavioral plasticity, fitness, how suggestions and priorities we believe will be of

238 10. OUTLOOK FOR USING STABLE ISOTOPES IN ANIMAL MIGRATION STUDIES
help in closing key knowledge gaps and in fur- bulk tissues for determining nonexchangeable
ther advancing the field. H, both on the analytical development side,
and by the introduction of new keratinous
reference materials to ensure results among
10.1 SAMPLES AND ISOTOPIC laboratories are comparable (Chapter 2:
ANALYSES Introduction to Conducting Stable Isotope
Measurements for Animal Migration Studies).
10.1.1 Isotopic Sampling Hydrogen isotopes, by far, remain the most
promising of the light stable isotopes for deter-
Stable isotopic information from migrant mining migratory connectivity. Yet, knowledge
animal tissues are useful only if they are gaps remain, particularly in our understanding
unambiguous indicators of the geographic of the H isotope diet-tissue discrimination for
location at which the tissues were grown and most terrestrial and aquatic organisms.
can be linked to a relevant isoscape for infer- Many of the analytical methods for bulk
ring movement. It is crucial to choose an tissue isotope analyses of CNS are nowadays
appropriate fixed or dynamic tissue (bulk tis- straightforward and highly cost-effective.
sue or specific compound) that reflects the Oxygen isotopes (18O), however, remain rela-
appropriate period of dietary integration of tively unexplored and hampered by analytical
interest. Critically, we need to know the eco- challenges and poor knowledge of O flow in
logical and physiological factors that deter- dietary food webs. We hypothesize that O iso-
mine diet-tissue isotopic discrimination topes will poorly mirror H isotopes, but it
corresponding to that tissue and how that remains to be seen whether new and other
discrimination varies both within and among valuable information is to be gained by tissue
individuals at a given location or on a given 18
O analyses (Bowen et al., 2009; Hobson &
diet. Accordingly, knowledge of the life history Koehler, 2015; Wolf, Newsome, Fogel, &
and biology of the species of interest is Martinez del Rio, 2013).
paramount. The analysis of compound-specific C, N, or
New models that link physiological pro- H isotope analysis of amino acids holds the
cesses of consumers to the processes and pat- tantalizing promise of obtaining less ambigu-
terns of stable isotope ratios in nature will be ous isotopic connections to dietary isoscapes
key to improving the accuracy and precision compared to using bulk tissues. However, the
of geospatial assignments. Improving our analytical aspects of compound-specific iso-
knowledge of diet-tissue isotopic discrimina- tope analysis (CSIA) remain difficult, costly,
tion and elemental turnover for organisms of and require the highest levels of isotope
interest will help refine field methods and analytical expertise and investment. Analytical
experimental design. This will likely require improvements (ease, lower cost) in CSIA anal-
controlled captive studies and sampling of yses of H and C isotopes of proteins are
wild populations whose diets and nutritional urgently needed, along with the need for
physiology are known, for multiple isotopes, amino acid H and C isotope reference materi-
bulk tissues, and specific compounds. als to ensure comparable results among
laboratories.
Trace element isotopes (i.e., 87Sr) have
10.1.2 Stable Isotope Analyses shown good promise for migratory studies,
Tremendous progress has been made over but barriers remain to routine use, such as the
the past 10 years for 2H isotope analysis of need for specialized laboratories, and the high

10.2 MIGRATORY SYSTEMS 239
risk of external contamination without careful communication across laboratories, with
precautions and handling. Furthermore, iso- exchange of reference materials and protocols
scapes for trace element isotopes are only to improve QA/QC and make the data and
beginning to be developed, and to date have associated interpretations comparable across
been characterized for few regions. studies. We also need continued method devel-
New laser-based isotope technologies show opment to make these CSIA techniques faster,
promise for reducing the cost and improving cheaper, and more efficient if we ever hope to
accessibility of isotope analyses (including see these powerful tools become more than a
field-based analyses), however, the current fringe endeavor pursued by a limited number
lack of automated sample preparation devices of labs. As the analytical, methodological, and
for bulk tissue C or H isotope analyses ham- theoretical commitment to understanding
pers any progress in this area, and will hope- CSIA-AA grows in the ecological community,
fully be overcome in the next decade. so will the power of these approaches to study-
ing movement and foraging ecology.
10.1.3 Compound Specific Isotope

Analyses 10.2 MIGRATORY SYSTEMS
The ability to disentangle the signals of
10.2.1 Avian and Insect Migratory
baseline isotope variability from trophic
Systems
dynamics with compound-specific
stable isotope analysis of individual amino Stable isotope applications in tracking bird
acids (CSIA-AA) has opened new doors to movement are clearly an area where the most
studying animal migration, habitat use, and progress has been made over the past 20 years.
foraging ecology. To realize the full potential Nonetheless, there continues to be a host of
of CSIA-AA in ecological studies, we need sev- areas of concern and needed developments
eral major improvements on both the analyti- (Hobson, 2011; Martinez del Rio, Wolf,
cal and theoretical fronts. Carleton, & Gannes, 2009). Fortunately, birds
There is an obvious need for controlled lab- molt at predictable periods of the annual cycle
oratory and field studies examining the pat- and these patterns are generally known for
terns, magnitudes, and variability of isotopic most species. So, feather isotope values can
discrimination of individual amino acids, as usually be associated with breeding, wintering,
well as the rates at which those isotopic signals or molt migration regions. However, the varia-
turnover in consumers. With this comes a need tion we see in δ2H values within single wild
to improve the CSIA-AA community’s collec- populations (i.e., growing tissues at the same
tive understanding of the underlying physiol- general location) is poorly understood.
ogy and biochemistry driving patterns in Understanding contributions of isotopic vari-
isotopic discrimination at the compound- ance within populations related to diet, age,
specific level (O’Connell, 2017). individual physiology, and microhabitats will
As more and more studies use CSIA-AA be important. On the other hand, the establish-
approaches, the need for standardization of ment of reasonable estimates of overall isoto-
procedures and corrections, reference materials, pic variance in feathers and other tissues and
and chromatography interpretations to enhance how this variance might be attributed to
QA/QC becomes even more important (e.g., predictable factors like habitat and feeding
Ohkouchi et al., 2017). This will require more and migratory guild will be of tremendous use

in moving toward better assignments of indivi- More effort is needed to quantify and pre-
duals and populations. Associated with this dict the extent and scale of temporal variability
are refinements to tissue-to-precipitation iso- in baseline aquatic isoscapes and the implica-
scape discrimination based on calibration algo- tions of temporal variability for isotope assign-
rithms used in all assignment approaches. ment methods. Mechanistic modeling is likely
Application of isotopic methods for tracking to be critical to cover the scale of marine iso-
small migratory insects are still in its infancy scapes, but field data are crucial to testing the
(Hobson et al., 2018) even though the size of validity of model simplifications and assump-
most insects makes them ideal candidates for tions. We repeat the longstanding call for more
this technique. Unlike birds and other animals, experimental and modeling studies quantify-
insects are amenable to controlled experiments ing and explaining physiological fractionation
whereby isotopic variance and calibrations of C and N isotopes within taxonomically and
between chitinaceous tissues and water functionally varied marine and aquatic organ-
sources can be established. Thus we expect to isms and their tissues.
see rapid advances in the use of isotopes for
migration insect research in the next 510
years. Future experiments should investigate
how isotopic patterns can be affected by life-
10.2.3 Mammal Migratory Systems
history strategies such as within-soil versus Isotopic tracking of terrestrial mammals has
above-soil larval stages. Many migratory been conducted successfully over the past dec-
insects of interest are associated with crops ades, yet the application of stable isotopes for
(i.e., pests) and the use of irrigation waters tracking mammals is impaired by low spatial
may be a confounding factor in establishing mobility of most terrestrial mammals, at least
appropriate water-based isoscapes. compared to migratory birds or insects. Owing
to their lower mobility, migratory mammals
do not cross necessarily wider areas of distinc-
tive isotopic patterns, which is a key prerequi-
10.2.2 Marine Migratory Systems
site in isotopic tracking. The spatial resolution
The use of isotopic tracking methods in of the isotopic tracking approach could be
marine and aquatic systems has been limited greatly improved if more reference data on the
by a lack of reference isoscapes. With isotopic composition of nonmigratory mam-
increased interest and effort invested in devel- mals were available. This would help in estab-
oping both sample-driven and mechanistically lishing isoscapes that are specific for certain
modeled CNS isoscapes, we anticipate wider mammal taxa or feeding guilds. Further, cap-
use of isotopic methods in marine migration tive studies may inform us about the causes of
ecology, particularly in fishes. Methodological variation in stable hydrogen isotopic composi-
advances lowering sample mass required for C tion in keratinous and nonkeratinous material.
and N isotopic analyses open the potential for This will eventually lead to improved isotopic
incremental sampling of the macromolecular transfer functions that support more fine-
framework of otoliths. This would allow retro- scaled tracking of mammal movements.
spective analysis of lifetime fish movements Approaches that combine isotopic tracking
drawing on the millions of otoliths routinely with complementary tracking methods such as
sampled (and often archived) as part of fisher- GPS or geolocators might help in evaluating
ies surveys. Studies on freshwater fish migra- the accuracy and precision of the isotopic
tion are also limited. methods in mammals. First and foremost, we

10.3 ISOSCAPES AND ASSIGNMENT 241
need more hypothesis-driven experimental seamless geological maps are in process that
studies conducted under controlled conditions promise additional potential for advances.
to learn more about the kinetics of stable iso- A predictive understanding of the isotope
topes, in particular those of hydrogen and discrimination processes that occur between
oxygen in mammal body tissues. To this end, ingestion of the animal’s water or food and the
we need more isotopic studies that address production of the animal tissues should be
physiological questions in an ecological con- developed and incorporated into the isotopic
text and vice versa. This will inform us about assignment process. As emphasized earlier,
promising research directions in the spatial this cross-cutting need will require integration
tracking of mammals based on stable isotopes. of behavioral and physiological theory with
data and experiments. Because this under-
standing would address potential sources of
10.3 ISOSCAPES AND both bias and uncertainty it will be integral to
ASSIGNMENT advancing all isotopic assignment work,
whether based on environmental isoscapes or
The success of using isoscapes in migratory more directly on known-origin data. In an ideal
studies rests on several interdisciplinary pillars world, isoscape-based assignments conducted
which will require more attention in the com- 10 years hence might allow consideration of an
ing years. The development and application of individual’s age, sex, and physiological status, as
theory, geospatial data, and quantitative tools well as other accessible properties such as
to predictively describe isotopic variation in weather, phenology, and environmental het-
the environment has been a focus of research erogeneity at potential places/times of origin,
for over a decade, but significant opportunities to provide the most accurate and robust
remain for improvement. Recent work has con- assessment of origin possible.
tinued to advance our capabilities with respect Significant scope remains for the isoscape
to modeling environmental waters, and signifi- community and quantitative ecologists
cant developments for others isotope systems involved in methods development of tools for
such as carbon and strontium have been made. isoscape-based assignment to broaden the
However, the isoscape research community is accessibility of these methods. The statistical
just beginning to scratch the surface in terms methods available for building isoscapes and
of the information resources available to sup- performing the assignment of migratory ani-
port such work. Large amounts of isotopic mals lie at the forefront of modern applied sta-
data remain unpublished, unavailable in tistics. Users can choose advanced frequentist
reports and institutions, or in papers, and with methods based on mixed models or advanced
the right tools and could be gathered from the Bayesian methods based on MCMC sampling,
literature and used to support isoscape model- which both consider and propagate many
ing. The capture and use of new data to sup- sources of uncertainty previously neglected.
port these efforts needs to become efficient Challenges include the distribution of existing
(Pauli et al., 2017). Modeling tools from the and upcoming methods, and the distribution
atmospheric and Earth science communities of reference datasets. Publications, no matter
could be leveraged to support improved how detailed, are insufficient to allow one to
predictions (Gibson et al., 2010; Risi et al., implement methods developed by others.
2012), remote-sensing data continue to grow in Generalizing the exchange of computer scripts,
scope and coverage and could inform many using R, e.g., would help solve this limitation.
isoscapes, and data resource projects such as It would allow users to implement different

methods with minimal difficulties and would book, stable isotopes offer a powerful compli-
offer the opportunity for developers to study mentary tool with distinctive advantages. The
the different methods in detail. The practice of most important of these is the absence of bias to
publishing code exists in the field (e.g., Vander site of marking of individuals, low cost, no
Zanden et al., 2014; Chapter 9: Isoscape interference with animal behavior, and the
Computation and Inference of Spatial Origins extremely small sample sizes required for analy-
With Mixed Models Using the R package sis. Every capture is a recapture.
IsoriX) but is not widespread. However, there are numerous intrinsic and
Sharing isoscape reference datasets is also extrinsic markers that add a degree of geo-
crucial. Combining efforts to share data of ref- graphic information to stable isotope
erence calibration samples for different taxa approaches and all recognize that the clever
would lead to more accurate assignments than blending of numerous tools will be the only
sharing the calibration estimates alone. Relying path forward. These include, but are not
on common calibrations would also help to restricted to, measurements related to the use
control an important source of variation of Bayesian priors in assignment models such
among assignment studies. Sharing reference as genetic structure, fatty acid composition,
data would further allow users to learn how to contaminant load, morphometrics, ring recov-
use a given method from the replication of ery vectors, and density distributions (Royle &
known results and developers to compare dif- Rubenstein, 2004; Rundel, Wunder, Alvarado,
ferent methods under the same conditions. For Ruegg, & Harrigan, 2013; Rushing, Ryder,
the sharing of computer code and data to be Saracco, & Marra, 2014; Van Wilgenburg et al.,
efficient, it is necessary that they come with 2011). It follows that the certainty of assign-
nonrestrictive copyright. Since publication of ment of a migrant’s origin may (but not
the first edition of this book several platforms always!) be improved by linking isotopic
supporting such analyses have been released, patterns with geographic information from
including the web-based IsoMAP toolkit other sets of nonisotopic data.
(http://isomap.org) and the IsoriX R package While the ability of other markers to pro-
(https://cran.r-project.org/web/packages/ vide geographic information is well estab-
IsoriX/index.html). Other researchers remain lished, there has been little work on defining
involved in development of purpose-specific the limits of resolution for each marker or
but potentially generalizable software tools for assessing the possible benefits and accuracy of
such analyses, and it is both clear and exciting combining multiple (isotopic and nonisotopic)
that the research community is beginning to markers or related information. Evaluating the
engage in a process of documenting and limitations of combinations of isotopic intrinsic
disseminating these research tools. markers will be important for understanding
which types of questions are best addressed
with which combinations of markers.
10.4 LINKING STABLE ISOTOPES
TO OTHER SPATIAL MARKERS
10.5 RESEARCH IN SUPPORT OF
There is no doubt that the revolution in the CONSERVATION
miniaturization of electronic devices for tracking
migrant animals will continue to have a tremen- Finally, while we have not emphasized
dous influence in studies of animal movements. important conservation efforts to protect
However, as we have stressed throughout this threatened migratory phenomena, in the next

REFERENCES 243
decades we may witness catastrophic declines Evolution. Available from http://dx.doi.org/10.1002/
of many migratory taxa (Wilcove & Wikelski, ece3.1383.
Hobson, K. A., Doward, K., Kardynal, K., & McNeil, J.
2008). We share a responsibility to conduct (2018). Inferring origins of migrating insects using iso-
meaningful research quickly and effectively to scapes: A case study using the true armyworm, Mythimna
define migratory connections, and to use this unipuncta, in North America. Ecological Entomology.
information and data to inform conservation Available from https://doi.org/10.1111/een.12505.
actions (Faaborg, Holmes, Anders, Bildstein, & Knight, A. T., Cowling, R. M., Rouget, M., Balmford, A.,
Lombard, A. T., & Campbell, B. M. (2008). Knowing
Dugger, 2010; Knight et al., 2008). It is insuffi- but not doing: Selecting priority conservation areas and
cient to endlessly refine and debate the merits the research-conservation gap. Conservation Biology, 22,
of calibration algorithms and isotopic variance, 610617.
or the pros and cons of this model versus that Martinez del Rio, C., Wolf, N., Carleton, S., & Gannes, L.
model. While these pursuits are important and (2009). Isotopic ecology ten years after a call for more
laboratory experiments. Biological Reviews of the
help to refine our approaches, conservation Cambridge Philosophical Society, 84, 91111.
action depends on focusing on answering O’Connell, T. C. (2017). ‘Trophic’ and ‘source’ amino acids
larger scale questions such as are northern in trophic estimation: A likely metabolic explanation.
populations declining faster than southern Oecologia, 184(2), 317326.
populations and can these trends be related to Ohkouchi, N., Chikaraishi, Y., Close, H. G., Fry, B., Larsen,
T., Madigan, D. J., . . . Ogawa, N. O. (2017). Advances
factors operating on the breeding or wintering in the application of amino acid nitrogen isotopic anal-
grounds or both? Stable isotopes will continue ysis in ecological and biogeochemical studies. Organic
to be a powerful advisory tool for conserva- Geochemistry, 113, 150174.
tion, and it is our duty to ensure that meaning- Pauli, J. N., Newsome, S. D., Cook, J. A., Harrod, C.,
ful and actionable results are conveyed to Steffan, S. A., Baker, C. J., . . . Cerling, T. E. (2017).
Opinion: Why we need a centralized repository for iso-
conservation managers. topic data. Proceedings of the National Academy of
Sciences, 114(12), 29973001.
Risi, C., Noone, D., Worden, J., Frankenberg, C., Stiller, G.,
& Sturm, C. (2012). Process-evaluation of tropospheric
References humidity simulated by general circulation models using
Bowen, G. J., Ehleringer, J. R., Chesson, L. A., Thompson, water vapor isotopologues: 1. Comparison between
A. H., Podlesak, D. W., & Cerling, T. E. (2009). Dietary models and observations. Journal of Geophysical Research:
and physiological controls on the hydrogen and oxygen Atmospheres, 117, D05303. Available from https://doi.
isotope ratios of hair from mid-20th century indigenous org/10.1029/2011JD016621.
populations. American Journal of Physical Anthropology, Royle, J. A., & Rubenstein, D. R. (2004). The role of species
139, 494504. Available from https://doi.org/10.1002/ abundance in determining breeding origins of migra-
ajpa.21008. tory birds with stable isotopes. Ecological Applications,
Faaborg, J., Holmes, R. T., Anders, A. D., Bildstein, K. L., 14, 17801788.
Dugger, K. M., et al. (2010). Conserving migratory land Rundel, C. W., Wunder, M. B., Alvarado, A. H., Ruegg,
birds in the New World: Do we know enough? K. C., Harrigan, R., et al. (2013). Novel statistical meth-
Ecological Applications, 20, 398418. ods for integrating genetic and stable isotope data to
Gibson, J. J., Fekete, B. M., & Bowen, G. J. (2010). Stable infer individual-level migratory connectivity. Molecular
isotopes in large scale hydrological applications. In J. B. Ecology, 22, 41634176.
West, G. J. Bowen, T. E. Dawson, & K. P. Tu (Eds.), Rushing, C. S., Ryder, T. B., Saracco, J. F., & Marra, P.
Isoscapes: Understanding movement, pattern and process on (2014). Assessing migratory connectivity for a long-
earth through isotope mapping (pp. 389406). Springer. distance migratory bird using multiple intrinsic mar-
Hobson, K. A. (2011). Isotopic Ornithology: A perspective. kers. Ecological Applications, 24, 445456.
Journal of Ornithology, 152, 4966. Van Wilgenburg, S. L., & Hobson, K. A. (2011). Combining
Hobson, K. A., & Koehler, G. (2015). On the use of stable- stable-isotope (δD) and band recovery data to improve
oxygen isotope (δ18O) measurements for tracking probabilistic assignment of migratory birds to origin.
avian movements in North America. Ecology and Ecological Applications, 21, 13401351.

Vander Zanden, H. B., Wunder, M. B., Hobson, K. A., system could do for experimental biologists. Journal of
Van Wilgenburg, S. L., Wassenaar, L. I., Welker, Experimental Biology, 210, 181186.
J. M., & Bowen, G. J. (2014). Contrasting assignment Wilcove, D. S., & Wikelski, M. (2008). Going, going, gone:
of migratory organisms to geographic origins using Is animal migration disappearing? PLoS Biology, 6(7),
long-term versus year-specific precipitation isotope e188. Available from https://doi.org/10.1371/journal.
maps. Methods in Ecology and Evolution, 5(9), pbio.0060188.
891900. Wolf, N., Newsome, S. D., Fogel, M. L., & Martinez del Rio,
Wikelski, M., Kays, R. W., Kasdin, J., Thorup, K., Smith, C. (2013). The relationship between drinking water and
J. A., Cochran, W. W., & Swenson, G. W., Jr (2007). the hydrogen and oxygen stable isotope values of tissues
Going wild—What a global small-animal tracking in Japanese Quail (Cortunix japonica). Auk, 130, 323330.
Index
Note: Page numbers followed by “f” and “t” refer to figures and tables, respectively.
A trophic discrimination factors, samples and isotopic analyses,

AA. See Amino acid (AA) 181182 238239
Active extrinsic markers, 812 Analytical instrumentation, 17 Animals of unknown origin
Active radar systems, 12 Anas acuta. See Northern pintails (Anas migratory carry-over questions
Aequipecten opercularis. See Scallops acuta) about, 199200
(Aequipecten opercularis) Anas platyrhynchos, 203 migratory catchment and dispersal
Aggregation Anax junius. See Common Green questions about, 199
aggregating source isotope data, Darner (Anax junius) migratory connectivity questions
216218 Animal migration, 3, 173, 178 about, 199
factor and aggregation function, 222 conservation, 58 Anthropogenic factors, 9394
Agriculture, 9394 migratory Aquaculture, 137138
Akaike Information Criteria (AIC), 220 connectivity, 58 Aquatic environments, migration in,
Alanine (Ala), 175176 movement, 1 138
δ2H analysis, 182 populations, 58 Aquatic organism tissue type, 154
Altitudinal migration, 34 patterns in different taxa, 2f ArcGIS, 66
American Redstarts (Setophaga scientific tools used to study, 816 Arctic Tern (Sterna paradisaea), 23,
ruticilla), 6, 99100 techniques used to track 137, 173
Amino acid (AA), 174 migration in animals, 9t Arginine (Arg), 176
AA C isotope fingerprinting technical advances and outlook, ARGOS system, 1213
approach, 176177 1617 Aspartic acid (Asp), 175176
discrimination factors, 185186 Animal migration studies, stable Assign-time calibrations,
essential amino acids, 176177 isotope measurements for 200201
isotope analysis isotopes Assignment models
accounting for consumer fractionation and discrimination, creation without sampling known-
physiology, 178179 28 origin material, 198199
case studies in movement and global-spatial, 3844 sampling considerations for
foraging ecology using light stable, 2628 creating, 196199
CSIA-AA, 180184 local-spatial, 4447 models directly linking tissue
CSIA-AA methodology, mass spectrometry chemistry and geography,
184185 instrumentation, 2830 196197
primer on amino acid sample collection and preparative Assignment samples, 209
biochemistry and isotope methods, 3038 Assignment to bins, 99100
discrimination, 174178 Animal migration studies, stable Asynchronous fur growth,
nonessential amino acids, 176177, isotopes in 123124
182 isoscapes and assignment, 241242 Atlantic salmon (Salmo salar), 96,
primer on amino acid biochemistry, linking stable isotopes to other 157158
174178 spatial markers, 242 Atmosphere, 2627
carbon metabolism, 176177 migratory systems, 239241 Avian migratory systems, 239240
hydrogen metabolism, 177178 research in support of conservation, Aythya affinis. See Lesser scaup
nitrogen metabolism, 174176 242243 (Aythya affinis)
245
246 INDEX
B fixation in plants, 5657 developing CSIA-based isoscapes,

Balaena mysticetus. See Bowhead isotopic composition of marine 180181
whales (Balaena mysticetus) phytoplankton, 147148 disentangling movement and
Barn Swallow (Hirundo rustica), 5 isotopic discrimination foraging ecology,
Baseline data, 209 factors, 93 181183
Basic assignment problem, calibration metabolism, 176177 methodology, 184185
and, 200201 essential amino acids, Computer modeling and simulation,
Bayes’ Rule, 204 176177 162166
Bayesian assignments, 205 nonessential amino example output of simulation
models, 104 acids, 177 modeling, 165f
Behavioral plasticity, 237 Carbon, hydrogen, nitrogen, oxygen, relationship between tissue δ13C
Biogeochemical mechanisms, 139 and sulfur elements (CHNOS values and modeled δ13CPOM
Biogeochemical processes, 95 elements), 2627, 53, 85 values, 163f
Biomineral C, O and Sr isotopes, 47 Carbon isotope(s), 37, 9394, 238 Conservation, 58
Biomineral δ18O values, 158 carbon isotope distributions, C3 and research in support, 242243
Biomineralized tissues, 156158 C4 plants, 57 Consumer physiological conditions,
Biosphere, 2627 in migration, 158160 185186
Black-legged Kittiwakes (Rissa of plant tissues, 5657 Consumer physiology, accounting for,
tridactyla), 14 Carbonate minerals, 58 178179
Black-throated Blue Warbler Carcharodon carcharias. See White Consumerresource relationships,
(Setophaga caerulescens), sharks (Carcharodon carcharias) 185186
100101, 100f Caretta caretta. See Loggerhead turtles Contaminants, 1415
Bone collagen, 154 (Caretta caretta) Continuous isoscapes, 62
Bowhead whales (Balaena mysticetus), Catharus ustulatus. See Swainson’s Continuous-flow isotope-ratio mass
9596 Thrush (Catharus ustulatus) spectrometers (CF-IRMS),
Breeding populations, 5 Cell phone technology, 812 2829
Broad-scale automated telemetry CF-IRMS. See Continuous-flow analytical methods, 44
array, 12 isotope-ratio mass methods, 4546
Brown stingray (Dasyatis lata), 181 spectrometers (CF-IRMS) Conventional DI-IRMS,
Bulk tissue isotope analysis, 174 CHNOS elements. See Carbon, 4546
Butterflies, 9899 hydrogen, nitrogen, oxygen, Coregonus clupeaformis. See Whitefish
and sulfur elements (CHNOS (Coregonus clupeaformis)
C elements) Coturnix japonica. See Quail
C3 plants, 5657 Classification trees, 203 (Coturnix japonica)
C4 plants, 5657 Claws, 106107, 124 Coupled ocean physics-
Calibration, 6364, 9195, 200201 Clustered sampling designs, 201202 biogeochemistry models,
algorithms, 105 Coastal and marginal marine 154
and basic assignment problem, environments, 138 Crossband transponders, 12
200201 Coasts, multiple isotopic variation in, Crossbills, 34
data preparation for, 225226 149151 Crustacean zooplankton, 153
estimation, 194 Color dyes, 8 CSIAs. See Compound-specific isotope
fit, 225, 227f Common Green Darner (Anax junius), analyses (CSIAs)
samples, 209 812 Cysteine, 45
Calidris mauri. See Western sandpipers “Comparative equilibration” approach
(Calidris mauri) to organic δ2H analyses, D
California Gray Whales (Eschrichtius 4042 Danaus plexippus. See Monarch
robustus), 9596 Compound-specific isotope analyses Butterflies (Danaus plexippus)
CAM “nectar corridors”, 97 (CSIAs), 16, 93, 238239 Data
CAM-based photosynthetic pathways, case studies in movement and analysis, 200203
95 foraging ecology using, basic statistical assumptions,
“Canopy effect”, 122 180184 201202
Carbohydrates, 86, 93, 184 constraining movement across calibration and basic assignment
Carbon, 121122 ecosystem, 183184 problem, 200201
INDEX 247
characteristics of isotope barcoding, 1617 Fisheries, 137138, 167
data, 201 fragment analyses, 1415 Fitting
population vs. individual-level Dragonflies, migration of, 812 calibration model, 226
geographic assignments, Drying methods, 40 isoscape, 219220
202203 Dual-inlet IRMS (DI-IRMS), 2829, isotope calibration model, 225226
preparation 4546 fitting calibration model, 226
for assignment, 227 Dynamic tissues, 32, 3536 preparing data for calibration,
for calibration, 225226 225226
scrutiny, 35 E models, 220
Deamination, 174175 EA-IRMS. See Elemental analyzer- Fixed tissues, 32, 3536
δ2H. See Hydrogen isotope isotope ratio mass spectrometry Food webs, 46, 8990, 9394
(δ2H isotope) (EA-IRMS) Foraging ecology, using CSIA-AA,
δ C values of primary production,
13
Ecogeochemistry, 173174 case studies in movement and,
147148 CSIA-AA, 181 180184
δ15N Ecosystem, constraining movement Fraction of exchangeable H (fex),
analysis, 174175 across, 183184 3839
discrimination factors, 93 Ecosystem Sr isoscapes, 7476 Fulmarus glacialis. See Northern
isoscapes, 93 Ehrenberg snapper (Lutjanus Fulmars (Fulmarus glacialis)
δ18O ehrenbergii), 182 Fur and molting patterns, 123124
isoscapes, 74 Elemental analyzer-isotope ratio mass
isotope, 200 spectrometry (EA-IRMS), 46, G
measurements, 9495 184 Gas chromatograph-combustion-IRMS
values, 157 Elephant (Loxodonta africana), 9697 (GC-C-IRMS), 184185
Dependencies, 211213, 212f Empirical lipid correction models, GC. See Gas chromatography (GC)
Desmodus rotundus. See Vampire bats 3536 GC-C-IRMS. See Gas chromatograph-
(Desmodus rotundus) Endogenous lipids, 178179 combustion-IRMS (GC-C-IRMS)
DI-IRMS. See Dual-inlet IRMS Escherichia coli (E. coli), 177178 Geographic Information System (GIS),
(DI-IRMS) Eschrichtius robustus. See California 66, 211212
Diazotrophs, 148 Gray Whales (Eschrichtius Geographical/geography
DIC. See Dissolved inorganic carbon robustus) accuracy of geographic model
(DIC) Essential amino acids, 176177 calibrations, 200201
Dichotomous assignment to bins, δ13C fingerprinting approach, 182 area, 196197
99100 Estuaries, multiple isotopic variation assignment(s), 130131, 208
Diet in, 149151 models, 200201
diet-tissue isotope discrimination Eubalaena australis. See Right whales plotting, 229230
factor, 86 (Eubalaena australis) running, 227229
quality, 176 Exchangeable H, 3839 gradients, 200
“Diet switch” experiments, 8687, organic reference materials for, isotopic variation, 192193
124126 4243 models directly linking tissue
Dietary δ15N, 7374 Extrinsic markers, 8. See also Intrinsic chemistry and, 196197
Differential migration, 3 markers models indirectly linking tissue
Diomedea exulans. See Wandering chemistry and, 198
Albatrosses (Diomedea exulans) F patterns, 200
Discriminant analysis, 204 Fatal round-trip migration, 3 Geospatial data, 6466
Disentangling movement and Feathers, 15, 32, 4647, 97 GIS. See Geographic Information
foraging ecology, 181183 laboratory intercomparison of δ2H System (GIS)
Dispersal, 12 for, 41f Global Network for Isotopes in
Dissolved inorganic carbon (DIC), Field costs for sampling animal Precipitation (GNIP), 5354,
147148 tissues, 195 129, 192, 214215
Distribution of C3 and C4 plants, Filter-feeding mussel tissue (Mytilus dataset, 5354
natural, climate-induced californianus), 180 Global plant δ13C isoscapes, 7173
variation in, 62 Fish otoliths, 47 Global-scale hydrologic processes,
DNA δ18O values (δ18Ooto), 158 3032
248 INDEX
Global-spatial assays, 3032 analysis of individual AAs, 182 R environment, 210211

Global-spatial isotopes of essential AAs, 182 workflow, 214230, 214f
hydrogen isotopes, 3840 IRMS analysis, 43 assigning samples to location of
organic reference materials for Hydrosphere, 2627 origin, 226230
nonexchangeable H, 4243 Hylocichla mustelina. See Wood Thrush fitting isotope calibration model,
oxygen isotopes, 4344 (Hylocichla mustelina) 225226
Glossophagasoricina, 121122 modeling H isoscape, 215225
Glutamate, 93 I preparing R environment for
Glutamic acid (Glu), 175176 IAEA. See International Atomic IsoriX workflow, 215
standards, 45 Energy Agency (IAEA) Isoscape(s), 34, 53, 86, 98105, 140f,
Glycine (Gly), 175 ICARUS. See International 150f, 200201, 208
Glycolytic nonessential AA alanine, Cooperation for Animal and assignment, 241242
177178 Research Using Space assignment to bins, 99100
GNIP. See Global Network for (ICARUS) computation and inference of
Isotopes in Precipitation ICP-MS. See Inductively-coupled- spatial origins with mixed
(GNIP) plasma mass spectrometry models
(ICP-MS) IsoriX package, 210214
H Incorporating multiple sources of IsoriX workflow, 214230, 214f
H isoscape information, 103105 construction
modeling, 215225 difference between feather δ2H and of marine isoscapes, 152154
fitting isoscape, 219220 predicted precipitation δ2H, and use of marine isoscapes,
plotting predicted H isoscape and 105f 151152
associated rasters, 223225 Incremental tissues and movement, ecosystem Sr isotopes, 58
predicting H isoscape, 222223 161162 gas exchange, photosynthetic
preparing raw source H isotope Individual compounds, 16 pathway, and plant C isotopes,
data, 215216 Inductively-coupled-plasma mass 5657
preparing structural raster, spectrometry (ICP-MS), 15, 47 hydrological mixing, evaporation,
221222 Insect migratory systems, 239240 and surface water δ2H and
selecting and aggregating source Insects, 8, 1213, 34, 240 δ18O, 55
isotope data, 216218 International Atomic Energy Agency isotopic variability in marine
prediction, 222223 (IAEA), 5354 systems, 139
H isotopes, 4142, 208209, 209f, 238 International Cooperation for Animal for marine migration research,
Hair keratin, 122123 Research Using Space 156162
High-pressure liquid (ICARUS), 13 modeling, 65, 67, 7677
chromatography-C-IRMS International Satellite Land-Surface movements inferred without,
(HPLC-C-IRMS), 184185 Climatology Project, 65 9598
High-Temperature Thermochemical Interpolation models, 6364 multiple isotopic variation in shelf
reactors (HTC reactors), 2829, Intersample isotopic heterogeneity, 32 seas, coasts, and estuaries,
44 Intrasample isotopic heterogeneity, 32 149151
Hirundo rustica. See Barn Swallow Intratropical migration, 34 nitrogen isotopes, 148149
(Hirundo rustica) Intrinsic markers, 1416 nitrogen isotopes in soils and
HPLC-C-IRMS. See High-pressure contaminants, parasites, and plants, 5758
liquid chromatography-C-IRMS pathogens, 1415 oxygen, hydrogen, and strontium
(HPLC-C-IRMS) stable isotope approaches, 1516 isotopes, 139147
HTC reactors. See High-Temperature trace elements, 15 plant water and organic H and O
Thermochemical reactors (HTC Invasion, 34 isotopes, 5556
reactors) Isoleucine (Ile), 175176 processes driving spatial patterns in
Humpback whales (Megaptera IsoMAP system, 7071, 211, 219 δ13C values, 147148
novaeangliae), 159160 IsoriX, 208 Rayleigh distillation and
Hydrogen, 74, 122123, 200 package precipitation δ2H and δ18O,
metabolism, 177178 dependencies, 211213, 212f 5455
Hydrogen isotope (δ2H isotope), internal structure, 213214 spatially explicit assignments,
3840, 9495, 139147 IsoriX and R, 211 100103
INDEX 249
temporal variability in baseline isotope-enabled global models directly linking tissue
isoscapes, 154156 biogeochemical models, chemistry and geography,
terrestrial migration research, 164166 196197
6976 isotope-fractionating processes,
ecosystem Sr isoscapes, 7476 5960 L
vegetation C isoscapes, 7173 local-spatial, 4447 Laboratory costs for stable isotope
vegetation H and O isoscapes, 74 Isotopic analyses, samples and analysis, 195
vegetation N isoscapes, 7374 compound specific isotope analyses, Landscape-level stable isotope
Water H and O isoscapes, 6971 239 variation, 6061, 61f
uncertainty of isoscape predictions, stable isotope analyses, 238239 Lanius ludovicianus. See Loggerhead
6769 Isotopic assays, 26 shrikes (Lanius ludovicianus)
Isoscapes mapping, 6369 tissue sample considerations for, Laser isotope analyzers, 30
geospatial data, 6466 3233 Lasiurus cinereus (L. cinereus), 123
model calculations, 66 Isotopic assignment of origins, 8695 Leaf water isotopic composition, 56
model selection and calibration, Isotopic discrimination factors, 9195, Leap-frog migration, 3
6364 185 Leptonycteris curasoae.
optimization of residuals, 67 Isotopic retention in tissues See Nectarivorous bat
uncertainty of isoscape predictions, fur and molting patterns, 123124 (Leptonycteris curasoae)
6769 keratinous body products, 124 Lesser scaup (Aythya affinis), 3233,
Isotope discrimination, 28, 86 tissue turnover rates, 124126 33f
primer on, 174178 Isotopic tracking, 117 Leucine (Leu), 175176
carbon metabolism, 176177 of marine animal movement Light logging geolocators, 195
hydrogen metabolism, 177178 computer modeling and Light stable isotopes, 2628, 8990
nitrogen metabolism, 174176 simulation, 162166 Likelihood-based methods, 204205
Isotope mass spectrometry examples of isotopes and Linear mixed-effects models (LMMs),
instrumentation, 2830 isoscapes for marine migration 210, 219, 234
Isotope-based studies of migratory research, 156162 Linear models, 201202
animals, 191 mechanism, spatial structure, and Lion’s mane jellyfish (Cyanea capillata),
assignment model types isoscapes, 139156 180181
classification trees, 203 migration in aquatic Lipid(s), 86, 179
likelihood-based methods, environments, 138 correction approach, 3536
204205 using stable isotopes to infer extraction with polar solvents, 184
data analysis and modeling migration in marine settings, lipid-rich foods, 126128
considerations, 200203 138139 LMMs. See Linear mixed-effects
basic statistical assumptions, studies in mammals, 123124 models (LMMs)
201202 of terrestrial mammals, 240241 Local-spatial assays, 3032
calibration and basic assignment Isotopic turnover, 8689 Local-spatial isotopes. See also Global-
problem, 200201 Isotopic variability in marine systems, spatial isotopes
characteristics of isotope data, 201 139 biomineral C, O and Sr isotopes, 47
population vs. individual-level Isotopic variation, 162, 202 isotopes of trace elements
geographic assignments, mapping, 54 (87Sr/86Sr), 4647
202203 stable carbon and nitrogen isotopes,
planning your study, 192195 K 4445
sampling considerations, Keratin, 36, 4142, 47, 117 sulfur isotopes, 4546
195200 Kinetic isotope fractionation, 28, 185 Localized isotopic gradients,
Isotope(s) Kirtland’s Warbler (Setophaga 139147
evaluate sources of isotope kirtlandii), 5 Loggerhead shrikes (Lanius
variance, 193 Known-origin material ludovicianus), 106107
global-spatial, 3844 creating assignment models without Loggerhead turtles (Caretta caretta),
isotope-based assignment workflow sampling, 198199 9596, 157
in IsoriX, 209 sampling considerations for Long-distance migrants, 5
isotope-based migration research, creating assignment models Longitudinal migration, 3
78 using, 196199 Loop migration pattern, 3
250 INDEX
Loxodonta africana. See Elephant Migrant isoscape, 194 N

(Loxodonta africana) Migration, 24, 137 Nearctic migratory birds, 97
Lysine (Lys), 175 in aquatic environments, 138 Nectarivorous bat (Leptonycteris
carbon and nitrogen isotopes, curasoae), 97, 121122
M 158160 Nitrogen, 93, 121122
Macrophytes, 149150 research applications, 6162 cycle openness, 7374
Mallards (Anas platyrhynchos), 203 scientific tools to study, 816 excretion, consumer mode of, 176
Mammal Migratory animals fixation, 148
biology, 117 characteristics of isotope data for isotopic effects, 5758
migration, 121 use in studying, 201 Nitrogen isotopes, 37, 9193, 148149
migratory systems, 240241 isotope-based studies, 191 in migration, 158160
tissue δ2H values, 126128 assignment model types, 203205 in soils and plants, 5758
MAPS. See Monitoring Avian data analysis and modeling Nitrogen metabolism, 174176
Productivity and Survivorship considerations, 200203 heavily fractionating “trophic”
(MAPS) planning your study, 192195 amino acids, 175176
Marine isoscapes, 47 sampling considerations, 195200 minimally fractionating “source”
assignment to, 160161 Migratory bats, 121 amino acids, 175
case study, 128131, 208, 225226 15
construction N-labeling experiments, 175
global models, 154 Migratory carry-over questions, NMR. See Nuclear magnetic resonance
reference organisms, 152154 199200 (NMR)
and use, 151152, 152f Migratory catchment and dispersal Noctule bat (Nyctalus noctula),
Marine migration research, examples questions, 199 128129
of isotopes and isoscapes for, Migratory connectivity questions, 199 Nomadic movement, 4, 138
156162 Migratory movement(s), 4, 117121 Nonessential amino acids, 176177,
assignment to marine isoscapes, stable isotopes application to 182
160161 migratory movements study, Nonexchangeable H, 38
carbon and nitrogen isotopes in 126131 North Sea isoscape, 180181
migration, 158160 Migratory populations, connectivity Northern Fulmars (Fulmarus glacialis),
incremental tissues and movement, and conservation, 58 14
161162 Migratory systems Northern pintails (Anas acuta), 203
oxygen isotope records of migration avian and insect, 239240 Nuclear magnetic resonance (NMR),
from biomineralized tissues, mammal, 240241 16
156158 marine, 240 Numbered neck collars, 8
Marine migratory systems, 240 Migratory terrestrial animals Nyctalus noctula. See Noctule bat
Mass spectrometers, 2728 isotope tracking, 53 (Nyctalus noctula)
Matrix equivalent organic reference MixSIAR, 210211
materials, 42 MOC. See Methoxycarbonyl AA ester O
Mechanistic isotope models, 162164 (MOC) One-way movement, 4
Mechanistic modeling, 240 Molting patterns, 123124 Ontogenetic migrations, 138
Megaptera novaeangliae. See Humpback Monarch Butterflies (Danaus Organic reference materials for
whales (Megaptera novaeangliae) plexippus), 23, 98 nonexchangeable H, 4243
Met. See Methionine (Met) Monitoring Avian Productivity and Organic tissues, 2930, 202
Metabolic pools, 196 Survivorship (MAPS), 100101 Organic δ2H analyses, comparative
Metabolic turnover, 8687 Movements inferred without equilibration approach to,
Metabolically active tissues, 8586, 154 isoscapes, 9598 4042
Metabolically inactive keratin, 15 Multiisotope Otolith(s), 158
Metabolically inert tissues, 196 approach, 96 fish, 32
Meteoric precipitation, isotopic structure of breeding populations of organic and inorganic fractions, 96
composition of, 6061 animals, 9798 otolith trace-element-based
Methionine (Met), 175 Multiple isotopic variation in shelf approaches, 156157
Methoxycarbonyl AA ester (MOC), seas, coasts, and estuaries, values of δ18O, 157
184185 149151 Ovenbirds (Seiurus aurocapillus),
Microbalance weighing, 36 Mycorrhizal fungi, 9192 9798
INDEX 251
Oxygen, 74, 200 Probability maps, 229, 230f Sample(s)
Oxygen isotopes, 4344, 9495, of geographic assignment, 231f cleaning and storing, 3536
139147, 238 Processes driving spatial patterns in heterogeneity, 33
estimates of global-scale spatial δ13C values of primary and isotopic analyses
variation, 146f production, 147148 compound specific isotope
records of migration from Proline (Pro), 175176 analyses, 239
biomineralized tissues, isotopic sampling, 238
156158 Q stable isotope analyses, 238239
Oyamel Fir (Abies religiosa), 98 Quail (Coturnix japonica), 94 isotopic uncertainty, expressing,
3334
P R sample-based isoscapes, 162164
Pacific bluefin tuna (Thunnus R environment, 210211 weighing, 3638
orientalis), 186 preparing for IsoriX workflow, 215 procedure for weighing tissues
Pacific leatherback sea turtles R package, 213 for stable isotopic analysis, 39t
(Dermochelys coriacea), 182183 “IsoriX”, 208 Samples assigning to location of
Pacific white shrimp (Litopenaeus Radar technology, to track migratory origin, 226230
vannamei), 186 movement, 814 data preparation for assignment, 227
Pacific yellowfin tuna (Thunnus Radio frequency transmitters, 812 Sampling animal tissues, field costs
albacares), 181 Radiogenic 87Sr isotopes, 2930 for, 195
Parasites, 1415 Reaction progress variables, 8788 Sampling considerations, 195200
Partial migration, 3 Reference organisms, 152154 for creating assignment models
Particulate organic matter (POM), 148, Reference samples. See Calibration, using known-origin material,
153 samples 196199
Passive extrinsic markers, 1314 Remote sensing technologies, 12 models indirectly linking tissue
Patagial tags, 8 Residual(s) chemistry and geography, 198
Pathogens, 1415 dispersion tissue amount, type, and timing,
PEP carboxylase. fit, 219 195196
See Phosphoenolpyruvate model, 219 for typical research questions about
carboxylase (PEP carboxylase) optimization, 67, 68f animals of unknown origin,
Phenylalanine (Phe), 175 variation, 228 199200
Philomachus pugnax. See Ruffs Right whales (Eubalaena australis), Satellite, 1213, 195
(Philomachus pugnax) 9596 technology, 814
Phosphoenolpyruvate carboxylase Rissa tridactyla. See Black-legged tracking, 1314
(PEP carboxylase), 5657 Kittiwakes (Rissa tridactyla) systems, 1213
Photosynthetic pathways, 5657, 85, Rosy Starlings (Sturnus roseus), 8687 transmitters, 1213
9394 Round-trip migration, 3 Savannah sparrows (Passerculus
Phytoplankton, 147150, 153, 180 Ruffs (Philomachus pugnax), 97 sandwichensis), 196
Pinnipeds, 138, 156, 161162 Scallops (Aequipecten opercularis),
Pivaloyl-isopropyl ester (Pv), 184185 S 160161
Plant C isotopes, 5657 34
S analysis, 45 Scientific tools used to study
POM. See Particulate organic matter Salmo salar. See Atlantic salmon (Salmo migration, 816
(POM) salar) SD. See Standard deviation (SD)
Precipitation δ2H and δ18O, 5455 Sample collection and preparative SE. See Standard error (SE)
Predicted H isoscape, plotting, methods, in migration studies, Sea turtles, 138, 160161
223225 3038 Sea-surface temperature (SST),
Prediction expressing sample isotopic 147148
error, 6769 uncertainty, 3334 Seabirds, 91, 138
locations, 221 other sample collection issues, Seiurus aurocapillus. See Ovenbirds
variance 3435 (Seiurus aurocapillus)
of calibration fit, 228 sample cleaning and storing, 3536 Serine (Ser), 175
of mean fit, 228 sample weighing, 3638 Setophaga caerulescens. See Black-
Primary producers, 180, 183184 tissue sample considerations for throated Blue Warbler
Primary reference material, 2728 isotopic assays, 3233 (Setophaga caerulescens)
252 INDEX
Setophaga coronata. See Yellow- approximate elemental application of stable isotopes to

rumped Warbler (Setophaga abundances as mass fraction, study of migratory movements,
coronata) 26t 126131
Setophaga kirtlandii. See Kirtland’s global-spatial isotopes, 3844 isotopic retention in tissues,
Warbler (Setophaga kirtlandii) isotope fractionation and 123126
Setophaga ruticilla. See American discrimination, 28 isotopic tracking, 240241
Redstarts (Setophaga ruticilla) isotope mass spectrometry stable isotopes and movements of
Shallow marine environments, instrumentation, 2830 terrestrial mammals, 121123
150151 light stable isotopes, 2628 Terrestrial migration research,
Shelf seas, multiple isotopic variation local-spatial isotopes, 4447 isoscapes for, 6976
in, 149151 sample collection and preparative ecosystem Sr isoscapes, 7476
Solvent cleaning, 35 methods, 3038 isoscapes mapping, 6369
Spatial organization, 5961, 60f and movements of terrestrial nitrogen isotopes in soils and
Spatial origin assignment models, mammals plants, 5758
198 carbon and nitrogen, 121122 plant water and organic H and O
Spatially continuous isoscapes, 6263 hydrogen, 122123 isotopes, 5556
Spatially discrete isoscapes, 6263 transfer functions, 126 Rayleigh distillation and
Spatially explicit assignments, Stable nitrogen isotopes, 4445 precipitation δ2H and δ18O,
100103, 101f Standard deviation (SD), 33 5455
migratory connectivity determined Standard error (SE), 33 vegetation C isoscapes, 7173
using stable isotope analyses, Static electricity, 37 vegetation H and O isoscapes, 74
100f Statistical assumptions vegetation N isoscapes, 7374
relationship between stable identically distributed (process Water H and O isoscapes, 6971
hydrogen isotope ratios, 102t homogeneity), 202 TFA. See Trifluoroacylisopropyl ester
Species-specific molting patterns, statistical independence, (TFA)
123124 201202 Thickbilled Murres (Uria lomvia), 14
87
Sr/86Sr. See Strontium isotope ratios Sterna paradisaea. See Arctic Tern Threonine (Thr), 175
(87Sr/86Sr) (Sterna paradisaea) Tissue
SST. See Sea-surface temperature (SST) Stochastic modeling extension, 205 amount, type, and timing, 195196
Stable-carbon isotopes, 4445 Strontium isotope ratios (87Sr/86Sr), CNS isotopic data, 159
Stable isotope(s), 25, 85, 152153, 174, 58, 139147 isotopic retention in, 123126
191 isoscape, 90 isotopic turnover, 8689
analyses, 47, 238239 isotopes of, 4647 conceptual depiction of way, 87f
in animal migration studies spatial variability, 58 physiology of tissue dynamics, 196
isoscapes and assignment, Strontium isotopes, 59, 139147 sample considerations for isotopic
241242 values of bedrock, 90 assays, 3233
linking stable isotopes to other Sturnus roseus. See Rosy Starlings intrasample hydrogen isotopic
spatial markers, 242 (Sturnus roseus) heterogeneity, 33f
migratory systems, 239241 Sulfur isotopes, 4546, 149 Tissue turnover rates, 124126
research in support of Swainson’s Thrush (Catharus trace element isotopes, 35
conservation, 242243 ustulatus), 8 tissue chemistry and geography,
samples and isotopic analyses, linking, 198
238239 T To-and-fro migration, 34
applications, 239240 Tachycineta bicolor. See Tree swallows Trace element, 15, 2627, 156157, 203
approaches, 1516 (Tachycineta bicolor) analysis, 195
assays, 2728 Taxon-or guild-specific isotopic isotopes, 238239
causes and consequences of transfer functions, 126128 Tracing
variation, 126128 TCA. See Tricarboxylic acid (TCA) carbon isotopes, 9394
to infer migration in marine Temporal variability in baseline hydrogen and oxygen, 9495
settings, 138139 isoscapes, 154156 nitrogen isotopes, 9193
laboratory costs for analysis, 195 Terrestrial animal movements, Tracking animal movements, isotopic
measurements for animal migration 139147 methods to
studies Terrestrial mammals, 117, 118t challenges, 105107
INDEX 253
using isoscapes, 98105 Uria lomvia. See Thickbilled Murres Water
isotopic assignment of origins, 8695 (Uria lomvia) H and O isoscapes, 6971
movements inferred without Uric-acid, 183184 water-dietary-tissue isotope
isoscapes, 9598 discriminations, 35
Transamination, 174176 V Western sandpipers (Calidris mauri),
Transmitters technology, to track Valine (Val), 175176 15
migratory movement, 814 Vampire bats (Desmodus rotundus), Whiskers, 124, 125t
Tree swallows (Tachycineta bicolor), 90 122 White sharks (Carcharodon carcharias),
Tricarboxylic acid (TCA), 177179 Vegetation 159
Trifluoroacylisopropyl ester (TFA), C isoscapes, 7173 Whitefish (Coregonus clupeaformis), 96
184185 H and O isoscapes, 74 Wilson’s Warbler (Wilsonia pusilla),
Trophic 2H discrimination effects, 105 N isoscapes, 7374 100f
Trophic transfer, 185186 VHF technology, 12 Wood Thrush (Hylocichla mustelina),
Tuna, 138 VSMOW. See Vienna Standard Mean 104
Turtles, 153154, 156 Ocean Water (VSMOW)
Tyrosine (Tyr), 176 Y
Yellow-rumped Warbler
U W (Setophaga coronata), 8
Uni-Prep device, 29f, 4041, 43 Wandering Albatrosses
Urey (Ur), 28 (Diomedea exulans), 1213

Tracking Animal Migration With Stable Isotopes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tracking Animal Migration With Stable Isotopes

Uploaded by

Copyright:

Available Formats

TRACKING ANIMAL MIGRATION WITH

British Library Cataloguing-in-Publication Data

Library of Congress Cataloging-in-Publication Data

For Information on all Academic Press publications

Publisher: Andre Gerhard Wolff

List of Contributors vii 3.2 Process 54

6. Isotopic Tracking of Marine Animal 8.5 Assignment Model Types 203

6.1 Introduction 137 9. Isoscape Computation and Inference

1.1 INTRODUCTION Some key proximate questions related to

Tracking Animal Migration with Stable Isotopes

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

Extrinsic 1. Can apply to a broad 1. Requires initial capture and

2. High spatial resolution. 2. Biased toward initial capture

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TABLE 1.1 (Continued)

2. Light weight (,1 g).

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TABLE 1.1 (Continued)

Technique Advantages Disadvantages References

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

2.1 INTRODUCTION commercial, government, or university labora-

Tracking Animal Migration with Stable Isotopes

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

Elemental Analyzer for δ13C and δ13N

Elemental Analyzer for δ2H or δ18O

Ceramic outer Open split H2 CO

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

“one-size-fits-all” cleaning or lipid correction 2.2.5 Sample Weighing

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

–105 FIGURE 2.4 The dependence of instrumental

0 0.1 0.2 0.3 0.4 0.5 0.6

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

To reiterate, inaccurate or imprecise weigh- species ranging from birds to mammals to

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

TRACKING ANIMAL MIGRATION WITH STABLE ISOTOPES

0 FIGURE 2.5 Laboratory intercom-

Moose hair 2 163.5 2.1 84 2 164.7 2.4 54