Professional Documents
Culture Documents
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.
ISBN: 978-0-12-814723-8
v
vi CONTENTS
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
This page intentionally left blank
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
This page intentionally left blank
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
This page intentionally left blank
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
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.
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
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
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
(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).
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, migra- tions 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
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
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
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.
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.
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.
2
Introduction to Conducting
Stable Isotope Measurements for Animal
Migration Studies
Leonard I. Wassenaar
International Atomic Energy Agency, Vienna, Austria
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).
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)
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
Auto sampler
Oxygen setup
He flow in He
Hydrogen setup Reference gas
injectors
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
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
–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.
–115
–120
–125
–130
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.
–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
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).
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,
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
2.5 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.
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
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
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
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
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
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
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 environ- ment 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
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
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.
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
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
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.
–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)
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
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
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
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.
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.
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.
4
Application of Isotopic Methods to
Tracking Animal Movements
Keith A. Hobson
University of Western Ontario, London, ON, Canada
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
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).
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
(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
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
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
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
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)
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)
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
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
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.
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
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
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
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)
(Continued)
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)
(Continued)
Ectophylla alba
Local movements Chiroptera Tropics Central America More than 10 species Rex et al.
(2011)
R. madagascariensis
Latitudinal migration Proboscidea Tropics Africa; Kenya Loxodonta africana H; C; N Cerling et al.
(2009)
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
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
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
(Continued)
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
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
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.
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.
6
Isotopic Tracking of Marine Animal
Movement
Clive N. Trueman and Katie St John Glew
University of Southampton, Southampton, United Kingdom
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
δ13C isoscape
δ15N isoscape
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
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)
TABLE 6.1 (Continued)
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?
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)
TABLE 6.1 (Continued)
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?
(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
146 6. ISOTOPIC TRACKING OF MARINE ANIMAL MOVEMENT
0 2
–2 0
–4 –2
–6 –4
–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
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 composi- tion 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
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.
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
POM
–25
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;
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.
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,
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
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
(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
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
(C) (D)
0
−20
25
−40
50
(E) (F)
−20.0
−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.
166 6. ISOTOPIC TRACKING OF MARINE ANIMAL MOVEMENT
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
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.
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.
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
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
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
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
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
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 deriva- tion 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
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
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.
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.
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
8.2 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,
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
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
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
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
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.
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
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
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 pro- of 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),
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
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.
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
This table shows the first three rows of a larger data frame.
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
This table shows the first three rows of a data frame containing a total of 327 source locations.
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.
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.
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
–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 sta- tion. 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.
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.
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.
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,
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
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.
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
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
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
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
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.
245
246 INDEX