Experimental Studies on Species Relationships in Erythrina (Leguminosae:
Papilionoideae)
David A. Neill
Annals of the Missouri Botanical Garden, Vol. 75, No. 3. (1988), pp. 886-969.
Stable URL:
http://links.jstor.org/sici?sici=0026-6493%281988%2975%3A3%3C886%3AESOSRI%3E2.0.CO%3B2-O
Annals of the Missouri Botanical Garden is currently published by Missouri Botanical Garden Press.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/mobot.html.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.
http://www.jstor.org
Thu Oct 25 13:35:33 2007
EXPERIMENTAL STUDIES ON
SPECIES RELATIONSHIPS
IN ER YTHRTYA
(LEGUMINOSAE:
PAPILIONO1DEAE)I
David -4. -IiezLL2
Erythrina I,. conaprises about 1 1 2 specirs distributed throughout the tropics arzd subtropics. Most species are
trees or shrubs, and most are diploids with n = 21. -411 are adapted to bird pollinntion, some b y pnssrrinr birds
arzd others b y hrtrr~mingbirds.Erythrina is subdivided intojiive subgenera and 2 7 sections. Research corlcrrztrnted
orz sect. Erythrina, with 36 sprcies crnterrd in Alilrxico and Ce~ltralAmrricn; srlectrd sprcirs in other sectiorzs
were also studied. Bxperinaental interspecijc hybridizntiorzs and se(fconapatibi1ity trinls Tarre corzductrd using
cultivatrd trees at severnl botarzical gardrrzs in Hazoaii. Corr~parntivrnaorphological analyses zoere rrlade o f t h e
hybrids arzd their parrnts. Studies ofpopulntiorz structurr and natural hybridization were cnrrird out irz natural
populatiorzs oj'hunanairzgbird-polli~~ntrd
srct. Erythrina. Erphrina specirs nre selj:conapatible, but some irzbreedirlg
depression is nssocintrd with seljng. Within sect. Erythrina, interspecijc hybrids are obtained just as readily
ns are progeny ji-ona within-species outcrosses. The hybrids are vigorous, fertile, and b y srveral measures exhibit
interspecijc hetrrosis. At greater tnxorzonaic distances betzoeerz the pnrerltnl specirs (brtzoeen srctiorzs and
s ~ ~ b g e r ~ e,r crossability,
a)
viability, nrld,firtility oj'the hybrid progeny nre gerzerally lower than in intrnsectional
hybridizntions. Some hybrids were obtained brtwern species qf dlgerer~t subgrrzera irldigerzous to difrrent
cor~tir~ents.
I%-ereare probably no absolute irlterrlul bnrrirrs to hybridizatiorl nrr~orzgall t/ar diploid species oj'
Erythrina. Tlzr grrzus rr~aybe characterized tnxogenrtically as n honaogamic corr~plex.Interspecijc hybrids are
intermediate betzoeerz their parrrztal species in morphological tmits, irzcluding naacroscopic fentures of the
ir~jlorescrncea n d j o ~ a e rarid rr~icroscopicjkatures oj'the leuf epirl~rrr~is.
The inheritance oj'particrtlar,fintrtres qf
the male parent in thr progrrzy allows ,for cor~jrmatiorz oj'hybridity. Species of srct. Erythrina nre generally
allopatric, b u t j r l d studies of rzatural poprtlations in southern lIexico rrvenlrd srvrral localitic~szohere two species
do occur sympatrically arzd where rzatural hybrids are jbrtnd. Trnplirzirzg hummingbirds, the pollinators oj'sect.
Er)-thrina, are implicated as thr agerzts oj'intrrspecijc hybridization among synapatric species. Tlze results qf'
experinaerztal hybridization, together with studirs oj'comparati~!emorphology and distribution pnttrrns, suggest
thnt sorrle species ofEr)-thrina are stabilizrd hybrid derivntiues.
Experimental hybridization studies h a r e beer1 a
cornerstone of research in ~ l a nbiosystematics
t
since
the emergence of this synthetic field. Much of the
work of early biosystematists was directed toward
efforts to define taxa and taxonomic categories on
the basis of reproductive barriers as revealed by
experiment, a n approach exemplified by the studies
of Clause11 et al. (1939, 1940) and their proposal
' Tlzis study wns undertaken as part of n doctoral dissertation at IVashirzgton lnil!ersity, St. Louis, ~ V I i s s o ~ ~ r i .
I'eter Raven j r s t suggested the research on Erphrina biosystenaatics as a dissertation topic arid helped in
irznurr~erablr ways during the research. His ideas on plarzt evolution arzd the role of hybridization irz the
evolrctionury process provided the concrptual jiarr~eworh.upon zohich this study is based. Rupert Barrzeby and
the late B . A. Krrtkoff i ~ ~ t r o d ~ L cnae
e d to the the systerr~aticsoj'Erythrina nrzd suggested rrlatiorzships nnaong the
t a r n oj'particular interest jbr the rxprrimental studies. Tlzis work laas rrlnde possible b y the generous support
o j ' t h ~three Hawniinrz botarlicnl gardens, nrzd especially b y the efforts o f t h e i r respective directors and chief
horticultrtrists: Ei'lliarr~ Throbald and Scott Lrtcns nt PaciJic Tropical Botarzicnl Garderz; Keith fiolliarns and
a
Cecilia L ~ b n oat Wnimen ..irboretum; arzd I'aul Weissich arzd Darziel 1WcGuire at f l o ' o m n l ~ ~ h iRotnrzic
Garden.
Gerald Carr nt the lniversity oj'l$awaii provided usr oj' his cytological laborntory and excrllrrzt advicr on the
har2dlir2g oj'chromosonaes. AIIicl~aelVeith provided assistnrzce and advice with the scr~rznirzgelectron naicroscope
at lTushingtorz lniversity. Hiroshi Tobe prepnred the lenf sectiorls illustrated in Figures 1 9 , 2 1 , and 23.
Conversatior~s~ a i t hand conamrrzts b y HGctor Hrrncindez arzd Peter Hoch helped to improvr the mnrzctscript.
Alinn Chaccir~arzd IIn,;l illere110 nssistrd in the preparation o f t h e rr~arzuscriptn n d j g u r r s . As a grndrtntr sturlerlt
I zoas supportrd b y j + l l ~ ~ ~ s / / i p ~the
, f iIlartforth
or~~
Foundation. LVashingtorz lniversity Ijivision oj'Riology and
Riorr~rdicalSciences, and ~VlissouriBotnrzicnl Garden. The research zoas ji~nded b y grants frorrl the Natiorzal
Science Ebrtndatior~ (IjBR 81-20386) a r ~ dEliznbeth ib-eill.
? AIIissouri Botarzicnl Garden, St. Louis, AIIissouri 6 3 1 6 6 , 11S.A.
Volume 75, Number 3
1988
Neill
Ery thrina
to establish ecotypes, ecospecies, coenospecies, and
comparia as universal units of classificatiorl to replace the traditional ones. The criteria used to
define taxa on the basis of fertility relationships
and the enormous labor required to obtain the
experimental results on a broad scale proved impractical for a general-purpose classification system, and most present-day biosystematists have
rightly abandoned the earlier efforts to "meddle"
with the traditional taxonomic hierarchy.
Experimental hybridization studies continue to
play a central role in research on the nature of
species relationships, however, and their usefulness
extends far beyond the requirements of formal
taxonomy. They provide the material for a broad
spectrum of integrated studies in the genetics of
evolutionary divergence. Research on long-lived
perennials, although it requires patience and a longterm commitment of labor and resources, is particularly amenable to hybridization programs because the parentals and several generations of
offspring can be grown side-by-side, and many comparative studies can be carried out with the living
plants.
This paper describes the results of experimental
investigations into the biosystematics and reproductive biology of Erythrinrc (Phaseoleae), a widespread genus comprising more than 110 species,
most of them tropical trees. The research was
concentrated on species of sect. Erythrinrc, a complex of 36 species centered in southern Mexico
and Central America (hlesoamerica), but other taxa
of Erythrinrc were included in some phases of the
investigation. The results of the research a r e used
to establish a series of hypotheses regarding species
relationships and evolutionary history of Erythrinrc. The hypotheses are presented here in sequence: each is dependent on the validity of the
previous hypotheses.
gamic complexes with weak to moderate reproductive barriers between the complexes.
Hypothesis 1. The species of sect. Erythrinrc
can all hybridize freely with each other, and the
resulting hybrids are as fertile as the parents. The
section is a homogamic complex, and internal, postmating isolating barriers between the species are
absent.
Hypothesis 2. T h e interfertile homogamic
complex encompassing sect. Erythrinn extends, to
a greater or lesser degree, to species in other sections and other subgenera of the genus. Any diploid
Erythrina species can mate with any other to form
a viable F, hybrid, but hybrids between widely
divergent species may exhibit varying degrees of
sterility. The genus as a whole may be characterized taxogenetically as a series of interfertile homo-
Hypothesis 3. The widely foraging hummingbirds that pollinate trees in sect. Erythrina ensure
effective outcrossirlg even in the isolated populations of small neighborhood size and low density
characteristic of these species. Self-compatibility
and occasional autogamy allow establishment of a
population from a single founder individual. W h e n
two species of sect. Erythrinrc are sympatric, the
pollinating birds do not discriminate between them,
and interspecific pollen transfer is likely.
Hypothesis 4 . The species in sect. Erythrinn
are mostly restricted in geographic range and are
usually allopatric, separated by habitat differences.
For the most part, these factors are effective barriers to interspecific gene flow. However, sometimes different species do come into contact in
nature, and then fertile hybrids are formed.
Hypothesis 5. Patterns of distribution and
phenetic variation in sect. Erythrincl indicate that
some distinct forms recognized as species are stabilized derivatives resulting from hybridization of
two parental species. I n the changing climates and
dynamic geomorphologic landscape that have characterized Mesoamerica since the Miocene, and with
the consequent migration of vegetation types and
mixing of floristic elements, formerly allopatric
species may have come into contact a number of
times. With the temporary breakdown of external
isolating barriers, the interfertile species hybridized, and the subsequent segregation and stabilization of hybrid derivatives has contributed to the
proliferation of Erythrinn species in h'Iesoamerica.
The first two hypotheses can be tested directly
by experimental hybridization programs. The third
and fourth can be substantially confirmed by obserrations of mating behavior and patterns of variation in natural populations. The fifth hypothesis
is historical and can only be inferred by drawing
on information obtained by testing the first four.
The "level of confidence" (Gottlieb, 1972) in the
final hypothesis of hybrid speciation in Erythrinn
is dependent upon the strength of the evidence
presented in this paper in support of the four antecedent hypotheses.
This research was made possible by the existence
of the extensive living collections of Erythrina a t
three botanical gardens in Hawaii: Pacific Tropical
Botanical Garden in Lawai, Kauai; Kraimea Arboretum in Haleiwa, Oahu; and Ho'omaluhia Botanic Garden in Kanehohe, Oahu. The cultivated
Erythrinrc collections were assembled, beginning
Annals of the
Missouri Botanical Garden
in the early 1970s, through the efforts of the
Erytl7rirzn morlographer B. A. Krukoff. The gardens collectively now have in cultivation more than
9 0 of the 1 12 recognized species in the genus, and
the remaining species a r e gradually being obtained
through requests to botanists around the world for
seed.
The following section provides a n overview of
the conceptual framework of the experimental work
on Erythrina and a literature review of the role
of hybridization in the evolution of homogamic
complexes.
Grant (1 9 5 3 ) coined the term "hybrid complex"
for groups of related species linked by occasional
or frequent hybridization, and he classified different
types of hybrid complex based on their reproductive mode and the means of stabilization of the
hybrids. In two of these complexes the hybrid derivatives are mostly or entirely apomictic: in a
clonal complex the hybrids are sterile and reproduce vegetatively, and in an agamic complex they
reproduce by agamospermy. In the remaining three
types of hybrid complexes, the hybrid derivatives
reproduce sexually: (1) in a heterogamic complex
they are permanent structural heterozygotes or
permanent odd polyploids; (2) in a polyploid complex they a r e amphiploid with respect to the parental species; and ( 3 ) in a homogamic complex
the hybrid derivatives exhibit normal meiosis and
are sexual diploids, homoploid with respect to the
parental species.
I n some groups forming homogamic complexes,
internal r e p r o d u c t i ~ ebarriers may be present and
the hybrid derivatives may be partially intersterile
with the parents and with each other, as revealed
by Grant's studies of annual Gilin (summarized in
Grant, 1 9 8 1). More frequently, though, particularly in complexes of perernlials and woody plants,
the derivatives are highly interfertile with the parents, with each other, and with all the other species
in the complex: the only barriers to gametic exchange between any populations or any pair of taxa
in the group are external.
Grant ( 1 9 5 3 ) and Gottlieb ( 1 9 7 2 ) pointed out
a paradox inherent in the recombinational system
of the homogamic complex that sets it apart from
the other types of hybrid complex. In clonal, agamic, heterogamic, and polyploid complexes, the cytogenetic features or reproductive systems of the
hybrid derivatives a r e important criteria of hybridity and distinguish them from the parents. In
homogamic complexes the derivatives are fertile
and cytogenetically homogeneous with the parents,
so these criteria cannot be used as a test of hybridity. Consequently, homogamic complexes a r e
more difficult to analyze and may pass undetected.
Grant ( 1 9 5 3 ) contended that in the long term
the "evolutionary potential" of homogamic complexes is much greater than in other types of hybrid
complex. I n the clonal, agamic, and heterogamic
complexes, and to a lesser extent in polyploid complexes, favorable gene combinations are stabilized
in the hybrid derivatives a t the cost of a severe
restriction in recombination. When environmental
conditions change, the derivatives a r e less flexible
in their capacity for genetic adaptation than are
sexual diploids. Should the progenitors of the complex, the original sexual diploids, become extinct,
an important source of new variation in the complex is lost.
These restrictions do not apply to homogamic
complexes, however. Since the derivatives are sexual diploids, recombination is unrestricted. They
are able to backcross freely with the parentals, and
the original species may become extinct without
jeopardizing the evolutionary potential and flexibility of the complex. Relative to the other types
of hybrid complex, the homogamic complex is, in
the words of Grant, a n "open-ended genetic system."
The maintenance of the ability to hybridize gains
importance because it extends the pool of natural
variation available for recombination and selection
far beyond that present in any single species. I n
environments undergoing climatic a n d / o r geologic
change, that "extra" genetic pool may be crucial
for adaptive adjustment of the organisms, and hybrid recombinants from two or more species may
have greater fitness in the newly created habitats
than either of the parental species. Evolution in a n
open-ended homogamic complex may follow a reticulate pattern, with cycles of divergence and differentiation alternating with hybridization and recombination as environmental conditions change
(Raven & Raven, 1 9 7 6 ; Raven, 1980).
The paradox is that while hybridization is most
difficult to detect and analyze in homogamic complexes, on a broad scale homogamic complexes may
be much more important in plant evolution than
other types of hybrid complex. Grant ( 1 9 5 3 ) even
speculated that hybridization in homogamic complexes may account for much of the diversity of
the angiosperms, and for the reticulate nature of
variation and lack of clear discontinuities between
the major phyletic lines of flowering plants. Grant
concluded that "the ancestral stocks may have
Volume 75, Number 3
1988
Neill
Erythrina
been hybridizing on the diploid (or diploidized) level
since the earliest stages of angiosperm evolution."
Whether or not homogamic complexes have
played such a n important role in the evolutionary
history of flowering plants, it is now well accepted
that they a r e characteristic of the genetic structure
of many large and ecologically dominant genera of
trees and shrubs, a t least in temperate regions. The
only really thorough biosystematic study of a homogamic complex in a genus of woody plants, combining fossil evidence, experimental hybridizations,
and careful field studies, is Nobs's ( 1 9 6 3 ) exemplary work on Cenrzoth/~sin California. h'Iany
are
~ s dominant shrubs in the
species of C e r c n o t f ~ ~
chaparral vegetation of that region, and all are
diploid with n = 1 2 . Nobs showed that since the
Miocene, certain wide-ranging species in Cennothus sect. Cernstes have formed hybrid swarms in
areas where they have intermixed. In novel habitats
created by a n increasingly arid climate and by the
exposure of new substrates such as serpentine outcrops, some of the hybrid derivatives have become
stabilized as new self-perpetuating species.
Numerous studies have also been carried out on
natural hybridization in Quercus, and this enormous hornoploid genus (12 = 12), which dominates
the forests of much of the north-temperate zone,
is generally agreed to comprise a homogamic complex (h'Iuller, 1 9 5 2 ; Hardin, 1 9 7 5 ; Van Valen,
1 9 7 6 ) or, perhaps more accurately, several homogamic complexes corresponding to its subgenera,
with strong but incomplete barriers between them.
The results of Cottam's long-term Quercus hybridization program (Cottam et al., 1 9 8 2 ) considerably strengthen the experimental evidence (most
of the previous hybridization studies in the genus
merely analyzed morphological variation in natural
populations).
Among other genera of trees and shrubs that
probably comprise extensive homogamic complexes a r e E u c n l y p t / ~ s(Pryor, 1 9 5 9 ) , Prosopis
(Simpson, 1977), and Kibes (Keep, 1 9 6 2 ) .
Most of the world's flora is made up of tropical
woody plants, and the role of hybridization and the
presence of homogamic complexes in these groups
is largely unknown and remains a matter of dispute.
Many systematists who work on tropical woody
genera evidently believe that hybridization is absent
or unimportant in the organisms they study, e.g.,
Ashton's ( 1 9 6 9 ) comments on Dipterocarpaceae
in Southeast Asia. Ehrendorfer ( 1 9 7 0 ) thought that
narrower "niche width" restricted gene flow, and
a higher incidence of polyploidy and apomixis in
tropical tree species made them much less likely
to hybridize than their temperate-zone counter-
parts. It does appear logical that in species-rich
tropical forests population density is low and neighborhood size is small for any one species, as well
as for groups of sympatric congeners, so the opportunities for hybridization may be fewer and the
hybrids harder to detect than, for example, in a
temperate forest with large populations of sympatric Quercus species.
At any rate, the critical experimental hybridization trials have not been carried out for tropical
woody plants, except for a few economically important genera. For example, strong sterility barriers have been found between Amazonian species
of Theobrorrln (Addison & Tavares, 1952). I n contrast, Hevea in the same region is probably a homogamic complex. In this homoploid genus (12 = 18),
fertile hybrids were easily obtained in experimental
gardens, and numerous natural hybrids were reported (Seibert, 1 9 4 7 ) .
I n many genera of tropical woody plants, all or
most species share the same relatively high chromosome number and may be considered diploidized
paleopolyploids. Thus, most genera of Bignoniaceae have the same chromosome number of 12 =
20; they are probably paleohexaploids based on
x = 7 (Goldblatt & Gentry, 1 9 7 9 ) . Such plant
groups may be prime candidates for the formation
of homogamic complexes. Until the present study,
however, a thorough biosystematic investigation on
the scale of Nobs's work on Cercrtothus had not
been carried out on any large tropical woody genus
or, in fact, on any other woody genus.
Erythrinrc L. comprises about 1 1 2 species distributed throughout the tropical regions of the world
and extending into warm-temperate areas such as
South Africa, the Himalayas and southern China,
the Rio de La Plata region of Argentina, and the
southern United States (Krukoff & Barneby, 1 9 7 1 )
(Fig. 1). Most species are trees or shrubs, but about
1 0 species occurring in climates with pronounced
dry and/or cool seasons are perennial herbs with
large woody rootstocks. Erythrinn species occur
in a very wide variety of habitats, from lowland
tropical rainforest to very arid subtropical deserts
to highland coniferous forests above 3 , 0 0 0 m.
The distinctiveness of Erythrinn has long been
recognized by legume systematists. Following Bentham (1865), the genus has been placed traditionally in the subtribe Erythrininae of the tribe Phaseoleae, a relationship based principally on the
characteristic trifoliolate leaves that Erythrina
Annals of the
Missouri Botanical Garden
ASIA & O C E A NIA
e
NEOTROPICS
70
Species
31 Species
FIGURE 1 .
k
Gi
dlv
Distribution of' Erythri~ia.
shares with the rest of the Phaseoleae. In a recent
generic treatment of Phaseoleae, Lackey ( 1 9 8 1 )
maintained the traditional classification but remarked that "the relationship of [Erythrir~a]to
the remainder of the Papilionoideae is an absolute
mystery . . . the genus would have long ago been
accommodated outside the Phaseoleae had not the
foliage suggested this tribe. In Inany significant
characters, Erythrir~a stands alone arnong the
Phaseoleae."
Besides Erythrirza, the subtrihe Erythrininae
w,
B ucontains the genera I ~ ~ r c u r ~Stror~g-ylodon,
tea, Apios, Spatlzolobus, Cochlianthus, Khodopsis, and n'eorl~dolphiw. The relationships arnong
these genera are not close and the erection of the
subtribe is largely a matter of convenience to accommodate a loose assortment of genera not easily
placed in other subtribes of the Phaseoleae.
Butea alone, the only other arborescent genus
in the Phaseoleae, appears to have some true affinities with Erytlzrinw. Baretta-Kuipers ( 1 9 8 2 )
found that the unusual wood anatomy of Erythrina
is very similar to that of Br~tew.In many other
important traits, however, Erythrina is distinct
from B l ~ t e aand from all other legumes. The base
chromosome number of x = 2 1, shared by all 86
Erythrir~aspecies that have been counted, is unique
in the Legurninosae and indicates no direct relationship with Blrtew with 11 = 9. The unusual, high
activitylow affinity nitrate reductase system present in all Erythrir~aspecies that have been examined differs in some respects from known nitrate
reduction patterns in other angiosperms (Orebamjo
et al., 1982). Neither IrI~rcl~na
nor Hutea ( G . R.
Stewart, pers. comm. to P . Raven, 1 9 8 4 ) shares
this trait with Brythrina. The Erythrina alkaloids,
structurally complex isoquinolines nearly universal
in the seeds of the genus, are found in no other
legumes (Mears & Mabry, 1971).
Although Erythrir~ais quite distinct from the
rest of the Legurninosae, and despite its great ecological and morphological diversity, the cytological
and phytochemical evidence cited above and the
interfertility relationships presented in this paper
indicate that the genus is unusually close-knit for
its size (Raven, 1974,), and there is no doubt that
it is monophyletic. As such, the genus is a n ideal
subject for the biosystematic study of diversification
of a n entire evolutionary clade.
The origin of Erythrir~a,like its relationship to
the rest of the Leguminosae, is obscure. No fossil
record of the genus has been reported. In the light
of its distribution patterns, pollination, and dispersal
mechanisms, and the known history of Leguminosae as a whole, Raven ( 1 9 7 4 ) postulated a n
Upper Eocene to Upper Oligocene origin for the
genus (430-30 1n.y. BP), follo~+-ed
by ocean-drift
a n d i o r other long-distance dispersal among the
Volume 75, Number 3
1988
Neill
Erythrina
three principal tropical regions of America, Africa,
and Asia-Oceania. Much diversification of Erytlzrirza has occurred independently in Africa and
America and to a lesser extent in Asia.
The place of origin of Erythrir~wis unknown,
but South America appears most likely since the
majority of the putative ancestral groups (as considered by Krukoff &- Barneby, 1974,) within the
genus are found there. Africa is also a possible
candidate, since it likewise contains a number of
endemic groups. Although Erythrina almost certainly originated ?+-ellafter the breakup of West
Gond~+-analand,
it has a basically South ArnericanAfrican distribution that is shared by Inany angiosperrnous groups, including the Leguminosae
itself (Raven &- Axelrod, 1974'). The Erythrir~a
taxa in "tropical Laurasia," i.e., Asia and Mesoamerica, are clearly derived groups. In Mesoamerica the genus has undergone extensive recent speciation within a single lineage.
ones that hover while feeding. The corolla standard
of hummingbird-pollinated Erythrir~ais narrow and
conduplicately folded to form a "pseudotube," concealing the wing and keel petals as well as the
reproductive parts. The flower resembles the tubular corollas of many gamopetalous hu:nmingbirdpollinated plants, but in Erythrir~athe pseudotube
is not sealed on the ventral side where the margins
of the corolla standard meet. The inflorescence
axis of the hummingbird-pollinated species is erect,
and the flowers are oriented outward, providing no
perch for the hovering hummingbirds.
Only a srnall number of the 3 1 5 neotropical
hummingbird species are Erythrir~apollinators, and
these are all similar in size, bill length, and behavior.
The Erythrina pollinators are principally specialized "high-reward trapliners," nonterritorial species
that follow regular daily foraging routes between
widely separated individual plants (Neill, 1 9 8 7 ) .
The flowers of the hummingbird-pollinated Erythrina species are much more uniform in size and
shape than those of the passerine-pollinated species,
and this probably reflects the relative uniformity
of pollination rnechanisrns arnong the former group.
PATTERNS OF Dl\ ERSIFIC ITIOh: POLLIN 4TION
Erythrina species exhibit a great diversity in
floral structure, inflorescence orientation, fruit
morphology, seed coat coloration, and vestiture and
epidermal ornamentation of foliage and calyces.
The infrageneric classification of Erythrina is based
principally on these characters.
T h e diversity of floral structure reflects adaptive
radiation in Erytlzrina with respect to pollination
mechanisms. All Erytlzrir~w species have red or
orange flowers and copious nectar, and are adapted
to pollination by nectarivorous birds. There are two
distinct syndromes of ornithophily in the genus. All
4 2 Old World species and 1 5 of the 7 0 New World
species are pollinated by "perching birds" of seuera1 families in the order Passeriformes. Passerine
birds cannot hover efficiently or for any length of
time, and the inflorescences of passerine-pollinated
Erythrina are oriented in such a way that the birds
can perch while feeding on floral nectar. The corolla standard is usually broad, and the flowers are
open, with exposed reproductive parts. Pollen is
deposited on the feeding bird's breast. The flowers
of passerine-pollinated species of Erytlzrinw are
diverse in size, forrn, and orientation, which appears to reflect the variation in size, morphology,
and behavior of the pollinators, which range from
sunbirds and white-eyes weighing 8-1 0 g to orioles
weighing over 3 5 g.
T h e remaining 5 5 New World species of Erythrina (nearly half the genus) are pollinated by
hummingbirds (Trochilidae). Hurnrningbirds are the
most specialized of nectarivorous birds and the only
FRTIITS, SEEDS, AN11 IIISPERSAL
The diverse fruit and seed characteristics of
Erythrina species are indicative of adaptation for
different dispersal mechanisms. T h e putative ancestral species (Krukoff &- Barneby, 1 9 7 4 ) inhabit
coastal, estuarine, or riverine environments and
have dull brown floating seeds transported by
oceanic or fluvial currents. These species are effective colonizers: Erythrir~a,firsca and E. variegata both became established on the island of
Krakatoa a few years after the cataclysmic eruption of 1 8 8 3 (Guppy, 1 9 0 6 ) . Since the review of
Raven (1974), new anecdotal evidence has come
to light concerning the dispersal of these seeds and
their viability following long exposure to salt water.
A drift seed of Erytlzrir~an;arirgwtw was recorded
after a storm on the beach of Canton Island, a low
coral atoll a t 3's latitude in the western Pacific,
where Erytlzrir~adoes not occur. The nearest possible source for the drift seed is Sarnoa, 7 0 0 km
to the south. The seed, planted in Hawaii, grew
into a 2 0 - m tree (frorn herbarium label of negrnrr
35066, BISH).
Alone in Erythrir~a,E. subl~mbransof AsiaOceania has winged, wind-dispersed fruits. The
Tanzanian endemic B.greenu'ayi has unusual fruits
with narrow winglike ridges, but the fruits are heavy
and do not appear to be effectively wind dispersed.
Most of the putative derived species of Ery-
Annals of the
Missouri Botanical Garden
thrir~ahave bright red seeds, which persist in conspicuous display on the pods after dehiscence. Red
seeds have evidently evolved independently in seuera1 lineages of Erythrir~w;one or two species in
each lineage have -bicolored red and black seeds.
The red or red-and-black seeds are presumed to
be "imitation arils" (Ridley, 1 9 3 0 ) or mimetic berries. According to this theory, the? are eaten by
frugivorous birds attracted by the bright colors and
are dispersed when they pass through the digestive
tract unharmed, but there are few actual reports
of such "mistake" dispersals. Skutch ( 1 9 7 1 ) recorded a n overwintering yellow-throated vireo (Vireojlavifrons) eating red Erytlzrina seeds in Costa
Rica. I have seen this phenomenon on only one
occasion, ?+-henin March 1 9 8 3 in Chiapas, Mexico,
I observed a migrant wood thrush (Hylocichla
mustellir1a) ingest several red seeds of Erythrinw
folkersii displayed on the pods. A major auturnn
food itern of this bird in eastern North America,
before it migrates south, is the bright red, fleshy
fruit of Cornus jlorida, which E ~ y t h r i r ~ sweeds
resemble quite closely (E. Morton, pers. comm.).
Thus the dispersal of Erythrina seeds as mimetic
berries by "naive" migrant birds does seem to be
a real, though perhaps infrequent, phenomenon.
The "rnirnetic berry" theory is fraught with all
of the conceptual difficulties cornrnon to considerations of the evolution of mimicry. The alkaloids
in the seeds of Erythrina are toxic, and the deceived bird must survive the passage of the seed
through its gut if it is to produce subsequent generations of birds that will disperse subsequent generations of Erythrir~w.(The alkaloids are not released unless the seed coat is broken, and
frugivorous birds do not have strong gizzards to
grind seeds.) Additionally, the mirnic should be rare
relative to the model, and the deception must occur
frequently enough so that natural selection can act
upon it. The question of mimetic seed coloration
in Erytlzrina and other legumes is discussed in
McKey (1975).
FEATURES OF THE EPIIIERRIIS
A great variety of special epidermal structures
occurs in Erythrir~a,particularly on the abaxial
leaf surfaces. These include hairs of many types,
epidermal papillae and various "larnellae," and epicuticular wax deposits. The adaptive significance
of these features is not known, but they are often
diagnostic for particular species or species groups
and often aid in identification of sterile material.
Patterns of leaf epidermal features and their in-
heritance in interspecific hybrids are discussed later.
SIJBDIVISIONS OF ERYrI1RIY.I
The first formal subdivision of Erythrinw was
established by Harvey ( 18 6 l ) , with subsequent
treatments by Harrns ( 19 15), Louis ( 19351, and
Krukoff ( 1 9 3 9 ~ 1for
, the American species; 1939b,
for the Asiatic-Polynesian species). In the 19th
century a number of generic segregates were proposed based on the distinctive floral morphs of
certain groups of species: e.g., Chirocalys Meisn.,
Jficropterys Walp., Dl~chassaingia Walp., and
Hypwphorus Hassk. These segregates were treated
as sections or subgenera of Erythrina by later
monographers. The modern classification of the
genus ?+-as established by Krukoff &- Barneby
(1974), who recognized 5 subgenera and 26 sections. I accept their treatment as the systematic
basis for the present work; a few taxonomic changes
to be published later are anticipated in this paper
prior to their formal designation. The infrageneric
classification of Erytlzrina is sumrnarized in Table
1 . A list of the currently recognized species, with
authorities and with changes in synonymy made
since Krukoff & Barneby's (1974,) conspectus of
the genus, is included in Appendix I.
The sections of Erythrinn are well delimited
rnorphologically and biogeographically, and each
appears to be monophyletic. The subgenera also
are delimited by several good characters and appear monophyletic, except for the large and heterogeneous subg. Erythrinw, which includes 70%
of the species in the genus. The relationships of
the sections comprising subg. Erythrir~a to one
another still present a number of unresolved taxonomic and phylogenetic questions.
T h e following is a short narrative s)-nopsis of
the infrageneric classification of Erytlzrir~aand a n
outline of evolutionary and biogeographical trends;
in the discussion, the taxa used in the experimental
studies are emphasized.
Subgenus Ilicroptrryx is restricted to South
America, except for Erythrina,firsca in the rnonotypic sect. Duchassair~g-iw. With floating seeds
dispersed by ocean currents, E. Jirsca is the only
species in the genus to occur in both the Old World
and the New World. It is widely distributed along
coasts and rivers in the Neotropics and AsiaOceania, as well as in Madagascar and the Mascarene Islands, but its present native range does
not include continental Africa. It often occurs in
extensive pure stands in seasonal swamps. With its
Volume 75, Number 3
1988
TABLE1.
Infingeneric classijcation of' Erythrina
Sections
I.
Neill
Erythrina
Subg. 4licropteryx
1. Duchassaingia
2. Cristae-galli
3. lWirropteryx
11.
Subg. Erythrina
4. Suberosae
5. Arborescentes
6 . Hypaphorus
7. Brevijorae
8 . Edules
9. Stenotropis
10. Pseudo-edules
11. Leptorhizae
12. Erythrina
13. Gibbosae
14. Corallodendra
14a. Fidelenses
15. Cubenses
16. Olivianae
17. Caffrae
1 8 . Hurneanae
19. Acanthocarpae
111.
Subg. Tripterolobus
20. Tripterolobus
IV.
Subg. Chirocalyx
2 1. Bruceanae
22. Macrocyrnbiurn
23. Dilobochilus
24. Dichilocraspedon
25. Chirocalyx
V.
Subg. Erythraster
26. Erythraster
Distribution
Nurnber of
Species
America
Africa
Asia-Oceania
1
2
4
X
X
X
X (Madagascar)
X
wide distribution and presumably primitive features
(Krukoff & Barneby, 1974), 6.
f u s c a or a , f r ~ s c a like ancestor may represent the original progenitor
of the entire genus.
Section Cristwe-galli includes two species, E.
crista-galli, which forrns extensive populations
along the estuary of the Rio de La Plata in extratropical South America, and B.,falcata, which inhabits the "Yungas" forest of the eastern Andean
foothills and similar subtropical forest vegetation
in southeast Brazil. T h e four species of sect. 111c r o p t e r y x inhabit riverine or upland forests of the
Amazon and Orinoco basins and the Planalto of
Brazil.
Subgenus E r y t h r i r ~ a w
, ith 7 9 species in 17 sec-
tions, is distributed throughout the three major
tropical regions of America, Africa, and Asia, but
no single section occurs in more than one of these
areas. The subgenus includes all 55 of the American hummingbird-pollinated species in six different
sections which I believe to have heen derived from
passerine-pollinated groups by convergent evolution in several independent lineages.
Erythrirza speciosa of coastal Brazil, in the
rnonotypic sect. S t e n o t r o p i s , is geographically and
phylogenetically isolated from the rest of the hummingbird-pollinated species. The herbaceous, hurnrningbird-pollinated species of sect. L e p t o r h i z a e ,
endemic to central Mexico, are probably derived
directly from the passerine-pollinated shrubbyiar-
Annals of the
Missouri Botanical Garden
FIGURE
2 . Distributiort of Erythrina sect. Erythrina. The numbers ir~dicatethe numbers of species h.nou,n to
occur i r ~each geopolitical regior~bourtdrd b y the h r a l ; ~ black
.
lir~es.
borescerlt sect. Brev~floraeendemic to the same
of species is in nuclear Central -4merica: particuregion. In a parallel nianner, the Andean hunilarly in the Mexican state of Chiapas and in Guamingbird-pollinated sect. Pseudo-edules may be
teniala. Geologically, nuclear Central Arlierica is
derived fro111the Andearl passerine-pollinated sect.
much older than southern Central America. It has
E d u Les.
been conrlected to the North -4merican continent
The large Mesoanierican-centered, huniniing- since the Cretaceous, whereas southern Central
bird-pollinated sect. Erythrirzn (36 species) is closely
America was only a chain of volcanic islands until
allied with the remaining humniingbird-pollinated
the close of the Panamanian isthmus in the Pliocene
sections; Corulloclerztlra, with 9 species in South
(Raven 8r Axelrod, 1 9 7 4 ; Coney, 1982). It is most
Anierica and the -4ntilles: and the monotypic sec- probable that sect. Rrythrintr originated in nuclear
tions Gitbosne in southern Central Anierica, and
Central America following migration of its progenCubenses endemic to Cuba. The relationship of
itor from South Anierica, an event that could have
these advanced arborescent hurnniingbird-pollinatoccurred either before or shortly after the final
ed groups to the rest of the genus is not clear,
formation of the southern Central Arlierican larld
however.
bridge. Species of this section, which comprises
nearly one-third of the entire genus, inhabit nearly
Figure 2 shows the distribution of sect. Eryevery forested hahitat in the geologically active
thrinu and the number of species known to occur
in geornorphologically and politically delir~iited arld climatically corliplex hfesoarnerican region. In
contrast to Erythrirztr fusctr and other species that
subregions of its range. The greatest concerltration
Volume 75, Number 3
1988
Neill
Erythrina
form extensive rlionospecific stands, the species of
sect. B r y / h r i n u generally occur at low population
densities. Many have a restricted geographic range
and occur in a single vegetation type, in a rather
narrow altitudinal helt, or only on particular suhstrates: such as outcrops of calcareous rock. Synipatry among species in the section is rare, hut
when it does occur, natural hybrids a r e generally
found. -411 available evidence indicates that sect.
Ery/hrirzn is a n outstanding example of rapid adaptive radiation and speciation in the recent geological
past.
The remaining sections of subg. Erythrirztr occur in the Old World, and their affinities to the
Anierican sections are not apparent. The South
African endeniic sects. C u f r n e , Hurneanae, and
Acurzthoctrrptre are the only representatives of the
suhgenus on that continent. Certain floral, fruit,
and seed features of sect. Ccqffrue do suggest a n
affinity with the rlionotypic Mexican sect. Oliviarzae, but a plausible explanation of such a connection is difficult to imagine.
The Asian sects. Suberosue, Arborescerztes, and
Hyptrphorus are a n autochthorlous group with
niostly "primitive" features and do not appear to
be closely allied with the Arlierican and -4frican
sections of subg. Erythrirztr. The species of sect.
Suberosne possess one singularly "advanced" feature: complex reticulate "larnellae" formed hy epidermal cells of the abaxial leaf surfaces (this paper,
Section 5).
The monotypic suhg. fiipterolobus, consisting
of B. greenwuyi and endemic to a sniall area in
the Rift Valley of Tanzania, is a n evolutionary
anonialy. The three-winged follicular pod is unique
in the genus, while the flower, as Krukoff & Barnehy ( 1 9 7 4 ) indicated, seenis constructed from
disparate elements of different subgenera.
Subgenus Chirocnlyx, with 5 sections and 1 9
species, is restricted to sub-Saharan -4frica. Section
Chiroc,aly.v comprises 1 4 species which inhabit
environments as diverse as the Kalahari Desert,
the lowland rainforests of Cameroon, the vast savannas of the Sahel, and the montane forests of
eastern Zaire-a
radiation rerliiniscent of sect. Ervthrintr in Mesoamerica, although with fewer
species. The remaining sections in subg. Chiroctr1y.v are mono- or ditypic: each quite distinct
morphologically.
The h a 1 suhgenus is Erythruster, with 1 3 species
in the sole sect. Erythraster. It is basically a n Old
World group with two disjunct, derived species in
the Neotropics. Erythrirltr znriegatu, the coastalstrand, ocean-dispersed species, occurs from T a n zania and Madagascar around the shores of the
Indian Ocean and westward through Indonesia,
New Guinea, Polynesia, and Micronesia to the Marquesas. The rerliaining species inhabit upland areas,
irlcluding four in East Africa and one in Australia.
There is one enderliic species on each of the islands
or island groups of Madagascar, Java-Bali, New
Guinea, Tahiti, and the Hawaiian archipelago, and
each of these niay he derived independently fro111
E. cnriegtrta. The disjunct E. celutina, widely
distributed in dry forests of northern South -4merica, the Galapagos, and the Antilles, and its Cuban
endemic derivative, E. grisebuchii, form a distinct
species coniplex together with the Tahitian E. tnhitensis and the Hawaiian E. suncl~c,icensis.All the
species of this Polynesian-Neotropical corliplex have
"niinietic berry" red seeds: unlike E. t ~ n r i e g n t u
and most of the other species in sect. Bryt/zru,ster.
Brythrinu t'uriegtrtn is present on Tahiti but not
in Hawaii or the Neotropics. With this pattern of
distribution, it appears most likely that these Polynesia~~-Neotropicaldisjuncts were established following long-distance dispersal hy birds across the
Pacific, and not by ocean-drift of B. variegnlu or
a t'nriegtrta-like ancestor.
Erythrirla is well known to be relatively uniform
cytologically; polyploidy is rare, and aneuploidy is
unknown (Lewis, 1 9 7 4 ; Goldblatt, 1 9 8 1a, 1984).
The basic chromosonie nuniber of the genus is s =
2 1 , unique in Leguminosae. Of the 6 5 species
counted prior to the present study, 6 1 are diploid
( 2 n = 42), two are tetraploid ( 2 n = 8 4 ) , one has
reports of both diploid and tetraploid races: and
one has reports of hexaploid (212 = 1 2 6 ) and octoploid (2n = 1 6 8 ) races.
The base number for Phaseoleae and prohahly
for subtribe Erythrinineae is s = 1 1 , and reduction
to n = 1 0 is common in the tribe. E r y / h r i n n is
likely either a n allotetraploid based on n = 1 1
n = 1 0 or a hypotetraploid n = (1 1 x 2 ) - 1
(Goldblatt, 1 9 8 l b ) , and thus a paleopolyploicl ge-
+
IlUS.
\I ATERI 4L5 4ND RILTHOrIl
Floral buds were collected from trees in cultivation a t three botanical gardens in Hawaii: Pacific
Tropical Botanical Gardens in Lawai (PT); Waimea
Arboretum in Haleiwa (WA); and Ho'ornaluhia Botanic Garden in Kaneohe (HO). Floral buds and;
or seeds were collected from wild populations of
certain species in Mexico and Costa Rica.
Annals of the
Missouri Botanical Garden
For gametic counts and meiotic analyses, floral
buds in developmental series were fixed either in
3 : 1 ethanol: acetic acid or in 6 : 3 : 1 chloroforni:
ethanol: acetic acid, which generally provided better fixation. After 1-2 weeks in the fixative a t
room temperature, buds were transferred to 7 0 %
ethanol and stored below 5°C. Anthers were
squashed in acetocarmine with Hoyer's solution
added (Beeks, 1 9 5 5 ) to make permanent slides.
For somatic counts, seeds obtained from wild
populations were germinated on filter paper. The
priniary root tips were pretreated in 0 . 0 0 3 M
8-hydroxyquinoline for 4 hours a t room temperature: fixed in 3 : 1 ethanol: acetic acid for 2-12
hours, and hydrolyzed in lo'% HC1 for 1 0 minutes
at 60°C. Root tips were squashed in FLP orcein
(Jackson, 1973).
Slides were exaniined under phase contrast with
a Zeiss Universal microscope; chromosonial configurations were photographed with Zeiss MC63
equipment using Kodak Technical Pan film developed for high contrast.
Chron~osome-1'urnbers. Chroniosome counts and
voucher data are listed in Table 2 . For cultivated
material the original wild-collected voucher is cited
if it exists; if not, a voucher made from the garden
progeny is cited. All vouchers are deposited a t
Missouri Botanical Garden (MO) unless otherwise
noted.
The gametic count of n = 4 2 for E. umazorzicu
reconfirms earlier somatic counts of 212 = 8 4
(.4tchison, 1 9 4 7 ; Goldblatt & Davidse, 1 9 7 7 ) for
this tetraploid. This species is distributed throughout the northern -4mazon basin and in the Guianas,
but all the chromosome counts to date have been
obtained from populations in the Brazilian state of
MaranhZo. A more complete sampling of the species
range may reveal infraspecific variation in ploidy
level, as has been determined for other species with
polyploid strains.
My count of n = 2 1 for the tropical Asian
Erythrinn suberosa is diploid and agrees with 1 3
previous reports for the species. Mehra (1976),
however, reported n = 4 2 in three populations in
the western Himalayas, a t its geographic niargin
and altitudinal upper limit. As with E. nmtrzonic,tr,
a cytogeographic survey of ploidy level in E. suberosu is desirable.
With the exception of Erytf~rintrrnac,ropf~ylla,
the reniaining 2 2 chrorliosorlie counts listed in T a ble 2 are all first reports for species. All are diploid
( n = 21 or 2n = 4 2 ) except the octoploid E.
burtrrzu (TI= 8 4 ) . This Ethiopian endemic is closely
related to E. burttii, which ranges fro111 Ethiopia
south to Tanzania, and for which both hexaploid
( 2 n = ca. 126; Atchison: 1 9 4 7 ) and octoploid
( 2 n = ca. 168; Goldblatt, 1 9 8 l a ) counts have been
obtained.
Chromosome numbers are now known for 8 6 of
the 1 1 2 species of E r y t f ~ r i n arecognized here.
Eighty-one species ( 9 4 % ) are diploid, with the remaining 5 species (6%) polyploid or variable in
ploidy level. The polyploid species are all in different sections and are not closely related to one
another, with the exception of E. burttii and E.
burnnu. Polyploidization has thus occurred at least
four times independently in Erythrinu. Given the
rarity of polyploidy in the genus, it seems likely
that the closely related K b u r t t ~and
~ i
Y burarza
were derived from a common polyploid ancestor.
Chromosome counts have yet to be obtained
fro111 2 5 species of Grythrirztr. Eleven of these are
from Africa where three of the five known polyploids occur. Seven uncounted species a r e South
American, where E. a m t r z o n i c . ~is the only polyploid known. Polyploidy is unknown for Erythrirztr
in North and Central .4merica, where 4 2 of the
4 5 native species have been counted.
!lileiosis i n n i p l o i d Spec,ies. Chromosome size,
morphology, and nieiotic behavior were similar in
all species exaniined. Observations of individuals
of two typical diploid species, Erythrirztr berenices
(W.4 8 1 ~ 5 0 5 ) a n d E. m n c r o p h y l l a ( P T
7 5 0 4 2 0 0 0 1 ) , are described and illustrated here.
Observations of chromosome pairing at zygotene
and pachytene are desirable in nieiotic analyses
(Jackson, 1984), but these were not feasible in
Erythrirzn because of its high chroniosonie numbers. -4t diakinesis, 2 1 bivalents were regularly
formed. Each bivalent had either one or two terminal chiasniata. I n K. mtrcrol)hylla, the average
number of chiasmata per cell was 3 1 . 5 ? 1 . 8 4
out of 1 0 cells sampled. This accords with the
figures of 3 1 . 3 5 k 0 . 5 4 , 3 1 . 2 5 k 0.61, and
3 2 . 0 8 i 0 . 7 chiasmata per cell reported by Jalil
et al. ( 1 9 8 2 ) fcr, respectively, E. cariegntn, E.
resz~pinntu,a nd their F , hybrid E. x resupurc~ellii.
-4t early to niid diakinesis: the bivalents were
generally well separated, thus the gametic chrorliosorlie counts listed in Table 2 were usually made
a t this stage. Toward the later stages of diakinesis
and as the nucleolus began to disintegrate, groups
of two or more bivalents appeared clumped together in the cell (Fig. 3). Thin strands of chromatin
were frequently observed to stretch between bivalent~.
Volume 75, Number 3
1988
Neill
Ery thrina
TABLE
2 . Chromosome counts oJ' Erythrina species reported in this paper. Vouchers are housed at 1l!lO unless
otheru,ise indicated.
Species
E. amazonica Krukoff
E. batolobium Krukoff &
Barneby
E. berer~icesKrukoff &
Barneby
E. breviJora A. DC.
E. burana R. Chiovenda
E. cochleata Standlep
E. elenae Howard &
Briggs
E. Jorenciae Krukoff &
n=
2n=
Voucher Data
WA 7 6 ~ 4 4 9 cultivated.
,
Brazil. Maranhio: Lapela, N. 7: Silva
4238 (NY).
Missouri Botanical Garden, cultivated (from wild-collected rootstock).
Mexico. Guerrero: Filo de Caballo, 6 , 3 0 0 ft., oak forest, Neil1
5647.
WA 8 1 ~ 5 0 5 cultivated.
,
Mexico. Veracruz: Tlalnelhuayocan, H.
Perales s.n. in 1 9 8 1 (NY).
Mexico. Guerrero: H. Iltis 28655.
P T 740435001, cultivated. Ethiopia. Locality unknown: F. lMeyer
s.n. in 1974. Voucher from cult.: Neill 5716.
Costa Rica. Heredia: La Virgen, 1 0 km S W of Puerto Viejo, 2 0 0
m, Neill 5102.
WA 8 0 ~ 6 1 4 ,cultivated. Cuba. Cienfuegos: Centro de Investigacio'n
Forestal s.n. Voucher from cult.: Neil1 5078.
Mexico. Chiapas: Motozintla, Cerro Mozotal, 6 , 6 0 0 ft., Neill 5600.
Barneby
E. gibbosa Cuf.
E. globocalyx Porsch &
Costa Rica. Alajuela: Cordillera de Tilarin, upper Pefias Blancas
valley, below Monteverde reserve, 1 , 3 0 0 m, Neil1 5028.
Costa Rica. San Josb: Las Nubes, 1,700 m, hreill 5033.
Cuf.
E. hondurensis Standley
E. horrida A. DC.
E. leptorhiza A. DC.
E. macrophylla A. DC.
E. mexicana Krukoff
E. oaxacana (Krukoff)
Krukoff
E. pudica Krukoff &
Barneby
E. sacleuxii Hua
E. sigmoidea Hua
E. smithiana Krukoff
E. sousae Krukoff
E. suberosa Roxb.
E. tahitensis Nad.
E. tuxtlana Krukoff &
Barneby
HO 80.037, cultivated. Honduras. Tela: Hazlett s.n. in 1980.
Voucher from cult.: Neil1 5709.
Mexico. Oaxaca: 2 km E of Ixtlin, road to Yavesia, 2,030 m, 1M.
Sousa 12634 (MEXU).
Mexico. Mkxico: Municipio Ixtapaluca, old Hwy. 190, km 25,
8 , 3 0 0 ft., hreill 5646.
P T 750420001, cultivated. Guatemala. Sololi: Godinez, 6 , 1 4 5 ft.,
B. A. Krukoff 1975-4 (NY).
Mexico. Oaxaca: 1 4 mi. S W of San Jeronimo Miahuatlin, 4,800
ft., Neil1 5423.
Mexico. Oaxaca: 9 km N of Diaz Ordaz, road to Cuajimolaya,
7,700 ft., Neill 5409.
Mexico. Chiapas: 1 5 mi. E of Cintalapa, Hwy. 1 9 0 , 2,000 ft., Neill
5440.
WA 7 4 ~ 1 2 9 6 cultivated.
,
Kenya: Arabuko forest, near coast, Lavranos 11225 (NY).
P T 740192001, cultivated. India: locality unknown, cultivated, D.
A. Millington s.n. in 1974. Voucher from cult.: Neill 5715.
P T 740329001, cultivated. Ecuador. Guayas: Manglaralto. MacBryde & Herrera-1MacBryde 690 (NY).
Mexico. Oaxaca: 1 4 km S of San Miguel Suchixtepec, 2,100 m,
Neil1 5425.
WA 7 5 ~ 9 6 0 cultivated.
,
India: Matrimandir Gardens, cultivated.
Voucher from cult.: hreill 5273.
P T 770442001, cultivated. Tahiti: Manupa Ridge, 2,000 ft., Perlman s.n. in 1977. Voucher from cult.: Neill 51 77.
Mexico. Chiapas: 2 6 km N of Ocozocuautla on road to Malpaso,
2,100 ft., Neill 5486.
The characteristic clumping of two or three bivalents became even more common in metaphase
I. At anaphase I, long strands of chromatin were
frequently observed stretching between disjoining
chromosomes, even after the main bodies of the
chromosomes were separated by a considerable
distance on either side of the equatorial plate (Fig.
4).
Annals of the
Missouri Botanical Garden
1Oum
c
FIGURES3-8. 3-6. Meiosis in pollen mother cells, diploid species of Erythrina.-3. Late diakinesis, E.
berenices, W A 8 1 ~ 5 0 5 (n = 2 1 ) . Clumping of biva1ents.-4. Anaphase, E. berenices, W A 8 1 ~ 5 0 5 .Sticky
chromatin bridges stretch between disjoined chromosomes.-5, 6 . Anaphase, E. macrophylla, PT 750420002 (n
= 2 1 ) . Late disjunction ofsome biva1ents.-7.
1Vitosis in root tip cell ofdiploid E. horrida, Sousa 12634 (2n
= 4 2 ) . Somatic chromosomes are not clumped.-8.
Diakinesis in pollen mother cell oftetraploid E. amazonica,
W A 7 6 ~ 4 4 9(n = 4 2 ) . Two quadrivalents are circled.
Volume 75, Number 3
1988
Neil1
Ery thrina
The chromatin connections between disjoining
chromosomes appeared to be in all cases simply
"matrix bridges" caused by chromosome "stickiness" (Beadle, 1932). Based on observations at
subsequent stages, it is unlikely that any of the
observed "bridges" were true dicentric bridges or
any other configuration resulting from chrornosomal inversions or translocations.
Another frequently observed phenomenon was
late disjunction of one or several bivalents at anaphase I (Figs. 5, 6). One or two lagging bivalents
were often present even at late anaphase I when
most chromosome pairs were completely separated.
However, observations of cells at later stages revealed no evidence of nondisjunction or unequal
assortment of chromosomes.
In contrast to the chromosome "stickiness" and
secondary association of bivalents at meiosis, somatic pairing of homologues was not observed in
mitotic root-tip cells. A typical mitotic configuration
of the diploid species Erythrina horridn (2n =
4 2 ) is shown in Figure 7, where there is no evidence
of pairing or of sticky chromatin connections between chromosomes.
Meiosis in pollen mother cells of the Asian E.
vnriegntn (as E. irzdicn Lam.) was depicted by
Sundar Rao (1945) and in several additional Asian
species by Mehra (1976). Both reported postsynaptic secondary association of chromosomes at
m e t a ~ h a s eI and subsequent stages to be common
in some species. hlehra ( 1 9 7 6 ) reported aberrant
meiosis with 13-2 1 bivalents, 0-16 uniualents,
and 0-2 B chromosomes in a diploid strain of E.
suberosa, while a tetraploid strain of the same
species exhibited normal meiosis with 42,,. Jalil et
al. ( 1 9 8 2 ) reported normal meiosis with 21,, in the
artificial hybrid E. x resupnrcellii (E. re.~upinntn x E. vnriegntn).
Pollen fertility, as estimated by Alexander's double stain technique (Alexander, 1 9 6 9 ) was uniformly high in all Erythrirzn species examined.
Seventeen individuals belonging to eight species,
used as parentals in the experimental hybridization
trials, had a mean pollen fertility of 9 5 % (at least
5 0 0 grains counted per sample). Such high fertility
suggests that the chromatin "bridges" and late
disjunction of bivalents observed in most cells at
anaphase I are not indicative of meiotic aberrations,
do not result in a high frequency of aborted cells,
and therefore most cells receive the correct complement of 2 1 chromosomes following meiosis I
and 11.
of E. amnzonica was available for analysis (WA
7 6 ~ 4 4 9 ;P T 760356001). In several individuals
of this strain, 4 2 bivalents were observed in some
cells at diakinesis and metaphase I, while in others
one or two quadrivalents were clearly visible (Fig.
8). In contrast to the merely "sticky" postsynaptic
associations seen in the diploids, the quadrivalents
in E. anznzonic.~appeared to be true multivalents
resulting from synaptic pairing at prophase. The
configuration of this species at meiosis I, then, is
38-42,, and 2-O,,. The formation of occasional
quadrivalents did not disrupt normal disjunction,
however, as no cells at telophase I or subsequent
stages were observed with other than 4 2 chroinosomes.
At diakinesis and metaphase I of the octoploid
E. hurnnn, considerable clumping of chromosomes
was evident in all cells examined. With a large
number of chromosomes crowded together. the
configurations were not completely resolvable and
it was not possible to determine whether synaptic
rnultivalents were actually formed.
LVIeiosis in Polyploit1 Species. Among the tetraploid species of Erythrirrn, only a single strain
The postsynaptic secondary association of biv a l e n t ~at meiotic metaphase observed in diploid
Erytizrir~nspecies has been reported from many
other plant groups. According to a theory introduced by Darlington (1930) and amplified by Lawrence ( 1 9 3 l ) , secondary pairing is due to attraction
of homologous or homeologous chromosomes when
the degree of homology is not close enough to result
in synaptic pairing, and is presumed to be indicative
of allopolyploidy. In a recently formed autotetraplaid, homology between the two pairs of chromosomes will be nearly complete and a multivalent
will he formed at pachytene. In a n allotetraploid
or in a n "old" tetraploid in which the genes of
homeologous chromosomes have diverged to some
extent, two bivalents result; they may later form
a secondary association at metaphase I or late
diakinesis due to attraction between the homeologous chromoson~esmaking up the two bivalents.
Secondary pairing does not always occur, however;
it is a relatively loose association and does not affect
disjunction at anaphase I.
This interpretation of secondary pairing and its
relation to allopolyploidy has been borne out by
quantitative studies of the spatial distribution at
metaphase I of marked homeologous chromosomes
in the allohexaploid 7i-iticum nesticum (Kempanna
8. Riley. 1964). In other plant groups such a
rigorous quantification of secondary pairing has not
been obtained. but a number of workers have in-
Annals of the
Missouri Botanical Garden
ferred a history of polyploidy in groups with meiotic
secondary pairing, particularly for those with high
chromosome numbers suspected to be paleopolyploids. For example, Venkatasubban (1944), in a
cytological study of Bignoniaceae, found a base
number of n = 2 0 for the family and a presumed
ancestral base number of x = 1 0 , since up to 1 0
"secondarily associated" pairs of bivalents were
present at metaphase I in many species. The euidence from secondary pairing is not unequivocal,
however; in the case of Bignoniaceae, Goldblatt &
Gentry ( 1 9 7 9 ) believed n = 2 0 to be a paleohexaploid of a base number x = 7, with aneuploidy
from 12 = 21.
Both Sundar Rao ( 1 9 4 5 ) and Mehra ( 1 9 7 6 )
noted secondary pairing in Erythriraa, and the
latter author cited it as evidence for a n ancestral
lower base number for the genus. On the basis of
present knowledge, however, it is not possible to
state unequivocally that the observed meiotic patterns in diploid E r y t h r i r ~ aspecies are due to secondary pairing of specific homologous or homeologous chromosomes, and not simply to random
nonhomologous "stickiness" of chromosomal matrix material. Multivalent formation in E. rrmazorrica, a neopolyploid, does appear to be a result of
true synaptic pairing of homologous chromosomes.
The hybridization trials carried out in Erythrirrrr
(described below) reveal a high degree of structural
and genic homology in the chromosomes of all
species, and it is probable that virtually any Erythrina genome can combine with that on the same
ploidy level of any other species in the genus to
form a viable F, hybrid. Whether tetraploid E.
arnrrzonica is of autoploid or alloploid origin, then,
it must have two highly homologous sets of chromosomes. I t is somewhat surprising that more than
two quadrivalents are not usually formed in meiosis
I of E. r~rnazonicrr.It is possible that the species
contains a specific gene that suppresses multivalent
formation and promotes strict homologous pairing
of bivalents, similar to the P h gene which performs
this function in hexaploid Triticum aestivur7z (Riley
& Chapman, 1958).
hf4TER14LS AND XIETHODS
Experimental hybridizations and self-compatibility trials were conducted at Pacific Tropical Botanical Garden and Waimea Arboretum FebruaryJuly 1 9 8 2 and February-April 1 9 8 4 . Although
the living Erythriraa collections at the two gardens
share many accessions from the same sources, the
species complement of mature, flowering individuals was different at each garden. The use of both
gardens allowed a broader inclusion of taxa in the
experimental studies than would have been possible
otherwise.
In addition, self-compatibility trials and interspecific hybridizations were conducted with natural
populations of Erythrina chiapasaraa and E. goldtnrrraii at El Sumidero Canyon National Park in
Chiapas, Mexico in February 1 9 8 3 . The two species
are parapatric at El Sumidero and hybridize naturally (this paper, Section 6).
In all, 3 2 species were used in the interspecific
hybridization trials, in 1 5 5 hybrid combinations
including reciprocals. Species from throughout the
worldwide distribution of Erythrirzrr were used in
the trials; four of the five subgenera and 1 2 of the
2 7 sections were represented. The monotypic African subg. Tripterolobus was the only subgenus
not included. Eighteen species were tested for selfcompatibility. All species used in the trials are
diploids (12 = 2 1 ) except E. arnazorrbca, a tetraploid
(n = 42). Attempts were made to hybridize E .
rrmazoraica as the pollen parent with several diploid
species.
The hybrid combinations were selected to represent different "taxonomic distances" between the
female and male parental species: "narrow hybridizations" between species of the same section,
"medium hybridizations" between species of different sections in the same subgenus, and "wide
hybridizations" between species of different subgenera. The narrow hybridizations involved mostly
species within sect. Erythrbna. The medium and
wide hybridizations included crosses of sect. Erythriraa to other sections and subgenera, as well as
representative hybridizations not involving sect. Erythrbrra, selected to include the maximum taxonomic diversity and geographic range of the genus.
Constrabrlts on the Experimeratrrl Protocol. Shortly after the initiation of the pollination trials, certain
constraints imposed by Erythriraa breeding systems
became apparent. Other constraints were imposed
by the fact that the experimental subjects were
trees exposed to the vicissitudes of the weather and
to local, uncontrolled variation in other factors that
may affect reproductive success, such as soil fertility and moisture, and insolation. These considerations required a somewhat different experimental protocol and a different statistical treatment of
the results than has been customary with biosystematic studies of greenhouse-grown herbaceous
plants.
Volume 75, Number 3
1988
Neill
Erythrina
The proportion of fruit set in intraspecific and
interspecific pollinations was quite low (see Results,
below) and the incidence of postfertilization abortion of young fruits was very high. Pollir~ationsuccess and fecundity varied greatly among individuals
of the same species. Some trees were effectively
"female sterile": they produced no fruits either
spontaneously (i.e., from "open-pollinated" flowers) or from controlled pollinations. At the same
time, conspecifics and even individuals from the
same accession, which were presumably at least
half-siblings of the "female sterile" individuals, produced fruits spontaneously in abundance and produced fruits quite readily from both intraspecific
and interspecific controlled pollinations. For many
species only a single tree was available, so the use
of intraspecific outcrossing success rate as a control
was not possible for those species.
Another constraint was the often limited number
of flowers per tree that were accessible each day.
On many trees only one or a few inflorescences
producing three or four new flowers each day were
accessible for hand-pollination. The time required
to emasculate, isolate, and pollinate each flower
individually also limited the number of flowers that
could be treated each day. Another practical consideration was the amount of land and labor required for growing the hybrid progeny at the
Hawaiian botanical gardens. It was desirable to
obtain progeny of many different hybrid combinations, so with limited resources large F, families
of any particular combination could not be accommodated.
One positive aspect of Erythriraa reproductive
systems that influenced the experimental protocol
was the relatively high viability of the seed. Among
the hybrids, 45% of the seeds germinated and
produced healthy F, plants. This high viability meant
that large seed lots of any particular combination
were not necessary to ensure that at least some
progeny would survive to maturity.
The above considerations and constraints led to
the development of procedures designed to maximize the number of "narrow," "medium," and
"wide" hybrid combinations without undue emphasis on any particular combination. Once several
well-formed maturing fruits were produced for any
combination, pollinations of that combination were
ceased and new combinations were attempted.
When possible, species combinations were repeated
using several different individuals as female and/
or male parents. For self-compatibility trials as well,
pollinations within an individual were terminated
once several semimature fruits had developed.
I n the course of the pollination trials it soon
became apparent that certain individual trees of
several species were more fecund, successful female parents than others. To the extent possible,
pollination trials were concentrated on the more
successful females, within the limitations imposed
by the number of flowers available. Hybrid combinations or self-pollinations that failed to set fruit
were repeated up to 50 times or more, but for
many combinations fewer than 10 flowers were
pollinated due to limitations of time and available
flowers.
Pollir~atiorlTechniques: H y b r i d i z a t i o n . T h e
development of suitable techniques to emasculate,
isolate, and pollinate the flowers involved considerable trial and error. Nylon mesh bags of several
types were used initially to isolate the flowers, but
these proved to be too unwieldy, requiring elaborate, heavy wire frames or other means of support
so the mesh did not touch the flowers. Also, in
rainy weather the high humidity within the mesh
bags tended to cause all the flowers to abort.
A simple alternative technique that proved successful was to isolate each flower individually. Floral buds were emasculated at the latest possible
stage of development, i.e., on the day before anthesis and pollen release. The tightly closed corolla
standard was carefully peeled open, and the anthers
were excised with dissecting scissors sterilized in
95% ethanol between each emasculation. If a n
anther released pollen before removal, the flower
was not used in the experiment. Following emasculation, the corolla standard was folded back over
the pistil and sealed with plastic Scotch tape. This
effectively protected the stigma from any chance
pollen deposition and also prevented it from drying
out. The following day the corolla was reopened
and the standard excised. After pollination a small
cone of aluminum foil, formed over the point of a
pencil, was placed over the stigma and pinched
lightly onto the style. This helped to hold the pollen
on the stigma in the face of rain and wind, and
isolated the stigma from any other pollen deposition. The cap remained on the stigma throughout
the development of the fruit.
For the open-corolla, homogamous species of
Erythrirac~adapted to pollination by passerine birds,
this technique ensured that the pollen was applied
while the stigma was receptive. In most species
receptivity was signalled by presence of a wet,
sticky exudate on the stigmatic surface on the day
of anthesis. For the closed-corolla, protandrous
species (primarily sect. E r y t h r i n r ~adapted
)
to hummingbird pollination, the style had not elongated
fully and the stigma was not yet receptive on the
Annals of the
Missouri Botanical Garden
day following emasculation. The stigma was thus
pollinated prematurely by this method, but the
pollen held in place by the aluminum cap evidently
remained viable at least until the next day when
the stigma became receptive, and these species did
set fruit with premature pollination. An alternative
method, to wait two days following emasculation
to pollinate the protandrous species, yielded no
better results and was logistically more complicated.
Stephenson ( 198 1 ) presented evidence from a n
extensive literature review that in many plant
species, particularly massively blooming trees, only
a small fraction of pollinated flowers produce mature fruits; the majority are aborted at a n early
stage of growth before large amounts of nutrients
are channelled into them. The number of fruits
that can be matured is usually limited by resource
availability, not by pollination. Furthermore, flower
and fruit abortion is selective: some species selectively shed self-pollinated flowers. They mature
fruits from self-pollinated flowers only when fruit
set is low and/or when "higher quality" fruits from
cross-pollinated flowers are removed. Therefore
there may be "mate competition" within a plant
among fruits of different paternity.
In this study attempts were made to reduce, to
the extent practical, the effects of resource limitation and competition on mating success. All flowers except the hand-pollinated ones were removed
from the inflorescence. Once a few fruits were set
on a n inflorescence, all flower buds were removed.
Until fruits were set, buds were left to develop into
flowers available for further pollination trials. Most
inflorescences bloomed continuously for several
weeks, producing a few new flowers each day, so
failed matings could be attempted repeatedly.
An individual inflorescence was treated with pollen from a single source. This eliminated mate
competition among the flowers within the inflorescence. An individual tree often had several inflorescences, each pollinated with a different species
of male parent, so there could have been interinflorescence competition among mates.
To further reduce resource competition and
channel available nutrients into the hand-pollinated
flowers, most untreated inflorescences and spontaneous, open-pollinated fruits (those accessible with
clipper poles) were removed from the crowns of
the trees.
In all species, seeds matured approximately 60
days after pollination. At maturity the hybrid fruit
was removed and the number of mature seeds,
aborted seeds, and undeveloped ovules was recorded. Length and width of each mature seed
were measured for comparison with seeds produced
from intraspecific matings.
Testsfor S e y Cotnpatibility, Autogamy, and d p o tnixis. For self-pollinations and intraspecific outcrosses, anthers were not emasculated, but the
corolla standard of the flower bud was sealed with
tape prior to dehiscence to prevent chance deposition of nonself pollen on the stigma. When the
stigma became receptive, pollen from the same
tree (for selfs) or from different conspecific trees
(for outcrosses) was applied, and a cap of aluminum
foil was placed over the stigmas in the same manner
as in the intraspecific hybridizations.
The pollen for the outcrosses was a mixture from
all available conspecific trees in the botanical garden, including individuals from the same accession
as the female parent as well as from different accessions. For the self-compatibility trials carried out
in the natural populations of Erythriraa chiapasans and E. goldtnanii at El Sumidero, the pollen
for the outcrosses was a mixture of at least five
different individuals in the population.
The treated flowers of a n individual inflorescence
were either all selfed or all outcrossed to eliminate
within-inflorescence mate competition.
The abortion of young fruits during the first two
to three weeks following fertilization was very high
for selfs and intraspecific outcrosses, as well as for
interspecific hybridizations. Fruit set data were taken at least four weeks following pollination, after
which abortion of the developing fruits was negligible. Complete data on intraspecific reproductive
success, including mature seed production and seed
size, germination success, and viability of the progeny, were obtained only for cultivated Erythrina
and E. cristr~-galli.Several indignate~naler~,si,s
viduals of these species from different accessions
were available for the trials, and they were the
most successful female parents in the interspecific
hybridizations. For comparative analyses, then, the
intraspecific data were particularly desirable for
these two species. Because of space and labor limitations, intraspecific progeny could not be raised
for all species.
In the flowers of sect. Erythriraa and the other
hummingbird-pollinated sections of the genus, the
anthers and stigma are positioned close to one
another. Initial observations indicated that, although the flowers are protandrous, autogamy may
sometimes take place. Autogamy was tested by
isolating entire inflorescences in wire-framed nylon
mesh bags. After all the flowers had either aborted
or set fruit, the mesh was removed. Six species
were tested this way at Pacific Tropical Botanical
Volume 75, Number 3
1988
Garden during a period of relatively dry weather
(May 1 9 8 2 ) to minimize abortion of flowers caused
by high humidity inside the mesh bags.
Autogarnous fruits were obtained only on the
most distal flowers (the last to open) on inflorescences of two individuals of Erythrir~r~
gnatetnrrlensis (Results, Table 4). These individuals were
tested for agamospermy. On three inflorescences
of each plant, all the flowers on the distal one-third
of the inflorescence were emasculated before dehiscence, and the stigmas were covered with aluminum foil caps to prevent any pollen deposition
on the stigma. Fruit set was monitored in the same
manner as in the pollination trials.
Statistical Analysis of Results. The protocol described above was necessitated by the flowering
patterns and reproductive traits of Erythriraa, by
practical limitations of breeding trees in the heterogeneous environments of open-air botanical gardens, and by the goal of obtaining viable progeny
of as many "narrow," "medium," and "wide"
hybrid combinations as possible. The resulting small
and very unequal sample sizes for different hybrid
combinations as well as for selfings and intraspecific
outcrosses meant that outcomes of particular combinations could not be compared statistically. Instead, for statistical analyses hybrid combinations
and intraspecific matings were pooled into broad
categories based on taxonomic distance (assumed
for the purposes of the study to be a true representation of relative genetic and phylogenetic distance) between the female and male parents. The
five experimental treatments are: self-matings, intraspecific outcrosses, and the three categories of
hybrid combinations-"narrow,"
"medium," and
"wide" hybridizations.
For statistical analyses, mating success for each
treatment was expressed as the proportion of handpollinated flowers producing mature fruit (i.e., a
fruit with at least one fully developed, normal-sized
seed). The commonly used analysis-of-variance
(ANOVA) tests (e.g., Sokal & Rohlf, 1969; Statistical Analysis Institute, 1 9 8 2 ) are designed to
test the significance of differences between means
of continuously variable data. ANOVA tests are
inappropriate for categorical (either/or) data, such
as mating success, where the outcome of a pollination attempt falls into one of only two categories.
A multiple comparison test for differences between
proportions, appropriate for categorical data, was
devised by Alan R. Templeton for these analyses.
Templeton's test allows for pairwise comparisons
of all combinations of the five treatment categories;
also, categories can be pooled to test various hy-
Neill
Ery thrina
potheses regarding mating success (e.g., all intraspecific us. all interspecific matings).
The null hypothesis for the test was that there
is no difference in proportion of mature fruits produced among any of the pollination treatments.
This is a corollary of the central hypothesis of this
research: that there are no interspecific or selfincompatibility barriers to mating within Erythriraa, that any pair of gametes from any species in
the genus are equally likely to pair successfully,
form a viable zygote, and grow into a healthy adult
sporophyte regardless of the infrageneric taxonomic position or putative phylogenetic distance between the parents.
Templeton's test is an inequality that compares
the differences between proportions with their variances. Proportions are subjected to a n arcsinesquare root transformation to set the variance independent of the mean; the variance is inversely
proportional to the sample size. The 9 5 % confidence limits of the proportion are:
where F = number of mature f r u ~ t s(successful
matings); 1 = number of flowers pollmated (attempted matmgs).
For small sample sizes (1< 50), the arcsinesquare root transformat~onIS corrected:
For the general case, the null hypothesis is rejected at P = 0 . 0 5 if the inequality is true:
where X, = the arcsine-square root transformed
proportion of successful matings in the bth category; 17; = sample size (total number of flowers
~ollinatedin the bth category); and a, = a weighting
a, I = 0 .
factor set so that
12
The CL, for each category is proportional to the
sample size :I):.
For pairwise comparisons between categories i
and j, the inequality is simplified; the H,is rejected
at P = 0 . 0 5 if it is true that
X X , l > 0 . 9 8 d i + i .
For a test of "highly significant" difference at
Annals of the
Missouri Botanical Garden
P = 0.01, the term "0.98" on the right side of
the inequality is replaced by the value "1.28."
Several individuals of Erythrirza gnatetnrrlrrzsis
and E. crista-grrlli were the most fecund, successful females in the interspecific hybridizations
as well as the intraspecific matings. For all categories employing these two species as female parents, separate multiple comparison tests were used
to compare mating success. These comparisons
included both selfings and intraspecific outcrosses
for E. guate~nalensisbut only selfings for E. crista-galli, which had only one individual in flower
at each garden (Pacific Tropical and Waimea) when
the pollinations were conducted.
Separate statistical tests were carried out for the
self-compatibility trials of individual species. For
those species with analyzable data on fruit set of
self-pollinations vs. intraspecific outcrosses, the data
were ordered into 2 x 2 contingency tables. With
small sample sizes and values of less than 5 in
many cells of the contingency tables, the standard
chi-square test was not appropriate; so Fisher's
exact probability was computed for the outcomes
(Sokal & Rohlf, 1969). For the pooled self-compatibility data including all species tested, the sample size was large enough for a chi-square test.
F, Hybrid Vir~l~ility.The F, hybrid seeds were
planted within a few weeks after harvest. To the
extent possible, seed lots of each hybrid combination were divided for propagation at two sites.
The F, plants were raised by the horticulturists at
Pacific Tropical and Ho'omaluhia Botanical Gardens; who monitored germination success, growth,
and vigor of the hybrids. Evaluations were made
approximately once each six months using a standardized form. Survivorship, growth rates, and indications of chlorosis or other abnormalities were
recorded.
The multiple comparison test described above
for analysis of fruit set was employed for a statistical evaluation of hybrid viability, defined as the
proportion of seeds in each category that germinated and survived as healthy plants for six months
(after which mortality in the garden was negligible).
Viability of the "narrow," "medium," and "wide"
F, hybrids was compared, together with that of the
"narrow" F, hybrids in sect. Erythrirza (see below).
Seeds from controlled intraspecific matings of Erythrirla guate~nalensisand E. crista-galli were
planted along with the hybrids. For each of these
species a separate multiple comparison test was
conducted for viability of "intraspecific" seed vs.
hybrid seed having these two species as female
parents.
F, Hybrid Fertility. Rlany of the narrow hybrids
between species in sect. Erythrirza, produced in
1 9 8 2 , grew to be 4-m trees and produced flowers
by February 1 9 8 4 . By several different measures
of fertility, these F,s were compared with the parental species and their meiosis examined. As a n
estimate of pollen fertility, percentage of stainable
(nonaborted) pollen was determined for the F, hybrids and their parents using Alexander's double
stain technique (Alexander, 1 9 6 9 ) (at least 5 0 0
grains counted per sample) and compared with a
one-tail t-test.
Fecundity o f F, Hybrids. As discussed earlier,
one goal of the hybridization study was to assess
the relative fitness of the hybrids in comparison
with their parents. The viability, vigor, meiotic
regularity, and pollen fertility of the hybrids are
indicators of fitness, but a more direct measure is
their relative reproductive success vis-B-vis that of
the parental species. During the period of this study
the F, hybrids, although some of them produced
flowers and fruits, did not grow into full-sized adult
trees, so a thorough assessment of hybrid fecundity
and fitness was not possible. However, a preliminary indication of reproductive success was obtained from the two-year-old narrow hybrids in
sect. Erythrinrc that flowered in the spring of 1 9 8 4 .
Controlled self-pollination of some of these F,
hybrids was conducted in order to obtain seed for
a limited number of F, families. The multiple comparison test was employed for pairwise comparisons
of mating success (proportion of hand-pollinated
flowers producing mature fruits) between the selfed
F,s and their parental species. Different categories
of parental matings varied among themselves in
mating success, and several types of parental matings were compared with the selfed F,s. The pairwise comparisons of fruit set included: 1 ) selfed
F,s vs. their own parental matings, i.e., the original
hybridizations that produced the F,s used in the
trials; 2) selfed F,s vs. all paired combinations of
the parental species, including the reciprocals that
failed to produce F, hybrids; 3) selfed F,s vs. selfed
parental species. The logic for using these particular groupings of parental matings in the comparative assessment of F, reproductive success is discussed in the Results.
Viability of F, Hyhrids. Many of the F,s also
produced fruit and mature seed spontaneously on
open-pollinated inflorescences, almost certainly the
result of autogamy. F, seed lots from selfed flowers,
and some from open-pollinated flowers, were planted along with the 1 9 8 4 F, hybrids. The viability
Volume 75, Number 3
1988
Neill
of the F, progeny was compared with that of the
F,s.
percentage of mating success in the pooled data
for all species-6% fruit set for selfings and 10%
for outcrosses-is due to a high incidence of postzygotic abortion of young fruits and to failure of pollen
tubes to reach the ovules, but the relative importance of these two factors is not known. For the
pooled totals, the difference in fruit set between
selfs and outcrosses is nonsignificant.
Statistical comparison of fruit set in selfs vs.
outcrosses was possible in four species. In only one
of these, the natural population of Er3.thriraa ,goldmanii at El Sumidero, was fruit set significantly
higher in outcrosses than in selfs, and then but
marginally so at P = 0.05.
In six additional species, self-pollinated flowers
set fruit, but only one individual of the species was
available, so the outcrossing control was not possible. In nine species, no fruits were set from selfpollinated flowers, but in four of these the outcrossing controls yielded no fruits either. Failure
of fruit set, then, is evidently a consequence of
overall low fecundity in Erythrirra and not of selfincompatibility per se.
Self-incompatibility has previously been reported for seven species of Erythrina: E. serzegalensis
and E. speciosa (East, 1940); E. crista-galli
(Fryxell, 1957); E. mitis and E. poeppigiana (Arroyo, 1981); E. leptorhiza (Hernindez & Toledo,
1979); and E. montana (Hernindez, 1982). Only
for E. montana was the assertion of self-incompatibility supported by evidence from experimental
self-pollinations and outcrossing controls. Calculation of Fisher's exact probability for the data
presented in Hernindez (1982), however, reveals
that the difference in fruit set between selfs and
outcrosses in E. montana is nonsignificant ( P =
0.25). My evidence for self-compatibility in E. senegalensis and E. crista-galli contradicts the earlier reports of self-incompatibility in these species,
which were based merely on the failure of isolated
cultivated trees to produce seed spontaneously.
Feinsinger et al. (1979) provided evidence from
experimentally controlled pollinations that E. j i ~ s c a
and E. pallida are self-compatible.
There is thus no reliable evidence for genetic
self-incompatibility in any species of Erythrina. It
appears safe to assume that genetic self-incompatibility-at least the classical single-locus, multiple S-allele, stigma- or style-mediated model of
self-incompatibility (Nettancourt, 1 9 7 7 ) p i s completely absent from the 1 1 2 species in the genus.
If this is true, it would invalidate some of the
evidence that Arroyo (1981) advanced to support
her assertion that tropical woody Papilionoideae
are predominantly self-incompatible. Five of the
Studies of Prcvionsly Synthcsizcd Hybrids. A
few Erythrirza hybrids have been produced in the
past by horticulturists and are commonly grown in
tropical and subtropical regions. Two were available
at the Hawaiian gardens: Erythrina x bidwillii
and E. x sykesii. Studies of meiosis, pollen fertility,
and fruit set from controlled self-pollinations were
conducted on these plants with the methods described above.
Hyhrid Nr~mes. In this paper horticultural convention (Brickell et al., 1980) is followed for the
hybrid names. For artificial hybrids when the female parent is known, the female parent is first in
the hybrid formula name. When the female parent
is not known, as in natural hybrids, the order of
the constituent species names is alphabetical.
RESULTS AND DISCUSSION
The results of the experimental hybridizations,
self-compatibility trials, and studies of viability and
fertility of the progeny are presented in summary
form for the statistical analyses of the data. In
addition, more complete data sets, listing the results
obtained from individual plants, are presented in
certain of the tables below.
There are several reasons for this more thorough
reporting of the data. The first is to provide the
most complete information available on the ancestry of each individual in the F, and subsequent
hybrid generations. The full documentation is necessary for the studies on the inheritance of various
traits in the hybrids. Studies of morphological inheritance were initiated in this paper (Section 5),
and research on the inheritance of micromolecular
and macromolecular traits in the interspecific hybrids is anticipated for the future. In addition, some
of the hybrid plants with their colorful flowers are
likely to be propagated widely as ornamentals; the
tables presented here serve as public documentation of the parentage of these cultivars. Finally, it
is hoped that some of these experiments will be
repeated with the same parental and hybrid trees
in the Hawaiian botanical gardens. The documentation of the results for individual plants of traits
probably indicative of reproductive success, such
as pollen fertility and fruit set, will allow investigation of the possibility that such traits may change
through time with the maturation of the plant.
SeFCompatibility. The results of the self-compatibility trials are shown in Table 3. The low
Erythrina
Annals of the
Missouri Botanical Garden
TABLE
3.
Selfcompatibility trials in Erythrina.
Self
Species'
E. berteroana (4)
E. chiapasana (2)
E. chiapasana3 (6)
E. crista-galli (1)
E. elenae (1)
E. falcata (1)
E. folkersii ( 1 )
E. fusca (2)
E. goldmanii3 (6)
E. guatemalensis (5)
E. latissima (1)
E. lysistemon (1)
E. perrieri (2)
E. sandwicensis (1)
E. senegalensis (1)
E. speciosa (2)
E. standleyana (1)
E. tahitensis (2)
E. oariegata (1)
Total
Outcross
Flowers
Fruits'
27
14
32
27
27
5
2
82
27
33
10
25
27
49
28
21
9
64
6
515
0
0
4 (0.13)
7 (0.26)
0
1 (0.20)
2 (1.0)
l(O.01)
3 (0.11)
3 (0.09)
0
4 (0.16)
0
2 (0.04)
2 (0.07)
0
0
0
0
29 (0.06)
Flowers
Fruits'
9
0
-
-
15
l(0.07)
Probability,
Self vs. Cross
-
-
0.48"
-
-
-
-
-
-
-
-
-
-
-
85
23
28
1 (0.01)
8 (0.35)
7 (0.25)
-
> 0.05""
0.05"
0.09*
-
-
-
-
-
-
1
0
-
-
-
-
-
-
8
0
-
6
175
-
-
-
-
0
17 (0.10)
-
> 0.05""
' In parentheses: number of individuals used in trials.
' In parentheses: proportion of pollinated flowers producing mature fruits.
' Pollinations conducted in natural population at El Sumidero, Chiapas, Mexico.
* Fisher's exact probability.
** Chi-square probability.
27 species Arroyo listed in that habitat/life form
category as self-incompatible were E r y t h r i n a
species. There are very few comparable studies on
other genera of tropical woody Papilionoideae. With
the information presently available, it is not known
if E r y t h r i n a is a n anomaly, or if self-compatibility
is common in this group of plants. Because low
fecundity and high rates of flower and fruit abortion
are probably characteristic of these plants, greater
caution is required in carrying out and interpreting
self-incompatibility tests than has customarily been
taken.
It is true that fruit set is frequently lower in
self-matings than in outcrosses. This may be due
not to genetic self-incompatibility, but rather to
multiallelic inbreeding depression, expressed either
in the progamic phase as failure of pollen tubes to
reach the ovules (Mulcahy & Mulcahy, 1 9 8 3 ) or
as postzygotic abortion of young fruits.
Although E r y t h r i n a species are genetically selfcompatible, the production of seed from selfed flowers in natural populations may be quite limited. A
flowering tree visited by pollen-bearing birds will
receive many geitonogamous pollinations (pollen
from a different flower on the same individual) as
well as xenogamous pollinations (pollen from a dif-
ferent individual). The reduced fruit production
from self-pollinations, as well as the relatively poor
viability of selfed seed (see section on F, viability
below) suggests that progeny derived from selfed
flowers are low in "quality" relative to progeny
derived from outcrossed flowers. The selective
abortion of low-quality selfed fruits, cited by Stephenson (1981), may be operative in E r y t h r i n a .
Interfruit competition may be very intense under
natural conditions, since such a small proportion
of pollinated flowers develops into mature fruits.
Therefore it is possible that most successful progeny are derived from outcrossing, and that the
level of inbreeding in most Erythrirza populations
is quite low in spite of self-compatibility and a large
proportion of geitonogamous pollinations. This is
still speculative; the significance to mating success
of mate competition among male parents has not
been explored in E r y t h r i n a .
I n regard to flower and fruit abortion, the attempts to increase mating success by eliminating
the effects of competition and resource limitation
in the experimental pollination trials were only partially successful. Certainly the fruit maturation rates
of 25%: or more obtained in some of the outcrossing
trials represent a n increase in fruit production over
Volume 75, Number 3
1988
TABLE4.
Neill
Erythrina
Tests for autogamy and agamospermy in Erythrina at Pacijc Troplcal Botanical Garden.
I. Test for autogamy
Species
E.
E.
E.
E.
E.
E.
E.
E.
E.
abysscnica
berteroana
crista-galli
guatrmalenscs
"
guatemalensis
gz~atemalensis
humeana
macrophylla
saloiijora
Accession Number
Inflorescences
Bagged
770034001
700044001
740283001
720999001
720999002
750419001
740283001
750420001
721000002
3
3
3
2
3
3
4
4
2
Flowers
Fruits
Flowers
Emasculated
Fruits Set
Seeds
11. Test for agamospermy (in individuals exhibiting autogamy)
Species
E. guatemalensis
E. guatemalensis
Accession Number
Inflorescences
Treated
750419001
720999001
3
3
the percentages found in natural populations. Usually the percentages of fruit maturation were much
lower, however, and in all cases the majority of
pollinated flowers were aborted early in development. The factors promoting flower and fruit abortion are several, including nutrition and resource
limitation, competitive effects, possible damage to
the flowers caused by emasculation, and factors
such as adverse weather conditions, in addition to
the factor under consideration here: the genetic
compatibility of the female and male parents. Neither for the self-compatibility trials nor for the
experimental hybridizations was it possible to sort
out all of these variables.
30
66
0
0
produce some fruits spontaneously in the absence
of evident pollen vectors.
The fact that autogamous fruits a r e produced
only on the latest flowers of an inflorescence suggests that the breakdown of protandry may be an
adaptive mechanism that allows some seed set in
the absence of the appropriate avian pollen vectors.
Each inflorescence, although it may produce 7 5
or more flowers, will mature only a few fruits, so
fruit set on the lower, earlier-blooming flowers of
the inflorescence must inhibit the formation of fruits
on the upper, later-blooming portion. It is likely
that autogamous fruits from the ultimate flowers
of the inflorescence will be produced only if some
allogamous fruits (from either xenogamous or geiAz~togamyand Agamos~rermy. The results of
tonogamous pollinations) have not already been
the tests for autogamy and agamospermy are shown
produced on the lower portion of the inflorescence.
in Table 4. Autogamous fruits were produced only
The two Erythrirra gr~atemalensis trees that
on two individuals of Erythrirra gr~atemalensis.
produced autogamous fruits were tested for agamoSignificantly, these were rather "fecund" trees with
spermy (Table 4). No fruits were produced when
relatively high mating success from controlled handstigmas were isolated from pollen deposition, so
pollinations. The autogamous fruits were produced
agamospermy is not indicated. Agamospermy is
only from the uppermost three fascicles of flowers
almost unknown in the Leguminosae (Arroyo, 1 9 8 1 )
on a n inflorescence-the
last flowers to bloom.
and it is unlikely to occur in Erythrirra.
They evidently were produced, in part, because of
the occasional breakdown of protandry, which pre- Hybridizatiorr 7iials: Mating Sz~c,cess (Fruit
vents autogamy on most flowers of E. gratema- :I!Zat~~ratiorr)
. The complete results of the hylensis and the other species that a r e adapted to bridization trials a r e listed in Appendix 11. From
hummingbird pollination (Neill, 1 9 8 7 ) .
1 , 6 7 1 hybridization attempts in 1 5 5 hybrid comAlthough autogamous fruits were produced in binations, 9 8 mature fruits were produced in 4 7
my limited trials only on Erythrinr~gz~atemalerrsis, hybrid combinations, for a n overall hybrid mating
I believe that occasional autogamy is widespread success of 6%, in 30%' of the attempted combiin the genus. Cultivated trees of many species nations.
Annals of the
Missouri Botanical Garden
TABLE
5.
Proportion of hand-pollinated powers producing mature fruit: all diploid Erythrina species.
Pollination Treatment
Flowers
Pollinated
Fruits
Matured
Proportion
Fruit Set
Selfed
Intraspecific outcross
Narrow hybridization (within section)
Medium hybridization (between sections, within subgenus)
Wide hybridization (between subgenera)
Total
515
175
540
350
705
2,285
29
17
50
22
25
143
5.6%
9.7%
9.3%
6.3%
3.6%
6.3%
Multiple comparison test for differences between treatments in proportion of fruit set
Self vs. outcross
Self vs. narrow hybrid
Self vs. medium hybrid
Self vs. wide hybrid
Outcross vs. narrow hybrid
Outcross vs. medium hybrid
Outcross vs. wide hybrid
Narrow hybrid vs. medium hybrid
Narrow hybrid vs. wide hybrid
Medium hybrid vs. wide hybrid
Intraspecific vs. hybrids
Outcross + narrow hybrids vs. self
' N.S.
=
+ medium + wide
N.S.'
P < 0.05
N.S.
N.S.
N.S.
N.S.
P < 0.01
N.S.
P < 0.01
N.S.
N.S.
P < 0.01
not significant ( P > 0.05).
For the statistical analysis of mating success in
selfs, intraspecific outcrosses, and hybridizations
(Table 5), data from the diploid species only were
included. The tetrapoloid Erythrirza amazonica as
male parent, after numerous pollination attempts
with the diploids E. g~~atemalerrsis
and E. cristagalli as female parents, produced one hybrid seed
with each of the females. Neither of the seeds
germinated, however, so there are no successful
hybrids between Erythrirza species of different ploidy levels. Since the results from E. amazonica are
not germane to the hypotheses of the interfertility
of diploid species and the formation of homogamic
complexes, they were excluded from the statistical
analysis.
The results for the diploid species in Table 5
indicate that the highest mating success was obtained with intraspecific outcrosses and "narrow"
hybridizations (within sections). "Medium" hybridizations (intersectional, intrasubgeneric) and selfings were intermediate in mating success, and
"wide" (intersubgeneric) hybrids were the least
successful of the five treatment classes. A general
trend, then, is evident: interspecific matings between closely related species (within sections) a r e
just as likely to succeed as intraspecific matings.
Mating success diminishes with increasing "taxonomic distance" between the parents (intersectional and intersubgeneric hybridizations). Mating suc-
cess is also somewhat lower in selfings than either
intraspecific outcrosses or hybridizations between
closely related species.
This overall trend, shown by the percentages of
fruit maturation in Table 5 , is not a strong one;
the differences between treatments a r e for the most
part nonsignificant. The multiple comparison test
revealed only three significant differences among
the ten possible pairwise combinations. Fruit maturation in narrow hybridizations was significantly
higher than in self-mating (P < 0.05). Intraspecific
outcrosses and narrow hybridizations also had higher
fruit maturation than wide hybridizations; in these
comparisons the difference was highly significant
(P < 0.01).
Because the pooled fruit maturation data for all
species may obscure the heterogeneity in results
among different species, it is instructive to examine
the patterns of mating success in a few selected
species. Erythrirza gz~atemalensisas female parent
accounted for 30% of all mature fruits in the
pollination trials and 33%' of all the hybrid fruits.
Thirty-three hybrid fruits were produced from E.
gz~atemalensisas female; of these, 3 0 ( 9 0 % ) were
from a single genetic individual, a clone represented
by one tree at each garden ( P T 7 0 0 0 1 8 0 0 1 , WA
7 4 ~ 1 4 5 3 ) The
.
pattern of mating success for Erythrina gr~atemalensis(Table 6 ) is very similar to
the overall results for the combined species trials.
Volume 75, Number 3
1988
Neill
Erythrina
TABLE6. Proportiorl of hand-pollir~atedjozilers producirlg mature fruit: female parerlt
malensis; male parerlts = diploid species.
Pollination Treatment
Selfed
Intraspecific outcross
Narrow hybridization (within section)
Medium hybridization (between sections, within subgenus)
Wide hybridization (between subgenera)
Total
=
Erythrina guate-
Flowers
Pollinated
Fruits
Matured
Proportion
Fruit Set
33
28
86
54
185
3
7
21
5
7
43
9%
25%
24%
9%
4%
386
11%
Multiple comparison test for differences between treatments in proportion of fruit set
Self vs. outcross
Self vs. narrow hybrid
Self vs. medium hybrid
Self vs. wide hybrid
Outcross vs. narrow hybrid
Outcross vs. medium hybrid
Outcross vs. wide hybrid
Narrow hybrid vs. medium hybrid
Narrow hybrid vs. wide hybrid
Medium hybrid vs. wide hybrid
Outcross
narrow hybrid vs. self
+
+ medium + wide
T h e E. g~~ntemnlerrsis
trees were unusually fecund;
fruit maturation was much higher than the overall
average for intraspecific outcrosses (25%) and narrow hybridizations (24%'). As in the combined
species results, fruit maturation was significiantly
higher in intraspecific outcrosses and narrow hybridizations than in wide hybridizations ( P < 0 . 0 1
for both pairwise comparisons), and for narrow
hybridizations vs. medium hybridizations the difference was marginally significant ( P < 0.05).
The data set for Erythrina crista-galli as female parent (Table 7), although not so extensive,
shows that the neat congruence of mating success
and taxonomic distance evidenced by E. guntemalensis does not always apply. Erthyrina cristagalli as female, represented by one genetic individual a t each of the two gardens, produced 1 6 %
of all the hybrid fruits in the trials, but it produced
49% of the "medium" and "wide" hybrid fruits.
Fruit maturation was higher in the medium and
wide hybridizations than in the few narrow hybridizations. (Only one species, Erythrirra fnlcnta, is
in the same section with E. crista-galli, so the
opportunities for narrow hybridization were limited.) There are no significant differences, however,
in any of the pairwise comparisons between pollination treatments for E. cristn-gnlli; the variances
are large because the sample sizes are rather small.
For intersectional hybridizations (the "medium"
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
P < 0.01
N.S.
P < 0.05
N.S.
P < 0.01
and "wide" categories combined) mating success
in Erythrina crista-galli as female parent was
significantly higher than in E. gr~ntemnlensis(multiple comparison test for proportions, P < 0.05).
This is illustrated by the results of attempted reciprocal hybridizations between these two species,
which are in different subgenera. Sixty-four pollination attempts to produce the hybrid E. guatemalensis 9 x E. crista-galli 8 yielded a single
fruit with two seeds, neither of which germinated.
Only ten attempts a t the reciprocal cross of E.
8 yielded four
crista-galli 9 x E. g~~ntemalerrsis
fruits, 1 5 seeds, and eight vigorous F, plants. I n
all, E. cristn-galli as female parent produced seeds
from seven hybrid combinations with species in six
sections and three subgenera. Five of the& combinations in all three subgenera survived as healthy
F, plants. Erythrina crista-gnlli, in short, was a
singularly successful "wide hybridizer."
Summing up the contrasting results in fruit maturation for Erythrirza gr~atemalensisand E. cristn-galli, E. gz~aternalensishybridized very readily
with species {n the same section, much less so with
more distantly related species. Erythrirrn cristngnlli, in contrast, hybridized with a number of
species in different subgenera with apparently equal
facility, regardless of the formal taxonomic relationships and presumed phylogenetic affinities between E. crista-galli and the male' parents. These
Annals of the
Missouri Botanical Garden
TABLE7. Proportion of hand-pollinated jfoluers
producing maturefruit: j6male parent = Erythrina crista-galli; male parents = diploid species.
Pollination Treatment
Selfed
Narrow hybridization
(within section)
Medium hybridization (between sections, within
subgenus)
Wide hybridization (between subgenera)
Total
Flowers
Pollinated
Fruits
Matured
Proportion
Fruit
Set
27
7
26%
22
2
9%
69
11
16%
64
182
10
30
16%
16%
Multiple comparison test for differences
between treatments in proportion of fruit set
Self vs. narrow hybrid
Self vs. medium hybrid
Self vs. wide hybrid
Narrow hybrid vs. medium hybrid
Narrow hybrid vs. wide hybrid
Medium hybrid vs. wide hybrid
N.S.
N.S.
N.S.
N.S.
N.S.
N.S.
differences in mating success probably reflect individual variation rather than real and consistent
interspecific differences in crossability.
F, Hybrid Viability.
8 . Viability ofhybrid Erythrina: proportions
TABLE
of seeds germinating and growing into healthy plants
(at 6 months).
Viability of the F, hybrids
was high and equal to or higher than normal viability of progeny within species. Overall, 1 4 3 (52%)
of the 2 7 3 F, hybrid seeds germinated; 1 2 0 of
these (44% of the total) survived as healthy F,
plants. In other words, most of the F, seeds either
germinated and grew vigorously or they did not
germinate a t all: the survival rate of those that
germinated was 84%'.
There were few instances of weakness in the
hybrids. Two individuals completely lacked chloroplasts and died soon after germination: a narrow
hybrid, Erythrinr~rnacro~rhyllax E. berteroana,
and a wide hybrid, E. crista-galli x E. speciosu.
I n both cases, though, siblings from the same cross
grew into healthy green plants. About 1 0 other
hybrid plants in different combinations were chlorotic, with yellow-green foliage, and died within 12 months. Several others a t first appeared chlorotic, but after several months they recovered the
normal green color and grew vigorously.
The comparative viablity of narrow, medium,
and wide hybrids is summarized in Table 8. Among
the F, hybrids viability (defined as successful ger-
Type of Hybrid Seed
Narrow F, hybrid
Narrow F, hybrid (all
within sect. Erythrina)
Medium F, hybrid
Wide F, hybrid
Total
Seeds
Sown
Live
Plants
Proportion
Viability
167
86
51%
66
62
44
339
7
16
18
127
11%
26%
41%
37%
Multiple comparison test for differences in
viability between types of hybrid
F,
F,
F,
F,
F,
F,
narrow vs. F1 narrow
narrow vs. F, medium
narrow vs. F , wide
narrow vs. F, medium
narrow vs. F, wide
medium vs. F , wide
All F, hybrids vs. F1 narrow hybrids
P < 0.01
P < 0.01
P
N.S.
N.S.
< 0.01
N.S.
P < 0.01
mination and survival of the plant for at least six
months) was highest for narrow hybrids, intermediate for wide hybrids, and lowest for medium hybrids. T h e difference between viability of narrow
and medium hybrids was statistically significant
( P < 0 . 0 1 ) , but between narrow and wide hybrids
it was not.
Included in Table 8 is the viability data for the
"narrow F?" hybrid seed produced in 1 9 8 4 from
the two-year-old narrow F,s within sect. Erythrirra.
Germination success of the F2s was very low, a n
unexpected and anomalous result; the difference
in viability between both the narrow F, hybrids
( 5 1 % ) and the wide F , s (41%') vs. the narrow F2s
(1 1%') was highly significant (both comparisons,
P < 0.01).
Viability data for the intraspecific and hybrid
progeny of maternal Erythrirza gz~atemalerrsisand
E. crista-galli, respectively, are presented in T a bles 9 and 1 0 . The intraspecific progeny of these
two species were grown for two purposes: to compare their viability with that of the hybrids from
the same female parents, and to carry out a study
of intraspecific variation of morphological traits
among siblings. The second goal was thwarted,
however, because of the poor germination of the
intraspecific seeds. Although the seed lots were not
large to begin with, viability of the seed derived
from selfings was particularly low: all seven selfed
Volume 75, Number 3
1988
Neil1
Ery thrina
TABLE9. Viability of'seed producedfrom Erythrina
guatemalensis as female parent (intraspecijc and hy-
TABLE10. Viability of' seed produced fiom Erythrina crista-galli as female parent (se@ and hybrids).
brids) . Proportion of seed germinating and growing
into healthy plants (at 6 months).
Proportion o f seeds germinating and grozctng into
healthy plants (at 6 months).
Paternity of Seed
Proportion
Seeds Live
ViaSown Plants bility
12
14
1
4
8%
29%
93
19
138
39
2
42%
11%
33%
Selfed
Intraspecific outcross
Narrow hybrid (within sect.
Erythrina)
Medium and wide hybrids
Total
46
Multiple comparison test for differences in
viability between seed of different paternity;
female parent = Erythrina gz~atemalensis
Self vs. outcross
Self vs. narrow hybrid
Self vs. medium & wide hybrid
Outcross vs. narrow hybrid
Outcross vs. medium & wide hybrid
Narrow hybrid vs. medium 92 wide hybrid
Self vs. all hybrids
Outcross
narrow hybrid vs. self
medium
wide hybrids
N.S.
P < 0.05
N.S.
N.S.
N.S.
P < 0.01
P < 0.01
Paternity of
Seed
Seeds
Sown
Live
Plants
Proportion Viability
Selfed
Narrow hybrid
Medium hybrid
Wide hybrid
7
4
33
24
70
0
1
7
10
18
0%
25%
21%
42%
26%
Total
Multiple comparison test for differences in
viability between seed of different paternity;
female parent = Erythrina crcsta-galli
Self vs. narrow hybrid
Self vs. medium hybrid
Self vs. wide hybrid
Narrow hybrid vs. medium hybrid
Narrow hybrid vs. wide hybrid
Medium hybrid vs. wide hybrid
Self vs. all hybrids
N.S.
P < 0.01
P < 0.01
N.S.
N.S.
N.S.
P < 0.01
wide hybrid combinations (1 1% of 6 4 attempts).
The number of individual F , hybrids for each comP < 0.01
+
bination ranges from one to nine.
Nineteen of the 2 2 narrow hybrid combinations
are between species in sect. Erythrina. There is
one narrow hybrid combination in each of sections
seeds of E. crista-gallc failed to germinate, as d ~ d
Cristae-galli, Chirocalyx, and Erythrnster. The
all but one of 1 2 selfed seeds of E. gr~atemalerrscs.
medium and wide hybrids include species combiThis could be a n expression of inbreeding depresnations in nine of the 2 6 sections of Erythrirrn and
sion in the self-progeny, but this possibility must
four of the five subgenera. I n seven of the hybrid
be corroborated with larger samples.
combinations, one parental species is native to the
Among the hybrids derived from Erythrirr n guaNew World and the other is native to the Old
temr~lerrsisand E. crista-gnlli females, the pattern
World.
of F, viability (Tables 9 , 10) and its relationship
I n summary, the viable F, hybrids obtained beto "taxonomic distance" between the parents was
tween the diploid species of Erythrirrn include repsimilar to the pattern of mating success discussed
resentative crosses that bridge the entire range of
earlier ( ~ a b l e s6, 7) for the same two species.
taxonomic diversity and geographic distribution of
Among the progeny of E. gz~atemalerrsis,viability
the genus. Interspecific crossability appears to be
of narrow hybrids was significantly higher than that
largely a function of individual variation in fecunof medium and wide hybrids. Among the progeny
dity of the female parent and only partially a funcof E. crista-galli, by contrast, there was no cortion of taxonomic/phylogenetic distance between
relation between F, hybrid viability and the degree
male and female parents. Given the results obtained
of relatedness of the parentals.
in these experiments, it may be expected that with
A complete listing of the F, hybrid plants prosufficient time, perseverance, and selection of comduced during 1 9 8 2 - 1 9 8 4 is contained in Tables
patible and fecund individual genotypes, any diploid
11-13. In all, there are 1 2 0 individuals in 3 3
Erythrina species could be crossed with any other
hybrid combinations (21% of the 1 5 5 attempted
to produce a viable F , hybrid.
combinations): 2 2 narrow hybrid combinations (34%
of 6 5 attempted combinations); four medium hy- Sexual :k'atr~ration nrrd Fertility of F, H y brid combinations ( 1 5 % of 2 7 attempts); and seven
brids. Some F, hybrids not only were very rapid
+
+
Annals of the
Missouri Botanical Garden
TABLE11. Artijcial Erythrina hybrids: rlarrolu (irltrasectior~a/J
Hybrid
E. crista-galli x E. jalcata
E. americana x E. berteroana
E. berteroana x E. gz~atemalensis
E. chiapasana
E. goldmanii
x
E. berterocrna
E. chanpnsarln
x
E. guatemalensis x E. berteroarln
E. guatemalensis
x
E. berteronrlcr
E. gz~aternalensis x E. berteronrla
E. gz~atemalensis x E chic~pc~sczr~c~
E. guatemalensis
x
E. chcnpnsar~cr
E. guatemalensis
x
E. folkersii
E. guatemalensis x E, mcrcrophylln
E. gz~atemalensis x E. mncrophylla
E. guatemalerlsis x E, mczcrophylln
E, gz~atemczler~sisx E. snl1:li'orn
E.
E.
E,
E.
gz~atemnlerlsis X E, tczjurnulcer~s~s
guatemcrlensis X E. tnjumulcerlsis
gz~atemczler~s~s
X E. tnjurnulcerlsis
herbncen x E. americc~r~c~
E herbacecr x E berterocrr~cz
E herbncea
x
E guntemnler~scs
E mncrophjlla x E c~rnerccc~r~c~
E rncrcrophylla
x
E berteronrln
E macrophylla
E mncrophyller
x
x
E berteroczrlcr
E folhrrsc~
E. mncrophylla x E. gunterncrlrr~s~s
E macrophylla
x
E. grrnternaler~s~s
Hybrid
Number
Live
Plants
Sect. Cristar-galli
2 x 3-1
1
Sect. Erythrirla
25 x 53-1
1
Hybrid
Accession
Numbers*
Parental Accession
Numbers*
PT 740283001 (F)
PT 750086001 (M)
WA 7 5 ~ 1 1 7 1(F)
WA 7 4 ~ 8 6 4(M)
WA 7 8 ~ 5 6 4(F)
WA 7 4 ~ 1 4 5 3(M)
PT 721005001 (F)
PT 700044002 (M)
)Veil1 561 7 (F)
!%ill 5497 (M)
PT 700018001 (F)
PT 700044001 (M)
PT 750419001 (F)
PT 73071 1001 (M)
PT 720999001 (F)
PT 700044001 (M)
PT 700018001 (F)
PT 721005001 (M)
PT 700018001 (F)
PT 730710001 (M)
PT 700018001 (F)
PT 700010001 (M)
PT 700018001 (F)
PT 750420002 (M)
PT 750420002 (M)
PT 750420002 (M)
PT 750419001 (F)
PT 721000002 (M)
WA 7 4 ~ 1 4 5 3(F)
WA 7 6 ~ 1 0 5 6(M)
WA 7 4 ~ 1 4 5 3(F)
WA 7 4 ~ 1 4 4 8(M)
WA 7 4 ~ 1 4 4 8(M)
WA 7 4 ~ 1 4 4 8(M)
WA 7 4 ~ 1 4 4 8(M)
WA 7 5 ~ 1 1 0 3(F)
WA 7 5 ~ 1 1 7 1(M)
WA 7 5 ~ 1 1 0 3(F)
WA 7 4 ~ 8 6 4(M)
WA 7 5 ~ 1 1 0 3(F)
WA 7 4 ~ 1 4 5 3(M)
WA 7 5 ~ 1 1 3 6(F)
WA 7 5 ~ 1 1 7 1(M)
PT 750420002 (F)
PT 700044001 (M)
PT 700044001 (M)
PT 750420002 (F)
PT 700010001 (M)
PT 750420002 (F)
PT 700018001 (M)
WA 7 5 ~ 1 1 3 6(F)
WA 7 4 ~ 1 4 5 3(M)
Volume 75, Number 3
1988
Neill
Erythrina
Hybrid
E. tajumulcensis
x
E. gz~atemalensis
Hybrid
Number
Live
Plants
40 X 43-1
4
Hybrid
Accession
Numbers*
Parental Accession
Numbers*
HO 82.761
PT 820599
WA 7 4 ~ 1 4 4 8(F)
WA 7 4 ~ 1 4 5 3(M)
HO 82.867
PT 770034001 (F)
PT 750281004 (M)
HO 82.768
PT 820550
HO 82.769
PT 820551
HO 82.770
WA
WA
WA
WA
WA
Sect. Chcrocalyx
E. abyssinica
x
E. latissimn
95 ~ 9 4 - 1
1
E. perrieri x E. variegntn
Sect. Erythraster
106 x 96-1
5
E. perrieri
x
E. variegata
106 x 96-2
3
E. perrieri
x
E. variegatn
1 0 6 x 96-3
1
* HO
Ho'omaluhia Botanic Garden; PT
female parent; M = male parent.
=
=
Pacific Tropical Botanical Garden; WA
in growth rates, but they also produced flowers at
a n exceptionally early age. Many of the narrow
hybrids between species in sect. Erythrinn grew
to be 4-m trees within two years after the seeds
were sown, and most flowered within that time.
Such sexual precocity is unknown in the parental
species. Seeds from intraspecific matings were sown
concurrently and in the same nursery with the
hybrids; none grew as rapidly or flowered as early
as most of the hybrids. None of the parental species,
in cultivation, have been known to produce flowers
in less than three years from seed.
An even more phenomenal case of precocious
flowering was the intersectional (medium) hybrid
Erythrinn crista-gnlli x E. fi~sca. Two sibling
individuals of this combination ( P T 8 4 0 2 3 1 0 0 1
and - 0 0 2 ) produced flowers when they were still
small plants in nursery pots less than five months
after the seeds were sown. The parentals are both
large- to medium-sized trees and are not known to
flower in the wild or in cultivation before at least
several years of growth.
Twenty-five of the two-year-old F, hybrid plants
flowered during February-March 1984. All were
narrow hybrids within sect. Erythrina, and represented nine hybrid combinations. These are listed
in Table 1 4 with the pollen fertility of each individual. Also included in Table 1 4 is the pollen
fertility of each of the parental individuals from
which these hybrids were derived.
With two exceptions, the pollen fertility of the
F , hybrids was above 95%. The pollen fertility of
the hybrids ( X = 97.6%) was slightly but significantly higher (P < 0 . 0 5 ) than the fertility of the
parentals
= 95.0%'). Eighteen of the 2 5 hybrid
(x
=
7 5 ~ 8 5 7(F)
7 4 ~ 8 9 2(M)
7 4 ~ 8 9 2(M)
7 4 ~ 8 9 2(M)
7 4 ~ 8 9 2(M)
Waimea Arboretum; F
=
individuals had pollen fertilities higher than either
of their parents. For this trait at least, the narrow
hybrids in sect. Erythrirra clearly exhibited interspecific heterosis.
Meiosis in pollen mother cells was examined in
several of the F, hybrids in sect. Erythrina. An
example is Erythrina g~~nternalensis
x E. macrophylln, HO 8 2 . 2 8 8 - A (Figs. 9, 10). Meiotic behavior in this hybrid can be compared with meiosis
in its male parent E. macrophylla, P T 7 5 0 4 2 0 0 0 2
(Figs. 5, 6).
As in the parental species, meiosis in the hybrids
was characterized by clumping of bivalents at late
diakinesis and metaphase I and by "sticky" chromatin bridges and late disjunction of some bivalents
at anaphase I. The normal meiotic process was,
however, not disrupted. Nondisjunction or unequal
assortment of chromosomes during- meiosis I was
not observed, and all cells examined at telophase
I or subsequent stages had the expected number
of 2 1 chromosomes. Meiotic behavior in the F,
hybrids within sect. Ervthrina, in short, was identical to the behavior described above for the parental species.
The only intersectional F, hybrid to flower by
November 1 9 8 4 was the five-month-old Erythrirr n
cristn-galli x E. fi~scn. Pollen fertility in this
hybrid ( P T 8 4 0 2 3 1 0 0 1 ) was 8 1 % . This was lower
than the pollen fertility of either parent ( E . crista,
E. j i ~ s c a ,W A 7 4 ~ 9 9 ,
galli, WA 7 4 ~ 8 4 0 96.1%';
96.3%') but probably not low enough to affect substantially fertility and mating success of the hybrid.
Only limited material was available for analysis of
meiosis in pollen mother cells of E. cristn-gnlli x
E. $LSC(L.In some cells, several quadrivalents ap-
Annals of the
Missouri Botanical Garden
TABLE12.
Artljclnl Erythrina hybrlds. mrdcum (brtwrrn sectcons, wcthcn sz~bgenercr)
Hybrid
Number
Hybrid'
E. crista-galli (2) x E. ji~sca(1)
Live
Plants
Subg. Dt~chassaingia
2~ 1-1
2
E. cristn-gnlli (2) x E. ji~sru(1)
2x1-2
5
Hybrid
Accession
Numbers
Parental
Accession
Numbers
HO 84.234
PT840232
HO 84.235
PT 840231
PT 740283001 (F)
PT740230005(M)
WA 74p840 (F)
WA 7 4 ~ 9 9(M)
WA 7 6 ~ 1 8 7(F)
WA 7 4 ~ 1 3 8 2(M)
PT 750280003 (F)
PT 730708001 (M)
PT 750280002 (F)
PT 730708001 (M)
PT 730708003 (F)
PT 750280003 (M)
PT 730742002 (F)
PT 750280003 (M)
Subg. Erythrincz
E. hrrbacecz (12) x E. hrrmrann (18)
22 x 73-1
2
E. lysistemon (17) x E. speccoscz (9)
72 x 16-1
1
HO 82.863
PT 820697
HO 84.238
E. lysistemon (17) x E. specioscz (9)
72 x 16-2
2
HO 84.243
E. speciosa (9) x E. lysisternorl (17)
16x72-1
1
HO 84.236
E. speciosa (9) x E. lysisternorl (17)
16x72-2
3
HO 84.237
I
Number in parentheses after each species denotes section (see Table 1).
peared to be formed at metaphase I (Fig. 11). In
other cells, meiosis was normal with 21 bivalents
at metaphase I. Without more thorough cytological
analyses, it is not possible to state whether or not
meiosis is significantly disrupted in this hybrid.
Nondisjunction and unequal segregation of some
chromosomes may contribute to the partial reduction in fertility of the pollen.
low, less than 374, and nine of the 1 2 F,s produced
no fruits from controlled selfing. In common with
the usual pattern of results in experimental pollinations of Erythrina parentals, much of the failure
in fruit maturation was due to postzygotic abortion
of young fruits, within one or two weeks after
fertilization. Most of the F,s did produce a few
fruits spontaneously, on open-pollinated inflorescences. Animal pollen vectors were not present in
the garden plots, and it is most likely that these
open-pollinated fruits were produced by autogamy.
&
Fecundity oj'F, Hybrids.
Fruit maturation from
the controlled self-pollinations of the two-year-old
F, hybrids in sect. Erythrina (Table 15) was very
TABLE13.
&
Artijcial Erythrina hybrids: wide (intrrsubgrnrric)
Hybrid'
Hybrid
Nurnber
Live
Plants
Hybrid
Accession
Numbers
2
PT 820422
Parental Accession
Numbers
-
E. caffra (17) x E. fusca (1)
71 x 1-1
E. crista-galli (2) x E. guatemalensis (12)
2 x 43-3
8
E. crista-galli (2) x E. speciosa (9)
2~16.1
1
HO 82.758
PT 820598
HO 82.860
E. crista-galli (2) x E. c;ariegata (26)
2 x 96-2
1
HO 82.495
E. guatemalensis (12) x E. abyssirzica (25)
4 3 x 95-1
1
HO 84.287
E. guatemalensis (12) x E. senegalensis (22)
4 3 x 79-2
1
HO 82.766
E. herbacea (12) x E. fusca (1)
22~1.1
4
HO 82.634
PT 820542
I
Number in parentheses after each species denotes section (see Table 1).
WA 7 4 ~ 1 4 5 6(F)
WA 7 4 ~ 9 9(hI)
WA 7 4 ~ 8 4 0(F)
WA 7 4 ~ 1 1 5 3(R.1)
WA 740283001 (F)
PT 730708001 (hI)
WA 7 4 ~ 8 4 0(F)
WA 7 6 ~ 9 9 6(R.1)
PT 700018001 (F)
PT 731006002 (hI)
WA 7 4 ~ 1 4 5 3(F)
WA 7 4 ~ 1 0 0(M)
WA 7 5 ~ 1 1 0 3(F)
WA 7 4 ~ 9 9(M)
Neill
Erythrina
Volume 75, Number 3
1988
TABLE14. Pollerz fertility of artijcinl F, hybrids
within sect. Erythrina and oftheir parents. At least 500
grains courzted for all samples.
TABLE14.
-
Continued.
-
11. Parentals
I. Hybrids
Hybrid Combination
E. nmericana x
E. berteroana
E. berteroarza x
E. guaternalensis
E. berteronna x
E. guatemalensis
E. chiapasnrza x
E. berteroann
E. chinpnsana x
E. berteroarza
E. chiapasana x
E. berteronna
E. chiapasnrza x
E. berteroann
E. guatemalensis X
E. berteroana
E. guaternnlerzsis X
E. berteronna
E. guatemnlensis X
E. chiapnsarza
E. guaternalensis X
E. chinpasana
E. gunternnlerzsis X
E. chiapasarza
E. guaternalensis X
E. chiapasnnn
E. guntemnlerzsis x
E. chiapasnna
E. guatemalerzsis x
E. folkersii
E. guaternalensis X
E. folkersii
E. guatemalensis x
E. macrophylla
E. guatemalensis x
E. rnncrophylla
E. guntemnlerzsis x
E. macrophylla
E. guatemalensis X
E. standleynrza
E. guatemalensis x
E. tajumulcensis
E. macrophylla x
E. berteroarza
E. rnncrophylla x
E. berteroarza
E. macrophylln x
E. guatemalensis
E. mncrophylla x
E. ,quatemalensis
Accession
Nurnber
Percent
Normal
Grains
Species
Percent
Normal
Grains
Accession
Nurnber
E. arnericana
E. berteroann
E. berteroann
E. berteroann
E. berteroann
E. berteronna
E. chiapasana
E. chinpnsana
E. folkersii
E. guatemalensis
E. guntemalensis
E. guatemalensis
E. rnacrophylla
E. macrophylln
E. starzdleyarza
E. tajumulcerzsis
*
Hybrids: mean pollen fertility = 97.6%
3.0%.
Parentals: rnean pollen fertility = 95.0% k 4.1%.
Differences in pollen fertility, hybrids vs. parentals: t =
2.28; DF = 39; P < 0.05.
Indicates hybrids with higher pollen fertility than either
parent.
"
Also in Table 1 5 a r e comparisons of fruit maturation in the selfed F , hybrids vs. their parents.
Several cornhinations of parental matings a r e included in the analyses. Fruit maturation was rnuch
lower in the selfed F,s (3%))than in the original
hybridizations which produced these F,s (22%) (P <
0.01).
The second pairwise comparison of fruit maturation in Tahle 1 5 , selfed F , hybrids vs. all of their
parental hyhrid comhinations (including reciprocals), rnay he more biologically meaningful than
the first cornparison for the following reason: the
fernale parents of the F , hybrids were very fecund,
with higher than average fruit maturation. T h e
inale parents (pollen donors) of the F,s generally
had lower fruit rnaturation when einployed as females in the hyhridization trials; rnany of the reciprocal crosses produced no hybrid fruit at all. If
it is assurned that fecundity (fruit maturation) is a
quantitatively heritable trait, then a n F , hyhrid
might he expected to be intermediate in fecundity
between its two parents, providing there is no reduction in fruit rnaturation in the hybrid caused
by incompatibilities between its constituent genornes. T h e proportion of fruit rnaturation expected
in the F,s, then, should approximate the proportion
in all the parental hybrid cornhinations, including
Annals of the
Missouri Botanical Garden
1 Oum
FIGURES
9-1 1. Meiosis i n Erythrina hybrids (pollen rr~othercells) .-9, 10. Late anaphase, E. guatemalensis x E. macrophylla, HO 82.288 (11 = 2 1 ) . Sticky
chromatin bridges and late disjunction of sorrLe bivalents (corr~parewith meiosis i n rnale parent E . macrophylla, Figs. 5, 6) .-11. Metaphase, E. crista-galli x
E. fusca, PT 8 4 0 2 3 1 0 0 1 (11 = 2 1 ) . At least two quadrivalents are visible.
the failed reciprocal hybridizations. By this measure, the second pairwise comparison in Table 15,
fruit maturation in the F, hybrids, is still much
lower than the 15% fruit maturation in the parental
generation; the difference is highly significant (P <
0.01).
There are several possible reasons for the reduced fruit maturation in the F, hybrids in sect.
Erythrina. The first is that the low fecundity is in
fact a consequence of hybridity caused by genic
incompatibility between the parental genomes. It
is evidently not, however, a matter of "hybrid
sterility" in the usual sense of the term, in which
the microgametophytes (pollen) and/or megagametophytes (embryo sacs) borne on the F, sporophyte are abortive and nonfunctional (Grant, 1953;
Stebbins, 1958). The pollen fertility of the F, hybrids, as discussed above, was exceptionally high;
the pistil and ovules also appeared to develop normally in the hybrids. Much of the failure of fruit
set in the selfed F,s was at the postzygotic stage
(abortion of young fruits). If the reduced fecundity
was truly a consequence of hybridity and intergenomic incompatibility, it is probably best considered as a case of "hybrid breakdown" (Grant,
1953; Stebbins, 1958) expressed as low viability
of the F, embryos.
There are other possible explanations for the
low fecundity of the selfed F,s that do not invoke
hybrid breakdown or other effects of hybridity. The
first is that it may be a consequence of self-mating,
the opposite effect from the apparent heterosis
evidenced by the exceptional vigor of the F, plants.
Fruit set in the selfed F,s was significantly lower
than the one in the selfed parentals, which in turn
were significantly lower in fruit set than the hybridizations. For both parental and F, selfings, the
high incidence of fruit abortion may be an expression of inbreeding depression, a result of the homozygous pairing of deleterious recessive alleles in the
genomes of the embryos. This possibility could be
tested by controlled cross-pollinations between F,s,
a step that was not taken initially because the goal
of the F, selfings was to produce F, plants with no
more than two constituent genomes.
Another possible reason for the low fecundity in
the F,s may simply be the juvenility of the F, plants
themselves. Although the F,s were very vigorous
and flowered precociously at two years of age, they
were not yet full-sized trees. At their size, they
might not be able to draw on sufficient resources
for the full fruit crop of a larger adult.
In short, the variables accounting for reduced
fecundity in the narrow F, hybrids still need to be
sorted out. This should be possible once the F, trees
attain their full adult size and several categories
of matings within and between individuals are carried out.
Viability of F, Hybrids. The viability of the F,
hybrid seed, obtained from selfed and open-pollinated F, plants in sect. Erytlzrina (Table 16), was
Volume 75, Number 3
1988
Neill
Erythrina
TABLE1 5. Fruit and seed maturntionjrorn controlled self:pollinations oj'narrou; F, I~ybridsin sect. Erythrina.
Hybrid Combination
Accessiori Number
E. berteroana x E. guatemalensis
E. guaternalensis x E. berteroana
E. guaterr~nlerlsis x E. chinpnsana
E. gunterr~alensis x E. chiapasana
E. guaternalensis x E. chiapasnrln
E. gunterr~alerlsis x E. chiapnsnna
E. guntemalensis x E. chiapasann
E. gunternalensis x E. rnarrophylln
E. guatemalensis x E. rnncrop/zylln
E. guntemalensis x E. tnjumulcensls
E. rnacrophylla x E. berteronrln
E. rnacroplzylla x E. guntemalensis
HO 82.674-A
PT820493002
HO 82.284-A
HO 82.284-B
HO 82.283-A
PT 820254002
PT820278002
HO 82.288-A
HO 82.285-8
H 0 820547001
HO 82.281-B
HO 82.763-A
Total selfed F, hybrids
Parental hybridizatioris
All parerital hybrid combinations (including reciprocals)
Selfed pareritals (sect. Erytlzrina)
Flowers
Pollinated
Mature Fruits
18
13
32
6
38
r,
10
27
16
20
23
15
225
51
171
144
Total
Number
of Seeds
0
0
7
0
0
3
0
0
0
1
0
0
0
0
3
0
0
2
0
0
0
1
0
0
6 (370)
21 (22%)
25 (15%)
12 (8%)
11
Multiple comparison test for differences in fruit maturation
Selfed F,s vs. parental hybridizations
Selfed F,s vs. all parental hybrid combinations
Selfed F,s vs. selfed parentals
Parental hybridizations us. selfed parentals
All parental hybrid combinatioris vs. selfed parentals
significantly lower than viability of the F , hybrids.
This was show11 in Table 8, where the F,s were
compared with all the narrow F, hybrids; the difference was highly sigrlificarlt ( P < 0.01). In Table
1 6 the viability of the F, seed is compared specifically with that of their OTVII parents, i.e., with the
F , seed lots producirlg the parents of the F,s. The
viability of the F2s (13%) was sigrlificarltly lower
(P < 0.01) than that of their F, parent generation
(6170 viability).
I n Table 1 6 the viability of the F, seed and of
their F , parents is also compared with seed from
irltraspecific matings in Erythrirra gz~atenlnlen,si,s
(including seed from selfirlgs and intraspecific outc r u s e s , the only intraspecific viability data arailable for sect. E r y t h r i n a ) . The F, hybrid seed was
significantly higher in viability than the intraspecific seed ( P < 0.05). The viability of the intraspecific seed (19%)was somewhat higher than that
of the F, seed, but the difference \\;as nonsignificant.
In summary, the viability of F , hybrid seed was
significantly higher than F, seed derived from selfed
F , matirlgs and higher than seed derived from intraspecific matings. If the very high F, viability is
truly an expression of interspecific heterosis, this
hybrid advantage is not retained in the F, gener-
P
P
P
P
< 0.01
< 0.01
< 0.05
< 0.01
N.S.
ation, when the F,s are derived from selfed F,
hybrids.
It is possible to interpret the reduction in F L
viability with respect to F , viability as "hybrid
breakdown." However, with the data presently
available, the reduced viability of the F,s derived
from selfed F,s could also be interpreted as a n
expression of irlbreedirlg depression. It could also
be interpreted simply as a n absence of the heterotic
advantage possessed by the F,s, since the viability
of the F,s was not significantly lower than that of
the irltraspecific progeny. These three alternatives
carlrlot be differentiated with the presently available
information. Additional progeny trials of F , , F,,
and intraspecific seed lots are needed to test the
possibility that hybrid breakdowrl may be expressed
in the F2 generation of E r y t h r i n a hybrids.
In any case, the lowered average viability of the
Fls was a function only of poor germination of the
seed. The seeds that did germinate produced healthy
plants with normal growth and vigor at six morlths
of age. There were no indications of chlorosis or
other debilities in the F2 plants.
S t u d i e s o f P r e v i o r ~ s l yS y n t h e s i z e d H y b r i d s . Nine
artificially produced E r y t h r i n a hybrids, all between
Annals of the
Missouri Botanical Garden
TABLE16.
Viability of F, hybrids withcn sect. Erythrina
F, Hybrid Comhinatiori
Accessiori
Number
Female Parent
(F, Hybrid)
Paternity
E. guatemalensis x E. bcrteroana
E. guaterr~alensis x E. chiapasar~a
Open
Self
E.
E.
E.
E
E.
E.
Self
Open
Self
Open
Operi
Operi
Open
guaterr~alerlsis x E. rnacrophylla
guatemalensis x E. taiumulcerlsis
hrrbacra x E guatemalcnsls
macrophylla x E berteroana
ma~rophylla x E. bcrtrroana
rnacrophylla x E. guatemalenscs
Total F, hybrids in sect. Erythrina
F, hybrid parents of F,s
Intraspecific progeny, Erythrina guatemalerlsis
Seeds
Sown
Live F,
Plants
56
28
26
7(13%)
17(61%)
5 (19%)
hlultiple comparisori test for differences in viability of seed
F, hybrids vs. their F, hybrid parents
F, hybrids vs. intraspecific E. guaternalensis
F, hybrids vs. intraspecific E. guaternalensis
species o f different sections, have been reported
prior to this study (Table 17). Krukoff & Barneby
( 1 9 7 4 )described most o f these; in the same paper
they described some putative natural hybrids between sympatric Mesoamerica11and African species.
The parentage o f only two o f the artificial hybrids
is known for certain; both o f these F,s are "wide"
intersubgeneric hybrids and are reported to be fertile.
The oldest and best-known Erytlzrinn hybrid is
E. x bidwillii Lindley, synthesized from E. herbacea (sect. Erythrina) 9 and E. crista-galli (sect.
Cristae-galli)8 in Australia in the 1840s and since
spread around the world as a cultivar by propagation o f cuttings. Krukoff & Barneby ( 1 9 7 4 ) reported E. x bidwillii to produce viable seed and
also that "no Merldeliarl segregation o f phenetic
characters is observed in the F, or subsequent
generations." They further claimed that this hybrid
had naturalized in Fiji and was therefore a stabilized
"neospecies."
I examined E. x bidwillii in cultivation at Foster
Garden, Honolulu (FG 64.2035). Meiosis in pollen
mother cells was normal with 2 1 bivalents at metaphase I . Pollen fertility was 6376, comparable to
Graham & Tomb's ( 1 9 7 4 ) report o f 7 6 % normal
pollen for this hybrid. I attempted to produce an
F, generation by controlled self-pollination o f 6 0
flowers over a period o f several weeks. Young fruits
P < 0.01
N.S.
P < 0.01
were obtained but they invariably aborted before
two weeks o f development. I have not seen mature
sporltaneously produced fruits on any cultivated
plants or herbarium specimerls o f E. x bitlu>illii,
so the reports o f its viable seed productiorl are
questionable.
I made limited attempts ( 1 2 trial pollinations)
to backcross E. x bidwillii to one o f its parents,
E. crista-gnlli. The pollinations all failed, but given
the reasonably high pollen fertility o f E. x bidwillii,
it is likely that with perseverance some backcross
progeny could be obtained.
The other previously reported hybrid o f known
parentage is Erythrina x resuparcellii Srivastava
not
(a rlomerr r ~ u d r ~ m
, validly published), a hybrid
between the perennial herb E. resupinata (sect.
Suberosae) 9 and E. z'ariegata (sect. Erythraster)
8 (Jalil et al., 1982). The F , is a branched shrub,
and in other morphological traits is also intermediate between the two parents. The flowers, however, resemble those o f the female parent much
more closely than those o f the male. This hybrid
was not available to me, but Jalil et al. ( 1 9 8 2 )
reported that it had normal meiosis in pollen mother
cells with 21,, at metaphase I , pollerl fertility o f
6276, and viable seed.
Erythrina x sykesii Barrleby & Krukoff was
the only other hybrid among those listed in Table
17 available to me for experimental studies. This
Volume 75, Number 3
1988
TABLE17.
Neill
Ery thrina
Previous reports ofartijcial Erythrina hybrids.'
1. Erythrina x bidwillii Lindley, Bot. Reg. 33: pl. 9. 1849.
E. herhacea a ( 1 2 ) x E. crista-galli 6 ( 2 )
2. Erythrina x bellangeri Focke, Die Pflanzen-mischlinge. 110. 1881.
? E. crista-galli 9 ( 2 ) x E. herbacea 6 ( 1 2 )
3. Erythrina x c r a s s ~ o l i aKoorders ex Backer, Schoolflora voor Java 1: 360. 1911.
? E. suburnbrarls ( 6 ) x E. variegata ( 2 6 )
? E. fusca ( 1 ) x E. variegata ( 2 6 )
4. Erythrirla xpurninensis Barneby & Krukoff, Lloydia 37: 446. 1974.
? E. speciosa ( 9 ) x E. sp. (subg. JVIicropteryx)
5 . Erythrina x herlrlesyae Barneby & Krukoff, Lloydia 37: 448. 1974.
? E. humeana ( 1 8 ) x E. lysistemon ( 1 7 )
6 . Erythrina xorba Barneby &. Krukoff, Lloydia 37: 449. 1974.
E. lysistemon ( 1 7 ) x E. speciosa ( 9 )
7 . Erythrina xsykesii Barneby & Krukoff, Lloydia 37: 447. 1974.
? E. americana ( 1 2 ) X E. lysisterr~on( 1 7 )
? E. speciosa ( 9 ) x E. lysisternon ( 1 7 )
8. Erythrina xvlissirzgensis Waby ex Barneby & Krukoff, Lloydia 37: 446. 1974.
? E. jusca ( 1 ) x E. varirgata ( 2 6 )
? E. fusca ( 1 ) x E. suberosa ( 4 )
9 . Erythrirla x resuparcellii Srivastava, Allertonia 3: 19. 1982. norrlen nudum.
E. resupinata 9 ( 4 ) x E. variegata 6 ( 2 6 )
I Known or presumed parental species combinations are listed below each hybrid binomial; question mark preceding
hybrid conlbination indicates uncertain parentage. Numbers in parentheses following species refer to sections to which
species belong (Table 1).
hybrid was reputedly produced under cultivation
in Australia in the 19th century, but its parentage
is unknown. Krukoff Bi Barneby ( 1 9 7 4 ) believed
the parents to be E. Lysistemon (sect. Caffrae) and
E. nmericnnn (syn. E. coralloitles) (sect. Erythrinn). Based on study of floral and leaf morphology, I believe instead that the parents are E.
lysistemon and R. speciosa (sect. Sterrotropis).
Since I have obtained both reciprocal hybrids of
E. speciosa x E. lysisternon (Table 12), these F,s
can be compared with E. x sykesii when they come
into flower.
I examined cytologically several individual ramets of E. x sykesii (WA 7 6 ~ 8 6 4 WA
,
75~1706,
Foster Garden FL.669) and attempted to produce
F, plants by controlled self-pollination. Meiosis in
PMCs was apparently normal, with 21 bivalents
a t metaphase I. Pollen fertility was 81-84%, which
agreed closely with results reported earlier for the
same taxon by Graham & Tomb (1974). However,
no mature fruits were obtained from 6 5 attempts
at selfing. Young fruits with partially developed
seeds were produced in abundance as with E. x bidwillii, but these always aborted within two to three
weeks following pollination.
E r y t l ~ r i n ax bidzuillii, E. x resuparcellii, and
E. x sykesii are the only intersectional or intersubgeneric hybrids that have been tested for fecundity at this time. An F, generation was reportedly obtained from E. x resz~parcellii(Jalil et al.,
1982), but the other two may be incapable of
producing F, progeny, at least from selfing of the
F,. Zygotes, embryos, and young fruits are formed,
but the fruits abort before maturity. These "wide"
hybrids, then, may be subject to hybrid breakdown
expressed as inviability of the F, hybrid embryos
borne on the F, hybrid sporophyte. As discussed
in the previous section, the cause of mating failure
in the F, hybrids is subject to different interpretations. Whether or not F, hybrid breakdown is a
general phenomenon in Er3,thrina remains to be
investigated.
CONCLlI5IONS: EXPERIhIENThL WYDRIDIZ.4'rION5
AND SELF-COSIPATIRILITY TRIAL5
From the information presented in this section,
a series of generalizations regarding breeding systems and species relationships in Brythrina can be
outlined:
1. Even under the most carefully controlled
conditions, mating success (proportion of pollinated
flowers producing mature fruits) is low in all Br'r-ythrina species. This is true even when the effects
of "resource competition" are eliminated by removing most flowers from a n inflorescence as well
as all of the spontaneously produced fruits on the
tree, and pollinating only a few selected flowers
per inflorescence. Mating failure results partly from
prefertilization abortion of pollinated flowers, but
also to a large extent from postfertilization abortion
of young fruits.
Annals of the
Missouri Botanical Garden
Volume 75, Number 3
1988
Neill
Erythrina
2. Gametophytic or sporophytic self-incompatibility systems of the "classical" model, mediated
by inhibition of pollen tubes in the style or stigma
and governed by a single-locus, multiallelic S-gene,
are evidently not present in any of the E r y t h r i n a
species examined. Self-incompatibility in this strict
sense is probably absent from the entire genus.
There is considerable individual variation in fecundity, and some individuals may be cryptically female-sterile, but if an individual produces seed from
outcrossing it will also produce some seed from
self-mating. For some individuals and some populations, mating success is lower in selfing than in
outcrosses, but much of the mating failure is expressed postzygotically by abortion of young fruits.
This is probably a n expression of inbreeding depression, the consequence of increased homozygosity
for any number of deleterious recessive alleles,
rather than the action of a specific S-allele. Inbreeding depression may also be expressed in the
progamic stage as inhibition of pollen tubes.
3. Spatial separation of anthers and stigma in
some Erytfzrina species, and protandry in other
species, limits autogamous pollinations. For the protandrous species at least, this mechanism is not
absolutely effective; autogamous fruits are occasionally produced. Autogamy occurs only with the
ultimate flowers on a n inflorescence and may be
a n adaptive feature of the breeding system to produce some seed as a "last resort" if no "highquality" (i.e., outbred) seed was produced on earlier
flowers of the inflorescence.
4. The hybridization trials indicate that matings
between closely related species (within sections) are
just as likely to produce viable progeny as are
matings within species. Mating success is usually
higher, in fact, in interspecific hybridizations within
sections than in self-mating. At increasing taxonomic distance between parental species, there is
a general trend to lower mating success in the
hybridization trials. This trend is not universally
applicable, however. Viable F , hybrids have been
produced between the most distantly related groups
of species in the genus-between
species of different subgenera indigenous to different continents.
It is probable that viable F , hybrids can be obtained
between any two diploid species in the genus Erythrina.
5. F , hybrids between the closely related species
in sect. Erythrina exhibit interspecific heterosis by
several measures: viability of the F, seed is higher,
and the F , plants are more vigorous, sexually precocious, and have higher pollen fertility than the
parental species. Pollen fertility is somewhat lower
in hybrids between more distantly related species,
but these hybrids are generally comparable in viability and vigor with the parental species.
6. A reduction in fecundity is exhibited by the
F , hybrids, a t least when the F,s are selfed. This
lowered mating success is not due to "hybrid sterility" per se, since the gametes produced by the
F, hybrid function normally. Mating failure is expressed postzygotically by abortion of young fruits
and evidently is a consequence of inviability of the
F2 hybrid embryo. An alternative explanation may
be that mating failure in the selfed F,s is a consequence of inbreeding depression.
These experiments support the first two hypotheses presented in the introduction: 1)the species
in sect. E r y t h r i n a can hybridize freely with each
other, and there are no internal qualitative or quanitative postmating isolating barriers between the
species; and 2) hybridization between more widely
divergent species is also possible; there is generally
a quantitative reduction in mating success in the
wider hybridizations, but this probably does not
constitute a n absolute barrier to the formation of
F , hybrids. The only major unanswered question
regarding interspecific compatibility among diploid
Erythrina is the possibility of hybrid breakdown
in the F, and subsequent generations. F2 breakdown, if it exists, does not form a n absolute isolating
barrier within sect. E r y t h r i n a , but it may form a n
absolute barrier in hybridizations between more
widely divergent species.
The fact that plant species with large morphological discontinuities can be hybridized, and that
large hybrid progenies can be grown together in a
common garden, has allowed for analyses of the
genetic basis of these phenetic differences (review
in Gottlieb, 1984). A thorough genetic analysis, of
course, requires at least the study of segregating
FIGURES
12-17. SEMimages oj'abaxial leafsurface ofErythrina.-12. Epicuticular waxplatelets, E. suberosa,
W.4 45~960.-13. Epicuticular wax platelets, E. berteroana, PT 700044001.-14. Two-armed hairs, E . chiapasana, WA 74~876.-15. Dendritic hairs, E. perrieri, W.4 74~876.-16. Balloonlike hairs, E. arborescens, W.4
78~225.-17. Ribbonlike hairs, E . leptorhiza, Neill 5646.
Annals of the
Missouri Botanical Garden
Volume 75, Number 3
1988
Neil1
Erythrina
FIGURES24-27. SEiW images, abaxial leafsurfaces ofErythrina. All images at 60" tilt.-24,
E. stricta, W A 7 4 ~ 8 9 7 . ~ 227.
6 , Lamellae arzd two-armed hairs, E . suberosa, W A 7 5 ~ 9 6 0 .
e
FIGURES
18-23.
25. Lamellae,
Epidermal features ofabaxial leafsurfaces of Erythrina. 18, 2 0 , 22.-SEibl images. 19,
21, 23.-Anatomical sectiorzs.-18. Papillae, E. guatemalensis, PT 750419001.-19. Papillae, E . folkersii, PT
700010001.-20. Lamellae, E . salviiflora, PT 721346001.-21. Lamellae, E. suberosa, WA 75~960.-22.
Glandular hair, E. salviiflora, PT 721346001.-23. "Glandular" hair, papillae, and lamellae, E. berteroana,
E4 7 4 ~ 8 6 4 .
Annals of the
Missouri Botanical Garden
TABLE18.
Comparison of leaf epidermal characters i n Erythrina hybrids and parents.
Female Parent
F, Hybrid
Male Parent
E. berteroana
PT 700044002
E. chiapasana
P T 721005001
Figures
Hairs
28-30
dense covering of two-armed
hairs
31-33
sparse two-armed hairs
34-36
hairs absent
Epidermal
sculpturing
low ridges around stomata;
no lamellae
dense network of discontinuous lamellae, up to 40 gm
tall
Epicuticular
wax
wax absent
open, irregular network of
discontinuous lamellae; un.
even in height, less than
15 pm tall
wax present
wax present
E. berteroarza
PT 700044001
Figures
Hairs
Epidermal
sculpturing
37-39
hairs absent
dense papillae, to 40 prn tall
Epicuticular
wax
wax present
40-42
hairs absent
sparse covering of papillae
and 2-6-celled lamellae;
short, less than 25 pm tall
wax present
E. chiapasarza
PT 721005001
E. guatemalensis
PT 700018001
Figures
Hairs
46, 47
hairs absent
Epidermal
sculpturing
dense papillae to 40 pnl tall
Epicuticular
wax
wax present
48, 4 9
sparse scattering of twoarmed hairs
incipient papillae: low crescent-shaped ridges outlining anticlinal walls of epi.
dermal cells
wax present
PT 700018001
52-54
hairs absent
Epidermal
sculpturing
Epicuticular
wax
dense papillae to 4 0 prn tall
wax present
dense covering of two-armed
hairs
low ridges around stomata: no
papillae
wax absent
E. abyssirzica
PT 731006002
E. guatemalerzsis
Figures
Hairs
43-45
hairs absent
dense network of discontinuous lamellae, to 4 0 pm
tall
wax present
55-57
sparse covering of two-armed
hairs
low epidermal ridges, less
than 1 0 pm tall
wax present
58-60
dense covering of two-armed
hairs
low epidermal ridges, less than
10 pm tall
wax present
E. senegalensis
E. guatemalerzsis
WA 745100
WA 7 4 ~ 1 4 5 3
Figures
Hairs
61-63
hairs absent
64-66
sparse scattering of balloonlike hairs, up to 5 0 pm x
100 pm in size
67-69
sparse scattering of balloonlike
hairs, up to 50 pm x 100
pm in size
Epidermal
sculpturing
dense papillae, to 4 0 prn tall
low, crescent-shaped epiderma1 ridges, less than 5 prn
tall
low stellate papillae, less than
1 0 pm tall
Volume 75, Number 3
1988
Neill
Erythrina
TABLE18. Contirzued.
F, Hybrid
Female Parent
Epicuticular
wax
wax present
wax present
Male Parent
wax absent
E. speciosa
PT 730708001
Figures
Hairs
70, 71
hairs absent
Epidermal
sculpturing
Epicuticular
wax
low papillae, less than 10 fim
tall
wax present
72, 73
sparse scattering of twoarmed hairs
low epidermal ridges, less
than 5 fim tall
wax present
E. crista-galli
(4 individuals)
WA 74~840
Figures
Hairs
Epidermal
sculpturing
Epicuticular
.,vax
76,82
hairs absent
low, discontinuous lamellae,
less than 10 wm tall,
forming reticulate pattern
wax absent
78-81, 84-87
hairs absent
variable: papillae or discontinuous lamellae, to 15
fim tall
wax present in all
74, 75
sparse scattering of two-armed
hairs
low epidermal ridges, less than
5 fim tall
wax present
E. guatemalerzsis
WA 7 4 ~ 1 4 5 3
77, 83
hairs absent
dense papillae, to 40 fim tall
wax present
PT 840231
E. crista-galli
HO 84.235
(3 individuals)
WA 74~840
Figures
Hairs
Epidermal
sculpturing
Epicuticular
wax
88,94
hairs absent
low discontinuous lamellae,
less than 10 fim tall,
forming reticulate pattern
wax absent
90-92, 96-98
hairs absent
variable: scattered low papillae, 3-4-celled lamellae
or nearly flat
wax present in all
progeny in the F2 generation. In the absence of
large F, families, however, prelirninary characterization of the inheritance of morphological characters c a n be obtained from F, hybrids.
A study of the inheritance of phenetic traits in
artificially produced hybrids serves several purposes beyond that of genetic analysis. Firstly, it
allows for confirmation of hybridity in the hybrid
progeny. I n any experimental hybridization, there
exists the possibility that the cross may be spurious;
the progeny could result from contamination of
self-pollen on the stigma, or from agarnospermy.
However, if the progeny possess a character present in the male parent but absent in the female,
their hybrid nature is reasonably confirmed.
A second purpose for studying the inheritance
of morphological traits in artificial hybrids is to
generate information on the patterns of variation
89, 95
hairs absent
irregular convoluted surface
with deep cavities, knobs,
and protrusions
wax absent
to be expected when hybridization occurs in nature.
If, as Raven ( 1 9 8 0 ) and Grant ( 1 9 8 1 ) have suggested, there is a great deal of hybridization in
flowering plants that passes undetected as such,
then study of the products of artificial hybridization
may help in the discovery and confirmation of
hybrids in natural populations. This method was
used effectively, for example, by Nobs ( 1 9 6 3 ) in
his biosystematic study of Ceunothus. Some of the
us
artificial F, hybrid segregates of C ~ a r ~ o t l z closely
resembled stabilized populations with restricted
ranges recognized as species. Nobs used this evidence to support his hypothesis that these species
were of hybrid origin and were derived from pairs
of extant, more wide-ranging species.
A similar study, combining artificial hybridization and analysis of natural hybridization, conducted by Gillett & 1,irn ( 1 9 7 0 ) o n Bidetzs in
Annals of the
Missouri Botanical Garden
FICI.RES28-36. SEM images, abaxial leafsurjaces of Erythrina chiapasana x E. berteroana and parents.
Each horizontal row at equal magn$cation. 29, 32, 35 at 60' tilt.-28-30.
E. chiapasana, PT 721005001,
female parent.-31-33.
E. chiapasana x E. berteroana, HO 82.278.-34-36. E. berteroana, PT 700044002,
male parent.
Volume 75, Number 3
1988
Neill
Erythrina
FIGI~RES
37-45. SEA4 images, aE axial leaj'surjkces of Erythrina guatenlalensis x E. berteroana and parents.
Each horizontal row at equal magnsl{cation. 38, 4 1 , 4 5 at 60' tilt.-37-3 i9. E . guatemalensis, PT 700018001,
fernale parent.-40-42.
E. guatemaltsnsis x E. berteroana, HO 82.289.-4. 3-45. E. berteroana, PT 700044001,
male parent.
Annals of the
Missouri Botanical Garden
FIC~JRES
46-5 1. SEM i m q e s , abazial lrafsurfaces oj'Erythrina guatemalensis x E. chiapasana artd parents.
Each horizontal row at equal rnagrtLjcatiort.-46, 4 7 . E. guatemalensis, P?' 700018001, fernale parer~t.-48,
49. E. guatemalensis x E . chiapasana, HO 82.283.-50, 5 1 . E . chiapasana, PT 721005001, mmle parent.
Hawaii, has been questioned by Ganders & Nagata
(1984). They showed that some of the putative
natural hybrids were merely intraspecific variants,
and Ganders & Nagata concluded that adaptive
divergence was more important than hybridization
in the evolution of R i d e n s in the Hawaiian Islands.
Although all of the 1 7 Hawaiian R i d e n s species
are in fact interfertile, natural hybridization is rare
because they are mostly allopatric. Evidently Gillett
& Lim used too few characters and ignored intra~ o ~ u l a t i ovariation.
n
An important caveat is to
avoid a too facile interpretation of hybridization
results when applying them to the study of processes in nature.
In this research, two sets of phenetic traits were
examined in the Erytlzrinrx hybrids and their parent
species: 1) features of the epidermis of abaxial leaf
surfaces, and 2) morphology and color of the flowers.
Characters of the leaf are second only to those
of flowers in their use and value in taxonomic
studies (Stace, 1 9 8 4 ) . Studies of the inheritance
of leaf surface characters in intrrspecific hybrids
have recently been carried out in A l o e and G u s terirx (Liliaceae) (Cutler, 1972), in l l e r (Baas,
1978), and in Qrrercns (Cottam et al., 1982).
Brytlzrinrx species possess a wide variety of leaf
surface characters. These have figured prominently in the taxonomic delimitation of the species
(Krukoff, 1939a, b; Krukoff &- Rarneby, 1974).
In these p r e ~ i o u sworks the surface characters
Volume 75, Number 3
1988
Neill
Erythrina
S E V rmages, abaxlul leaf surfaces ofErythrlna guatemalens~s x E abysslnlca arzd parents
F~c,rRE\ 52-60
Each horrzontal rol* at equal magnrjicatlon -52-54
E guatemalens~s,PT 700018001, female parent -55-E abysslnlca, PT 731006002, male parent
J , E guatemalens~s x E abysslnlca, HO 8 2 6 4 7 -58-60
Annals of the
Missouri Botanical Garden
of Erythrina guatemalensis X E.senegalensis and parents.
FI(.TIRE'
61-69, SE.\.I images, aba.x~allec~f'surfi~ces
Each horizontc~lrow at equal n~agni'cation.-61-63. E. guaternalensis, 1K-1 7 4 ~ 1 4 . 5 3ferraale
,
parent.-64-66
E . guatenlalensis x E. senegalensis, H O 82.766 -67-69. E . senegalensis, K X 7 4 ~ 1 0 0 mc~le
,
parent.
Volume 75, Number 3
1988
Neill
Erythrina
FIGURES
70-75. SEM images, abaxial leaf surfaces of Erythrina lysistemon x E. speciosa and parents. Each
horizontal row at equal magni$cation.-70,
71. E. lysistemon, PT 750280003, female parent.-72,
73. E.
lysistemon x E. speciosa, HO 84.283.-74, 75. E . speciosa, PT 730708001, male parent.
were not studied with high magnification or anatomical sectioning, however, and the structure of
some of the surface characters was misinterpreted.
This is discussed below in the description of epidermal characters. Leaf epidermal features of a
few species of E r y t h r i r ~ ahave also been surveyed
using scanning electron microscopy (Ayensu, 1977).
Materials and Methods
In this study, only the abaxial surfaces of leaves
were examined. All samples were obtained from
mature, fully expanded leaves. which were pressed
and dried as in preparation of herbarium specimens. The specimens were gold-coated with a Polaron E 5 0 0 0 sputter-coater and observed with a n
Hitachi 4 5 0 - S scanning electron microscope. For
a few selected species, anatomical sections of par-
affin-embedded leaves were prepared by Dr. Hiroshi Tobe of Chiba University, Japan.
Results
S u r v e y of L e a f E p i d e r m a l Features i n Erythrina
Epicuticular Wan-. Platelets of epicuticular wax
cover the abaxial leaf surfaces of many E r y t h r i n a
species. The wax gives a whitish, glaucous appearance to the leaf observed without magnification. The platelets are 1-3 p m in size, are oriented
randomly on the leaf surface, and vary in density
(Figs. 12, 13). Epicuticular wax is consistently
present in some species: but in others its presence
or absence is variable even among individuals of a
single population.
Multicellular Branched Hairs.
Several types of
Annals of the
Missouri Botanical Garden
Volume 75, Number 3
1988
Neill
Erythrina
hair occur on the abaxial leaf surfaces of E r y t h r i n c ~
species. The most common, and the only type
found in sect. E r y t h r i n a , is a multicellular, twoarmed hair (Fig. 14). This consists of several short
basal cells, one or two longer cells forming the
stalk, and one cell forming each of the arms, which
may be 1 , 0 0 0 pm or more in length. Two-armed
hairs are present on the young leaves of most
species in sect. Erythrina and in species of other
sections as well, but in many species they are
deciduous and are absent from fully expanded
leaves. In the other species, the hairs a r e retained
on mature leaves and form a dense tomentum.
Multiple-branched, dendritic hairs (Fig. 1 5 ) are
restricted to sect. Erythraster. They occur in all
species of that section and a r e found on calyces,
inflorescence branches, and on leaves. Each branch
of the dendritic hair is 50-100 ym long and is
formed by a single cell. The dendritic hair has
about 8-12 branches and extends up to 3 0 0 Fm
above the surface of the epidermis.
Papillae are single-celled, fingerlike trichomes.
Each papilla is formed by the protrusion of a n
epidermal cell above the leaf surface. Papillae are
up to 4 0 hrn tall and 1 5 hrn in diameter (Figs. 18,
19). Papillae and the epidermal surface between
them are usually covered with epicuticular wax
platelets. Lnder low magnification, a leaf surface
with papillae appears covered with whitish granules. Krukoff (Krukoff & Barneby, 1 9 7 4 ) termed
such leaf surfaces "farinose-ceriferous" or "granular-ceriferous," but the "granules" he described
are wax-covered papillate cells, not individual particles of wax.
Papillae similar to the ones found in Erythrina
occur on leaf surfaces in many groups of plants,
but the structures I have termed "lamellae" have
not to my knowledge been reported from leaf epidermis of any angiosperm besides E r y t h r i n a . Lamellae, like papillae, are formed by protrusions of
epidermal cells, but in lamellae the cells are joined
edge-to-edge to form continuous "walls" one cell
thick that stand above the surface of the leaf (Figs.
2 0 , 2 1). Leaf surfaces with lamellae are also usually
covered with epicuticular wax.
Lamellae occur in several species of sect. Ery t h r i n a . I n these species the lamellae a r e discontinuous; each lamella is composed of several to
twenty cells standing edge-to-edge. The lamellae
form a dense, discontinuous network with a characteristic pattern when observed a t low magnification (Fig. 34). Krukoff (Krukoff & Barneby,
1 9 7 4 ) referred to leaves with wax-covered lamellae
as "reticulately ceriferous."
A unique pattern of lamellae occurs only in the
Asian sect. Suberosae (Figs. 24-27). The lamellae
are tall ( 5 0 pm) and continuous. Parallel rows of
several lamellae, each leaning a t a different angle
with respect to the leaf surface (Fig. 25), are joined
to form a n open network of interconnected polygons (Fig. 24). Shorter lamellae extend into the
center of the polygons. The distribution of the
polygons is associated with the vascular tissue of
the leaf. E r y t h r i n a suberosa (Figs. 2 6 , 2 7 ) has
both polygon-forming lamellae and two-branched
hairs.
Trichome characters are generally quite constant within a species and a r e useful taxonomic
markers, often allowing species identification from
"Glandular" H a i r s . Multicellular, uniseriate
hairs occur sporadically on leaf surfaces of many
E r y t h r i n a species (Figs. 22, 23). They appear to
be glandular, but what substance these hairs secrete, if any, is not known. They a r e squat, rounded
hairs about 5 0 pm long and comprised of five or
six cells. Observed with a microscope, they glisten
with a translucent amber color.
Vnicellular Hairs. The most common type of
multicellular hair in E r y t h r i n a is formed by a
rounded or elliptic, thin-walled cell which loses its
cytoplasm and collapses a t leaf maturity or upon
drying (Fig. 16). These I refer to as "balloonlike"
hairs. They occur in several sections of E r y t h r i n a .
Long, flat, ribbonlike hairs 5 0 0 pm or more in
length a r e found particularly along the principal
veins of the leaf in some species (Fig. 17). This
type of hair is predominant in the Mexican sects.
BrevilfEorae and Leptorhizae.
E p i d e r m a l S c u l p t z ~ r i n g : Papillae a n d Larnellae. T h e remaining types of Erythrina trichomes
a r e considered separately from hairs because they
a r e more integrally a part of the foliar epidermis.
These are papillae and lamellae, which I refer to
collectively as "epidermal sculpturing."
FIGURES
76-81. SEM Images, abaxial leaf surfaces of E . crista-galli x E. guatemalensis and parents. All
photos at equal magn$cation.-76.
E. crista-galli, WA 7 4 ~ 8 4 0fernale
,
parent.-77. E. guatemalensis, WA
7 4 ~ 1 4 5 3rnale
,
parent.-78-81.
E. crista-galli x E. guatemalensis, four F, siblings, PT 820548 and WA 82.278.
Annals of the
Missouri Botanical Garden
sterile material. In contrast, presepce or absence
of epicuticular wax is variable within populations
and is not a useful marker. As will be seen in the
discussion of the hybrids below, epicuticular wax
is evidently a simply inherited trait. Krukoff (Krukoff &- Barneby, 1 9 7 3 ) separated the Mexican
species Erythrina americana Miller and E. coralloides A. DC. (sect. Erythrina) solely on the
basis of presence or absence of epicuticular wax.
This trait is variable and is not well correlated with
geographic distribution. For this and other reasons,
Erythrinn coralloides is here considered a synonym of E. americnna.
Inheritance of LeafSurface Characters in
Interspecijc Hybrids
Each of the six plates comprising Figures 287 5 illustrates the leaf surface features of a single
F , hybrid and its two parents. On each plate, the
female parent is illustrated on the left, the male
parent on the right, and the hybrid in the center.
Each horizontal row of photographs is a comparison
of the three individuals at equal magnification (indicated by the bar in the left-hand photograph).
Table 1 8 summarizes the features present in the
parents and hybrids.
Four of the six hybrids illustrated were derived
from the same genetic individual as female parent,
Erythrinn guntemnlensis P T 7 0 0 0 1 8 0 0 1 and W-A
7 4 ~ 1 4 5 3 It
. is particularly instructive to note the
pattern of inheritance in the hybrids produced from
the combination of this genome with those of four
different species: E. berteroann, E. chiapasana,
E. abyssinica, and E. senegalensis.
Erythrina guatemalensis has well-developed
papillae on the abaxial leaf surface, each composed
of a single epidermal cell; the male parent E. berteroana has well-developed lamellae, each composed of about 4-5 cells forming a "wall-like"
structure. The F, hybrid has lamellae intermediate
in length between the two parents, composed of
2-3 cells, but these a r e lower in stature and less
developed than the epidermal sculpturing of either
parent (Figs. 37-45).
Erythrina guatemalensis lacks hairs on mature
leaf surfaces. The male parents E. chiapnsnnn
and E. abyssinicn have dense covering of twoarmed hairs. The hybrids derived from these males
with E. guatemalensis as female also possess twoarmed hairs, but at a much lower density than in
the male parents (Figs. 46-5 1, 52-60). Erythrinn
senegalensis has scattered balloonlike hairs, and
these are also inherited in the F , hybrid E. guatemalensis x E. senegalensis (Figs. 61-69). T h e
male parents E. chiapasana and E. senegalensis
lack epicuticular wax; this trait is present in the
female E. gz~ntemnlensisand in the hybrids. (Other
individuals of E. chiapasana than the one used in
this cross do have epicuticular wax.)
Similar patterns of inheritance are exhibited by
the other F, hybrids, for example, Erythrina chiapasana x E. berteroana (Figs. 28-36). The female parent E. chiapnsnnn has a dense covering
of hairs and lacks lamellae and epicuticular wax.
The male parent E. berteroana lacks hair but
possesses lamellae and wax. The F, hybrid has
scattered hairs, lamellae reduced in stature, and
epicuticular wax.
I n the interspecific hybridization that produced
Erythrina lysistemon x E. speciosa, the male
parent possesses two-armed hairs, which are lacking in the female parent; hairs a r e present in the
F , hybrid, but again, a t a rather lower density than
in the male parent (Figs. 70-75).
These results demonstrate that it is possible to
confirm hybridity in the progeny by examination
of leaf epidermal characters. Many of the hybrids
possess characters present in the male parent but
absent in the female parent. There is no evidence
FIGURES
82-87. SEJI images, abaxial leaf surfaces of E. crista-galli x E. guatemalensis and parents. All
photos at equal magnijcation.-82. E. crista-galli, W A 7 4 ~ 8 4 0female
,
parent.-83.
E. guatemalensis, W A
E. crista-galli x E. guaternalensis, four F, siblings, PT 820548 and W A 82.278.
7 4 ~ 1 4 5 3male
,
parent.-84-87.
FIGURES
88-93. SEhl images, abaxial leafsurfaces ofErythrina crista.galli x E. fusca and parents, and E.
dominguezii. All photos at equal magnijcation.-88. E. crista-galli, W A 7 4 ~ 8 4 0female
,
parent.-89.
E, fusca,
W A 7 4 ~ 9 9 male
,
parent.-90-92.
E. crista-galli x E. fusca, three F, siblings, PT 840231 and HO 84.235.93. E. dorninguezii, PT 740234001.
FIGURES
94-99. SEM images, abaxial leafsurfaces of Erythrina crista-galli x E. fusca and parents, and E.
E. fusca,
dorninguezii. All photos at equal magnijcation.-94. E. crista-galli, W A 74,11840,female parent.-95.
W A 7 4 ~ 9 9 male
,
parent.-96-98.
E. crista-galli x E. fusca, three F, siblings, PT 840231 and HO 84.235.99. E. dominguezii, PT 740234001.
Volume 75, Number 3
1988
Neil1
Erythrina
Annals of the
Missouri Botanical Garden
Volurne 75, Number 3
1988
Neill
Erythrina
Annals of the
Missouri Botanical Garden
Volume 75, Number 3
1988
Neil1
Erythrina
TABLE19. Comparison o f l o w e r s o f F , hybrids and their parents in sect. Erythrina. Color code in "Standard
Color" refers to Berlin & K a y color chart; see Croat (1983).
F, Hybrid
Female Parent
Male Parent
E. berteroana
Figure 103
E. guatemalensis
WA 7 8 ~ 5 6 4
HO 82.647
WA 7 4 ~ 1 4 5 3
CALYX
Shape
Apex shape
Length (em)
Vexillar side
Carinal side
Width (em)
Greatest
Middle
Indumentum
Texture
Color
oblong
oblique; longer on carinal side
oblong to elliptic
oblique; longer on carinal side
elliptic
irregular; bilabiate or
oblique
glabrous
smooth
pale red
sparsely puberulent
minutely papillate
red to reddish brown
glabrous
minutely papillate
reddish brown
red 8/7.5
8.7
1.6
red 8/7.5
7.5
1.6
red 6/7.5
6.5
1.8
apex acute
1.2
0.3
apex acute to rounded
1.1
0.4
apex rounded
1.3
0.3
emarginate; each half
acute at apex
1.O
0.8
apex short-apiculate
apex rounded
1.1
1.O
1.3
1.0
2.2-2.3
2.6-2.7
COROLLA
Standard
Color
Length (em)
Greatest width (cm)
Wings
Shape
Length (em)
Width (cm)
Keel
Shape
Length (em)
Width (cm)
E. guatemalensis
Figure 104
WA 7 4 ~ 1 4 5 3
E. tajumulcensis
P T 820547001
WA 7 4 ~ 1 4 4 8
CALYX
Shape
Apex shape
Length (em)
Vexillar side
Carinal side
Width (em)
Greatest
Middle
Indumentum
Texture
Color
elliptic
variable; bilabiate or
oblique
oblong
oblique; longer on carinal side
oblong
oblique; longer on carinal
side
glabrous
minutely papillate
reddish brown
glabrous
smooth
glabrous
smooth
red
red 6/7.5
6.5
1.8
red 7/7.5
7.2
1.5
red 6/7.5
9.3
1.3
COROLLA
Standard
Color
Length (em)
Greatest width (em)
Annals of the
Missouri Botanical Garden
TABLE
19.
Continued.
Female Parent
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (cm)
Width (cm)
Figure 105
F , Hybrid
Male Parent
oblong; apex rounded
1.3
0.4
oblong; apex rounded
1.o
0.4
oblong, apex rounded
1.o
0.3
apex rounded
emarginate; each half
short-apiculate at
apex
1.1
1.8
apex long-acuminate
1.3
1.0
E. guatemalensis
PT 700018001
HO 82.282
1.3
1.0
E. folkersii
PT 700010001
CALYX
Shape
Apex shape
Length (cm)
Vexillar side
Carinal side
Width (crn)
Greatest
Middle
Indurnentum
Texture
Color
elliptic; broadest in middle
irregular; oblique or bilabiate
1.2-1.3
1.2-1.3
glabrous
minutely papillate
reddish brown
cuneate; broadest at
apex
oblique to truncate
cuneate; broadest at apex
oblique; longer on side
sparsely puberulent
minutely papillate
reddish brown to purplebrown
densely puberulent
minutely papillate
light brown
red 6/7.5
6.5
1.8
red 6/10
7.0-7.8
2.2
red 7/7.5
8.0
2.4
apex rounded
1.3
0.4
apex rounded
1.1-1.3
0.5
apex rounded
1.0
0.5
apex rounded
apex variable; rounded
or ernarginate
1.2-1.3
0.9
emarginate, each half obtuse at apex
1.o
1.3
COROLLA
Standard
Color
Length (cm)
Greatest width (cm)
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (crn)
Width (cm)
E. guatemalensis
Figure 106
PT 700018001
HO 82.284
PT 820278002
(2 individuals)
E, chiapasana
PT 73071001
CALYX
Shape
Apex shape
Length (cm)
Vexillar side
Carinal side
elliptic
variable; bilabiate or
oblique
2.3-2.7
2.6-2.9
oblong to elliptic
truncate to oblique; 5
apical lobes in bud,
absent in anthesis
oblong
truncate; 5 apical lobes in
bud, absent at anthesis
Volume 75, Number 3
1988
TABLE
19.
Neill
Erythrina
Continued.
Female Parent
Width (cm)
Greatest
Middle
Indumentum
Texture
Color
F , Hybrid
Male Parent
1.2-1.3
1.2-1.3
glabrous
minutely papillate
sparsely puberulent
smooth
reddish brown
reddish brown to green
densely puberulent
smooth to indistinctly
5-angled
green
red 6/7.5
6.5
1.8
red 5/5 to 5/7.5
6.5
1.5-1.8
red 5/7.5
7.4
1.4
oblong; apex rounded
1.3
0.4
oblong; apex rounded
1.5
0.4
oblong; apex rounded
1.3-1.4
0.4
apex rounded
variable; apex rounded
or emarginate
1.0-1.3
0.9-1.0
emarginate; each half
short-apiculate
1.3
0.9
COROLLA
Standard
Color
Length (cm)
Greatest width (cm)
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (cm)
Width (cm)
Figure 107
1.3
1.O
E. guaternalensis
P T 700018001
HO 82.285
HO 82.288
PT 820276001
E. rnacroplzylla
P T 750420002
CALYX
Shape
Apex shape
elliptic; broadest in middle
variable; bilabiate or
oblique
oblong to elliptic
cuneate; broadest at apex
truncate to slightly
oblique; 5 irregular
apical lobes
truncate; 5 prominent,
blunt lobes
1.2-1.3
1.2-1.3
glabrous
1.2 (at apex)
0.9
densely puberulent
minutely papillate
reddish brown
1.0-1.2
1.0-1.2
sparsely to densely puberulent
obscurely 5-angled
reddish brown to green
longitudinally 5-angled
brown to green
red 6/7.5
6.5
1.8
red 6/7.5 to 6/10
6.4-7.0
1.7-2.1
red 5/7.5
6.4
1.7
oblong; apex rounded
1.3
oblong; apex rounded
1.3-1.6
0.4-0.5
oblong; apex rounded
1.3
0.4
apex rounded
1.3
1.o
apex short-apiculate
1.1-1.3
1.1-1.2
apex short-apiculate
1.3
1.1
Length (cm)
Vexillar side
Carinal side
Width (cm)
Greatest
Middle
Indumentum
Texture
Color
COROLLA
Standard
Color
Length (cm)
Greatest width (cm)
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (cm)
Width (cm)
Annals of the
Missouri Botanical Garden
TABLE1 9 .
Continued.
F, Hybrid
Female Parent
-
E. macrophylla
Figure 1 0 8
Male Parent
-
P T 750420002
HO 8 2 . 7 6 3
( 2 individuals)
E. guatemalensis
P T 700018001
CALYX
Shape
Apex shape
Length (cm)
Vexillar side
Carinal side
Width (cm)
Greatest
Middle
Indumentum
Texture
Color
elliptic
elliptic; bilabiate in middle
variable; bilabiate or
truncate; 5 indistinct
apical lobes present or
lacking
variable; bilabiate or
oblique
glabrous
minutely papillate
brown to green
densely puberulent
smooth to minutely papillate
reddish brown to green
reddish brown
red 5/7.5
6.4
1.7
red 5 / 7 . 5
6.5-6.7
1 .8
red 6/7.5
6.5
1.8
oblong; apex rounded
1.3
0.4
oblong; apex rounded
1.3
0.4-0.5
oblong; apex rounded
1.3
0.4
apex short-apiculate
1.3
1.1
apex rouuded
1.0-1.1
1.1-1.2
apex rounded
1.3
1 .o
cuneate; broadest at
apex
truncate; 5 prominent
apical lobes
1 . 2 (at apex)
0.9
densely puberulent
longitudinally 5-angled
COROLLA
Standard
Color
Length (cm)
Greatest width (cm)
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (cm)
Width (cm)
Figure 1 0 9
E. chiapasana
P T 721005001
HO 82.278
(2 individuals)
E. beteroana
P T 700044002
CALYX
Shape
Apex shape
Length (cm)
Vexillar side
Carinal side
Width (cm)
Greatest
Middle
Indumentum
Texture
Color
oblong
truncate
oblong
oblique; longer on carinal
side
indistinct longitudinal
ridges
green
indistinct longitudinal
ridges
pale green to red
smooth
red 5/7.5
6.8
1.4
red 6 / 5 to 8 / 7 . 5
8.5
1.6-1.7
red 8 / 7 . 5 to 9/7.5
8.3
1.6
oblong
truncate; 5 indistinct
apical knobs
1.5-1.6
1.5-1.6
0.8
0.7
pale red
COROLLA
Standard
Color
Length (cm)
Greatest width (cm)
Volume 75, Number 3
TABLE19.
Neill
Erythrina
943
Continued
-
F, Hybrid
Female Parent
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (cm)
Width (cm)
Figure 1 1 0
oblong; apex rounded
-
Male Parent
oblong; apex acute
1.5
0.4
oblong; apex acute to
rounded
1.2
0.4
emarginate, each half
short apiculate
1.3
0.9
apex rounded to emarginate
1.0
0.9
deeply emarginate; each
half acute at apex
0.8
0.9
E. macrophylla
PT 750420002
1 .O
0.4
E. beteroana
P T 700044001
HO 82.281
( 2 individuals)
CALYX
Shape
Apex shape
Length (cm)
Vexillar side
Carinal side
Width (cm)
Greatest
Middle
Indumentum
Texture
Color
cuneate; broadest at
apex
truncate; 5 prominent
apical lobes
oblong
oblong, narrow
truncate; 5 indistinct
apical lobes
oblique; longer on carinal
side
glabrous
smooth
brown to green
sparsely puberulent
indistinct longitudinal
ridges
green to red
red 5/7.5
6.4
1.7
red 6 / 7 . 5
7.3-7.4
1.5-1.6
red 8 / 7 5
8.3
1.3
oblong; apex rounded
oblong; apex acute to
rounded
1.0-1.2
0.4
oblong; apex acute
1.0
0.4
apex apiculate or emarginate
1.0-1.1
0.8-1.0
deeply emarginate; each
half acute at apex
0.8
0.9
2.0
2.0
1.2
0.9
densely puberulent
longitudinally 5-angled
pale red to green
COROLLA
Standard
Color
Length (cm)
Greatest width (cm)
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (cm)
Width (cm)
1.3
0.4
apex short-apiculate
1.3
1.1
of matrocliny or female-dominant inheritance in
the Erythrina hybrids.
The results, although preliminary, also suggest
a difference in the genetics of inheritance of hairs
on the one hand, and papillae and lamellae on the
other. Hairs are inherited in the hybrids as discrete
characters-that
is, they are fully formed and of
normal size, although they occur at low densities,
in crosses between hairy and hairless parents. The
formation of hairs may thus be controlled by a
single gene or supergene, with modifiers controlling
density of the hairs. On the other hand, in crosses
between papillate (or lamellate) and nonpapillate
parents, papillae (lamellae) may be present but they
are much reduced in stature. This suggests that
the stature of papillae and lamellae are continuously variable, typical of morphometric traits, and
controlled by many genes, each of small effect.
Annals of the
Missouri Botanical Garden
TABLE20. Comparison ofjowers of Erythrina atitlanensis and F, hybrid, E . macrophylla
(Photo, Fig. 11 1.)
x
E. berteroana.
E. atitlanensis
E. macroplzylla x E. berteroana
WA 7 4 ~ 9 8
WA 7 5 ~ 1 1 4 1
HO 82.281
CALYX
Shape
Apex shape
oblong
truncate; 5 indistinct apical lobes
oblong
truncate; 5 indistinct apical lobes
Length (cm)
Vexillar side
Carinal side
Width (cm)
Greatest
Middle
Indumentum
Texture
Color
1.0
0.7
sparsely puberulent
indistinct longitudinal ridges
pale green to red
1.0-1.1
0.7-0.8
sparsely puberulent
indistinct longitudinal ridges
green to red
red 6/7.5
6.5-6.8
1.7
red 6/7.5 to 7 / 1 0
7.3-7.4
1.5-1.6
oblong; apex rounded
1 .o
0.4
oblong; apex acute or rounded
1.0-1.1
0.4
apex short-apiculate
0.8
apex apiculate or emarginate
1.0-1.1
0.8-1.0
COROLLA
Standard
Color
Length (cm)
Greatest width (cm)
Wings
Shape
Length (cm)
Width (cm)
Keel
Shape
Length (cm)
Width (cm)
This suggestion requires confirmation by analysis
of segregation in the F, generation.
The variation in the four F, hybrid siblings derived from a single cross, Erythrina crista-galli x
E. guatemalensis, is illustrated in Figures 76-87.
The female parent has an irregular network of low
lamellae less than 10 ym tall and lacks epicuticular
wax. The male parent has a dense covering of
unicellular papillae up to 40 ym tall and has epicuticular wax. The F, hybrids exhibit a narrowly
segregating array of these characters: they have
papillae and/or lamellae intermediate in form and
stature between the two parents. Some F,s resemble
the female parent more closely and some resemble
the male parent. All the F,s have epicuticular wax,
evidently derived from the male parent. (Other
individuals of E. crista-galli, besides the one used
in this cross, do possess wax.)
The leaf epidermis of three F, siblings derived
from the cross Erythrina crista-galli x E. fusca,
together with their parents, are illustrated in the
two plates comprising Figures 88-99. Also includ-
ed on these plates are photos of E. domingr~ezii,
a species that, based on floral characters, may be
a stabilized derivative of hybridization between E.
crista-galli and E. f i s c a (see discussion of hybrid
flowers below). The male parent Erythrina f i s c a
has a very unusual epidermal surface with deep
and irregularly sized and shaped cavities, knobs
and protrusions, appearing under SEhiI much like
the surface of a limestone cavern. The stomata are
at the bottom of the cavities.
The F, hybrids vary somewhat in surface configuration, but none of them possess either the
regular network of lamellae present in E. cristagalli or the complex, irregular cavity structure of
E. jusca. Two of the F,s have scattered low papillae
and one is nearly flat on the abaxial surface. All
three F,s have epicuticular platelets, which are not
present in either of the parents.
Erythrina domingr~ezii(Figs. 9 3 , 99) has scattered balloonlike hairs, not present in E. cristagalli or E. jusca. Otherwise, the epidermal surface
on E. dominguezii has no distinctive features. Low
Volume 75, Number 3
1988
Neill
Erythrina
FIGURE103. Flowers and buds of Erythrina berteroana x E. guatemalensis and parents. Left: E. berteroana,
WA 7 8 ~ 5 6 4female
,
parent. Center: E. berteroana x E. guatemalensis, HO 82.647. Right: E. guatemalensis, WA
7 4 ~ 1 4 5 3male
,
parent.
epidermal convolutions similar to the lamellae of
E. crista-galli are visible at high magnification
(Fig. 99) but these are not organized into a regular
reticulate pattern.
vember 1984. These included the hybrids within
sect. Erythrina and the intersectional hybrid Erythrina crista-galli x E. fusca.
Materials and Methods
FLORAL FEATURES: INHERITANCE IN
INTERSPECIFIC HYBRIDS
The inheritance of floral morphology and color
was examined in the hybrids that flowered by No-
Fresh flowers of the hybrids and parents were
fixed in FAA, which preserved their three-dimensional shape, and later measured, described, and
photographed, each hybrid together with its par-
FIGURE104. Flowers and buds of Erythrina guatemalensis x E. tajumulcensis and parents. Left: E. guatemalensis, WA 7 4 ~ 1 4 5 3female
,
parent. Center; E. guatemalensis x E. tajumulcensis, PT 820547. Right: E.
tajumulcensis, WA 7 4 ~ 1 4 4 8male
,
parent.
Annals of the
Missouri Botanical Garden
FIGURE105. Flowers and buds of Erythrina guatemalensis x E. folkersii and parents. Left: E. guatemalensis,
PT 700018001, female parent. Center: E. guatemalensis x E. folkersii, two F, siblings, HO 82.282. Right: E.
folkersii, PT 700010001, male parent.
ents. Color was determined from fresh flowers at
the time of collection. A Berlin & Kay color chart
(Berlin & Kay, 1969) was used for color descriptions of corolla standards. For use of the Berlin &
Kay color chart in botanical descriptions see Croat
(1983). Colors are reported in the form "Red
5/7.5." The number preceding the slash refers to
brightness (1-9; 1 is brightest) and the number
after the slash refers to the hue.
Inflorescences and floral details were photo-
FIGURE
106. Flowers and buds of Erythrina guatemalensis x E. chiapasana and parents. Left: E. guatemalensis,
PT 700018001, female parent. Center: E. guatemalensis x E. chiapasana, two F, siblings, HO 82.284 and PT
820278002. Right: E. chiapasana, PT 730710001, male parent.
Volume 75, Number 3
1988
Neill
Erythrina
FIGURE
107. Flowers and buds ofErythrina guatemalensis x E. macrophylla andparents. Left: E. guatemalensis,
PT 700018001, female parent. Center: E. guatemalensis x E. macrophylla, three F, siblings, HO 82.285, HO
82.288, PT 820276001. Right: E. macrophylla, PT 750420002, male parent.
FIGURE
108. Flowers and buds ofErythrina macrophylla x E. guatemalensis and parents. Left: E. macrophylla,
PT 750420002, female parent. Center: E. macrophylla x E. guatemalensis, two F, siblings, HO 82.763. Right:
E. guatemalensis, PT 700018001, male parent.
Annals of the
Missouri Botanical Garden
FIGURE
109. Flowers and buds of Erythrina chiapasana x E. berteroana and parents. Left: E. chiapasana, PT
721005001, female parent. Center: E. chiapasana x E. berteroana, two F, siblings, HO 82.278. Right: E.
berteroana, PT 700044002, male parent.
FIGURE
110. Flowers and buds of Erythrina macrophylla x E. berteroana and parents. Left: E. macrophylla,
PT 750420002, female parent. Center: E. macrophylla x E. berteroana, two F, siblings, HO 82.281. Right: E.
berteroana, PT 700044001, male parent.
Volume 75, Number 3
1988
Neill
Erythrina
FIGURE
111. Comparison ofjowers of Erythrina macrophylla x E. berteroana and Erythrina atitlanensis. Left:
E. macrophylla x E. berteroana, HO 82.281, two F, siblings. Top right: E. atitlanensis, WA 7 4 ~ 9 8Bottom
.
right:
E. atitlanensis, WA 7 5 ~ 1 11
4.
graphed at the time of collection. Herbarium
vouchers are deposited at Missouri Botanical Garden (MO).
Results
Hybrids within Sect. Erythrina
Injorescence and Flower Orientation. Species
of sect. Erythrina all have erect inflorescences,
but they differ in the arrangement of the flowers
on the inflorescence axis (congested or open), length
of the axis, and orientation of the flowers (ascending, horizontal, or descending). These traits are
generally intermediate in the hybrids (e.g., Figs.
100-102). In Erythrina guatemalensis, the female parent, the flowers are horizontal on an open
inflorescence. In E. folkersii, the male parent, the
flowers descend to nearly vertical on a congested
inflorescence. The F, hybrid is intermediate in both
these traits.
A comparison of floral characters of the hybrids and their parents within sect.
Erythrina is summarized in Table 19. The flowers
are illustrated in Figures 103-1 10.
In all characters-color,
indumentum, shape,
and morphometric dimensions-the F, hybrids are
intermediate between the two parents. The F, siblings from a single cross vary to some extent. There
is no evidence of matrocliny or maternal dominance.
Subjectively, some of the hybrids resemble the
male parent more closely than the female parent.
This is evident in the progeny of the reciprocal
hybridizations between Erythrina guatemalensis
and E. macrophylla (Figs. 107, 108). The vestiture and shapes of the calyces of the hybrids are
Floral Characters.
Annals of the
Missouri Botanical Garden
TABLE2 1.
Comparison ofjZowers ofErythrina crista-galli x E. fusca, its parents, a n d E. dominguezii.
E cr~sta- gall^
WA 7 4 ~ 8 4 0
Orientation
E cr~sta- gall^ X
E fusca
PT 84021001
E jusca
PT 840231001
E dom~nguezi~
WA 7 4 ~ 8 6 5
resupinate; standard beneath,
open
standard semi-cleistogamous, folded
over reproductive parts
standard reflexed
standard semi-cleistogamous, folded
over reproductive
parts
red
bowl shaped
reddish brown
bowl shaped; slightly asymmetric
brown
asymmetrically
bowl shaped
pale orange-pink
asymmetrically bowl
shaped
1.8 ern
1.0 cm
large subulate
tooth, carinal
side
1.6 ern
1.4 cm
narrow tooth, carinal side
1.4 cm
2.1 cm
large blunt tooth,
carinal side
1.5 cm
1.4 cm
blunt tooth, carinal
side
red
5.0 cm
3.3 cm
orange
6.1 cm
4.9 cm
orange-pink
6 . 5 cm
5.5 cm
asymmetric; broadest at base
1.3 cm
0.9 cm
obovate, rounded,
cucullate
1.8 cm
1.0 cm
ohovate, cucullate,
broadly rounded
2.9 cm
1.8 cm
small, obovate, cucullate
0.6 cm
0.4 cm
red
pale red
pale red
falcate, acute at
apex
ovate-falcate,
rounded at apex
ivory at base, red
at apex
ovate-falcate,
broadly rounded
at apex
3.9 cm
1.5 cm
CALYX
Color
Shape
Length
Carinal side
Width at apex
Apex ornamen.
tation
COROL.1.A
Standard
Color
Length
Width
Wings
Shape
Length
Width
Keel
Color
Shape
Length
Width
intermediate and variable, but in both reciprocals
the hybrids resemble the male parent somewhat
more than the female.
In crosses involving species with tomentose and
glabrous calyces, the hybrids are tomentose, but
sometimes sparsely so. In color values and morphometric dimensions, the hybrids are generally
intermediate between the two parents.
The flower of the hybrid Erythrina macrophylla x E, berteroana is intermediate between
the two parents, and it closely resembles a third
recognized species, E, atitlanensis (Fig. 1 1 1, Table 20). The principal difference is that the calyx
of the F, hybrid is somewhat longer than that of
E. atitlanensis. The natural distribution of Erythrina atitlanensis is confined to a small area near
Lake Atitlan in western Guatemala, and it is geo-
falcate, acute at apex
graphically and ecologically intermediate between
E. macrophylla and E. berteroana. The possibility
that the form known as E. atitlanensis represents
either a hybridizing population or a stabilized species
of hybrid origin will be discussed below.
Flowers of Erythrina crista-galli x E. fusca
The inflorescence and flowers of this intersectional hybrid and its parents are illustrated in Figures 1 12- 1 1 7 and described in Table 2 1 . A third
species, E. dominguezii, is included in the illustrations and descriptions for reasons discussed below.
The morphometric dimensions and proportions
of the floral parts of the two parental species are
relatively similar, considering the total range of
Volume 75, Number 3
1988
Neill
Erythrina
FIGURES
112-1 15. Inj9orescences in natural position ofErythrina crista-galli x E. fusca, its parents, and E.
dominguezii.-112. E. crista-galli, WA 7 4 ~ 8 4 0 . - 1 1 3 . E. fusca, WA 74~99.-114. E. crista-galli x E. fusca, PT
840231001.-115. E. dominguezii, PT 740234001.
variation of these traits in the genus Erythrina,
but the way in which these parts are arranged and
the overall appearance of the flowers are very
different. The flower of E. crista-galli is resupinate
(inverted from the usual position, with the standard
below the keel) and the red standard is flattened
out, an unusual trait in Erythrina. The orange
standard of E. fusca is reflexed from the clawed
base, exposing the reproductive parts, and is broadly folded down the middle.
The flower of the F, hybrid E. crista-galli x
E.fusca is different from either parent. The hybrid
flower is semicleistogamous, with the standard tightly
folded over the wings, keel, and reproductive parts.
In this semicleistogamous form, in the orientation
of the flowers and the inflorescence, and in the
pale
. pink-orange
.
- color of the corolla standard, E.
crista-galli x E. fusca bears a striking resemblance to E. dominguezii (Figs. 1 14-1 17). Certainly the hybrid resembles E. dominguezii more
closely in overall appearance than either of its
parents. The dimensions of the floral parts are not
identical in the F, hybrid and in E. dominguezii
(Fig. 117, Table 21). The overall similarity between the two could be a coincidence, but it is so
striking and so unexpected that it raises the possibility that Erythrina dominguezii is in fact a
hybrid derivative of E. crista-galli and E. fusca.
The distribution of E. dominguezii is geographically intermediate but ecologically distinct from E.
Annals of the
Missouri Botanical Garden
FIGI.RE116. Flowers ofErythrina crista-galli x E. fusca, its parents, and E. dominguezii, showing approximate
natural orientation. Top roul, left to right: E. crista-galli, WA 7 5 ~ 8 4 0female
,
parent; E . crista-galli x E. fusca,
PT 840231001; E. fusca, WA 7 4 ~ 9 9male
,
parent. Bottom center: E. dominguezii, PT 740234001.
crista-galli and E. fusca. This will be discussed in
greater detail in Section 6.
Conclusions: Inheritance of Phenetic
Traits in Interspecific Hybrids
The results of the morphological studies of the
F, hybrids clearly demonstrate that the progeny
are indeed of hybrid origin. Almost universally, the
F, progeny meet the criterion of intermediacy, and
frequently they possess traits present in the male
parent but absent in the female parent. Matrocliny
is not indicated in E r y t h r i n a hybrids. Some of the
F, hybrids closely resemble forms occurring in natural populations and recognized as species.
6. NATURAL
HYBRIDIZATION
A ~ D
SECTIOA
HYBRID
S PECI~TION
The Mexican state of Chiapas has great geographical diversity and complexity and a very large
flora for an area its size. Climate ranges from
semidesert to rainforest, and elevation from sea
level to over 4,000 m. The flora of Chiapas contains
more than 8,000 plant species and 13 major vegetational formations recognized by Breedlove (1981).
Chiapas, together with adjacent western Guatemala, is also the center of diversity of Erythrina
sect. E r y t h r i n a . Eleven species are known to occur
in the state, and six of these are endemic or shared
only with western Guatemala. Although they are
not found in abundance or in large populations,
species of Erythrina occur in virtually every vegetation type in Chiapas except the upper belts of
the cloud forest and elfin forest on the highest
peaks.
In common with the usual pattern of distribution
in E r y t h r i n a , the Chiapas species of the genus are
mostly allopatric. However, at some localities, particularly at the margins of distribution of the species,
different species do come into contact, and there
natural hybrids are formed.
One phenomenon that has apparently occurred
with E r y t h r i n a in Chiapas and perhaps elsewhere
is spontaneous hybridization in man-made populations. Throughout Mesoamerica many species of
E r y t h r i n a trees are used by the local populace a s
"living fenceposts." Erythrinas take root readily
from woody cuttings and the trunks are ideal posts
for stringing barbed wire. Extensive fencerows of
the plants line roads and fields in many areas.
Volume 75, Number 3
1988
Neill
Ery thrina
FIGURE
1 17. Dissected petals ofErythrina crista-galli x E. fusca, its parents, and E. dominguezii (eachjower,
left to right: standard, wings, keel). Top: E . crista-galli, WA 7 4 ~ 8 4 0jkrnale
,
parent. Center left: E. crista-galli x
E. fusca, PT840231001. Center right: E. dominguezii, PT 740234001. Bottom: E. fusca, WA 7 4 ~ 9 9rnale
,
parent.
Sometimes two species are cultivated together, and
hybrids, apparently produced spontaneously in situ,
are occasionally found in these fencerows.
An analysis of hybridizing populations involving
three species of Erythrina in central Chiapas, Errthrina chiapasana, E. goldmanii, and E. pudica, is presented below. Distributions of these
species and their hybrid populations a r e shown in
Figure 1 1 8 .
Erythrina chiapasana x E. goldmanii
Erythrina chiapasana is a tree of the pineoak forests of the Central Plateau of Chiapas, occurring primarily above 1 , 5 0 0 m . Erythrina goldmanii inhabits the dry tropical deciduous forests
of the Central Depression of Chiapas, formed by
the highland-rimmed valley of the Rio Grijalva. At
El Sumidero National Park a few km north of the
city of Tuxtla Gutierrez, where the Rio Grijalva
cuts through the limestone of the Central Plateau
on its way to the Atlantic Ocean and forms a
spectacular 800-m-deep canyon, the two species
occur parapatrically and a hybrid zone is found
about 2 km wide and extending about 3 0 0 m along
a n elevational gradient (Fig. 119).
Throughout their respective distributions, Erythrina chiapasana and E. goldlnanii exhibit some
intraspecific variation, but the two species are readily
distinguishable morphologically. The leaves of E.
chiapasana are densely tomentose with two-armed
hairs on the abaxial surface (Fig. 120). The leaves
of E. goldmanii are glabrous or nearly so a t maturity and are aculeate along the midvein and primary veins of the abaxial surface (Fig. 1 2 2 ) . The
calyx of E. chiapasana is green to reddish, densely
puberulent, and truncate at the margin without a
prominent tooth on the carinal side; the corolla
standard is dark red. The calyx of E. goldlnanii
is broader, dark purple brown to nearly black,
glabrous, and provided with a prominent apical
tooth on the carinal side; the corolla standard is
usually pale red.
At El Sumidero both species are a t the altitudinal
and geographical limits of their ranges. Only individuals with the "pure" E. chiapasana phenotype are found in the oak-dominated forest at the
plateau summit above 1 , 1 0 0 m; only individuals
Volume 75, Number 3
1988
Neill
Erythrina
C A R O N DEL SUMIDERO, CHIAPAS
FIGURE119. Cross section of slope at El Sumidero, Chiapas, Mexico, showing distribution of Erythrina
chiapasana, E. goldmanii, and hybrid zone.
alongside accessions of both parental strains (Table
11). When this artificial hybrid flowers it will be
possible to compare it with specimens of the putative natural hybrids from El Sumidero.
As discussed in a separate paper (Neill, 1987),
the hummingbird Heliomaster constantii pollinates Erythrina chiapasana and F. goldmanii at
El Sumidero and is therefore implicated as the
agent directly responsible for interspecific gene flow
in the hybridizing Erythrina population.
E. goldmanii and E. pudica, and occasional intermediate and evidently hybrid individuals occur
there. These intermediates are similar in appearance to the trees occurring in the adjacent natural
hybrid populations. The fencepost hybrids are very
likely the progeny of other fencepost trees that
received interspecific pollen from foraging hummingbirds moving down the line of mixed species
fenceposts. The hybrid seed thus probably germinated directly below its female parent and grew up
to become part of the fencerow itself.
Erythrina goldmanii x E. pudica
Erythrina pudica is a locally endemic species
that is restricted to the dry valley of the Rio de
La Venta, a tributary of the Rio Grijalva, at the
western end of the Central Depression of Chiapas.
This is an unusual species, with the flowers drooping
to nearly parallel with the erect axis of the inflorescence (Fig. 126). The calyx is truncate without
a n apical tooth and covered with a dense grayish
tomentum; the corolla is very pale pink or orangepink.
In the vicinity of Ocozocuautla, Chiapas, at the
eastern margin of its small range, Erythrina pudica occurs sympatrically with Erythrina goldmanii. In disturbed scrub forest small hybrid populations are found, with individuals of both parental
species as well as intermediates (Figs. 124- 126).
Along the highway 5 km east of Ocozocuautla
are living fencerows of Erythrina containing both
Erythrina berteroana
X
E. folkersii
On the Atlantic coastal plain of northern Chiapas
and adjacent states the natural vegetation has been
almost entirely destroyed and replaced with pastures. There, as elsewhere in Mesoamerica, the
pastures and roadsides are commonly lined with
living fencerows of Erythrina trees. On the Atlantic plain the most frequently used species are E.
berterounu and E. folkersii, which are both native
to the region.
Trees morphologically intermediate between Erythrina berteroana and E. folkersii in shape and
vestiture of the calyx and orientation of the flower
occur in northern Chiapas. None of the intermediates set seed. Pollen stainability from four collections of the intermediates (Alexander's stain;
5 0 0 grains per sample: k i l l 5533, 5540, 5 5 4 3 ,
5 5 4 4 ) was 73.2% (range 60.9-86.1%), an un-
Annals of the
Missouri Botanical Garden
usually low figure for Erythrina. These individuals
a r e almost certainly hybrid E r y t h r i n a bertero a n a x E. folkersii. The reason for the low level
of stainable pollen and lack of fruit set is not known;
the experimentally produced hybrids within sect.
Erythrina (Section 4) all had very high pollen
fertility.
These intermediates closely match the type
specimen of E r y t h r i n a caribaea Krukoff Sr Barneby as well as other collections determined by
Krukoff as this species. Despite a protracted search,
I never found this form occurring in a natural
population and never found any seed set on the
fencepost trees. It seems reasonable that Erythrina
caribaea is in fact a hybrid E. berteroana x E.
folkersii and probably occurs only a s a cultivated
fencepost tree.
HYBRID SPECIATION
120-122'
images' abaxiaz leaf surfaces ofErythrina chiapasana, E. goldmanii, and ltybrid
from a ~ o' ~ u l a t i o
atnEl Sumidero. Chiaaas.
' . Mexico,120. E. chiapasana, Neill 5617.-121. E. chiapasana x
E. goldmanii, Neill 5618.-122. E. goldmanii, Neill 5616.
Scale bars = 0.5 mm.
.
I n this paper it has been demonstrated that
diploid Erythrina species are interfertile, that the
hybrids are viable and fertile, and that hybridization
sometimes occurs in natural populations. W h a t has
not yet been shown is the validity of the final
hypothesis set forth in the introductory chapter:
that hybrid speciation has taken place in Erythrin u , that some distinct forms recognized as species
are stabilized derivatives resulting from hybridization of two ~ a r e n t a lspecies, and that this process
has been a n important element in the evolutionary
history of the genus.
Direct and unequivocal evidence relating to this
hypothesis is difficult to obtain. Phylogenies based
on molecular data of the taxa involved, including
studies on isoenzymes and on nucleic acid restriction sites, might in the future provide such evidence. The best evidence available a t present is
the morphological congruence between certain artificially produced hybrids and certain naturally
occurring forms that evidently are stabilized and
self-perpetuating populations.
I n considering the hypothesis of hybrid speciation, there is no reason to assume that the stabilized
derivatives, especially if they became stabilized several generations or more after the original hybridization event, should be precisely intermediate between the parental species or should resemble closely
the F, hybrids. A limited number of F, hybrids is,
however, generally the only material available for
comparison,
I n E r y t h r i n n some of the artificial F, hybrids
do resemble naturally occurring forms recognized
as species, according to the results of the morphological studies presented in Section 5. Fry-
Volume 75, Number 3
1988
Neill
Erythrina
FIGURE
1 23. Flowers of Erythrina chiapasana, E. goldmanii, and hybrid from a population at El Sumidero,
Chiapas, ~Wexico.Left: E. chiapasana, Neill 5455. Center top: E. chiapasana x E. goldmanii, Neill 5493. Center
bottom: E. chiapasana x E. goldmanii, Neill 5466. Right: E. goldmanii, Neill 5495.
thrina macrophylla x E. berteroana bears a close
resemblance to E. atitlanensis, and E. cristagalli x E. fusca resembles in certain features E.
dominguezii. Field studies, which would be valuable for determining whether hybrid speciation could
have occurred, were not conducted in either of
these situations. The known geographical and ecological distribution of the taxa involved is outlined
below.
Erythrina macrophylla is distributed throughout the highlands of Guatemala and western El
Salvador, growing in the pine-oak forests above
1,500 m elevation. Erythrina berteroana, the most
widespread species in sect. Erythrina, is common
in the Pacific coastal plain of Guatemala and on
the lower slopes of the volcanic range that lead up
from the plain to the highlands. The intermediate
known as Erythrina atitlanensis is known only
from the vicinity of Lake Atitlan on the southern
edge of the highlands. In terms of geography and
elevational distribution, E. atitlanensis is precisely
intermediate between the putative parental species.
If hybridization is really implicated in this case, E.
atitlanensis could be merely an early generation
segregate rather than a stabilized, self-perpetuating
derivative. Based on comparison of herbarium
specimens, the progeny cultivated in Hawaii grown
from seed obtained from the population in Guatemala closely resemble the parents. Therefore stabilization of the hybrid form may have taken place.
The case of Erythrina dominguezii and its putative parental species E. crista-galli and E, fusca
is more problematic because the three taxa are so
morphologically distinct. They are also ecologically
distinct. Erythrina crista-galli and E. fusca are
both riparian or estuarine species. Erythrina crista-galli is common along the estuary of the Rio
de La Plata and its tributaries and along the coast
of southern Brazil. The more tropical E. fusca is
distributed widely throughout the Amazon basin
and south along the coast of Brazil. The ranges of
the two species evidently do overlap in southern
Brazil. The putative derivative Erythrina dominguezii also occurs in southern Brazil and westward
through Paraguay and northern Argentina to eastern Bolivia, but it is an upland species of the dry
Chaco forest and cerrado. Erythrina dominguezii
would never have been suspected as a hybrid derivative of E. crista-galli x E. fusca were not its
resemblance to the artificially produced F, so compelling. This situation appears to merit further investigation.
In the introduction, a set of five hypotheses was
stated regarding the species relationships and evolutionary history of Erythrina: 1) The numerous
species of sect. Erythrina can all cross freely with
one another, producing fully fertile hybrids. The
section forms a homogamic complex in which in-
Annals of the
Missouri Botanical Garden
F I G ~ ~ R124-126.
ES
lnjlorescences of Erythrina goldmanii, E. pudica, and hybrid from a population near
Ocozocuautla, Chiapas, .llexico.- 124. E . goldmanii, Neil1 5 5 10.- 125. E. goldmanii x E. pudica, Neil1 5586.126. E. pudica, Neil1 5585.
ternal barriers to hybridization are absent. 2) The
interfertile hornogamic conlplex of sect. Erythrincr
extends, to a greater or lesser degree, to species
in other sections and subgenera of E r ~ , t h r i n a A
. ny
diploid BrJ.thrina species can hybridize with any
other, but crosses between widely divergent taxa
a r e generally difficult to obtain and the resulting
F,s may exhibit varying degrees of sterility. The
genus as a whole rnay be characterized as a series
of interfertile hornogarnic cornplexes with weak to
moderate reproductive barriers between the conl-
plexes. 3) The widely foraging hurnrningbirds that
pollinate species of sect. Erythrincc a r e capable of
effecting interspecific pollen flow between syrnpatric species of sect. E r y t h r i n a . 12) Sympatry a t the
local conlnlunity level is rare among species of
sect. B r y t h r i n a . Most species a r e restricted in geographic range and ecological amplitude and a r e
allopatric, separated by habitat differences. HOT\-ever, sornetirnes different species do corne into contact in nature, and then hybridizing populations
a r e forrned. 5 ) Patterns of distribution and phenetic
Volume 75, Number 3
1988
Neill
Ery thrina
variation in sect. Erythrincr indicate that some
distinct forms recognized as species are stabilized
derivatives resulting from hybridization of two parental species. As a consequence of changing climates and dynamic geomorphological processes,
and the consequent migration of vegetation types
and mixing of floristic elements, formerly allopatric
species may have come into contact a number of
times. With the temporary breakdown of external
isolating barriers, the interfertile species hybridized
and the subsequent segregation and stabilization of
hybrid derivatives have contributed to the proliferation of species of Erythrina.
The data presented in this paper have been
marshalled in support of this set of hypotheses.
The cytological studies (Section 3) and the experimental hybridization and self-compatibility trials
(Section 4) present evidence in support of the first
two hypotheses. In spite of the considerable rnorphological, ecological, and geographic differentiation of Erythrina, the species have retained a high
degree of chromosomal (structural and genic) homology. Within sect. Erythrir~a,this homology, as
evidenced by interspecific compatibility, is virtually
complete: there is no detectable difference in the
success of interspecific matings as compared with
intraspecific matings. At greater taxonomic distances between the two parents (intersectional and
intersubgeneric rnatings), mating success declines
to some extent, but the number of successful "wide
hybridizations" obtained in the experimental trials
indicates that even the most morphologically and
ecologically divergent of diploid Er?.t/zrina species
have retained their ancestral chromosonial and genic
homology and have not evolved substantial barriers
to hybridization in concert with rnorphological differentiation. Erythrina forms a homogamic cornplex of interfertile species, or perhaps a series of
homogamic complexes with weak to moderate barriers between the complexes. Brythrina shares this
pattern of species relationships with many ternperate-zone genera of trees and shrubs. The evidence from Erythrina suggests that the patterns
of species relationships in predominantly or exclusively tropical groups of woody plants may not
differ significantly from the patterns found in their
better-knorzrn temperate-zone counterparts. Formation of homogamic cornplexes may be a comrnorl
phenomenon in tropical woody plants and may be
a n important factor in the evolution of these taxa.
The patterns of inheritance of phenetic traits in
the artificially produced hybrids (Section 5) confirm
the true hybrid nature of these plants and demonstrate that matricliny, a potential con~plicating
factor in the inheritance of these traits and in the
interpretation of hybridization patterns, is not indicated in Erythrina hybrids. The patterns of inheritance in the artificial hybrids reveal the patterns to be expected in the detection and analysis
of natural hybridization: for morphometric characters, a rather narrowly segregating array of intermediate types arnong the hybrids; and for discrete characters such as trichomes, possessed
exclusively by either the female or male parent,
the inheritance of the character in some of the
hybrid offspring, the character being often reduced
in size or density.
Evidence for the third hypothesis, concerning
the pollination of sect. Erythrincc by relatively specialized, widely foraging hummingbirds and the relation of this pollination system to Br.ythrina breeding systems, is presented in a separate paper (Neill,
1987). The pollination studies indicate that interspecific pollen flow and potential natural hybridization are likely to occur arnong sympatric species
of sect. Brythrina.
Evidence for the fourth and fifth hypotheses,
concerning natural hybridization and hybrid speciation in Erythrina, is presented in Section 6.
Natural hybridization was detected among several
co-occurring species of sect. Erythrirla in Chiapas,
Mexico, at the geographical center of diversity of
the section. The natural hybrids display the same
patterns of inheritance of phenetic traits as the
artificial hybrids described earlier. The evidence
for hybrid speciation itself is somewhat more equivocal. As stated in the introduction to this paper,
the final hypothesis is historical and cannot be
tested directly, but can be inferred only by drawing
on information obtained by testing the first four.
The information presented throughout this paper does make plausible the hypothesis of hybrid
speciation in Brythrina. Moreover, the research
reported here provides a unique base of information
for further studies of species relationships and the
evolutionary history of h'rythrina, a s a model of
evolutionary processes in flowering plants that may
be cornrnon to many tropical woody genera. Perhaps the most incisive research that could be carried out at this point in the continuing biosystematic
investigation of Brythrincc ~vouldentail studies of
isoenzymes and particularly of nucleic acid restriction sites among the taxa, as well as the inheritance
of these molecular character states in the hybrids
of known origin, followed by the construction of
phylogenies combining rnolecular data with the
presently available evidence on morphological and
biogeographic patterns and the data from crossing
experiments. The collection of Erythrincc species
and hybrids now available in cultivation at the
Annals of the
Missouri Botanical Garden
Hawaiian botanical gardens provides a n ideal resource for such studies, and it is my hope that my
colleagues specializing in chemosystematics and
molecular phylogenetics do take advantage of this
resource to investigate further the patterns of evolution in this interesting genus.
LITERATURE
CITED
.
Hybridization
ADDISON,G. A. & R. T A V ~ R E S1952.
and grafting in 'pecies of Theobroma which Occur
in Amazonia. Evolution 6: 380-386.
ALEXANDER,
M. P. 1969. Differential staining of aborted
and non-aborted pollen. "ain
14:
122.
M. T. K. 1981. Breeding systems and pollination biology in the Leguminosae. Pp. 723-770 in
R. M. Polhill & P . H. Raven (editors), Advances in
Legume Systematics. Royal Botanic Gardens, Kew.
ASHTON,P. S. 1969. Speciation among tropical trees:
some deductions in the light of recent evidence. Biol.
J. Linn. Soc. 1: 155-196.
ATCHISON,E. 1947. Studies in the Leguminosae. I.
Chromosome numbers in Erythrirza L. Amer. J . Bot.
34: 407-414.
of
AyENSU,E' S' 1977' Scanning
epidermal features in Erythrina (Fabaceae). Lloydia
40: 436-453.
BAAS,P. 1978. Inheritance of foliar and nodal anatomical characters in some llex hybrids. Bot. J . Linn.
SOC.71: 41-52.
B~RETTA-KUIPERS,
T. 1982. w o o d structure of the
genus Erythrirza. Allertonia 3: 53-69.
BEADLE,
G. W . 1932. A gene for sticky chromosomes
in Zea mays. Z. Indukt. Abstammungs-Vererbungsl.
63: 195-217.
BEEKS,R. M. 1955. Improvements in the squash technique for plant chromosomes. Aliso 3: 131-134.
BENTHARI,
G. 1865. I ~ LG.
: Bentham & J. D. Hooker,
Genera Plantarum, Volume I.
BERLIN,B. & P. KAY. 1969. Basic Color Terms, their
Universality and Evolution. Univ. California Press,
Berkeley.
BREEDLO\.E,
D. E. 1981. Introduction to the Flora of
Chiapas. California Academy of Sciences, San Francisco.
F. SCHNEIBRICKELL,
C. D., A. F. KELLY,R. H. RICHENS,
DER & E. G. VOSS (editors). 1980. International
Code of Nomenclature for Cultivated Plants-1980.
Regnum Veg. 104.
CLAUSEN,J., D. P. KECK & W. HIESEY. 1939. The
concept of
based On experiment. Arner. J.
Bot. 26: 103-106.
- -& . 1940. Experimental
studies on the nature of species. I. Effect of varied
environments on western North American plants.
Publ. Carnegie Inst. Wash. 520.
CONEY,P. J. 1982. Plate tectonic constraints on the
biogeography of Middle America and the Caribbean
Region. Ann. Missouri Bot. Gard. 69: 432-443.
COTTAR^, W . P., J. M. TUCKER& F. S. SANTALIOUR,
JR.
1982. Oak hybridization at the University of Utah.
State Arboretum of Utah Publ. No. 1.
CROAT,T. B. 1983. A revision of the genus Anthurium
(Araceae) of Mexico and Central America. Part I:
Mexico and Middle America. Ann. Missouri Bot.
Gard. 70: 211-420.
CTITLER,
D. F. 1972. Leaf anatomy of certain Aloe and
Gasteria species and their hybrids, Pp, 103-122
A. K. M. Ghouse (editor), Research Trends in Plant
Anatomy. McGraw-Hill, New Delhi.
& P. E. BR4NDH4h.l. 1977. Experimental evidence for the genetic control of leaf surface characters in hybrid Aloineae (Liliaceae). Kew Bull. 32:
23-32.
D~RLINGTO
C.ND.
, 1930. Recent Advances in Cytology.
Churchill, London.
EAST, E, M, 1940, The distribution of self-sterility in
flowering plants. Proc. Amer. Phil. Soc. 82: 449518.
EHRENDORFER,
F, 1970, Evolutionary patterns and
strategies in seed plants. Taxon 19: 185-195.
L, A, SX.ARII & J . A.
FEINSIN(;ER,
P , , Y, B, LINHART,
WOLFE. 1979. Aspects of the pollination biology
of three ~
~ species on
~ ~ ~ i ~~ i d~ ~h d ~
Ann. Missouri Bot. Card. 66: 451-471.
P, A, 1957. ~~d~ of reproduction in higher
FRYYELL,
~ l a n t s .Bot. Rev. (Lancaster) 25: 135-233.
GANDERS,
F. R. & K. M. NAGATA.1984. The role of
hybridization in the evolution of Bidens on the Hawai(editor), plant
ian Islands, pp, 179-194 i= W,
Biosystematics. Academic Press, Toronto.
GII,LETT,
G, W , & E, K, S, L111, 1970, An experimental
study of the genus Bidens in the Hawaiian Islands,
Univ. Calif. Publ. Bot. 56: 1-63.
P, 1981a, Chromosome numbers in leGOLDBL4TT,
gumes, 11, Ann, Missouri Bat, Gard, 68: 551-557,
. 1981b. Index to plant chromosome numbers
1975-1978. Monogr. Syst. Bot. Missouri Bot.
Gard. 5.
. 1984. Index to plant chromosome numbers
1979-1981. Monogr. Syst. Bot. Missouri Bot.
Gard. 8 .
& G. DA\ IDSE. 1977. Chromosome numbers
in legumes. Ann. Missouri Bot. Gard. 64: 121-128.
& A. GENTRY. 1979. Cytology of Bignoniaceae. Bot. Not. 132: 475-482.
L, D, 1972, ~~~~l~ of confidence in the
GOTTLIEB,
analysis of hybridization in plants. Ann. Missouri Bot.
Gard. 59: 435-446.
,
1984. Genetics and morphological evolution
in plants, Amer, Naturalist 123: 681-709,
G R ~ H AA.
~ I&
, A. S. TOMB. 1974. Palynology of Erythrina (Leguminosae: Papilionoideae): preliminary
survey of the subgenera, ~ l ~ 37:
~ 465-481,
d i ~
&
. 1977. Palynology of Erythrina
(Fabaceae: Faboideae): the subgenera, sections, and
generic relationships, ~ l ~ 40:~ 413-435.
d i ~
GRANT,V. 1953. The role of hybridization in the evolution of the leafy-stemmed gilias. Evolution 7: 5 1 64.
. 1981. Plant Speciation, 2nd edition. Columbia
Univ. Press, New York.
GUPPY,H. B. 1906. Observations of a Naturalist in the
Pacific between 1 8 9 6 and 1899. MacMillan & Co.,
Ltd., London.
HARDIN,
J. W . 1975. Hybridization and introgression
in Quercus alba. J. Arnold Arbor. 56: 336-363.
: G. A. Engler &
HARLIS,H. 1915. Erythrina. I ~ LH.
0 . Drude, Die Vegetation der Erde 9(3): 656-659.
HAR\EY, W . H. 1861. Flora Capensis, Volume 2.
Hodges, Smith & Co., Dublin.
~ b
Volume 75, Number 3
1988
Neill
Erythrina
HERNANDEZ,
H. M. 1982. Female sterility in Erythrina
montana. Allertonia 3: 72-76.
& V. M. TOLEDO. 1979. The role of nectar
robbers and pollinators in the reproduction of Erythrirza leptorhiza. Ann. Missouri Bot. Gard. 66:
512-520.
JACKSON,
R. C. 1973. Chromosomal evolution in Haplopappus gracilis: a centric transposition race. Evolution 27: 243-256.
,
1984. Chromosome pairing in species and
hybrids. Pp. 67-86 in W. Grant (editor), Plant Biosystematics. Academic Press, Toronto.
& T. N. KHOSHOO.
JALIL,R., M. P ~ LG.
, S. SRI\.ASTAVA
1 9 8 2 . Cytogenetics of Erythrina x resuparcellii
Srivastava. Allertonia 3: 19-24.
KEEP, E. 1962. Interspecific hybridization in Ribes.
Genetica 33: 1-23.
KERIPANNA,
C. & R. RILEY. 1964. Secondary association between genetically equivalent bivalents. Heredity 19: 289-299.
KRUKOFF,B. A. 1939a. The American species of Erythrirza. Brittonia 3: 205-337.
. 1939b. Preliminary notes on Asiatic-Polynesian species of Erythrirza. J. Arnold Arbor. 20: 225233.
& R. C. BARNEBY.1973. Notes on the species
of Erythrirza. VII. Phytologia 27: 108-141.
&. 1974. A conspectus of the genus
Erythrirza. Lloydia 37: 332-459.
LACKEY,
J. A. 1981. Phaseoleae. Pp. 301-328 irz R.
M. Polhill & P. H. Raven (editors), Advances in
Legume Systematics. Royal Botanic Gardens, Kew.
LATRENCE,W. J. 1931. The secondary association of
chromosomes. Cytologia 2: 352-384.
LETIS, W. H. 1974. Chromosomes and phylogeny of
Erythrina (Fabaceae). Lloydia 37: 460-464.
LOUIS,J. 1935. Revision des esphces congolaises du
genre Erythrina L. Bull. Jard. Bot. Etat 13: 295319.
MCKEY,D. 1975. The ecology of co-evolved seed dispersal systems. Pp. 159-191 in L. E. Gilbert & P.
H. Raven (editors), Coevolution of Animals and Plants.
Univ. Texas Press, Austin.
MEARS,J. A. & T. J. MABRY. 1971. Alkaloids in the
Leguminosae. Pp. 73-178 in J. B. Harborne, D.
Butler & B. L. Turner (editors), Chemotaxonomy of
the Leguminosae. Academic Press, London.
M E H R ~P., 1976. Cytology of Himalayan Hardwoods.
Sree Saraswaty Press, Calcutta.
MTILCAHY,
D. L. & G. B. MULCAHY.1983. Gametophytic self-incompatibility reexamined. Science 220:
1247-1251.
MULLER,
C. H. 1952. Ecological control of hybridization
in Quercus: a factor in the mechanism of evolution.
Evolution 6: 147-161.
NEILL,D. A. 1987. Trapliners in the trees: hummingbird pollination of Erythrina sect. Erythrlna (Leguminosae: Papilionoideae). Ann. Missouri Bot. Gard.
74: 27-41.
NETTANCOURT,
D. DE. 1977. Incompatibility in Angiosperms. Springer-Verlag, New York.
NOBS, M. A. 1963. Experimental studies on species
relationships in Ceanothus. Publ. Carnegie Inst. Wash.
623: 1-94.
OREB~MJO,
T. O., G. PORTEOUS
& G. R. STETTART.1982.
Nitrate reduction in the genus Erythrina. Allertonia
3: 11-18.
PRYOR,L. D. 1959. Species distribution and association
in Eucalyptus. Monogr. Biol. 8: 461-471.
RAYEN,P. H. 1974. Erythrirza (Fabaceae): achievements and opportunities. Lloydia 37: 321-331.
. 1977. Erythrina (Fabaceae: Faboideae): introduction to symposium 11. Lloydia 40: 401-406.
. 1980. Hybridization and the nature of species
in higher plants. Canadian Bot. Assoc. Bull. Suppl.
Vol. 13: 3-10.
& D. I. AXELROD. 1974. Angiosperm biogeography and past continental movements. Ann.
Missouri Bot. Gard. 61: 539-673.
& T . E. R,wF.N. 1 9 7 6 . The genus Epilobium
in Australasia: a systematic and evolutionary study.
New Zealand Dept. Sci. Industr. Res. Bull. 216: 1321.
RIDLEY,
H. N. 1930. The Dispersal of Plants Throughout the World. L. Reeve & Co., Ashford, Kent.
.
The genetic control
RILEY,R. & V. C H A P X I ~ N1958.
of cytologically diploid behavior of hexaploid wheat.
Nature 182: 713-715.
SEIBERT,R. J. 1947. A study of Hevea (with its economic aspects) in the republic of Peru. Ann. Missouri
Bot. Gard. 34: 261-353.
SIXIPSON,
B. B. (editor). 1977. Mesquite: its biology in
two desert ecosystems. Dowden, Hutchison & Ross,
Stroudsberg, Pennsylvania.
SKTITCH,
A. 1971. A Naturalist in Costa Rica. Univ.
Florida Press, Gainesville.
SOKAL,R. R. & F. J. ROHLF. 1969. Biometry. W . H.
Freeman & Co., San Francisco.
STACE,C. A. 1984. The taxonomic importance of the
leaf surface. Pp. 67-94 in V. H. Heywood & D.
M. Moore (editors), Current Concepts in Plant Taxonomy. Academic Press, London.
ST~TISTICAL
ANALYSISINSTITCTE. 1982. SAS User's
Guide: Statistics. Statistical Analysis Institute, Cary,
North Carolina.
STEBBINS,
G. L., JR. 1958. The inviability, weakness
and sterility of interspecific hybrids. Advances Genet.
9: 147-216.
STEPHENSON,
A. G. 1981. Flower and fruit abortion:
proximate causes and ultimate functions. Ann. Rev.
Ecol. Syst. 12: 253-280.
RAO,
STIN~A
R Y. 1945. Chromosomes of Erythrina
indica Lamk. J . Indian Bot. Soc. 24: 42-44.
VANVALEN,L. 1976. Ecological species, multispecies,
and oaks. Taxon 25: 233-239.
V E N K ~ T ~ S U B BK.A N
R., 1944. Cytological studies in
Bignoniaceae. Annamalai Univ., Annarnalaingar, India.
Annals of the
Missouri Botanical Garden
APPENDIX
I. Species, sectiorzs, and subgerzera of
Erythrina. All recogrzized taxa in Erythrina are included
in this list. I do not recognize infiaspecijic taxa irz
Erythrina. Proposed taxonomic changes are anticipated
here, prior to their formal designatiorz. The numbering
sequence of Krukoff & Barneby ( 1 974) is followed for
reference to that work, and because the rzumbers were
used to designate the hybrids. There are gaps in the
rzumber sequerzce because of reduction of species to
synonymy. Species reduced to synonymy since Krukoff
& Barrzeby (1974) are indicated at the erzd of this
list.
Erythrina L.
I.
Subgenus Micropteryx (Walp.) F. G. Baker
1. Sect. Duchassaingia (Walp.) Krukoff
1. Erythrina fusca Lour.
2. Sect. Cristae-galli Krukoff
2. Erythrina crista-galli L.
3. Erythrina falcata Benth.
3. Sect. hficropteryx
4. Erythrina domirzguezii Hassler
5. Erythrina ulei Harms
6. Erythrina verna Velloso
7. Erythrir~apoeppigiana (Walp.) 0.
F. Cook
11.
Subgenus Erythrina
4. Sect. Suberosae Krukoff
8. Erythrina suberosa Roxb.
9. Erythrina microcarpa Koord. &
Valeton
10. Erythrina stricta Roxb.
11. Erythrina resupinata Roxb.
5. Sect. Arborescentes Krukoff
12. Erythrina arborrscer~s(Roxb.)
Walp.
6 . Sect. Hypaphorus (Hassk.) Krukoff
13. Erythrina subumbrans (Hassk.)
Merr.
7. Sect. Breuijorae Krukoff
14. Erythrina brrvijora A. DC.
14a. Erythrir~apetraea Brandegee
14b. Erythrir~aoaxacana (Krukoff) Kru.
koff
14c. Erythrina batolobium Barneby &
Krukoff
8. Sect. Edulrs Krukoff
15. Erythrina edulis Triana ex M .
Micheli
15a. Erythrina mrgistophylla Diels
9. Sect. Strr~otropis(Hassk.) Krukoff
16. Erythrir~asprciosa Andrews
10. Sect. Pseudo-edules Krukoff & Barneby
17. Erythrina polychaeta Harms
18. Erythrina schimnpji Diels
1 1. Sect. Leptorhizar Krukoff
19. Erythrina montana Standley
APPENDIXI.
Continued.
20. Erythrirza leptorhiza A. DC.
21. Erythrina horrida A. DC.
2 l a . Erythrina sousae Krukoff
12. Sect. Erythrina
22. Erythrina herbacea L.
23. Erythrirza standleyana Krukoff
24. Erythrirza /labelliformis Kearney
25. Erythrina americana Miller
27. Erythrina pudica Krukoff & Barneby
27a. Erythrirza krukoviana Neill, sp.
nov. ined.
28. Erythrina lanata Rose
29. Erythrina goldmanii Standley
31. Erythrina folkersii Krukoff &
Mold.
32. Erythrina tuxtlana Krukoff & Barneby
33. Erythrina smithiana Krukoff
34. Erythrina cochleata Standley
35. Erythrirza hondurensis Standley
36. Erythrina ch,iapasana Krukoff
37. Erythrina atitlanrnsis Krukoff &
Barneby
38. Erythrina cobar~er~sis
Krukoff &
Barneby
39. Erythrir~azuilliamsii Krukoff &
Barneby
40. Erythrina tajumulcrr~sisKrukoff &
Barneby
41. Erythrina chiriquensis Krukoff
42. Erythrir~amacrophylla A. DC.
43. Erythrina gc~atrrnalensis Krukoff
44. Erythrir~aglobocalyx Porsch &
Cuf.
45. Erythrina steyrrmarkii Krukoff &
Barneby
46. Erythrina jorenciae Krukoff &
Barneby
47. Erythrina berenices Krukoff & Barneby
Kru48. Erythrir~ahuehueter~anger~sis
koff & Barneby
49. Erythrina lar~ceolataStandley
50. Erythrina costaricrnsis M. Micheli
51. Erythrina barqueroana Krukoff &
Barneby
53. Erythrina berteroana Urban
54. Erythrina rubrir~erviaH.B.K.
55. Erythrir~amexicana Krukoff
56. Erythrina saluiijora Krukoff &
Barneby
56a. Erythrina santamartensis Krukoff
& Barneby
57. Erythrir~acastillejijora Krukoff &
Barneby
57a. Erythrina thyrsijora G6mez &
G6mez
Volume 75, Number 3
1988
APPENDIXI.
Neill
Erythrina
Continued.
Gibbosar Krukoff & Barneby
Erythrir~agibbosa Cuf.
Cornlloder~draKrukoff
Erythrinn nrnazor~icaKrukoff
Erythrir~asimilis Krukoff
Erythrina prruviana Krukoff
Erythrina mitis Jacq.
Erythrina pallidn Britton & Rose
Erythrinn corallodendrum L.
Erythrir~aeggrrsii Krukoff
Erythrina buchii Urban
Erythrir~aleptopoda Urban & Ek
man
Sect. Fidelenses Neill, sect. nov. ined.
68. Erythrinn rlenae Howard & Briggs
Sect. Cuber~sesKrukoff
69. Erythrir~acubrnsis C. Wright
Sect. Olivianae Krukoff & Barneby
70. Erythrina oliviae Krukoff
Sect. Caffrne Barneby & Krukoff
7 1. Erythrir~acaffra Thunb.
72. Erythrir~alysistrmor~Hutchinson
Sect. Humennne Barneby & Krukoff
73. Erythrinn humeana Sprengel
a
Harvey
74. E r y t h r ~ r ~zeyhrri
Sect. Acnnthocarpae Barneby & Krukoff
75. Erythrina acar~thocarpnE. Meyer
APPENDIXI.
13. Sect.
58.
14. Sect.
59.
60.
61.
62.
63.
64.
65.
66.
67.
14a.
15.
16.
17.
18.
19.
111.
Subgenus Tripterolobus Barneby & Krukoff
20. Sect. Tripterolobus Barneby & Krukoff
76. Erythrir~agrernzunyL Verdcourt
IV.
Subgenus Chirocalyz (Meisner) Harvey
21. Sect. Brucear~aeBarneby & Krukoff
77. Erythrina brucei Schwein.
22. Sect. ~Wacrocymbium(Walp.) Barneby &
Krukoff
78. Erythrinn vogelii Hooker f.
79. Erythrina senegalensis A. DC.
23. Sect. Dilobochilus Harms
80. Erythrina rzcelsa Baker
24. Sect. D~chilocraspedonHarms
81. Erythrinn mildbraedii Harms
25. Sect. Chirocalyz
82. Erythrinn pygrnaea Torre
83. Erythrir~amendrsii Torre
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
V.
Continued.
Erythrina baumii Harms
Erythrinn drcorn Harms
Erythrir~alivir~gstoninnnBaker
Erythrir~atholloninr~nHua
Erythrina addisoniae Hutchinson &
Dalziel
Erythrina droogmansinnn DeWild.
& T. Durand
Erythrina orophiln Ghesq.
Erythrinn sacleuzii Hua
Erythrina harrdii Verdc.
Erythrinn sigrnoidea Hua
Erythrir~alatissima E. Meyer
Erythrina abyssinicn Lam.
Subgenus Erythraster Barneby & Krukoff
26. Sect. Erythraster
96. Erythrina variegata L.
97. Erythrina tahitensis Nad.
is
97a. Erythrinn s a n d ~ ~ i c e n s Degener
98. Erythrir~aeuodiphylln Hassk.
99. Erythrir~avrsprrtilio Benth.
100. Erythrinn mrrrillinr~nKrukoff
101. Erythrina velutina Willd.
103. Erythrir~agrisrbnchii Urban
104. Erythrina burtii Baker f.
105. Erythrina burnnn R. Chiovenda
106. Erythrina prrrirri R. Viguier
107. Erythrina schlieber~iiHarms ex
Mildbr .
108. Erythrina melanncantha Taubert
ex Harms
Species reduced to synonymy since Krukoff & Barneby
(1974):
Erythrir~acaribaea Krukoff & Barneby
= E. berteroar~aUrban x E. folkrrsii Krukoff &
Mold.
Erythrina coralloides A. DC. = E. amrricnna Miller
Erythrina ir~sulnrisF. M. Bailey = E. vespertilio
Benth.
The first two reductions to synonymy are proposed for
the first time in this paper. The third reduction follows
Krukoff's treatment in his post-1974 publications on Erythrina.
Annals of the
Missouri Botanical Garden
APPENDIX11. Erythrina hybridization trials. This appendix surnrnarizes the results of the interspecijc hybridizatior~trinls for each species combination. For m a n y o f t h e species combinntior~s,more than or2e nccessior~
was employed as the male n r ~ d l o female
r
parents. The identity a f t h e ir~dividualparents is presented in Tables
11-1.3 o n l y f o r the successful trinls resulting i n viable hybrid plants.
The hybridization trials are grouped into jive categories:
I.
Nnrrou hybridizntior~swithin sect. Erythrina.
11. Aiarrow hybridizations, excluding sect. Erythrina.
111. Intersectioraal hybridizations: fernale parent ira sect. Erythrina.
IV. Intersectior~alhybridizatior~s:male parent i n sect. Erythrina.
V. Intersectional hybridizations: excluding sect. Erythrina.
In categories III-V, the number in parentheses after the species nnrne refers to the section to ~ ~ h i cthe
h species
belongs (see Table 1 ) . A I Lasterisk indicates a uide (intersubgeneric) hybridization.
Hybrid # = a number assigned to each hybrid combination, made u p jiom the numbers assigned to the
parental species as listed in Krukoff & Bnrneby (1974)
rs
in the hybrid cornbir~ation
Pol = number o f J l o ~ ~ e hand-pollinated
Frt = number ofpollinations producing mature fruits
Sds = total nurnber of normal-sized seeds produced i r ~the hybrid cornbinatior~
Ger = number of'seeds that germinated
Liv = survivir~gprogeny; nurnber of live F, plants in the hybrid cornbir~ation
Female Parent
Male Parent
I. Narrow hybridizatioins within sect. Erythrinn
americann
berteroann
americann
herbncea
atitlnnensis
berteroar~a
ntctlanensis
guaternnlensis
berteronna
chiapasana
berteroana
folkersii
berteroana
guatemalensis
berteroann
rubrinervia
berteroann
snlviiJlora
stnndleynna
berteroann
berteroar~a
tajurnulcensis
chinpnsana
berteroar~a
chiapasann
fblkersii
guatemalensis
chinpasana
chiapasar~a
macrophylln
chiapasnna
tnjurnulcer~sis
costaricensis
berteronr~a
folkersii
berteronnn
folkersii
guatemalensis
goldmanii
chiapasar~a
berteroann
guatemalensis
guatemalensis
chinpnsnna
jolkersii
guatemalensis
herbncea
guatemalensis
rnacrophylla
saluiiJora
guaternalensis
stnndleyana
guatemalensis
tajurnulcensis
guatemalensis
americana
herbacea
berteroana
herbncea
chiapasar~a
herbacen
guatemalensis
herbacea
stnndleyana
herbacea
tajumulcensis
herbacea
nmericana
rnacrophylla
macrop hylln
atitlnnensis
rnacrophylla
berteroann
Hybrid #
Pol
Frt
Sds
Ger
Liv
Volume 75, Number 3
1988
A P P E N D I11.
X
Corztinued.
Female Parent
mncrophylla
macrophylln
macrophylla
rnacrophylla
mncrophylla
macrophylln
macrophylla
rubrznervzn
saluzzflora
salvzzj7ora
starzdleyana
stnrzdleynna
standleynrza
tnjumulcer~scs
tajumulcerzscs
tajumulcrnscs
tajumulcenscs
I Total
Neill
Erythrina
Male Parent
chcapasana
folherszc
gunternnlenscs
herbncen
snluzzflora
standleyana
tajumulcenszs
berteronrzn
brrteroana
guatemnlrnszs
berteroana
guatrmalenszs
herbac~a
brrteroana
guntemalrnszs
herbac~a
macrophylla
Hybrid #
Pol
Frt
Sds
Ger
Lir
42 x 36
42x31
42 x 4 3
42 x 22
42 x 56
42 x 23
42x40
54 x 5 3
56 x 5 3
56x43
23 x 53
23x43
23 x 22
40 x 5 3
40 x 4 3
40 x 22
40 x 42
6
7
17
8
8
2
9
2
19
3
19
8
10
12
8
2
5
417
1
1
2
0
0
0
1
0
0
0
0
0
0
0
1
39
3
2
6
0
0
0
1
0
0
0
0
0
0
0
4
0
0
142
0
1
4
0
0
0
0
0
0
0
0
0
0
0
4
0
0
90
0
1
4
0
0
0
0
0
0
0
0
0
0
0
4
0
0
75
0
0
11. Narrow (intrasectional) hybridizations: excluding sect. Erythrina
crista-galli
falcata
falcata
crista-galli
1. Sect. Cristar-galli
2x3
3x2
22
14
2
4
4
7
1
0
1
0
abyssirzica
abyssinica
latissirnn
saclruxii
2. Sect. Chirocnlyx
95 x 94
95x91
3
2
1
0
2
0
1
0
1
0
prrrirri
sandzuicensis
tahitrnsis
tahiterzsis
tahiterzsis
unriegata
uariegnta
11. Total
unrirgata
uariegatn
sandujicensis
uariegatn
uelutina
prrrirri
uespertilio
3. Sect. Erythraster
106x96
97a x 96
97 x 97a
97 x 96
97 x 102
9 6 106
~
96 x 99
10
8
7
26
19
6
6
123
4
12
0
0
0
0
0
0
25
10
0
0
0
0
0
0
9
0
0
0
0
0
0
12
11
111. Intersectional hybridizations: female parent in sect. Erythrirza
berteroana*
chzapnsnrza*
folkrrsii*
guatemalenszs*
herbncea*
macrophylla*
berteroarza*
guntemalerzszs*
herbacea*
rnacrophylla*
herbacen*
guatemalenszs
guatrmnlenscs
guatemnlrnsis
herbacea
macrophylla
brrteroana
fusca (1)
fusca
fusca
fusca
fisca
juscn
crista-gnlli (2)
crista-gnlli
crista-gnlli
crista-gnlli
dominguezii (3)
strictn (4)
arborescens (5)
speciosn (9)
speciosn
speciosa
pallida (14)
0
0
0
0
0
0
11
Annals of the
Missouri Botanical Garden
APPENDIX11.
Cor~tir~ued.
Female Parent
guatemalensis
guatrmalrnsis
guaternalensis
guatemnlrnsis
hrrbacea
guatrmalrnsis
herbacea
guatrmalensis*
guatrmalrnsis*
guatemnlrnsis*
macrophylla*
macrophylln*
guatrmalrnsis*
guatemalensis*
guatrmnlrnscs*
guntemalensis*
herbacea*
herbacrn*
macrophylla*
rnncrophylln*
rnncrophylln*
111. Total
Male Parent
Hybrid #
arnazonica (14)
corallodendrurn (14)
caffra ( 1 7 )
lysistemon ( 1 7)
caffra
hurnrana ( 18 )
humrnna
ser~egalensis( 2 2 )
nbyssinica ( 2 5 )
latissimn ( 2 5 )
abyssir~ica
lntissima
perrieri ( 2 6 )
s a n d ~ ~ i c r n s i(26)
s
uariegata (26)
vespertilio (26)
perrieri
uariegata
vnrirgata
vrsprrtilio
IV. Intersectional hybridizations: male parent in sect. Erythrina
,fusca (1)*
fisca*
,fusca*
crista-galli (2)*
stricta ( 4 )
arborescer~s( 5 )
speciosn ( 9 )
corallodrndrurn ( 14 )
pnllidn ( 1 4 )
pallida ( 1 4 )
humenna ( 1 8 )
abyssir~ica(25)*
perrirri (26)*
perrieri (26)*
uariegata (26)*
uariegata (26)*
IV. Total
brrteroar~n
folkersii
guntemalrnsis
guntemalrnsis
guatemalrnsis
guntemalensis
brrteroana
berteroana
brrteroana
fuscn
berteroana
guatemalensis
bertrroana
guatrmalrnsis
guntemalensis
hrrbacen
V. Intersectional hybridizations: excluding sect. Erythrina
fusca ( 1 )
fisca*
fuscn*
crista-galli ( 2 )
crista-galli
crista-galli*
crista-galli*
crista-galli*
crista-galli*
crista-galli*
cristn-gnlli*
crista-galli*
dorninguezii ( 3 )
cristn-gnlli ( 2 )
lysistemor~( 1 7 )
uariegata ( 2 6 )
fusca ( 1)
dorninguezii ( 3 )
nrborescrns ( 5 )
speciosa ( 9 )
nmazonica ( 1 4 )
nbyssir~ica( 2 5 )
prrrirri ( 2 6 )
s a n d ~ ~ i c e n s(i2s 6 )
vnriegatn ( 2 6 )
cristn-gnlli ( 2 )
Pol
Frt
Sds
Ger
Liv
Volume 75, Number 3
1988
APPENDIX
11.
Neill
Ery thrina
Continued.
Female Parent
arborescens (5)*
arborrscens
arborescer~s*
sprciosa (9)*
speciosa*
speciosn
caffra (17)*
lysisternor~(17)*
lysistemon
lysisternon*
lysistemon*
senrgalrnsis (22)*
abyssinicn (25)*
abyssinica*
nbyssinicn*
nbyssinicn*
abyssinicn*
lntissimn (25)*
latissima*
perrieri (26)*
uariegata (26)*
unriegata*
uariegata*
unriegata*
V . Total
Male Parent
crista-gnlli ( 2 )
hurneana ( 18 )
sandwicrnsis ( 2 6 )
fusca ( 1 )
cristn-gnlli ( 2 )
lysistemor~( 1 7 )
fuscn ( 1 )
fusca ( 1 )
speciosa ( 9 )
abyssinica ( 2 5 )
latissima ( 2 5 )
fusca ( 1 )
juscn ( 1 )
cristn-galli ( 2 )
humeana ( 1 8 )
sar~dwicensis( 2 6 )
unrirgata ( 2 6 )
lysistemon ( 1 7 )
humeann ( 18 )
fusca ( 1 )
jusca ( 1 )
crista-galli ( 2 )
speciosa ( 9 )
srnegalrnsis ( 2 2 )
Hybrid #
Pol
Frt
Sds
Ger
Liv
Annals of the
Missouri Botanical Garden
APPENDIX111. Sources of cultivated Erythrina used as parer~tnls in successjul intrrsprc$c hybridizatior~s.
L ild populations; (:IiW) : Accession obtair~edfrom cultivated source, or
( W ) : Accession obtained from ~ I L O I L I w
o t h e r ~ ~ i snot
e from a known wild population.
Vouchers jfrorn plar~tscultivatrd in Hawaiian gardens) are deposited at rW0. Location ofvoucher jiom original
uild collection of'seed is indicated here f k n o w n .
Erythrina abyssinica Lam. PT 770034001
Kenya: Nairobi. E. Taylor 17. (NW)
Erythrina abyssinica Lam. PT 731006002
Kenya: Nairobi, cultivated tree in yard of Cunningham van Someren. (NW)
Erythrina americar~aMiller W.4 7 5 ~ 1 1 7 1
Mexico: Mexico City, cultivated tree. L. S. Ayres s.11. (Waimea received as cutting from Los Angeles State &
County Arboretum # 565874) (NW)
Erythrina berteroana Urban PT 730311001
Guatemala: Suchitepequez. Nahualate, Finca El Salvador. B . A. Krukoff 1973-13 (NY). (W)
Erythrina berteroar~nUrban PT 700044001, -002
Panama: Canal Zone. Tree cultivated at Summit Gardens. IF', S. Stewart s.n. (NW)
Erythrina berteroar~nUrban WA 7 4 ~ 8 6 4
Guatemala: Suchitepequez. Municipio Chicacao. B . A. Krukoff 1968-508 (NY). ( W )
Erythrina brrteroana Urban WA 7 8 ~ 5 6 4
Panama: Canal Zone. Between Madden Dam and Chilibre. J . Folsor~3661 (MO). (W)
Erythrina caffrn Thunb. WA 7 4 ~ 1 4 5 6
South Africa: Cape Province, Grahamstown, elev. 2,400 ft. R o y Bayliss s.11. (Waimea received as cutting from
Foster Garden # 69.265) (W)
Erythrina chiapasar~nKrukoff PT 7 2 100500 1
Guatemala: Huehuetenango, near La Estancia. B. A. Krukoff 1969-68 (NY). (W)
Erythrinn chinpasana Krukoff PT 730710001
Guatemala: Huehuetenango, near La Estancia. B. A. Krukoff 1973-16 (NY). (W)
Erythrina crista-galli L. PT 740283001
Paraguay: near Asuncibn. Conrad & Die 2191. (W)
Erythrina crista-galli L. WA 74p840
South Africa. Cultivated tree; seed received from Wm. J. Tijmens, Univ. of Stellenbosch. (Waimea received as
live plant from PT 7 2 ~ 3 5 2 (NW)
)
Erythrina julcnta Benth. PT 750086001
Argentina. Thays Botanical Garden, cultivated tree. E. Pingitorr s.n. (NW)
Erythrinn folkersii Krukoff & Mold. PT 700010001
Guatemala: Izabal, at junction of road to Puerto Barrios and Mathias Calves. B. A. Krukoff 1969-109 (NY).
(W)
Erythrina ,fusca Lour. P T 740230005, WA 74599
Guatemala: Escuintla. B. A. Krukoff 1972-12 (NY). ( W )
Erythrina guatemalensis PT 700018001, WA 7 4 ~ 1 4 5 3
Guatemala: Alta Verapaz, along Cobin-Salama road, near Santa Cruz, elev. 1,280 m. B . A. Krukoff 1969-195
(NY). (W) Note: The tree at Waimea Arboretum WA 7 4 ~ 1 4 5 3was grown from a cutting taken from PT
70001 8 0 0 1, so the two accessions are genetically identical.
Erythrir~nguntemalensis Krukoff P T 720999002
Guatemala: Huehuetenango, near Barillas. B. A. Krukoff'1969-200 (NY). (W)
Erythrinn guatrmaler~sisKrukoff P T 750419001
Guatemala: Huehuetenango, near Barillas. B . A. Krukoff s.n. (W)
Erythrina herbacea L. P T 7 5 ~ 1 1 0 3
California: Los Angeles State & County Arboretum # 5 4 ~ 1 2 0 1 ,cultivated. (NW)
Erythrina herbacea L. WA 7 6 ~ 1 8 7
Florida: Miami, Fairchild Garden, cultivated. (NW)
E r ~ t h r i n nhumennn S ~ r e n g e l WA 7 4 ~ 1 3 8 2
South Africa: locality unknown. D. Ilillingtor~s.n. (NW)
Erythrina latissirna E . Meyer PT 750281004
South Africa: Natal. Cultivated tree at Pine Town Gardens. l a n It7hitton 750401. (NW)
Erythrinn lysisternon Hutchinson PT 750280002, -003
South Africa: Natal, Durban. Inr~IVhitten s.n. (1975) (W)
Erythrina macrophylla A. DC. PT 750420002, WA 7 5 ~ 1 1 3 6
Guatemala: Sololi, near Godinez. Elev. 6 , 1 4 5 ft. B . A. Krukoff 1975-4 (NY). ( W )
Volume 75, Number 3
1988
APPENDIX
111.
Neill
Erythrina
Continued.
Erythrina perrieri R. Viguier WA 7 4 ~ 8 5 7
Madagascar: Maintirano, near Bekopaka. Fred llileyrr s.n. ( W )
Erythrinn salviiflorn Krukoff & Barneby PT 721000002
Guatemala: Suchitepequez, Municipio Chicacao, Finca El Naranjo. Elev. 1.070 m. B. A. Krukoff 1969-58 (NY).
(W)
Erythrina senegaler~sisA. DC. WA 7 4 ~ 1 0 0
Nigeria: Coastal area. Seeds received from B. A. Krukoff; collector unknown. ( W )
Erythrina speciosa Andrews P T 730708001, P T 730742002
Brazil: S i o Paulo. Cultivated tree at S i o Paulo Botanical Garden. B. A. Krukoff 197.3-20 (NY). (NW)
Erythrinn standleyana Krukoff WA 7 6 ~ 1 0 5 6
California: Escondido. Cultivated tree. (NW) Originally collected as seed by Fred Meyer from wild tree, Yucatin,
Mexico.
Erythrina tajurnulcensis Krukoff & Barneby WA 7 4 ~ 1 4 4 8
Guatemala: San Marcos, near Aldea Feria, along road from San Marcos to San Rafael de La Costa. B. A.
Krukoff-1969-249 (NY). (Waimea received as cutting from PT 700015001) ( W )
Erythrina variegata L. W A 7 4 ~ 8 9 2
Hawaii: Honolulu, Mid-Pacific Country Club, cultivated (white-flowered form). Beatrice Krauss s.11. (NW)
Erythrir~avariegatn L. WA 7 6 ~ 9 9 6
Mariana Islands: Saipan Unai, Laulau Beach. Derral Herbst s.11. ( W )