Experimental Studies on Species Relationships in Erythrina (Leguminosae: Papilionoideae)

David A. Neill

Erythrina L. comprises about 112 species distributed throughout the tropics and subtropics. Most species are trees or shrubs, and most are diploids with n = 21. All are adapted to bird pollinntion, some by passerine birdsvand others by hummingbirds. Erythrina is subdivided into five subgenera and 27 sections. Research concentrated on sect. Erythrina, with 36 sprcies centered in Mexico and Central America; selected speciess in other sections were also studied. Experimental interspcific hybridizations and self-conapatibility trials were Tarre conductedd using cultivated trees at several botanical gardens in Hawaii. Compartive morphological analyses were made of the hybrids an their parents. Studies ofpopulation structure and natural hybridization were carried out in natural populations of hummingbird-pollinated sect. Erythrina. Erythrina species are self-compatible, but some inbreeding depression is associated with selfng. Within sect. Erythrina, interspecifc hybrids are obtained just as readily as are progeny from within-species outcrosses. The hybrids are vigorous, fertile, and by several measures exhibit interspecific heterosis. At greater taxonomic distances between the parental species (between sections and subgenera, crossability, viability, and fertility of the hybrid progeny are generally lower than in intrasectional hybridizations. Some hybrids were obtained between species of different subgenera indigenous to different continents. There are probably no absolute internal barrierrs to hybridization among all the diploid species of Erythrina. The genus may be characterized taxogenetically as a homogamic complex. Interspecifc hybrids are intermediate between their parental species in morphological traits, including macroscopic fentures of the inflorescence and flowers and microscopic features of the leaf epidermos. The inheritance of the male parent in the progeny allows for confirmation of hybridity. Species of sect. Erythrina are generally allopatric, but field studies of natural populations in southern Mexico revealed several localities where two species do occur sympatrically and where natural hybrids are found. Traplining hummingbirds, the pollinators of sect. Erythrina, are implicated as the agents of interspecific hybridization among sympatric species. The results of experimental hybridization, together with studies of comparative morphology and distribution patterns, suggest that some species of Erythrina are stabilized hybrid derivatives.

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 )

RELATED PAPERS

RELATED TOPICS

Log In

Experimental Studies on Species Relationships in Erythrina (Leguminosae: Papilionoideae

Experimental Studies on Species Relationships in Erythrina (Leguminosae: Papilionoideae

Experimental Studies on Species Relationships in Erythrina (Leguminosae: Papilionoideae

Experimental Studies on Species Relationships in Erythrina (Leguminosae: Papilionoideae

Experimental Studies on Species Relationships in Erythrina (Leguminosae: Papilionoideae

Related Papers

RELATED PAPERS

RELATED TOPICS