Evolution, 48(6), 1994, pp, 1914-1932 INTRASPECIFIC MOLECULAR PHYLOGENY IN THE YELLOW WARBLER (DENDROICA PETECHIA), AND IMPLICATIONS FOR AVIAN BIOGEOGRAPHY IN THE WEST INDIES NEDRA K. KLEIN'AND WESLEY M. BROWN Museum ojZoology and Department ofBiology, University ojMichigan, Ann Arbor, Michigan 48109 E-mail: Wesley.Brownenon.cc.umich.edu Abstract.-A phylogenetic analysis of mitochondrial DNA (mtDNA) restriction sites was used to examine the evolutionary history of populations of yellow warbler (Dendroica petechia) sampled from North America, Central America, South America, and the West Indies. Thirty-seven hap­ lotypes were identified, and only one was found in more than one of these regions. Estimated sequence divergence among haplotypes ranged from 0.14 to 3.17%, and mtDNAs from North American migratory populations clearly were differentiated from those ofmost tropical sedentary populations. Parsimony analysis ofhaplotypes suggested multiple colonizations ofthe West Indies archipelago and of individual Caribbean islands. The inference of multiple colonizations has important implications for studies ofavian ecology and evolution in this region. Key words.-Bird, Caribbean region, Dendroica petechia, island biogeography, mtDNA, phylogeny, West Indies. Received September 28, 1992. Accepted March 18, 1994. Studies of island biogeography have contrib­ er taxa, and most likely have arrived at their uted greatly to our understanding of ecological present distributions via dispersal from sur­ andevolutionary processes (MacArthurand Wil­ rounding continental areas and from other is­ son 1963, 1967; Lack 1976; Berry 1983; Wil­ lands (Bond 1963, 1978; Seutin et al. 1993). liams 1989). Such inferences about process can Nevertheless, testable predictions of sources become even more powerful, however, when they of colonists can be made based on distances to are made within an historical framework pro­ the nearest land masses. In a series ofpapers on vided by estimates of phylogenetic relationship WestIndian birds, Bond (1948, 1961, 1963, 1978, among island populations and species (Rosen 1979, 1993) documented distributions and dis­ 1975; Eldredge and Cracraft 1980; Wiley 1981; cussed taxonomic relationships of West Indian Guyer and Savage 1986; Kluge 1988; Miller and birds. Basing his hypotheses on current faunal Miller 1989). distributions, phenotypic similarities, and in­ Caribbean islands have acquired their fauna ferred affinities to mainland groups, he proposed eitherfrom overwaterdispersal (Darlington 1938; two main sources of the Caribbean avifauna: Bond 1963, 1978; Pregill198l; Koopman 1989), northern South America for many ofthe Lesser or as relicts through fragmentation ofa formerly Antillean species, and Central and North Amer­ widespread distribution (Rosen 1975; Kluge ica for Greater Antillean and some Lesser An­ 1988). Vicariance explanations of distribution tillean species. He proposed multiple coloniza­ and evolutionary diversification have been pro­ tions from these sources for island populations posed for several faunal groups on Caribbean of widespread species. islands (Rosen 1975; Kluge 1988; Miller and Whether single or multiple colonizations have Miller 1989). However, many avian species are occurred in the West Indies is an important dis­ too recent in origin (Brodkorb 1971; Feduccia tinction for inferences about evolution within and Martin 1976; Feduccia 1977; Olson 1985) this region. The implication ofa widespread spe­ for the vicariance explanations proposed for oth- cies having colonized the archipelago multiple times from different sources is that island pop­ ulations would not constitute a monophyletic group and would thus not be each others' closest I Present address: Department ofOrnithology, Amer­ relatives. Monophyly ofisland populations is an ican Museum of Natural History, Central Park West at 79th, New York, NY 10024-5192, E-mail: implicit assumption of many evolutionary and [email protected]. ecological studies of species on archipelagoes 1914 © 1994 The Society for the Study ofEvolution. All rights reserved. YELLOW WARBLER MTDNA PHYLOGENY 1915 (MacArthur and Wilson 1967; Ricklefs and Cox We present here an estimate of the phyloge­ 1972, 1978). netic relationships among populations ofyellow Another assumption of many island biogeo­ warbler (Dendroica petechia), a widespread, pol­ graphic studies ofcolonizing species is that pop­ ytypic avian species. The phylogenetic recon­ ulations on adjacent islands are each others' clos­ struction is based on a restriction endonuclease est relatives; that is, their distributions are the analysis of mitochondrial DNA (mtDNA). We result ofa stepping-stone process ofcolonization use the mtDNA phylogeny to examine the evo­ (Darlington 1938; MacArthur and Wilson 1967; lutionary history of resident populations in the Ricklefs and Cox 1972, 1978). An alternative is West Indies and tropical continental areas. We a colonization pattern that is random with re­ test the hypothesis that the West Indies has been spect to geography. Under the stepping-stone colonized multiple times by different source pop­ model, islands arranged linearly (e.g., the Lesser ulations ofyellow warbler, as suggested by Bond Antilles) are colonized sequentially, beginning (1963, 1978) for widespread Caribbean birds. If with that closest to the mainland source (Wil­ multiple colonizations have occurred, the Carib­ liams 1989). Individual island populations would bean samples should not form a monophyletic be derived from those on other islands or the group relative to othersamples. The analysis also adjacent mainland, and heritable differences is used to infer colonization sources, as has been among them should thus be due to selection or done for other groups ofisland organisms (Ash­ random differentiation processes rather than to ley and Wills 1987; Phillips et al. 1989), and to different phylogenetic histories. If, however, a test the critical prediction of the stepping-stone widespread species has colonized the archipelago colonization model that populations on adjacent multipletimesfrom different source populations, islands are closely related to each other. there may be areas where populations on adja­ cent islands are descendants ofdifferent ancestral METHODS colonizers. Differences among such populations Specimens were collected by shotgun or mist may thus be due to their different genealogies. net from 1986-1990. A total of 194 yellow war­ A phylogenetic inference of relationships blers was used in this study. General localities among island and mainland populations of a and sample sizes were: Michigan (17), North widespread taxon is useful both because it allows Carolina (9), Washington (6), Baja California (1), testing the models of single and multiple colo­ Florida Keys (1), Jamaica (9), Dominican Re­ nizations of archipelagoes (different patterns of public (10), Puerto Rico (17), U.S. Virgin Islands relationship are predicted under each) and be­ (24), Montserrat (8), Guadeloupe (5), Dominica cause potential interpretive problems associated (6), Martinique (4), St. Lucia (4), Venezuela (40), with missing taxa (islands that were never col­ Panama (13), Costa Rica (20). Populations from onized or whose populations have been extir­ Michigan, North Carolina, and Washington are pated) can be eliminated. Under the model of a wholly migratory. The rest are wholly sedentary single colonization ofan archipelago, monophyly tropical residents. of island populations is predicted. Polyphyly is Although a few estimates of historical rela­ predicted under the multiple-colonization mod­ tionships among paruline warblers exist (Mengel el, provided founders came from different source 1964; Barrowclough and Corbin 1978; Raikow populations. It is difficult to test models ofisland 1978; Bledsoe 1988; Sibley and Ahlquist 1990), colonization pattern within archipelagoes, how­ none of the studies include Dendroica petechia. ever, because (1) any tree topology is consistent Preliminary allozyme evidence (Avise et al. 1980) with a random colonization model (i.e., the ran­ suggests that the yellow warbler may fall outside dom model is not falsifiable), and (2) although a ofa group consisting ofother Dendroica. In the pectinate tree is predicted under the stepping­ absence ofan explicit hypothesis ofrelationship stone model, if colonization has occurred rap­ among D. petechia and other warblers, choice of idly, or if there has been any postcolonization outgroup taxa was done with the goal ofencom­ movement among islands, the comblike pattern passinga wide range ofinferred divergence levels ofhistorical relationship will not be recoverable. relative to the ingroup (Maddison et al. 1984). Nevertheless, one can test the prediction of the Outgroup species (with sample sizes and locali­ stepping-stone model that populations on adja­ ties) were: D. pensylvanica (2, Michigan), D. dis­ cent islands should be each others' closest rela­ color (2, Florida), Setophaga ruticilla (1, Mich­ tives. igan), Parula americana (1, North Carolina), 1916 N. K. KLEIN AND W. M. BROWN Geothlypis triehas (2, Michigan), and Basileute­ passing the geographic range ofsamples) digested rus eulicivorus (I, Venezuela). All outgroups ex­ with that enzyme were run together on one set ceptB. eulicivorus are migratory NorthAmerican of agarose and polyacrylamide gels to facilitate species. Basileuterus is a wholly sedentary, Neo­ comparison of fragment sizes among different tropical genus that is the probable sister taxon haplotypes. Other gel sets