We explored the evolution of geographic distributions in archipelagos by comparing mitochondrial DNA (mtDNA) sequences and morphometric characters within and among conspecific populations of Adelaide's Warbler (Dendroica adelaidae), Plumbeous Warbler (D. plumbea), and Olive-capped Warbler (D. pityophila). Phylogenetic reconstructions were based upon 1,455 nucleotides of protein-coding mtDNA sequence from 53 individual warblers; morphological analyses employed three external measurements from a larger number of museum specimens. Of the three taxa studied, Adelaide's Warbler occupied the broadest and most fragmented geographical distribution and exhibited the greatest inter-population differentiation in both mtDNA and morphology. Phylogenetic analyses demonstrated that the three Adelaide's Warbler populations are each reciprocally monophyletic with the Puerto Rican lineage basal to sister clades on Barbuda and St. Lucia. Genetic distances among these populations were comparable with those between some continental species. In contrast to the mtDNA pattern, the Puerto Rican and Barbudan Adelaide's Warbler populations were most similar in morphometry. We observed considerably less mtDNA and morphometric differentiation among populations of the two species with more restricted and less fragmented distributions, the Plumbeous Warbler of Dominica and Guadeloupe and the Olive-capped Warbler of the Bahamas and Cuba. High levels of molecular and morphological differentiation among the geographically disjunct Adelaide's Warbler populations and low differentiation in the two species with less fragmented ranges suggest that range disjunctions indicate the long-term evolutionary independence of geographically isolated island populations. Received 21 August 1997, accepted 18 February 1998.

Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Panama; 2 Department of Biology, University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA; 3 Department of Geography, McGill University, Montreal, Quebec, H3A 2K6, Canada; and 4 Department of Biology, University of Missouri St. Louis, 8001 Natural Bridge Road, St. Louis, Missouri 63121, USA INTERPRETATIONS OF BIOGEOGRAPH1C PAT- TERNS in avian taxa traditionally have relied on estimates of relationship based upon morpho- logical characters such as size and plumage col- oration. Morphological divergence among lin- eages is summarized in the hierarchical taxo- nomic classification of a group, and such clas- sifications have often been accepted as indices of phylogenetic relationships (e.g. Ricklefs and Cox 1972). Molecular phylogenies now provide an alternative source of information on biogeo- graphic patterns and the evolutionary process- es that have molded them (Avise 1994). Species that occupy archipelagos are particularly ame- nable to biogeographic and phylogenetic inves- tigations. The general distribution of a species s Address correspondence to: Smithsonian Tropi- cal Research Institute, Unit 0948, AP0 AA 34002, USA. E-mail: ilovette@sas.upenn.edu and the boundaries of particular populations are readily defined by a taxon's presence or ab- sence on a given island. Water barriers or un- occupied islands situated between occupied is- lands might retard gene flow between popu- lations and thereby enhance their genetic subdivision. Finally, archipelagos offer the op- portunity to make comparisons among taxa with different evolutionary histories superim- posed upon a relatively simple common geog- raphy. To explore evolutionary and geographic pat- terns in one archipelago, we investigated intra- specific variation in mitochondrial DNA (mtDNA) sequences and external morphomet- ric characters in three West Indian Dendroica species, Adelaide's Warbler (D. adelaidae), Plum- beous Warbler (D. plumbea), and Olive-capped Warbler (D. pityophila). Adelaide's Warbler has both the broadest and most fragmented distri- bution, with populations on Puerto Rico, Bar- 0 km 500 Puerto Rico. I" '- Barbuda Adelaide's./x. Warbler Plumbeous.. .- Guadeloupe I Warbler -- Dominica $ St. Lucia FIG. 1. Distribution of Adelaide's Warbler, Plumbeous Warbler, and Olive-capped Warbler in the eastern Caribbean. Note large disjunctions between islands occupied by Adelaide's Warbler relative to the continuous (Plumbeous Warbler) or nearly continuous (Olive-capped Warbler) distributions of the other two species. buda, and St. Lucia (Fig. 1). The Plumbeous Warbler occupies a restricted and continuous distribution on the adjacent islands of Guade- loupe and Dominica in the Lesser Antilles (Fig. 1). The Olive-capped Warbler, found on the two northernmost islands in the Bahamas and on Cuba, is intermediate in both geographic breadth and range fragmentation (Fig. 1). Here, we describe the geographic structure of variation in morphometry and mtDNA with- in each species and determine whether mor- phometric differentiation is proportional to ge- netic differentiation. We then test whether the phylogeographic data are consistent with the hypothesis that gaps in the distributions of Lesser Antillean bird species result from the extinction of geographically intermediate pop- ulations (Ricklefs and Cox 1972). Based on the assumption that both population extinctions and genetic divergence accumulate over evo- lutionary time, we predicted that the ancestral nodes of intraspecific phylogenetic trees should be oldest in the highly disjunct Ade- laide's Warbler and youngest in the continu- ously distributed Plumbeous Warbler. STUDY SPECIES AND METHODS Adelaideõ Warbler.--Adelaide's Warbler occurs on Puerto Rico (and nearby Vieques Island), Barbuda, and St. Lucia (Fig. 1). Each island population has been accorded specific status in the past (e.g. Ridg- way 1902), but the three populations currently are considered to be subspecies (AOU 1983). Birds from Puerto Rico (D. a. adelaidae) and Barbuda (D. a. subita) are similar in plumage and size (Riley 1904, Hell- mayr 1935), whereas those from St. Lucia (D. a. deli- cata) are about 10% larger than individuals from the two northern populations and have a diagnostically yellow supercilium and extensively yellow under- parts (Ridgway 1882, 1902; Riley 1904, Curson et al. 1994). Bond (1930, 1956, 1965) also noted differences in habitat selection, foraging height, and song types among the three islands. Adelaide's Warbler is most abundant in lowland dry forest and scrub except on St. Lucia, where it ranges into humid montane forest (Bond 1956, Cruz and Delaney 1984, Raffaele 1989). Because there are no records of the species elsewhere in the Caribbean or on the mainland (Bond 1956), Adelaide's Warbler apparently is nonmigratory. Plumbeous Warbler.--The Plumbeous Warbler is en- demic to the Lesser Antilles, where it is known only from the adjacent islands of Dominica and Guade- loupe, and from Guadeloupe's satellite islands, Ma- rie-Galante and Terre-de-Haut (Fig. 1). The species usually has been considered monomorphic (Ridg- way 1902, Bond 1956, Peters 1968), but Hellmayr (1935) and Kepler and Parkes (1972) suggested that the Dominican and Guadeloupean populations each be accorded subspecific status (D. p. plumbea and D. p. guadeloupensis, respectively)based on slight differ- ences in the coloration of the underparts. The Plum- beous Warbler is a common permanent resident of both dry and wet forests throughout its limited range. Olive-capped Warbler.--The Olive-capped Warbler is restricted to the pine forests of the northernmost islands in the Bahamas (Abaco and Grand Bahama Islands) and to similar habitat in eastern and possi- bly western Cuba (Fig. 1; Bond 1956, 1958). Ridgway (1902) noted slight plumage differences between the Bahamian and Cuban forms, which he separated into subspecies (D. p. bahamensis and D. p. pityophila, respectively), but later workers have not considered these differences substantial enough to warrant even subspecific distinction (Hellmayr 1935, Bond 1956, Peters 1968). The lack of extralimital records of Ol- ive-capped Warblers (e.g. none from Florida, which lies 100 km west of Grand Bahama and 140 km north of Cuba) indicates that this species is sedentary. mtDNA sample collection.--We obtained samples for genetic analyses from three Adelaide's Warblers on Puerto Rico (20 to 30 October 1993), five on Barbuda (6 to 9 May 1993), and eight on St. Lucia (18 to 26 July 1991); from 14 Plumbeous Warblers on Dominica (27 July to 3 August 1991) and 13 on Guadeloupe (21 to 27 April 1993); from two Olive-capped Warblers on Abaco, Bahamas (19 October 1993); and from one Ar- row-headed Warbler (D. pharetra; used for outgroup rooting of Plumbeous Warbler haplotypes) on Ja- maica (11 December 1995). Muscle biopsies and/or blood samples were collected nondestructively (Ba- ker 1981) from mist-netted individuals. Muscle tis- sue was preserved in a DMSO/EDTA/NaC1 solution, and blood was preserved in Queen's lysis buffer (Seutin et al. 1991, 1993). DNA extracts of these sam- ples are available upon request from the authors. We obtained frozen muscle samples from four addition- al Puerto Rican Adelaide's Warblers and two North American Yellow-throated Warblers (D. dominica; used for outgroup rooting of Adelaide's Warbler hap- lotypes) from the Museum of Natural Science at Lou- isiana State University, and one muscle sample from a Cuban Olive-capped Warbler (collected near Pinas Del Rio in western Cuba) from the Academy of Nat- ural Sciences of Philadelphia. All specimens were collected and transported under appropriate per- mits. mtDNA laboratory procedures.--Total cellular DNA was extracted from each sample following the pro- tocols of Seutin et al. (1991, 1993). We used the poly- merase chain reaction (PCR) to amplify two regions of the mitochondrial genome from all individuals. The primer pair CO2GQL and CO3HMH (Seutin and Bermingham unpubl. primer sequences; all primer sequences available upon request from E. Berming- ham) was used to amplify a 1,074 base-pair region that spanned the full tRNA Ly', ATPase 8, and ATPase 6 genes. A 681 base-pair portion of the cytochrome oxidase I (COI) gene was similarly amplified using primers COIa and COIf (Kessing et al. 1989). PCRs for both ATPase and COI amplifications were con- ducted for 25 cycles at an annealing temperature of 54øC. We cleaned all amplification products electrophor- etically on low-melting-point agarose gels and pu- rified them using the Geneclean procedure. We then conducted Dyedeoxy terminator cycle sequencing reactions (Applied Biosystems Division of Perkin El- mer, Inc.) following the manufacturer's protocol. We sequenced the light strand of the ATPase region with the primers CO2GQL, A6PWL, and A6TPL (Seutin and Bermingham unpubl. primer sequences). The COI region was seqtenced using the two amplifica- tion primers. The cycle sequencing reactions were then electrophoresed in Applied Biosystems model 373A or model 377 automated DNA sequencers. On average, 42% of the ATPase coding region from each individual was confirmed by overlapping sequences generated from two light strand primers. In the COI region, overlap between the light-strand sequences from primer COIf and the heavy-strand sequences from primer COIa ranged between 80 and 100%, and we found no nucleotide differences between overlap- ping complementary sequences. Genetic data analysis.--Analyses were conducted upon the concatenated ATPase and COl sequences because mitochondrial genes are fully linked and thus constitute a single phylogenetic marker, and be- cause the combinability test of Farris et al. (1995), as implemented in test version 4.0d56 of David L. Swof- ford's Paup* package, identified no significant dif- ferences between the trees generated using the ATP- ase 8, ATPase 6, and COI partitions of the combined data (P > 0.65 for all conspecific data sets). Each con- catenated sequence included the entire 842 nucleo- tide coding region of the overlapping ATPase 6 and ATPase 8 genes and 613 nucleotides of the COI gene corresponding to nucleotides 7342 to 7954 in the chicken mitochondrial genome (GertBank accession number X52392; Desjardins and Morais 1990). We re- fer to each unique concatenated sequence as a mt- DNA "haplotype." Sequences were imported into the program Se- quencer 3.2 (B. Kessing pers. comm.) to generate de- scriptive statistics about nucleotide variation. We then used PAUP* 4.0d56 to estimate genetic distanc- es between individuals using the LogDeterminant (LogDet; Steel 1994) distance metric. Swofford et al. (1996) discuss the advantages of this metric; in the present study, all intraspecific LogDet distances were almost identical to the corresponding distances based on uncorrected percent nucleotide difference or upon the Kimura two-parameter substitution model (Kimura 1980). In order to compare intrapop- ulation mitochondrial diversity across populations, we calculated the haplotype-diversity index h (Nei 1987: equation 8.5) and nucleotide-diversity index (Nei 1987: equation 10.5) for all populations where n > 4. The h metric estimates the probability that two individuals sampled at random from a population will have different haplotypes based on the observed haplotype frequencies, whereas incorporates the distribution of pairwise sequence differences into an estimate of whether randomly selected individuals from a population will differ at a nucleotide site. We used three phylogenetic methods--maximum- likelihood (ML), neighbor-joining (NJ), and maxi- mum-parsimony (MP)--to reconstruct the relation- ships between the three Adelaide's Warbler popula- tions. ML analyses were conducted using the pro- gram PUZZLE 3.1 (Strimmer and von Haeseler 1997), which employs the quartet puzzling search al- gorithm of Strimmer and von Haeseler (1996) to re- construct phylogenetic trees. These ML searches were conducted for 10,000 puzzling steps using the Hasagawa-Kishino-Yano substitution model (Hasa- gawa et al. 1985), and with transition:transversion ratios, nucleotide frequencies, and gamma rate het- erogeneity parameters determined from the se- quence data using PUZZLE's "exaci' function. NJ and MP analyses were conducted using PAUP* 4.0d56. NJ analyses were based on the LogDet dis- tance matrix, and 1,000 bootstrap replicates were performed on the NJ tree. MP analyses were con- ducted using the branch-and-bound search option with all characters weighted equally, and with tran- sitions assigned a weight of 1 and transversions a weight of 15 to reflect the empirically determined bias (see Results); 1,000 bootstrap replications were performed under both weighting schemes. Alterna- tive topologies among Adelaide's Warbler popula- tions were compared using the Kishino-Hasegawa test (Kishino and Hasegawa 1989) as implemented in PAUP* 4.0d56. Because the three warbler species considered here are endemic to the West Indies, we assume that pop- ulations of each species are more closely related to each other than to any non-conspecific population; molecular systematic analyses that include addition- al Dendroica support this assumption (Lovette and Bermingham unpubl. data). In all phylogenetic anal- yses of Adelaide's Warbler, two Yellow-throated War- bler sequences were specified as the outgroup taxon; this species was chosen because plumage and mor- phology place it in the same "superspecies" complex as Adelaide's Warbler (Bond 1956, Mengel 1964, Mayr and Short 1970). The use of the other members of this complex (D. pityophila and D. graciae) or other Dendroica warblers as outgroups did not change the phylogenetic topology among Adelaide's Warbler populations. We investigated the relationships of Plumbeous Warbler haplotypes through ML, NJ, and MP anal- yses as described above, except that trees were root- ed to sequences from the Arrow-headed Warbler ow- ing to the proposed superspecies affinity between these taxa (Bond 1956, Kepler and Parkes 1972). We did not conduct phylogenetic reconstructions for the Olive-capped Warbler because it was represented by only three individuals with very similar mtDNA haplotypes. Morphometric measurements and analyses.--All mor- phological measurements were made by Lovette on museum specimens from the National Museum of Natural History, Washington, D.C.; American Mu- seum of Natural History, New York; Museum of Nat- ural Science at Louisiana State University, Baton Rouge; and the Academy of Natural Sciences of Phil- adelphia. The lengths of the wing (chord of unflat- tened wing), bill (distal tip of the maxilla to the proximal edge of the exposed culmen), and tarsus were measured (_+0.1 mm) using dial calipers. The island of origin and, when indicated, the sex of each specimen were recorded from the museum label. In some cases, Adelaide's Warbler skins with no indi- cated gender were sexed using plumage criteria (Ridgway 1902, Curson et al. 1994). Because sample sizes of female warbler specimens from some pop- ulations were small, we were unable to test whether these populations exhibit sexual dimorphism in wing, bill, or tarsus length. Therefore, analyses were restricted to male specimens. All measurements were log-transformed prior to analysis. The general structure of mensural variation among populations of each species was examined via ANOVA where island of origin was specified as the independent variable. The relative magnitude of morphometric divergence among populations of the three species was assessed by comparing: (1) the Eu- clidian distance between island centroids in the log- transformed measurement space; and (2) the Mahal- anobis distance between populations, a measure that represents the squared distance in the canonical dis- criminant space normalized by the pooled within-is- land variance along the discriminant axis. The SAS statistical analysis package (SAS Institute 1987) was used to conduct both the ANOVA (PROC GLM) and the multivariate (PROC DISCRIM and CANDISC) analyses. RESULTS We obtained the entire 842-bp coding se- quence of the overlapping ATPase 6 and ATPase 8 genes and 611 bp of COI coding sequence from 20 Adelaide's Warblers, 27 Plumbeous Warblers, 3 Olive-capped Warblers, 2 Yellow-throated Warblers, and 1 Arrow-headed Warbler (Gen- Bank accession numbers U91961, AF018094 to AF018145 inclusive, and AF018200 to AF018252 inclusive; see Appendix for sequence align- ments showing all nucleotide sites that varied among intraspecific haplotypes). We found no insertions or deletions in either gene region; hence, sequence alignments were unambiguous. Base frequencies were biased in both regions, es- pecially at third-codon positions, as is typical of the avian mitochondrial genome (e.g. Desjardins and Morais 1990, Zink and Blackwell 1996). Of the 118 nucleotide sites that varied among con- specific haplotypes, 104 involved synonymous substitutions; we found five nonsynonymous changes among ATPase 8 sequences, six among ATPase 6 sequences, and three among COI se- quences. Empirical transition:transversion ra- tios among all conspecific Adelaide's Warbler and Plumbeous Warbler haplotypes were 15:1 and 17:1, respectively; too few nucleotide differ- ences were found among Olive-capped Warbler haplotypes (see below) to allow a meaningful transition bias calculation for this species. Recent discoveries of nuclear copies of avian mitochondrial genes ("pseudogenes;" e.g. Arc- tander 1995) have highlighted the importance of assessing the homology of the mtDNA PCR products used for phylogenetic analyses. In the present study, the presence of a single ampli- fication product in our PCR reactions, the ab- sence of indels and stop codons within the ATPase and COI coding regions, the high pro- portion of silent substitutions, the similarity to Dendroica sequences obtained from highly pu- rified mtDNA samples (Lovette unpubl. data), and the complete congruency of the phyloge- netic reconstructions from the two mtDNA gene regions separated by 981 nucleotides pro- vided evidence that the sequences we obtained were mitochondrial in origin. mtDNA diversity within West Indian warbler populations.--We found 11 haplotypes among the 20 Adelaide's Warblers from Barbuda, St. Lucia, and Puerto Rico. Two haplotypes differ- ing by a single synonymous transition were present among the five Barbudan individuals; thus, the maximum pairwise genetic difference within this population was only 0.1%. We found slightly more variation within the St. Lu- cian population, where four haplotypes were distributed among the eight individuals ex- amined. Four nucleotide sites varied among these St. Lucian haplotypes, yielding a maxi- mum within-population divergence of 0.3%. The seven birds from Puerto Rico had five hap- lotypes that differed at a total of 11 nucleotide sites and had a maximum pairwise divergence of 0.6%. Although the number of samples avail- able to us was small, our survey suggests that mtDNA diversity is proportional to island area, with the Puerto Rican (9,104 km 2) popu- lation showing the highest level of diversity (h = 0.86, r = 0.0015) relative to those from St. Lucia (h = 0.75, r = 0.0005; 620 km 2) and Bar- buda (h = 0.40, r = 0.0001; 160 km2). Plumbeous Warbler populations on Domini- ca and Guadeloupe also had high levels of hap- lotype diversity but exhibited only moderate levels of nucleotide differentiation. We distin- guished nine mtDNA haplotypes (h = 0.88) in the sample of 14 birds from Dominica and 10 haplotypes (h = 0.96) in the 13 birds from Gua- deloupe. Because all haplotypes within each population were closely related, these high haplotype diversities were not paralleled by high levels of nucleotide diversity: the nine Do- minican haplotypes differed at a maximum of seven nucleotide sites (r = 0.0012; maximum divergence 0.7%) and the 10 Guadeloupean haplotypes differed at a maximum of nine nu- cleotide sites (r = 0.0018; maximum diver- gence 0.8%). We can say little about mtDNA diversity in Olive-capped Warblers based on our sample of only three individuals. The two Olive-capped Warblers from Abaco had identical haplotypes. Haplotype diversities in the Puerto Rican and St. Lucian Adelaide's Warbler populations and in the two Plumbeous Warbler populations are among the highest reported for birds, ex- ceeding diversities reported among continental taxa that must support much larger effective population sizes (e.g. Seutin et al. 1993, 1995; Bermingham et al. 1996; Zink 1996). We cau- tion, however, that direct comparisons of h be- tween studies employing different molecular techniques are problematic because h is depen- dent upon the sensitivity of the methodology used to assay haplotype diversity (Nei 1987). Nucleotide diversities are less influenced by this potential source of bias. Values of r ob- served on the larger islands (0.0012 to 0.0018) were somewhat lower than those observed in some continental populations (0.0017 to 0.0033; Ball et al. 1988, Zink 1991, Bermingham et al. 1992, Seutin et al. 1995) and in Bananaquits (Coereba fiaveola) on larger West Indian islands (0.0019 to 0.0037; Seutin et al. 1994). Genetic divergence and phylogenetic relationships among conspecific warbler populations.---Genetic distances among islands in Adelaide's Warbler were high relative to the modest variation with- in islands in this species. Pairwise divergences between Barbudan and St. Lucian haplotypes varied between 2.2 and 2.5%, and distances be- tween Puerto Rican and Lesser Antillean hap- lotypes varied between 3.7 and 4.7%. ML, NJ, and MP reconstructions of the relationships be- Adelaide's Warbler r [Baoud a lOO 7.11.1 10 1.5 + 0.5 1.8ñ0.5 100 I 1.6+0A 0.7+0.3 St. Lucia Puerto Rico r D. dominica 1% nucleotide substitution FIG. 2. Maximum-likelihood tree depicting intra- specific phylogenetic relationships in Adelaide's Warbler, based on a comparison of 1,455 nucleotides of mtDNA sequence. Numbers above branches indi- cate reliability values as calculated by the maximum- likelihood analysis program PUZZLE. Values below branches give branch lengths and their associated standard errors. Neighbor-joining and maximum- parsimony analyses identified an identical topology among Adelaide's Warbler populations, with 100% bootstrap support for all internal branches connect- ing the three haplotype groups. Scale bar shows 1% nucleotide substitution (2% sequence divergence). tween the three Adelaide's Warbler populations identified almost identical trees that invariably supported the monophylly of each island pop- ulation. All reconstructions placed the Puerto Rican population basal to a clade comprised of the Barbudan and St. Lucian populations (Fig. 2). Kishino-Hasegawa tests indicated that this topology was significantly better than alterna- tive topologies in which the St. Lucian popu- lation (tree 14 steps longer; t = 3.31, P < 0.001) or the Barbudan population (tree 13 steps lon- ger; t = 2.99, P < 0.003) was constrained to be basal. The same highly supported topology was obtained when the ATPase and COI regions were analyzed separately and under the two MP character-weighting schemes; to- pological differences among these reconstruc- tions involved only the relationships of very similar haplotypes within single island popu- lations. In the analyses of the combined ATPase and COI sequences, the internal branches con- necting the three Adelaide's Warbler popula- tions were invariably supported by reliability (ML) and bootstrap (NJ and MP) values of 100% (Fig. 2). No haplotypes were shared by the Domini- can and Guadeloupean Plumbeous Warbler populations, but interisland genetic divergence was modest: pairs of haplotypes from Domin- ica and Guadeloupe differed at 5 to 11 nucleo- tide sites (0.3 to 1.0% LogDet divergence). Al- though the overall magnitude of within- and among-population haplotype divergences was similar in this species, phylogenetic analyses suggested that the Dominican and Guadelou- pean populations are monophyletic with re- spect to one another When an Arrow-headed Warbler sequence was employed as an out- group, ML, NJ, and MP analyses all placed the basal root in the internal branch separating the Dominican and Guadeloupean haplotypes (Fig. 3). Unrooted reconstructions (not shown) similarly split the Dominican and Guadelou- pean haplotypes into monophyletic clades, and reliability scores (ML) and bootstrap analyses (NJ and MP) indicated high support for the in- ternal branch separating the two island-specific clades (ML reliability = 98%; NJ bootstrap = 86%; unweighted MP bootstrap = 90%). Divergence between the Abacan and Cuban Olive-capped Warblers was small: the Cuban haplotype differed from the Abacan haplotype by five nucleotide substitutions (0.4% LogDet difference). Without additional samples, we cannot determine whether this difference rep- resents divergence between genetically isolat- ed populations or diversity within a single panmictic population. Morphometric variation.--Table 1 summarizes numbers of individuals measured and means and variances in wing length, exposed culmen length, and tarsus length. ANOVA revealed significant morphometric differences among males from different populations of each of the three warbler species: (1) the three Adelaide's Warbler populations differed in all three men- sural characters, (2) the two Plumbeous War- bler populations differed only in tarsus length, and (3) the two Olive-capped Warbler popula- tions differed in both wing length and tarsus length (Table 2). Multivariate analyses similarly identified morphometric differences within each of the three warbler species (Table 3). In Adelaide's Plumbeous Warbler Guadeloupe 62 0.3+0.1 . D. pharetra 8.96 + 1.2 Dominica 1% nucleotide substitution FIG. 3. Maximum-likelihood tree depicting phy- logenetic relationships among Plumbeous Warbler haplotypes. Number above branch indicates reli- ability value; number below branch gives branch length and the associated standard error. Only one branch had a length >0.25% nucleotide substitution; values for branches <0.25% are not shown. Branch leading to outgroup taxon not drawn to scale. Un- rooted maximum-likelihood, neighbor-ioining, and maximum-parsimony analyses similarly separated all haplotypes into two island-specific clades. Scale bar shows 1% nucleotide substitution (2% sequence divergence). Warbler, the Puerto Rican and Barbudan pop- ulations had overlapping distributions in the multivariate measurement space and were not significantly different; the St. Lucian popula- tion, however, had the greatest degree of mor- phological distinction among the intraspecific comparisons (Table 3). The overall magnitude of population differentiation was more modest in the Plumbeous Warbler and the Olive-cap- ped Warbler, but nonetheless each island pop- ulation was significantly distinct (Table 3). Considered across species, the general pattern TABLE 1. Wing length, culmen length, and tarsus length of males in three species of West Indian warblers. Values are : _+ SD. Island n Wing Culmen Tarsus Adelaide's Warbler Puerto Rico 41 51.2 _+ 1.4 9.6 _+ 0.4 16.7 -+ 0.6 Barbuda 19 51.6 -+ 2.1 9.9 _+ 0.6 17.2 _+ 0.6 St. Lucia 27 57.2 _+ 1.1 10.0 _+ 0.4 17.9 -+ 0.4 Plumbeous Warbler Dominica 10 65.1 _+ 1.5 10.4 _+ 0.6 20.3 _+ 0.3 Guadeloupe 20 62.2 _+ 2.4 10.5 _+ 0.3 19.5 -+ 0.7 Olive-capped Warbler Bahamas 12 58.7 -+ 1.9 10.0 +_ 0.4 16.8 -+ 0.5 Cuba 30 59.4 _+ 1.4 9.5 -+ 0.7 16.0 +_ 0.5 of multivariate variation did not differ between comparisons normalized by the within-popu- lation variances and unstandardized compari- sons (Table 3). Comparison of molecular and morphometric dif- ferentation.--The pattern of interisland genetic differentiation across the three species was par- alleled in part by the corresponding pattern of morphological variation. Overall, morphologi- cal diversity was highest in Adelaide's Warbler (but see below), the species with the highest in- terisland genetic divergence. Less interisland morphological variation was present in Plum- beous Warblers and Olive-capped Warblers, species whose populations were less genetical- ly distinct. Nonetheless, we found significant differences among islands in morphology in each of the three species, suggesting either that TABLE 2. Results of ANOVA of interisland variation (islands noted in Table 1) in wing length, culmen length, and tarsus length of males in three species of West Indian warblers. Species R 2 F P Adelaide's Warbler (df = 2 and 70) Wing length 0.81 146.9 *** Culmen length 0.17 7.1 * Tarsus length 0.50 34.2 *** Plumbeous Warbler (df = 1 and 37) Wing length 0.03 1.0 ns Culmen length 0.10 4.1 ns Tarsus length 0.33 17.9 *** Olive-capped Warbler (df = 1 and 25) Wing length 0.35 13.5 ** Culmen length 0.05 1.4 ns Tarsus length 0.43 19.0 ** ns, P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.1301. TABLE 3. Comparisons of multivariate morphometric differences among populations of three species of West Indian warblers. Mahalanobis distance (D 2) is the squared distance in the canonical discriminant space normalized by the within-population variance. E 2 represents the squared Euclidean distance between population centroids in the log-transformed measurement space. Species Islands compared df D 2 F P E 2 (x 100) Adelaide's Warbler Puerto Rico/Barbuda 1, 59 0.73 2.4 0.078 0.016 Adelaide's Warbler Puerto Rico / St. Lucia 1, 67 25.49 115.9 <0.0001 0.385 Adelaide's Warbler Barbuda / St. Lucia 1, 45 18.86 53.2 <0.0001 0.268 Plumbeous Warbler Dominica/Guadeloupe 1, 29 5.66 10.9 <0.0001 0.094 Olive-capped Warbler Bahamas/Cuba 1, 41 2.55 7.0 <0.001 0.075 gene flow between the various pairs of conspe- cific populations has not been frequent enough to override morphological differentiation be- tween them, or that differences between is- lands reflect individual phenotypic responses to different ecological conditions on each is- land. When all pairwise comparisons among con- specific island populations are treated sepa- rately, four of the five possible comparisons be- tween conspecific populations show a general correlation of genetic and morphometric diver- gence (Fig. 4). The comparison between the Puerto Rican and Barbudan Adelaide's Warbler populations is anomalous in that these popu- lations are similar morphologically in mensu- ral traits but highly divergent genetically. DISCUSSION The comparison of molecular and morpho- logical differentiation within Adelaide's, Plum- 25 5 ß Euclidian ß Mahalanobis ß PR-SL I BA-SL PR-BA CU-BH ß 0 , , , , 0.0 0 1 2 3 4 5 mtDNA Distance 0.3 m 0.2 0.1 FIG. 4. Comparison of mean interisland mtDNA divergence with corresponding mean Mahalanobis and Euclidean distances in multivariate morpho- metric measurement space. beous, and Olive-capped warblers demon- strates that the disjunct populations of Ade- laide's Warbler have been evolutionarily inde- pendent far longer than have populations of the two more continuously distributed species. In- deed, our most striking result was the high lev- el of mtDNA divergence and phylogenetic structure in Adelaide's Warbler, a species with three island-specific and highly distinct mono- phyletic groups of closely related haplotypes (Fig. 2). In contrast, the low mtDNA divergence between the two Plumbeous Warbler popula- tions (Fig. 3) indicates a relatively recent com- mon ancestry, but the presence of reciprocally monophyletic haplotype groups on Dominica and Guadeloupe suggests that even these geo- graphically adjacent populations have attained evolutionary independence. Our reconstruction of phylogeographic variation among popula- tions of Olive-capped Warbler was less detailed owing to small sample sizes. Nevertheless, the very low level of mtDNA divergence between the Cuban and Bahamian samples (Fig. 3) dem- onstrates a recent or continuing evolutionary connection between these populations. The low intraspecific genetic divergences in Plumbeous Warblers and Olive-capped War- biers were paralleled by correspondingly little morphometric variation in these taxa. Nonethe- less, we found significant differences in mor- phology among islands within both species, indicating either that gene flow has been insuf- ficient to override morphological differentia- tion, or that individuals respond develop- mentally to different ecological conditions on each island. The relationship between genetic and morphometric differentiation in Adelaide's Warbler was more complex, because the Puerto Rican and Barbudan populations exhibited highly divergem mtDNA haplotypes but sim- ilar external measurements. The morphological similarity between northern Adelaide's Warbler populations is probably not restricted to size traits, because previous workers (Riley 1904, Curson et al. 1994) have noted plumage simi- larities between Barbudan and Puerto Rican Adelaide's Warblers. Several processes could account for the lack of congruency between mi- tochondrial divergence and morphological di- vergence in this species. First, if the mtDNA tree accurately reflects the evolutionary rela- tionships of the three Adelaide's Warbler pop- ulations, then either morphological evolution in the St. Lucian lineage has been accelerated rel- ative to that in the two northern lineages, or the Puerto Rican and Barbudan populations have evolved similar morphologies independently. Alternatively, morphological similarity be- tween the Puerto Rican and Barbudan popu- lations may result from male-mediated gene flow, which would not be reflected in the ma- ternally inherited mitochondrial genome. Al- though a lack of records of Adelaide's Warblers from other islands and the great distances be- tween islands argue for the genetic isolation of the three extant populations, confirmation of an absence of male-mediated gene flow would require additional study based on paternally or biparentally inherited genetic markers. The 2.2 to 4.7% mitochondrial divergences among Adelaide's Warbler populations equals or exceeds those between many pairs of closely related bird species (e.g. Bermingham et al. 1992, Helbig et al. 1995, Klicka and Zink 1997). Our ongoing studies of parulid warbler rela- tionships suggest that the large divergences among Adelaide's Warbler populations are not a result of high rates of nucleotide substitution in the genes we sequenced. For example, the in- terpopulation divergences in Adelaide's War- bler are higher than corresponding divergences among species in the "Black-throated" group (i.e.D. nigrescens, D. occidentalis, D. townsendi, and D. virens; see Bermingham et al. 1992) in which combined ATPase and COI sequence di- vergence ranges between 0.9 and 4.2% (Lovette, Bermingham, and S. Rohwer unpubl. data). The much lower magnitudes of divergence among island populations in Plumbeous War- biers and Olive-capped Warblers are more typ- ical of genetic divergence within continental species (Bermingham et al. 1992, Helbig et al. 1995, Zink 1997). Several independent calibrations have found that avian mtDNA sequences diverge at a rate of approximately 2% per million years (Shields and Wilson 1987, Tarr and Fleischer 1993, Nunn et al. 1996, Randi 1996, Klicka and Zink 1997). This rate calibration must be applied cautiously because it was derived primarily from RFLP- based estimates of genetic divergence and be- cause the process of nucleotide substitution is not perfectly clock-like (e.g. Ayala 1986, Gilles- pie 1986). Nonetheless, the high degree of dif- ferentiation among Adelaide's Warbler popu- lations provides strong evidence that they have been isolated for a long time; the split that iso- lated the Puerto Rican Adelaide's Warbler lin- eage probably occurred in the late Pliocene (ca. 1.8 to 2.4 million years ago), whereas the more recent split between the Barbudan and St. Lu- cian lineages occurred in the mid-Pleistocene (ca. 1.1 to 1.3 million years ago). In contrast, the small mtDNA divergences within Plumbeous Warblers and Olive-capped Warblers suggest that populations of these species have been connected by gene flow within the past 150,000 to 200,000 years. The range of intraspecific divergence among the three warblers is interesting given the pos- sibility that Antillean bird distributions have been influenced by recent changes in habitat distribution and quality. Fossil remains from the Bahamas suggest a pattern of late Pleisto- cene extinctions in West Indian vertebrates typ- ical of xeric habitats (Pregill and Olson 1981). Pregill and Olson attributed these extinctions to a regional increase in rainfall over the past 10,000 to 20,000 years stemming from climate changes associated with the end of the last gla- cial period. Because Adelaide's Warblers favor dry lowland habitats (Bond 1956, Cruz and De- laney 1984, Raffaele 1989), this species may have been susceptible to the loss of xeric forests in the Lesser Antilles. Biogeographic consid- erations support the recent extinction of at least one population of Adelaide's Warbler Barbuda and Antigua formed parts of a single large is- land during the last glacial maximum (ca. 20,000 years ago), and presumably a popula- tion of Adelaide's Warbler on what is now An- tigua went extinct after the present-day islands were separated by rising sea levels. This ex- tinction may have resulted from anthropogenic causes; Steadman et al. (1984) and Pregill et al. (1988) characterized a 4,300 to 2,500 year-old faunal assemblage from Antigua and found the remains of a number of vertebrates, including seven bird species, that are no longer found on the island, possibly due to habitat degradation (Steadman et al. 1984). Evidence against a more general recent ex- tinction scenario is twofold. First, the presence of extensive dry forest on several islands be- tween St. Lucia and Puerto Rico argues against the hypothesis that recent loss of suitable hab- itat caused the extinction of Adelaide's Warbler populations on these islands. Second, our ge- netic data strongly demonstrate that the evo- lutionary separation of the three extant Adelai- de's Warbler lineages greatly predates the end of the most recent glaciation. The high levels of molecular divergence between the three extant Adelaide's Warbler populations show that they have persisted as evolutionarily isolated units through several Pleistocene glaciation cycles. What biogeographic and evolutionary pro- cesses might account for distributional gaps such as those seen in Adelaide's Warbler? In general, range disjunctions might be created ei- ther by haphazard dispersal events that pass over intervening suitable islands, or by the extinction of populations on intermediate islands. Cases in which disjunct populations lack genetic divergence would support the long-distance dispersal scenario, because the absence of divergence between disjunct pop- ulations would allow little time for the extinc- tion of intervening populations. Genetic sim- ilarity among disjunct populations is unlikely to be maintained by gene flow, because contin- ued movement between disjunct populations should facilitate the colonization of intervening islands and hence close distributional gaps. Support for the alternative extinction-based scenario could come from paleontological evi- dence of a species' presence on an island where it does not presently occur, but avian fossils are not known from most West Indian islands. The proposition that gaps result from extinctions over evolutionary periods of time could also be supported statistically; under the extinction scenario, we would expect a positive relation- ship between range disjunction and interisland genetic divergence. In any particular case, how- ever, large genetic divergences among disjunct populations could also result from old haphaz- ard colonization events with subsequent differ- entiation in isolation. Because we found a greater degree of genetic divergence among the highly disjunct popula- tions of Adelaide's Warbler than among the less-disjunct populations of Plumbeous War- biers and Olive-capped Warblers, our data sup- port the hypothesis that distributional gaps arise in long-established taxa by the extinction of geographically intermediate populations (Ricklefs and Cox 1972). Although other mt- DNA-based studies of West Indian birds have documented a variety of phylogeographic his- tories superimposed on a common geography (Seutin et al. 1993, 1994; Klein and Brown 1994; Bermingham et al. 1996; Ricklefs and Ber- mingham 1997), Adelaide's Warbler is the only taxon with a large West Indian range disjunc- tion for which genetic data have been pub- lished. A number of other Lesser Antillean spe- cies, including Mimocichla plumbea, Cichlhermi- nia lherminieri, Myadestes genibarbis, Troglodytes aedon, and the Icterus "dominicensis" orioles also have conspicuous gaps within their distri- butions, and studies of these taxa are currently in progress. TAXONOMIC iMPLICATIONS The large mitochondrial differentiation be- tween the three Adelaide's Warbler populations raises the issue of whether each population warrants species status. As noted above, mt- DNA differences between the three popula- tions exceed those between some North Amer- ican Dendroica species, and each Adelaide's Warbler population appears to be a monophy- letic entity. Although the pattern of morpho- logical differentiation is not completely con- gruent with the mtDNA pattern, each popula- tion is also characterized by unique morpho- logical differences. Thus, the criteria of the phylogenetic species concept (Cracraft 1983, Zink and McKittrick 1995) argue for elevating the three Adelaide's Warbler populations to species-level status. The genetic distinctiveness of the three populations also suggests that each should be considered an evolutionarily signif- icant unit (Ryder 1986, Moritz 1994) for pur- poses of conservation. As is common in situations where differen- tiated populations exist in allopatry, their po- tential classification under the biological spe- cies concept (Mayr 1963, 1969) or the recogni- tion species concept (Patterson 1985) is prob- lematic due to a lack of evidence on how individuals from different islands would be- have if they came into contact. Song recogni- tion may play a key role in reproductive isola- tion or the lack thereof in birds, and Bond (1930, 1965) noted qualitative differences in song structure among Adelaide's Warbler pop- ulations. Even within an island, however, Ade- laide's Warbler song varies on a microgeo- graphic scale (Stacier 1996). Nonetheless, we found that Barbudan Adelaide's Warblers re- sponded to recordings of St. Lucian individuals (Seutin pers. obs.); unfortunately, logistical constraints and a lack of recordings from the Barbudan population precluded controlled tests. Formal song-recognition experiments could provide important information on the de- gree of behavioral differentiation among pop- ulations of Adelaide's Warbler. If the three Ade- laide's Warbler populations are accorded spe- cies status, the Puerto Rican population should remain Dendroica adelaidae Baird 1865, the Bar- budan population should be referred to Den- droica subita Riley 1905, and the St. Lucian pop- ulation to Dendroica delicata Sclater 1871. ACKNOWLEDGMENTS Our laboratory work was supported by grants from the Smithsonian Institution and National Sci- ence Foundation (DEB-9419645). Field work was sup- ported by grants from the American Ornithologists' Union, the National Geographic Society, and the Smithsonian Institution. Lovette was supported by an NSF predoctoral fellowship and a Smithsonian Tropical Research Institute fellowship, and Seutin was supported by an NSERC postdoctoral fellow- ship and a Smithsonian Scholarly Studies postdoc- toral fellowship. This study would have been impos- sible without the assistance of the national and local authorities on the islands where these warblers oc- cur. We are grateful to the Agriculture, Forestry, and Environment ministries of St. Lucia, Dominica, Gua- deloupe, Antigua and Barbuda, Puerto Rico, Jamaica, and the Bahamas for granting the permits that have facilitated our studies of Caribbean bird evolution. We thank F. Sheldon and D. Dittmann of the Louisi- ana State University Museum of Vertebrate Zoology and E Gill, R. Ridgely, and D. Agro of the Academy of Natural Sciences of Philadelphia for the generous loan of tissue samples, and S. Cardiff, E Gill, P. Mar- ra, J. V. Remsen, and K. Rosenberg for collecting those tissues and associated voucher specimens. We also thank G. Graves, S. Olson, and J. Angle of the U.S. National Museum of Natural History; G. Bar- rowclough and P. Sweet of the American Museum of Natural History; J. V. Remsen and S. Cardiff of the Louisiana State Museum of Natural Sciences; and R. Ridgely and D. Agro of the Academy of Natural Sci- ences of Philadelphia for making the skin collections under their care available to us. D. Swofford kindly allowed us to use a pre-release version of Paup*. We thank V. Apanius, D. Wechsler, W. Shew, and P. Siev- ert for their assistance in the field; J. Hunt for invalu- able laboratory assistance; B. Kessing for his many helpful suggestions on laboratory techniques and data analysis; and H. Lovette, T. Price, S. Olson, R. Zink, and an anonymous reviewer for their com- ments on an earlier version of the manuscript. LITERATURE CITED AMERICAN ORNITHOLOGISTS' UNION. 1983. Check- list of North American birds, 6th ed. American Ornithologists' Union, Washington, D.C. ARCTANDER, P. 1995. Comparison of a mitochondrial gene and a corresponding nuclear pseudogene. Proceedings of the Royal Society of London Se- ries B 262:13-19. AVISE, J. C. 1994. Molecular markers, natural history, and evolution. Chapman and Hall, New York. AYALA, E J. 1986. On the virtues and pitfalls of the molecular evolutionary clock. Journal of Hered- ity 77:228-235. BAKER, M. C. 1981. A muscle biopsy procedure for use in electrophoretic studies of birds. Auk 98: 392-393. BALL, R. M., JR., F. C. JAMES, S. FREEMAN, E. BER- MINGHAM, AND J. C. AVISE. 1988. Phylogeo- graphic population structure of Red-winged Blackbirds assessed by mitochondrial DNA. Proceedings of the National Academy of Sci- ences USA 85:1558-1562. BERMINGHAM, E., S. ROHWER, S. FREEMAN, AND C. WOOD. 1992. Vicariance biogeography in the Pleistocene and speciation in North American wood warblers: A test of Mengel's model. Pro- ceedings of the National Academy of Sciences USA 89:6624-6628. BERMINGHAM, E., G. SEUTIN, AND R. E. RICKLEFS. 1996. Regional approaches to conservation bi- ology: RFLPs, DNA sequence, and Caribbean birds. Pages 104-124 in Molecular genetic ap- proaches to conservation (T. B Smith and R. K. Wayne, Eds.). Oxford University Press, Oxford. BOND, J. 1930. The resident West Indian warblers of the genus Dendroica. Proceedings of the Acade- my of Natural Sciences of Philadelphia 82:329- 337. BOND, J. 1956. Check-list of birds of the West Indies, 4th ed. Academy of Natural Sciences of Phila- delphia, Philadelphia. BOND, J. 1958. Third supplement to the check-list of birds of the West Indies (1956). Academy of Nat- ural Sciences of Philadelphia, Philadelphia. BOND, J. 1965. Tenth supplement to the check-list of birds of the West Indies (1956). Academy of Nat- ural Sciences of Philadelphia, Philadelphia. CRACRAFT, J. 1983. Species concepts and speciation analysis. Current Ornithology 1:159-187. CRUZ, A., AND C. A. DELANEY. 1984. Ecology of the Elfin Wood's Warbler (Dendroica angelae) 1. Dis- tribution, habitat usage, and population densi- ties. Caribbean Journal of Science 20:89-96. CURSON, J., D. QUINN, AND D. BEADLE. 1994. Warblers of the Americas. Houghton Mifflin, Boston. DESJARDINS, P., AND R. MORAIS. 1990. Sequence and gene organization of the chicken mitochondrial genome--A novel order in higher vertebrates. Journal of Molecular Biology 212:599-634. FARRIS, J. S., M. KALLERSJO, A. G. KLUGE, AND C. BULT. 1995. Constructing a significance test for incongruence. Systematic Biology 44:570-572. GILLESPIE, J. H. 1986. Natural selection and the mo- lecular clock. Molecular Biology and Evolution 3:138-155. HASAGAWA, g., H. KISHINO, AND K. YANO. 1985. Dating of the human-ape splitting by a molec- ular clock of mitochondrial DNA. Journal of Mo- lecular Evolution 22:160-174. HELBIG, A. J., I. SEIBOLD, J. MARTENS, AND M. WINK. 1995. Genetic differentiation and phylogenetic relationships of Bonelli's Warbler Phylloscopus bonelli and Green Warbler P. nitidus. Journal of Avian Biology 26:139-153. HELLMAYR, C. E. 1935. Catalogue of birds of the Americas. Field Museum of Natural History Publications in Zoology Series, vol. 13. KEPLER, C. B., AND K. C. PARKES. 1972. A new species of warbler (Parulidae) from Puerto Rico. Auk 89: 1-18. KESSING, B., H. CROOM, A. MARTIN, C. MCINTOSH, W. O. MCMILLAN, AND S. P. PALUMBI. 1989. The sim- ple fool's guide to PCR, version 1.0. Special Pub- lication of the Department of Zoology, Univer- sity of Hawaii, Honolulu. KIMURA, M. 1980. A simple method for estimating the evolutionary rate of base substitutions through comparative studies of nucleotide se- quences. Journal of Molecular Evolution 16:111- 120. KISHINO, H., AND M. HASEGAWA. 1989. Evaluation of the maximum-likelihood estimate of the evolu- tionary tree topologies from DNA sequence data, and the branching order in Hominoidea. Journal of Molecular Evolution 29:170-179. KLEIN, N. K., AND W. M. BROWN. 1994. Intraspecific molecular phylogeny in the Yellow Warbler (Dendroica petechia), and implications for avian biogeography in the West Indies. Evolution 48: 1914-1932. KLICKA, J., AND R. M. ZINK. 1997. The importance of recent ice ages in speciation: A failed paradigm. Science 277:1666-1669. MAYR, E. 1963. Animal species and evolution. Har- vard University Press, Cambridge, Massachu- setts. MAYR, E. 1969. Principles of systematic zoology. Mc- Graw-Hill, New York. MAYR, E., AND L. L. SHORT. 1970. Species taxa of North American birds. Publications of the Nut- tall Ornithological Club No. 9. MINGEL, R. M. 1964. The probable history of species formation in some northern wood warblers (Pa- rulidae). Living Bird 3:9-43. MORITZ, C. 1994. Applications of mitochondrial DNA analysis in conservation: A critical review. Molecular Ecology 3:401-411. NEI, M. 1987. Molecular evolutionary genetics. Co- lumbia University Press, New York. NUNN, G. g., J. COOPER, P. JOUVENTIN, C. J. R. ROB- ERTSON, AND G. G. ROBERTSON. 1996. Evolution- ary relationships among extant albatrosses (Pro- cellariiformes: Diomedeidae) established from complete cytochrome-b gene sequences. Auk 113:784-801. PATTERSON, H. E. H. 1985. The recognition concept of species. Transvaal Museum Monographs 4: 21-29. PETERS, J. L. 1968. Check-list of birds of the world, vol. 14. Museum of Comparative Zoology, Cam- bridge, Massachusetts. PREGILL, G. K., AND S. L. OLSON. 1981. Zoogeogra- phy of West Indian vertebrates in relation to Pleistocene climatic cycles. Annual Review of Ecology and Systematics 12:75-98. EREGILL, G. K., D. W. STEADMAN, S. L. OLSON, AND E V. GRADY. 1988. Late Holocene fossil vertebrates from Burma Quarry, Antigua, Lesser Antilles. Smithsonian Contributions to Zoology No. 463. RAFFAELE, H. A. 1989. A guide to the birds of Puerto Rico and the Virgin Islands. Princeton Univer- sity Press, Princeton, New Jersey. RANDI, E. 1996. A mitochondrial cytochrome b phy- logeny of the Alectoris partridges. Molecular Phylogenetics and Evolution 6:214-227. RICKLEFS, R. E., AND E. BERMINGHAM. 1997. Molec- ular phylogenetics and conservation of Carib- bean birds. E1 Pitirre 10:85-92. RICKLEFS, R. E., AND G. W. COX. 1972. Taxon cycles in the West Indian avifauna. American Naturalist 106:195-219. RIDGWAY, R. 1882. Description of a new warbler from the island of Santa Lucia, West Indies. Pro- ceedings of the United States National Museum 5:525-526. RIDGWAY, R. 1902. Birds of North and Middle Amer- ica, part II. Bulletin of the United States National Museum 50:587-588. RILEY, J. H. 1904. Catalogue of a collection of birds from Barbuda and Antigua, British West Indies. Smithsonian Miscellaneous Collections 47:277- 291. RYDER, O. A. 1986. Species conservation and system- atics: The dilemma of subspecies. Trends in Ecology and Evolution 1:9-10. SAS INSTITUTE. 1987. SAS/STAT user's guide, ver- sion 6.03. SAS Institute, Inc., Cary, North Caro- lina. SEUTIN, G., J. BRAWN, g. E. RICKLEES, AND E. BER- MINGHAM. 1993. Genetic divergence among pop- ulations of a tropical passerine, the Streaked Sal- tator (Saltator albicollis). Auk 110:117-126. SEUTIN, G., N. K. KLEIN, R. E. RICKLEES, AND E. BER- MINGHAM. 1994. Historical biogeography of the Bananaquit (Coerebafiaveola) in the Caribbean re- gion: A mitochondrial DNA assessment. Evolu- tion 48:1041-1061. SEUTIN, G., L. M. RATCLIFFE, AND P. t. BOAG. 1995. Mitochondrial DNA homogeneity in the phe- notypically diverse redpoll finch complex (Aves: Carduelinae: Carduelis fiammea- hornemanni). Evolution 49:962-973. SEUTIN, G., B. N. WHITE, AND P. t. BOAG. 1991. Pres- ervation of avian blood and tissue samples for DNA analyses. Canadian Journal of Zoology 69: 82-90. SHIELDS, G. E, AND A. C. WILSON. 1987. Calibration of mitochondrial DNA evolution in geese. Jour- nal of Molecular Evolution 24:212-217. STACIER, C. A. 1996. Acoustical features of song cat- egories of the Adelaide's Warbler (Dendroica ade- laidae). Auk 113:771-783. STEADMAN, D. W., G. K. PREGILL, AND S. L. OLSON. 1984. Fossil vertebrates from Antigua, Lesser Antilles: Evidence for late Holocene human- caused extinction in the West Indies. Proceed- ings of the National Academy of Sciences USA 81:4448-4451. STEEL, M. 1994. Recovering a tree from the Markov model. Applied Mathematics Letters 7:19-23. STRIMMER, K., AND A. VON HAESELER. 1996. Quartet puzzling: A quartet maximum-likelihood meth- od for reconstructing tree topologies. Molecular Biology and Evolution 13:964-969. STRIMMER, K., AND A. VON HAESELER. 1997. PUZZLE 3.1: Maximum-likelihood analysis for nucleotide and amino acid alignments (computer pro- gram). Universitaet Muenchen, Muenchen, Ger- many. SWOFFORD, D. L., G. J. OLSEN, P. J. WADDELL, AND D. M. HILLS. 1996. Phylogenetic inference. Pages 407-514 in Molecular systematics, 2nd ed. (D. M. Hillis, C. Moritz, and B. K. Mable, Eds.). Sinauer Associates, Sunderland, Massachusetts. TARR, C. L., AND g. C. FLEISCHER. 1993. Mitochon- drial-DNA variation and evolutionary relation- ships in the Amakihi complex. Auk 110:825-831. ZINK, R. M. 1991. The geography of mitochondrial DNA variation in two sympatric sparrows. Evo- lution 45:329-339. ZNK, R. M. 1996. Comparative phylogeography in North American birds. Evolution 50:308-317. ZNK, R. M. 1997. Phylogeographic studies of North American birds. Pages 301-324 in Avian molec- ular evolution and systematics (D. P. Mindell, Ed.). Academic Press, San Diego, California. ZINK, g. M., AND g. C. BLACKWELL. 1996. Patterns of allozyme, mitochondrial DNA, and morphomet- ric variation in four sparrow genera. Auk 113: 59-67. ZINK, g. M., AND M. C. MCKITRICK. 1995. The debate over species concepts and its implications for or- nithology. Auk 112:701-719. Associate Editor: R. M. Zink APPENDIX. Variable nucleotide sites among conspecific mtDNA haplotypes in three West Indian Dendroica warblers. Dots indicate a match to the uppermost conspecific sequence. Sample size (n) indicates the num- ber of individuals with identical haplotypes. PR = Puerto Rico; BA = Barbuda; SL = St. Lucia; DO = Dominica; GU = Guadeloupe; BH = Bahamas; CU = Cuba. Island and haplo- type n ATPase 8 ATPase 6 Cytochrome oxidase I PR A 3 PR B 1 PR C 1 PR D 1 PR E 1 BA A 4 BA B 1 SL A 2 SL B 1 SL C 4 SL D 1 DOA 1 DO B 1 DO C 1 DOD 1 DO E 1 DO F 2 DOG 1 DOH 1 DO I 5 GU A 1 GU B 1 GU C 2 GUD 1 GU E 1 GU F 2 GUG 1 GUH 2 GU I 1 GU J I BH A 2 CU A 1 Adelaide's Warbler CTTAGATGT TTCCTTGC GCGGAGTGATTCCATTCATTATCACACTAGTG CTTGC CACTTGTTTTGGCCTTATCTTGGAC . . .G ........................................................ C .............. A..T ß . .G ................. G ...................................... C ................. T ß . .G ........................ T ............................... C .................. ß . .T ....................... C ........ G .......... T...C .................. . . .G..C.. TCC. AG. A. TCC. AG. A. ß C.. AGCA. . C.. AGCA. . C.. AGCA. ß C.. AGCA. AGATG C...A C .... C.G. CA, C C.G. . CT. CCAAATAA. T. A. CC. AGC. TGCC. CTG. G. C. A. A . CT. CCAAATAA. T. A. CC. AGC. TGCC. CTG. G. C. A. A ß CT.. CA. A. AA. TC. GCC... C. T. CCGC.. TGT. GACA CCT.. CA. A. AA. TC. GCC... CCT. CCGC.. TGT. GACA ß CTT. CA. A. AA. TC. GCC... C. T. CCGC..TGT. GACA CCT.. CA. A. AA. TC. GCC... CCT. CCGC.. T. T. GACA Plumbeous Warbler GCCAGTCCGGTGGCCTAT ........ A ....... G. ß . .G,C .... C.AT...C ß . .G,C...AC.A .... C ß . .G.C .... C.A .... C ß .A..C .... C .... C.C .T..AC .... C ..... C .T...C.T, .C...T .C ..... C .... CA .... C ß . .G.C .... C.AT. .C ..... CT...C ..... C ..... C .... C ..... C Olive-capped Warbler ACTA TTCG TCCTAT..TCACCCCAATTCCG.TCC.A.. TCCTAT...CACCCCAATTCCG.TCC.A.. TCC.A.G...ACCCCA.T.CCGC.CC.AG. TCC.A.G...ACCCCA.T.CCGC.CC.AG. TCC.A.G...ACCCCA.T.CCGC.CC.AG. TCC.A.G...ACCCCA.T.CCGC.CC.AG. CAGATCATTACTT .G.G ..... G..C ......... G.CC ß - .G ..... G..C ß .A ...... G..C ............ C ..... T...G..C ..... T...G..C ..... T...GT.C ..... T...G..C ..... T...G..C ..... T..CG..C ..... T...G..C T .... T..CG..C .... CT.C.G..C