HISTORICAL BIOGEOGRAPHY of the BANANAQUIT (<I>COEREBA

Evolution, 48(4), 1994, pp. 1041-1061

HISTORICAL BIOGEOGRAPHY OF THE BANANAQUIT (COEREBA FLA VEOLA) IN THE CARIBBEAN REGION: A MITOCHONDRIAL DNA ASSESSMENT

GILLES SEUTIN,'.4 NEDRA K. KLEIN,2 ROBERT E. RICKLEFS,' AND ELDREDGE BERMINGHAM'" , Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Republic ofPanama 2 Museum ofZoology. University ofMichigan. Ann Arbor, Michigan 48109 3 Department ofBiology, University ofPennsylvania. Phi/adelphia. Pennsylvania 19104-6018

Abstract.-We analyzed mitochondrial DNA (mtDNA) restriction-site variation in bananaquit (Coerebajlaveola; Aves, Coerebinae) populations sampled on 12 Caribbean islands and at 5 con tinentallocalities in Central America and northern South America. Multiple fixed restriction-site differences genetically defined several regional bananaquit populations. An mtDNA clade repre senting all Jamaican bananaquits was the most divergent; the estimated average sequence diver

gence (dxy) between Jamaican and all other mtDNA haplotypes surveyed was 0.027. Three groups of populations, representing Central America, northern South America, and the eastern Antilles

(Puerto Rico to Grenada) were nearly equally differentiated among themselves (average dxy = 0.014), and may represent a single, recent range expansion. Within the eastern Antilles, three geographically restricted haplotype groups were identified: Puerto Rico, north-central Lesser An tilles (U.S. Virgin Islands to St. Lucia), and Grenada-St. Vincent. The evolutionary relationships ofthese groups were not clear. Genetic homogeneity ofthe island populations from the U.S. Virgin Islands to St. Lucia suggested a recent spread ofa specific north-central Lesser Antillean haplotype through most of those islands. Haplotype variation across this region indicated that this spread may have occurred in two waves, first through the southernmost islands ofSt. Lucia, Martinique, and Dominica, and more recently from Guadeloupe to the north. The geographic distribution of mtDNA haplotypes, and ofbananaquit populations, suggests periods ofinvasiveness followed by relative geographic quiescence. Although most genetic studies of bird populations have revealed homogeneity over large geographic areas, our findings provide a remarkable counterexample of strong geographic structuring of mtDNA variation over relatively small distances. Furthermore, although the mtDNA data were consistent with several subspecific distinctions, it was clear that named subspecies do not define equally differentiated evolutionary entities.

Key words.-Bird, Caribbean region, Coerebajlaveola, genetic divergence, genetic diversity, his torical biogeography, mtDNA, RFLP, West Indies.

Received June 15, 1993. Accepted September 15, 1993.

Historical biogeography is concerned with re al. 1987; Avise 1992; Bermingham et al. 1992). constructing the history of geographic distribu In addition, application of the molecular-clock tions of populations and species. Such recon concept permits a rough dating of divergence structions help us to understand the dynamics of times, which may then be compared to histories evolutionary changes in species and ofecological ofclimate change and thedates ofgeologic events changes in communities. Contemporary distri having biogeographic consequence. butions and inferences about genetic divergence Although most terrestrial species are distrib obtained from phenotypic characters provide the uted on continental land masses, archipelagoes basis for most biogeographic analyses at the tax provide excellent opportunities for historical onomic level of the species (Brooks and Me biogeographic analyses. For biogeographers, the Lennan 1991). DNA restriction-site analysis and primary advantage of island groups over conti DNA sequencing now provide important data nents lies in the subdivision ofspecies into dis for historical biogeography because they permit crete, geographically isolated populations. Thus, direct measurement of genetic divergence be distributions can be defined in terms ofpresence tween populations or genotypes and increased or absence on an island-by-island basis, and phe certainty in the inference ofthe history ofgenetic notypic or genotypic variation can be character changes (Bermingham and Avise 1986; Avise et ized by within-population and between-popu lation variance components that clearly match • Mailing address: Smithsonian Tropical Research discrete populations. The analysis of genetic Institute, Unit 0948, APO AA 34002-0948, USA. variation within and among island populations

1041 1042 GILLES SEUTIN ET AL. provides a powerful tool for assessing the his individuals migrate readily between islands, torical development of biogeographic patterns forming a single, genetically homogenous pop and the processes responsible for these patterns ulation. (Rosen 1976, 1978; Kluge 1988). We previously used those biogeographic prem One island chain that has been the focus of ises to interpret the history of the streaked sal much attention from biogeographers, and which tator (Saltator albicollis; Seutin et al. 1993) and is central to the present study, is the Lesser An ofthe yellow warbler (Dendroica petechia; Klein tilles. The geologic and biogeographic histories 1992) in the Caribbean. Populations of the of those Caribbean islands have been summa streaked saltator on the adjacent islands ofDom rized in several recent studies (Pregill 1981; Sykes inica, Martinique, and St. Lucia in the Lesser et al. 1982; Rosen 1985; Roughgarden 1990). Antilles share most mitochondrial DNA (mt The island arc that forms the present-day Lesser DNA) restriction-site haplotypes and exhibit a Antilles is believed to have cleared the south maximum genetic divergence among haplotypes eastern edge ofthe Bahama Platform on its east (dx y ) of 0.006. Recent derivation of the popula ward movement across what is now the Carib tions from a common ancestor, ratherthan a high bean Basin approximately 35 mya (Burke 1988). level of migration between islands, is indicated The formation and present-day positioning of by the limited Caribbean distribution ofthe spe the Lesser Antilles are undoubtedly more recent, cies to the central part of the Lesser Antilles. that is, 20+ mya, but certainly antedate the ap High migration levels would probably further pearance of the study species, the bananaquit colonization of the island chain. Antillean hap Coerebaflaveola, in the archipelago. Lowered sea lotypes differ from continental haplotypes ofthe levels during the glacial maxima of the Pleisto same species by d., values exceeding 0.06, fur cene did not substantially alter the area and iso ther implying low vagility. On some of the is lation ofmost Lesser Antillean islands, although lands occupied by saltators, Klein (1992) ob larger banks exposed during glacial maxima pro served two coexisting mtDNA haplotypes in the vided terrestrial connections between some is yellow warbler that differed by dx y = 0.0135, one lands (e.g., Puerto Rico and some of the Virgin of West Indian origin and the other of Venezu Islands; St. Kitts and Nevis; Antigua and Bar elan derivation. This suggests a more complex buda) (Donnelly 1988). origin for Lesser Antillean populations of the Knowledge about biogeographic and evolu yellow warbler, compared to those ofthe streaked tionary processes allows the generation of his saltator, with invasion from two genetically dif torical biogeographic premises. First, space and ferentiated source areas. history should be generally related; one expects Here we continue an appraisal of historical individuals orgenotypes that are genetically more biogeography and evolution of Antillean land divergent to be geographically more distant. Sec birds by reporting on levels of intrapopulation ond, genetic divergence increases with time and variability and interpopulation differentiation in with the strength ofbarriers to dispersal between mtDNAamongisland populations ofC.jIaveola. populations. Third, when one population is a The bananaquit is a small (8-12 g; Faaborg and genetic subset of another population, the latter Winters 1979), nectarivorous and frugivorous is likely to be ancestral, assuming that the two emberizine (Passeriformes) of uncertain affini populations had similar demographic histories. ties. In recent treatments, it has been variously These premises may be used to interpret the his placed in a monotypic subfamily (Coerebinae; torical development of biogeographic patterns. e.g., American Ornithologists' Union 1983) or For example, ifgenetic relationships over island with the tanagers (Thraupini; e.g., Sibley and chain A-B-C were 1-3-2, a more complex history Monroe 1990). It is widely distributed, mainly of colonization than a simple island-by-island in lowland habitats, from southern Mexico, stepping-stone model could be inferred. In this through Central America and the West Indies, particular case, island B would have been colo to most of humid-tropical South America (fig. nized last from island C, indicating either hap I). Its abundance varies greatly through its range; hazard dispersal or recolonization following ex it is most numerous in semiopen to open habitats tinction ofthe population on B. Ifthe distribution and is scarce or even absent from extensively of genotypes was I-I-lover the same island forested regions, particularly in continental ar group, probably either colonization was recent eas. The bananaquit is especially common on and also rapid compared to genetic change, or islands and probably is the most abundant bird MTDNA BIOGEOGRAPHY OF CARIBBEAN BANANAQUITS 1043 species in the West Indies (Lack 1976, Cox and Ricklefs 1977, Wunderle 1985). According to Bond (1963, p. 93), C. j/aveola is of South American origin, and West Indian populations have resulted from two invasions: "one entering via Grenada and spreading north and west to Jamaica, the other from Central America, spreading north and east to the Ba hamas." Thus, two subspecies groups are rec ognized in the West Indies (Bond 1956, 1963): the bahamensis group from the Bahamas (ba hamensis), Cayman Islands (sharpei), coastal is lands off Yucatan (eaboti), Isla Providencia (tri color) and Isla San Andres (oblita); and the An tillean group of II subspecies considered in this article (table I). In total, Paynter (1968) rec ognized 41 bananaquit subspecies, based pri marily on variation in plumage coloration. In FIG. I. Geographic distribution of Coereba j/aveola addition to this geographic variation in plumage, with collection localities indicated. Acronyms for lo the species exhibits plumage dimorphism on the calities are defined in table I. islands of Grenada (subspecies aterrima), St. Vincent (atrata), Los Testigos (laurae), and Los Roques (lowiz), where individuals with both nor tribution which includes the Lesser Antilles, a mal and melanistic plumage occur. This poly chain oflong-isolated, oceanic islands that could morphism has been described in detail by Wun have been colonized only by overwaterdispersal. derle (l981a,b,c, 1983). This study offers insights on avian colonization Our genetic analysis of C. j/aveola in the Ca of the West Indies that address the relationship ribbean region thusconsidersa species with high between history and biogeographic patterns. Our ly variable phenotypes having a widespread dis- results also provide additional evidence suggest-

TABLE I. Subspecies and samples of the bananaquit from the West Indies and continenta1locations around the Caribbean Basin. ns, not sampled.

Sample Haplotype Nucleotide Subspecies Location Code size diversity diversity j/aveola Jamaica JA 10 0.889 0.00369 bananivora Hispaniola ns nectarea Tortue ns portoricensis Puerto Rico PR 25 0.853 0.00213 sanctithomae St. Thomas ST I St. John SJ I newtoni St. Croix SC 10 0.378 0.00054 bartholemica Montserrat MO 5 Guadeloupe GU 6 Dominica DO 19 0.790 0.00274 martinicana Martinique MA 8 St. Lucia SL 18 0.758 0.00188 barbadensis Barbados ns atrata St. Vincent SV 15 0.848 0.00202 aterrima Grenada GR 17 0.544 0.00080 luteola Venezuela (Sucre) VS 12 0.758 0.00258 Venezuela (Distrito Federal) VD 4 Venezuela (Falcon) VF 2 mexicana Panama (Bocas del Toro) PA 8 Costa Rica CR I cerinoclunis Panama (Pearl Islands) PE 8 Total 170 1044 GILLES SEUTIN ET AL. ing that Neotropical bird species exhibit more ever, liver, brain, kidney, heart, or lung tissues phylogeographic structure than their temperate were also used. The DNAs were not dialyzed counterparts. before restriction analysis, because we have found this step unnecessary. Purified mtDNA was ob MATERIAI.S AND METHODS tained from 24 individuals from five sampling Field Methods. - We obtained 170 tissue sam locales (Jamaica, Puerto Rico, Guadeloupe, ples ofbananaquits from Jamaica, Puerto Rico, Dominica, and Venezuela) by ultracentrifuga the three U.S. Virgin Islands, seven Lesser An tion in cesium chloride-propidium iodide (CsCI tillean islands, three locations in Venezuela, the PI) density gradients, following the protocols of Pearl Islands off the Pacific coast ofPanama, the Dowling et aI. (1990) and Klein (1992). province ofBocas del Toro on the Atlantic coast For the Southern blotting analyses, approxi of Panama, and the Pacific coast of Costa Rica mately 2 JLg of total DNA or I ng of purified (table I, fig. I). Detailed information on sam mtDNA was digested with 5-10 units ofenzyme, pling localities is available from the authors. Both following the manufacturer's recommendations. light-phase and dark-phase color morphs were We used 2 enzymes with r values equal to 5.3 represented in our samples from Grenada and (Ava I and Hinc II) and 14 enzymes with rvalues St. Vincent. Each sample consisted oftissues col equal to 6 (BamH I, Bgl I, Bgl II, Cia I, Dra I, lected primarily from adult birds; judging from EcoR I, EcoR V, Hind III, Nco I, Nde I, Pst I, the presence orabsence ofa cloacal protuberance Pvu II, Sac I, and Stu I). Restriction fragments in the biopsied birds, or from gonadal inspection were separated electrophoretically in 1.0% aga on collected specimens, both sexes were repre rose gels and blotted by capillarity onto Zeta sented in most samples. bind® membrane as detailed in Seutin et aI. Three Puerto Rican specimens were obtained (1993). as whole corpses that had been stored at - 20°C Two to four blots were prehybridized at once for up to several years; we extracted genomic in canisters rotating at approximately 5 rpm. DNA from the pectoral muscle of these speci Prehybridizations were conducted at 65°C for 1 mens. Most samples from Dominica, Marti 3 h, in 20-30 ml ofsolution (10% dextran sulfate, nique, St. Lucia, Grenada, Venezuela, and the 0.5M NaCI, 1% SDS). We used a bananaquit Pearl Islands consisted of pectoral muscle bi mtDNA preparation purified two times on CsCI opsies of mist-netted birds. The biopsy proce EtdBr gradients as a probe. Traces of nuclear dure followed Baker (1981), except that we ex DNA were removed from this probe by running cised a triangular piece of the lower part of the native or EcoR I-digested aliquots in 0.9% low breast muscle; birds were released within an hour melting point agarose gels and extracting the of the procedure. We preserved the samples at mtDNA fragments with the GeneClean® pro ambient temperature in the field in a salt-di cedure (BiolOl, LaJolla, Calif.). In random methyl sulfoxide (DMSO) solution (Seutin et aI. priming reactions, a few nanograms ofprobe were 1991). All other tissue sampIes were collected radioactively labeled with [a-32P]dCTP to very from sacrificed birds, stored in the field in liquid high specific activity(IOcI09 dpm/JLg). Blots were nitrogen, and later transferred to ultracold hybridized for 12-18 h, and then washed as de (-70°C) freezers. All samples were collected and scribed by Seutin et aI. (1991). Scorable bands imported under appropriate permits and licens on autoradiographs were obtained in 16-120 h es. Voucher specimens for the bananaquits col on Kodak XAR® film at -70°C using one in lected by Klein have been deposited in the fol tensifying screen. No attempt was made to score lowing research collections: Museum ofZoology, fragments smaller than 300 base pairs (bp). University ofMichigan; Department ofNatural Mapping. - We physically mapped the posi Resources, Puerto Rico; Field Museum of Nat tion ofmost restriction sites in single individuals ural History, Chicago; Fundacion La Salle, Ca from Puerto Rico and Venezuela (haplotypes racas, Venezuela. PR12 and VE7), using a double-digestion strat DNA Extraction, Restriction Digests, Electro egy (Dowling et aI. 1990). Typically, 1-2 ng of phoresis, Southern Blotting, and Probing.- We purified mtDNA were digested simultaneously extracted total cellular DNA from most samples or sequentially with two enzymes, fragments were using the CTAB extraction procedure described end-labeled with the appropriate [a- 32P]dNTP, in Seutin et aI. (1993). We typically used pectoral and electrophoresed in polyacrylamide and aga muscle for these extractions; on occasion, how- rose gels that were then dried and exposed to MTDNA BIOGEOGRAPHYOF CARIBBEAN BANANAQUITS 1045 autoradiographic films (for details, see Klein ,?-~(.~.f'~~\..t' i-I'~ ~-:,~' ~o/,~ ,, 1992). Mapped restriction sites (fig. 2) are pre ? m sented relative to the single Cia I site found in <.J'ct<}'.i-!'~I'~:i' ~01~,~~~~~~~

...... ~ JA PR 5T 5J SC MO 00 ~ 5L 5V V5 PE PA CR ~ u ------"0 ------"0 j ZA ABC CCCCCCC j CCBB HHFF FFF j 5

FIG. 3. Restriction-fragment-Iength-polymorphism analysis ofCoerebaflaveola from various localities around the Caribbean basin, using the endonuclease Stu I. Two-letter acronyms at the top refer to sampling localities as presented in table I; single letters below refer to restriction-site patterns. A l-kilobase ladder was used as a molecular size marker.

Panama (table I, fig. I). Fifty-eight mtDNA hap MtDNA Relationships among Populations. lotypes were observed (table 2), and for two of Multiple fixed restriction-site differences genet those (PRI2 and VE7) most restriction sites were ically defined six regional bananaquit popula physically mapped (fig. 2). A total of 108 sites tions (fig. 4): Jamaica (JA), Central America were identified, representing 3.8% of the ap (Costa Rica and Panama; hereafter CAl, Vene proximately 16.8 kilobase pair bananaquit zuela (VE), southern Lesser Antilles (Grenada mtDNA genome. On the average, 66 sites were assayed for each individual, of which 46 were and St. Vincent; GSV), north-central Lesser An shared by all bananaquits. No obvious mtDNA tilles and the U.S. Virgin Islands (St. Lucia to St. size variation was noted across the birds sur Thomas; LA), and Puerto Rico (PR). In most veyed; we did, however, observe one possible cases, thegeographic structuringofmtDNA vari example ofHinc II restriction-site heteroplasmy ation was well documented in neighbor-joining in our unique sample from Costa Rica. (fig. 5), UPGMA (results not shown) and Wagner TABLE 2. Geographical origin and frequency ofoccurrence ofmtDNA haplotypes in Coerebaflaveola. Codes for sample sites are from table I, and locations are shown in fig. 1. Letters indicate restriction-site patterns corresponding to the following enzymes: BamH I, Bgi I, Bgi II, Cia I, Dra I, EcoR I, EcoR V, Hind III, Nco I, Nde I, Pst I, Pvu II, Sac I, Stu I, Ava I, and Hinc II.

Ta- JA PR ST SJ SC MO GU DO MA SL SV GR VS VD VF PA PE CR tal JAI ABCDCCCBCBDECADL I I JA2 ABCDCFGBCBDECZDL 2 2 JA3 ABCDCFGBCBDECYDM 1 I JA4 ABCECFGBCBDECZDM 3 3 JA5 ABCECEGBCBDECZDM 2 2 JA6 ABCCCFGACBDECZCM 1 1 PRI CBDCCCECCCCCCBCD I 1 PR2 CBDCCCEDCCCCCBCE I 1 PR3 CBDCCDECCCCCCBCC 1 1 PR4 CBDCCDECCCCCCBCD 8 8 PR5 CBDCCDECCCCCCBCF 1 1 PR6 CBDCCDECCCCCCBCI 1 I PR7 CBDCCDECCCCCCBBD I 1 PR8 CBDCCDECCCCCCACD 6 6 PR9 CBDCCDECCACCCACD I I PRIO CBDCCDECCCCECACD 2 2 PRII CBDCCDICCCCCCACD I 1 PR12 CBDCEDECCCCECACC I 1 LAI CCCCCCCCCCCCCCCC 8 5 6 8 3 9 39 LA2 BCCCCCCCCCCCCCCC 1 2 3 LA3 BCCCDCCCCCCCCCCC 2 2 LA4 CCCCDCCCCCCCCCCC 1 LA5 CCCCFCCCCCCCCCCC 1 LA6 CDCCCCCCCCCCCCCC 1 LA7 CDCCCCGCCCCCCCCC 4 2 7 LA8 CDCCCCGCCCCCCCCW 1 1 LA9 CCCCCCGCCCCCCCCC 1 LA10 CDCCCDGCCCCCCCCC 1 LAII CCCCCCCECCCCCCCC 1 LAI2 CCCCCCCCBCCCCCCC 1 LAl3 CCCCCCCCCDCCCCCC 1 LAI4 CCCCCCCCCCCBCCCC 1 LA15 CCCCCCCCCCCCBCCC 2 LAI6 CCCCCCCCCCCCCDCC 2 2 LA17 CCCCCCCCCCCCCCCG I LA18 CCCCCCCCCCCCCCCH 1 LAI9 CECCECCCCCCCCCCC 1 GSVI CBCCCCDCCCCCCBCC 5 2 7 GSV2 CBCCCCDCCCCCCBCZ I 1 GSV3 CBCCCCDCCCCDCBCC 3 II 14 GSV4 CBCCCCECCCCDCBCC 1 I GSV5 CBCCCCDCCCCDCBCB 3 3 GSV6 CBCCCCDCCCCDDBCC I I GSV7 CBCCCCCCCCCDDBCC 1 1 GSV8 CBCCCCDCCCCFCBCC 4 4 VEl CCCCCCFCDCBCCHDZ 2 2 VE2 CCCCCCHCDCBCCIDZ I I VE3 BCCCCCHCDCBCCHDZ 1 I VE4 CCCCCCHCDCBCCHDZ 6 2 8 VE5 DCCCCCHFDCBCCHDZ 1 1 VE6 CCCCCCHFECBCCHDZ 1 I VE7 CCCCCCHCECACCHDZ 3 3 VE8 CGCCCCHCECACCHDZ I 1 CAl BBECCCFCCCBCCGDX I 1 CA2 BBECCCFCCCBCCFDX 4 4 CA3 BBECCCFCCCBDCFDX 3 3 CA4 BBECCCFCCCBCCFCA 8 8 CA5 CFECCCFCCCBCCFEZ I 1048 GILLES SEUTIN ET AL.

FIG. 4. Relationships between mtDNA haplotypes within groups of populations identified on the basis of multiple restriction-sitedifferences (see text). Minimal distance betweengroups of populations are also shown. Each tic mark represents the loss or gain of a restriction site. parsimony (fig. 6) analyses. However, these anal tillean bananaquit populations was 0.014 (range yses were not in full agreement regarding the 0.011-0.022), a mean value identical to that ob relationships of the three Antillean population served between the Panama and Venezuela pop groups (PR, LA, and GSV; see below). ulations. These levels of divergence are higher The mtDNA clade representing the 10 Ja than those typically observed between conspe maican individuals was the most divergent; six cific avian populations and are closer to levels fixed restriction-site differences distinguished the seen between closely related species (Avise and Jamaican bananaquits from all others. In pair Zink 1988; Seutin et al. in press). wise comparisons between Jamaican birds and Within Panama, a minimum of three restric all other samples, we observed a minimum of 15 tion-site differences distinguished the single restriction-site changes (fig. 4), and an estimated mtDNA haplotype carried by the eight Pearl Is mean mtDNA sequence divergence (dx y ) of0.029 lands bananaquits (c. f cerinoclunis) from the (range 0.027-0.035). In birds, such levels ofdif three haplotypes assayed in the eight birds from ferentiation are typical of interspecific relation Bocas del Toro (C.f mexicana; fig. 4); the mean ships (see table 3 in Seutin et al. 1993). Numer genetic distance (d x y ) between these populations ical analyses, assuming a mid-point root, further was 0.004. A minimum of five site differences indicated that a continental mtDNA clade, rath were inferred between bananaquits from Bocas er than the Jamaican clade, was the sister taxon del Toro and our unique Costa Rica specimen, to a clade comprised of the three eastern Antil which all belong to the mexicana subspecies. lean mtDNA haplotype groups (PR, LA, GSV; Thus, mtDNA differentiation was higher among figs. 5, 6). The mean mtDNA sequence diver individuals within the mexicana subspecies than gence separatingcontinental from all eastern An- it was between this and the cerinoclunis subspe- MTDNA BIOGEOGRAPHY OF CARIBBEAN BANANAQUITS 1049

r---c=:-f»f.i JA5 JA2

,------,.;;.;;.;;~PR9 PRIO ,,# ,,'$'.... ,,'$'''- ~--PRI2 '----'----' mtDNA distance (d,y) GSVI

11---~:!....-VE8

CA2 CAl CA4 CA3 CA5

FIG. 5. Neighbor-joining cluster analysis of 58 Coerebajlaveola mtDNA haplotypes. cies. Within Venezuela (c. f luteo/a), the two moplasious characters in the restriction-site data individuals collected in the Estado de Falcon had set. Mean mtDNA distances (dx y ) between the the same mtDNA haplotype (VE4) as six indi three groups were PR-LA, 0.009; PR-GSV, viduals from Estado de Sucre. However, mtDNA 0.006; LA-GSV, 0.005. haplotypes from individuals collected in the Dis The limited geographic distribution of mt trito Federal (VE7 and VE8) differed by a min DNA haplotypes found in Puerto Rico, Grenada imum of two restriction sites (d x,) = 0.003; fig. and St. Vincent, Jamaica, and most continental 4) from all other VE haplotypes. Sampling lo locations contrasted sharply with the widespread calities in Falcon and Sucre were coastal and at distribution of north-central Antillean (LA) sea level; our collecting site in the Distrito Fede mtDNA haplotypes. The numerically predomi ral, which lies between Estado de Falcon and nant LA haplotype (LAl), observed in 39 ofthe Estado de Sucre, was inland and at an altitude 68 samples collected from the region (table 2), of 600 m. was in high frequency or fixed on all islands from Within the eastern Antilles, the three geo- . St. Lucia north to St. Croix (subspecies marti graphic units defined earlier (PR, LA, and GSV) nicana, bartholemica, and newtonii. Other LA shared no mtDNA haplotypes among them mtDNAs found in two or more individuals had selves or with other population groups. This more restricted distributions (table 2); for ex clearly indicates the genetic distinctiveness of ample, the second most numerous LA mtDNA those regional populations, but their phyloge haplotype (LA7), seen in seven individuals, was netic relationships could not be clearly deter restricted to St. Lucia, Martinique, and Dom mined either through the analysis of haplotype inica (subspecies martinicana and bartholemi relationships (fig.4) orby numerical analyses (fig. ca). There were no fixed differences between St. 5, 6). This probably resulted from the similar Lucia, Martinique, and Dominica and the unique minimum numbers of site differences distin haplotypes on each ofthe islands had frequencies guishing these three groups (PR-LA, 6; PR-GSV, that were too low to reject the hypothesis that 3; LA-GSV, 3), and the existence ofseveral ho- the three islands represent either a single pan- 1050 GILLES SEUTIN ET AL.

jAI types and large geographic distances within is jA2 jA3 lands. Nonetheless, certain island samples (i.e., jA4 JAS Jamaica, Puerto Rico, St. Lucia, and St. Vincent) jA6 PRI comprised individuals collected over much of PR2 PR3 the island or over a wide range of altitudes. PR4 PRS Jamaica had both the highest mtDNA hap ~~===~~~ lotype and nucleotide diversity (0.89 and0.0037, PRB PR9 respectively), and St. Croix had both the lowest PRIO estimated mtDNA haplotype and nucleotide di PRI2 PRII versity (0.38 and 0.0005, respectively). The other LAI LA2 populations had roughly similar diversity values LA3 LA4 (table 1). Islandarea and both mtDNAhaplotype LAS LA6 and nucleotide diversity were significantly relat LA7 LAB ed (Kendall's rank correlations; in both cases: T LA9 ~==:::= = 0.810, P = 0.011). Klein (1992) also found ::: LA11LA10 LAI2 island area and haplotype diversity to be corre LA13 lated in the yellow warbler. Overall, intrapopu LAI4 LAIS lation mtDNA variability in C.jlaveola was sim LAI6 LAI7 ilar to levels observed in other tropical avian LAIB LAI9 species. Seutin et al. (1993) reported that nucle GSVI GSV2 otide diversity in five insular populations ofSal GSV3 GSV4 tator albicollis ranged from 0.0000 to 0.0028, GSVS GSV6 and was 0.0017 in a mainland population. Most GSV7 GSVB values for bananaquits are in the range reported VEl for local populations ofcontinental North Amer VE2 VE3 ican passerines (0.0008-0.0027; summarized in VE4 VES Seutin et al. 1993, in press). VE6 VE7 We observed reduced mtDNA variability in VEB CAl two groups of bananaquits. The eight assayed CA2 CA3 individuals from Isla Chapera, in the Pearl Is CA4 CAS lands, GulfofPanama, carried a single mtDNA haplotype (table 2). Isla Chapera has been iso FiG. 6. Strict consensus of unrooted Wagner parsi mony trees (length: 98; CI: 0.622) of the relationships lated from the mainland ofPanama for less than among 58 Coerebajlaveo/a mtDNA haplotypes. 10,000 yr (Bartlett and Barghoorn 1973; Fair banks 1989) and supports a large population of bananaquits assigned to a subspecies endemic to mictic population or a recent range expansion the Pearl Archipelago (C. f cerinoclunis; Wet within the Lesser Antilles, as we had previously more et al. 1984). Reduced genetic variability in inferredfor Saltatoralbicollis (Seutin et al. 1993). the Chapera population could be due either to One mtDNA haplotype (LAI5) was assayed in recent colonization of the island by a small two bananaquits collected from the two ends of founder population or, more probably, the loss the region (St. Lucia and St. Croix). The single of haplotype diversity through genetic drift. birds sampled on St. Thomas and St. John each Reduced mtDNA variability was also noted had a unique mtDNA haplotype, but these dif in the northern LA islands (St. Croix south to fered from the common LA haplotype by only Guadeloupe): St. Croix had both the lowest es one restriction site. timated mtDNA haplotype and nucleotide di MtDNA Diversity within Island and Mainland versity (table 1), and we observed no variation Populations. - We sampled 10 or more individ in the combined samples from Guadeloupe (six uals from each ofseven islands and one mainland individuals) and adjacent Montserrat (five in location (table I) and restricted our analysis of dividuals). The probability ofsampling by chance haplotype and nucleotide diversity to these eight II LAI haplotypes from a population showing sites. Because our study focused on larger-scale the levels ofvariability seen in Dominica, Mar patterns ofgeographic variation in bananaquits, tinique, and St. Lucia (frequency of LA I, 0.44) we did not attempt to collect birds across habitat is only 0.000I. Because Guadeloupe is the largest MTDNA BIOGEOGRAPHY OF CARIBBEAN BANANAQUITS 1051 of the Lesser Antillean islands, its apparent re from all other subspecies studied. Furthermore, duced mtDNA variability is most readily ex we discovered no cases where the mtDNA data plained by a founder effect resulting from recent were inconsistent with named subspecies, al colonization or from a post-founding bottleneck. though the other West Indian mtDNA groupings On Grenada and St. Vincent, our collections (GSVand LA) encompassed two and four sub included both the light-phase (n = 9) and dark species, respectively (table 1). To the degree that phase (n = 22) bananaquits. There was no rela mtDNA in these bananaquits evolves in a rough tionship between mtDNA haplotype and bana ly clocklike manner, it appears thatnotall named naquit color phase, whether the island samples subspecies are historically and evolutionarily were considered separately (for Grenada: G = equivalent. This has important consequences for 2.316, df = 2, P > 0.25; for St. Vincent: G = the studyofbiogeographic patterns. Forinstance, 3.165, df= 3, P > 0.30) or jointly (G = 5.222, in the characterization ofthe so-called taxon cy df = 4, P > 0.25). cle for Antillean birds, Ricklefs and Cox (1972) relied on taxonomic distinctions as indicators of DISCUSSION evolutionary divergence and the age of island Our study of mtDNA polymorphism among populations. Genetic analyses will undoubtedly Caribbean bananaquits revealed a degree of'phy modify this and other biogeographic character logeographic structuring and divergence among izations. populations that was high by avian standards. An important finding ofourstudy is that levels The geographic distribution of bananaquit of mtDNA differentiation between regional mtDNA haplotypes indicates that the species has groups of bananaquit populations were higher had a complex history in the Caribbean region. and showed greater phylogeographic structure Our results suggest that certain island popula than those typically observed among avian con tions underwent phases ofgeographic expansion specifics in temperate regions. Data showing el within the Antilles, possibly following either ex evated genetic divergence among populations of tinction in parts of the range, the competitive Neotropical birds have now been obtained for exclusion oflocal populations by invaders, or the species representing a wide range of unrelated replacement of local haplotypes through in passerine families (Capparella 1988; Hackettand trogression. Colonization of the continental Rosenberg 1990; Escalante-Pliego 1991; Peter mainland from the West Indies cannot be re son et al. 1992; Seutin et al. 1993; this study) jected by our data. We consider these issues in suggesting that the tendency is generalized in more detail below. tropical species and due not only to taxon-spe Taxonomic Distinctions and Genetic Diver cific variation in rates of mtDNA and protein gence.- As Banks and Hole (1991) pointed out, evolution. Other factors, such as decreased va early taxonomic work on Caribbean birds "was gility, more stable population sizes, and possibly based mainly on small samples from island pop lower rates of population extinction in the trop ulations in an age when supposed isolation led ics, have to be considered. A careful assessment to the expectation of differentiation, and when of the generality and details of increased phy individual variation within populations was not logeographic structuring of tropical bird popu taken into account." On closer examination, they lations is likely to modify our understanding of found little morphological support for the many the process ofavifaunal diversification. named subspecies ofthe mangrove cuckoo (Coc The Origin of West Indian Bananaquit Pop cyzus minor) in the West Indies and suggested ulations.-Of the bananaquit populations that that the species has been highly vagile within the we sampled, the Jamaican is the mostdistinctive, archipelago, its movement aided by hurricanes and it is almost equally differentiated from all and storms. other populations (fig. 4). Without samples from On the other hand, in the bananaquit, which other Greater Antillean and continental popu in the West Indies occupies much of the same lations we cannot resolve its history. Additional geographical range and habitats as the mangrove data might distinguish whether the Jamaican cuckoo, there was mtDNA support for genetic population is part ofan older West Indian taxon distinctiveness ofsome ofthe named subspecies. that is ancestral to all other bananaquit clades, For example, fixed restriction-site differences or whether it represents an older invasion ofthe separate the subspecies jlaveola (Jamaica) and West Indies from established continental popu portoricensis(Puerto Rico) from one another and lations. All the other populations that we studied 1052 GILLES SEUTIN ET AL. clearly belong to a clade of C. jIaveola that di the Grenada Bank probably to within 10 km of versified subsequent to the isolation of the Ja St. Vincent during Pleistocene periods oflow sea maican population. level (Fairbanks 1989; Pregill and Olson 1981) Three groups ofbananaquit populations, rep allowed gene flow between these islands at times, resenting southern Central America, northern accounting for their presentgenetic closeness. To Venezuela, and the eastern Antilles taken as a the north, the genetic homogeneity of the pop whole (Puerto Rico to Grenada), are approxi ulations between St. Lucia and the U.S. Virgin mately equally differentiated from each other Islands suggests a relatively recent spread of a

(average dxy = 0.014; figs. 4, 5). Assuming that specific north-central Lesser Antillean haplotype mtDNA sequences of these birds diverge at a through most of those islands. Replacement of rate ofapproximately 2% per million years (e.g., older mtDNA haplotypes could have occurred Brown et al. 1979; Shields and Wilson 1987; but through lineage sortingassociated with introgres see Avise et al. 1992; Martin et al. 1992), these sion, or through the competitive exclusion oflo groups became isolated less than I mya. The cal populations by invaders; alternatively the almost equal levels ofdivergence between these colonizers may have replaced island populations groups suggest that they may be products of a that had previously gone extinct. single range expansion. With other models of Iffounder effects resulting in reduced within dispersal (e.g., sequential stepping-stone, non population genotypic variability accompany col synchronized invasions from a single source, in onization, the pattern ofhaplotypic diversity in dependent invasions from differentiated sources), the north-central Lesser Antillean islands sug a hierarchy ofrelationships betweengroups would gests that the later spread of C. jIaveola within be expected. the Lesser Antilles occurred in two waves: an Judging from the distinctiveness of the Ja older one in the southern islands (St. Lucia, Mar maican bananaquit population and the absence tinique, Dominica), which now show moderate of the species from the Yucatan Peninsula and levels ofmtDNA variability, and a more recent Cuba, it appears more likely thatthe colonization progression, probably from Dominica, through ofthe eastern Caribbean islands proceeded from the northern islands (Guadeloupe through St. South America rather than from northern Cen Croix and the Virgin Islands), which today still tral America through the Greater Antilles. Al show little mtDNA variation. Reduced mtDNA ternatively, the expansion may have originated variability on Montserrat and St. Croix may also in the West Indies and spread to the continent. be associated, at least in part, with the probable More extensive sampling, especially in South smaller size of the populations on those small America and northern Central America, should islands. help resolve these possibilities. Alternatively, there may not have been Ci fla History of the Bananaquit in the Antilles. veolapopulations in the northern Lesser Antilles Within the eastern Antilles, three geographically until recently. During an ancient expansion, the restricted groups of haplotypes were identified species may have bypassed these islands and on the basis ofmultiple mtDNA restriction-site reached Puerto Rico directly from the southern differences (fig. 4): Puerto Rico (PR); north-cen end of the archipelago. tral Lesser Antillean islands (the U.S. Virgin Is Invasiveness and Invasibility. -Our analyses lands south to St. Lucia; LA); and Grenada-St. of mtDNA variation indicated that C. jIaveola Vincent (GSV). Distance and parsimony analy is not highly vagile within the West Indies as ses (figs. 5, 6) did not clearly resolve evolutionary fixed genotypic differences often are maintained relationships among these groups, probably be between adjacent islands. Multiple fixed restric cause of the presence of homoplasious charac tion-site differences were also found between ters. continental populations separated by only a few A number of historical scenarios relating the hundred kilometers, suggesting that reduced va Antillean groups ofhaplotypes are thus possible. gility is also characteristic of continental bana The simplest begins with the spread of an an naquit populations. cestral eastern Antillean population throughout A remarkable feature ofthis data is the mtDNA the West Indies, at least as far as Puerto Rico, distance observed between bananaquits on Puer but not so far as Jamaica. The island populations to Rico and those on the U.S. Virgin Islands, in then began to diversify, producing distinctive spite of the land connection that united these haplotypes on each of the islands. Exposure of sites within the past 20,000 years. The banana- MTDNA BIOGEOGRAPHY OF CARIBBEAN BANANAQUITS 1053 quit population on St. Croix clearly belongs to a establishment of C. flaveola. Avian malaria and north-central Lesser Antillean mtDNA clade, as pox have been implicated in the extinction of apparently do those ofSt. John and S1. Thomas. many native Hawaiian birds (Warner 1968; Van Those three populations probably represent the Riper et al. 1986). Susceptibility or resistance to end pointofa wave ofexpansion ofcentral Lesser haematozoa appears to vary among species of Antillean birds through the small islands of the Lesser Antillean passerines and perhaps even northern Lesser Antilles (e.g., Nevis, St. Kitts, among island populations of the same species St-Barthelemy, St. Maarten and Anguilla). After (V. Apanius et al. unpubl. data). Cuban native having successfully colonized those islands, there species may harbor strains ofdisease organisms is no evidence of these birds continuing to col to which bananaquit colonizers are susceptible. onize across the short distance between the V.S. The possible role of disease organisms in con Virgin Islands and Puerto Rico. straining the geographic distribution of the ba To explain the apparent absence ofPuerto Ri nanaquit and other insular birds deserves atten can mtDNA haplotypes in the V.S. Virgin Is tion. lands, it is possible (e.g., Banks and Hole 1991) The geographical distribution ofmtDNA hap that dispersal within the West Indies occurs lotypes, and more generally that of the popula mostly from south and east to north and west tions ofbananaquits in the West Indies, suggest because of the prevailing winds and the general that populations may go through phases of in direction ofhurricanes. However, depending on vasiveness, as north-central Antillean birds ap the track ofthe center ofa hurricane, winds be pear to have done recently, and relative geo tween two relatively close points may blow either graphic quiescence, as evidenced by the birds of direction. Thus, we have no simple explanation Puerto Rico and those ofGrenada and St. Vin for the lack ofPuerto Rican mtDNA haplotypes cent. We concluded earlier (Seutin et al. 1993), on islands to the east and south. from an analysis ofmtDNA variability in insular The fact that Lesser Antillean bananaquits did and continental populations ofthe streaked sal not expand to Puerto Rico from the V.S. Virgin tator, thatan Antillean population ofthisspecies, Islands suggests that invasion from a small land after a long period ofisolation in the archipelago, mass to a larger one is difficult. This idea is sup has probably recently expanded its range through ported by the observation that Bahamian ba at least three of the four islands now occupied nanaquits have not invaded Cuba, even though by the taxon. Genetic analyses ofadditional in they occur on small keys to the north (e.g., Cayo sular bird species will be required to assess the Coco), and that populations in the far western generality of a pattern of alternation between Caribbean (c. f caboti) have not invaded the geographical expansion and quiescence within Yucatan Peninsula. Considering the ability of archipelagoes. Nevertheless, these data suggest bananaquits to establish populations on small that different populations of the same species islands across substantial oceanic distances, it is may be in different phases ofcolonizing activity unlikely that failure to colonize results from a at a specific time, a possibility that was not con lack ofmigrants. There have been many cases of sidered in the theoretical development of taxon vagrants reported outside the normal range (Bond cycles (Wilson 1961; Ricklefs and Cox 1972). 1963, 1979) and flying at sea. In summary, our study of mtDNA variation We believe it is unlikely that bananaquits are in Caribbean bananaquits revealed greater phy absent from Cuba and the Yucatan Peninsula logeographic structure than has been observed because ofunsuitable physical conditions or the in other avian species studied thus far. In addi presence of strong competitors. The environ tion, levels of mtDNA divergence among re ment ofeastern Cuba (Oriente Province) resem gional populations ofbananaquits were generally bles that of Hispaniola and Jamaica, where the higher than those found among conspecific tem species is abundant. Like most ofthe Caribbean perate bird populations. Hence, generalizations islands Cuba has fewer species ofpotential com concerning population genetic architecture in petitors than continental Central and South birds (i.e., low-to-moderate levels of genetic di America, over which the bananaquit is widely vergence among conspecifics and low-to-mod distributed. Within the Greater Antilles the ba erate geographic structuring of the variation nanaquit is one ofthe most abundant landbirds present) have probably been biased by an almost (e.g., Lack 1976; Cox and Ricklefs 1977). A dis exclusive focus on temperate species. Further ease organism present on Cuba may prevent the more, our studies of Caribbean birds have in- 1054 GILLES SEUTIN ET AL. dicated that taxonomic distinctions are, at best, We greatly appreciate the assistance in the field onlyloosely correlatedwithgenetic distance, sug ofW. Schew, V. Apanius, E. Benito-Espinal, C. gesting that presently recognized Antillean bird Gauveca, P. Hautcastel, S. Lanyon, P. Lau, P. subspecies should not be considered of equiva Sanchez-Baracaldo, G. Tayalay, D. Wechsler and lent phylogenetic rank in historical biogeograph J. Wunderle, and that of S. McCafferty and M. ic analyses. Walker in the laboratory. R. Zink, S. Lanyon, T. Parsons, and R. Waide generously provided samples from the collections of the Louisiana State University Museum of Natural Sciences, the Field Museum of Natural History, and the ACKNOWLEDGMENTS National Museum of Natural History. The G. Seutin was supported by a Natural Sciences manuscript benefitted from the critical reading and Engineering Research Council (Canada) and suggestions ofS. Rennerandtwo anonymous postdoctoral fellowship and R. E. Ricklefs by a reviewers. Regents' Fellowship from the SmithsonianTrop ical Research Institute. Grants from the National Geographic Society and the Smithsonian Insti LITERATURE CITED tution (Scholarly Studies Program and Abbott American Ornithologists' Union. 1983. Check-list of Fund) made the field and laboratory work pos North American Birds, 6th ed. American Orni sible. N. K. Klein collected specimens while sup thologists' Union, Lawrence, Kans. Avise, J. C. 1992. Molecular population structureand ported by a National Science Foundation Dis the biogeographic history of a regional fauna-a sertation Research Grant (BSR-881544S), and case history with lessons for conservation biology. grants from the Frank M. Chapman Memorial Oikos 63:62-76. Fund ofthe American Museum ofNatural His Avise, J. c., and R. M. Zink. 1988. Moleculargenetic divergence between avian sibling species: king and tory, Sigma Xi Scientific Research Society, Rack clapper rails, long-billed and short-billed dowitch ham Graduate School ofthe University ofMich ers, boat-tailed and great-tailed grackles, and tufted igan, the Department ofBiology at the University and black-crested titmice. Auk 105:516-528. of Michigan, and the Hinsdale and Walker Fel Avise, J. c., J. Arnold, R. M. Ball, E. Bermingham, T. lowships ofthe Museum ofZoology, University Lamb, J. E. Neigel, C. A. Reeb, and N. C. Saunders. 1987. Intraspecific phylogeny: the mitochondrial of Michigan. We are grateful to these organiza DNA bridge between population genetics and sys tions. tematics. Annual Review of Ecology and System We thank the following for granting research, atics 18:489-522. scientific collecting, and importation permits: In Avise, J. c., B. W. Bowen, T. Lamb, A. B. Meylan, and E. Bermingham. 1992. Mitochondrial DNA stituto Nacional de Recursos Naturales Renova evolution at a turtle's pace: evidence for low genetic bles (Panama); G. Canessa Mora, Ministerio de variability and reduced microevolutionary rate in Recursos Naturales, Energia, y Minas (Costa the Testudines. Molecular Biology and Evolution Rica); G. Colonello, Fundacion La Salle de Cien 9:457-473. cias Naturales, Venezuela; PROFAUNA, Minis Baker, M. C. 1981. A muscle biopsy procedure for use in electrophoretic studies ofbirds. Auk 98:392 terio del Ambiante y de los Recursos Naturales 393. (Venezuela); B. Nyack, Ministry ofAgriculture, Banks, R. c., and R. Hole, Jr. 1991. Taxonomic re Land, Forestry and Fisheries (Grenada); N. view of the mangrove cuckoo, Coccyzus minor Weekes, Department of Agriculture (St. Vin (Gmelin). Caribbean Journal ofScience 27:54-62. Bartlett, A. S., and E. S. Barghoorn. 1973. Phytoge cent); G. Charles, Department of Forestry (St. ographic history of the Isthmus ofPanama during Lucia); M. Lagarigue, Office National des Forets the past 12,000 years: a history of vegetation, cli (Martinique); F. Gregoire and C. Christian, Min mate and sea-level change, Pp. 203-299 in A. Gra istry of Agriculture (Dominica); E. Benito-Es ham, ed. Vegetation and vegetational history of pinal, Ministere charge de I'Environnement northern Latin America. Elsevier, New York. Bermingham, E. 1990. Mitochondrial DNA and the (Guadeloupe); G. A. L. Gray, Ministry ofAgri analysis of fish population structure, Pp. 197-221 culture, Trade, and the Environment (Montser in D. Whitmore, ed. Electrophoretic and isoelectric rat); A. Smith and D. Nellis, Division of Fish focusing techniques in fisheries management. CRC and Wildlife (U.S. Virgin Islands); E. Cardona, Press, Boca Raton, Fla. Bermingham, E., and J. C. Avise. 1986. Molecular Departamento de Recursos Naturales (Puerto zoogeography offreshwater fishes in the southeast Rico); Y. Strong and S. Sinha, Ministry of Ag ern United States. Genetics 113:939-965. riculture (Jamaica). Bermingham, E., S. Rohwer, S. Freeman, and C. Wood. MTDNA BIOGEOGRAPHY OF CARIBBEAN BANANAQUITS 1055

1992. Vicariance biogeography in the Pleistocene Lack, D. 1976. Island biogeography illustrated by the and speciation in North American wood warblers: land birds ofJamaica. University ofCalifornia Press, a test of Mengel's model. Proceedings of the Na Berkeley. tional Academy ofSciences, USA 89:6624-6628. Martin, A. P., G. J. P. Naylor, and S. R. Palumbi. Bond, J. 1956. Check-list ofbirds ofthe West Indies. 1992. Rates of mitochondrial DNA evolution in Academy of Natural Sciences, Philadelphia. sharks is slow compared with mammals. Nature ---. 1963. Derivation of the Antillean avifauna. 357:153-155. Proceedings of the Academy of Natural Sciences, McElroy, D., P. Moran, E. Bermingham, and I. Korn Philadelphia 115:79-98. field. 1992. REAP: the restriction enzyme analysis ---. 1979. Derivation of the Lesser Antillean birds. package. Journal of Heredity 83:157-158. Proceedings of the Academy of Natural Sciences, Nei, M. 1987. Molecular evolutionary genetics. Co Philadelphia 131:89-103. lumbia University Press, New York. Brooks, E. R., and D. A. McLennan. 1991. Phylog Nei, M., and J. C. Miller. 1990. A simple method eny, ecology,and behavior. University of Chicago for estimating average number of nucleotide sub Press, Chicago. stitutions within and between populations from re Brown, W. M. 1985. The mitochondrial genome of striction data. Genetics 125:873-879. animals, Pp. 95-130 in R. J. MacIntyre, ed. Mo Nei, M., and F. Tajima. 1983. Maximum likelihood lecular evolutionary genetics. Plenum, New York. estimation of the number of nucleotide substitu Brown, W. M., M. George, Jr., and A. C. Wilson. 1979. tions from restriction site data. Genetics 105:207 Rapid evolution of mitochondrial DNA. Proceed 217. ings ofthe National Academy ofSciences, USA 76: Paynter, R. A., Jr., ed. 1968. Check-list of the birds 1967-1971. of the world, vol. XIV. Museum of Comparative Burke, K. 1988. Tectonic evolution ofthe Caribbean. Zoology, Harvard University, Cambridge, Mass. Annual Review ofEarth and Planetary Sciences 16: Peterson, A. T., P. Escalante P., and A. Navarro S. 201-230. 1992. Genetic variation and differentiation in Capparella, A. P. 1988. Genetic variation in Neo Mexican populations ofcommon bush-tanagers and tropical birds: implications for the speciation pro chestnut-capped brush-finch. Condor 94:244-253. cess. Acta XIX Congressus Internationalis Orni Pregill, G. K. 1981. An appraisal of the vicariance thologici XIX:1658-1664. hypothesis of Caribbean biogeography and its ap Cox, G. W., and R. E. Ricklefs. 1977. Species di plication to West Indian terrestrial vertebrates. Sys versity and ecological release in Caribbean land bird tematic Zoology 30:147-155. faunas. Oikos 28:113-122. Pregill, G. K., and S. L. Olson 1981. Zoogeography Desjardins, P., and R. Morais. 1990. Sequence and ofWest Indian vertebrates in relation to Pleistocene gene organization ofthe chicken mitochondrial ge climate cycles. Annual Review ofEcology and Sys nome-a novel order in higher vertebrates. Journal tematics 12:75-98. ofMolecular Biology 212:599-634. Ricklefs, R. E., and G. W. Cox. 1972. Taxon cycle Donnelly, T. W. 1988. Geologic constraints on Ca in the West Indian avifauna. American Naturalist ribbean biogeography, Pp. 15-37 in J. K. Liebherr, 106:195-219. ed. Zoogeography of Caribbean insects. Cornell Rohlf, F. J. 1990. NTSYS-pc: numerical taxonomy University Press, Ithaca, N.Y. and multivariate analysis system. Exeter Software, Dowling, T. E., C. Moritz, and J. D. Palmer. 1990. Applied Biostatistics, Setauket, N.Y. Nucleic acids II: restriction site analysis, Pp, 250 Rosen, D. E. 1976. A vicariance model ofCaribbean 317 in D. M. Hillis and C. Moritz, ed. Molecular biogeography. Systematic Zoology 24:431-464. systematics. Sinauer, Sunderland, Mass. ---. 1978. Vicariant patterns and historical ex Escalante-Pliego, B. P. 1991. Genetic differentiation planation in biogeography. Systematic Zoology 27: in yellowthroats (Parulinae: Geothlypis). Acta XX 158-188. Congressus Internationalis Ornithologici XX:333 1985. Geological hierarchies and biogeo 343. graphic congruence in the Caribbean. Annals ofthe Faaborg, J., and J. E. Winters. 1979. Winter returns Missouri Botanical Garden 72:636-659. and longevity and weights of Puerto Rican birds. Roughgarden, J. 1990. Origin of the eastern Carib Bird-Banding 50:216-233. bean: data from reptiles and amphibians. Pp. 10 Fairbanks, R. G. 1989. A 17,OOO-yearglacio-eustatic 26 in D. Larue and G. Draper, eds. Transactions sea level record: influences ofglacial melting rates ofthe 12th Caribbean Conference, Miami Geolog on the Younger Dryas event and deep-ocean cir ical Survey, Miami, Fla. culation. Nature 342:637-642. Saitou, N., and M. Nei. 1987. The neighbor-joining Hackett, S. J., and K. V. Rosenberg. 1990. Compar method: a new method for reconstructing phylo ison of phenotypic and genetic differentiation in genetic trees. Molecular Biology and Evolution South American antwrens (Formicariidae). Auk 107: 4:406-425. 473-489. Seutin, G., B. N. White, and P. T. Boag. 1991. Pres Klein, N. K. 1992. Historical processes and the evo ervation ofavian blood and tissue samples for DNA lution ofgeographic variation patterns in the yellow analyses. Canadian Journal ofZoology 69:82-90. warbler (Dendroica petechia). Ph.D. diss. Univer Seutin, G., J. Brawn, R. E. Ricklefs, and E. Berming sity ofMichigan, Ann Arbor. ham. 1993. Genetic divergence among popula Kluge, A. G. 1988. Parsimony in vicariance bioge tions of a tropical passerine, the streaked saltator ography: a quantitative method and a Greater An (Saltator albicollis). Auk II0: 117-126. tillean example. Systematic Zoology 37:315-328. Seutin, G., L. M. Ratcliffe, and P. T. Boag. In press. 1056 GILLES SEUTIN ET AL.

Mitochondrial DNA homogeneity in the pheno Wetmore, A., R. F. Pasquier, and S. L. Olson. 1984. typically diverse redpoll finch complex (Aves; Car The birds of the Republic of Panama. Passeri duelinae; Carduelis flammea-hornemannii. Evo formes: Hirundinidae (swallows) to Fringillidae lution. In press. (finches). Smithsonian Miscellaneous Collections Shields, G. F., and A. C. Wilson. 1987. Calibration 150(4). ofmitochondrial DNA evolution in geese. Journal Wilson, E. O. 1961. The nature ofthe taxon cycle in ofMolecular Evolution 24:212-217. the Melanesian ant fauna. American Naturalist 95: Sibley, C. G., and B. L. Monroe, Jr. 1990. Distri 169-193. bution and taxonomy of birds of the world. Yale Wunderle, J. M., Jr. 1981a. Colour phases of the University Press, New Haven, Conn. bananaquit Coereba flaveola on St.Vincent, West Sneath, P. H. A. and R. R. Sokal. 1973. Numerical Indies. Ibis 123:354-358. taxonomy. W. H. Freeman, San Francisco. ---. 1981b. An analysis ofa morph ratio cline in Swofford, D. L. 1993. PAUP: phylogenetic analysis the bananaquit (Ceorebaflaveola) on Grenada, West using parsimony, Version 3.1. Illinois Natural His Indies. Evolution 35:333-344. tory Survey, Champaign, 111. ---. 1981c. Movements ofadult and juvenile ba Sykes, L. R., W. R. McCann, and A. L. Kafka. 1982. nanaquits within a morph-ratio cline. Auk 98:571 Motion of the Caribbean plate during the last 7 577. million years and implications for earlier Cenozoic ---. 1983. A shift in the morph ratio cline in the movements. Journal ofGeophysical Research 87: bananaquit on Grenada, West Indies. Condor 85: 10,656-10,676. 365-367. Van Riper, c, III., S. G. Van Riper, M. L. Goff, and ---. 1985. An ecological comparison of the avi M. Laird. 1986. The epizootiology and ecological fauna ofGrenada and Tobago, West Indies. Wilson significance ofmalaria in Hawaiian land birds. Eco Bulletin 97:356-365. logical Monographs 56:327-344. Warner, R. E. 1968. The role ofintroduced diseases in the extinction ofthe endemic Hawaiian avifauna. Condor 70:101-120. Corresponding Editor: G. Vermeij

ApPENDIX 2 Matrix of presence (I) and absence (0) of 108 restriction sites in 58 Coereba flaveola mitochondrial DNA haplotypes, Haplotype acronyms are from table 2. Sites are presented by restriction enzymes: BamH I, Bgl I, Bgi II, Cia I, Dra I, EcoR I, EcoR Y, Hind III, Nco I, Nde I, Pst I, Pvu II, Sac I, Stu I, Ava I, and Hinc II.

JA1 11100 1110000 1100 110 111000 111000 110000 1111110010 100 111100 1111 11101 110 11111111110000 11111110 100111100000000011 JA2 11100 1110000 1100 110 111000 111010 110001 1111110010 100 111100 1111 11101 110 11111111100000 11111110 100111100000000011 JA3 11100 1110000 1100 110 111000 111010 110001 1111110010 100 111100 1111 11101 110 11111011100000 11111110 100111100000010011 JA4 11100 1110000 1100 111 111000 111010 110001 1111110010 100 111100 1111 11101 110 11111111100000 11111110 100111100000010011 JA5 11100 1110000 1100 111 111000 111011 110001 1111110010 100 111100 1111 11101 110 11111111100000 11111110 100111100000010011 JA6 11100 1110000 1100 100 111000 111010 110001 1111100010 100 111100 1111 11101 110 11111111100000 11111100 100111100000010011 PR1 10010 1110000 1110 100 111000 111000 111100 1111111010 100 111110 1110 11100 110 11111110110000 11111100 111111110110000010 PR2 10010 1110000 1110 100 111000 111000 111100 1111111110 100 111110 1110 11100 110 11111110110000 11111100 111111110010000010 PR3 10010 1110000 1110 100 111000 111100 111100 1111111010 100 111110 1110 11100 110 11111110110000 11111100 111111110100000010 PR4 10010 1110000 1110 100 111000 111100 111100 1111111010 100 111110 1110 11100 110 11111110110000 11111100 111111110110000010 PR5 10010 1110000 1110 100 111000 111100 111100 1111111010 100 111110 1110 11100 110 11111110110000 11111100 111111100110000010 PR6 10010 1110000 1110 100 111000 111100 111100 1111111010 100 111110 1110 11100 110 11111110110000 11111100 111111110110100010 PR7 10010 1110000 1110 100 111000 111100 111100 1111111010 100 111110 1110 11100 110 11111110110000 11111000 111111110110000010 PRS 10010 1110000 1110 100 111000 111100 111100 1111111010 100 111110 1110 11100 110 11111111110000 11111100 111111110110000010 PR9 10010 1110000 1110 100 111000 111100 111100 1111111010 100 110110 1110 11100 110 11111111110000 11111100 111111110110000010 PR10 10010 1110000 1110 100 111000 111100 111100 1111111010 100 111110 1110 11101 110 11111111110000 11111100 111111110110000010 PR11 10010 1110000 1110 100 111000 111100 110100 1111111010 100 111110 1110 11100 110 11111111110000 11111100 111111110110000010 PR12 10010 1110000 1110 100 111010 111100 111100 1111111010 100 111110 1110 11101 110 11111111110000 11111100 111111110100000010 LA1 10010 1111000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA2 10100 1111000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA3 10100 1111000 1100 100 111100 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA4 10010 1111000 1100 100 111100 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA5 10010 1111000 1100 100 111001 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA6 10010 1111100 1100 100 111000 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA7 10010 1111100 1100 100 111000 111000 110001 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LAS 10010 1111100 1100 100 111000 111000 110001 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000110 LA9 10010 1111000 1100 100 111000 111000 110001 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA10 10010 1111100 1100 100 111000 111100 110001 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA 11 10010 1111000 1100 100 11iooo 111000 110000 1111111011 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA12 10010 1111000 1100 100 111000 111000 110000 1111111010 000 111110 1110 11100 110 11111110111000 11111100 111111110100000010 LA13 10010 1111000 1100 100 111000 111000 110000 1111111010 100 111111 1110 11100 110 11111110111000 11111100 111111110100000010 LA14 10010 1111000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11000 110 11111110111000 11111100 111111110100000010 LA15 10010 1111000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11100 100 11111110111000 11111100 111111110100000010 ApPENDIX 2. Continued.

LAI6 10010 1111000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11100 110 11111110111100 11111100 111111110100000010 LAI7 10010 1111000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 110111110100000010 LAIS 10010 1111000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11100 110 11111110 111000 11111100 111111110101000010 LAI9 10010 1111010 1100 100 111010 111000 110000 1111111010 100 111110 1110 11100 110 11111110111000 11111100 111111110100000010 GSVI 10010 1110000 1100 100 111000 111000 111000 1111111010 100 111110 1110 11100 110 11111110110000 11111100 111111110100000010 GSV2 10010 1110000 1100 100 111000 111000 111000 1111111010 100 111110 1110 11100 110 11111110110000 11111100 111111110000000010 GSV3 10010 1110000 1100 100 111000 111000 111000 1111111010 100 111110 1110 11110 110 11111110110000 11111100 111111110100000010 GSV4 10010 1110000 1100 100 111000 111000 111100 1111111010 100 111110 1110 11110 110 11111110110000 11111100 111111110100000010 GSV5 10010 1110000 1100 100 111000 111000 111000 1111111010 100 111110 1110 11110 110 11111110110000 11111100 111111100100000010 GSV6 10010 1110000 1100 100 111000 111000 111000 1111111010 100 111110 1110 11110 III 11111110110000 11111100 111111110100000010 GSV7 10010 1110000 1100 100 111000 111000 110000 1111111010 100 111110 1110 11110 III 11111110110000 11111100 111111110100000010 GSVS 10010 1110000 1100 100 111000 111000 111000 1111111010 100 111110 1110 11111 110 11111110110000 11111100 111111110100000010 VEl 10010 1111000 1100 100 111000 111000 110010 1111111010 110 111110 1100 11100 110 11111010110000 11111110 111111110000000010 VE2 10010 1111000 1100 100 111000 111000 IlOIlO 1111111010 110 111110 1100 11100 110 11111010110010 11111110 111111110000000010 VE3 10100 1111000 1100 100 111000 111000 IlOIlO 1111111010 110 111110 1100 11100 110 11111010110000 11111110 111111110000000010 VE4 10010 1111000 1100 100 111000 111000 IlOIlO 1111111010 110 111110 1100 11100 110 11111010110000 11111110 111111110000000010 VE5 10011 1111000 1100 100 111000 111000 IlOIlO 1111101010 011 111110 1100 11100 110 11111010110000 11111110 111111110000000010 VE6 10010 1111000 1100 100 111000 111000 IlOIlO 1111101010 010 111110 1100 11100 110 11111010110000 11111110 111111110000000010 VE7 10010 1111000 1100 100 111000 111000 ilOIlO 1111111010 010 111110 1000 11100 110 11111010110000 11111110 111111110000000010 VES 10010 1111001 1100 100 111000 111000 IlOIlO 1111111010 010 111110 1000 11100 110 11111010110000 11111110 111111110000000010 CAl 10100 1110000 1101 100 111000 111000 110010 1111111010 100 111110 1100 11100 110 11111010111011 11111110 111111111000000010 CA2 10100 1110000 1101 100 111000 111000 110010 ! 111111010 100 111110 1100 11100 110 11111010111010 11111110 111111111000000010 CA3 10100 1110000 1101 100 111000 111000 110010 1111111010 100 111110 1100 11110 110 11111010111010 11111110 111111111000000010 CA4 10100 1110000 1101 100 111000 111000 110010 1111111010 100 111110 1100 11100 110 11111010111010 11111100 111111100000000010 CA5 10010 1110100 1101 100 111000 111000 110010 1111111010 100 111110 1100 11100 110 11111010111010 11111111 111111110000000010