Relationships of West Indian Anolis (Sauria: Iguanidae): an Approach Using Slow-Evolving Protein Loci

Relationships of West Indian Anolis (Sauria: Iguanidae): An Approach Using Slow-Evolving Protein Loci

K RISTI L. BURNELL AND S. BLAIR H EDGES

Department of Zoology, University of Maryland, College Park, Maryland 207421

ABSTRACT . – Protein variation in 49 West Indian species of the iguanid lizard genus Anolis was examined by sequential electrophoresis at 12 slow-evolving loci. The use of this technique nearly doubled the total number of alleles detected (121 before, 233 after). Genetic distance and parsimony analyses identified intra-island radiations on Cuba, Jamaica, Hispaniola, and Puerto Rico, and found little evidence of close inter-island relationships. In most cases, these island radiations (series) defined by protein data were supported by other data (morphology, immunology, chromosomes). No support was obtained for previously defined higher-level groupings (sections and subsections) within the genus. A revised classification is proposed that recognizes 21 series of West Indian Anolis (131 spp.) with distributions centering on the following islands or island groups: Cuba (alutaceus, argillaceous, carolinensis, equestris, lucius, and sagrei), Jamaica (grahami), Hispaniola (chlorocyanus, christophei, cuvieri, cybotes, darlingtoni, distichus, hendersoni, monticola, semilineatus, and sheplani), the Puerto Rican Bank (cristatellus, occultus), the northern Lesser Antilles (bimaculatus), and the southern Lesser Antilles (roquet). No categories above the level of series are recognized due to conflicting evidence for higher-level relationships. Although the existence of Anolis on the North Island (Hispaniola) in the Eocene or Oligocene indicates an early arrival of the genus in the West Indies, molecular dating suggests that mid-Tertiary dispersal and not early-Tertiary vicariance best explains the present distribution of the group. Extensive intra-island radiation occurred during the late-Tertiary (Miocene-Present) with relatively little inter-island dispersal among the Greater Antilles.

The iguanid lizard genus Anolis (sensu The subsections are based on the shape of Williams, 1976a, b) comprises over 300 de- the interclavicle, which is arrow-shaped scribed species. In addition to being the (punctatus subsection) or T-shaped (caroli- largest amniote genus, it has figured prom- nensis subsection). Subsections contain se- inently in the ecological, ethnological, and ries that are defined, again, by osteological systematic literature. However, the phy- characters as well as by karyological and logeny of this group has been a continuous scale characters. Series are further broken challenge to systematists since the initial into subseries, species groups and work done by Etheridge (1960). Williams subgroups. Most of the systematic work on (1976a, b) published the first comprehen- Anolis has involved West Indian taxa (131 sive taxonomy of these animals. His clas- spp.). sification relied primarily on osteology and Albumin immunological data presented divided the genus into two sections, alpha by Wyles and Gorman (1980a) and Shochat and beta. These sections originally were and Dessauer (1981) did not support the established by Etheridge (1960) and are de- alpha-beta dichotomy. They found that fined osteologically by the absence (alpha) members of the grahami series (beta sec- or presence (beta) of transverse processes tion) and cristatellus series (alpha section) on posterior caudal vertebrae. Williams clustered more closely with each other than further divided the alpha section into two either did with any other series within their subsections, also osteologically defined. own sections. Shochat and Dessauer erect- ed the “Central Caribbean series complex” to include series of alpha and beta anoles 1 Current address: Department of Biological Sci- that appeared to form a monophyletic ences, University of California, Santa Barbara, Cali- fornia 93106, USA (KLB); and Department of Biology, group. 208 Mueller Lab, Penn State University, University Guyer and Savage (1986) proposed a re- Park, Pennsylvania 16802, USA (SBH). vised classification of Anolis after reana- 7 8 K. L. BURNELL AND S. B. HEDGES lyzing several studies. These studies in- locus. However, it was shown recently in cluded karyological, immunological and a study of West Indian frogs of the genus osteological data, but the final phylogeny Eleutherodactylus (Hedges, 1989) that this was based primarily on osteology. Wil- constraint can be avoided by simply using liams (1989) provides a detailed critique of only slow-evolving loci. In that case, 84 Guyer and Savage in which many serious species were examined in a single study. errors are discussed. Additional problems The specific loci to be used are chosen by with the data and methods of analysis are first running all species at a suite of loci discussed by Cannatella and de Queiroz and then choosing those that have the few- (1989). Both critiques recommended rejec- est alleles, up to a number that can be re- tion of the phylogenetic conclusions and solved accurately on a gel. Sequential elec- classification proposed by Guyer and Sav- trophoresis also can be used to reduce allelic age. Because no new data were presented convergence by detecting “hidden” vari- in Guyer and Savage, and given the serious ation (Coyne, 1982). It involves using a problems associated with their paper, their succession of electrophoretic conditions revised classification will not be used here, (usually different gel and electrode buff- and their study will not be discussed fur- ers) on the same samples. As much as 85– ther. 100% of the actual amino acid sequence Previous molecular studies of West In- variation can be detected using this meth- dian Anolis have focused primarily on rel- od (Lewontin, 1985). atively small groups of species from Puerto This study examines the phylogenetic Rico and the Lesser Antilles (Gorman and relationships of West Indian Anolis using Dessauer, 1965, 1966; Gorman and Atkins, sequential electrophoresis of slow-evolv- 1969; Yang et al., 1974; Gorman and Kim, ing loci. Forty-nine species from all four 1975, 1976; Gorman et al., 1980a, b, 1983; Greater Antillean islands, Guadeloupe in Wyles and Gorman, 1980a). Those studies the Lesser Antilles, and North America have been useful in clarifying the rela- were compared. These new molecular data tionships of eastern Caribbean Anolis. provide further insight into the relation- However, two-thirds of West Indian Anolis ships of this large genus. occur on Cuba and Hispaniola (42 and 39 species, respectively; Schwartz and Hen- M ATERIALS AND M ETHODS derson, 1988). Aside from population stud- Forty-nine species of West Indian Anolis ies of one or a few species (Webster and were collected (Appendix 1): all seven Ja- Burns, 1973; Webster, 1975; Perez-Beato and maican species, 26 of the 38 Hispaniolan Berovides, 1982; Case and Williams, 1984) species, 9 of 13 from the Puerto Rican Bank, and the immunological distance data for 1 of 20 from the Lesser Antilles, and the cybotes (Wyles and Gorman, 1980b), no mo- single North American species. Cuban lecular studies have been published con- Anolis were attainable only through Guan- cerning the phylogenetic relationships of tanamo Bay Naval Station, thus only 5 of species on these two major islands. The the 42 Cuban species were collected. Cha- only comprehensive molecular study of maelinorops barbouri from Hispaniola was West Indian Anolis relationships (Shochat included to provide a root for the parsi- and Dessauer, 1981) examined two Cuban mony trees (see below). Because electro- and two Hispaniolan species, none of phoretic differences between species and which was represented by antisera. As stat- species groups usually are complete with ed by Gorman et al. (1984), the definition no shared alleles (Avise, 1975; Gorman and of the finer divisions within Anolis (series, Renzi, 1979), only one individual per species groups) remains an unfinished species was used. Errors that might result taxonomic task. from this sampling strategy (e.g., missing Electrophoretic studies rarely involve some shared alleles) would most likely af- more than 25 species (Avise and Aquadro, fect close relationships and not the com- 1982) because of the technical aspect of position of species groups or larger clus- comparing a large number of alleles at a ters. RELATIONSHIPS OF WEST INDIAN ANOLIS 9

TABLE 1. Protein loci and electrophoretic conditions

Electrophoretic Enzyme conditions Commission Protein Locus Number a 1234 Stainc Alcohol dehydrogenase Adh 1.1.1.1 5 3 Aspartate aminotransferase Aat 2.6.1.1 5 2 Carboxylesterase-D Esd 3.1.1.1 5 1 Glucose-6-phosphate isomerase Gpi 5.3.1.9 547 2 Lactate dehydrogenase Ldh-1 1.1.1.27 31 3 Lactate dehydrogenase Ldh-2 1.1.1.27 3127 3 Lactoyl-glutathione lyase Lgl 4.4.1.5 36 1 Phosphoglucomutase Pgm 5.4.2.2 1235 2 Protein 1 Pt-1 — 432 3 Protein 2 Pt-2 — 421 3 Protein 3 Pt-3 — 46 3 Pyruvate kinase Pk 2.7.1.40 51 1 a Nomenclature Committee of the International Union of Biochemistry (1984). b (1) Tris-citrate pH 8.0, 130 v, 6 h; (2) Tris-citrate pH 6.7, 150 v, 6 h; (3) Poulik, 300 v, ca. 5.5 h; (4) Lithium hydroxide, 350 v, ca. 7 h; (5) Tris-versene-borate, 250 v, 6 h; (6) Tris-HCl, 250 v, 4 h; (7) Tris-citrate EDTA, 300 V, 6 h. c (1) Harris and Hopkinson (1976); (2) Selander et al. (1971); (3) Hedges (1986).

Lizards were transported live to the lab- Differences and similarities in electro- oratory for processing or were processed phoretic mobility were confirmed in com- in the field. Blood was collected and tissue parison runs. By alternating samples that samples (heart, liver, kidney, and leg mus- presumably represented the same allele on cle) were returned to the laboratory in liq- the same gel, very small differences in mo- uid nitrogen. Samples were prepared for bility were detected. This procedure was electrophoresis following the methods of repeated for all pairs of samples repre- Hedges (1986, 1989). Preserved voucher senting the same presumed allele. specimens (Appendix 1) were deposited in Alleles and multiple loci were ordered the United States National Museum of Nat- from cathode to anode. Alleles detected ural History (USNM) and the Museum of during the first electrophoretic run were Comparative Zoology (MCZ). assigned lower-case letters. Additional al- The 50 species of West Indian anoles were leles detected during the second, third and examined using sequential starch gel elec- fourth runs were assigned numbers, up- trophoresis of 12 slow-evolving loci. Hor- per-case letters and lower-case letters, re- izontal starch gel electrophoresis was em- spectively. Thus, subdivided alleles retain ployed using Connaught starch and sucrose their initial designation, but are further at concentrations of 12.5% and 7.5%, re- defined by additional designations (Ap- spectively. Buffers were prepared follow- pendix 2). ing the methods of Selander et al. (1971). Genetic Distance Analyses.— UPGMA phe- The primary variable chosen for sequential nogram (Sneath and Sokal, 1973) was gen- electrophoresis was buffer type, because it erated using modified Cavalli-Sforza dis- has substantial effects on mobility (Coyne, tances (Nei et al., 1983) and a distance 1982). However, not all loci were resolv- Wagner tree was produced from Cavalli- able on all buffer systems or in all taxa. Sforza and Edwards (1967) chord distances. Therefore, no more than four conditions A discussion of the preferential use of these (usually fewer) were used with each locus. distances and methods is detailed in The loci examined, electrophoretic condi- Hedges (1986). In particular, the distances tions, and stain recipes used are listed in of Cavalli-Sforza and Edwards (1967) have Table 1. optimal properties for systematic, com- 10 K. L. BURNELL AND S. B. HEDGES pared with other measures such as Rogers most in the literature (i. e., typical data sets (1972) and Nei’s (1972) distances (Rogers, in systematic) fall short of the needed 1984; Felsenstein, 1985b). BIOSYS-1 (Swof- characters for statistical significance. This ford and Selander, 1981), modified to in- study is no exception, but we believe that corporate the Cavalli-Sforza distance of Nei bootstrapping is a useful method to apply et al. (1983), was used to produce trees from in all situations, if only to show relative genetic distance data. levels of support for the different clusters Character Analyses.— Character analyses on a tree. were performed on the allelic data using PAUP (Phylogenetic Analysis Using Par- RESULTS simony; version 3.0) computer software. There were 233 alleles detected at the 12 Each locus was treated as a character and presumptive loci. The number of alleles alleles as unordered character states. Where per locus ranged from six to 33 with an heterozygotes were encountered in which average of 19.4. There were 121 alleles de- one allele was shared with other species tected before sequential electrophoresis, and the other allele was unique (2.0% of thus the use of additional buffer systems data set), the unique allele was not used in nearly doubled the total number of alleles the analysis because it did not provide in- detected. In one species (ricordii) at one lo- formation relevant to tree topology. In cases cus (Pt-1), no resolution was obtained be- where both alleles of a heterozygote were yond the first buffer system; those initial shared with two or more species, the allele alleles (Pt-1 b,i) were treated as unique in was chosen that would result in the least the analyses. Only 14 heterozygotes were amount of homoplasy. Because this special observed, accounting for 2.3% of the total coding of alleles was only rarely done (3 data set. out of 600 pairs of alleles scored), PAUP Genetic Distance Analyses.— The tree of was considered to be more appropriate for modified Cavalli-Sforza distances (Fig. 1) data analysis than the computationally-in- has a cophenetic correlation coefficient of tensive frequency parsimony program 0.81 and Prager and Wilson’s (1976) F-val- FREQPARS (Swofford and Berlocher, 1987). ue of 7.26. In general, species form intra- The global branch swapping option of island groups that are concordant with PAUP was used to find the most-parsi- morphology (taxonomy based on Williams monious tree (MPT) or trees. [1976a] and Gorman et al. [1980b, 1983], or Confidence Limits.— A bootstrapping defined herein [see below]). Examples are method (Felsenstein, 1985a) was used to representatives of the chlorocyanus, crista- obtain confidence estimates on groupings tellus, cuvieri, cybotes, distichus, grahami, hen- in the two distance trees and the character dersoni, and semlineatus series. Some pre- analysis. In each case, the loci were treated viously defined morphological groups are as characters and sampled randomly with unsupported or split into separate units. replacement to obtain individual boot- These include the monticola series (here the strapped trees. For the distance analyses, christophei and monticola series), the occultus this was accomplished with a modified ver- series (here the occultus and sheplani series), sion of BIOSYS-1, and the percentage of and the sagrei species group (here the sagrei bootstrapped trees supporting each cluster series). The Hispaniolan aquatic anole eu- were placed directly on the genetic dis- genegrahami, previously of uncertain affin- tance tree (thus preserving branch length ities, clusters with a group from the same information). In PAUP, the bootstrap op- island, the distichus series. Two species in tion generated a majority-rule consensus the carolinensis species group (here the caro- tree (Margush and McMorris, 1981) of 50 linensis series), carolinensis and porcatus, do bootstrapped trees. not cluster. As Felsenstein (1985a) pointed out, a rel- Higher-level relationships in Fig. 1 are atively large number of characters is need- less concordant with morphological data ed to obtain 95% significance for a phylog- and more concordant with geography and eny. The data set used in his example and previous molecular data (Gorman et al., RELATIONSHIPS OF WEST INDIAN ANOLIS 11

FIG. 1. Phylogenetic tree of 49 species of Anolis and Chamaelinorops barbouri constructed by UPGMA clustering of modified Cavalli-Sforza distances. Geographic abbreviations are: C, Cuba; N, Hispaniola—North Island; S, Hispaniola—South Island; J, Jamaica; L, Lesser Antilles; P, Puerto Rico; U, United States. Numbers on tree are the proportion of bootstrapped trees defining each group.

1980a, b, 1983; Shochat and Dessauer, 1981). verse morphological groupings cluster to- For example, Chamaelinorops barbouri of His- gether. paniola clusters with the cybotes series, also The distance Wagner tree (Fig. 2) has a from Hispaniola. The beta section anoles cophenetic correlation coefficient of 0.88 (grahami and sagrei series) do not form a and Prager and Wilson’s (1976) F-value of monophyletic group, and the remaining 3.07 (after branch-length optimization; anoles (alpha section) are not partitioned Swofford, 1981). In general, the groupings into the carolinensis and punctatus subsec- are similar to those in the phenogram, al- tions (Williams, 1976a). Except for sagrei, though part of the cristatellus series clusters the Cuban species representing several di- with the hendersoni series. 12 K. L. BURNELL AND S. B. HEDGES

FIG. 2. Phylogenetic tree of 49 species of Anolis and Chamaelinorops barbouri constructed by the distance Wagner method using Cavalli-Sforza and Edwards (1967) chord distance and rooted with Chamaelinorops. Numbers on tree are the proportion of bootstrapped trees defining each group. RELATIONSHIPS OF WEST INDIAN ANOLIS 13

Character Analyses.— Only three hetero- lar results (Rohlf and Wooten, 1988; Sour- zygotes (0.5% of data set) were found where dis and Nei, 1988). However, it is evident both alleles were shared with other species from these studies and others (Tateno et (Esdn/t in pulchellus and cooki and Pkc1/d in al., 1982; Nei et al., 1983; Fiala and Sokal, occultus). In these cases, the allele used in 1985; Sourdis and Krimbas, 1987; Kim and the analysis was chosen to minimize ho- Burgman, 1988) that no single method is moplasy (Esdn and Pkc1, respectively). A superior to all others in all situations. character analysis was performed on the Outgroup Rooting.— The use of Chamaeli- data set containing all 50 species and a large norops as a root in Figs. 2 and 3 is open to number (> 2000) of MPT’s was found. Each debate. It has been recognized as a separate MPT had a length of 257 and a consistency genus since its description (Schmidt, 1919), index (CI) of 0.80. Although this large and was considered by Etheridge (1960) number of MPT’s was expected due to the and Williams (1976a) to be phylogenet- small number of characters, large number ically outside of Anolis. However, albumin of unordered character states, and large immunological distances (AID’s) to some number of species, it presents a problem Hispaniolan species of Anolis suggested a in that the topology of those MPT’s is biased close relationship (Wyles and Gorman, by the initial order of species. This was 1980b). More recently, Case and Williams confirmed by comparing consensus trees (1987) have reinterpreted the immunolog- generated from different orderings of ical data of Wyles and Gorman, based on species. The bootstrap tree of the PAUP their own electrophoretic data, as evidence analysis (Fig. 3) provides a more unbiased that albumin is “slowly evolving in one or representation of the groupings defined in several of these species.” Evaluation of the the parsimony analysis. The groups de- morphological evidence pertaining to Cha- fined in this tree also show considerable maelinorops led Case and Williams to con- agreement with morphology and geog- clude that there was no evidence for a close raphy and are identical to most of the relationship with “any specific group of groups defined in the distance analyses. Anolis.” Finally, two recent cladistic anal- However, two large island radiations de- yses of morphological characters (Ether- fined in the distance analyses and previous idge and de Queiroz, 1988; Cannatella and studies, the cristatellus series (Puerto Rico), de Queiroz, 1989) came to different con- and the grahami series (Jamaica), each are clusions regarding the phylogenetic posi- broken into smaller clusters in the char- tion of Chamaelinorops. Considering all of acter parsimony analysis. Thus, the dis- the above, the relationship of Chamaelino- tance analyses, and specifically the UPGMA rops to the major anole lineages remains an tree (Fig. 1), show slightly better agree- unsettled question. ment with morphology, geography, and However, among the many alternative previous molecular and chromosomal placements of the root within Figs. 2 and studies on West Indian Anolis. 3 (e.g., midpoint, or using any combination Working with an artificially generated of series as the outgroup), none would “known” phylogeny, Sokal (1983) also change the major conclusions of this study: found that a phenetic analysis resulted in the definitions of the series. This is because a better estimate of phylogeny than a cla- the parsimony algorithm results in a single distic analysis when a relatively small network (undirected tree) which is unaf- number of characters was used. Specifical- fected by placement of the root. The to- ly, when the number of characters divided pology of the resulting tree maintains the by two times the number of taxa (minus branching pattern of the network except three) results in a value less than one (e.g., for the immediate vicinity of the root. As 0.12 in this study), then a phenogram will long as the root is not placed within a ter- provide a better estimate of the true phy- minal series, all series in Figs. 2 and 3 will logeny than a cladogram (Sokal, 1985). Re- remain defined regardless of the place- cent computer simulation studies using a ment of the root. Thus, although the dif- stochastic model of evolution found simi- ferent roots possible for Figs. 2 and 3 may 14 K. L. BURNELL AND S. B. HEDGES RELATIONSHIPS OF WEST INDIAN ANOLIS 15 affect some higher-level relationships, they idge, 1960), those two series have been would have essentially no effect on the se- grouped together in the punctatus subsec- ries defined in this study. tion (Williams, 1976a). Immunological (Wyles and Gorman, 1980b; Shochat and D ISCUSSION Dessauer, 1981) and previous allozyme data The most controversial aspect of the sys- (Gorman et al., 1983) indicated that the cy- tematic of Anolis has been the higher-level botes series is not close to the cristatellus relationships within the genus. One of the series. These electrophoretic data (Fig. 1) major conflicts concerns the monophyly of support that finding. the beta section, which includes approxi- Many of the higher-level relationships mately 120 species from Jamaica, Cuba, and defined in this study are associated with the mainland (Middle and South America). relatively low bootstrapped confidence Williams (1976a, 1989), considered it to be limits and therefore should be treated with monophyletic based on a single osteolog- caution. However, some of the patterns are ical character (transverse processes on pos- concordant with other data and worthy of terior caudal vertebrae; Etheridge, 1960). mention. Perhaps the most interesting is Albumin immunological data, however, the clustering of the Cuban series. The suggest that the Jamaican beta anoles (gra- species that form this group are morpho- hami series) are most closely related to al- logically diverse, belonging to the argilla- pha anoles (bimaculatus and cristatellus se- ceus, equestris, and sagrei series (Figs. 1 and ries) in the eastern Caribbean (Gorman et 3). This “Cuban group” is defined by Pt- al., 1980b, 1984; Shochat and Dessauer, 2 b1, which is shared by all four species and 1981). not found elsewhere. Additionally, Ldh- The results of this study provide addi- 2e3Ab clusters centralis, homolechis, and ophi- tional molecular evidence that the beta sec- olepis, and Pka further defines the subgroup tion is not monophyletic. The grahami se- of homolechis and ophiolepis. ries (beta section) clusters with alpha Another higher-level pattern involves section anoles from Hispaniola and the the cohort of series from the eastern Ca- eastern Caribbean more closely than with ribbean (bimaculatus, cristatellus, and occul- most of the Cuban alpha and beta anoles tus). These three series form a cluster that examined (Fig. 1). This lends support to in turn joins with the distichus series of Shochat and Dessauer’s (1981) “Central Ca- Hispaniola in Fig. 1 (although not in Fig. ribbean series complex,” but with the in- 3). Allele Esdn occurs in nearly all of the clusion of several series of Hispaniolan al- species of the cristatellus and distichus series pha anoles that were not examined by those and nowhere else. The single species in authors. With those Hispaniolan species, the occultus series shares Aate with mar- the complex now forms a more geograph- moratus (bimaculatus series) and evermanni ically continuous group. However, there (cristatellus series). Besides the obvious geo- are no defining alleles for this complex and graphic concordance, a close relationship thus its recognition as a monophyletic between the bimaculatus and cristatellus se- group remains to be established. ries derives support from both osteology The new protein data also address (Etheridge, 1960) and chromosomes (Gor- another controversy in the systematic of man et al., 1968, 1983; Gorman and Atkins, Anolis: the affinities of the cybotes series. 1969). Although the albumin immunolog- These Hispaniolan species are similar in ical distances between species in these two morphology (arrow-shaped interclavicle) series are relatively low, the cristatellus se- to Puerto Rican species in the cristatellus ries is closer to the Jamaican grahami series series. On the basis of osteology (Ether- based on those data (Shochat and Dessauer,

4--

FIG. 3. Phylogenetic tree of 49 species of Anolis and Chamaelinorops barbouri constructed by PAUP (parsimony analysis) and rooted with Chamaelinorops. Numbers on tree are the proportion of bootstrapped trees defining each group. Dots indicate clusters also defined in Figs. 1 or 2. 16 K. L. BURNELL AND S. B. HEDGES

1981). Immunological distances are not accord (but not complete agreement) with available for nearly all of the Cuban and the immunological data. In particular, both Hispaniolan species and therefore such re- molecular data sets indicate that the sec- lationships are difficult to assess. The tions (alpha and beta) and subsections (ca- weight of the evidence (osteology, chro- rolinensis and punctatus) are not monophy- mosomes, and electrophoresis) argues for letic. For that reason, we propose that these an eastern Caribbean clade that includes higher-level categories be abandoned. the bimaculatus, cristatellus, and distichus se- Although little support was found for ries. Whether or not the occultus series is the monophyly of Williams’ (1976a) series, part of that clade (as suggested here) will the intra-island radiations defined in this have to be addressed by additional data. study largely correspond to his species groups. We propose that all of these gen- A REVISED CLASSIFICATION erally well-supported clades of West In- The classification of West Indian Anolis dian Anolis be recognized as series. Al- has enjoyed considerable attention since though the actual category used is the initial osteological study of Etheridge somewhat arbitrary (but see Williams, (1960). Williams (1976a) proposed a de- 1976a:5-6), it is necessary that all of the tailed classification, based largely on Eth- clades be recognized at the same taxonomic eridge’s work. The finer divisions of Wil- level. This is because their relationships liams’ classification (series and levels presently are unresolved, or at best, poorly below) for the most part have not been resolved and should not be reflected in the challenged. This is unfortunate, because classification (i.e., sedis mutabilis). Some of most of the recent phylogenetic studies of the clades are large enough that the cate- West Indian Anolis have used series as the gory of series is more appropriate than unit of analysis, assuming them to be species group, allowing additional cate- monophyletic. In this study, nine of the 13 gories to be used for finer divisions within series recognized by Williams (1976a) were the clade. represented by more than one species and In this revised classification of West In- seven were found not to be monophyletic. dian Anolis (Table 2), series are listed al- The higher-level categories of Williams’ phabetically by island or island group (for classification (sections and subsections), convenience) and species described since osteologically defined, and the relation- Williams’ classification are included (no ships of his series have been contested pri- changes are proposed for the mainland se- marily by immunological data (Gorman et ries; Etheridge, 1960; Williams, 1976b). al., 1980a, b, 1983, 1984; Shochat and Des- Taxonomic and distributional data for each sauer, 1981) and partially by karyological of the species are listed in Schwartz and data (Gorman, 1973; Peccinini-Scale, 1981; Henderson (1988) and distributions of the Gorman et al., 1983). The lack of agreement series are shown in Fig. 4. Clearly-defined among these data sets is exemplified by the clades within series are treated as species almost completely unresolved consensus groups and subgroups (the category of su- tree of osteological, karyological, and im- perspecies is not used). The following dis- munological data (Cannatella and de Quei- cussion will not present morphological def- roz, 1989:Fig. 5). However, consensus initions of the series and species groups, methods may not be appropriate in this but instead will focus on the new allelic case because the lack of agreement pene- data presented in this study. In fact, it is trates even some of the terminal OTU’s (se- possible that some series defined by mo- ries). Also, the comparison is confounded lecular data never will be uniquely defined further by confusions over the composi- on morphological grounds. tion of series used by Cannatella and de Queiroz, many of which do not correspond Cuba to the series of Williams (1976a) and none The Alutaceus Series (10 spp.).—This is one of which were defined (see Williams, 1989). of three series that were unrepresented The results of this study are in better here. Garrido (1980) revised this series of RELATIONSHIPS OF WEST INDIAN ANOLIS 17 18 K. L. BURNELL AND S. B. HEDGES

1 TABLE 2. Revised classification of West Indian Anolis.

Cuba alutaceus series (C) [carolinensis * (NA)] vermiculatus group (C) alutaceus group (C) fairchildi (B) bartschi (C) alutaceus (C) longiceps (NI) vermiculatus (C) anfiloquioi (C) maynardi (CI) clivicola group (C) porcatus * (C) sagrei series (B, C, CA, CI, J, NA) clivicola (C) smaragdinus (B) allogus group (C) cyanopleurus group (C) ahli (C) cupeyalensis (C) isolepis subgroup (C) allogus (C) cyanopleurus (C) isolepis (C) homolechis group (C) fugitivus (C) angusticeps group (B, C) homolechis (C) juangundlachi (C) * jubar (C) mimus (C) angusticeps (B, C) quadriocellifer (C) spectrum group (C) guazuma (C) spectrum (C) paternus (C) imias group (C) vanidicus (C) imias (C) argillaceus series (C) equestris series (C) mestrei group (C) argillaceus (C) equestris group (C) mestrei (C) centralis* (C) baracoae (C) ophiolepis group (C) loysiana (C) equestris (C) ophiolepis * (C) pumilus (C) luteogularis (C) noblei (C) rubribarbus group (C) carolinensis series (B, C, CI, NI, smallwoodi * (C) rubribarbus (C) NA) sagrei group (B, C, CA, CI, J, carolinensis group (B, C, CI, NI, pigmaequestris group (C) NA) NA) pigmaequestris (C) bremeri (C) allisoni subgroup (B, C) delafuenti (C) allisoni (B, C) lucius series (C) luteosignifer (CI) carolinensis subgroup (B, C, lucius group (C) nelsoni (SI) CI, NI, NA) lucius (C) sagrei* (B, C, CA, CI, J, brunneus (B) argenteolus (C) NA)

Hispaniola chlorocyanus series (N, S) cybotes series (N, S) hendersoni series (S) aliniger group (N, S) armouri * (S) bahorucoensis * (S) aliniger (N, S) cybotes * (N, S) dolichocephalus * (S) * singulars (S) haetianus (S) hendersoni * (S) chlorocyanus group (N, S) longitibialis (S) * monticola series (S) chlorocyanus (N) marcanoi (N) * * koopmani (S) coelestinus (S) shrevei (N) monticola* (S) strahmi (S) christophei series (N) * rupinae (S) christophei* (N) whitemani (N, S) * semilineatus series (N, S) etheridgei (N) darlingtoni series (S) * * alumina (S) fowleri (N) darlingtoni (S) * * olssoni (N, S) insolitus (N) distichus series (B, N, S) semilineatus * (N, S) rimarum (N) altavelensis group (S) sheplani series cuvieri series (N, P, S) altavelensis (S) placidus (N) cuvieri group (P) brevirostris group (N, S) * * sheplani (S) cuvieri (P) brevirostris (N, S) ricordii group (N, S) caudalis (N, S) baleatus * (N) marron (S) barahonae * (S) websteri * (N) ricordii* (N, S) distichus group (B, N, S) roosevelti group (P) distichus* (B, N, S) roosevelti (P) eugenegrahami group (N) eugenegrahami * (N) RELATIONSHIPS OF WEST INDIAN ANOLIS 19

T ABLE 2. Continued.

Jamaica grahami series (J) grahami * (J) reconditus* (J) conspersus (CI) lineatopus* (J) valencienni* (J) garmoni * (J) opalinus* (J)

Puerto Rican Bank cristatellus series (B, P, MI) pulchellus group (P) occultus series (P) cristatellus group (B, P, MI) gundlachi * (P) occultus * (P) Cooki * (P) krugi* (P) cristatellus* (P) poncensis* (P) desechensis (P) pulchellus * (P) ernestwilliamsi (P) stratulus group (P, SC) monensis (MI) acutus (SC) scriptus (B) evermanni * (P) stratulus * (P)

Lesser Antilles bimaculatus series (L) marmoratus* (L) roquet group (L, SA) bimaculatus group (L) occulatus (L) griseus subgroup (L) bimaculatus subgroup (L) wattsi group (L) griseus (L) bimaculatus (L) forresti (L) richardi subgroup (L, SA) gingivinus subgroup (L) pogus (L) richardi (L, SA) gingivinus (L) schwartzi (L) roquet subgroup (L) nubilus (L) wattsi (L) aeneus (L) sabanus (L) roquet series (L, SA) extremus (L) leachi subgroup (L) luciae group (L, SA) roquet (L) leachi (L) [blanquillanus (SA)] trinitatus subgroup (L) marmoratus subgroup (L) [bonairensis (SA)] trinitatus (L) ferreus (L) luciae (L) lividus (L)

1 Series are grouped by island or island group in which most of the species occur (no relationships among the series are implied). Natural distributions are indicated by the following abbreviations: B = Bahamas, C = Cuba, CA = Central America, CI = Cayman Islands, J = Jamaica, L = Lesser Antilles, MI = Mona Island, N = North Island (Hispaniola), NA = North America, NI = Navassa Island, P = Puerto Rican Bank, S = South Island (Hispaniola), SA = South America and associated islands, SC = St. Croix, and SI = Swan Island. Species used in this study are indicated with an asterisk and brackets denote extralimital species.

grass anoles, which he considered a sub- species groups, and his two subgroups (an- genus, and defined four species groups. gusticeps and carolinensis) are raised to Unlike Williams (1961, 1976a), he did not species group level. Only carolinensis and include the three Hispaniolan grass anoles porcatus were examined here. Two alleles (alumina, olssoni, and semilineatus) here shared by these two species, Aatd (also placed in the semilineatus series. found in opalinus, but presumably conver- The Argillaceus Series (4 spp.).—Williams gent) and Ldh-1j likely are synapomor- (1976a) treated this series as a species group phies, but three symplesiomorphies (Lglg1, within his carolinensis series. Only one of Pgmb3b, and Adhf) force a clustering of gund- the four species (centralis) was examined, lachi and porcatus in the distance trees (Figs. and it was found to cluster with other Cu- 1, 2). The character-parsimony tree (Fig. 3) ban species (equestris and sagrei series) in correctly groups carolinensis and porcatus, all three trees (Figs. 1-3). which are closely related species (Wil- The Carolinensis Series (12 spp.).—As de- liams, 1976a). The relatively low AID (31) fined here, this series of “green anoles” is between angusticeps and carolinensis (Wyles equal to Williams' (1976a) carolinensis and Gorman, 1980a) compared with other 20 K. L. BURNELL AND S. B. HEDGES distances from carolinensis supports their fowleri a diploid chromosome number (2N placement in the same series. = 44) not found in any other species of The Equestris Series (6 spp.).—Considered Anolis examined (Webster et al., 1972; Gor- the equestris species group by Williams man, 1973; Peccinini-Seale, 1981; Williams, (1976a), this series contains the crown-giant 1989). Williams (1976a) placed etheridgei and anoles of Cuba. Only smallwoodi was ex- fowleri in the monticola series and insolitus amined here, and it was found to cluster in the darlingtoni series, but later (Williams, with species in other Cuban series of ano- 1983:353), he associated fowleri with dar- les. lingtoni. However, South Island monticola is The Lucius Series (4 spp.).—As proposed very different electrophoretically from by Williams, this series of long-limbed Cu- either etheridgei or fowleri. Likewise, South ban anoles contains two species groups (lu- Island darlingtoni is electrophoretically dis- cius and vermiculatus). None of the species tinct from insolitus. Although morpholog- was examined here. ically divergent, the species in this series The Sagrei Series (5 spp.).—This series cor- all occur on the North Island of Hispaniola responds to Williams (1976a) sagrei species and at least three (etheridgei, fowleri, and group. Of the three species examined here, insolitus) share alleles EsdO (also found in homolechis and ophiolepis form a group with sheplani but considered here to be conver- other Cuban species, and sagrei clusters with gent) and Lglb. The clustering of carolinensis Jamaican species in the grahami series. and christophei (Fig. 1) is due to the sharing However, immunological data (Shochat of a single, presumably convergent, allele and Dessauer, 1981) and additional elec- (Esdg). Otherwise, christophei belongs in the trophoretic data (Hedges and Burnell, series of North Island species to which it 1990) indicate that the grahami series is is morphologically and geographically as- monophyletic and does not include sagrei. sociated (christophei also shares allele Ldh-1b Also, sagrei shares GpibC with ophiolepis sug- with fowleri). Although not examined, the gesting that it is misplaced on the tree. North Island species rimarum, placed in the etheridgei subgroup by Williams (1976a), is assumed to be a member of this series. Hispaniola The Cuvieri Series (5 spp.).—This series of The Chlorocyanus Series (4 spp.).—This crown-giant anoles consists of five species morphologically well-defined series of from Hispaniola and the Puerto Rican Bank. Hispaniolan green anoles consists of four Only the three Hispaniolan species (bale- species (Williams, 1965, 1976a) of which atus, barahonae, and ricordii), morphologi- three were examined in this study: chlo- cally a well-defined group (Williams, 1962), rocyanus, coelestinus, and singulars. The pro- were examined here. They share alleles tein data (allele Esdf; coelestinus has auta- Adhh (also shared by etheridgei), Gpii2B (ex- pomorphic allele Esdm ) also support the cept ricordii), Ldh-2i2 (except barahonae) and monophyly of this group. In addition, the Pt-2d (except baleatus). In addition, alleles chlorocyanus group (considered a superspe- Esdc and Lgla are found only in these species. cies by Williams, 1976a) of coelestinus (South Although Williams (1976a) placed one Island) and chlorocyanus (North Island) is of the two Puerto Rican Bank species (cu- supported both by protein data (Pt-3 hl [al- vieri) in its own group and the other (roose- though shared convergently with christo- velti) with the three Hispaniolan species in phei] and Ldh-2bA) and morphology (Wil- the ricordii group, no corroborative data liams, 1965). were presented. That arrangement likely The Christophei Series (5 spp.).—This com- is based on differences in the inscriptional plex of North Island anoles is morpholog- rib formulae noted by Etheridge (1960). ically diverse. Schwartz’s (1973) original However, some features of scalation (scales description of fowleri suggested a relation- across snout, lamellae under 4th toe) listed ship with etheridgei based on blue eyes and by Williams (1962:Table 2) would appear crossbanding. Morphologically, insolitus is to unite the two Puerto Rican Bank species a twig anole which does not resemble either in a more geographically plausible group. of the other species, however, it shares with Here, we take a conservative position and RELATIONSHIPS OF WEST INDIAN ANOLIS 21

place roosevelti in its own group without of its unique lifestyle and morphology, the reflecting relationships of the three species affinity of eugenegrahami has remained a groups until more data become available. question since its discovery (Schwartz, The Cybotes Series (8 spp.).—This complex 1978). Based on osteology, Williams (in of eight largely allopatric Hispaniolan Schwartz, 1978) considered it to be in its species of the trunk/ground ecomorph was own subseries and species group within placed in the cybotes subseries of the bi- the bimaculatus series of the alpha section. maculutus series by Williams (1976a), but Upon reanalysis of the osteological data, was raised to full series by Gorman et al. Williams (1989) recently revised his as- (1980b) based on albumin immunological sessment of eugenegrahami and placed it in data (Shochat and Dessauer, 1981; Wyles the carolinensis subsection. The protein data and Gorman, 1980a, b). The four species (allele Adhi is unique to the four species examined in this study (armouri, cybotes, examined), however, associate it most shrevei, and whitemani) form a group in all closely with the distichoids. This arrange- three trees (Figs. 1-3). This series does not ment is supported by the similar habitus show affinities with any of the species of of those species: long limbs and a relatively the old bimaculatus series in Figs. 1 and 2, short snout. although it clusters with sheplani and the The Hendersoni Series (3 spp.).—This com- distichus series in Fig. 3. Alleles Pgmd3 and plex of three long-snouted South Island Esdd were found only in these four species bush anoles is well-defined morphologi- of the cybotes series. cally (Williams, 1963; Schwartz, 1978) and The Darlingtoni Series (1 sp.).—As pro- electrophoretically (Ldh-2e2). Additionally, posed by Williams, this series included two the cluster of dolicocephalus and hendersoni twig anoles, darlingtoni and insolitus. The is defined by allele Adhj (bahorucoensis has electrophoretic data indicate that the two the primitive allele Adhf). species are morphologically convergent, The Monticola Series (3 spp.).—As pro- and that the North Island insolitus is related posed by Williams, this series included four to other North Island species (christophei North Island species (christophei, etheridgei, series). The South Island darlingtoni is not fowleri, and rimarum) here placed in the close to any other species and therefore is christophei series. As defined here, the mon- placed in its own series. ticola series includes only the three South The Distichus Series (7 spp.).—This com- Island species that formed Williams’ mon- plex of Hispaniolan species, except for eu- ticola subgroup. Although only one of the genegrahami, initially was placed in the acu- three species (monticola) was examined in tus series by Gorman and Atkins (1969). this study, the three form a morphologi- Williams (1976a) later placed the complex, cally and chromosomally well-defined along with two Puerto Rican species (ev- clade (Webster et al., 1972; Williams and ermanni and stratulus), in the stratulus sub- Webster, 1974). Phonetically (Fig. 1), mon- series of the bimaculatus series. Based on an ticola is the most divergent species of Anolis electrophoretic study, Gorman et al. (1980a, examined. However, the parsimony trees 1983) placed the distichoids in the crista- place it with either the Puerto Rican species tellus subseries of the cristatellus series. The (Fig. 2) or the South Island species of the only other molecular data for this group hendersoni series (Fig. 3). In the latter case, (Shochat and Dessauer, 1981) showed it to this is the result of sharing Aatj with doli- be equally close to both the bimaculatus and cocephalus and Esdr with dolicocephalus, ba- grahami series. The electrophoretic data now horucoensis, and darlingtoni. This suggests a warrant the recognition of the distichus possible relationship with these two South complex as a separate series of Hispaniolan Island series (hendersoni and darlingtoni). species with possible affinities to the bi- The Semilineatus Series (3 spp.).—This se- maculatus and cristatellus series (Fig. 1). ries of Hispaniolan grass anoles consists of An important addition to this series is three species (Williams, 1976a) of which the species eugenegrahami. This species is two were examined in this study: olssoni one of two aquatic West Indian anoles (the and semilineatus. Both species cluster in Figs. other is the Cuban vermiculatus). Because 1 and 2 because they have a low genetic 22 K. L. BURNELL AND S. B. HEDGES distance, but do not cluster in Fig. 3 because they have no derived alleles in com- The Puerto Rican Bank mon. According to Williams (1961, 1976a), The Cristatellus Series (13 spp.).—Most of they are related to a large group of Cuban the Puerto Rican Bank Anolis and one grass anoles, here considered the alutaceus species each from Mona Island and the Ba- series, of which there were no other rep- hamas are believed to form a single radia- resentatives in this study. tion (Williams, 1972) and have been placed The Sheplani Series (2 spp.).—A closely- in the cristatellus subseries of the cristatellus related North Island/South Island pair of series (Williams, 1976a). Chromosomal twig anoles are placed in this series. Al- (Gorman et al., 1968; Gorman and Atkins, though only sheplani was examined here, 1969; Gorman et al., 1983), immunological it was compared with placidus in another (Wyles and Gorman, 1980a; Shochat and study using 38 protein loci and found to Dessauer, 1981), and electrophoretic (Gor- have a low genetic distance typical of man et al., 1983) data have added support closely related species (Hedges and Thom- for this clade, here treated as a series. In as, 1989). Williams (1976a) originally placed this study, six of the eight species from the sheplani with occultus in the occultus series, cristatellus series formed a group (Fig. 1): but recently (Williams, 1989) considered cooki, cristatellus, krugi, poncensis, pulchellus, them to be quite dissimilar. In this study, and stratulus. Although Esdn occurs only in sheplani also was not found to be close to these species, evermanni, and three species occultus or to any other species of Anolis. of the distichus series, there are no defining For that reason, it is placed in its own series alleles for the cristatellus series. The species with placidus. groups recognized here are based largely on the results of Gorman et al.’s (1983) study Jamaica using karyotypic and electrophoretic data. The Grahami Series (7 spp.).—This series The Occultus Series (1 sp.).—Immunolog- is clearly-defined chromosomally (Gor- ical data (Shochat and Dessauer, 1981) sug- man, 1973), immunologically (Shochat and gested that the Puerto Rican twig anole Dessauer, 1981) and geographically (all occultus is not close to any particular lin- species occur on Jamaica and the Cayman eage of West Indian Anolis, although the Islands). The Cuban species sagrei (sagrei lowest distance reported (36; Wyles and series) occurs on Jamaica but presumably Gorman, 1980a) is to a Puerto Rican species, it was introduced based on recent range cuvieri (cuvieri series). Also, its primitive expansion in Jamaica (Underwood and karyotype would appear to exclude it from Williams, 1959; Williams, 1969). It was the chromosomally well-defined Puerto found to be outside of the Jamaican radia- Rican Bank radiation, the cristatellus series tion in a more detailed electrophoretic (Gorman and Atkins, 1969). Morphologi- study of the Jamaican species using addi- cally, it resembles the Hispaniolan twig tional loci and individuals (Hedges and anole sheplani in some respects, leading Burnell, 1990). The inclusion of valencienni Williams (1976a) to place the two together in the grahami series is in agreement with in the occultus series. However, his position chromosomal (Gorman, 1973) and immu- recently has been revised in favor of con- nological (Shochat and Dessauer, 1981 ) data vergence for those similarities (Williams, and indicates that it is morphologically convergent with other West Indian twig 1989). anoles (Williams, 1983; Hedges and Thom- In this study, occultus was found to clus- as, 1989). Three alleles define this series: ter with species in the bimaculatus series Esd1 is shared by all members except linea- (marmoratus) and cristatellus series (ever- e topus and reconditus, GpibA is common to all manni) due to a single shared allele (Aat ). members except lineatopus and valencienni This suggests a possible affinity with those (which both share allele GpibB), and Ldh- two geographically proximal series in 2 e4b is shared by all members except gra- Puerto Rico and the northern Lesser An- hami. tilles. RELATIONSHIPS OF WEST INDIAN ANOLIS 23

Buskirk, 1985; Williams, 1989). However, The Northern Lesser Antilles anything more than conjecture is difficult The Bimaculatus Series (13 spp.).—The im- because the geologic history still is incom- munological data of Shochat and Dessauer pletely known, the West Indian Tertiary (1981) provide strong evidence for the vertebrate fossil record is poor, and we monophyly of this series, which occupies know very little about the phylogenetic the islands north of Martinique. The species relationships of most groups (Hedges, 1989; groups and subgroups listed in Table 2 cor- Perfit and Williams, 1989; Williams, 1989). respond to the results of an electrophoretic The biogeographic history of Anolis is as analysis of this series (Gorman and Kim, controversial as its relationships. Previous 1976:Fig. 2). The single species examined discussions based on morphology have here, marmoratus, showed affinities with stressed inter-island dispersal while at the cristatellus and occultus series of the adja- same time have acknowledged intra-island cent Puerto Rican Bank. radiation and convergence (Williams, 1969, 1972, 1983). The results of this study (Fig. The Southern Lesser Antilles 1) support previous molecular studies of The Roquet Series (7 spp.).—This series of Anolis (Gorman et al., 1980a, b, 1983, 1984; anoles appears to be distantly related to Shochat and Dessauer, 1981) in identifying other West Indian species and has affinities a general pattern of agreement between with South American taxa (Williams, 1976b; phylogeny and geography. A similar pat- Gorman et al., 1980a, b, 1983; Shochat and tern was found in an equally diverse West Dessauer, 1981). The species groups and Indian group, frogs of the genus Eleuthero- subgroups listed here are based on the re- dactylus (Hedges, 1989). This pattern sug- sults of electrophoretic (Yang et al., 1974: gests that inter-island dispersal was less Fig. 3) and immunological (Shochat and frequent in the past than was previously Dessauer, 1981) data. No species from this believed. series was examined here. An important aspect of any biogeographic analysis is the establishment of a BIOGEOGRAPHY time frame. The only pre-Quarternary West The complex geologic history of the Ca- Indian Anolis fossils are from Eocene or ribbean region undoubtedly has influ- Oligocene amber deposits on the North Is- enced different Antillean groups in differ- land of Hispaniola (Rieppel, 1980; new dat- ent ways. Current models support ing in Lambert et al., 1985). Unfortunately, considerable plate tectonic movement of they do not provide the crucial informa- the islands, with a proto-Antilles situated tion on times of divergence. For that, we in the position of present-day Middle must turn to the only other source avail- America in the late Cretaceus (Pindell and able for dating: albumin immunological Dewey, 1982; Sykes et al., 1982; Mann and distances (Wyles and Gorman, 1980a; Sho- Burke, 1984; Ross and Scotese, 1988; Perfit chat and Dessauer, 1981; Gorman et al., and Williams, 1989). The vicariance model 1984). Using the immunological clock (1 suggests that the proto-Antillean biota, AID = 0.60 million years) calibrated for continuous with that of North and South diverse groups such as iguanid lizards America, fragmented along with the is- (Gorman et al., 1971), frogs (Maxson et al., lands during the Tertiary (Rosen, 1976, 1975), and mammals (Wilson et al., 1977), 1978, 1985). Before the wide acceptance of these data allow us to make some biogeo- plate tectonics, dispersal was the primary graphic inferences. model of West Indian biogeography (Arl- The break-up of the proto-Antilles oc- ington, 1938, 1957; Simpson, 1956; Wil- curred about 60–65 mya (Pindell and Dew- liams, 1969), and this view has been held ey, 1982; Duncan and Hargraves, 1984), by some recent authors (Pregill, 1981; which corresponds to an immunological Briggs, 1984). Others have combined dis- distance of about 100. The highest AID’s persal and vicariance scenarios (Mac- reported within Anolis (81–82) are between Fadden, 1980, 1981; Hedges, 1982, 1989; Middle American species (Gorman et al., 24 K. L. BURNELL AND S. B. HEDGES

1984). Among Antillean species, and be- The radiations of Anolis on the Puerto tween Antillean and Middle American Rican Bank (cristatellus series) and the species, AID’s range from 0-67 with many northern Lesser Antilles (bimaculatus se- inter-island distances less than 30 (Wyles ries) are relatively recent ( <10 my; Shochat and Gorman, 1980a; Shochat and Dessauer, and Dessauer, 1981:Table 1). Lesser Antil- 1981; Gorman et al., 1984). Unless the rate lean Anolis on islands south of Dominica of albumin evolution in AnoliS is greatly (roquet series) have affinities with groups reduced relative to other vertebrate groups in South America (Gorman et al., 1980b). studied (Wilson et al., 1977; Wyles and Gor- Thus, it appears that the Anolis of the man, 1980a), this indicates that dispersal, northern Lesser Antilles originated by dis- and not vicariance, best explains the bio- persal from Puerto Rico or Hispaniola and geographic history of West Indian Anolis. not from South America or the southern The immunological data suggest a scenario Lesser Antilles. Also, the affinities of the involving inter-island dispersal in the mid- distichus series (Hispaniola) with the bi- Tertiary followed by late-Miocene to Pres- maculatus and cristatellus series suggests ent intra-island speciation. Miocene dispersal from Hispaniola to Jamaica apparently was completely sub- Puerto Rico. The North Island and Puerto merged during the Oligocene, and has been Rico have been closely associated through- above water only for about the last 20 mil- out the Tertiary (Pindell and Dewey, 1982; lion years (references in Buskirk [1985] and Mann and Burke, 1984; Pindell and Barrett, Hedges [1989]). Thus, the immunological 1989), which probably explains these con- dating of the Jamaican radiation (<10 my; cordant patterns of inter-island relation- Shochat and Dessauer, 1981:Table 1) is in ships. agreement with that geological constraint. The Bahamas are relatively flat lime- Hispaniolan Anolis do not exhibit the stone islands that have changed greatly in major North Island/South Island dichot- size with fluctuating sea levels of the Plio- omy found in Eleutherodactylus (Hedges, cene and Pleistocene. There are only four 1989). Although the christophei series (North endemic Bahamian Anolis: three in the ca- Island) and the hendersoni and monticola se- rolinensis series and one in the cristatellus ries (South Island) follow that pattern, the series. There is little doubt that dispersal remaining five series contain species from from Cuba (in the former case) and dis- both regions. This is not unexpected, how- persal from Puerto Rico (in the latter case) ever, because most presumably are recent led to the origin of these species. Thus, it radiations post-dating the joining of the would appear that the large Bahamas Plat- North and South Islands and contain form, affixed to the North American plate closely related species. This especially is since its origin in the Mesozoic (Dietz et evident in the allopatric distributions and al., 1970), has played only a minor role in high degree of morphological similarity the biogeographic history of West Indian seen in species of the cuvieri, cybotes, disti- Anolis. chus, hendersoni, and sheplani series. The Any comprehensive synthesis of Carib- North Island/South Island sister species of bean biogeography (e.g., Rosen, 1976, 1985; the chlorocyanus series (Williams, 1965) and Buskirk, 1985) is only as strong as the data sheplani series (Hedges and Thomas, 1989) or phylogenies used. Few West Indian provide a further example of post-colli- groups have been as well studied as Anolis sional dispersal and divergence probably associated with sea level and climatic fluc- and Eleutherodactylus yet many aspects of tuations in the late Pliocene and Pleisto- the relationships of those two groups still cene (Pregill and Olson, 1981). The con- are unclear. The reanalysis or synthesis of siderable morphological differences and previously published data sets may glean sympatry of species in the christophei series, new insights, but major advances in our most restricted to montane areas in the understanding of biogeography in this Cordillera Central of the North Island, complex region likely will come from the suggest that it is an older radiation. accumulation of new data. RELATIONSHIPS OF WEST INDIAN ANOLIS 25

Acknowledgments.— Richard Thomas protein variation. Isozymes: Current Topics in Bi- provided invaluable assistance and guid- ological and Medical Research 6:1-32. Darlington, P. J. 1938. The origin of the fauna of ance in the field aspect of this study. Others the Greater Antilles, with discussion of dispersal who generously assisted SBH in collecting of animals over water and through the air. Quart. animals were: M. Coggiano, D. Hardy, C. Rev. Biol. 13:274-300. A. Hass, R. Highton, M. Londner, C. May- ———. 1957. Zoogeography: the geographical dis- er, J. Pinero, K. and S. Schindler, M. and tribution of animals. John Wiley and Sons, New York. 675 pp. W. Stephenson, L. White, and G. Zustack. Dietz, R. S., J. C. Holden, and W. P. Sproll. 1970. Collecting permits and general support Geotectonic evolution and subsidence of Bahama were received from P. Fairbairn and A. Platform. Geol. Soc. Amer. Bull. 81:1915-1928. Haynes (Jamaica), G. Hermatin, E. Magny, Duncan, R. A., and R. B. Hargraves. 1984. Plate tectonic evolution of the Caribbean region in the P. Paryski, R. Pierre-Louis, and F. Sergile mantle reference frame. Geol. Soc. Amer. Mem. (Haiti), S. and Y. Inchaustegui (Dominican 162:81-93. Republic), E. Cardona (Puerto Rico) and J. Etheridge, R. 1960. The relationships of the anoles Fifi (Guadeloupe). H. Dowling provided (Reptilia: Sauria: Iguanidae): an interpretation based on skeletal morphology. Ph.D. Thesis, Uni- use of the Dowling House, Jamaica. BIO- versity of Michigan, Ann Arbor. SYS-1 and PAUP computer programs were ———, and K. de Queiroz, 1988. A phylogeny of generously made available by D. Swofford. the Iguanidae. In R. Estes and G. K. Pregill (eds.), D. Cannatella, S. Case, L. Chao, R. Ether- Phylogenetic relationships of the lizard families, idge, O. Garrido, C. A. Hass, R. Highton, pp. 283-368. Stanford University Press, Palo Alto. Felsenstein, J. 1985a. Confidence limits on phylog- G. Mayer, A. Schwartz, G. Wilkinson, and enies: an approach using the bootstrap. Evolution E. Williams offered helpful comments on 39:783-791. earlier drafts of the manuscript. Eileen Bar- ———. 1985b. Phylogenies from gene frequencies: nett assisted with word processing. Finan- a statistical problem. Syst. Zool. 34:300-311. Fiala, K. L., and R. R. Sokal. 1985. Factors determin- cial support was provided by The Univer- ing the accuracy of cladogram estimation: evalu- sity of Maryland Computer Science Center ation using computer simulation. Evolution 39: and the National Science Foundation 609-622. (grants BSR 83-07115 to Richard Highton Garrido, O. H. 1980. Revision del complejo Anolis alutaceus (Lacertilia: Iguanidae) y descripcion de and 89-06325 to SBH). una nueva especie de Cuba. Poeyana 201:1-41. Gorman, G. C. 1973. The chromosomes of the Rep- tilia, a cytotaxonomic interpretation. In A. B. Chia- relli and E. Capenna (eds.), Cytotaxonomy and LITERATURE CITED vertebrate evolution, pp. 349-424. Academic Press, Avise, J. C. 1975. Systematic value of electrophoretic London. data. Syst. Zool. 23:465-481. ———, and L. Atkins. 1969. The zoogeography of ———, and C. Aquadro. 1982. A comparative sum- Lesser Antillean Anolis lizards—an analysis based mary of genetic distance in the vertebrates. Evol. upon chromosomes and lactic dehydrogenase. Bull. Biol. 15:151-185. Mus. Comp. Zool. 138:53-80. Briggs, J. C. 1984. Freshwater fishes and biogeog- ———, D. G. Buth, M. Soule, and S. Y. Yang. 1980a. raphy of Central America and the Antilles. Syst. The relationships of the Anolis cristatellus species Zool. 33:428-434. group: electrophoretic analysis. J. Herpetol. 14:269- Buskirk, R. 1985. Zoogeographic patterns and tec- 278. tonic history of Jamaica and the northern Carib- ———, ———, ——— and ———. 1983. The rela- bean. J. Biogeogr. 12:445-461. tionships of the Puerto Rican Anolis: electropho- Cannatella, D. C., and K. de Queiroz. 1989. Phylo- retic and karyotypic studies. In A. G. J. Rhodin and genetic systematic of the anoles: is a new tax- K. Miyata (eds.), Advances in herpetology and onomy warranted? Syst. Zool. 38:57-69. evolutionary biology, pp. 626-642. Museum of Case, S. M., and E. E. Williams. 1984. Study of a Comparative Zoology, Harvard University, Cam- contact zone in the Anolis distichus complex in the bridge. central Dominican Republic. Herpetologica 40:118- ———, ———, and J. S. Wyles. 1980b. Anolis lizards 137. of the eastern Caribbean: a case study of evolution. ———. 1987. The cybotoid anoles and Chamaelino- III. A cladistic analysis of albumin immunological rops lizards (Reptilia: Iguanidae): evidence of mo- data, and the definition of species groups. Syst. saic evolution. Zool. J. Linn. Soc. 91:325-341. Zool.29:143-158. Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. ———, and H. C. Dessauer. 1965. Hemoglobin and Phylogenetic analysis: models and estimation pro- transferring electrophoresis and relationships of is- cedures. Evolution 21:550-570. land populations of Anolis lizards. Science 150: Coyne, J. A. 1982. Gel electrophoresis and cryptic 1454-1455. 26 K. L. BURNELL AND S. B. HEDGES

———, and ———. 1966. The relationships of Anolis ———. 1981. Comments on Pregill’s appraisal of of the roquet species group (Sauria: Iguanidae)—I. historical biogeography of Caribbean vertebrates: Electrophoretic comparison of blood proteins. vicariance, dispersal, or both? Syst. Zool. 30:370- Comp. Biochem. Physiol. 19:845-853. 372. ———, and Y. J. Kim. 1975. Genetic variation and Mann, P., and K. Burke. 1984. Neotectonics of the genetic distance among populations of Anolis liz- Caribbean. Rev. Geophys, Space Phys. 22:309-362. ards on two Lesser Antillean island banks. Syst. Margush, T., and F. R. McMorris. 1981. Consensus Zool. 24:369-373. n-trees. Bull. Math. Biol. 43:239-244. ———, and ———. 1976. Anolis lizards of the east- Maxson, L. R., V. M. Sarich, and A. C. Wilson. 1975, ern Caribbean: a case study in evolution. II. Ge- Continental drift and the use of albumin as an netic relationships and genetic variation in the evolutionary clock. Science 255:397-400. bimaculatus group. Syst. Zool. 25:62-77. Nei, M. 1972. Genetic distance between populations, ———, C. S. Lieb, and R. H. Harwood. 1984. The Amer. Nat. 106:283-292. relationships of Anolis gadovi: albumin immuno- ———, F. Tajima, and Y. Tateno. 1983. Accuracy of logical evidence. Carib. J. Sci. 20:145-152. estimated phylogenetic trees from molecular data. ———, and J. Renzi. 1979. Genetic distance and het- J. Mol. Evol. 19:153-170. erozygosity estimates in electrophoretic studies: Nomenclature Committee of the International Union effects of sample size. Copeia 1979:242-249. of Biochemistry. 1984. Enzyme Nomenclature ———, R. Thomas, and L. Atkins. 1968. Intra- and 1984. Academic Press, New York. interspecific chromosome variation in the lizard Peccinini-Seale, D. 1981, New developments in ver- Anolis cristatellus and its closest relatives. Breviora tebrate cytotaxonomy IV. Cytogenetic studies in (Mus. Comp. Zool.) 293:1-7. reptiles. Genetica 56:123-148. ———, A. C. Wilson, and M. Nakanishi. 1971. A Perez-Beato, O., and V. Berovides. 1982. Estudio biochemical approach towards the study of rep- electroforetico de las proteinas del plasma en dos tilian phylogeny: evolution of serum albumin and especies de Anolis (Sauria: Iguanidae). Poeyana 242: lactate dehydrogenase. Syst. Zool. 20:167-185. 1-7. Guyer, C., and J. M. Savage. 1986. Cladistic rela- Perfit, M., and E. E. Williams. 1989. Geological con- tionships among anoles (Sauria: Iguanidae). Syst. straints and biological retrodictions in the evo- Zool. 35:509-531. lution of the Caribbean sea and its islands. In C. Harris, S. B., and D. A. Hopkinson. 1976. Handbook A. Woods (ed.), Biogeography of the West Indies: of enzyme electrophoresis in human genetics. past, present, and future. E. J. Brill, The Nether- North Holland Publishing, Amsterdam, The lands. In press. Netherlands. Pindell, J. L., and Barrett, S. F. 1989. Geologic evo- Hedges, S. B. 1982. Caribbean biogeography: im- lution of the Caribbean: a plate tectonic perspec- plications of recent plate tectonic studies. Syst. tive. In J. E. Case and G. Dengo (eds.), The geology Zool. 31:518-522. of North America. Vol. H, The Caribbean Region. ———. 1986. An electrophoretic analysis of Hol- Geological Society of America, Boulder, Colorado. arctic hylid frog evolution. Syst. Zool. 35:1-21. ———, and J, F. Dewey, 1982. Permo-Triassic re- ———. 1989. Evolution and biogeography of West construction of Western Pangaea and the evolu- Indian frogs of the genus Eleutherodactylus: slow- tion of the Gulf of Mexico/Caribbean region. Tec- evolving loci and the major groups. In C. A. Woods tonics 1:179-211. (ed.), Biogeography of the Caribbean: past, pres- Prager, E. M., and A. C. Wilson. 1976, Congruency ent, and future, pp. 305-370. Sandhill Crane Press, of phylogenies derived from different proteins. A Gainesville, Florida. molecular analysis of the phylogenetic position of ——— , and K. L. Burnell. 1990. The Jamaican ra- cracid birds. J. Mol. Evol. 9:45-57. diation of Anolis (Sauria: Iguanidae): an analysis Pregill, G. K, 1981. An appraisal of the vicariance of relationships and biogeography using sequen- hypothesis of Caribbean biogeography and its ap- tial electrophoresis. Carib. J. Sci. 26:31-44. plication to West Indian terrestrial vertebrates, Syst. ———, and R. Thomas. 1989. A new species of Anolis Zool. 30:147-155. (Sauria: Iguanidae) from the Sierra de Neiba, His- paniola. Herpetologica 45(3):330-336. ———, and S. L. Olson. 1981. Zoogeography of West Kim, J., and M. A. Burgman. 1988. Accuracy of phy- Indian vertebrates in relation to Pleistocene climatic cycles. Annu. Rev. Ecol. Syst. 12:75-98. logenetic estimation methods under unequal evolutionary rates. Evolution 42:596-602. Rieppel, O. 1980. Green anole in Dominican amber. Lambert, J. B., J. S. Frye, and G. O. Poinar, Jr, 1985. Nature 286:486-487. Amber from the Dominican Republic: analysis by Rogers, J. S. 1972. Measures of genetic similarity and nuclear magnetic resonance spectroscopy. Ar- genetic distance. Studies Genet. VII. Univ. Texas chaeometry 27(1):43-51. Publ., 7213:145-153, Lewontin, R. 1985. Population genetics. Ann. Rev. ———. 1984. Deriving phylogenetic trees from al- Genet. 19:81-102. lele frequencies. Syst. Zool. 33:52-63. MacFadden, B. 1980. Rafting mammals or drifting Rohlf, F. J., and M. C. Wooten. 1988. Evaluation of islands?: biogeography of the Greater Antillean restricted maximum-likelihood method for esti- insectivores Nesophontes and Solenodon. J. Bio- mating phylogenetic trees using simulated allele- geogr. 7:11-22. frequency data. Evolution 42:581-595. RELATIONSHIPS OF WEST INDIAN ANOLIS 27

Rosen, D. E. 1976. A vicariance model of Caribbean years and implications for earlier Cenozoic move- biogeography. Syst. Zool. 24:431-464. ments. J. Geophys. Res. 87:10656-10676. ———. 1978. Vicariant patterns and historical ex- Tateno, Y., M. Nei, and F. Tajima. 1982. Accuracy of planation in biogeography. Syst. Zool. 27:159-188. estimated phylogenetic trees from molecular data. ———. 1985. Geological hierarchies and biogeo- I. Distantly related species. J. Mol. Evol. 18:387- graphic congruence in the Caribbean. Ann. Mo. 404. Bot. Gard. 72:636-659. Underwood, G., and E. E. Williams. 1959. The ano- Ross, M. I., and C. R. Scotese. 1988. A hierarchical line lizards of Jamaica. Bull. Instit. Jamaica, Sci- model of the Gulf of Mexico and Caribbean region. ence Series 9:1-48. Tectonophysics 155:139-168. Webster, T. P. 1975. An electrophoretic comparison Schmidt, K. P. 1919. Descriptions of new amphibi- of the Hispaniolan lizards Anolis cybotes and A. ans and reptiles from Santo Domingo and Navas- marconoi. Breviora 431:1-8. sa. Bull. Amer. Mus. Nat. Hist. 41:519-525. ———, and J. Burns. 1973. Dewlap color variation Schwartz, A. 1973. A new species of montane Anolis and electrophoretically detected sibling species in (Sauria: Iguanidae) from Hispaniola. Ann. Car- a Haitian lizard Anolis brevirostris. Evolution 27: negie Mus. 44:183-195. 368-377. ———. 1978. A new species of aquatic Anolis (Sauria: ———, W. P. Hall, and E. E. Williams. 1972. Fission Iguanidae) from Hispaniola. Ann. Carnegie Mus. in the evolution of a lizard karyotype. Science 177: 47:261-279. 611-613. ———, and R. W. Henderson. 1988. West Indian Williams, E. E. 1961. Notes on Hispaniolan herpe- amphibians and reptiles: a checklist. Milwaukee tology. 3. The evolution and relationships of the Publ. Mus. Contrib. Biol. Geol. 74:1-264. Anolis semilineatus group. Breviora 136:1-8. Selander, R. K., M. H. Smith, S. Y. Yang, W. E. John- ———. 1962. Notes on Hispaniolan herpetology. 6. son, and J. B. Gentry. 1971. Biochemical poly- The giant anoles. Breviora 155:1-15. morphism and systematic in the genus Peromys- ———. 1963. Notes on Hispaniolan herpetology 8. cus. I. Variation in the old-field mouse (Peromyscus The forms related to Anolis hendersoni Cochran. polionotus). Studies in Genetics VI. University of Breviora 186:1-13. Texas Publications 7103:49-90. ———. 1965. The species of Hispaniolan green ano- Shochat, D., and H. C. Dessauer. 1981. Comparative les (Sauria: Iguanidae). Breviora 227:1-16. immunological study of albumins of Anolis lizards ———. 1969. The ecology of colonization as seen in of the Caribbean islands. Comp. Biochem. Physiol. the zoogeography of anoline lizards on small is- 68A:67-73. lands. Quat. Rev. Biol. 44:345-389. Simpson, G. G. 1956. Zoogeography of West Indian ———. 1972. Origin of faunas: evolution of lizard land mammals. Amer. Mus. Novitates 1759:1-28. congeners in a complex island fauna—a trial anal- Sneath, P. H. A., and R. R. Sokal. 1973. Numerical ysis. Evol. Biol. 6:47-89. taxonomy. W. H. Freeman, San Francisco. ———. 1976a. West Indian anoles: a taxonomic and Sokal, R. R. 1983. A phylogenetic analysis of the evolutionary summary. 1. Introduction and a caminalcules. IV. Congruence and character sta- species list. Breviora 440:1-21. bility. Syst. Zool. 32:259-275. ———. 1976b. South American anoles: the species ———. 1985. The continuing search for order. Amer. groups. Papeis Avulsos de Zoologia 29:259-268. Nat. 126:729-749. ———. 1983. Ecomorphs, faunas, island size, and diverse end points in island radiations of Anolis. Sourdis, J., and M. Nei. 1988. Relative efficiencies R. B. Huey, E. R. Pianka, and T. W. Schoener of the maximum parsimony and distance-matrix In (eds.), Lizard ecology, pp. 326-370. Harvard Univ. methods in obtaining the correct phylogenetic tree. Press, Cambridge. Mol. Biol. Evol. 5:298-311. ———. 1989. A critique of Guyer and Savage (1986): ———, and C. Krimbas. 1987. Accuracy of phylo- cladistic relationships among anoles (Sauria: Igua- genetic trees estimated from DNA sequence data. nidae). In C. A. Woods (ed.), Biogeography of the Mol. Biol. Evol. 4:159-166. West Indies: past, present, and future, pp. 433- Swofford, D. L. 1981. On the utility of the distance 478. Sandhill Crane Press, Gainesville, Florida. Wagner procedure. In V. A. Funk and D. A. Brooks ———, and T. P. Webster. 1974. Anolis rupinae new (eds.), Advances in cladistics, pp. 25-43. Proceed- species, a syntopic sibling of A. monticola Shreve. ings of the First Meeting of the Willi Hennig So- Breviora 429:1-22. ciety. New York Botanical Garden, New York. Wilson, A. C., S. S. Carlson, and T. J. White. 1977. ———, and S. H. Berlocher. 1987. Inferring evolu- Biochemical evolution. Annu. Rev. Biochem. 46: tionary trees from gene frequency data under the 573-639. principle of maximum parsimony. Syst. Zool. 35: Wyles, J. S., and G. C. Gorman. 1980a. The albumin 293-325. immunological and Nei electrophoretic distance ———, and R. B. Selander. 1981. BIOSYS-1: A FOR- correlation: a calibration for the saurian genus TRAN program for the comprehensive analysis of Anolis (Iguanidae). Copeia 1980:66-71. electrophoretic data in population genetics and Wyles, J. S., and G. C. Gormam. 1980b. The classi- systematic. J. Hered. 72:281-283. fication of Anolis: conflict between genetic and os- Sykes, L. R., W. R. McCann, and A. L. Kafka. 1982. teological interpretation as exemplified by Anolis Motion of Caribbean plate during last 7 million cybotes. J. Herpetol. 14:149-153. 28 K. L. BURNELL AND S. B. HEDGES

Yang, S. Y., M. Soule, and G. C. Gorman. 1974. Anolis eridgei—Dom. Rep., Elias Pina, Loma Nalga lizards of the eastern Caribbean: a case study in de Mace, N slope (3.2 km S, 4.0 km E Rio evolution, I. Genetic relationships, phylogeny, and colonization sequence of the roquet group. Syst. Limpio), 1270 m (USNM 286883); eugene- Zool. 23:387-399. grahami—Haiti, Nord, Plaisance, 9.0 km NE, 130 m (USNM 286901 ); fowleri-Dom. Rep., Accepted: 12 February 1990. Peravia, La Horma, 13 km NW, 1770 m (USNM 266303); hendersoni–Haiti, Sud Est, Sequin, 15 km SW, 600 m (USNM 286902); A PPENDIX 1 insolitus—Dom. Rep., Peravia, La Horma, Localities and Voucher Specimens 10.5 km NW, 1645 m (tissue voucher CUBA—centralis-Guantanamo Bay Na- 102956); monticola—Haiti, Sud, Camp Per- val Station, John Paul Jones Hill, at peak, rin, 13.5 km N, 680 m (USNM 286905); ols- 151 m (USNM 286816); homolechis—Guan- soni—Haiti, Artibonite, Ca Soleil, 10.4 km tanamo Bay Naval Station, golf course, 25 NW, 130 m (USNM 286906); ricordii—Haiti, m (USNM 286817); ophiolepis—Guantana- Nord Ouest, Ballade, 1 km N (7.8 km S Port- mo Bay Naval Station, golf course, 25 m de-Paix) (USNM 286907); semilineatus— (USNM 286818); porcatus—Guantanamo Dom. Rep., El Seibo, El Valle, 22 km WNW Bay Naval Station, golf course, 25 m (USNM (16 km to Trepada Alta, ca. 6 km by trail 286819); smallwoodi—Guantanamo Bay Na- to Montebonito), 76 m (USNM 286890); val Station, horse corral, 20 m (USNM sheplani—Dom. Rep., Barahona, Cabral, 20.8 286820). km S, 975 m (tissue voucher 102530); GUADELOUPE—marmoratus—Guade- shrevei— Dom. Rep., Vega, Constanza, ca. loupe, Basse Terre, Pointe de la Grande 37 km SE (via new road to San Jose de Anse, near Trois Rivieres, 0 m (USNM Ocoa), 2300 m (USNM 286894); singularis— 286911). Haiti, Sud Est, Gros Cheval, ca. 15 km W HISPANIOLA—Chamaelinorops bar- via logging roads (NE slope of Pic La Selle), bouri-Dominican Republic, Barahona, Te- 2020 m (USNM 286908); websteri-Haiti, junde (13 km SSW of La Guazara), 975 m Artibonite, Ca Soleil, 10.4 km NW, 130 m (USNM 286895); A. armouri-Haiti, Sud Est, (USNM 286909); whitemani-Haiti, Arti- Gros Cheval, ca. 15 km W via logging roads bonite, Ca Soleil, 10.4 km NW, 130 m (NE slope of Pic La Selle), 2020 m (USNM (USNM 286910). 286896); bahorucoensis—Dom. Rep., Bara- JAMAICA—garmani—Trelawny, Dun- hona, Cabral, 15.3 km S, 6.7 km E, 1340- cans, 0.3 km W at jct. with Silver Sands 1370 m (USNM 286871); baleatus—Dom. access road, 80 m (USNM 286836); gra- Rep., La Vega, Jayaco, 7 km W on road to hami—St. Mary, Port Maria, 2.9 km NW Constanza, 760 m (USNM 286872); bara- (Dowling House), 5 m (USNM 286841); li- honae—Dom. Rep., Barahona, Barahona, neatopus—St. Mary, Port Maria, 2.9 km NW 19.5 km SW, 880 m (USNM 286873); bre- (Dowling House), 5 m (USNM 286847); virostris—Dom. Rep., Pedernales, Oviedo, opalinus—Clarendon, Portland Cave 5 m (USNM 286874); chlorocyanus—Dom. (USNM 286850); reconditus–Portland, Sec- Rep., Elias Pina, Rio Limpio, 700 m (USNM tion, ca. 1-3 km WNW (USNM 286852); 286875); christophei—Dom. Rep., La Vega, sagrei—St. Mary, Port Maria, 2.9 km NW Jayaco, 7 km W on road to Constanza, 760 (Dowling House), 5 m (USNM 286858); va- m (USNM 286876); coelestinus—Haiti, Sud, lencienni—Trelawny, Quick Step, vic. of, 395 Camp Perrin, 13.5 km N, 680 m (USNM m (USNM 286860). 286897); cybotes—Dom. Rep., Barahona, PUERTO RICO—cooki—Playa de Tama- Barahona, vie. of, 0 m (USNM 286880); dar- rindo, 5 m (USNM 286821); cristatellus— lingtoni—Haiti, Grande Anse, Marche Leon, Rio Piedras, University of Puerto Rico cam- 11.2 km S, 1.9 km E, 1360 m (MCZ 173207); pus (USNM 286823); evermanni–El distichus—Dom. Rep., Barahona, Puerto Yunque, vic. of peak, 1000 m (USNM Alejandro, vic. of, 425 m (USNM 286882); 286824); gundlachi—El Yunque, vic. of peak, dolicocephalus—Haiti, Sud, ca. 5-6 km NW 1000 m (USNM 286825); krugi-El Yunque, Les Platons, ca. 900 m (USNM 286900); eth- vic. of peak, 1000 m (USNM 286826); oc- RELATIONSHIPS OF WEST INDIAN ANOLIS 29 cultus — El Yunque, vic. of University of Rio Piedras, University of Puerto Rico cam- Puerto Rico Biological House (USNM pus (USNM 286833). 286829); poncensis - Parguera, vic. of, 10 m UNITED STATES — carolinensis — South (USNM 286831); pulchellus – Ponce, 11 km Carolina, Jasper County, near Tillman W on rt 2, 0 m (USNM 286832); stratulus— (USNM 286915). 30 K. L. BURNELL AND S. B. HEDGES

APPENDIX 2. Protein variation for 50 taxa of anoline lizards at 12 Slow-Evolving Loci.

Locus Taxon Aat Adh Esd Gpi Ldh-1 Ldh-2 Lgl Pgm Pk Pt-1 Pt-2 Pt-3 centralis c f s Clc i2 e3Ab i blAc f2 b3A bl al homolechis bf j C1F c3 e3Ab g1 b2b a h2 bl f3 ophiolepis cf 1 bC a, C2 e3Ab g1 d2 a hl bl d porcatus df q h3A j j g1 b2b fl a3 C 2 h4 smallwoodi ff h clC cl c g1 b2b C2 g bl a2 rnarmoratus ff 1 f3 c3 e1a g1 blAc cl b3B b2 h5 C. barbouri fk k clC f e1a e a, b2a gl il b2 f9 armouri fe b,d al g2 e4c g1 d3 gl i3B b2 f8 bahorucoensis gf r h3D c2 e2 g1 blAb gl d b2 C 3 baleatus fh c i2B bl i2 a, c b2b gl b3B b2 i3 barahonae fh c i2B c1 d3 a b2b gl blC d i2 brevirostris hi n c1C c4 e4c g1 blAb cl b3B b2 f4 chlorocyanus ff f h3D c2 bA g1 b2b b blA b2 hl christophei ca i bB b2 e5 g1 blAb f4 blC b2 hl coelestinus hf m gB c2 bA g1 blAb d blB b2 hl cybotes fe d d2 g2 e4c g1 d3 gl i3B b2 f4 darlingtoni ff r i2D gl fl h blAb d — b2 h2 distichus hi n c1C e2 g g1 blAb cl a2 b2 f6 dolicocephalus jj r h3B c3 e2 g1 blAb gl alC b2 f7 etheridgei c h o c1E c2 f2A b blAb gl e b2 C 3 eugenegrahami fi t c1C c2 h g1 blAb c1 blC b2 f6 fowleri c g o d3 b2 e3B b blAb gl c b2 f7 hendersoni gj P h2 c2 e2 g1 blAb gl alB b2 f2 insolitus c f o e c2 d2 b b2b g2 i2B b2 fl monticola jf r f3 d e1a f blAa c2 blC cl fl0 olssoni c f a c2 b2 e4b e blAb gl b2A b2 C 2 ricordii fh c i2A bl i2 a b2b gl b,i d il semilineatus c f g c1B c2, i 1 e4b d c gl b2B b2 f2 sheplani fc o gA e1 f2Ba g1 blAe h i4 b2 cl shrevei fe d a2 h e4c d d3 gl i3B b2 fl singulars ff f bD c2 bB g1 blAc d alA b2 h6 websteri fi n c1C c4 e3Aa g1 blAb gl b3B b2 f2 whitemani fe d d2 g3 e4d g1 d3 gl i3B b2 f8 garmani ff 1 bA b3 e4b g1 blAb h b3C bl gl grahami ff 1 bA c3 e4a g1 blAb h b4C b2 g2 lineatopus ce v bB c3 e4b g1 blAb i i3C b2 c1 opalinus df 1 bA c3 e4b g1 blAb e b4C b2 g2 reconditus ce v bA c3 e4b g1 blAb i f b2 C 3 sagrei c f 1 bC c3 e4b g1 blAb e i2A b2 a3 valencienni a d 1 bB c3 e4b g2 blAb f3 i3C b2 e cooki gf n, t il d e4b g1 blAd gl i3A b2 f5 cristatellus gf n d1, i2C d e4d g1 blAd gl b3B b2 fl1 evermanni e f e, n cID c3 i1 g1 blAc cl b3B b2 h3 gundlachi e f g cIA c3 e4d g1 b2b h b4A a j krugi if n f2 c3 bB g1 b2b gl b3B b2 f2 Occultus e b, f u c1C c3 e1a g1 blAc cl, d i3A b2 b poncensis kf n i2D d f2Bb g1 blB,dl gl b4B b2 f7 pulchellus if n, t f1 c3 a, e1b g1 b2b gl b4B b2 f6 stratulus ff n h1 c3 d1 g1 blAc gl b3B b2 f2 carolinensis df i h3C i k g1 blAb e alA b2 fl2