Adaptive radiation of day-geckos (Phelsuma) inthe Seychelles archipelago: a phylogenetic analysis.
Although each system may be the result of a unique combination of factors, the factors themselves are general and should be studied in a standard way that facilitates comparisons. One essential step in this method should be the comparison of robust phylogenetic hypotheses to estimates of physical history. Concordance between the two indicates invasion into the system through speciation; discrepancies indicate invasions into the island community via colonization from other local or regional biogeographic areas (see Brooks and McLennan 1991 for general examples). Integration of ecological information with historical analyses allows the definition of the role ecological interactions play in the evolutionary process (Williams 1972, 1983). Identifying physical invasions allows the application of theory defining how communities are assembled and structured. Identifying the ecological context of speciation events allows the definition of evolutionary mechanisms that involve ecological processes. In this study I used phylogenetic inference, together with ecological data, to determine what combination of these two processes is responsible for the three taxa of day-geckos (Phelsuma) seen in the Seychelles archipelago.
Arboreal day geckos of the genus Phelsuma, with about 40 species and 60 described "taxonomic units," occur from southwestern Africa to the Bay of Bengal (Loveridge 1942). Within this range, the majority of species are found on islands in the southwest Indian Ocean, including Madagascar, the Seychelles, Mascarene, and Comoros Islands [ILLUSTRATION FOR FIGURE 1 OMITTED]. Two points of disjunction exist in the distribution, however. First, the only species to occur to the east of India is P. andamanensis, which is endemic to the Andaman and Nicobar Island archipelagos of the Bay of Bengal (Loveridge 1942; Biswas and Sanyal 1977, 1980). No species occurs between these archipelagos either on islands in the central Indian Ocean (Chagos, Lakshadweep, Maidives, Sri Lanka) or the Indian subcontinent. Second, one native species from the Richters-veld of the southwestern African mainland, P. ocellata, has been placed in the genus by Schmidt (1934) and Russell (1977).
The geologic history through the range of Phelsuma is well known. Madagascar, for instance, was tectonically rifted from the African land mass over 130 mya (Coffin and Rabinowitz 1987) and had a close association with southern Africa that goes back at least to the Miocene (Forster 1975; Tarling 1981). At the northern end of the distribution, the Andaman and Nicobar island archipelagos are considered extensions of the Himalayas uplifted during the Miocene (Karunakaran 1962, 1967), although they have never been connected to the Indian subcontinent.
Between the extremes of the Phelsuma distribution [ILLUSTRATION FOR FIGURE 1 OMITTED] the remaining islands of the Indian Ocean have a varied past. For instance, the Aldabra group (Aldabra, Assumption, Cosmoledo, and Astove) are raised coralline islands. The last total submergence of Aldabra due to eustatic sea-level changes was approximately 125,000 years ago (Thompson and Walton 1972; Taylor et al. 1979). The other islands in the Aldabra group, along with the Farquhars and Amirantes (sandy cays to the west and north of Madagascar, respectively) have probably emerged only within the last 15,000 years. Northwest of Madagascar lie the Comoros islands and to the east the Mascarenes. Both groups are volcanic islands high enough to remain emergent since their origin. Finally, the granitic Seychelles are high islands of continental origin representing fragments left by the Indian plate as it moved north. The Seychelles [ILLUSTRATION FOR FIGURE 2 OMITTED] and the Comoros have formed separate large banks within a period of 15,000 years (Peake 1971).
Systematic studies within the genus Phelsuma are limited, with the majority of the work being phenetic (Russell 1977; Crawford and Thorpe 1981; Cheke 1982, 1984; Thorpe 1983; Thorpe and Giddings 1983; Gardner 1984, 1987). Russell (1977) recognized two groups in Phelsuma based on toe-pad morphology. His group I contains most Phelsuma species, the mainly-arboreal forms. His group II lizards (P. barbouri, breviceps, dubia, mutabilis, standingi, guttata, and ocellata) possess slightly different habits (e.g., some ground dwelling and preferring drier climates), smaller body size, and duller coloration. The only phylogenetic work on species within Phelsuma has focused almost entirely on species thought to be closely related to P. madagascariensis (Borner and Minuth 1984; Thorpe 1985, 1986) and will be discussed below in relation to my current results.
Ecological Motivation of Phylogenetic Hypotheses
Each of the granitic Seychelles supports either zero, one, or two species of Phelsuma. Under current taxonomy (Gardner 1984, 1987; Thorpe 1985), P. astriata and P. sundbergi are the only species involved in this simple ecological setting. The low species richness of Phelsuma in the Seychelles is likely due to the isolation of this archipelago (Losos 1986). Although several small-scale ecological studies (Crawford and Thorpe 1979; Thorpe and Crawford 1979; Evans and Evans 1980) have been conducted, Gardner (1984) has provided the most extensive look at the reproductive tactics, micro- and macrohabitat utilization of these two species in sympatry and in allopatry.
An intriguing summary of body size information, a combination from those of Gardner (1984) and my own measurements, is given in Figure 3. The figure shows the distribution of snout-vent lengths (mean of the three largest males) for the species across the granitic Seychelles islands with the islands (1-19) placed in approximate south to north orientation. Figure 4 shows explicit distributions of snout-vent (SVL) lengths for both species from six of the largest islands. Several patterns emerge from these figures. First, when ordered from south to north, SVL of P. sundbergi generally increases while SVL of P. astriata shows no clear pattern of variation with respect to island. Second, the mean difference (25.76 mm) of SVL between P. astriata and P. sundbergi within Praslin and its associated islands is significantly different from the corresponding mean difference (7.35 mm) in the Mahe group of islands (t-test, P [less than] 0.0001). Third, on Silhouette, one of the larger and more isolated islands associated with Mahe, a convergence of relative body size is observed.
Assuming the habitats are similar across this small archipelago (see below), it seems inexplicable that one set of islands has coexisting congeners with overlapping body size distributions and another set has the same two species but with very distinct body size distributions. The situation is complicated further by the fact that these islands have historically been united into a single large bank several times during glacial maxima (Braithwaite 1984; Montaggioni and Hoang 1988). Given this intimate setting, why does one set of islands contain two species with significantly less difference in maximum body size than another set of islands with the same two species?
MacArthur and Levins (1967) and May and MacArthur (1972) have shown analytically that a limit exists to the similarity competing species can attain. If the resource overlap between two species is too large they must either diverge in use or one species must go extinct. Where species occur in the absence of competitors, density-dependent natural selection should work to maximize carrying capacity thus leading to intermediate correlates of niche use (e.g., body size) under certain conditions (Schoener 1969; Roughgarden 1976, 1983). Since then it has been shown that evolutionary divergence is predicted under a wide variety of conditions of species interactions (see reviews in Taper and Case 1992a,b).
In the present case, Gardner (1984) argues that, although ecological processes are important in determining population densities of Phelsuma, a relationship between body size and competition is not the causal factor in body size evolution in the Seychelles Phelsuma. He bases his argument on the fact that on Mahe, where P. astriata and P. sundbergi are very similar in maximum SVL (57 vs. 63 mm), the two species currently exhibit little microhabitat partitioning. Is this the correct interpretation of the body-size information? Below I offer an alternative argument that is consistent with character displacement driving body size evolution and speciation in this system.
To test whether the differences in body size observed within the Seychelles Phelsuma are a result of character displacement, I will use the six necessary criteria for competitive character divergence provided by Schluter and McPhail (1992) and Taper and Case (1992a). The six criteria stated in relation to this system are: (1) body size differences between sympatric species relative to allopatric species cannot be ascribed to differential colonization (size assortment); (2) body size differences between sympatric species pairs are statistically greater than between paired allopatric populations; (3) differences in body size are realized in differences in resource utilization; (4) Competition for resources increases as individual Phelsuma become more similar in size; (5) size differences are not correlated to resource distribution; and (6) body size differences are genetic (assumed in this case).
The criteria outlined above suggest alternative hypotheses to character displacement for observed body size patterns. The focus of this study is to use a phylogeny generated from changes of mtDNA nucleotide sequence to test alternative phylogenetic hypotheses that would have resulted under the size assortment hypothesis of Criterion 1. If it can be shown using a phylogeny that no body size evolution has taken place, then further analysis becomes unnecessary. For phylogenetic testing I will refer to the two different sized forms of P. sundbergi using subspecific designations found under current taxonomy (Gardner 1984, 1987; Thorpe 1985). The intermediate form (mean maximum SVL = 62.23 mm) from the Mahe island is referred to as P. s. longinsulae; the larger form (mean maximum SVL = 80.63 ram) from the Praslin island group is referred to as P. s. sundbergi.
After using the mtDNA phylogeny to show that the observed body size differences of Phelsuma in the Seychelles are due to evolutionary adjustments within the archipelago, I will address the other criteria for character displacement. First I will use within species genetic differences between islands to infer that the body size distribution of P. s. longinsulae, which falls between P. astriata and P. s. sundbergi, was likely generated in allopatry from other Phelsuma species. I will then use ecological information to show that body size is related to resource use and that similar sized individuals compete for resources. Finally I will argue against the alternative hypothesis that body size differences are the result of differential resource availability. The alternative hypothesis that body size differences are not genetic is assumed to be false.
Testing for Size Assortment
Three possibilities exist for the phylogenetic relationships of the Seychelles Phelsuma [ILLUSTRATION FOR FIGURE 5 OMITTED] relative to other species in the genus. The first possibility, as a result of size assortment, could be generated by three physical invasions leading to the three Phelsuma taxa observed today. Without prior knowledge of the genus the three taxa would be expected to be placed randomly in a larger phylogeny of Phelsuma species. Populations of P. sundbergi inhabiting these two groups of islands would appear as different lineages that show closer relationships to their potential source pools than between adjacent island groups.
The published phylogenetic hypothesis of Borner and Minuth (1984) and other published taxonomic affiliations (Loveridge 1942; Mertens 1962, 1966; Borner 1972; Cheke 1982; Gardner 1984, 1987) provide a more focused size assortment hypothesis [ILLUSTRATION FOR FIGURE 5 OMITTED]. Proposed sister relationships resulting from three colonizations are astriata-lineata, s. longinsulae-abbotti and s. sundbergi-madagascariensis. If the original colonization of the Seychelles is a recent event (100,000 years before present, as suggested by Cheke ), populations of P. sundbergi of the Praslin and Mahe groups of islands could be closer phylogenetically to different sized subspecies of P. abbotti from Aldabra, Assumption, or Madagascar than they are to each other. The size assortment hypothesis further implies that the two subspecies of P. sundbergi are in fact separate species.
If P. s. longinsulae and P. sundbergi are really only one morphologically variable species, then two colonizations could have led to the observed species present in the Seychelles. This scenario would likely produce two pairs of sister relationships: astriata-lineata, and either sundbergi-abbotti or sundbergi-madagascariensis. If individuals of P. abbotti had colonized, an evolutionary increase of body size for the Praslin forms would have to be invoked; if P. madagascariensis were the colonist then a decrease of body size occurred. Cheke (1984) proposed a P. abbotti colonization from the Aldabra group due to favorable ocean currents. A two-invasion hypothesis implies that the observed body size distributions result from the interaction between separate founder events, an undetermined evolutionary force and the time required for variability of body-size to build up in a population. Character displacement could still be operating in this scenario.
Under the two-invasion hypothesis, founder effect from isolation due to rising sea level can also be invoked to explain the body size patterns. In this case, original founders of P. astriata left in each island group would all fortuitously be the same size, while P. sundbergi founders were different. Such a scenario seems unlikely but would be difficult to separate from character displacement.
The last general phylogenetic hypothesis is that the Seychelles Phelsuma form a monophyletic unit. If species interactions are responsible for driving evolution of these species, and therefore represent adaptations, observed changes must have occurred in the taxa under investigation while the interactions were taking place (Coddington 1988, Baum and Larson 1991). Both of these would be established with a result of monophyly; body size evolution would necessarily have taken place in a lineage present continuously from colonization. I am therefore accepting monophyly as a necessary condition to verify character displacement in this system.
To test for monophyly of the Seychelles Phelsuma, I reconstructed the phylogenetic relationships for a set of species within the genus Phelsuma that included the taxa from the Seychelles. I used restriction enzyme site mapping and sequencing of mitochondrial DNA (mtDNA) (360 bp. of the cytochrome b gene) to provide discrete characters for two estimates of phylogenetic relationships in this genus.
Restriction enzyme maps were constructed for the following species: P. abbotti, P. astriata, P. laticauda, P. leiogaster, P. lineata, P. madagascariensis grandis, P. quadriocellata, P. sundbergi longinsulae, and P. sundbergi sundbergi. In addition to the above species, nucleotide sequences were determined for P. andamanensis and P. dubia. Between the restriction enzyme and sequence surveys ten different islands in the Seychelles were sampled: Silhouette, Therese, Mahe, Cerf, St. Anne, Praslin, La Digue, Sister, Curieuse, and Bird [ILLUSTRATION FOR FIGURE 2 OMITTED]. Single individuals were used for each species except for P. astriata and P. sundbergi where single individuals were used for each island. P. leiogaster was provided by the U.S. National Museum (Field Series #59410); P. andamanensis was provided by the British Museum (#220.127.116.11). All other species were provided by private collectors and breeders. The Seychelles species were collected by me during two trips (1989, 1991). All animals were brought back alive.
Whole mtDNA was extracted from freshly killed individual geckos using a modified alkaline lysis procedure described by Timura and Aotsuka (1988). After isolation, whole mtDNA was subjected to either restriction enzyme digestion or PCR amplification, or both.
Restriction Enzyme Analysis
Individual samples were subjected to single digestion with each of the following restriction enzymes: BamH I, Bcl I, Bgl II, BstE II, Cla I Nco I, Pvu II, Sst II, and Xba I. All enzymes were used according to the manufacturer's (GIBCO-BRL) recommended reaction conditions. Digestions were carried out for approximately 4 h. For restriction site mapping, it was necessary to perform double digestions. The first double digestion performed used Sst II and Cla I. Carr et al. (1987) have shown previously that Sst II produced two sites about 1.7 kb apart that were conserved between frogs in the genus Xenopus and salmonid fishes (Berg and Ferris 1984). These sites were also found in all Phelsuma surveyed. Cla I acted on a site that was nearly invariant within Phelsuma. Because Sst II produces blunt-ended restriction fragments that do not end-label efficiently, Cla I was used to anchor the majority of double digestions. Restriction fragments were end-labeled with the large fragment of DNA polymerase I (klenow) and separated on agarose gels. The gels were dried and placed on x-ray film at -70 [degrees] C for 24-72 h.
After double digestion and mapping of restriction sites, the presence or absence of restriction sites was coded into a matrix. Taxonomic units for P. sundbergi and P. astriata consisted of every unique haplotype (i.e., each unique signature of presence or absence of restriction sites).
For phylogeny reconstruction a maximum parsimony criterion was adopted with restriction site gains weighted three times heavier than losses (Debry and Slade 1985; Swofford and Olsen 1990). For both restriction enzyme and sequence characters, trees were searched using the branch-and-bound algorithm (Hendy and Penny 1982) implemented in the computer program PAUP 3.0s (Swofford 1991), thus assuring the discovery of all equally most parsimonious trees. Confidence of individual tree nodes was established using the bootstrap data resampling technique (Efron 1982; Felsenstein 1985; Penny and Hendy 1985).
After total DNA isolation, sequencing templates were generated using PCR methodology. The procedure for amplifying double-stranded DNA (dsDNA) follows procedures found in Innis et al. (1990). The standard reaction was run in a 50 [[micro]liter] volume and contained 1 ng of target mtDNA plus the following components (final concentration): 50 mM KCl, 10 mM Tris-HCl pH 9.0, 0.1% Triton X-100, 3 mM MgCl, 0.2 mM each dNTP, 0.4 mM each primer, and 2.5 units (total) Taq DNA Polymerase. The following primers were used for PCR reactions and sequencing: GLUDG-L and CB3-H (Palumbi et al. 1991), CB2-H (Kocher et al. 1989). The following cycling parameters were used: initial melting - 94 [degrees] C for 2.5 min, [melting - 94 [degrees] C for 1 min, annealing - 49 [degrees] C for 1 min, extension - 72 [degrees] C for 1 min] 35 cycles, final extension - 72 [degrees] C for 7 min.
Once dsDNA PCR products were obtained, a small aliquot (5 [[micro]liter] of a 50 [[micro]liter] reaction) was carried over to another PCR reaction to which only one primer was added. For this reaction, the annealing temperature was raised to 55 [degrees] C and the number of cycles was reduced to 25, otherwise parameters were unchanged. This second PCR reaction produced a single-stranded DNA (ssDNA) product which was purified using microcentrifuge filtration units to remove unused components of the PCR reaction. Template DNA was produced for both strands of the dsDNA product. The chain termination method of Sanger et al. (1977) was used to produce sequencing reactions and the products electrophoresed on 6% polyacrylamide gels.
Individual nucleotide sites were used as discrete characters and the phylogeny was estimated using PAUP with a maximum parsimony criterion. Felsenstein (1978) showed that taxa separated by large evolutionary distances might be falsely placed together on phylogenies. The reason is that the misinformation generated by parallel changes overwhelms the information contained in nonparallel changes. This phenomenon of long phylogenetic branches attracting one another represents a problem for cytochrome b sequences because of their relatively rapid rate of evolution in vertebrates. To minimize the chance of multiple parallel events, only transversional differences were used for phylogenetic reconstruction. Further, the phylogenies are not rooted because of the same problem. Rooting the phylogenies is not necessary to test whether the Seychelles' Phelsuma are monophyletic.
Body Size Analysis
Snout-vent length was used as a measure of body size. Data are presented in two ways. First, the average of the largest three individuals is used to allow the combination of my measurements with those previously reported by Gardner (1984). Second, for a smaller set of islands, I give frequency distributions of SVL for all individuals measured. Gardner (1984) did not provide similar distributions. Pairwise comparisons of means were conducted using t-tests. In the case of the comparison of P. astriata means between the Praslin and Mahe island groups the nonparametric Mann-Whitney U-test was used because the Fisher's exact test showed the variances of these two populations to be unequal. Comparisons of the body size distributions of P. astriata (combined across all islands), P. s. longinsulae (Mahe island group), and P. s. sundbergi (Praslin island group) were made using ANOVA and Scheffe tests.
Restriction Enzyme Analysis
The restriction site maps produced from the endonuclease reactions are shown in Figure 6; the 65 restriction sites derived from the maps are presented in Table 1. From the maps and the presence/absence matrix it can be seen that two Sst II sites (sites 25-26) are invariant across all haplotypes present. In addition, most haplotypes also possessed a common Xba I site (4) and Cla I site (34). Minor size variation of the entire mtDNA genome for each haplotype was mapped but was not used for phylogenetic analysis because within-species variation lead me to believe that homology would be impossible to verify.
Four P. astriata and five P. sundbergi haplotypes were found among the seven islands surveyed. A matrix giving the number of restriction sites found in each taxon along the diagonal, number of shared restriction sites in the upper triangle, and estimates of genetic distance (Nei and Tajima 1983) in the lower triangle is shown in Table 2. For P. astriata unique haplotypes were found on Silhouette and La Digue. A third widespread haplotype was found on Therese, Mahe, Cerf, Praslin and Curieuse. The La Digue haplotype possesses two variable restriction sites (6 and 32) defining this haplotype. The common haplotype lacks site 6, whereas the Silhouette haplotype lacks both. Estimates of percent nucleotide substitution are 0.48% (common-La Digue), 0.51% (common-Silhouette) and 1.01% (La Digue-Silhouette).
Five unique haplotypes in P. sundbergi were found on Silhouette, Mahe, Therese/Cerf, Praslin/Curieuse, and La Digue. Haplotypes found in the P. s. longinsulae individuals from the islands associated with Mahe vary at sites 27 and 29; haplotypes found in P. sundbergi individuals were defined by variable site 63. Percent nucleotide substitution was 0.46-0.48% between P. s. longinsulae haplotypes, 0.51% between P. sundbergi haplotypes, and 1.50-2.75% between P. s. longinsulae and P. sundbergi haplotypes.
The 65 restriction sites found in the survey provided 41 phylogenetically informative characters. The single most parsimonious tree (mp-tree) is shown in Figure 7. The length of this tree based on the relaxed Dollo criterion is 178 steps. The tree contains three monophyletic units: (1) the Seychelles taxa; (2) P. abbotti and madagascariensis; and (3) P. lineata, leiogaster, quadriocellata, and laticauda. Branch lengths are drawn proportional to the number of changes between nodes. The bootstrap value testing the monophyly of the Seychelles taxa is 89% and is shown in bold.
Five alternative phylogenetic hypotheses [ILLUSTRATION FOR FIGURE 8 OMITTED] were compared to the mp-tree. The first tree, showing a sister relationship between P. leiogaster and the P. sundbergi clade, was the closest tree (tree length of 186 vs. 178) found from the parsimony analysis hypothesizing a paraphyletic relationship [TABULAR DATA FOR TABLE 1 OMITTED] for the Seychelles taxa. Alternatives 1-5 are the hypotheses motivated by previous studies of Phelsuma and generalized in Figure 5.
The results of the tree comparisons are given in Table 3A. The first column gives the tree lengths. The best nonmonophyletic Seychelles tree is eight steps longer than the mp-tree. All of the alternatives are substantially longer (190-206). The maximum likelihood algorithms (RESTML81 and DNAML81) of Felsenstein (1992) implement the paired sites test of Kishino and Hasegawa (1989) for comparing tree topologies. The standard deviations and likelihood differences to the mp-tree are given in the last two columns. The significance criterion is a In likelihood difference of 1.96 SD or more. All trees are significantly worse than the mp-tree.
The analysis of cytochrome b mtDNA produced sequences 360 bp in length (available in GenBANK). These sequences resulted in a total of 129 phylogenetically informative transversions. The distance matrix in Table 4 is provided to summarize differences between taxa. Values in the lower triangle are the overall proportion of nucleotide substitutions; the proportion of differences considering only transversions are shown in the upper triangle.
Nucleotide substitution varied from 0.3% between various P. astriata and sundbergi haplotypes to 28.9% between P. abbotti and P. dubia. The smallest between-species distance is 8.9% between P. s. longinsulae and P. astriata. Proportion of transversions varied from zero within P. astriata and within P. sundbergi, to 11.9% for several taxa. The distance between P. astriata and P. sundbergi was 3.0%.
The phylogeny based on only transversions is shown in Figure 9. Only one haplotype of each taxonomic unit from the Seychelles was used (astriata-Mahe, sundbergi-Praslin, s. longinsulae-Mahe). Branch lengths are proportional to the number of changes. The bold value (98%, 400 replicates) is the bootstrap percentage supporting the hypothesis that the taxa from the Seychelles form a monophyletic unit. The bootstrap value for the Seychelles is also 98% when all characters are weighted equally. The transversion tree [ILLUSTRATION FOR FIGURE 9 OMITTED] is similar to the restriction enzyme tree [ILLUSTRATION FOR FIGURE 7 OMITTED] in several ways. First, the Seychelles taxa form a monophyletic group supported by high bootstrap values. Second, P. laticauda, leiogaster, lineata, and quadriocellata fall within a second monophyletic group. Finally, P. abbotti and madagascariensis form a sister relationship. P. dubia and andamanensis were not considered in the restriction enzyme analysis. In Table 3B the mp-tree is compared to the general alternatives depicted in Figures 5 and 8. All alternatives were significantly worse than the mp-tree.
To understand divergence patterns within the Seychelles, I constructed a phylogeny [ILLUSTRATION FOR FIGURE 10 OMITTED] for this subset of taxa using all variable sites. Two monophyletic groups are defined for P. sundbergi corresponding to the nominate subspecies s. sundbergi (2 haplotypes) and s. longinsulae (3 haplotypes). Bootstrap values supporting each of these monophyletic groups are shown in bold. Very little variation was observed in P. astriata haplotypes (4) across islands.
The phylogenies from both the restriction enzyme and sequence analysis are not concordant with the size assortment hypothesis of the origin of the Seychelles Phelsuma [ILLUSTRATION FOR FIGURE 5 OMITTED], nor are they concordant with a two colonization hypothesis. The monophyly of P. astriata, P. s. sundbergi and P. s. longinsulae is consistent with the size adjustment (character displacement) hypothesis; therefore, the necessary condition (Schluter and McPhail 1992; Taper and Case 1992a) for character displacement as an adaptation to species interactions has been met.
Body Sizes of Sympatric versus Allopatric Species Pairs
The next step for testing whether character displacement is responsible for the variation of body size in P. sundbergi is to compare body size differences of sympatric species pairs with paired allopatric populations. Differences of sympatric species pairs should be statistically greater than allopatric [TABULAR DATA FOR TABLE 2 OMITTED] pairs (Schluter and McPhail 1992; Taper and Case 1992a). Unfortunately few single species islands exist and four of the eight islands are recently emergent with human introductions likely. Meaningful statistical tests are therefore not possible. Is there any other way to address this criterion?
Figures 2-4 demonstrate why Gardner (1984) discounts the importance of frequency dependent natural selection. Some islands have sympatric populations with similar body sizes while others have widely differing body sizes. On the other hand, if these populations were all established by rising sea levels and interspecific competition is not driving evolution, we would expect to see random differences in body size between species pairs throughout the archipelago. Figures 2 and 3 show that the distribution of body sizes across islands is clearly not random. As stated above, the mean difference of maximum SVL between P. astriata and P. sundbergi among the islands of the Praslin group is significantly different from the corresponding mean difference in the Mahe island group. Further, differences in mean maximum SVL of P. astriata from the two island groups are not significant (t-test, P [greater than] 0.15) whereas the difference in P. sundbergi is significant (t-test, P [less than] 0.001).
Before interactions between two species could participate in the evolutionary process, the original colonist of the Seychelles must have undergone speciation. The cladogenesis within the Seychelles represents a minor radiation of species that was likely facilitated by eustatic sea level changes. Falls in sea-level of up to 120 m during glacial maxima created a single land area exceeding 125,000 [km.sup.2] (Braithwaite 1984; Montaggioni and Hoang 1988). In Figure 11, I give a historical scenario consistent with ecologically driven changes of body size. The scenario given minimizes both the number of colonization events and the number of body size changes necessary to explain the distribution of body sizes in the Seychelles Phelsuma. As shown here the original species founded the single large bank and then became isolated into separate populations during an interglacial period [ILLUSTRATION FOR FIGURE 11A OMITTED]. Slight variations exist for the exact timing of geological events but eventually two isolated populations [ILLUSTRATION FOR FIGURE 11B OMITTED] were formed. Each set of islands can be considered units because each group would have formed sub-banks when the sea level reached lower levels eustatically (e.g., 40 m, [ILLUSTRATION FOR FIGURE 3 OMITTED]) in the past.
[TABULAR DATA FOR TABLE 3 OMITTED]
Given enough time and lack of gene flow, the geographically isolated populations became reproductively isolated establishing P. sundbergi in the Mahe group and P. astriata in the Praslin group. Percent nucleotide substitution estimates of at least 8.9% between P. astriata and P. sundbergi along with several morphological characterizations (Rendahl 1939; Loveridge 1942; Mertens 1962) confirms the species status and relative timing of the event shown in Figure 11B. The lowest between species distance observed for non-Seychelles species was 16.5% (Table 2) between P. lineata and P. laticauda. Similar genetic distances have been observed between species of geckos in the genus Lepidodactylus (Radtkey et al. 1995).
Figures 11C-D show how differences in the order of secondary contact between the two sister species could have led to the puzzling distribution of P. sundbergi body sizes. Figure 11C shows how secondary contact between the two species may have first occurred in the Praslin group of islands. Subsequently P. astriata and P. sundbergi coevolved body sizes throughout the Praslin group. Meanwhile, on the Mahe group P. sundbergi evolved (or retained) a solitary body size predicted on the basis of density-dependent natural selection (i.e., a body size that gives the greatest net intake of the resources available) prior to invasion of P. astriata. If P. astriata had invaded the Mahe group after density-dependent selection on P. sundbergi had taken place [ILLUSTRATION FOR FIGURE 11D OMITTED], the body size of P. sundbergi is now in coevolutionary non-equilibrium, with natural selection driving it towards equilibrium.
Assuming that the rate of mtDNA evolution is constant within these closely related species, the two sets of species should show dissimilar degrees of mtDNA divergence if the two secondary contact events did not occur simultaneously. The prediction that follows from this hypothesis asserts P. sundbergi should have mtDNA genetic distances that reflect the older invasion event which occurred from the Mahe to the Praslin system; P. astriata mtDNA is predicted to be much less divergent between island groups reflecting the relative recency of P. astriata in the Mahe system.
Figure 12 shows that the observed pattern of genetic distances supports the predictions of a model of asynchronous secondary contact between island groups. The figure gives the average genetic distance obtained from both the restriction enzyme and sequence analysis between all island pairs surveyed. Between island genetic distances are about four times higher for P. sundbergi compared to P. astriata. Similar inter-island genetic distances are observed for both species within each island group and for P. astriata between island groups.
These results show the total population of P. sundbergi from each island group has been isolated much longer than P. astriata populations and P. s. longinsulae populations in the Mahe group occurred on single species islands until only recently. The results also imply a common geologic history in the form of isolation by the most recent rise in sea level. Prior to the separation of the two subbanks, P. astriata probably colonized the Mahe group [ILLUSTRATION FOR FIGURE 11D OMITTED]. As the sea level rose further the populations in the two sub-banks were likely experiencing relative panmixia.
Given the relatively shorter time period of interaction within the Mahe group, the body size distribution of P. s. longinsulae in this area is not representative of a two-species island. Body size distributions from the three taxa P. astriata, P. s. longinsulae and P. s. sundbergi are shown in Figure 13. Because the body size distributions of P. astriata between island groups were not significantly different (Praslin mean SVL = 49.98 mm, SD = 3.76 mm; Mahe mean SVL = 51.36 mm, SD = 5.37 mm; Mann-Whitney U-test P = 0.074) they were combined. The body size distribution (n = 166, mean SVL = 57.84 mm) of the recently "solitary" species, P. s. longinsulae, falls between the distributions of the two coevolved species P. astriata (n = 238, mean SVL = 50.56 mm) and P. s. sundbergi. (n = 119, SVL = 72.01) with all distributions significantly different from each other (ANOVA P [less than] 0.0001, Scheffe P [less than] 0.0001).
Relationships between Body Size and Resource Use
Although the body size distributions of P. astriata, P. s. longinsulae and P. s. sundbergi can be reconciled with competition [TABULAR DATA FOR TABLE 4 OMITTED] theory by the previous argument, I still must show that body size reflects resource use and also that similar sized Phelsuma compete for resources (Schluter and McPhail 1992; Taper and Case 1992a). Below I use ecological measurements and conclusions provided by Gardner (1984) to show the most intensive competition for resources occurs between similar sized Phelsuma.
Gardner (1984) verified a body size/prey size relationship by stomach content analysis of 65 P. astriata and 32 P. sundbergi from Praslin island in 4 different habitats. He generated frequency distributions of arthropod prey size for each species according to number and volume of each size class in the diet. He found these distributions significantly nonrandom with P. astriata taking smaller prey than P sundbergi. Because of morphological constraints of gape size, a prey size/body size relationship is no surprise and is common in reptiles (Schoener 1967, 1968; Roughgarden 1972; Case 1979; Pianka 1986). No prey size or prey category overlap values were computed.
Gardner reported explicitly measured macrohabitat niche breadths and overlaps (Simpson 1949; Pianka 1973) only from Praslin island where the difference between maximum SVL is large ([approximately equal to] 25 mm). Based on six habitat categories (coconut plantation, Lodoicea forest, Deckenia forest, coastal hardwood forest, eroded hillside scrub, and Casuarina plantation), macrohabitat niche breadth for P. astriata was 0.60 and for P. sundbergi 0.58. The overlap value between these species was 0.49. On Mahe, where body-size differences are small ([approximately equal to] 7 mm), Gardner concluded that macrohabitat overlap was nearly complete.
Gardner more convincingly demonstrated the importance of body size with observations of the differential use of palm tree pollen and nectar by individual Phelsuma. Palms from the genera Lodoicea, Deckenia, and Cocos all produce male inflorescences that are actively defended for highly nutritious resources. In all cases reported, the largest individual present displaced smaller individuals regardless of the intruding species. Indirect demonstrations of the effect of size on the competition for flower resources is illustrated in Figure 14 and Table 5. The graph shows the relative density of P. astriata in coconut plantations as a function of its body size relative to sympatric P. sundbergi. As the two species become more equivalent in body size the relative density of P. astriata increases. On Silhouette, where the relative body size is slightly greater than 1.0, P. astriata is able to dominate.
Gardner took detailed microhabitat measurements (niche breadths and overlaps, [Simpson 1949; Pianka 1973]) within the coconut plantation macrohabitat from Praslin island. Presumably this was done because of the directly observable interference competition associated with the nectar and pollen resources and the uniformity of the coconut plantations. Eight structural niche categories were used to divide the coconut trees (trunk, matting, leaf petiole, leaf midrib, leaf frond, inflorescence with open male flowers, inflorescence without open male flowers, coconuts). Based upon observations of 189 P. s. sundbergi and 48 P. astriata, niche breadth for P. astriata was 0.357 and for P. sundbergi niche breadth was 0.584; microhabitat niche overlap between the two species was 0.380. Gardner states that on Mahe "no microhabitat, temporal or food partitioning could be detected in the coconut plantations." Microhabitat niche width of P. astriata on Mahe, measured as on Praslin, is 0.504. The niche width of P. sundbergi was not reported nor was the overlap value.
The effect that size has on competition is also demonstrated by differential microhabitat utilization in coconut plantations between one and two species islands (Table 5). On Praslin and La Digue (where P. sundbergi are much larger than P. astriata) P. astriata are rarely seen on male inflorescence when present in coconut trees (0.0% and 4.8% respectively). In contrast, on nearby islands with no P. sundbergi present this percentage increases to a maximum of 40.0%. A similar phenomenon is observed on the islands of Silhouette and North. Phelsuma sundbergi has a slight size disadvantage on Silhouette. The percentage of P. sundbergi on male flowers drops to 20% on Silhouette but increases to 35% on North where P. astriata is absent.
Effect of Resource Distributions
Gardner's ecological data suggest that competition is ongoing. Density compensation and microhabitat shifts by a smaller species on single species islands are important results of these natural removal experiments. As pointed out by Diamond (1986) however one of the biggest problems with natural experiments is the lack of control over confounding variables. Criterion 5 (Schluter and McPhail 1992; Taper and Case 1992a) defines one of the variables that must be accounted for: resource distribution. Differences in body size distribution, density, and habitat use could all be a result of differences in resources across sites.
The major resources considered here are male inflorescences of palm trees. As suggested by Gardner (1984), large body size has likely evolved in response to the size associated interference competition for this nutritious and defendable resource. He believes, however, that differences in the distribution of palm trees led to the large and intermediate sized forms of P. sundbergi. He claims that because Lodoicea maldivica occurred only in the Praslin group, the selection pressure necessary to drive the evolution of larger body size was absent in the Mahe group. But as shown by all of Gardner's observations, competition in C. nucifera plantations is equally intense. This species along with Deckenia, the other palm used by Phelsuma, is distributed throughout the Seychelles.
Sauer (1967) discussed the origin of C. nucifera in the Seychelles and came to the conclusion that the variety present there colonized naturally long before humans arrived. Sauer cites several reports of fossilized palms closely resembling Cocos from Tertiary India (Sahni 1946; Kaul 1951; Rao and Menon 1964) and Pliocene New Zealand (Berry 1926). The Seychelles Cocos seeds are smaller and have thicker husks when compared to other more actively cultivated, exotic varieties (Durocher-Yvon 1953). Exotic varieties are also very susceptible to attack from native Seychelles beetles (Cooke 1958) which implies that the native C. nucifera varieties have been present long enough to evolve resistance to insect attack. The distribution of ancestral Cocos around the Indian Ocean along with the evolutionary changes observed in the native variety show that C. nucifera has provided a suitable context for density dependent natural selection to occur.
The difference in my interpretation compared to Gardner's doesn't lie in the large difference in body size of the two Phelsuma species in the Praslin group. Larger body size may have evolved as a result of intense interference competition for resources. The difference lies in the reason why P. s. longinsulae has an intermediate body size in the Mahe group. Gardner believes that the lack of the defendable male palm inflorescences eliminates the need for larger body size. Given the single evolutionary lineage present in the Seychelles I conclude, upon reexamination, that the ecological data are consistent with the interpretation that P. s. longinsulae evolved as a solitary species to use a wider resource distribution.
If P. s. longinsulae body-size is not at evolutionary equilibrium due to the selective effects of resource competition, why hasn't it reached equilibrium? Rapid body size shifts are not unknown. Large body size shifts have occurred in black tiger snakes (Notechis ater tiger) within a 10,000 year time-frame on islands off south Australia in response to a shift in prey size (Schwaner and Sarre 1988, 1990). One possible reason the rate of body-size evolution may be decreased could be an increase in negative fitness from another genetically correlated trait (Lande and Arnold 1984). Charlesworth (1984) provides one analytical derivation showing that this rate reduction is indeed possible.
A second reason P. s. longinsulae retains its ancestral size may be character convergence. It is possible to show (MacArthur and Wilson 1967; Schoener 1969; Cody 1973; Abrams 1987) that two species competition can lead to character convergence. There is, however, only a narrow window of character values within which this phenomenon can occur and the convergence end-state would be unstable (Case, pers. comm.). The result would be extinction of one of the species or evolution of resource use in another resource dimension.
TABLE 5. Population changes associated with single- and two-species islands. The two density columns are numbers per hundred trees. Lines indicate the species was absent. The % Flower columns represent the percentage of observations in coconut trees where individuals were first seen on male inflorescence (from Gardner 1984).
Density Density % Flower % Flower Island sundbergi astriata sundbergi astriata
Praslin 126 28 44.8 0.0 La Digue 403 117 58.9 4.8 Cousin - 120 - 21.2 Cousine - 153 - 28.6 Aride - 119 - 40.0 Round - 385 - 19.5 Mahe 183 198 48.6 45.1 Silhouette 52 643 20.0 53.2 North 410 - 35.0 -
The speciation mechanism depicted in Figure 11 is virtually the same as the models proposed for the origin and radiation of a number of species including the Darwin's Finches (subfamily Geospizinae) of the Galapagos islands (Lack 1945, 1947; Hamilton and Rubinoff 1963; Grant 1981, 1984), the origin of species in the lizard genus Anolis from Puerto Rico (Williams 1972) and several Pacific island bird radiations (Diamond 1977). In these systems the belief is that archipelagos were originally colonized by a single ancestral species and diversification was driven by a number of rounds of isolation by colonization, followed by evolutionary divergence in isolation and upon secondary contact. Differences were acquired in allopatry and were built upon by the effects of competition or ecological release. The radiation underway in the Seychelles is small compared to the radiations above and only evolutionary time can tell what will happen next. It is possible that another round of speciation could take place in the Seychelles system on more remote islands such as Silhouette.
The importance of history, in the form of eustatic sea-level changes, and ecology in driving speciation is reflected by other lizard taxa in the Seychelles. The endemic genus Ailuronyx is an arboreal, nocturnal gecko that occurs across the granitic islands. Previous lizard studies of the Seychelles recognize only one species, Ailuronyx seychellensis. The body size of this species is 110 mm or greater (Cheke 1984). My collections and others indicate that at least two species exist in the Seychelles and possibly a third. I collected two mature male individuals from La Digue with SVL of 78 and 85 mm. Henkel and Zobel (1987) and McKeown (pers. comm.) report collecting both A. seychellensis and an undescribed smaller species from Praslin. Preliminary cytochrome b sequence data for these large and small forms (Radtkey, unpubl. data) shows divergence levels similar to other gecko species ([approximately equal to] 19%).
Several skink genera are undergoing mini-radiations as well. Two monotypic genera, Janetaescincus and Pamalaescincus, are considered by Greer (1970 a, b) to have morphology convergent with the African genus Scelotes. Given the secretive habits of these small skinks, and the general morphological conservatism of skinks within genera (e.g., Hutchinson et al. 1990), it is reasonable to believe that other cryptic species exist in these genera. A second species of Pamalaescincus is suggested by the fact that the population of P. gardineri from Praslin is smaller and diurnal whereas the population from Cousin is larger and diurnal (Cheke 1984). Finally, one large and one small diurnal species of Mabuya, M. wrightii, and M. seychellensis, respectively, occur sympatrically on several islands.
The Phelsuma species in the Comoros and Mascarene archipelagos may also represent monophyletic units. The similarity to the Seychelles is increased even more by the historical presence of two skink species, Leiolopisma telfairii (small) and L. mauritianus (large), in the Mascarenes (Arnold 1980). Two final examples of within archipelago lizard radiation include the Mabuya/Macroscincus skinks of the Cape Verde Islands (Greer 1976) and the Lacertid genus Gallotia (4 species) from the Canary Islands. The existence of these and other monophyletic lizard radiations in small isolated archipelagos intimates the existence of an underlying species generating mechanism.
Convergence of independently evolved communities implies that similar, nonrandom forces are structuring the system (Recher 1969; Cody 1974; Pianka 1975, 1986). Schoener (1988) gives two criteria necessary for conducting meaningful convergence studies. First, the two systems must contain phylogenetically distinct species. Second, environmental conditions in the two systems must be similar. Both criteria above are met for a comparative study involving the well studied Anolis of the Caribbean and Phelsuma. Losos (1990a,b,c, 1992) presents a far more extensive combination of historical and ecological approaches than presented here in his integrated studies of the evolutionary ecology of Anolis lizards. He found cases of size assortment, character displacement, convergent evolution of species and convergent faunal build-up of community structure. More ecological information is needed from more species rich Phelsuma systems (e.g., the Comoros, Madagascar) before meaningful comparisons can be made.
Comparisons between the radiation of Phelsuma across the Indian Ocean and the Anolis radiation will be important in a time when some workers claim that community ecology is always system-specific and generalizations are dangerous or impossible. Parallel results will allow more concrete statements on the generality of the sequence of changes inferred to occur during faunal build-up (e.g., body size changes first, microhabitat changes second, macrohabitat third, etc.), the importance of species interactions for augmenting faunal build-up, and the evolutionary lability that exists to respond to ecological interactions.
1. Historical and ecological data were used to determine whether body size differences observed from three taxa of day-geckos (Phelsuma) from the Seychelles archipelago could be due to character displacement.
2. A phylogeny inferred from restriction enzyme and sequence analysis of mtDNA was used to show that the three taxa of Phelsuma found in the Seychelles archipelago (P. astriata, P. s. sundbergi, P. s. longinsulae) are monophyletic.
3. Phelsuma astriata, P. s. sundbergi, and P. s. longinsulae have significantly different body size distributions. Because these three taxa are monophyletic parsimony dictates that all body size evolution occurred in the Seychelles.
4. Comparisons of genetic distances between the Mahe and Praslin island groups suggests that P. astriata has invaded the Mahe island group very recently in geologic time. Therefore, the body size distribution of P. s. longinsulae can be considered to be that of a solitary species. Further, body size distributions observed in populations of P. astriata and P. s. sundbergi from the Praslin island group have been reached while sympatric.
5. Ecological information observed by Gardner (1984) was used to show that similar sized Phelsuma compete for resources and resource distributions do not appear to be correlated to body size distributions.
6. When these conclusions are considered in terms of the necessary criteria for character displacement (Schluter and McPhail 1992; Taper and Case 1992a) frequency dependent natural selection has likely played an important role in shaping the body size distributions and speciation of the Seychelles' day-geckos.
I am grateful to the government of the Republic of the Seychelles for their permission to conduct this research. I am also grateful to the Island Development Corporation for access and transportation to Silhouette, and the Seychelles Island Foundation for access to Cousin. Many individuals in the Seychelles provided assistance. These include D. Neufield, E. Ruku, M. St. Ange, N. Shah, K. Shah, and E. Smith. Technical and laboratory support were provided by K. Artzt, the late D. Bennett, T. Case, and H. Forrest. Specimens were provided by the British Museum (P. andamanensis), U.S. National Museum (P. leiogaster), R. Fisher, O. Pronk, and T. Tytle. Assistance with statistical analyses of body sizes was provided by E. Bruna and D. Vollmer. R. Buskirk, D. Hillis, and B. Milligan provided valuable comments on the manuscript. Financial support, scientific discussions, and philosophical advise were provided by T. Case and E. Pianka. This work was also supported by the National Science Foundation grant BSR-9016403.
ABRAMS, P. A. 1987. Alternative models of character displacement and niche shift. 2. Displacement when there is competition for a single resource. Am. Nat. 130:271-282.
ARNOLD, E. N. 1980. Recently extinct reptile populations from Mauritius and Reunion, Indian Ocean. J. Zool. Soc. Lond. 191: 33-47.
BAUM, D. A., AND A. LARSON. 1991. A phylogenetic methodology for studying character macroevolution. Syst. Zool. 40:1-18.
BERG, W. J., AND S. D. FERRIS. 1984. Restriction endonuclease analysis of salmonid mitochondrial DNA. Can. J. Fish. Aquat. Sci. 41:1041-1047.
BERRY, E. W. 1926. Cocos and Phymatocaryon in the Pliocene of New Zealand. Am. J. Sci. 12:181-184.
BlSWAS, S., AND D. P. SANYAL. 1977. Notes on the reptilia collection from the Great Nicobar Expedition in 1966. Rec. of the Zool. Survey of India 72:107-124.
-----. 1980. A report on the reptilia fauna of Andaman and Nicobar Islands in the collection of Zoological Survey of India. Rec. of the Zool. Survey of India 77:255-292.
BORNER, A. R. 1972. Revision of the Gekkonid genus Phelsuma Gray 1825. Saurol. (Cologne) 1:145.
BORNER, A. R., AND W. MINUTH. 1984. On the taxonomy of the Indian Ocean lizards of the Phelsuma madagascariensis species group (Reptilia, Gekkonidae). J. Bombay Nat. Hist. Soc. 81: 242-281.
BRAITHWAITE, C. J. R. 1984. Geology of the Seychelles. Pp. 17-38 in D. R. Stoddart, eds Biogeography of the Seychelles Islands. W. Junk, The Hague, Netherlands.
BROOKS, D. R., AND D. A. MCLENNAN. 1991. Phylogeny, ecology, and behavior: A research program in comparative biology. University of Chicago Press, Chicago.
CARR, S. M., A. J. BROWN, AND A. C. WILSON. 1987. Evolutionary inferences from restriction maps of mitochondrial DNA from nine taxa of Xenopus frogs. Evolution 41:176-188.
CASE, T. J. 1979. Character displacement and coevolution in some Cnemidophorus lizards. Fortschr. Zool. 25:235-282.
CASE, T. J., AND M. L. CODY. 1987. Testing theories of island biogeography. Am. Sci. 75:402-411.
CHARLESWORTH, B. 1984. The cost of phenotypic evolution. Paleobiology 10:319-327.
CHEKE, A. S. 1982. Phelsuma Gray 1825 in the Seychelles and neighboring islands: A reappraisal of their taxonomy and description of two new forms. Senkenbergiana. Biol. 62:181-198.
-----. 1984. Lizards of the Seychelles. Pp. 487-504 in D. R. Stoddart, ed. Biogeography and ecology of the Seychelles Islands. W. Junk, The Hague, Netherlands.
CODDINGTON, J. A. 1988. Cladistic tests of adaptational hypotheses. Cladistics 4:3-22.
CODY, M. L. 1973. Character convergence. Annu. Rev. Ecol. Syst. 4:189-211.
-----. 1974. Competition and the structure of bird communities. Princeton University Press, Princeton, NJ.
COFFIN, M. F., AND P. D. RABINOWITZ. 1987. Reconstruction of Madagascar and Africa: Evidence from the Davie Fracture Zone and Western Somali Basin. J. Geophys. Res. 92:9385-9406.
COOKE, F. C. 1958. Report on the coconut industry. Government Printing Office. Victoria, Mahe, Australia.
CRAWFORD, C. M., AND R. S. THORPE. 1979. Body temperatures of two geckos (Phelsuma) and a skink (Mabuya) in Praslin, Seychelles. Br. J. Herpetol. 6:25-31.
-----. 1981. Identification, morphometrics and cytogenetics of green geckos (Phelsuma) from Praslin, Seychelles (with notes on P. longinsulae, Mahe). Biol. J. Linn. Soc. 15:29-37.
DEBRY, R. W., AND N. A. SLADE. 1985. Cladistic analysis of restriction endonuclease cleavage maps within a maximum-likelihood framework. Syst. Zool. 34:21-34.
DIAMOND, J. M. 1977. Continental and insular speciation in Pacific land birds. Syst. Zool. 26:263-268.
-----. 1986. Overview: Laboratory experiments, field experiments, and natural experiments. Pp. 3-22 in J. Diamond and T. J. Case, eds. Community ecology. Harper and Row, New York.
DUROCHER-YVON, F. 1953. The coconut industry of the Seychelles. World Crops 5:437-441.
EFRON, B. 1982. The jackknife, the bootstrap, and other resampling plans. CNBS-NSF Regional Conference Series in Applied Mathematics, No. 38. Soc. for Ind. Appl. Math., Philadelphia, PA.
EVANS, P. G. H., AND J. B. EVANS. 1980. The ecology of lizards on Praslin Island, Seychelles. J. Zool., Lond. 191:171-192.
FELSENSTEIN, J. 1978. Cases in which parsimony or compatibility methods will be positively misleading. Syst. Zool. 27:401-410.
-----. 1985. Confidence limits on phylogenies: An approach using the bootstrap. Evolution 39:783-791.
-----. 1992. Phylogenies from restriction sites: A maximum-likelihood approach. Evolution 46:159-173.
-----. 1993. PHYLIP (Phylogeny Inference Package) vets. 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle.
FORSTER, R. 1975. The geologic history of the sedimentary basin of southern Mozambique, and some aspects of the origin of the Mozambique Channel. Palaeogeogr. Palaeoclimat. Palaeoecol. 17:267-287.
GARDNER, A. S. 1984. The evolutionary ecology and population systematics of day-geckos Phelsuma in the Seychelles. Ph.D. diss. University of Aberdeen, Scotland.
-----. 1987. The systematics of the Phelsuma madagascariensis species group of day geckos (Reptilia: Gekkonidae) in the Seychelles. Zoological. J. Linn. Soc. 41:93-105.
GRANT, P. R. 1981. Speciation and adaptive radiation of Darwin's Finches. Am. Sci. 69:653-663.
-----. 1984. Recent research on the evolution of land birds on the Galapagos. Biol. J. Linn. Soc. 21:113-136.
GREER, A. E. 1970a. A subfamilial classification of scincid lizards. Bull. Mus. Comp. Zool. Harvard 139:151-184.
-----. 1970b. The systematics and evolution of the subsaharan Africa, Seychelles, and Mauritius scincine Scincid lizards. Bull. Mus. Comp. Zool. Harvard 140:1-24.
-----. 1976. On the evolution of the giant Cape Verde scincid lizard Macroscincus coctei. J. Nat. Hist. 10:691-712.
HAMILTON, T. H., AND I. RUBINOFF. 1963. Isolation, endemism, and multiplication of the species in the Darwin Finches. Evolution 17:388-403.
-----. 1967. On predicting insular variation in endemism and sympatry for the Darwin's Finches in the Galapagos Archipelago. Am. Nat. 101:161-172.
HENDY, M. D., AND D. PENNY. 1982. Branch and bound algorithms to determine minimal evolutionary trees. Math. Biosci. 59:277-290.
HENKEL, F.-W., AND R. ZOBEL. 1987. Zur kenntbisn des bronzegeckos, Ailuronyx seychellensis (Dumeril and Bibron 1836). Herpetofauna 9:12-14.
HUTCHINSON, M. N., S. C. DONNELLAN, P. R. BAVERSTOCK, M. KREIG, S. SIMMS, AND S. BURGIN. 1990. Immunological relationships and generic revision of the Australian lizards assigned to the genus Leiolopisma (Scincidae:Lygosominae). Aust. J. Zool. 38:535-554.
INNIS, M. A., D. H. DELFAND, J. J. SNINSKY, AND T. J. WHITE, EDS. 1990. PCR protocols. Academic Press, San Diego, CA.
JUVIK, J. O., AND A. P. AUSTRING. 1979. The Hawaiian avifauna: biogeographic theory in evolutionary time. J. Biogeogr. 6:205-224.
KARUNAKARAN, C. 1962. Geology of the Andaman and Nicobar Islands. Andaman and Nicobar Info. (Port Blair) 10:64-75.
-----. 1967. Exhibition on the Great Nicobar Island. Geological Survey of India 1-14.
KAUL, K. N. 1951. A palm fruit from Kapurdi (Jodhpur, Rajasthan Desert): Cocos sahnii sp. nov. Current Sci. (Bangalore) 20:138.
KISHINO, H., AND M. HASEGAWA. 1989. Evaluation of the maximum-likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29:170-179.
KOCHER, T. D., W. K. THOMAS, A. MEYER, S. V. EDWARDS, S. PAABO, F. X. VILLABLANCA, AND A. C. WILSON. 1989. Dynamics of mitochondrial DNA evolution in animals: Amplification and sequencing with conserved primers. Proc. Nat. Acad. Sci. USA 86:6196-6200,
LACK, D. 1945. The Galapagos finches (Geospizinae): A study in variation. Occ. Pap. Calif. Acad. Sci. 21:1-159.
-----. 1947. Darwin's Finches. Cambridge University Press, Cambridge.
LANDE, R., AND S. J. ARNOLD. 1984. The measurement of selection on correlated characters. Evolution 37:1210-1226.
Losos, J. B. 1986. Island biogeography of day-geckos (Phelsuma) in the Indian Ocean. Oecologia 68:338-343.
-----. 1990a. Concordant evolution of locomotor behaviour, display rate and morphology in Anolis lizards. Anita. Behav. 39: 879-890.
-----. 1990b. Ecomorphology, performance capability, and scaling of West Indian Anolis lizards: An evolutionary analysis. Ecol. Monogr. 60:369-388.
-----. 1990c. A phylogenetic analysis of character displacement in Caribbean Anolis lizards. Evolution 44:1189-1203.
-----. 1992. The evolution of convergent structure in Caribbean Anolis communities. Syst. Zool. 41:403-420.
LOVERIDGE, A. 1942. Revision of the afro-oriental geckos of the genus Phelsuma. Bull. Mus. Comp. Zool. 89:438-482.
MACARTHUR, R. H., AND R. LEVINS. 1967. The limiting similarity, convergence, and divergence of coexisting species. Am. Nat. 101:377-385.
MACARTHUR, R. H., AND E. O. WILSON. 1963. An equilibrium theory of insular zoogeography. Evolution 17:373-387.
-----. 1967. The theory of island biogeography. Princeton University Press, Princeton, NJ.
MAY, R., M., AND R. H. MACARTHUR. 1972. Niche overlap as a function of environmental variability. Proc. Nat. Acad. Sci. USA 69:1109-1113.
MERTENS, R. 1962. The species and subspecies of the gecko genus Phelsuma. Senckenb. Bio. 43:81-127.
-----. 1966. The non-malagasian species and subspecies of the gecko genus Phelsuma. Senckenb. Bio. 47:85-110.
MONTAGGIONI, L. F., AND C. T. HOANG 1988. The last interglacial high sea level in the granitic Seychelles, Indian Ocean. Palaeogeogr. Palaeoclimat. Paleoecol. 64:79-81.
NEI, M., AND F. TAJIMA. 1983. Maximum likelihood estimation of the number of nucleotide substitutions from restriction site data. Genetics 105:207-217.
PALUMBI, S., A. MARTIN, S. ROMANO, W. O. MCMILLAN, L. STICE, AND G. GRABOWSKI. 1991. The simple fool's guide to PCR, ver. 2.0. Dept. Zool. Kewalo Marine Lab, Honolulu, HI.
PEAKE, J. F. 1971. The evolution of terrestrial faunas in the western Indian Ocean. Phil. Trans. Roy. Soc. Lond. B Biol. Sci. 260: 581-610.
PENNY, D., AND M. D. HENDY 1985. Testing methods of evolutionary tree construction. Cladistics 1:266-278.
PIANKA, E. R. 1973. The structure of lizard communities. Annu. Rev. Ecol. Syst. 4:53-74.
-----. 1975. Niche relations and desert lizards. Pp. 292-314 in M. Cody and J. Diamond, eds. Ecology and evolution of communities. Harvard University Press, Cambridge, MA.
-----. 1986. Ecology and natural history of desert lizards. Princeton University Press, Princeton, NJ.
PRESTON, F. W. 1962. The canonical distribution of commonness and parity. Ecology 43:185-215.
RADTKEY, R. R., S. C. DONNELLAN, R. N. FISHER, C. MORITZ, K. A. HANLEY, AND T. J. CASE. 1995. When species collide: The origin and spread of an asexual species of gecko. Proc. R. Soc. Lond. B Biol. Sci. 259:145-152.
RAO, A. R., AND V. K. MENON. 1964. Palmoxylon parthasarathyi sp. nov., a petrified palm stem from Mohgaon Kalan. Paleobotanist 12:1-6.
RECHER, H. F. 1969. Bird species diversity and habitat diversity in Australia and North America. Am. Nat. 103:75-80.
RENDAHL, H. 1939. The herpetology of the Seychelles. I. Reptiles. Zool. Jahrb. Abt. Syst., Oekol. Geogr. Tiere 72:157-328.
RICKLEFS, R. E. 1988. Speciation and diversity: the integration of local and regional processes. Pp. 599-622 in P.S. Giller and J. Gee, eds. Organization of communities: Past and present, Blackwell, London.
ROUGHGARDEN, J. 1972. Evolution of niche width. Am. Nat. 106: 683-718.
-----. 1976. Resource partitioning among competing species - a coevolutionary approach. Theor. Pop. Biol. 9:388-424.
-----. 1983. Competition and theory in community ecology. Am. Nat. 122:583-601.
-----. 1988. The structure and assembly of communities. Pp. 203-226 in J. Roughgarden, R. M. May, and S. A. Levin, eds. Perspectives of ecological theory, Princeton University Press, NJ.
ROUGHGARDEN, J., S. D. GAINES, AND S. W. PACALA. 1987. Supply-side ecology: The role of physical transport processes. Pp. 459-86 in P. S. Giller and J. Gee, eds. Organization of communities: past and present. Blackwell, London.
RUSSELL, A. P. 1977. The genera Rhoptropus and Phelsuma (Reptilia: Gekkonidae) in southern Africa: A case of convergence and a reconsideration of the biogeography of Phelsuma. Zool. Afr. 12:393-408.
SAHNI, B. 1946. A silicified Cocos-like palmstem, Palmoxylon (Cocos) sundaram, from the Deccan Intertrappean beds. J. Indian Bot. Soc. Iyengar Commemorative Volume:361-374.
SANGER, F. S., S. NICKLEN, AND A. R. COULSON. 1977. DNA sequencing with chain terminating inhibitors. Proc. Nat. Acad. Sci. USA 74:5463-5467.
SAUER, J. D. 1967. Plants and Animals on the Seychelles Coast: A study in historical biogeography. University of Wisconsin Press, Madison.
SCHLUTER, D., AND J. D. MCPHAIL. 1992. Ecological character displacement and speciation in sticklebacks. Am. Nat. 140:85-108.
SCHMIDT, K. P. 1934. The reptiles of the Pulitzer Angola expedition. Annu. Carnegie Mus. 22:1-15.
SCHOENER, T. W. 1967. The ecological significance of sexual dimorphism in size in the lizard Anolis conspersus. Science 155: 474-477
-----. 1968. The Anolis lizards of Bimini: resource partitioning in a complex fauna. Ecology 49:704-726.
-----. 1969. Models of optimal size for solitary predators. Am. Nat. 103:277-313.
-----. 1988. Ecological interactions. Pp. 255-297 in A. A. Myers and P. S. Giller, eds. Analytical biogeography, Chapman and Hall, London.
SCHWANER, T. D., AND S. D. SARRE. 1988. Body size of tiger snakes in Southern Australia, with particular reference to Notechis ater serventyi. J. Herpetol. 22:24-33.
-----. 1990. Body size and sexual dimorphism in mainland and island tiger snakes. J. Herpetol. 24:320-323.
SIMPSON, E. H. 1949. Measurement of diversity. Nature 163:688.
SWOFFORD, D. L. 1991. PAUP: Phylogenetic Analysis Using Parsimony, ver. 3.0s. Il. Nat. Hist. Sur., Champaign.
SWOFFORD, D. L. AND G. J. OLSEN. 1990. Phylogeny reconstruction. Pp. 411-501 in D. M. Hillis and C. Moritz, eds. Molecular systematics. Sinauer, Sunderland, MA.
TAPER, M. L., AND t. J. CASE. 1992a. Coevolution among competitors. Pp. 63-109 in D. Futuyma and J. Antonovics, eds. Oxford series in evolutionary biology, Vol. 8. Oxford University Press, Oxford.
-----. 1992b. Models of character displacement and the robustness of taxon cycles. Evolution 46:317-333.
TARLING, D. H. 1981. Models for the fragmentation of Gondwana. Pp. 261-266 in M. M. Cresswell and P. Villa, eds. Fifth international Gondwana symposium. Balkema, Rotterdam, Wellington, New Zealand.
TAYLOR, J. D., C. J. R. BRAITEWAITE, J. F. ARNOLD, AND E. N. ARNOLD. 1979. Terrestrial faunas and habitats of Aldabra during the late Pleistocene. Phil. Trans. Roy. Soc. Lond. B Biol. Ser. 286:47-66.
THOMPSON, J., AND A. WALTON. 1972. Redetermination of chronology of Aldabra Atoll by 230Th/234U dating. Nature, London 240:145-146.
THORPE, R. S. 1983. A biometric study of the effects of growth on the analysis of geographic variation: Tooth number in the Green geckos (Reptilia: Phelsuma). J. Zool. Lond. 201:13-26.
-----. 1985. A numerical analysis of isoelectric focused keratin monomers: Affinities of some western Indian Ocean green geckos. Biochem. Syst. Ecol. 13:63-69.
-----. 1986. Evolution and character congruence in some western Indian Ocean Phelsuma: Numerical analysis of biochemistry, shape and scalation. J. Zool. Lond (A) 208:429-441.
THORPE, R. S., AND C. M. CRAWFORD. 1979. The comparative abundance and resource partitioning of two species of green gecko species (Phelsuma) on Praslin, Seychelles. British J. Herpetol. 6:19-24.
THORPE, R. S., AND M. R. GIDDINGS. 1983. The biochemistry of keratin in green geckos (Phelsuma) and its systematic value. Herpetologica 39:399-410.
TIMURA, K. AND T. AOTSUKA. 1988. Rapid isolation method of animal mitochondrial DNA by the alkaline lysis procedure. Bio-chem. Genetics 26:815-823.
WILLIAMS, E. E. 1972. The origin of faunas: Evolution of lizard congeners in a complex island fauna: A trial analysis. Evol. Biol. 6:47-89.
-----. 1983. Ecomorphs, faunas, island size, and diverse end points in island radiations of Anolis. Pp. 326-370 in R. B. Huey, E. R. Pianka, and T. W. Schoener, eds. Lizard ecology: Studies of a model organism. Harvard University Press, Cambridge, MA.
|Printer friendly Cite/link Email Feedback|
|Author:||Radtkey, Ray R.|
|Date:||Apr 1, 1996|
|Previous Article:||The paradox of the phylogeny: character displacement of analyses of body size in island Anolis.|
|Next Article:||Genetic evidence for reproductive isolation and multiple origins of sympatric trophic ecotypes of whitefish (Coregonus).|