INTRODUCTION

Kluge (1987), based on the cladistic analysis of morphological characters, recognized “eyelid geckos” classified to the subfamily Eublepharinae of the family Gekkonidae to that date (Kluge, 1967), as constituting a single primitive lineage distant from all other members of the microorder Gekkota (Fig. 1). He thus revalidated family Eublepharidae Boulenger, 1883, to accommodate all eyelid geckos that sporadically occur in Eurasia, Africa and North America.

Fig. 1

Cladistic relationships of the major groups of Gekkota (exclusive of Pygopodidae) hypothesized by Kluge (1987). He also argued for (and depicted) the sister-group relationship of Pygopodidae with Diplodactylinae, but this account is not fully accepted (e.g., Estes et al., 1988).

Also on the basis of the cladistic analysis of morphological characters, Grismer (1988), while confirming the monophyly of the eublepharid geckos, formulated a detailed, comprehensive phylogenetic hypothesis for the family (Fig. 2) and revised the infrafamilial classification accordingly. Through this and two subsequent taxonomic works (Grismer, 1989, 1991), 18 species belonging to six genera of two subfamilies have been recognized for the family Eublepharidae.

Fig. 2

Cladistic relationships of the eublepharid genera hypothesized by Grismer (1988).

A few other studies demonstrated considerable chromosomal and eco-physiological variations among some eublepharid geckos (e.g., Murphy, 1974; Ota et al., 1987; Dial and Grismer, 1992), and these predict the effectiveness of approaches other than the external and anatomical analyses to the eublepharid phylogeny. For the improvement of a given phylogenetic hypothesis formulated on the basis of a certain set of characters, testing by using a different character set would be highly effective. Nevertheless, no approaches other than the above-mentioned morphological ones have ever been attempted toward the comprehensive phylogeny of this family or its relationship with other gekkotan groups.

We partially sequenced mitochondrial DNA (mtDNA) for representatives of all eublepharid genera and a few of other major groups of Gekkota (subfamilies Gekkoninae and Diplodactylinae of the family Gekkonidae), and analysed resultant sequences phylogenetically. In this paper, we compare our results with results of morphological analyses by previous authors, and discuss their systematic and biogeographical implications.

MATERIALS AND METHODS

A total of 17 specimens (three subspecies and seven species) representing all of the six eublepharid genera were used. Of these, six specimens of Goniurosaurus kuroiwae and two specimens of Aeluroscalabotes felinus were sampled in the field. The other specimens were obtained from pet dealers in life or in fresh dead state. These included two obviously conspecific eublepharid specimens of unknown origins that showed morphological features characteristic of Goniurosaurus (as defined by Grismer [1988]), but were distinct from either of the two known species of the genus (i.e., G. kuroiwae and G. lichtenfelderi: Grismer, 1988) (Fig. 3). We henceforth refer to these specimens as G. sp. One diplodactyline (Rhacodactylus trachyrhynchus) and two gekkonine geckos (Gekko gecko and Cyrtodactylus intermedius) of the family Gekkonidae were also subjected to this study. As outgroups, two scincids (Eumeces latiscutatus and Lygosoma bowringii) and one iguanid species (Iguana iguana) were chosen as representatives of Autarchoglossa (an extant sister-group of Gekkota) and a more basal lineage (Estes et al., 1988), respectively. Voucher specimens are deposited in the Department of Zoology, Kyoto University (KUZ: see Appendix-1).

Fig. 3

Apparently undescribed species of Goniurosaurus used in this study.

Livers, removed from anesthetized or dead specimens and stocked at −80°C, were used for DNA extraction. Each sample was processed for sequencing of 12S and 16S ribosomal RNA (rRNA) genes of mitochondrial DNA (mtDNA) following the method by Wada et al. (1992) as modified by Honda et al. (1999). These genes, functionally important and thus are supposedly fairly conservative, are expected to be informative for divergence back to about 300 million years ago (MYA) at maximum (Mindell and Honeycutt, 1990). Thus, sequence variation in these genes seems to be suitable for the phylogenetic reconstruction of extant Gekkota, which divergence is considered to have initiated some 180 MYA (Kluge, 1987).

Alignments for DNA sequences were determined based on maximum nucleotide similarity. Using the aligned sequences, we eliminated gap sites and prepared a pairwise matrix of distance by Kimura's (1980) two-parameter model for transition/transversion bias. The matrix was then subjected to the analysis by the neighbor-joining (NJ) method (Saitou and Nei, 1987) using Clustal W (Thompson et al., 1994). Degrees of supports for internal branches of each tree were assessed by 1,000 bootstrap pseudoreplications (Felsenstein, 1985). Parsimony analysis was also performed using the Heuristic algorithum of PAUP (version 3.11: Swofford, 1993). In this analysis, each nucleotide base was regarded as a character and four kinds of salts as different character states. The result of this analysis was expressed as a 50% majority-rule consensus tree. In both analyses, each specimen was set as constituting a single OTU by itself.

The interpretation of bootstrap proportions (BPs) is still controvertial (e.g., Felsenstein and Kishino, 1993; Hillis and Bull, 1993). We tentatively postulated BPs≥90% as strong support, 70≤BPs<90% as moderate support, 50≤BPs<70% as constituting limited evidence of monophyly, and BPs<50% as insufficent support for monophyly, following Shaffer et al. (1997).

RESULTS

Aligned sequences from two mtDNA genes are presented in Appendix-2. The 12S rRNA fragment consisted of 406 total aligned sites, of which 253 (62.3%) were variable. For the 16S rRNA fragment, there were 473 total sites, of which 259 (54.8%) were variable. Nucleotide replacements of the eublepharids with a diplodactyline and gekkonines varied from 174 base pairs (bp) (Holodactylus africanus vs Rhacodactylus trachyrhynchus) to 200 bp (Goniurosaurus sp. vs R. trachyrhynchus), and from 202 bp (H. africanus vs Cyrtodactylus intermedius) to 236 bp (G. kuroiwae kuroiwae vs Gekko gecko), respectively, whereas those between the latter two groups were from 197 bp (C. intermedius vs G. gecko) to 239 bp (R. trachyrhynchus vs G. gecko). Within Eublepharidae, the intergeneric and intrageneric replacements ranged from 137 bp (Hemitheconyx caudicinctus vs Holodactylus africanus) to 200 bp (Aeluroscalabotes felinus vs H. africanus), and from 7 bp (G. k. kuroiwae vs G. k. orientalis) to 159 bp (G. k. kuroiwae vs G. sp.), respectively.

Fig. 4 shows the NJ dendrogram derived from the OTU-distance matrix. In this analysis, no sufficient supports were obtained for the monophyly of the Eublepharidae, because of the inclusion of R. trachyrhynchus (Diplodactylinae of Gekkonidae) to the otherwise eublepharid cluster. Within Eublepharidae, Coleonyx and Aeluroscalabotes showed great, independent divergences, whereas the others appeared monophyletic (BP=76%). Sister-group relationships of the two African genera (Hemitheconyx and Holodactylus) of the latter was strongly supported (BP=98%), whereas their sister-group relationships with Eublepharis received a weaker support (BP=68%). On the other hand, monophyly of Goniurosaurus was even more weakly supported (BP=56%). All conspecific/consubspecific combinations received 98–100% BP supports. Of G. kuroiwae subspecies/populations examined here, G. k. splendens first diverged, and then did G. k. kuroiwae from the northern part of Okinawajima: G. k. kuroiwae from the southern part of Okinawajima and G. k. orientalis constituted a minimum monophyletic unit with high BP value (99%).

Fig. 4

NJ dendrogram deriving from Kimura's (1980) distance matrix from 12S and 16S rRNA sequence data. Numbers at branches indicate bootstrap proportions in 1,000 bootstrap pseudoreplications. Of the ingroup taxa, those belonging to the family Eublepharidae, and subfamilies Gekkoninae and Diplodactylinae of the family Gekkonidae are denoted by circles, triangles and a rectangle, respectively. N and S following G. k. kuroiwae indicate that those specimens were from the northern and the southern parts of Okinawajima, respectively (see Appendix-1 for further details). For each of other species/subspecies, different individuals, when examined, are discriminated by arabic numerals (1, 2). Scale bar equals 0.1 of the distance.

Results of the parsimony analysis showed no apparent inconsistensy with the relationships suggested by the NJ dendrogram. As are depicted in Fig. 5, relationships resulting from this analysis weakly but still substantially supported the closer diplodactyline affinity to eublepharids than to gekkonines (BP = 64%), and strongly supported the monophyly of African eublepharids (97%), greatest divergence of G. k. splendens among the conspecifics, and closer affinity of G. k. kuroiwae from southern Okinawajima to G. k. orientalis than to G. k. kuroiwae from northern Okinawajima (BP=100% for both).

Fig. 5

The majority-rule consensus tree deriving from parsimony analysis using nucleotide bases of 12S and 16S rRNA genes as characters and the four kinds of salts as different character states. See Fig. 4 for explanations of denotations.

DISCUSSION

As far as the intergeneric phylogeny within Eublepharidae is concerned, results of the present analyses showed no serious contradictions with the previous hypothesis formulated on the basis of a cladistic analysis of morphological data (Grismer, 1988) (Fig. 2). Our results, nevertheless, did not support the initial, most fundamental divergence of Aeluroscalabotes from the remaining eublepharids, which let Grismer (1988) establish the monotypic subfamily Aeluroscalabotinae against Eublepharinae. Further, more comprehensive analysis is needed to verify the validity of these subfamilies from the phylogenetic view point.

The relatively poor support for the closest affinity between the two Goniurosaurus species examined may be attributable to their relatively long isolation as is also implied by the highly relict distribution of the genus (Ota, 1998). This ad hoc assumption needs verifications by incorporating data for the other Goniurosaurus species, G. lichtenfelderi, into future analyses.

The relationship among populations of G. kuroiwae did not agree with their assignments to morphologically defined subspecies (Grismer et al., 1994), because it strongly suggests the genealogical proximity of the southern Okinawajima population of G. k. kuroiwae with G. k. orientalis rather than with the consubspecific population in the northern part of the island. The habitat of G. kuroiwae is largely confined to areas covered with humid vegetations, such as primary and well recovered secondary forests (Matsui and Ota, 1999). Thus, the broad, dry vegetational zone in the middle of Okinawajima (AJLSA, 1994), from which no G. k. kuroiwae have ever been found (Matsui and Ota, 1999), might have possibly been intervening the gene flow between populations on both sides. The small, but still substantial, genetic divergence between the G. k. kuroiwae populations from the northern and southern parts of the island seems to reflect a certain duration of such an isolation. Considerably close affinity of the southern Okinawajima population of G. k. kuroiwae with G. k. orientalis is puzzling, because, as long as the current geography is concerned, locality of the sample of orientalis examined here (Iejima) is closer to the northern part of Okinawajima. This may reflect a closer geohistorical relationship of Iejima with the southern part of Okinawajima than with the northern part.

The closer relationship of a diplodactyline species (R. trachyrhynchus) with eublepharids than with gekkonines (Gekko gecko and Cyrtodactylus intermedius), suggested by both NJ and parsimony analyses, is also puzzling because such a relationship needs to presume occurrences of either parallel derivations of a number of advanced character states shared by the diplodactyline and gekkonine geckos, or reversals of a number of characters to primitive states in the eublepharid lineage (Kluge, 1987; Grismer, 1988). Further analyses using more samples especially of representatives of Sphaerodactylinae (the other gekkonid subfamily: see Fig. 1), as well as of additional gekkonine and diplodactyline species, are strongly desired to confirm such diplodactyline relationship.

Kluge (1967) assumed that Southeast Asia was the center of origin of all extant gekkotan lineages and that diplodactyline geckos, having emerged there from eublepharid ancestor, dispersed into Australia through the Indo-Australian Archipelago during the late Mesozoic. With the subsequent general acceptance of the “mobile continents” and the revision of the cladistic hypothesis of Gekkota, he formulated a completely different scenario, in which divergences of eublepharids and diplodactylines from the others are, respectively, attributed to the breakup of Pangaea into Laurasia and Gondowanaland in 180 MYA, and more tentatively to the Africa-India and Australia-New Zealand separation of the latter in 85 MYA (Kluge, 1987). The secondary dispersal is assumed to explain the current occurrence of a few eublepharid genera in the Gondowana region (i.e., Hemitheconyx and Holodactylus in Africa) in this hypothesis. In the same year, King (1987a, b), based on the chromosomal and immunodistance data, and the fact that all major extant lineages occur in the Gondowana region, argued for the Gondowana origin of the whole Gekkota (Gekkonidae in his use). He considered the current Laurasian eublepharids (e.g., Aeluroscalabotes in Southeast Asia and Coleonyx in North America) as deriving from the secondary dispersal.

Based on a few morphological characters, Borner (1981) argued that Holodactylus in Africa is most primitive among the eublepharid genera. This account seems to lend some supports to King's (1987a, b) hypothesis. However, our results, along with results of morphological analysis by Grismer (1988), strongly suggest the most primitive status for the Laurasian Aeluroscalabotes and Coleonyx, and, contrary to the view of Borner (1981), most recent divergence of the Gondowanan Holodactylus, as well as of Hemitheconyx. These stongly suggest the Laurasian origin of the eublepharid geckos as was hypothesized by Kluge (1987). Closer affinity of Rhacodactylus with eublepharids suggested in our results may be indicative of the past dispersal of the diplodactyline ancestor to the Australia-New Zealand region directly from Laurasia (contra Kluge [1987] and King [1987a, b]).