Thecadactylus

Bergmann, Philip J. & Russell, Anthony P., 2007, Systematics and biogeography of the widespread Neotropical gekkonid genus Thecadactylus (Squamata), with the description of a new cryptic species, Zoological Journal of the Linnean Society 149 (3), pp. 339-370 : 356-357

THECADACTYLUS

One of the basic questions addressed here is whether there are any consistently identifiable clades within Thecadactylus . A number of groupings were identified repeatedly throughout the analyses presented. Yucatán and the Belize / Guatemala OTU were clustered by all data sets except the morphometric, which did not include the latter locality. Morphological, molecular, and total evidence data sets also clustered Guyana with Yucatán + Belize / Guatemala, which clearly represents a geographically incoherent grouping.

The south-western Amazonian clade was consistenly recovered by analyses of all data sets. In the morphological data set this clade included Bolivia, Madre de Dios, Ucayali, Loreto, Iquitos, Peruvian Amazonas and Amazonian Ecuador (including Caqueta, Colombia). The molecular data set did not sample Ucayali, Iquitos or Peruvian Amazonas, but added Brazilian Amazonas and Brazilian Acre to the cluster ( Kronauer et al., 2005). The total evidence data set included all of these OTUs in this clade ( Figs 5 , 6 ). Therefore, the evidence supports two identifiable evolutionary lineages: a south-western Amazonian lineage, recognized as T. solimoensis , and a cluster consisting of the remainder of Thecadactylus , recognized as T. rapicauda .

Relationships between the OTUs within these clades are poorly resolved. Although one explanation for this might be an inadequate data set that includes inappropriate characters, obscuring patterns of clustering, we posit that this is not the case. A wide geographical distribution and extensive morphological polytypy make phylogenetic reconstruction and species delimitation difficult ( Morando et al., 2003). Wilson & Brown (1953) noted that widely distributed, highly variable species are difficult to subdivide because genetically independent morphological characters often exhibit geographically independent patterns of variation, character states recur multiple times in different portions of the range, and microgeographical races are often present. Systematic hypotheses in these situations are primarily dependent upon the characters that are considered, and are obscured as more characters are added ( Wilson & Brown, 1953). All of these problems were observed with Thecadactylus . For example, specimens from St. Martin often have a highly unusual spotted pattern, while those from St. Lucia, Grenada and Guyana tend to be very small, with minimal patterning (our pers. observ.). Lack of resolution and poor clade support can be used as supportive evidence of a single species. Wiens & Penkrot (2002) noted that one would expect hierarchical patterns of relationships within a species to be obscured, poorly supported, and geographically discordant due to gene flow and character recombination between populations. This is exactly what is observed with Thecadactylus , with the exception of the two identified clades.

Despite poor phylogenetic resolution it is reasonable to argue that Thecadactylus arrived on the Lesser Antilles through dispersal. As a result of the volcanic ( Rosen, 1976), Eocene ( Hedges, Hass & Maxson, 1992) origin of these islands, it is universally accepted that they represent a separate biogeographical unit from the Greater Antilles ( Rosen, 1976; Breuil & Masson, 1991) and that their fauna arose through dispersal, primarily from South America via ocean currents ( Pregill & Olson, 1981; Hedges, 1996a,b,c). It is likely that Thecadactylus dispersed to these islands during the Quaternary ( Hedges, 1996b). Late Pleistocene fossils of Thecadactylus found in caves on Barbuda ( Etheridge, 1964) are consistent with this hypothesis. A South American dispersal origin to the Lesser Antilles has also been advocated for the yellow warbler, Denroica petechia L. ( Klein & Brown, 1994), Sphaerodacytlus geckos ( King, 1962; Hass, 1991) and Eleutherodactylus frogs ( Kaiser, Green & Schmid, 1994; Kaiser, 1996).

When identifiable distinct clades of Thecadactylus are considered, several other biogeographical scenarios enter into consideration. The northern Middle American clade, consisting of Yucatán and Belize / Guatemala, is also repeatedly identified, but its southern boundaries are not identifiable because of current sampling incompleteness. Thecadactylus belongs to the humid herpetofaunal assemblage in Middle America, and so may be limited to more mesic areas in the region ( Wilson & McCranie, 1998). Also in this region, our data suggest continuity in the range of T. rapicauda between Yucatán localities and those further south ( Russell & Bauer, 2002a), consistent with new locality data of Luja & Calderon-Mandujano (2005).

Mountains play a vicariant role in the distribution of Thecadactylus solimoensis and are a significant barrier to gene flow. The boundaries of this species are the Andes Mountains to the west, and the Colombian Cordillera Oriental to the north-west. The Andes as a barrier to gene flow is supported by the exclusion of the Pacific Ecuador OTU from T. solimoensis and its inclusion in T. rapicauda . Comparatively, the Andes have been implicated as a major vicariant agency for the viper Lachesis muta L. ( Zamudio & Greene, 1997), the toad Bufo marinus L. ( Slade & Moritz, 1998), lizards of the genus Liolaemus ( Schulte et al., 2000) , and a number of marsupial and rodent clades ( da Silva & Patton, 1998).

The southern extent of Thecadactylus solimoensis is in Bolivia and Rondônia, at the southern extreme of the range of the genus ( Fig. 1 ; Russell & Bauer, 2002a), correlating well with the area outside of the ‘north-west arc’ identified by Vanzolini (1968). By contrast, the eastern boundary of T. solimoensis is very uncertain. This is in part due to inadequate sampling within the central Amazon basin, and in part due to conflicting evidence from phylogenetic analyses. Morphological phylogenetic analyses did not group the Brazilian Amazonas OTU with T. solimoensis , restricting its range to extreme western Amazonia. By contrast, molecular analyses placed the Brazilian Amazonian locality within T. solimoensis , extending its eastern boundary as far east as the western border of Pará ( Fig. 1 ). More credence is lent to the conclusions drawn from the molecular analyses because these were supported by more characters and by extremely robust clade support values (bootstrap values up to 100%, Pp up to 1.00). A similar western Amazonian area was identified for anurans ( Ron, 2000) and lizards ( Avila-Pires, 1995). Both of these areas correlate very closely with the range of Thecadactylus solimoensis . Furthermore, Avila-Pires’s (1995) eastern boundary of this area was also at the Pará–Amazonas border. A western Amazonian area similar to the range of T. solimoensis has also been identified for primates ( da Silva & Oren, 1996), but its western boundary is slightly further to the east, defined by the Rios Trombetas and Tapajós.

From the findings presented here, Thecadacytlus represents a widespread and highly variable taxon that exhibits low taxonomic diversity, but that consists of two species. This is similar to the situation seen in the pantropical sea urchin ( Lessios et al., 1999), which also consists of relatively few species. However, unlike the urchin, Thecadactylus is much more variable, but this variation, although geographically correlated, is continuous across the landscape and has few discrete breaks. This is the situation described by Wilson & Brown (1953) as characteristic of widespread species.

Figures

Figure 1. Distribution of Thecadactylus, showing ranges of the two constituent species, T. rapicauda (diagonal lines) and T. solimoensis sp. nov. (cross hatching). Modified from Russell & Bauer (2002a).

Figure 5. Majority rule bootstrap consensus tree of cladograms produced by MP analysis of the total evidence data set. Bootstrap values are presented below branches. SWA indicates the south-western Amazonian clade, which herein is described as T. solimoensis.

Figure 6. Majority rule consensus summary of trees sampled by the BI analysis of the total evidence data set including all OTUs (A) and including only those OTUs for which a majority of the data was available (B). Posterior probabilities are presented below the branches. SWA indicates the south-western Amazonian clade, which herein is described as T. solimoensis.