As had been hypothesized based on allozyme data (Fisher, 1997), there are two cryptic lineages within G. mutilata. Levels of differentiation at the 16s rRNA marker (~10%) are higher or equivalent to differentiation observed between known distinct species of geckos (Jesus et al., 2005; Rocha et al., 2005; Weiss and Hedges, 2007).
Our data show that Western Indian Ocean populations (Seychelles, Mascarenes, Madagascar) belong to the same lineage as individuals analysed from the Philippines (Luzon Island) and mainland Asia (Myanmar and Malaysia).
Fisher (1997) observed the coexistence of two lineages in the Philippines (and Thailand) that possibly correspond to the ones that are now evident at the mtDNA level, with the Negros Island population belonging to the ‘southern mainland type’, also widespread across the Pacific, and the other two populations (Luzon and Caminguin) belonging to the same lineage as the two northern Thailand populations (the ‘northern mainland type’). The Philippine material included in our study is from Luzon Island, which is also the type locality of G. mutilata (Manila), and belongs to the same clade of the remaining Southern Asia and Indian Ocean individuals. On the basis of these data, we argue that they should all be referred to as Gehyra mutilata sensu stricto with Hemidactylus peronii Duméril and Bibron, 1836 (type locality Mauritius), Gecko pardus Tytler, 1865 (type localities Rangoon, Moulmein and Port Blair) and Peropus packardii Cope, 1869 (type locality Penang) as synonyms. Indonesian populations (on which the names Hemidactylus platurus Bleeker, 1859 and Gehyra beebei Annandale, 1913 are based) were not included in our study and their status cannot yet be assessed. If future work leads to the recognition of the Oceanian population as a separate species, priority shall be given to the name Dactyloperus (= Gehyra) insulensis Girard, 1857 (type locality Sandwich Islands). We suggest however, that a formal nomenclatural change be delayed pending further studies on the effective isolation of the two lineages and the geographic structure of genetic variation.
Remarkably, little or no variation was found within the ‘Indian Ocean and Southern Asia’ and ‘Pacific’ lineages, respectively, indicating a recent and fast spread of the species throughout these areas, probably from sources located in Southern Asia. The hypothesis of a recent, human-mediated, dispersal of G. mutilata across Indian Ocean islands is supported by the observation that records of G. mutilata are either recent (for Madagascar, the Mascarenes and the Seychelles - Glaw and Vences, 1994; Gerlach, 2007; Cheke and Hume, 2008) or absent (for the Comoros - Carretero et al., 2005) across this region.
Despite the near-complete absence of variation, the two haplotypes constituting the ‘Pacific lineage’ exhibit some geographic structure, being exclusive of either Western populations (comprising Palau, New Guinea and Micronesia – Mariana and Caroline) or Eastern Pacific populations (Polynesia, Vanuatu and Clipperton). This may indicate that two colonisation routes existed, although a single colonisation followed by limited gene flow would also explain the data. Interestingly, the ‘two routes’ pattern would fit Fisher’s (1997) ‘non-European’ human-mediated dispersal hypothesis involving two - Australoid and proto-Polynesian - Pacific colonisations. However, the Melanesian populations from Vanuatu, which share the Eastern and not the Western Pacific haplotype, do not fit the picture. The only specimen from Vanuatu used in this study was recently collected on Torres Islands, in northern Vanuatu, where the species was locally abundant, thus possibly reflecting a recent introduction. Previous records from Vanuatu are very scarce (only one record from a single specimen from Efate, collected in 1924 or 1925 - Medway and Marshall, 1975; Cranbrook and Pickering, 1981) and it is possible that both Western and Eastern lineages now occur on that archipelago, the Eastern one being very recently introduced. It would be interesting to compare Fijian and Solomon Islands populations with Vanuatu to test conflicting biogeographical hypotheses for that area (Bauer, 1988).
Clipperton Atoll populations belong to the Eastern Pacific lineage and were generally considered as being recently introduced from Mexico (Lorvelec and Pascal, 2006). However, a natural occurrence on the atoll cannot be excluded. Likewise, the origin of Mexican populations is still under debate (Ineich and Blanc, 1987).
It is well known that G. mutilata is an effective island colonist possessing several morphological and ecological adaptations that facilitate successful inter-island travel (see Ineich and Blanc, 1987). It even recently recolonized Krakatau Island in Indonesia, which was completely covered by lava after an historical volcanic eruption (O’Shea and Cook, 2006) and is also present in southern Japan (Ota and Yamashita, 1985). Genetic uniformity of both lineages, across the Indian Ocean on the one hand and the Pacific on the other, is classically interpreted as proof of their recent spread through human agency (see e.g., Bruna et al., 1996). However, given that human colonisation of Pacific islands is also extremely recent (Belwood, 1997; Blust, 1995; Kirch, 1997), the hypothesis that this species was already present on the islands before man cannot be rejected with present data, i.e., colonisation can be natural, and ‘pre-human’, but still recent, and thus without time for organisms to exhibit geographical patterns of genetic variation that would allow the trace of the colonisation routes. Across most of the Pacific Islands, Gehyra mutilata occupies a wide range of habitats, and not only the ones related to man. It is usually not very abundant but geographically widespread and there are no clear gaps in its distribution in French Polynesia, as would be expected for a very recent ‘anthropogenic’ species (Ineich and Blanc, 1988).
Use of faster evolving genetic markers and further sampling along south-eastern territories between Myanmar and Indonesia/Papua New Guinea and comparison with proposed human migration routes should clarify this question and may possibly shed light on the causes of speciation within this apparent species complex.
The fact that G. mutilata is clearly a ‘newcomer’ across the major oceanic island areas is relevant for consideration of its effect on island ecosystems. As already discussed, the impact of alien species on islands can be considerable. The impact of G. mutilata on other Indian Ocean gecko populations is unknown. Across the Pacific this species is less abundant than many other introduced house geckos (e.g., several species of the genus Hemidactylus Gray, 1825) but it is nowadays widespread and abundant across the Seychelles, with established populations on both coralline and granitic islands. Bringsøe (2006; see also Rösler, 2007) hypothesised that nocturnal activity of G. mutilata (that he considered as recently introduced) had a significant impact on, and led to nocturnal activity of some, Seychelles day geckos such as Phelsuma sundbergi Rendahl, 1939. We do not agree with that hypothesis and rather think that well known behavioural plasticity of island reptiles (Ineich and Blanc, 1987) better explains this recent shift of P. sundbergi to exploit an easily obtained and abundant food resource located around artificial lights. Across Indian Ocean islands this species seems to be rapidly spreading: earliest G. mutilata records date back from 1885 (1905 for the first precise locality record - Victoria, Mahé Island - Gerlach, 2007), but this species is nowadays present on almost all islands of the archipelago, including the outer ones (Gerlach, 2007). Not exclusive to urban areas, G. mutilata in the Seychelles can be also found in more pristine vegetation and montane areas, and it has even been observed co-existing with the native gecko Urocotyledon inexpectata Stejneger, 1893 at the very specific habitat/nesting sites of that species: rocks covered with empty wasp nests in which the eggs are laid (S. Rocha, pers. obs., 2007). Interspecific competition may have an enormous impact on endemic gecko populations (see Cole et al., 2005 and references therein), such as Hemidactylus mercatorius (sensu Vences et al., 2004) on Aldabra (S. Rocha, pers. obs., 2007) or Urocotyledon inexpectata in the granitic Seychelles, and should be the target of specific monitoring.
Concerning Pacific island populations, we have found that they clearly avoid artificial lights, contrary to Indian Ocean G. mutilata. It is also noteworthy that they avoid humid forest habitats were they are rarely observed. Instead, they are more abundant in littoral dry forests (e.g., the Tuamotu atolls in French Polynesia), even when isolated and far from human habitations, but also occur in and around houses. They shelter by day under loose bark on dead trees, and lay two adhesive eggs (sometimes in communal laying sites) in the same microhabitat. They occupy more xeric habitats than G. oceanica (Lesson, 1839), a more common Pacific island congener, thus explaining its colonization success on the American continent, even in Mediterranean climates such as in California, and on some remote and dry areas like Clipperton Atoll in Eastern Pacific.
Competition with other species in the Pacific islands apparently does not occur, or at least not at a significant level. The impact of G. mutilata on native species, if really recently introduced, seems limited. Perhaps this can be considered as evidence of its slightly ‘older’ arrival in the Pacific, and establishment of equilibrium with other sympatric species. The rarity of G. mutilata on Pacific islands compared to the initial explosive demography of H. frenatus Duméril and Bibron, 1836 (Case and Bolger, 1991), suggests that its current impact on other local species is not significant and that the species does not constitute a major threat to other native species.