We studied phylogeographic patterns of the giant orb web spider, Nephila, on the islands of the Indian Ocean, and investigated how these compare with related spiders and with other organisms known for long range, over ocean dispersal. We sampled Nephila on the islands Mayotte, Madagascar, Réunion, Mauritius and Rodrigues, and tested the route and timing of their oceanic origin, genetic structure among the islands, and the status of the enigmatic Rodrigues population, which is sometimes referred to as a separate species, N. ardentipes [19, 21]. We found phylogenetic and population genetic structures based on nuclear (ITS2) and mitochondrial markers (CO1) to be consistent with a dispersal model from the African mainland, and a recent and rapid colonization of the Mascarenes, or a recent termination of gene flow between Madagascar and Mascarene populations. Taxonomically, all targeted populations are thus best circumscribed as one species, N. inaurata, which invalidates N. ardentipes as a species. However, the Rodrigues population is monophyletic and has a unique ITS2 haplotype, and the other Mascarene islands have also started accumulating unique haplotypes. This might be consistent with a 'speciation in progress' scenario. Another possibility cannot be completely ruled out, that all haplotypes also occur on Madagascar but that we failed to sample Mascarene haplotypes from Madagascar. However, we find this unlikely given the structure of the data, the number of unique haplotypes on the Mascarenes, and the fact that Nephila is not synanthrophic in the area and thus not likely human transported.
Nephila has colonized at least 40 major islands and land masses globally (Table 1), and many species within this clade are extremely widespread. The best examples are the American N. clavipes spanning from North America through Central and South America into Argentina, the African mainland N. fenestrata and N. senegalensis covering most of the continent, and the Australasian N. pilipes, which ranges from India to Solomon Islands, and from Japan to Australia [20, 21, 24]. Such large ranges are unusual for invertebrates, and hint at excellent dispersal abilities and ecological success. Although not empirically observed to balloon [5, 25], we find ballooning, most likely at the earliest ontogenetic stages, to be the most logical explanation for Nephila colonizing remote islands. Since Nephila spiders are not synanthropic (as are, e.g., some species of the sister genus, Nephilengys), the travel among islands by human assistance is unlikely. This conclusion is supported by the timing of inferred colonization of the Indian Ocean islands, which vastly predates human settlement at slightly over 2000 years ago.
Perhaps the apparently high dispersal abilities have limited Nephila diversification by maintaining gene flow among even geographically distant populations. The taxonomy of Nephila is well studied and the genus is hypothesized to contain only 14 species globally [[21, 22]; this paper]. Although its sister genus, Nephilengys, is even less diverse globally, it invites a direct comparison within the region of study. We also studied the speciation patterns of the Nephilengys populations on the same islands , and found molecular, biogeographical and morphological evidence for three species: Ng. livida inhabiting Madagascar and Comoros, Ng. borbonica endemic to Réunion, and Ng. dodo from Mauritius. Nephila and Nephilengys show comparable biogeographic histories - both lineages occupied Madagascar from Africa between 2 and 5 Ma (although the upper bound estimates for Nephilengys are 13 Ma), and other smaller islands more recently, after which Nephilengys diversified through a lack of gene flow, while Nephila diverged less, with the only deep divergencies occurring between the African mainland and the population across the Indian Ocean as a whole (Table 2). Combined, the comparable timing of initial colonization of the Indian Ocean by both lineages and the subsequent speciation and lack of gene flow in Nephilengys rule out the potential human assisted travel to remote islands in Nephilengys.
The third nephilid spider genus inhabiting some of these islands is Clitaetra, known from Madagascar, Comoros, Sri Lanka and mainland Africa [26, 27]. Clitaetra are much smaller spiders that inhabit forest trees, and probably are poor dispersers. Consequently, Comoros are inhabited by an endemic species, as apparently the belt of sea between Mayotte and Madagascar 300 km wide presents enough of a barrier to prevent gene flow. Judging by the number of landmasses occupied by each lineage, Clitaetra is a poor disperser and Nephila a very successful one, while Nephilengys, as an intermediate disperser, is the most diverse of the three across the Indian Ocean archipelago. Better dispersers can colonize more islands, but also require larger distances to effectively prevent gene flow, while poor dispersers only rarely reach less isolated islands with few opportunities to diversify [6, 14]. Does this simple model hold for other organisms on these islands?
Recent literature provides compelling evidence for the origin of the majority of the Indian Ocean island biotas via Cenozoic dispersal rather than via vicariant origin on ancient Gondwanan landmasses [[2, 4, 18]; but, see ]. Thus, the best explanation for the origin of most biotas on the islands is that their ancestors must have arrived relatively recently, when the landmasses were in, or close to today's position, having mainly arrived from Africa, but also with elements from Asia and Australasia. The modes of dispersal must be either aerial, rafting on ocean, or a mix of both, and in some cases, assisted by human transport . We argue that Nephila and Nephilengys spiders fall in the category of aerial dispersers, with no evidence of human transport among islands in the region.
Aerial transport, either active or passive with wind, is probably the best understood mode of dispersal. Several groups of flying animals have colonized the Indian Ocean islands and speciated there, but their origin varies. Logically, the oceans present the least of a barrier to birds and bats. For example, oscine passerines dispersed from Australia to Asia, and on over the Indian Ocean to Africa, where they radiated . Parrots reached the Mascarenes from India , and Triaenops bats colonized Madagascar from Africa several times resulting in several independent lineages there . Another group of organisms that disperse by wind are flying insects; in allodapine bees for example, there is a pervasive pattern of African Miocene origin with several dispersal events onto Madagascar, to Asia, and to Australia . These authors concluded that the bees possess the ability to cross large expanses of ocean via west drift wind, and did not exclude the potential of over water rafting over the Mozambique Channel between Africa and Madagascar. Apparently, for flying insects, the Mozambique Channel (just over 400 km wide) presents only a moderate barrier to dispersal. To dragonflies over ocean wind dispersal presents little difficulty as evidenced by wind assisted colonization of the Indian Ocean islands from Asia .
Terrestrial and freshwater groups, both of presumed lower dispersal abilities compared with aerial dispersers, have also occupied most of the islands that we studied, e.g. lizards [34–38] and frogs . These groups probably used rafting on ocean as means of dispersal. Chameleons, once believed to be of Gondwanan origin on Madagascar, have in fact colonized Madagascar over the ocean where they subsequently radiated . They then spread to Comoros and Seychelles, where they also speciated. One species is thought to have recently colonized Réunion where it has accumulated morphological differences from the source population in Madagascar. In each case, over water dispersal events occur frequently enough to allow colonization of several islands not followed by extinction, but rare enough such that colonization events immediately restrict gene flow and eventually lead to speciation. Coastal lizards (genus Cryptoblepharus) are globally distributed in Asia and the islands around Madagascar, where they diversified, then separately colonized the East African coast, the Comoros islands and Mauritius, but not Réunion or Rodrigues . This suggests the occurrence of occasional over water dispersal, which is rare enough to lead to speciation even on islands separated by small bodies of water. Phelsuma geckos show higher speciation rates [35, 36] with a species rich radiation confined to Madagascar, from where a colonization event to the Mascarenes is dated at 4-5 Ma, followed by speciation on all the islands: three endemic species are known from Réunion, five on Mauritius and three on Rodrigues, of which two are extinct . These studies suggest that over ocean dispersal in lizards is possible but rare, and these relatively poor dispersing abilities facilitate speciation in the absence of recent gene flow.
The above examples of groups with good (aerial) dispersers versus moderate (rafting) dispersers provide us with the following comparison of radiation success: Birds as the best dispersers have colonized all the Mascarene islands, but have not radiated [4, 30]. Triaenops fruit bats, also good dispersers, have repeatedly colonized Madagascar and adjacent islands in the relatively recent past, but remain species poor , presumably due to continuous gene flow. Among the best insect dispersers, dragonflies, colonized all the islands but only diversified very moderately . Rafting dispersers, presumably of medium dispersal abilities, are present in almost all the islands, and exhibit some exceptionally diverse radiations (chameleons on Madagascar and Phelsuma geckos throughout the archipelago, see above). Terrestrial mammals and amphibians, presumably poor dispersers, are entirely absent from the Mascarenes , but have radiated in Madagascar after reaching it during rare dispersal events: lemurs, rodents, tenrecs and carnivores radiated on Madagascar (reviewed by ). The common pattern seems to be analogous to the Nephila-Nephilengys-Clitaetra example that we studied. Therefore, the model of intermediate dispersal abilities underlying diversification across archipelagos seems to be also supported by the data from organisms other than spiders. The well-known trend of good dispersers losing their dispersal ability subsequent to colonizing islands also may lead to speciation events, another example where somewhat reduced dispersal ability positively correlates with diversification across archipelagos. Of course, a broader comparison would be needed to better test the validity of this speciation model, which is beyond our scope here.