Genetic diversity of the Chaerephon leucogaster/pumilus complex from mainland Africa and the western Indian Ocean islands.
Chaerephon (Dobson, 1874), an Old World genus belonging to the family Molossidae, is part of the suborder Vespertilioniformes. Members of this genus are distributed across mainland Africa (sample sites; Tanzania, Yemen, Kenya, Botswana, South Africa and Swaziland), its offshore islands (Zanzibar, Pemba and Mozambique Island), Madagascar and the surrounding western Indian Ocean islands (Anjouan, Mayotte, Moheli, Grande Comore, Aldabra and La Reunion). A multifaceted approach was used to elucidate the phylogenetic and population genetic relationships at varying levels amongst these different taxa. Working at the subspecific level, I analysed the phylogenetics and phylogeography of Chaerephon leucogaster from Madagascar, based on mitochondrial cytochrome b and control region sequences. Cytochrome b genetic distances among C. leucogaster samples were low (maximum 0.35 %). Genetic distances between C. leucogaster and C. atsinanana ranged from 1.77 % to 2.62 %. Together, phylogenetic and distance analyses supported the classification of C. leucogaster as a separate species. D-loop data for C. leucogaster samples revealed significant but shallow phylogeographic structuring into three latitudinal groups (13º S, 15 - 17º S, 22 - 23º S) showing exclusive haplotypes which correlated with regions of suitable habitat defined by ecological niche modelling. Population genetic analysis of D-loop sequences indicated that populations from Madagascar have been expanding since 5 842 - 11 143 years BP. At the infra-generic level, I carried out analyses of sequences of the mitochondrial cytochrome b gene and control region, and the nuclear RAG2 region, to resolve the evolutionary history and taxonomy of the C. pumilus species complex from Africa and the western Indian Ocean islands. The nominate form comprised C. pumilus from Massawa, Eritrea, and this was genetically distinct from all other forms of Chaerephon. Our molecular evidence does not support that the syntype of C. limbatus and the holotypes of C. elphicki and C. langi and topotype of C. naivashae are specifically distinct from C. pumilus s.s. There is evidence of introgression of both C. pusillus and C. pumilus s.l. (south eastern Africa) mitochondrial haplotypes into C. leucogaster. The C. pumilus species complex has several attributes of a ring species, but appears to differ from this model in some important respects. It occurs on the African mainland and western Indian Ocean Islands, including Madagascar, ringing a potential barrier to gene flow, the Mozambique Channel. The taxa within the species complex form a ring in which the differentiated terminal forms, C. pusillus and C. leucogaster, occur in sympatry on Mayotte (Comoro Islands). Although there is evidence of isolation by distance around the ring, there is also a relatively high degree of genetic structure and limited gene flow. It appears that the island-based component species may have differentiated in allopatry, with some gene flow by over water dispersal, whereas the African mainland species may have differentiated through isolation by distance. A further study was aimed at re-examining the phylogeny of C. pumilus sensu lato from south eastern Africa based on a considerably larger sample set with a wider geographic range; I confirmed the previously-reported phylogenetic structure, and identified an additional strongly-supported control region clade. Discriminant Function Analysis based on four echolocation parameters could not discriminate between these clades. The hypothesised existence of cryptic species with distinct echolocation characteristics was not supported. Indices of diversity and neutrality, combined with a ragged multimodal mismatch distribution, are inconsistent with demographic expansion of a single C. pumilus south eastern African population and suggest that the control region lineages are stable populations at demographic equilibrium that were established during the late Pleistocene between 60 000 and 13 000 years ago. Further, more variable markers (microsatellites) were employed for finer-scale resolution of population genetic structure among the five genetic lineages of C. pumilus sensu lato found in the Durban area of KwaZulu-Natal, and to search for hybridization between these lineages. We recovered strong mitochondrial genetic structure, with 90% of the molecular variance occurring among four phylogenetically-defined groups, and a high significant Fst (0.897). Microsatellite data recovered three admixed populations with 3% of the nuclear variance occurring among populations, and global (Fst=0.037) and pairwise Fst values among populations were low and not significant. This is indicative of little genetic structure among the groups of C. pumilus s.l., which appear to comprise a single interbreeding population. Such high levels of mitochondrial genetic structure in the absence of significant nuclear structure are consistent with social isolation mechanisms such as female philopatry, and may reflect introgression of mitochondrial genes due to past hybridisation events with mitochondrially-distinct forms from outside the sampled area.