Relative importance of habitat connectivity in shaping the genetic profiles of two southern African elephant-shrews

Aim: When interpreting genetic patterns across a landscape it is surprisingly difficult to disentangle the effects of landscape connectivity from those of species biology. Here, the spatial distributions of genetic variation of two sympatric elephant-shrew species, the western rock elephant-shrew (. Elephantulus rupestris) and the round-eared elephant-shrew (. Macroscelides proboscideus), are determined and compared. We selected these species because they have similar biologies but differ markedly in habitat use, the rationale being that differences in their genetic structure should be a result largely of landscape variables directly or indirectly affecting dispersal rather than of the biology of the species. Location: South Africa and Namibia. Methods: Mitochondrial sequence data (control region and cytochrome b) were used to describe the phylogeographic structure of these elephant-shrew species across their distribution. To determine whether genetic variation is significantly structured, spatial analyses of molecular variation were performed. Isolation-by-distance versus alternative patterns of genetic structure was investigated using a Mantel test. Results: Our analyses indicated an overall structured genetic profile for E. rupestris, a species closely associated with rocky outcrops. This was in contrast to a pattern mostly of isolation-by-distance across the distribution of M. proboscideus, a species found on gravel plains. Main conclusions: Specific landscape features will differentially affect gene flow (both historical and current), and therefore also the spatial genetic structure, of species with markedly different habitat requirements. The genetic profiles for the two species included here support predictions based on the connectivity of their respective occupied habitats. The results also support the more general prediction that species with a naturally clustered distribution (such as E. rupestris) should have a more structured genetic pattern than those having a more continuous distribution (. M. proboscideus).