Rasmus Heller:
Demographic History and Phylogeography of Large African Mammals

Date: 15-12-2011    Supervisor: Hans R. Siegismund

Population genetics can be used to infer many different aspects of species biology, for example demographic history, dispersal patterns and conservations status. The studies presented in this thesis used population genetic methods to investigate some of these aspects in large African mammals. Almost two decades of population genetic research on African ungulates was evaluated in a review study, extracting cross-species phylogeographic patterns that increase our understanding of African biogeography. These show that the most prominent agent of genetic divergence in savannah adapted species is the periodical vicariance among northern and southern populations caused by extension of the equatorial forest belt during pluvial periods. We used complete mitochondrial genomes to infer the demographic history of two prominent African mammals, the African buffalo and the chimpanzee, in two related studies. Bayesian skyline plots were used, and the results were rigorously evaluated through simulation of data under a range of alternative scenarios and the application of approximate Bayesian computation. Results show that the buffalo population expanded contemporaneously with Palaeolithic humans, but that these synchronous dynamics changed dramatically during the Holocene, perhaps attributable to the Neolithic revolution in African humans. The chimpanzee shows a more heterogeneous demographic history, with two subspecies expanding in the late Pleistocene but two other subspecies and the bonobo apparently remaining stable in this period. In a study focusing on the genetic structure of the buffalo on a more local scale in eastern Africa, we found that there is a clear correlation between the genetic diversity of fragmented populations and the size of the conservancy in which they are found, indicating that even the vagile buffalo is not able to compensate for humanly induced fragmentation by maintaining gene flow among national parks.

In addition to these practical studies, we evaluated some common population genetic methods. We found that the very commonly used differentiation measure GST has sometimes yielded misleading estimates of population differentiation when applied to highly variable markers such as microsatellites. Finally, we found that population structure can severely bias results based on Bayesian skyline plots, hence leading to erroneous conclusions about demographic history. We identify the circumstances under which this confounding effect is most serious and propose ways of minimizing the potential bias.