Structure and history of African elephant populations: I. Eastern and southern Africa.

Patterns of restriction site variation within mitochondrial DNA (mtDNA) of 270 individuals were used to examine the current structure of savanna elephant populations and to infer historical patterns of gene flow across eastern and southern Africa. Elephants have a complex population structure characterized by marked subdivision at the continental level (Fst = 0.39; 95% confidence interval 0.19-0.58), and isolation by distance at the regional level. However, phylogeographic analysis revealed evidence of protracted gene flow across the continent. First, one relatively derived haplotype was found at all sampling locations. Second, haplotypes representing exceptionally divergent (up to 8.3%) mitochondrial clades were found to coexist at distant (> 2,000 km) sampling locations. In the few other species characterized by sympatric individuals bearing such divergent haplotypes, all such individuals were found to coexist within limited geographical regions. Accordingly, pronounced mitochondrial divergence within populations is often attributed to ancestral isolation in allopatry, followed by secondary contact. The patterns within elephants do not accord with ancestral isolation in allopatry. Given the exceptional mobility of elephants, a geographical barrier is unlikely to have obstructed gene flow between regions for long enough to produce the observed mitochondrial divergence. Rather, the patterns are consistent with the more parsimonious hypothesis, based on neutral coalescent theory, that gene flow has maintained a sufficiently large effective population size (> 50,000 females) for representatives of clades that diverged at least 4 million years ago to have persisted by chance within a population that was subdivided, but not strictly isolated in allopatry.