Human-mediated introduction of introgressed deer across Wallace's line: Historical biogeography of Rusa unicolor and R. timorensis

Abstract

In this study we compared the phylogeographic patterns of two Rusa species, Rusa unicolor and Rusa timorensis, in order to understand what drove and maintained differentiation between these two geographically and genetically close species and investigated the route of introduction of individuals to the islands outside of the Sunda Shelf. We analyzed full mitogenomes from 56 archival samples from the distribution areas of the two species and 18 microsatellite loci in a subset of 16 individuals to generate the phylogeographic patterns of both species. Bayesian inference with fossil calibration was used to estimate the age of each species and major divergence events. Our results indicated that the split between the two species took place during the Pleistocene, ~1.8 Mya, possibly driven by adaptations of R. timorensis to the drier climate found on Java compared to the other islands of Sundaland. Although both markers identified two well-differentiated clades, there was a largely discrepant pattern between mitochondrial and nuclear markers. While nDNA separated the individuals into the two species, largely in agreement with their museum label, mtDNA revealed that all R. timorensis sampled to the east of the Sunda shelf carried haplotypes from R. unicolor and one Rusa unicolor from South Sumatra carried a R. timorensis haplotype. Our results show that hybridization occurred between these two sister species in Sundaland during the Late Pleistocene and resulted in human-mediated introduction of hybrid descendants in all islands outside Sundaland.

Keywords: Cervidae; Phylogeography; Sundaland; Wallace's line; human introduction; hybridization.

Figure 1

Distribution map of both species and sampling location. Light gray indicates the distribution range of Rusa unicolor (a). Dark gray indicates the native distribution of Rusa timorensis, (b) and dashed dark gray areas indicate introduction range of R. timorensis (C). Filled circles and diamonds indicate R. unicolor and R. timorensis, respectively, and size is proportional to the number of samples. For detailed information about all samples see Table A1

Figure 2

Haplotypic network for all 46 haplotypes shared among the two species. Circle size is in accordance with frequency and color represents sampling location. Small black circles represent median vectors. All branches represent one mutation step, except when indicated otherwise by numbers on branches. Two major clades were recovered and are indicated by the species names. Haplotypes are described in Table A1

Figure 3

Mitogenome maximum likelihood tree of both species. Colors on tips represent sampling location (as in Figure 2) and stars represent split events with bootstrap values/Bayesian posterior probabilities lower than 90/0.95 (but bigger than 50/0.5). Red and green dots represent samples for which we obtained nDNA; red dot: assigned to the Rusa unicolor genotypic cluster and green dot: assigned to the Rusa timorensis genotypic cluster. Major mtDNA clades and subclades are labeled with curved brackets. Scale bar indicates number of substitutions per position

Figure 4

Mitochondrial DNA dated tree according to BEAST analyses, from 3 Mya to present. Blue bars represent associated deviations for the most important splits. A time scale in millions of years and a rough estimate of sea level changes through time (adapted from Patou et al. 2010) are presented below

Figure 5

Genotyping results from 16 individuals genotyped for 18 loci, showing a structure plot for K = 2, with R. timorensis samples in green and R. unicolor individuals in red. Each column represents a single individual, as identified below