Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 26 (2), 163-73

Worldwide Patterns of Genomic Variation and Admixture in Gray Wolves

Affiliations

Worldwide Patterns of Genomic Variation and Admixture in Gray Wolves

Zhenxin Fan et al. Genome Res.

Abstract

The gray wolf (Canis lupus) is a widely distributed top predator and ancestor of the domestic dog. To address questions about wolf relationships to each other and dogs, we assembled and analyzed a data set of 34 canine genomes. The divergence between New and Old World wolves is the earliest branching event and is followed by the divergence of Old World wolves and dogs, confirming that the dog was domesticated in the Old World. However, no single wolf population is more closely related to dogs, supporting the hypothesis that dogs were derived from an extinct wolf population. All extant wolves have a surprisingly recent common ancestry and experienced a dramatic population decline beginning at least ∼30 thousand years ago (kya). We suggest this crisis was related to the colonization of Eurasia by modern human hunter-gatherers, who competed with wolves for limited prey but also domesticated them, leading to a compensatory population expansion of dogs. We found extensive admixture between dogs and wolves, with up to 25% of Eurasian wolf genomes showing signs of dog ancestry. Dogs have influenced the recent history of wolves through admixture and vice versa, potentially enhancing adaptation. Simple scenarios of dog domestication are confounded by admixture, and studies that do not take admixture into account with specific demographic models are problematic.

Figures

Figure 1.
Figure 1.
Sample distribution. Solid circles are samples sequenced in this study. Open circles indicate sequences from Zhang et al. (2014). Triangles and boxes indicate sequences from Wang et al. (2013) and Freedman et al. (2014), respectively. Species memberships are indicated by color: gray wolf (red), domestic dog (blue), coyote (green), and golden jackal (yellow). The reference dog genome is from a boxer.
Figure 2.
Figure 2.
Total length of runs of homozygosity (ROHs) and heterozygosity. The black line is the total length of ROHs (Mb) in each genome, and the blue and red bars are the genome-wide heterozygosity with and without ROHs, respectively.
Figure 3.
Figure 3.
The maximum likelihood tree of 30 sequences. Numbers represent node support inferred from 100 bootstrap repetitions. The reference genome boxer was not included. The Israeli golden jackal is the outgroup.
Figure 4.
Figure 4.
Principal component analyses. (A) PC1 and PC2 of dogs and 20 wolves; (B) PC1 and PC2 of dogs and 18 wolves, excluding the Tibetan wolf 1 and Qinghai wolf 1; (C) PC3 and PC4 of dogs and 20 wolves; (D) PC3 and PC4 of dogs and 18 wolves, excluding the Tibetan wolf 1 and Qinghai wolf 1. (□) Highland Asian wolves; (▵) lowland Asian wolves; (○) Middle Eastern wolves; (■) European wolves; (▲) dogs; (●) North American wolves.
Figure 5.
Figure 5.
Demographic history inferred using PSMC. Following Freedman et al. (2014) and Zhang et al. (2014), we used a generation time = 3 and a mutation rate = 1.0 × 10−8 per generation. The Tibetan wolf 1 and Inner Mongolia wolf 4 are shown in all the plots for comparison purposes. (A) All the Asian wolves; (B) all the European wolves, Middle Eastern wolves, and Indian wolf; (C) dogs; (D) Mexican wolf and Yellowstone wolves.
Figure 6.
Figure 6.
Demographic model inferred using G-PhoCS. Estimates of divergence times and effective population sizes (Ne) inferred by applying a Bayesian demography inference method (G-PhoCS) to sequence data from 13,647 putative neutral loci in a subset of 22 canid genomes (because of limitations in computational power). Estimates were obtained in four separate analyses (Methods; Supplemental Table 6). Ranges of Ne are shown and correspond to 95% Bayesian credible intervals. Estimates are calibrated by assuming a per-generation mutation rate of μ = 10−8. Mean estimates (vertical lines) and ranges corresponding to 95% Bayesian credible intervals are provided at select nodes. Scales are given in units of years by assuming an average generation time of 3 yr and two different mutation rates: μ = 10−8 (dark blue) and μ = 4 × 10−9 (brown). The model also considered gene flow between different population groups (see Table 1).

Similar articles

See all similar articles

Cited by 36 PubMed Central articles

See all "Cited by" articles

Publication types

Substances

LinkOut - more resources

Feedback