Tracing Eastern Wolf Origins From Whole-Genome Data in Context of Extensive Hybridization

Abstract

Southeastern Canada is inhabited by an amalgam of hybridizing wolf-like canids, raising fundamental questions regarding their taxonomy, origins, and timing of hybridization events. Eastern wolves (Canis lycaon), specifically, have been the subject of significant controversy, being viewed as either a distinct taxonomic entity of conservation concern or a recent hybrid of coyotes (C. latrans) and grey wolves (C. lupus). Mitochondrial DNA analyses show some evidence of eastern wolves being North American evolved canids. In contrast, nuclear genome studies indicate eastern wolves are best described as a hybrid entity, but with unclear timing of hybridization events. To test hypotheses related to these competing findings we sequenced whole genomes of 25 individuals, representative of extant Canadian wolf-like canid types of known origin and levels of contemporary hybridization. Here we present data describing eastern wolves as a distinct taxonomic entity that evolved separately from grey wolves for the past ∼67,000 years with an admixture event with coyotes ∼37,000 years ago. We show that Great Lakes wolves originated as a product of admixture between grey wolves and eastern wolves after the last glaciation (∼8,000 years ago) while eastern coyotes originated as a product of admixture between "western" coyotes and eastern wolves during the last century. Eastern wolf nuclear genomes appear shaped by historical and contemporary gene flow with grey wolves and coyotes, yet evolutionary uniqueness remains among eastern wolves currently inhabiting a restricted range in southeastern Canada.

Keywords: admixture; eastern coyote; eastern wolf; great lakes wolf; red wolf; whole genomes.

Fig. 1.

Sample design for canid whole-genome sequence analysis. Triangles refer to samples newly sequenced in this study, while circles denote samples from the literature. The four Eurasian grey wolf samples and one golden jackal sample used as outgroups are not shown. The red wolf samples are placed according to vonHoldt et al. (2016) corresponding to captive breeding facilities in North Carolina, however, the original capture zone for the founders is approximately Texas/Louisiana (Sinding et al. 2018). Inset map highlights wolf-like canid samples from the Great Lakes region. Map layers were obtained from Commission for Environmental Cooperation Atlas.

Fig. 2.

(a) Splitstree network for 6,020,173 autosomal SNPs. Distances were expressed as “1-IBS”. (b) Multidimensional scaling (MDS) analysis of pairwise Fst values. (c) Admixture plot showing the estimated ancestry proportions of 46 individuals determined using Structure for K = 2 through K = 4 based on only high coverage genomes (>10×; 42,785 autosomal SNPs from 50 kbp windows). Each partitioned vertical bar represents an individual's proportional membership to the inferred populations. Asterisks next to sample names indicate samples sequenced in this study. K = 5–9 not shown, yielding no further biologically relevant partitions. GLW = Great Lakes wolf.

Fig. 3.

(a) MSMC2 population size estimates from four haplotypes per population. (b) Bar charts of heterozygosity estimates (Het) and inbreeding coefficients (Fis) per individual. Asterisks next to sample names indicate samples sequenced in this study. GLW = Great Lakes wolf; kyr = thousands of years in the past.

Fig. 4.

Migration profiles from MSMC-IM between eastern wolves and other canid groups. (a) Timing and dynamics of separation process between two groups visualized by the time-dependent-symmetric migration rate m(t). Time is represented as years in the past. The dashed lines represent the median, or the time when 50% of the ancestry between the two groups merged. Shading indicates 1–99% (lighter shade) and 25–75% (darker shade) percentiles of the cumulative migration probabilities. (b) Cumulative migration probabilities that estimate the proportion of ancestry already merged at time t, and represents proportions of gene flow through time. M_(t) values close to 0 denote complete separation between the two groups, while 1 shows a complete mix as one population. Dashed lines represent the relative cross-coalescent rate.

Fig. 5.

(a) Most supported model of ABC-RF analysis. Numbers on the left represent point (median) estimates of divergence times. (b) Admixture graph topology with highest posterior probability using AdmixtureBayes. Tip nodes indicate the sampled genomes used to fit the graph that are the same as those for the ABC-RF analysis. Percentage numbers on the branches represent admixture proportions. Convergence graphs for AdmixtureBayes analysis can be found in supplementary figures S10 and S11, Supplementary Material online.

Fig. 6.

(a) Genetic distance of the fragments painted as eastern wolf ancestry with other canid populations. (b) Observed distributions of introgressed fragments for eastern wolf individuals. Orange and blue represent introgressed fragments from grey wolves and coyotes, respectively. The abundance of introgressed fragments as a function of their length is represented for all eight eastern wolf individuals. Polar = grey wolf Polar; Eurasian = grey wolf Eurasia.