Mitochondrial genomes from modern horses reveal the major haplogroups that underwent domestication

Abstract

Archaeological and genetic evidence concerning the time and mode of wild horse (Equus ferus) domestication is still debated. High levels of genetic diversity in horse mtDNA have been detected when analyzing the control region; recurrent mutations, however, tend to blur the structure of the phylogenetic tree. Here, we brought the horse mtDNA phylogeny to the highest level of molecular resolution by analyzing 83 mitochondrial genomes from modern horses across Asia, Europe, the Middle East, and the Americas. Our data reveal 18 major haplogroups (A-R) with radiation times that are mostly confined to the Neolithic and later periods and place the root of the phylogeny corresponding to the Ancestral Mare Mitogenome at ~130-160 thousand years ago. All haplogroups were detected in modern horses from Asia, but F was only found in E. przewalskii--the only remaining wild horse. Therefore, a wide range of matrilineal lineages from the extinct E. ferus underwent domestication in the Eurasian steppes during the Eneolithic period and were transmitted to modern E. caballus breeds. Importantly, now that the major horse haplogroups have been defined, each with diagnostic mutational motifs (in both the coding and control regions), these haplotypes could be easily used to (i) classify well-preserved ancient remains, (ii) (re)assess the haplogroup variation of modern breeds, including Thoroughbreds, and (iii) evaluate the possible role of mtDNA backgrounds in racehorse performance.

Fig. 1.

Sequence variation within the horse mitochondrial genome. Nucleotide diversity (π) and total number of substitutions by considering windows of 200 bp (step size = 100 bp) centered at the midpoint. A schematic (linearized) genetic map of the mitochondrial genome is presented.

Fig. 2.

Schematic phylogeny of complete mtDNAs from modern horses. This tree includes 81 sequences and was rooted by using a published donkey (E. asinus) mitochondrial genome (not displayed). AMM indicates the reconstructed AMM, whereas horse reference sequence (HRS) is the newly proposed horse reference sequence (GenBank JN398377). The topology was inferred by a maximum parsimony (MP) approach, whereas an ML time divergence scale (based on synonymous substitutions) is shown on the bottom. Additional details concerning samples, mutational motifs, and ages are given in SI Appendix, Fig. S1 and Tables S1 and S2.

Fig. 3.

Age estimates (A) calculated on the protein coding genes (by considering only synonymous mutations) and the entire genome partitioned into five datasets (first, second, and third positions of the codons, RNAs, and noncoding region). B shows a comparison of BEAST and PAML ages relative to the synonymous estimates. Additional details are given in SI Appendix, Table S2.

Fig. 4.

Bayesian skyline plot (BSP) showing the horse population size trend with a generation time of 8 y (52). The y axis indicates the effective number of females. The thick solid line is the median estimate, and the gray shading shows the 95% highest posterior density limits. The time axis is limited to 100 ky; beyond that time, the curve remains linear.