Genetic structure and demographic history of house mice in western Europe inferred using whole-genome sequences

Abstract

The western house mouse, Mus musculus domesticus, is a human commensal and an outstanding model organism for studying a wide variety of traits and diseases. However, we have few genomic resources for wild mice and only a rudimentary understanding of the demographic history of house mice in Europe. Here, we sequenced 59 whole genomes of mice collected from England, Scotland, Wales, Guernsey, northern France, Italy, Portugal and Spain. We combined this dataset with 24 previously published sequences from southern France, Germany and Iran and compared patterns of population structure and inferred demographic parameters for house mice in western Europe to patterns seen in humans. Principal component and phylogenetic analyses identified three genetic clusters in western European mice. Admixture and f-branch statistics identified historical gene flow between these genetic clusters. Demographic analyses suggest a shared history of population bottlenecks prior to 20 000 years ago. Estimated divergence times between populations of house mice from western Europe ranged from 1500 to 5500 years ago, in general agreement with the zooarchaeological record. These results correspond well with key aspects of contemporary human population structure and the history of migration in western Europe, highlighting the commensal relationship of this important genetic model.

Keywords: commensal; demographic history; house mouse; population genetics; western Europe; whole-genome sequences.

Figure 1.

Sample distribution and population structure of western house mice. (a) Principal component analysis. (b) Sample map depicts locations (colour, year caught) in Portugal (green, 2005), southwest Spain (green, 2008), southern France (purple, 2005), Italy (purple, 1987 and 1991), northeast Spain (purple, 1991), England (red, ca 1990, 2001, 2004 and 2008), Wales (red, 2005), Scotland (red, 1989, 1996, 1997, 1999 and 2004), Guernsey (red, 2007), northern France (red, 2007), Germany (red, 2006) and Iran (cyan, 2006) from which western house mouse genomes were obtained or were previously available. Samples from southern France (purple), Germany (red) and Iran (cyan) are from [39]. (c) Maximum-likelihood phylogeny and (d) ADMIXTURE analysis identify four distinct groups corresponding to samples from northern Europe (red), the Mediterranean (purple), the Atlantic coast of Iberia (green) and Iran (cyan).

Figure 2.

Pairwise comparisons of genetic similarity in western house mouse populations. (a) Left: biplot of normalized f₃ statistics. Right: table of normalized f₃ statistics, with z-score computed using qp3Pop. (b) f₃ statistics found in (a) plotted against pairwise distances between averaged geographic coordinates of each population. (c) Average Euclidean distance between points in PC1 and PC2 of principal component analysis found in figure 1a plotted against average pairwise distance between populations. p-values reported are from Mantel’s test using Pearson correlation.

Figure 3.

Signals of gene flow between western European house mouse populations. Coloured cells in the matrix highlight excess allele sharing between populations along the x-axis and branch along the expanded phylogeny on y-axis. Cells are shared from white (f_b = 0) to dark red (highest value) according to f_b, the f-branch statistics, as described in Malinksy et al. [51]. The program sets f_b to 0 for any non-significant tests, since very short internal branches can lead to large but non-significant values even in the absence of gene flow. Grey squares correspond to cells whose value cannot be computed because of the structure of the tree. For example, there is no four-taxon test that enables comparison of the most basal internal lineages (e.g. the top row).

Figure 4.

Demographic history of western house mice inferred using smc++. Plot of effective population size changes in (a) all populations (with light blue shaded region indicating the period of the Last Glacial Maximum), (b) northern European populations, (c) Mediterranean populations and (d) Iberian populations.

Figure 5.

Summary of inferred divergence times of sampled house mouse populations based on smc++ (electronic supplementary material, table S5). This summary updates previous models that were based on microarrays, mtDNA, and archaeological evidence ( [9,22,23,31]). Each clade comparison found in electronic supplementary material, table S5 reports inferred split time used to construct the time intervals summarized here in black text next to the grey double-sided arrows. Intra-clade comparisons of neighbouring geographic regions connected by the black line are provided in red and highlighted in yellow.