Genomic Consequences of Fragmentation in the Endangered Fennoscandian Arctic Fox (Vulpes lagopus)

Abstract

Accelerating climate change is causing severe habitat fragmentation in the Arctic, threatening the persistence of many cold-adapted species. The Scandinavian arctic fox (Vulpes lagopus) is highly fragmented, with a once continuous, circumpolar distribution, it struggled to recover from a demographic bottleneck in the late 19th century. The future persistence of the entire Scandinavian population is highly dependent on the northernmost Fennoscandian subpopulations (Scandinavia and the Kola Peninsula), to provide a link to the viable Siberian population. By analyzing 43 arctic fox genomes, we quantified genomic variation and inbreeding in these populations. Signatures of genome erosion increased from Siberia to northern Sweden indicating a stepping-stone model of connectivity. In northern Fennoscandia, runs of homozygosity (ROH) were on average ~1.47-fold longer than ROH found in Siberia, stretching almost entire scaffolds. Moreover, consistent with recent inbreeding, northern Fennoscandia harbored more homozygous deleterious mutations, whereas Siberia had more in heterozygous state. This study underlines the value of documenting genome erosion following population fragmentation to identify areas requiring conservation priority. With the increasing fragmentation and isolation of Arctic habitats due to global warming, understanding the genomic and demographic consequences is vital for maintaining evolutionary potential and preventing local extinctions.

Keywords: bottleneck; conservation; fragmentation; inbreeding; mutational load; runs of homozygosity.

Figure 1

Map showing the distribution of Arctic tundra habitat (Scandinavia: alpine tundra [60]; Eurasia: oro-arctic tundra [61]) in relation to study sample areas in northern Sweden, northern Norway, Kola, Yamal, Taymyr, Indigirka, Faddeyevsky Island and Wrangel Island.

Figure 2

Map showing the distribution of Arctic tundra habitat (Scandinavia: alpine tundra [60]; Kola: oro-arctic tundra [61]) in relation to study sample areas in Vindelfjällen-Arjeplog, Saltfjellet, Øvre Dividal, Reisa nord, the Varanger Peninsula, and the Kola Peninsula.

Figure 3

Principal component (PC) analysis of population stratification across samples of arctic foxes on axes (a) PC1-PC2 and (b) PC1-PC3. Points represent genotypic data for 4,862,746 single nucleotide polymorphisms (SNPs) per individual. The first two principal components (PCs) explained 11% (PC1) and 6% (PC2) of the genotypic variation across all individuals and SNPs. A clear isolation-by-distance pattern is exhibited. PC3 explained 6% of the genotypic variation.

Figure 4

Genome-wide heterozygosity of individuals in northern Fennoscandian and Siberian populations (a) and separated into subpopulations (b). Heterozygosity measured as the number of heterozygote sites per 1 kb. The horizontal bar shows the median, the boxes show the 25–75% interquartile range and the whiskers show the whole range. Significant results marked with * P < 0.05, ** P < 0.01, *** P < 0.001.

Figure 5

Mean genomic inbreeding coefficients (F_ROH) with standard deviation for individuals in northern Fennoscandian and Siberian populations (a) and separated into subpopulations (b). Orange bars show inbreeding due to close relations deep back in history (45–850 generations, >100 kb–2 Mb), Green bars shows show inbreeding due to close relations 10–45 generations back (2–8 Mb) and blue bars show inbreeding due to recent ancestors less than 10 generations back (>8 Mb).

Figure 6

Proportion of deleterious variants in arctic foxes of the northern Fennoscandian and Siberian populations; separated into (a) homozygous loss of function variants (LoF), (b) heterozygous LoF variants, (c) homozygous missense variants, and (d) heterozygous missense variants. Proportion calculated as the number of each variant type divided by total variants. The horizontal bar shows the median, the boxes show the 25–75% interquartile range and the whiskers show the whole range. Significant results marked with *** P < 0.001.