The Finnish brown hare population has been growing in size and increasing its range northwards during the past few decades 6. This development offers an interesting natural setting to study the population genetics and evolutionary adaptation of expanding species. The small pioneering populations at the edges of the distribution are influenced by founder effect and genetic drift, causing the edge populations to have lower genetic diversity than in the core populations – a phenomenon known as expansion load 2. As asymmetric gene flow from mountain hare to brown hare is well established 30–34, we sought to see how this introgression contributes to the genetic makeup of the Finnish brown hares and whether the mountain hare alleles could have adaptive significance for the brown hares.
From our analysis, it was apparent that much of the genetic makeup of Finnish brown hares is heavily influenced by the introgression of alleles from mountain hare (Table 1, Figs. 2 and 3). For example, Finnish brown hares had higher genetic diversity (GD) than their allopatric Austrian counterparts and the difference could be attributed solely to the representation of mountain hare alleles in Finnish brown hares. In contrast, Austrian brown hares had a number of private alleles in all analyzed loci, which were not present in the Finnish hares (Table 1, Figs. 2 and 3). If only the alleles that are shared between Austrian and Finnish brown hares (Fig. 2, Table S2) were taken into account, the GD of Finnish brown hares would be 0.40 ± 0.37. This is notably lower value than observed for the Finnish (0.65 ± 0.45) or Austrian (0.53 ± 0.28) brown hare populations. It should be pointed out that this is an estimate, as the Austrian population is not directly ancestral for the Finnish brown hares and because the origin of the MHC alleles cannot be traced based on their phylogeny. As another tell-tale sign of a founder effect, the most diverse loci DQA and DQB loci were dominated by one common allele each in brown hares (Leeu-DQA*006 = 0.49; Leeu-DQB*001 = 0.56), while mountain hares showed more even genotypic representation and higher number of alleles (Fig. 3, Table 1, Table S3), resulting also in higher levels of heterozygosity in mountain hares than brown hares for both MHC class II loci. In contrast, the other nuclear loci showed higher levels of Hz in Finnish brown hares, which can be explained by an asymmetric flow of alleles at these loci from mountain hares to brown hares, as alleles dominant in mountain hare were commonly present in brown hares but not vice versa (Table 1, Figs. 2 and 3, Table S3).
Taken together, these findings show that brown hares commonly obtain genetic variants from mountain hares and suggest that this plays a role in suppressing the effects of expansion load. Interestingly, some mountain hare alleles were also detected, albeit very rarely, also in the allopatric Austrian brown hare population (Figs. 2 and 3).
Our prime candidates to detect potentially meaningful adaptive introgression in hares included the UCP1 gene, which could have a potential role in adaptation for a colder climate 35,36, MHC class II loci DQA and DQB, which confer general pathogen resistance and are known to be under strong balancing selection favoring genetic diversity in various chordates (e.g. 28,37−39 and TLR2, required for innate immunity 40. SDHa was chosen as potential neutral locus together with mitochondrial DNA.
The conducted outlier analyses detected moderate balancing selection acting on UCP1 and strong balancing selection at both MHC loci (Fig. 4A), while the maintenance of introgressed alleles at TLR2 and SDHa likely reflects neutral processes similar to those at SNP loci (Levanen et al. 2018). This is also evident in the AMOVA analysis, which showed that the majority of the variation in neutral alleles was between the species, while the loci under selection showed significant within individual variation (Table 2). However, the fact that the Bayesian outlier analysis suggests balancing selection for UCP1 is quite likely an artefact caused by the dynamics of the hybridization and selection progress. Alleles under purifying selection are identified from their tendency for fixation, while in the case of the Finnish brown hares, there is a constant input of ancestral alleles through hybridization, which are encountered by differential selection pressures. Revealingly, the mountain hare specific alleles were represented in Finnish brown hares at higher levels than observed for the neutral loci (Fig. 2, Table 1, Table S3). This results in rather even distribution of mountain hare alleles in the Finnish brown hares (Fig. 4B), which mimics the effects of balancing selection. In fact, it is well known that demographic factors of a population can specifically increase the false discovery of balancing selection in Bayesian outlier analysis 41.
Our interpretation is that although some brown hare alleles were detected also in mountain hares, the dominance of mountain hare alleles in both species suggests that selection is favoring locally adapted alleles at the expense of outbred alleles per se. This is evident also from the fact that the Austrian allopatric brown hares have a surprising allelic diversity at the UCP1 locus, including four alleles that are shared with the Finnish hares and six private alleles, while the UCP01 allele is almost the dominant allele in both species in Finland (Fig. 2).
Adaptive thermogenesis in brown fat is dependent on the UCP1 and is essential for all neonate mammals 42. At birth, the newly born mammal needs to rapidly adapt from the maternal body temperature to the environmental temperature and this process is developmentally controlled. We chose to investigate the variation in UCP1 specifically as both mountain hares and brown hares do not have nests or lairs, but leave their young unattended after birth and typically feed the leverets only once per day. This means that especially the spring generation, often born on snow, is exposed to cold and variable weather conditions of a boreal climate and is independent of the mother for shelter or thermoregulation. Under these circumstances, any cold adapted alleles for NST are likely to make a difference for early season breeding for brown hares. Therefore, we find it credible that positive selection explains the excess of mountain hare UCP1 alleles in the Finnish brown hare population. As most of Finland is at the northern edge of brown hare distribution, a steep allelic gradient of the mountain hare UCP1 in the brown hare population is not necessarily expected (Fig. 4B). However, there can be also trade-offs with the cold-adapted alleles, such metabolic inefficiency 43, which could balance out the benefits when food is scarce or poor quality. Unfortunately, we can currently only hypothesize about the physiological differences between the different alleles, but these might be possible to investigate in future studies.
While balancing selection favoring allelic diversity is well established for MHC loci, the directionality of the gene flow between the two species is difficult to dissect. As pointed out, while Finnish brown hares have a dominant allele in both DQA and DQB loci, the same alleles are also abundant in mountain hares (Fig. 3). Similarly, there is evidence for gene flow from brown hare to mountain hare at the studied nuclear loci and mtDNA, although this is dwarfed by the introgression to the opposite direction (Table 1, Fig. 2). MHC loci are also notable for examples of trans-species polymorphisms 26,44, which are known to occur even among different genera of leporids 45. For example, TSP could explain the existence of common brown hare DQA and DQB alleles in the Finnish mountain hare population (Fig. 3, Table S3) but without samples from highly isolated populations e.g., in the arctic islands, the distinction from recent introgression cannot be easily resolved. Similarly, our phylogenetic analysis was inconclusive as the DQA and DQB alleles do not cluster as species-specific (or one species dominated) clades, unlike UCP1. The timing of the nodes on the tree also gave rather young approximations for the ages of the clades when compared to the speciation estimate. This is likely due to the fact that the applicability of fast evolving sequences as molecular clocks saturates at deeper divergences 46. However, the allopatric Austrian brown hares had four private DQA alleles and seven DQB allelles, while sharing only four and three alleles respectively with the Finnish hares (Fig. 3). We take this as evidence that the alleles shared among the Finnish and Austrian populations could represent TSPs or a very ancient introgression event, whereas the Finnish hares have universal allele sharing via bidirectional gene flow, as is also seen for the other investigated loci. The fact that there were no DQA or DQB private alleles in the allopatric Finnish mountain hares (Fig. 3), suggests that the Finnish mountain hares form a uniform, genetically connected population 34 and that any beneficial alleles spread quickly in it 47. Similarly, balancing selection involving novel allele advantage has been suggested to have promoted the introgression of MHC alleles also in other vertebrates 48–51. As a follow-up, it would be interesting to correlate the MHC genotypes with susceptibility to specific pathogens or parasites.
In Finland, the distribution of brown hares is rather precisely limited by 150 days of snow cover, a boundary that has steadily retreated northward during the last three decades due to climate change 6. Although capable of replacing mountain hares through direct competition 4,52, brown hares are dependent on anthropogenic habitats in Finland and they are thought to be unable to penetrate the boreal forest ecosystems. The situation may be more complex regionally however, as brown hares can exploit road networks in their expansion and there are indications that the species may adapt to forested habitats in Nordic countries 53. Our data give some insight into how adaptation may be occurring at the genomic level and underlines the advantage conveyed by the ability to co-opt genic variants from the resident species to open up otherwise inaccessible habitats. Climate change may facilitate this process by relaxing local selection pressures, such as the ones caused by the snow cover.
The recent decline in mountain hare numbers in Southern Finland is likely due to not only increased competition with the adapting brown hare, but also because of the shortening of the snow-covered season 33, resulting in the camouflage mismatch of the white winter pelage and increasing predation mortality 54. Further work incorporating coat color loci 55 and gene expression may be informative on this front and indicate if plasticity exists which might assist the persistence of mountain hares. Although the mountain hare has shown remarkable resilience in the past 56, the extent of the combined effects of the current climate changes are likely to be complex and their outcome for the species is unknown.