Evidence for the parallel evolution of head and body shape morphology across ecomorphs has been shown in previous studies of Arctic charr . However, even disregarding the notably distinct Dughaill ecomorph pair, the other three ecomorph pairs show limited evidence of parallelism in shape morphology despite the parallel evolution of ecomorphs themselves. Evolutionary divergence and thus phenotypic trajectories are influenced by the interaction between environmental variation and adaptive genetic variation. As a result, repeated ecomorph divergences often have very different phenotypic trajectories for key components of phenotypes, as seen in three-spine stickleback . Thus, the lack of strict parallelism seen in our study is likely the result of known differences in the evolutionary histories of these pairs and differing selective pressures in their local environments, for example from differences in their ecosystems or diets . Parallelism can also be generated by other mechanisms, such as differences in gene expression, post-translational modifications, and/or alternate splicing . Indeed, differences in splicing and gene expression patterns showed parallel patterns across ecomorph pairs in a study on three of the ecomorph pairs we investigated and provide explanations of alternative adaptive paths.
Of the two traits we tested, head shape had longer phenotypic trajectory lengths and greater angles, suggesting that the ecomorphs are more differentiated in head shape from one another and that head shape has evolved in more distinct directions across lakes. Head shape is well recognised as an important phenotype for foraging and prey specialisation, while body shape is important for swimming behaviour and habitat complexity . The considerable divergence in head shape between ecomorphs within lakes suggests different prey specialisations (Garduño-Paz et al. 2012; Hooker et al 2016) while their body shapes and therefore perhaps swimming behaviours are more subtly different. The high trajectory angles for head shape suggest notable difference in foraging across lakes, whether that be due to the lake environments or the actual species available as prey . For both traits, Loch Tay and Dughaill showed notably higher trajectory lengths than Awe and na Sealga and more evolutionary divergence, also seen in the ecomorphs’ genomic divergence (FST ~1% between ecomorphs in Awe and na Sealga vs 9% in Dughaill and Tay). This reflects likely what were previously inferred to be recent sympatric divergences of ecomorph pairs in Awe and na Sealga while Tay and Dughaill each have complex histories of divergence and secondary contact between colonising lineages .

Genomic underpinnings of head and body shape across lakes:

From our total of 212 SNPs that showed high associations with head and/or body shape (Figure 2), we found more SNPs associated with body shape than for head shape. Head shape was controlled by more large-effect loci relative to body shape and may suggest that head shape is controlled by fewer genes/pathways. In both cases these will be an underestimate of actual associations because we have reduced representation of the genome captured. We found for that for both head and body shape only a small number of associated SNPs were diverged between ecomorphs in all four pairs (Figure 3). This is line with what has been suggested both in other Arctic charr studies and other salmonid species, in which genetic differentiation between ecomorphs is largely nonparallel across pairs bar at a few key genes . Further to previous work, we found that the SNPs shared across pairs were not highly differentiated between ecomorphs in all pairs suggesting that while present, they are not critical to underlying the phenotypic differences in each pair (Figure 4). These results also suggest that the genomic underpinnings of each phenotype varies across the lakes, likely contributing to the phenotypic differences we see between pairs. The polygenic genomic underpinnings of both phenotypes, as indicated by the numbers of associated SNPs identified, indicate that there are multiple pathways that can achieve the same phenotypes hence the lack of high divergence for the same SNPs across all lakes .
Loch Dughaill and in particular Loch Tay often showed notable high genomic divergence between ecomorphs for many of the associated SNPs for each trait. Additionally, the associated SNPs for both traits showed high DXY values compared to the background subsets at both of these lakes. While increased levels of DXY or FST compared to genomic background can be indicative of positive selection , they might also be expected for loci resisting introgression following secondary contact , as is likely the case in Loch Tay and Loch Dughaill . The associated SNPs we found are widespread across the genome (Figure 2) indicating these are not single linked regions of divergence as found in studies on Atlantic cod and rainbow trout but instead are diffused and highly polygenic, similar to patterns for body shape in lake whitefish .

Functional genomic regions for head and body shape:

Roughly half of the associated SNPs identified for each of head and body shape were found within or proximal to an annotated gene in the charr genome. A number of the GO terms that appeared as significantly overrepresented or enriched in our study have been identified in other studies investigating adaptive divergences or parallel evolution in various fish species. Odontogenesis (GO:0042476), sensory perception of sound (GO:0007605), blood vessel remodelling (GO:0001974), response to muscle activity (GO:0014850), ventricular trabecula myocardium morphogenesis (GO:0003222), common-partner SMAD protein phosphorylation (GO:0007182), cellular response to ethanol (GO:0071361), and neuromuscular synaptic transmission (GO:0007274) have shown significance in other Arctic charr studies investigating ecomorph divergence . The GO terms for associative learning (GO:0008306), regulation of cell shape (GO:0008360), and UDP-glucuronate biosynthetic process (GO:0006065) also appear in overrepresented groups in a study on the divergence of a sympatric lake whitefish species pair (Corgeonus clupeaformis ) in the USA . Finally, in pupfishes (Cyprinodon. sp. ), the divergent expression of a number of genes involved in cranial skeletal system development was seen between different trophic specialists with the GO term for this process (GO:1904888) significant in our study . Differences in ossification rate have been related to adaptive morphological differentiation in other freshwater fish and the over enrichment or overexpression of genes related to formation of various bones in our study indicates a similarly important role in adaptive divergences between ecomorphs of Arctic charr . Indeed, previous work has noted the importance of differences in bone structure and sizes between different ecomorphs of Arctic charr .
The QTL database that we have developed allows us to explore whether rapid replicated diversification of ecomorphs in different salmonid species is underlined by the use of the same functional regions as has been previously suggested for salinity tolerance . Our results suggest that this is true to some extent with QTLs related to body shape in lake trout and whitefish found in proximity to SNPs that we identified as being associated with phenotypic differences in Arctic charr ecomorphs. Whilst we only identified a small number of QTLs located near the associated SNPs, this is line with other work which suggests that shared basis for ecomorph divergence across species may be limited . This QTL marker database will be a valuable resource for future salmonid research.