Genome - environment association analyses
The number of candidate SNPs significantly correlated with each ecological variable ranged between 14 000 and 16 650 (Table 5), on 6000 - 7500 genes. Many of the candidate SNPs were the same across models, likely due to the highly correlated nature of the ecological variables. The model for zPC2 (which described the negative covariance ofRotifers and Alona sp. with other cladocerans and copepods) had the lowest DIC, suggesting that genomic variation was best predicted by invertebrate composition, followed by water depth and temperature (Table 5).
We found between 18 and 34 enriched molecular functions per ecological variable (Table 5). The biological process occurring most commonly in enriched terms across all environments was the development of the central nervous system, followed by other developmental processes (i.e., the development of circulatory system, kidneys, tissue, embryo development, fin development, and ossification) and metabolic processes (most commonly involving protein metabolism). For temperature, protein phosphorylation was the most significantly enriched; for depth, it was ion transmembrane transport; for bird density and zPC3, it was the cell surface receptor signalling pathway category (i.e., signal transduction); for CPUE, it was cell adhesion; for zPC2 it was neuron development; and for zPC4 it was nervous system development (Table 5). Interestingly, we found that terms associated with sensory systems (including response to abiotic stimulus and perception of sound) were enriched for zPC3 and zPC4, which each reflect different components of the community structure of potential prey species. Loci associated with aspects of kidney function, including renal filtration, were enriched for temperature, depth and bird density, and terms associated with immune system and function were enriched in relation to temperature, depth and zPC4.
We found that candidate SNPs identified in LFMM analyses were linked on the genome to putative QTLs (as identified in GWAs analyses) for DS1 (for all environments), DS2 (for water depth, zPC2, zPC3 and zPC4), and PS (for water depth, zPC2, zPC3 and zPC4). There was a total of 23 genes on regions of the genome where there were both candidate SNPs identified in environmental association analyses, and QTLs associated with phenotypic variance (see Table S5). Of those genes, ULK2 andCCNB1 genes had haplotypes causing changes in amino acid sequence.