3.2 | Genotype-environment association
The results of LFMM analysis linking environmental and phenotypic variables to SNP variants indicated similar patterns of association with latitude and collection date in both years (Table 4). Of the 76 SNPs associated with these variables in 2014 and 305 in 2015, ten were shared between years (Supplementar materials, Table 1). The loci common in both years were significantly associated with latitude and collection date only. However, because of the sampling being conducted in a generally south to north direction, sampling date and latitude are collinear. This may possibly indicate a temporal gradient of selection where less fit individuals, those with deleterious alleles die off during their first months of life. Therefore the fish collected at later dates may be a subset of the fitter individuals as compared to earlier collections. Or there may be a true latitudinal gradient, or a combination of both factors contributing in various proportions to a selection gradient. Chlorophyll concentration and seawater temperature did not appear to influence loci in 2014, but were associated with 100 loci in 2015. It is important to note the Gulf of Alaska experienced unusually warm temperatures in 2015 (Cavole et al. 2016), marked by large sea bird die-offs (Jones et al. 2018).
In 2015, the fish experienced poorer growing condition as compared to 2014 (Cavole et al. 2016). This was evident in their weight for a given length when examining the condition index graphs (Figure 4). Linear regression analysis indicated a significantly (p<0.05) lower intercept and steeper slope in 2015 suggesting that smaller fish had poorer condition in 2015, but larger fish appeared to be unaffected. Whether the smaller fish died off and only larger fish survived is uncertain, although there appears to be a genetic basis of selection where a number of loci were identified as being associated with fish body condition (% lipid and condition index). This was not observed for the fish collected in 2014.