(c) Nuclear diversity
In contrast to the mitochondrial results, there was no evidence for strong latitudinal or longitudinal diversity gradients in the nuclear dataset. Nuclear genetic diversity declined only weakly towards the poles and did not follow strong longitudinal patterns (Fig. 2C & 3C, Fig. S7). The null model performed better than any models that incorporated latitude or longitude predictors, and neither latitude nor longitude was a significant term in any of the models (Table 3). However, diversity was consistently lower for loci amplified with primers originally developed in another species (Fig. S14) and showed a negative, albeit non-significant, relationship with range position (Table 3, Fig. S10).
Nuclear diversity was also significantly associated with chlorophyll-a concentration; all models with chlorophyll-a as a predictor performed better than the null, and the model with only mean chlorophyll-a concentration performed best overall (Table 2). Similarly to the mitochondrial patterns, nuclear genetic diversity peaked at mid-to-upper chlorophyll-a concentrations (5-10 mg/m3) (Fig. 4F). Mean SST was not significantly related with nuclear genetic diversity (Table 2, Fig 4C).
As with mitochondrial genetic diversity, global patterns in nuclear genetic diversity also appeared to vary across families, although to a much diminished degree (Fig. S11-13).