Parallel adaptive evolution of Atlantic cod on both sides of the Atlantic Ocean in response to temperature

Abstract

Despite the enormous economic and ecological importance of marine organisms, the spatial scales of adaptation and biocomplexity remain largely unknown. Yet, the preservation of local stocks that possess adaptive diversity is critical to the long-term maintenance of productive stable fisheries and ecosystems. Here, we document genomic evidence of range-wide adaptive differentiation in a broadcast spawning marine fish, Atlantic cod (Gadus morhua), using a genome survey of single nucleotide polymorphisms. Of 1641 gene-associated polymorphisms examined, 70 (4.2%) tested positive for signatures of selection using a Bayesian approach. We identify a subset of these loci (n=40) for which allele frequencies show parallel temperature-associated clines (p<0.001, r2=0.89) in the eastern and western north Atlantic. Temperature associations were robust to the statistical removal of geographic distance or latitude effects, and contrasted 'neutral' loci, which displayed no temperature association. Allele frequencies at temperature-associated loci were significantly correlated, spanned three linkage groups and several were successfully annotated supporting the involvement of multiple independent genes. Our results are consistent with the evolution and/or selective sweep of multiple genes in response to ocean temperature, and support the possibility of a new conservation paradigm for non-model marine organisms based on genomic approaches to resolving functional and adaptive diversity.

Figure 1.

Map of sample locations of Atlantic cod. See electronic supplementary material, table S1 for sample details and figure S1 for fine-scale map of Newfoundland locations. Note that two locations are represented by a single symbol (i.e. Smith Sound A and B, and Georges Bank A and B). Scale bar, 0–500 km.

Figure 2.

Bayesian test for selection on individual SNPs. Approach as implemented in Bayescan for the (a) western and (b) eastern Atlantic separately. Outliers were defined as those loci that possess a Bayes factor > 100. Red symbols represent outliers significantly associated with temperature. The Ogac Lake and Baltic Sea samples were omitted from outlier analysis (see §2).

Figure 3.

Spatial changes in allele frequency and associations with ocean temperature. (a) Histogram of average annual bottom temperature and plot of allele frequencies at clinal outlier SNPs and (b) association between allele frequencies for parallel clinal outlier SNPs and average annual bottom temperatures. Data include all locations (see §2). Overall relationship: r² = 0.76; p < 0.001; y = −0.08 + 0.98x.

Figure 4.

Location of temperature-associated SNPs across the cod genome. Temperature-associated loci spanned three linkage groups (Hb, haemoglobin β1-associated SNPs). See Hubert et al. (2010) for details regarding mapping crosses and Borza et al. (2009) about the SNPs associated to Hb β1. Only clinal loci shown, and two SNPs could not be placed on the map as they were not informative in any of the families used for mapping. Linkage groups are arranged LG2, LG7 and LG12 from left to right.