Genomic analysis reveals neutral and adaptive patterns that challenge the current management regime for East Atlantic cod Gadus morhua L

Abstract

Challenging long-held perceptions of fish management units can help to protect vulnerable stocks. When a fishery consisting of multiple genetic stocks is managed as a single unit, overexploitation and depletion of minor genetic units can occur. Atlantic cod (Gadus morhua) is an economically and ecologically important marine species across the North Atlantic. The application of new genomic resources, including SNP arrays, allows us to detect and explore novel structure within specific cod management units. In Norwegian waters, coastal cod (i.e. those not undertaking extensive migrations) are divided into two arbitrary management units defined by ICES: one between 62° and 70°N (Norwegian coastal cod; NCC) and one between 58° and 62°N (Norwegian coastal south; NCS). Together, these capture a fishery area of >25,000 km² containing many spawning grounds. To assess whether these geographic units correctly represent genetic stocks, we analysed spawning cod of NCC and NCS for more than 8,000 SNPs along with samples of Russian White Sea cod, north-east Arctic cod (NEAC: the largest Atlantic stock), and outgroup samples representing the Irish and Faroe Sea's. Our analyses revealed large differences in spatial patterns of genetic differentiation across the genome and revealed a complex biological structure within NCC and NCS. Haplotype maps from four chromosome sets show regional specific SNP indicating a complex genetic structure. The current management plan dividing the coastal cod into only two management units does not accurately reflect the genetic units and needs to be revised. Coastal cod in Norway, while highly heterogenous, is also genetically distinct from neighbouring stocks in the north (NEAC), west (Faroe Island) and the south. The White Sea cod are highly divergent from other cod, possibly yielding support to the earlier notion of subspecies rank.

Keywords: White Sea cod; chromosomes; haplotype maps; inversions; managements.

FIGURE 1

Study area including sampling sites

FIGURE 2

Outlier scan. SNPs are ordered according to their positions on each of the 23 LGs. The estimated alpha coefficient indicates the strength and direction of selection, being positive for diversifying and negative for balancing. The alpha reported on the y‐axis: dark lines for p > .001 and red lines for p < .001

FIGURE 3

Bayesian clustering of the twelve samples with: (a) 5,854 neutral SNPs (Neutrals‐A), (b) 1,433 neutral SNPs within the four LG groups (Neutrals‐B), and (c) the pools of SNPs identified as under positive selection on LG1, LG2, LG7 and LG12 when including NEAC. Inferred ancestry of individuals was calculated after averaging ten STRUCTURE runs with CLUMPP; see Table S1 for Evanno test and StructureSelector results

FIGURE 4

Discriminant analyses of principal components (DAPC) of the samples from eleven sites (i.e. excluding the White Sea) based on neutral markers: (a) 5,854 SNPs (Neutrals‐A) and (b) 1,344 SNPs (Neutrals‐B). Plots containing the White Sea can be found in Figure S5a‐b in Supplementary Information

FIGURE 5

Discriminant analyses of principal components (DAPC) of the samples from the twelve sites based on SNPs under positive selection on: (a) LG1, (b) LG2, (c) LG7 and (d) LG12

FIGURE 6

Geographic distribution of haplotypes: (a) LG1: (b) LG2: (c) LG12: geographic repartition of the haplotypes with frequency ≥ 10% in at least one region. Each haplotype is identified by the same given colour across sampling sites. Black represents the sum of all other haplotypes displaying < 10% frequencies. (i.e. all frequencies always sum to 100%)

FIGURE 7

Geographic distribution haplotypes. The four plots represent different subregions of LG7 (cf. Table 2). (a) LG7 SNP frame 205–219, (b) LG7 SNP frame 252–266, (c) LG7 SNP frame 287–301 and (d) LG7 SNP frame 330–344. Each haplotype is identified by the same given colour across sampling sites. Black represents the sum of all other haplotypes displaying < 10% frequencies