Genetic analysis redraws the management boundaries for the European sprat

Abstract

Sustainable fisheries management requires detailed knowledge of population genetic structure. The European sprat is an important commercial fish distributed from Morocco to the Arctic circle, Baltic, Mediterranean, and Black seas. Prior to 2018, annual catch advice on sprat from the International Council for the Exploration of the Sea (ICES) was based on five putative stocks: (a) North Sea, (b) Kattegat-Skagerrak and Norwegian fjords, (c) Baltic Sea, (d) West of Scotland-southern Celtic Seas, and (e) English Channel. However, there were concerns that the sprat advice on stock size estimates management plan inadequately reflected the underlying biological units. Here, we used ddRAD sequencing to develop 91 SNPs that were thereafter used to genotype approximately 2,500 fish from 40 locations. Three highly distinct and relatively homogenous genetic groups were identified: (a) Norwegian fjords; (b) Northeast Atlantic including the North Sea, Kattegat-Skagerrak, Celtic Sea, and Bay of Biscay; and (c) Baltic Sea. Evidence of genetic admixture and possibly physical mixing was detected in samples collected from the transition zone between the North and Baltic seas, but not between any of the other groups. These results have already been implemented by ICES with the decision to merge the North Sea and the Kattegat-Skagerrak sprat to be assessed as a single unit, thus demonstrating that genetic data can be rapidly absorbed to align harvest regimes and biological units.

Keywords: SNPs; Sprattus sprattus; ddRADseq; fisheries; management; population structure.

Figure 1

Map of the sampling locations. The colored areas show the different management areas used by ICES for giving advice in 2019. Codes and associated full names for the sampling sites can be found in Table 1

Figure 2

Heat map of pairwise F _ST values in the bottom diagonal and the corresponding p‐values after 10,000 permutations in the top diagonal. Cells highlighted in pink depict values significantly different from zero after sequential Bonferroni correction. Greener colors indicate low differentiation (F _ST closer to zero), increasing toward red to indicate large differentiation. This matrix has also been included in the Supplementary Information to ease reading

Figure 3

Discriminant analysis of principal components (DAPCs) for sprat samples including (a) and excluding (b) the out‐groups (i.e., Adriatic Sea and Black Sea)

Figure 4

Barplot representing the proportion of individuals’ ancestry to cluster at K = 3 as inferred from Bayesian clustering in STRUCTURE using the total set of 91 SNP loci

Figure 5

Distribution of q (admixture proportion) values and 90% posterior probability intervals among individuals. Samples correspond to a random suite of 150 individuals from the North Sea locations, 150 random individuals from the Baltic Sea sites, and the F1 hybrids in silico‐obtained from these two groups, and all the individuals sampled in Uddevalla (UV), Great Belt (GB), and Øresund (ØS)

Figure 6

Plot of pairwise F _ST/(1−F _ST) between the northernmost location sampled in Norway (Holandsfjord—HOL) and each of the 37 remaining locations (excluding Adriatic Sea and Black Sea) versus the corresponding shortest water distance (in km). Plotted values fitted a regression with R ² = .6145 and p < .0001. The colors correspond to the clusters in STRUCTURE, whereas the squares depict the relation between HOL and the sites in the English Channel (EC), Celtic Sea (CEL), and Bay of Biscay (BoB)

Figure 7

Allele frequency per sample as a function of the geographic distance to the starting point of the transect (HOL, in northernmost Norway) for the loci identified as undergoing divergent selection after consensus between LOSITAN and BayeScan. Sites are represented by dots colored corresponding to the patterns of the STRUCTURE barplots, whereas the squares depict the samples from the English Channel, Celtic Sea, and Bay of Biscay