In recent years there has been an increase in the use of reduced representation sequencing data to develop cost-effective resources for monitoring genetic diversity (von Thaden et al 2020), with single nucleotide polymorphisms (SNPs) now used widely for non-model species such as salmonid fishes. Atlantic salmon (Salmo salar L.—hereafter referred to as salmon) is an iconic fish species that, due to its predominantly anadromous life history, is subject to multiple stressors in both its freshwater and marine habitats.

The chalk streams of the Hampshire Basin, southern England contain a genetically unique group of salmon (Finnegan et al. 2013; Ikediashi et al. 2018) which, like populations across the entire range of the species, have suffered severe declines in abundance over the last 40 years (ICES 2017). Within the Hampshire Basin chalk streams, salmon are found in only seven rivers and are threatened by agricultural pollution and over-abstraction of water (WWF 2014). Additionally, these populations have lower levels of genetic diversity (based on analysis of microsatellite loci) than neighbouring rivers in southwest England (Finnegan et al. 2013; Ikediashi et al. 2018). There is therefore a need for efficient genetic monitoring of these populations to aid in their management and conservation. Here, we develop a panel of SNP markers for use in population genetic and assignment studies in salmon populations from southern England and assess their variability in both southern English salmon and salmon from rivers outside of the target area.

DNA was extracted from adipose finclips using Qiagen Blood and Tissue kits for 84 individuals sampled from 11 rivers from southern England (Supplementary Table 1). RAD libraries were prepared in-house and 125 bp paired-end sequencing was undertaken on an Illumina HiSeq 2500 in rapid run mode (Supplementary Methods). Data was analysed using STACKS v2.41 (Rochette et al. 2019). Sequences were demultiplexed and trimmed to 122 bp using the process_radtags module. RAD loci were built using the pipeline using optimised parameters of M = 1 and n = 1 (Paris et al. 2017). The populations module was run to retain RAD loci found in at least 85% of the individuals in each of the 11 populations with a maximum observed heterozygosity of 0.7 and a minimum minor allele frequency of 0.05. A list of RAD loci containing only a single SNP were obtained and information on RAD locus length, sequence and variable SNP position were extracted from the population module outputs.

Loci were ranked based on G’’ST, calculated using the MMOD R library (Winter 2012). The top 700 loci were aligned against the Atlantic salmon reference genome (ICSASG_v2—Lien et al. 2016) using BLAST. To determine the genic nature of each RAD locus, i.e. non-coding, intron, exon etc., we used the Genome Browser on SalmoBase (, Samy et al. 2017). A whole genome duplication event in the ancestor of salmonid species approximately 80 MYA has resulted in high levels of gene duplication and synteny between different chromosomes in Atlantic salmon (Lien et al. 2016). We therefore retained only those loci aligning to a single genomic location.

SNP genotyping was undertaken on the Fluidigm EP1 Genotyping System using 96.96 Dynamic Genotyping Arrays and scored using the Fluidigm SNP Genotyping analysis software. We tested the loci on DNA extracted from both contemporary and archive tissues. Of the 106 loci tested, 12 did not give the expected genotype clusters (Supplementary Fig. 1) resulting in a final panel of 94 loci (Table 1, Supplementary Table 2). To demonstrate the utility of our SNP loci to detect variation in salmon populations, we screened loci for variation in salmon from eight rivers – five ‘target’ rivers from southern England (Tamar, Fowey, Exe, Frome and Test) and three ‘non-target’ rivers: Bresle (northern France), Wye (south Wales) and Eden (northwest England). The number of loci per linkage group ranged from 0 to 12 (Table 1). All loci were polymorphic in at least three populations. Observed and expected heterozygosity ranged from 0.054 to 0.536 and 0.064 to 0.491, respectively (Table 1). Minor allele frequency and FIS ranged from 0.035 to 0.497 and -0.293 to 0.358, respectively (Table 1). At the population level, observed and expected heterozygosity ranged from 0.289 to 0.332 and 0.275 to 0.327, respectively (Table 2). Percentage of polymorphic loci ranged from 86.17% to 94.68% (Table 2). Three significant cases of linkage between pairs of markers were found: two pairs of loci (Ssa_41748-Ssa_25077 and Ssa_25077-Ssa1354) on linkage group ssa19 in the Bresle sample and one pair in the Frome sample on linkage group ssa01 (Ssa_76064-Ssa_68872). Population pairwise G’’st (Supplementary Table 3) values ranged from 0.016 (Frome v Test) to 0.440 (Eden v Frome).

Table 1 Summary statistics for 94 single nucleotide polymorphisms developed for Atlantic salmon (Salmo salar)
Table 2 Basic diversity metrics for eight populations of British Atlantic salmon