Genetic population structure of harbour seals in the United Kingdom and neighbouring waters

Section for Evolutionary Genomics, Centre for Geogenetics, Natural History Museum of Denmark, University of Copenhagen, Copenhagen, Denmark Sea Mammal Research Unit, Scottish Oceans Institute, University of St Andrews, UK Puerto Ángel, Distrito de San Pedro Pochutla, Universidad del Mar, Ciudad Universitaria, Oaxaca, México School of Biology, University of St Andrews, UK Xelect Ltd, Scottish Oceans Institute, UK Centre d'Etudes Biologiques de Chize, Université de La Rochelle, La Rochelle, France Department of Ecology, IMARES, AD, The Netherlands 8 Institute of Marine Research, Norway Correspondence Ailsa J. Hall, Sea Mammal Research Unit, Scottish Oceans Institute, University of St Andrews, St Andrews, Fife, KY16 8LB, Scotland, UK. Email: ajh7@st‐andrews.ac.uk Funding information Scottish Government; Natural Environment Research Council, Grant/Award Number: SMRU 10001


| INTRODUCTION
It is well recognized that information on the genetic population structure and levels of genetic variation within and between populations of a species are critical to its successful conservation and management. In particular, it enables the identification of discrete units that may be evolutionarily or ecologically important and thus require specific conservation and management strategies to ensure demographic stability and maintain biodiversity (Moritz, 1994;Waples, 1998;Waples & Gaggiotti, 2006).
The harbour seal is a small phocid seal found in temperate and subarctic regions across the Northern Hemisphere. Harbour seals are relatively philopatric and form multiple discrete populations across their rangesome with geographic extensions measured in tens of kilometres, others in hundreds of kilometres, and often following seemingly distinct demographic trajectories (Andersen et al., 2011;Andersen & Olsen, 2010;Goodman, 1998;Olsen et al., 2014;Stanley et al., 1996;Westlake & O'Corry-Crowe, 2002).
Approximately 30% of the harbour seals in Europe occur in the UK (SCOS, 2014). They are widespread around the west coast of Scotland and throughout the Hebrides, Orkney and Shetland Islands (Supplementary Figure S1). In contrast, the distribution on the east coast is more or less restricted to the major estuaries of the Thames, The Wash, Firth of Tay and the Moray Firth. Overall, Scotland holds 79% of the UK harbour seal population, with 16% in England and 5% in Northern Ireland. The Irish population is about one-tenth the size of the UK population (SCOS, 2014). In the UK, harbour seal counts had been stable or increasing until about a decade ago, when declines were seen in Shetland by 30%, Orkney by 80% and the Tay Estuary by 90%. However, in other regions such as the Scottish west coast, the Outer Hebrides, Moray Firth and the English east coast counts have been stable or fluctuating. The causes of these declines are uncertain, but they are not thought to be related to the 2002 phocine distemper virus (PDV) epidemic (SCOS, 2014).
A critical first step in understanding and assessing the effects of changes in abundance is to identify groupings or units that, on both a temporal and spatial scale, are relevant for management and conservation efforts (Waples & Gaggiotti, 2006). Currently there are 11 defined harbour seal management units in the UK. The delineation of these was supported by extensive aerial survey count data collected during the harbour seal annual moult and information on movement patterns obtained from telemetry tagging programmes (SCOS, 2014;Sharples, Moss, Patterson, & Hammond, 2012). Still, there are questions that cannot be answered before a range-wide assessment of genetic population structure and diversity has been performed.
The only genetic assessment of harbour seal population structure to date that included the UK was based on seven microsatellite markers genotyped in samples from the 1988 PDV epidemic (Goodman, 1998). It investigated population structure on a pan-European scale and included four sampling sites in the UK (English east coast, Scottish east coast, Scottish west coast and Irish east coast).
The results suggested the existence of six harbour seal populations in Europe, of which Scottish and Irish localities comprised one population, and the English east coast another (the remaining four being Iceland, Wadden Sea, western Scandinavia and the Baltic Sea). Given the strong site-fidelity of European harbour seals documented by both tagging (Dietz, Teilmann, Andersen, Rigét, & Olsen, 2012;SCOS, 2014;Sharples et al., 2012) and genetic studies (Olsen et al., 2014) it is likely that additional population structuring exists within the UK, and that the two Scottish-Irish and English east coast populations reflect evolutionary significant units (senso Moritz, 1994) rather than populations on the ecological or demographic scale appropriate for management (Lowe & Allendorf, 2010;Palsbøll, Bérubé, & Allendorf, 2007;Waples & Gaggiotti, 2006). The aim of this study was to guide management and conservation efforts by assessing the population structure of harbour seals in the UK, and in Scotland in particular, on a wider and finer spatial scale than previously.  2.2 | DNA extraction and microsatellite genotyping Genomic DNA was extracted from the skin samples using a salt saturated DNA extraction method (Sunnucks & Hales, 1996

| Population structure
All analyses were conducted on two datasets; a full dataset with 12 loci, 299 animals and 13% missing data, and a reduced dataset with seven loci, 295 animals, and only 3% missing data. The five loci were excluded from the dataset because they were characterized by high levels of missing data and were flagged by the MICROCHECKER program to have a high probability of stuttering and/or null alleles.
The presence of genetic structure within UK and mainland European harbour seals was assessed by cluster analyses using the program STRUCTURE 2.3.4 (Hubisz, Falush, Stephens, & Pritchard, 2009;Pritchard, Stephens, & Donnelly, 2000). Analyses were performed under the admixture model, using the model of correlated allele frequencies between clusters and locations as priors. For each value of K from 1 to 10, five runs were performed, each with 100000 initial steps of burn-in followed by 1 000 000 iterations. To minimize the potential effects of isolation by distance, analyses were also conducted on two geographically defined subsets of the data comprising the northern UK localities and the southern UK and mainland Europe localities. Output data were processed in STRUCTURE HARVESTER (Earl, 2009) and CLUMPP (Jakobsson & Rosenberg, 2007) and graphically displayed using DISTRUCT (Rosenberg, 2004). As inference of the number of clusters K can be difficult under scenarios of extensive admixture and isolation by distance (IBD) (Falush, Stephens, & Pritchard, 2003;Pritchard et al., 2000) Evanno's ΔK was applied as an additional predictor of K (Evanno, Regnaut, & Goudet, 2005).

| Genetic diversity
For each of the clusters inferred by STRUCTURE expected and observed heterozygosity (H E and H O ) were estimated for each locus using ARLEQUIN ver. 3.5.1.2 (Excoffier & Lischer, 2010) and the allelic richness was calculated with FSTAT ver. 2.9.3.2 (Goudet, 1995), while GENEPOP (Rousset, 2008) was used to test for deviations from Hardy-Weinberg expectations and for linkage disequilibrium. Sequential Bonferroni corrections were applied to assess significance values (Rice, 1989).
Finally, given the recent decline in harbour seal populations in north-eastern UK the Wilcoxon and the sign tests implemented in the program BOTTLENECK v. 1.2 (Piry, Luikart, & Cornuet, 1999) were used to test for recent bottlenecks. Both tests were run under the stepwise mutation model (SMM), andsince microsatellite markers may not be strictly defined by the SMM, but may experience mutational jumps according to the infinite allele mutation model (IAM)the two-phase mutation model (TPM) was also applied allowing for 95% single-step mutations and 5% multi-step mutations following the recommendations of Piry et al. (1999). Tests were applied separately to samples from Orkney, Shetland, Moray Firth and Tay-Eden Estuaries using both the full and reduced datasets.

| Population structure
The results from the STRUCTURE analyses were similar for the full and the reduced datasets. They both suggested an initial division into two main groups consisting of localities in northern UK and southern UK-mainland Europe, respectively ( Figure 1a), with strong support from both likelihood and ΔK values (Supporting Table S1). Within these, the northern cluster was further divided into a north-western   Shetland and Orkney south to the Tay and Eden estuaries (Figure 1b,d).
Similarly, the southern UK-mainland Europe cluster was split into a south-eastern UK cluster consisting of haul-out sites from the Wash to Chichester harbour, including sites in France and the Dutch Wadden Sea, whereas the Norwegian harbour seals appeared to form a separate fourth cluster (Figure 1c). This Norwegian cluster appeared more closely genetically related to harbour seals at localities in south-eastern UK and mainland Europe, than the geographically  Genetic structure of harbour seals in the UK and neighbouring localities on mainland Europe estimated using the reduced dataset (7 loci; 295 animals; 3% missing data) in the program STRUCTURE ver. 2.3.4 (Hubisz et al., 2009;Pritchard et al., 2000).

| Genetic diversity
Genetic diversity in UK harbour seals was similar or slightly higher than those reported for other European harbour seal populations (Andersen et al., 2011;Olsen et al., 2014), and none of the Orkney, Shetland,

Moray Firth and Tay-Eden Estuaries sampling sites tested in BOTTLE-
NECK carried genetic signatures of recent bottlenecks. Genetic and archaeological material suggests that harbour seals colonized northern Europe 10 000 years ago (Goodman, 1998;Härkönen, Harding, Goodman, & Johannesson, 2005;Sommer & Benecke, 2003) and it is likely that associated and subsequent founder effects rather than recent population declines have shaped the main patterns of genetic variation.

| Implications for management
The Scottish and English harbour seal populations have been divided into 11 management areas based on the distribution of haul-outs and breeding sites, as well as information from tagging data (SCOS, 2014).
The results of the genetic analyses reported here lend support to the delineation of these areas, pointing to the existence of three major genetic clusters in the UK, as well as several more fine-scale genetic units ( Figure 2  The seven loci included in the reduced dataset (7 loci; 295 animals; 3% missing data) are marked with an asterisk, whereas the remaining five loci only were included in the full dataset (12 loci; 299 animals; 13% missing data). often used for the identification of management units, genetic patterns may not reflect the contemporary ecological and demographic patterns typically of interest for management (Lowe & Allendorf, 2010;Palsbøll et al., 2007;Waples & Gaggiotti, 2006). The two genetic datasets used in the present study are characterized by a moderate to relatively high proportion of missing data (12 loci; 299 animals; 13% missing data) or a relatively low number of loci (7 loci, 295 animals, 3% missing data), respectively. Thus the quality and/or power of the genetic analyses may not be adequate for a full understanding of the population structure of UK harbour seals. That said, a recent study on harbour seals in Denmark and Sweden showed a good match between the population structure inferred from genetic (15 microsatellite loci) and ecological (tagging) data, respectively (Olsen et al., 2014). It further showed that the inferred genetic populations could be regarded as separate demographic entities and thus serve as management units. Similarly, an overall agreement between the genetic results and the existing management areas for UK harbour seals was observed here. This suggests that an adaptive management approach (Holling, 1978;McLain & Lee, 1996) should be adopted for UK harbour seals, in which the delineation of the current management areas is maintained until further genetic and ecological information on movements and population dynamics has been accumulated.

| The next steps?
Future studies would certainly benefit from including more genetic data from north-east England and Ireland, as well as further neighbouring sites in Europe, to obtain a more complete understanding of harbour seal genetic structure in the British Isles and how that may connect with the north-eastern European population structure. This could include an integration of microsatellite data from published studies (Andersen et al., 2011;Olsen et al., 2014), and/or the generation of novel genome-wide single nucleotide polymorphism (SNP) data, as has been carried out for other marine mammals (Cammen et al., 2016). The latter approach would not only provide detailed information on harbour seal population structure and movement or migration patterns, but would also facilitate further research, exploring aspects such as local adaptation and pathogen susceptibility. For example, Hammond, Guethlein, Norman, and Parham (2012) found significant regional differences in UK harbour seal MHC class I genes, which are particularly important in regulating the immune response against viral infections. However, a limited number of individuals were included in their study and further work is required to determine how important such differences are at the population level and particularly in predicting their response to viral epidemics, such as phocine distemper. This level of genetic detail would thus allow for the development of bespoke conservation and management strategies, accounting for the specific characteristics of each population or management unit.

ACKNOWLEDGEMENTS
The authors would like to thank all staff at SMRU for their assistance in collecting the data, particularly Simon Moss and Ecomare for their help in collecting the samples in the Netherlands. Two anonymous reviewers are thanked for their valuable comments on earlier versions of the manuscript. Funding for this study was provided by the Scottish FIGURE 2 Population genetic structuring among harbour seals in the UK and neighbouring localities, showing the major genetic clusters identified in STRUCTURE (thick stippled lines) and minor but statistical significant levels of genetic differentiation estimated by F ST and the exact G-test (thin stippled lines). The major clusters can be interpreted as separate genetic lineages, whereas the finer scale genetic structure is likely due to some degree of demographic and/or ecological separation Government and the Natural Environment Research Council (Grant number SMRU 10001). MTO was partly funded by a postdoctoral grant from the Villum Foundation.