Contrasting genetic structure of sympatric congeneric gastropods: Do differences in habitat preference, abundance and distribution matter?

Ocean and Earth Sciences, National Oceanography Centre Southampton, University of Southampton, Southampton, UK Department of Life Sciences, Natural History Museum, London, UK Biological Sciences, University of Southampton, Southampton, UK The Marine Biological Association of the UK, The Laboratory, Plymouth, UK Centre for Ecological Genomics and Wildlife Conservation, University of Johannesburg, Johannesburg, South Africa

for the habitat specialist. Our study highlights the importance of considering ecological (e.g. abundance, habitat preferences) and abiotic variables (e.g. ocean currents and temperature) for understanding differences in genetic structure of sympatrically distributed congeners.

K E Y W O R D S
abundance, comparative population genetics, habitat specialist, intertidal ecology, Microsatellites, north-east Atlantic, Steromphala
To explore whether differences in species' abundance and habitat specificity affect population genetic structure, it is useful to compare closely related species. Ideally this comparison would involve co-distributed congeners with the same propagule dispersal type. Such species would experience broadly similar environmental conditions and dispersal potential, but differ in their habitat preferences and overall abundance. This will increase confidence that any population genetic differences between species are related to these targeted ecological or demographic factors, rather than differences in environment or dispersal (Schiebelhut & Dawson, 2018). For example, a habitat generalist species, which should have relatively high abundances and hence, larger population sizes, may exhibit lower population subdivision than a sympatric congener with more restrictive habitat preferences (i.e. more fragmented geographic distribution), and therefore comparatively lower abundances and population sizes. For the high abundance/habitat generalist species, comparatively more offspring are expected to be produced, leading to a greater chance of recruitment, higher gene flow, and ultimately lower population genetic differentiation than for the low abundance/habitat specialist (Scheltema, 1971). Moreover, this species is likely to have shorter dispersal pathways than the habitat specialist among populations due to its more continuous geographic distribution. However, if one or both species have experienced a recent range expansion (e.g. post-glacial), then its genetic signature could mask some of these predicted population genetic patterns (Marko & Hart, 2011).
The relationship between habitat specificity and abundance with genetic population structure has been explored in terrestrial systems, where habitat generalist species exhibited more genetic connectivity than sympatric congeners with more restrictive habitat preferences and therefore a more fragmented geographic distribution (Engler et al., 2014;Kierepka et al., 2016). There are fewer examples in marine systems, but a recent study showed a positive relationship between estimates of local abundance and genetic structure, although this varied between pairs of synchronously diverging co-distributed species (Dawson, Hays, Grosberg, & Raimondi, 2014). In a review of reef community population genetics, all species with "chaotic" genetic structure (highly variable levels of genetic structuring with no obvious spatial patterning) were habitat generalists, whereas habitat specialists showed genetic differentiation between regions (Selkoe, Gaggiotti, Bowen, & Toonen, 2014). Differing habitat preferences and subsequent variation in current and historic distribution in intertidal gastropod congeners have also been associated with population genetic structure (Marko, 2004).
The trochids Steromphala umbilicalis (da Costa, 1778) and Steromphala pennanti (Philippi, 1846) were chosen to explore these patterns because they are phenotypically similar and sympatrically distributed along the rocky intertidal zone of the north-east Atlantic. The recorded range of S. umbilicalis extends from north-west Scotland (Mieszkowska, Milligan, Burrows, Freckleton, & Spencer, 2013) to Morocco (Southward, Hawkins, & Burrows, 1995). Steromphala umbilicalis is a habitat generalist, found within the mid to low-intertidal zone in rockpools, cracks, above and under boulders/rock ledges/ platforms, and the fucoid zone (Mieszkowska et al., 2013;Muñoz-Colmenero et al., 2015). Steromphala pennanti is distributed from the Cherbourg peninsula in northern France to Morocco (Southward et al., 1995) on the low-intertidal zone, and is more closely associated with fucoid habitat, such as Fucus serratus (Linnaeus, 1753), than S. umbilicalis (Bordeyne et al., 2017). Steromphala pennanti is therefore a habitat specialist and largely restricted to the fragmented distribution of fucoids along the Bay of Biscay. Regarding dispersal potential, Steromphala species are widely held to produce lecithotrophic larvae and are estimated to have approximately the same maximum larval duration of 7 to 9 days (Keith, Herbert, Norton, Hawkins, & Newton, 2011;Underwood, 1972a,b).
Within the ranges of both species lies 230 km of almost uninterrupted sandy intertidal habitat in south-west France (Castelle et al., 2006). This habitat gap (Figure 1) results in the absence of both species along that shoreline and coincides with the northern range limit of the rocky intertidal gastropods Patella rustica (Linnaeus, 1758) (Lima et al., 2007) and Stramonita haemastoma (Linnaeus, 1767) (EJGW & PBF, personal observation). However, no studies to date assess whether the habitat gap affects population connectivity of rocky shore species that are found on both sides of the gap.
Here, we explored differences in the abundance, distribution, and population genetic structure of S. umbilicalis (habitat generalist) and S. pennanti (habitat specialist) within the sympatric portion of their ranges. In order to evaluate their genetic structure, we isolated and genotyped microsatellite loci of both species. We hypothesized that, if more abundant and continuously distributed (resulting in a larger overall population size), the habitat generalist will have lower-genetic differentiation than the habitat specialist. We also ask whether genetic breaks for both species occur in association with large habitat gaps and/or other physical variables.

| Sampling
Abundance estimates were made at 23 localities ( Figure 1) at low tide in November 2014, August 2015, April 2016 and August 2016, with multiple visits to several localities. These localities were roughly evenly spaced and equally sampled north and south of the habitat gap (11 north and 12 south). At each locality, four people searched the mid to low shore in varied microhabitats where both species occurred, for 2 to 4 minutes. To prevent sampling overlap, a distance of at least 10 m was kept between each person. Timed searches were part of a wider sampling effort to quantify the presence/absence of rocky shore gastropods at each locality (~3 hr per locality). This additional sampling effort was used to increase confidence of any locality-specific S. pennanti absences. For both species, the mean number of individuals found per minute was calculated at each locality where both species were present (n = 16). The normality of abundance distributions was tested using a Shapiro-Wilk test in RSTUDIO 1.1.383, which showed a non-normal abundance distribution in S. pennanti. Therefore, a Wilcoxon-Mann-Whitney test was used in RSTUDIO to compare abundances between species over their sympatric range. Abundance measurements were also compared between mean value of localities to the north of the habitat gap in south-west France against the mean value of localities to the south of the habitat gap for both species.  Table 2) and stored in 95% ethanol. Where it was not possible to collect enough specimens of each species from the same locality, the nearest locality was used instead for the under-sampled species (usually S. pennanti due to its fragmented distribution and often low abundances). Steromphala umbilicalis was also collected from Swanage (the UK) to investigate whether the English Channel was associated with a genetic break.

| Microsatellite identification and primer development
Microsatellite markers for S. umbilicalis were developed from the transcriptome sequence generated from a single individual collected at Southsea, Hampshire, the UK. Details of the transcriptome sequencing and assembly for S. umbilicalis are given in Appendix S1 in the Supporting Information. The assembled transcriptome of S. umbilicalis was mined for microsatellites using MISA (http://pgrc.ipkgatersleben.de/misa/) with the minimum number of repeats being: dinucleotide ≥8x, trinucleotide ≥6x and tetranucleotide ≥4x. Loci without >20 bp of sequence flanking the microsatellite were removed and longer microsatellites were preferred for primer design.  (Table 1).
PCR was performed in 25 μl reactions containing 1 μl of forward and reverse primers of each microsatellite locus at 10 μM concentration, 12.5 μl of GoTaq G2 Green Master Mix (Promega, Madison, WI), 2 μl of DNA solution between 50 and 500 ng/μl concentration measured using a Nanodrop, and sterile distilled water. Thermal cycling was performed with initial denaturation for 5 minutes at 95°C, 40 cycles of 60 s at 95°C, 60 s at the given annealing temperature (Table 1), 90 s at 72°C, and a final elongation step of 15 min at 72°C. PCR product was then diluted to 1:50 volume with sterile distilled water and sent to DBS Genomics (Durham University, the UK) for fragment length analysis (FLA) using a 3,730 DNA Analyser (Applied Biosystems) with the filter set DS-30, and ROX500 as the size standard. Primers were initially trialled with two samples from three localities using a temperature gradient PCR for annealing temperatures from 52°C to 62°C, and subsequent FLA. Where amplification of polymorphic loci was unsuccessful regardless of temperature or samples, the primers were excluded.

| Microsatellite population genetics
Microsatellite data were obtained for 187 S. umbilicalis and 157 S. pennanti individuals ( (and between groups), then we can discard random genetic drift as a cause of genetic differentiation among samples (Wright, 1931).
For both species, the optimum k was determined to be 3 based on the peak in Δk, with or without prior population information.
However, GENELAND analyses estimated that S. pennanti constituted a single genetic unit (k = 1) while two different units were found in S.
umbilicalis: the first was represented by the Swanage population, and the second by the continental populations. It is possible that STRUC-TURE is less useful (Appendix S3) due to high gene flow (Waples & Gaggiotti, 2006). The DAPCs for S. umbilicalis showed substantial overlap of genetic similarity for most populations (Fig. 2a), but Swanage (and to a lesser extent, Vilagarcía) is a clear outlier population.
By contrast, the DAPC for S. pennanti showed less overlap between populations (Fig. 2b). N E and their 95% confidence interval were cal-

| DISCUSSION
Until recently, studies infrequently accounted for the potential of differing abundance and distribution patterns to help explain the contrasting patterns of genetic structure between species. Our study joins a small, but growing literature that combines such field data with population genetics for marine species [e.g. Dawson et al. (2014)]. By integrating field data on abundance and distribution with knowledge of their respective ecologies, we are able to more clearly isolate the ecological and physical mechanisms behind differences in genetic structure of the studied species. Our results are in accordance with recent studies of rocky shore invertebrates and fish (Dawson et al., 2014;Selkoe et al., 2014), as well as butterflies and mammals (Engler et al., 2014;Kierepka et al., 2016), showing that specialist species (which are often less abundant with fragmented distributions) exhibit more genetic differentiation compared to generalist species (which are often more abundant with more continuous distributions).
Within the same region, a habitat generalist species with highlocal abundance and a continuously spread distribution should yield greater reproductive output than a phenotypically similar habitat specialist congener with lower abundance and a more fragmented distribution. Assuming juvenile mobility is similar, then this will result in shorter dispersal pathways among populations and more emigrant T A B L E 3 Pairwise comparisons of microsatellite genotypic differentiation (P-value) and F-statistics: localities north of the habitat gap in bold, sites ordered from south (top) to north (bottom); significant P-value following sequential Bonferroni correction marked with asterisk (*); F ST WC = F ST calculated according to Weir and Cockerham (1984) (Dawson et al., 2014). Other factors such as differences in larval dispersal could also be important [e.g. in congeneric chthamalid barnacles, Pannacciulli, Bishop, and Hawkins (1997)], but remain largely unexplored in rocky intertidal invertebrates. Here we present that the widespread and common S.
umbilicalis shows lower genetic differentiation among localities compared to the habitat specialist, S. pennanti, which is characterized by low abundances and a more fragmented distribution (Figure 1;  (Table 3). Thus, the English Channel (which S. pennanti does not cross) appears to be a larger barrier to gene flow than the Biscayan habitat gap for S. umbilicalis, as described in both the GENELAND and STRUCTURE analyses. In S. pennanti, the greatest number of significant pairwise differences are found between San Vicente de la Barquera and all other localities (Table 3). San Vicente de la Barquera (central north Iberian coast) coincides with the most fragmented part of its geographic range that we sampled (Figure 1). In addition, this region (south of the habitat gap) is associated with the lowest abundances (albeit not significant due to the small sample size) and estimates of genetic effective population size in S. pennanti. In contrast, this region coincides with highest abundances and estimates of genetic effective population size for S. umbilicalis. Below we will further discuss the biological and physical factors that may explain the differences and commonalities of genetic structure between the two species.

| Differences between species
Steromphala umbilicalis is continuously distributed within the Bay of Biscay and shows high-overall field (census population size) and genetic population sizes (i.e. effective population size). Thus, a panmictic metapopulation fits with predictions for a generalist species such as S. umbilicalis (Dawson et al., 2014;Engler et al., 2014;Kierepka et al., 2016). By contrast, S. pennanti has a fragmented geographic distribution, lower overall abundance both in field observations and genetic estimations, and is associated with fucoids largely on more sheltered shores (Bordeyne et al., 2017). Given these clear differences between the studied species, we can look more closely at locality specific patterns of genetic differentiation with the abundance and distribution data. For example, we find significant genetic differentiation between San Vicente de la Barquera (central north Iberian coast) and all other localities in S. pennanti.
Our surveys show that S. pennanti and Fucus were largely absent within the intervening section of coastline separating St Jean de Luz (SW France) and San Vicente de la Barquera (225 km). The absence of several fucoid species in this region of the Iberian Peninsula has been attributed to its relatively high sea surface temperature (SST) (Duarte et al., 2013;Southward et al., 1995;Zardi et al., 2015).
Fucoids act as a bioengineer once established (Pocklington et al., 2017;Seed & O'Connor, 1981) and are thought to facilitate the recruitment of certain gastropods, including S. pennanti (Bordeyne et al., 2017). Conversely, there is no such genetic differentiation between any of the mainland sampled S. umbilicalis populations. This may be largely attributable to the presence of S. umbilicalis at every sampling locality, reducing isolation among populations. Steromphala pennanti is also absent from two of five localities between Vilagarcia (NW Iberian coast) and San Vicente de la Barquera (Fig. 1). This may explain the significant genetic differentiation between the two localities. Interestingly, while S. umbilicalis shows no significant genetic difference between these two localities, there is a relatively high F ST value (0.011) and Vilagarcia appears as an outlier population in the DACP (Fig. 2).
Another major difference between species is the presence of S.
umbilicalis on the British Isles. As Steromphala is thought to have originated south of the British Isles (Southward et al., 1995) umbilicalis has a longer larval duration than S. pennanti, allowing T A B L E 4 AMOVA results for: (a) Steromphala umbilicalis with loci U34184 and U36148 removed from calculations as greater than 5% values missing; (b) Steromphala pennanti with loci HEX1, NED1 and FAM3 removed from calculations as greater than 5% values missing  (Pingree & Maddock, 1977).

| Patterns associated with both species
According to our hypothesis, there should be low-genetic differentia- has a sustained and well-established fucoid habitat (Southward et al., 1995;Zardi et al., 2015), which is mirrored by the relatively consistent presence and higher abundances of S. pennanti in this region relative to south of the habitat gap (albeit a non-significant difference due to low sample number) ( Figure 1). However, we acknowledge that potential range expansion dynamics (e.g. post-glacial retreat) could also have contributed to the lack of significant population differences within the northern range of S. pennanti (Figure 1).
Furthermore, the genetic differentiation between southern Britain (Swanage) and mainland European localities of S. umbilicalis may be associated with population divergence associated with glacial cycles (Marko, 2004). Future studies should explore these possibilities using a combination of mtDNA and nuclear markers (Marko & Hart, 2011).
We found that both Steromphala species exhibit no significant genetic differentiation between the sampled localities directly north and south of the habitat gap (F ST = 0.005, p = 0.079 for S. umbilicalis and F ST = 0.000, p = 0.474 for S. pennanti), which has also been found for similar scale sandy habitat gaps for some rocky intertidal species (Ayre, Minchinton, & Perrin, 2009). One possible scenario is southward transport of larvae of both species along the surface water layer of this region during late summer (Lazure et al., 2009). Larval transport spanning this gap of~230 km would require a mean long-shore current of approximately 0.4 m/s, which would be possible considering the estimated larval duration of ≤7 days for S. umbilicalis (Keith et al., 2011) and, we assume, S. pennanti. Southward-long shore current velocities measured and modelled on this sandy coast mostly range between 0.1 m/s and 0.5 m/s, sometimes reaching 1 m/s (Castelle et al., 2006).
These currents may therefore enable southward larval transport over the habitat gap during the spawning season, resulting in genetic connectivity. As such, our results suggest that currents, larval duration and spawning season are more important than absolute distances, when considering genetic differentiation across habitat gaps.

| CONCLUSION S
Our study supports the hypothesis that, over their sympatric range, a habitat generalist with a continuous distribution and high abundances will exhibit lower-genetic differentiation compared to a congener habitat specialist with a fragmented distribution and lower abundances. These genetic differences are likely a function of contrasting population sizes and population isolation, which have long been known to influence genetic differentiation in a broad sense (Frankham, 1996;Riginos & Nachman, 2001). However, our study goes further by providing field data supporting the ecological mechanisms that likely drive the observed differences in population genetics.
Compared to the largely panmictic and continuously distributed S.
umbilicalis, higher genetic differentiation in S. pennanti is suggested to be caused by an association between habitat availability (fucoid presence/abundance), which is spatially fragmented on the N. Iberian Peninsula where genetic differentiation is highest in this species. Our research broadly supports recent studies that suggest how distribution and abundance data can help the interpretation of comparative population genetic studies (Dawson et al., 2014;Engler et al., 2014;Kierepka et al., 2016). We therefore recommend incorporating a measure of abundance, distribution, and knowledge of the ecology and life history of species in population genetics studies. This can be combined with information on the spatial distribution of habitat availability and environmental variables to inform subsequent interpretation, particularly through targeted sampling of sympatric congeners.

ACKNOWLEDG EMENTS
The authors thank E. Michel and the staff at the Natural History

S U P P O R T I N G I N F O R M A T I O N
Additional supporting information may be found online in the Supporting Information section at the end of the article.