Genomic evidence of local adaptation in Scottish Margaritifera margaritifera

Genomic evidence of local adaptation in Scottish Margaritifera margaritifera | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Genomic evidence of local adaptation in Scottish Margaritifera margaritifera Victoria H. Gillman, Peter Cosgrove, Lesley T. Lancaster, Barbara Morrissey, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7363533/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 04 Dec, 2025 Read the published version in Conservation Genetics → Version 1 posted 9 You are reading this latest preprint version Abstract Climate change is a global threat that is already impacting populations and species across diverse ecosystems. Local adaptation of populations, identified through genotype-environment association, can be used to guide conservation management. The critically endangered freshwater pearl mussel Margaritifera margaritifera has a widespread Holarctic distribution across a range of habitats. Here, we used 156 samples from 18 populations of M. margaritifera across Scotland to characterize neutral population structure and identify signals of environment-associated adaptation in this endangered mussel. This study revealed a complex pattern of population structure, indicated by high genetic diversity, high genetic clustering and high inbreeding, although neutral structure was not shaped by distances or environments. Controlling for host species, freshwater pearl mussels show significant patterns of local adaptation to abiotic habitat variables, including in response to elevational range of their sub-catchment, their specific altitude, and bioclimatic factors such as isothermality. Locally adapted and generalist populations were identified which could guide genetic rescue and restocking conservation efforts, while a subset of 302 SNPs putatively under environmental selection were identified for further adaptation investigation. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Climate change is a global threat that is already impacting populations and species across diverse ecosystems. There is evidence that species with longer generation times exhibit poorer resilience and are at higher risk of extinction (Geist 2011 ; He et al. 2019 ; Albaladejo-Robles et al. 2023 ). Under the rapid pace of climate change, species with slower reproductive rates will have a reduced capacity for adaptation, as there will be fewer generations for selection (Peck 2011 ). These species will be reliant on migration, phenotypic plasticity and current genomic diversity (Peck 2011 ). Within a species, populations may have adaptive variants that confer resilience to climate change, and can therefore potentially rescue the species from extinction (Santos et al. 2023 ). Given the rapid pace of climate change, this quantification of adaptive genetic variation will aid predictions of long-lived species’ and potential response to climate change and inform management decisions (Waldvogel et al. 2020 ). High-throughput genomics quickly captures the standing genetic variation of populations across space through single nucleotide polymorphisms (SNPs). These SNPs are often spatially structured, reflecting shared ancestry of nearer neighbours. However, SNPs associated to adaptive genetic variants also can be targets of selection and therefore may turn over strongly along environmental (and not just spatial) gradients, indicating local adaptation. Testing for habitat-specific predictors in a genotype-environment association (GEA) analysis can thus identify putatively adaptive alleles (Fitzpatrick and Keller 2015 ; Capblancq et al. 2018 , 2020 ). GEA methods have been validated across multiple species using common garden approaches and historical records of population trends over time, for example rubber rabbitbrush ( Ericameria nauseosa ) and yellow warbler ( Setophaga petechia ) (Bay et al. 2018 ; Faske et al. 2021 ; Lotterhos 2024 ). Multivariate GEA models such as redundancy analysis (RDA), a type of constrained ordination, have high power in identifying multilocus adaptation (Capblancq et al. 2018 ; Capblancq and Forester 2021 ). Identifying alleles within a population significantly associated with habitat parameters is advantageous in guiding assisted gene-flow (i.e., aiding in assisting appropriate, adapted genotypes for translocation; Lotterhos, 2024). Defining population-specific adaptation across a landscape can also be used in further models to assess the genomic offset between current adaptation and predicted future climate, which allows quantification and evaluation of differences of future risk in individual populations across the species’ range (Fitzpatrick and Keller 2015 ; Capblancq et al. 2020 ). As such, GEA is a valuable tool to inform management decisions for endangered species. The freshwater pearl mussel Margaritifera margaritifera (Linnaeus, 1758) has a widespread Holarctic distribution, occurring across northern Russia, to the Iberian Peninsula and across the Atlantic to the west coast of North America (Smith 1980 ; Young et al. 2001 ). Once a significant source of pearls in Europe, these mussels also improve riverbed diversity and functioning by providing habitat and water filtration services (Strack 2006 ; Geist 2010 ). Despite their cultural and ecological importance, M. margaritifera are now critically endangered throughout Europe (Moorkens 2010 ), with up to 90% of Central European populations having been lost in the 20th century (Bauer 1988). Reasons for these declines have been attributed to a combination of pearl fishing, pollution, river engineering, changes to their host fish and increasingly the effects of climate change, such as precipitation and temperature shifts (Cosgrove et al. 2000 , 2012 , 2016 ; Hastie et al. 2003b ; Geist 2010 ). Scotland is a stronghold for actively recruiting M. margaritifera populations, although they are in rapid decline despite conservation efforts (Cosgrove et al. 2016 ). Pearl fishing in Scotland was banned in 1998, although evidence of continued illegal pearl fishing (i.e. piles of discarded shells) has been found at certain sites (Cosgrove et al. 2016 ). Freshwater pearl mussels in Scotland are found in a wide variety of locations across the country, in rivers with large/small catchments, close to/far from remote areas and within a range of climate regimes (Cosgrove et al. 2000 , 2016 ). While exploitation of M. margaritifera for their pearls has contributed to their decline, other life history traits also increase their vulnerability. Freshwater pearl mussels are obligate parasites in the early stages of their lifecycle and depend on the presence of host salmonid fish, such as brown trout ( Salmo trutta , Linnaeus, 1758) and Atlantic salmon ( Salmo salar , Linnaeus, 1758) (Bauer 1987 ). M. margaritifera larvae- termed glochidia - are released into the water column, inhaled by these fish and encyst on the gills. After an encystment period of 10–12 months (Young 1984 ), the juvenile mussels then drop off the gills to bury in the river sediment before emerging as adults and reaching sexual maturity at 12–20 years (Bauer 1987 ; Young et al. 2001 ). Adult mussels then display negligible senescence under cooler conditions where the oldest living freshwater pearl mussel was found to be 280 years old in a river in northern Sweden (Bauer 1992 ; Degerman et al. 2009 ). Consequently, freshwater pearl mussel migration and persistence is largely dependent on the host fish, and the slow generation time reduces their adaptive capacity. Freshwater pearl mussels have long been of conservation importance, so there have been multiple studies investigating genetic population structure. North, central and eastern European populations show discrete population structure and genetic diversity which are not always explained by geography (Geist and Kuehn 2005 ; Geist et al. 2010 ). However, southwestern Europe and Irish populations are more structured by geography, such as isolation-by-distance and river catchment (Cauwelier, et al. 2009 ; Stoeckle et al. 2017 ; Geist et al. 2018 ; Perea et al. 2022 ). In North America, populations are largely panmictic and display high genetic diversity (Zanatta et al. 2018 ; Farrington et al. 2020 ). Drainage-independent structure of M. margaritifera has been associated with life-history traits, such as facultative hermaphrodism and specific host use (Geist et al. 2010 ; Karlsson et al. 2014 ; Pritchard et al. 2025 ). However, adaptation to other local features such as climate has not been investigated at the genetic or genomic level. As a species with limited dispersal potential, M. margaritifera , persistence under climate change will depend on phenotypic plasticity and local adaptation. Previous translocation experiments in Central Europe have demonstrated that freshwater pearl mussels do show evidence of local adaptation (Denic et al. 2015 ). As a long-lived species, with slow reproductive rate, a GEA approach to capturing putatively adaptive alleles could be advantageous to guide conservation management (Waldvogel et al. 2020 ). Given the ongoing decline of this critically important species, urgent work is needed to better understand the adaptive landscape in M. margaritifera . First, this study will use genomic data to dissect neutral population structure of M. margaritifera across Scotland. Then, using a GEA approach, it will identify putatively adaptive loci and shed light on the environmental factors influencing genomic variation in M. margaritifera . Methods Sample collection Sample collection, DNA extraction, and DNA quality control is described in detail in Pritchard et al. (2025). Briefly, between 2019 and 2021, 396 samples for genomic analysis were collected across 20 Scottish rivers by licenced individuals. These comprised 15 samples of glochidia larvae (each produced by a different mother) and 368 visceral swab samples of adults. Due to the continued risk of pearl fishing in Scotland and subsequent legislation, the specific locations of pearl mussel rivers are treated as sensitive confidential information (Cosgrove et al. 2016). To provide informative locality data while maintaining confidentiality, a vice-county approach has been taken. Vice-counties for Great Britian are used often in museum collections (Watson 1852). Digitised 3-mile Watsonian Vice County boundary layers were obtained from the Biological Records Centre (2019). This method avoids identifying specific water bodies/catchments while still offering fine scale information. Anonymised population names use the initials of the vice-county and a number. Vice counties sampled are as follows: East Inverness-shire (EI1); East Ross (ER1); East Sutherland (ES1); North Ebudes (NE1); Outer Hebrides (OH1); West Inverness-shire (WI1, WI2, WI3, WI4); West Ross (WR1, WR2, WR3, WR4); West Sutherland (WS1, WS2, WS3, WS4, WS5). All populations showed signs of recent recruitment, other than OH1 and WI2. DNA extraction and sequencing Between November 2021 and October 2022, swabs were extracted using the Isohelix BuccalFix Plus DNA Isolation Kit while glochidia samples were extracted using Qiagen Blood and Tissue Kit. Extracts were quality checked using electrophoresis gels and quantified using a QIAxpert spectrophotometer. Those samples containing >10ng/ml DNA were further purified with Ampure XP beads following manufacturer protocol at a ratio of 40µl DNA to 16µl beads (1:0.4) to retain longer DNA fragments. Out of these, 175 samples across 23 rivers passed laboratory quality controls and were sent to SNPsaurus LLC, Eugene, Oregon for restriction site-associated DNA sequencing (RADseq), following the methods of Russello et al. (2015). Alignment, variant calling and filtering Bioinformatic analysis followed Pritchard et al. (2025). Untrimmed demultiplexed RADSeq files, supplied by SNPsaurus LLC, Eugene, Oregon, were used for alignment to minimise data loss. The untrimmed files were aligned to a concatenated genome file that included the most recent reference nuclear genome (Gomes-dos-Santos et al. 2023) and the mitochondrial genome (Gomes-dos-Santos et al. 2019). As the PHRED scores of the raw files were consistently high, alignment was completed using a Burrows-Wheeler alignment (BWA) algorithm as opposed to the alternative Bowtie2. Specifically, BWA MEM v0.7.17 under default settings was used (Li 2013) . Variants were called from the combined BAM alignment files using bcftools v1.14 (Danecek et al. 2021). The mpileup command created genotype likelihoods which were piped to bcftools call to call SNPs, with “-mv” flags to handle multiallelic variants and perform variant normalisation and “-f GQ” to provide genotype quality data for downstream filtering. An initial filter was performed using vcftools v0.1.14 (Danecek et al. 2011). Variants were filtered based on a minimum quality score of 30, a minimum depth of sequencing coverage of 20 at the site level. Genotypes below a quality threshold of 20 were excluded and finally, sites with more than 90% missing data were also excluded. Further filtering was performed using plink v1.9 (Chang et al. 2015). Variants with more than 20% missing genotypes were excluded and samples with more than 40% missing data were removed. Variants aligned to the mitochondrial genome were also removed to reduce the impact of double uniparental inheritance documented in M. margaritifera and because they exhibit different evolutionary dynamics compared to the nuclear genome (Gomes-dos-Santos et al. 2019). A minor allele frequency filter was set to 0.02 to balance the impact of false calls against retaining sensitivity for downstream association analysis (Marandel et al. 2020). Plink was also used to generate a Hardy-Weinburg equilibrium (HWE) report (Wigginton et al. 2005). Each variant was then manually inspected against HWE to identify any deviating unexpectedly from the equilibrium which could be a result of genotyping error. This was completed by hand to reduce data loss and maintain sensitivity. Finally for population genomic analysis, SNPs were pruned for linkage disequilibrium using an r2 threshold of 0.5 (within a window size of 50 SNPs and a step size of 5 SNPs) to reduce redundancy and minimise the impact of correlated variants. Environmental Predictors A suite of environmental data was extracted for use in GEA analysis. Catchment area, defined as the total area where surface water drains into a river, was extracted from publicly available catchment polygons (Cefas 2023). This was deemed relevant since larger catchment areas can act as a buffer for climate change, and particularly spate events (Hastie et al. 2003b). Altitude and freshwater data was downloaded using the sdmpredictors v0.2.15 package in R in 9.28x5.58km and 1x1km resolution respectively (Hijmans et al. 2005; Domisch et al. 2015; Bosch and Fernandez 2023). Freshwater hydroclimatic data derive from WorldClim terrestrial data (1950-2000, Hijmans et al. 2005) and the HydroSHEDs database (Lehner et al. 2008) following the framework described in Domisch et al. (2015). Elevation ranges, defined here as the difference between upstream maximum and minimum elevation within a sub-catchment, derived from HydroSHEDs (Lehner et al. 2008) were also included following the framework described in Domisch et al. (2015). For 7 small rivers for which freshwater data was unavailable, the points2nearestcell function in the rSDM v0.4.0 package was used to bump the coordinates to the nearest river raster, at a maximum of distance 10,050m (Rodriguez-Sanchez 2024). See Figure S1 for a histogram of distances moved. These freshwater hydroclimatic data were extracted using the extract function in the raster package v3.6-26 (Hijmans 2023). Finally, altitude was extracted from exact coordinates using the WorldClim terrestrial data (Hijmans et al. 2005) As pearl mussels have a long history of exploitation, a measure of isolation from human settlement was included within the association analysis, with the rationale that more isolated populations may be less impacted by poaching efforts. Wildness data was obtained from NatureScot with a spatial resolution of 25 x 25 metres per pixel (NatureScot 2014). This raster dataset gives “wildness” as a measure of distance from roads and ferries while accounting for the ruggedness of the terrain, slope, ground cover and barrier features such as cliffs or other rivers (Carver et al. 2008). Wildness for each sample was extracted using the original coordinates of the pearl mussel sites using the raster package v3.6-26 (Hijmans 2023). After the predictor variables were gathered, these were then further reduced to minimise issues of multicollinearity. This was completed by inspecting correlation plots using corrplot , initially with a correlation cutoff of 0.7, and confirmed post-hoc with variance inflation factors (tested using the vegan package) of less than ten (Wei and Simko 2021; Oksanen et al. 2024). The final predictor variables elevation range, catchment area, wildness, altitude, BIOCLIM3 (Isothermality), BIOCLIM5 (maximum temperature of the warmest month) and BIOCLIM17 (precipitation of the driest quarter) were included. Host species is known to drive some of the underlying population structure in the system (Pritchard et al. 2025). As such, host data was retrieved from Pritchard et al. (2025), who re-analysed previous glochidia count data from salmon and trout inhabiting rivers (Hastie et al. 2003a; Baum 2015, 2018; Clements et al. 2018). In some cases, populations derive from rivers where only trout were caught, and in others no host study had been performed. As such, populations were classified as “Salmon” specialist, “Trout” specialist, “Only Trout Available”, or “Unknown” as data is unavailable/inconclusive (Table 1). Data Analysis Neutral population structure Neutral population structure was first explored by Pritchard et al. (2025) as the backdrop for understanding host use and host specificity. Here, we examine neutral population structure as the backdrop for understanding environmental adaptation. Using the SNP dataset pruned for linkage disequilibrium, first the r package StAMPP v1.6.3 was used to generate fixation index (F ST ) values between each population/sample site, using 1000 bootstraps across loci (Weir and Cockerham 1984; Pembleton et al. 2013). All samples were included to minimise data loss, despite some populations having small sample sizes. Linearized F ST values (F ST /(1-F ST )) were then used to investigate isolation by distance (IBD) using both Euclidean distances and least cost distances between populations in Scotland, and isolation-by-environment (IBE). Least cost distances were calculated by first measuring distances from sampling point to river mouth using the riverdist R package with ordnance survey river network data (Ordnance Survey 2023; Tyers 2024). Next, least cost distances at sea between river mouths around Scotland were calculated using the costDistance function from gdistance v1.6.4 using a 1km resolution transition raster from Scotland’s coastline (Etten 2017). Both river distances and least cost ocean distance were summed to give a total “least cost distance” path between sampling sites, or as the “fish swims” (See Figure S2 for visualisation). IBE was tested using environmental distances calculated from a principal component analysis (PCA) of site-level environmental variables including wildness, altitude, elevation range, catchment area, and bioclimatic variables BIOCLIM3, BIOCLIM5, and BIOCLIM17. Both the full dataset of linearised F ST measurements and subsets of populations assigned to “Salmon” and “Trout” or “Only Trout Caught” were investigated for IBD and IBE using Mantel tests of Spearman’s rank correlation with 9999 permutations where appropriate. Individual-level observed and expected heterozygosity was calculated from the PLINK –het flag output (Chang et al. 2015). Population-level expected, observed heterozygosity and method-of-moments F IS (inbreeding) coefficients were also computed from the linkage-disequilibrium pruned SNPs using the “gl.report.heterozygosity()” command under default parameters in the dartR v2.9.7 package (Mijangos et al. 2022). Inbreeding in dartR is calculated using unbiased heterozygosity, which in turn is calculated using a weighted/effective sample size accounting for missing data per individual (Mijangos et al. 2022). The pi function from the radiator package (Gosselin et al. 2020) was used to generate nucleotide diversity (π) per individual and per population. Ancestry coefficients were estimated for each individual using a sparse non-negative matrix factorisation (sNMF) in the LEA v3.16.0 (Landscape and Ecological Associations) package in R (Frichot and François 2015). Cross-entropy criterion across ten repetitions was calculated to estimate the best value of K, or number of ancestral populations. A principal component analysis (PCA) was generated to infer population structure using the pcadapt v4.3.5 package in R (Privé et al. 2020). Finally, the F ST values were used to create a neighbour-joining tree using the nj function in ape v5.8 (Paradis and Schliep 2019) to further validate the clustering patterns shown in the PCA (Saitou and Nei 1987) . Genotype-environment association To test for genotype-environment association, a Redundancy Analysis (RDA) was employed using the vegan package (Oksanen et al. 2024). This method was selected because it has a low false positive rate, high true positive rate and can identify weak signatures of multilocus adaptation (Forester et al. 2018; Capblancq et al. 2020; Lind and Lotterhos 2024). As RDA cannot handle missing data, missing SNPs of individuals were imputed using the most common genotype for each SNP. The RDA was computed using the filtered predictor variables detailed above and population host specificity was used as a condition ( +Condition(Host) ). Controlling for population structure is a point of contention in RDA protocols (Forester et al. 2018). This is because controlling for population structure can reduce the impact of covariance of non-climate driven structure, but this approach can also overcorrect and reduce the sensitivity of the analysis (Forester et al. 2018). To compromise, here we controlled for known host specificity, as we know that host use is an important driver of population structure in these populations (Pritchard et al. 2025), using the categories shown in Table 1. Global R 2 of RDA models was adjusted using RsquareAdj() from the vegan package and significance of the RDA was tested using an ANOVA like permutation test (anova.cca) with 999 permutations at each the model, axis and term level (Oksanen et al. 2024). To identify putatively adaptive loci, several methods using the RDA loadings from each selected RDA axis were used. Axes were selected considering ANOVA significance and a scree plot. First, candidate loci were selected using three standard deviations from the mean loading along each selected axes following methods described by Forester et al. (2018). Second, a more robust p-value approach, selecting loci with raw p-values below a Bonferroni-corrected threshold (0.05 divided by the number of tests), was used following Capblancq et al. (2018). Finally, a q-value approach was used, selecting loci with q-values under 0.05 to control for a false discovery rate of 5%. (Storey et al. 2017; Capblancq et al. 2018). To provide an independent method of detecting local adaptation and test for robustness a separate from the RDA, pcadapt v4.3.5 was used to also identify outliers by selecting pcadapt generated p-values under a Bonferroni-corrected threshold (0.05 divided by the number of tests)(Privé et al. 2020). PCAdapt performs genome scans to identify loci under selection without the use of explanatory predictor variables (Privé et al. 2020). It is less powerful than the RDA approach and will also pick up putatively adaptive loci unrelated to local climates (Capblancq et al. 2018). However, SNP’s that overlap between RDA and pcadapt can improve confidence/validity of the results. The q-value outlier approach was selected going forward as an appropriate middle ground between sensitivity and an acceptable threshold for false positives. First the outlier SNPs were investigated using absolute Pearson correlation coefficient against the predictor values included in the RDA to identify their top correlated predictor. The outliers were then placed into a second RDA to investigate the “adaptively enriched genetic space” (Steane et al. 2014), using the same techniques as above to test significance. Finally, the outlier SNPs and their flanking 100, 500 and 1000bp nucleotides were extracted using bedtools to define the region and bedtools getfasta (v2.30.0) to extract the FASTA sequences (Camacho et al. 2009). These were then queried using blast-plus v2.14.1 using the remote command line interface. Both the full nucleotide database and a molluscan level database was searched, and the .xml results were downloaded. The results were compiled in R, where an e-value threshold of 1e -5 was applied and more general annotations (genome assembly, microsatellite, chromosome, sequence) were filtered out. Finally, top hits per SNP were retained, prioritising proximity to the SNP/smallest flank length and hits from the molluscan database over the full nucleotide database. Results Sequencing results and SNP diversity After bioinformatic filtering, a dataset of 5,486 SNPs from 156 individuals across 18 rivers was retained with an overall genotyping rate of 95.95% (Figure 1). Population sample counts ranged in number from 2-15 (average 8.667). From the linkage-disequilibrium pruned dataset of 3,456 SNPs, inbreeding values ( F IS , Table 1) ranged from -0.047 at sample site OH1 to 0.262 at site ER1 (x̄=0.0.127). Negative values, indicating outbreeding, occurred in EI1 and OH1. Observed heterozygosity (H O ) averaged at 0.155 between all individuals where population-level H O ranged between 0.050 at OH1 and 0.267 at EI1. Expected heterozygosity (H E ) had a range of 0.035-0.248 between OH1 and WR1 respectively, averaging at 0.230 between all individuals. Nucleotide diversity ranged from 0.017 at OH1 to 0.124 at WR1. Finally, OH1 had the highest private allele count at 3161, whereas WI3 had the lowest at 576. Neutral population structure A total of 3,456 SNPs pruned for linkage-disequilibrium from 156 individuals were used to investigate population structure of M. margaritifera in Scotland. The mean global F ST across all SNPs was 0.19. Pairwise F ST values between populations ranged from a maximum of 0.64 and a minimum of 0.01 where all p-values were <0.001 (Figure 2b). An isolation by distance analysis revealed no significant correlation between Euclidean distance and linearised F ST (Mantel r =0.21, p=0.055, Figure 2d), nor between least-cost-path distance and linearised F ST (Mantel r=-0.00, p=0.478, Figure 2c). Isolation by environment (PC1 = 28.50% of environmental variance) also showed no significant correlation with linearised F ST (Mantel r = –0.188, p = 0.887; Figure 2e). The analysis of ancestry coefficients using sparse non-negative matrix factorisation (sNMF) identified K=9 as the optimum number of ancestral populations for the 18 populations sampled across Scotland (Figure 3a). This was based on the lowest cross-entropy criterion of 0.526 (range: 0.520-0.538). A plot of the cross-entropy criterion for K values ranging from one to 18 can be viewed in Figure S3, along with admixture plots ranging from K=5-12 at Figure S4. Out of the 18 populations genotyped, 15 comprised of over 75% from a single ancestral cluster. Populations WR4, WS3, WS1, WS4, ER1, and WS2 had the most distinct genotypes, with each comprised of a single unique ancestral cluster and the least evidence of admixture. Population WR2 and WR3, which are a river/tributary pair share a dominant ancestral cluster. Populations WI1, WI2 and WI4, located in neighbouring small streams, were all dominated by the same ancestral cluster, while WI3 had a more admixed profile similar to EI1, NE1, WR1 and WS5. Populations ES1 and OH1 displayed the highest level of admixture. In the principal component analysis (PCA, Figure 3b), PC1 accounted for 10.39% of the genomic variance and PC2, 6.56%. Subsequent axes accounted for 4.91%, 3.50%, 3.15% and 2.75% of the variance, respectively. The PCA plot of PC1 and PC2 shows that most populations in Western Inverness-shire are clearly differentiated from the remaining populations, with the exception of WI3. Within the remaining populations, EI1, NE1, WI3, WR1and WS5 are genetically similar and form a single cluster in the PCA despite being distributed across different vice-counties, as is also the case for ER1, ES1, OH1, WR2, WR3, WS2 and WS3. Neighbouring populations WR4, WS1, WS4 also cluster together and appear quite distinct from the remaining populations. A separate PCA excluding the Western Inverness-shire populations (WI1, WI2 and WI4) demonstrated that WS1 is divergent from the remaining populations (Figure S5). The neighbour-joining tree of F ST values (Figure 3c) also revealed similar patterns of population structure to the sNMF and PCA. Again, three populations from Western Inverness-shire WI1, WI2 and WI4 formed a distinct clade, whereas the population WI3 shares a branch with EI1, NE1, WR1 and WS5, all from different vice-counties. Populations ER1, ES1, OH1and WS2 also share this same overall branch, however distances are much greater. Populations WR2 and WR3 remain similar whereas WR4, WS1, WS3 and WS4 are all have high F ST distances whilst still clustering on the same overall branch. Genotype-environment association The RDA model was performed with the full 5,486 SNP dataset as the response variable and elevation range, wildness, altitude, catchment area, BIOCLIM3 (Isothermality), BIOCLIM5 (maximum temperature of the warmest month) and BIOCLIM17 (precipitation of the driest quarter) as the predictor variables. These variables were retained since they were not strongly correlated (Figure S7). Host was used as a condition. The model was significant (F=3.711, p=0.001). The constrained part of the model (predictors) accounted for 13.700% of the variance, whereas the conditional variable (host use) accounted for 9.831%, leaving the unconstrained variance accounting for 76.470% of the variance. The overall adjusted R 2 of the RDA model was 10.206%. A further permutation test showed each of the predictor terms were significant in the model (p=0.001), where BIOCLIM17 explained most of the variation (162.3, 21.60%) followed by elevation range (144.0, 19.16%) and wildness (109.7, 14.60%). Individual axis testing showed that each RDA axis was also significant (RDA1-7, p=0.001), with maximum variation explained at 27.74% (RDA1) to a minimum of 7.19% (RDA7) (Figure 4). To identify putatively adaptive loci, multiple methods were applied to the RDA loadings. A subset of 278 SNPs was identified using a standard deviation approach, whereas 63 SNPs were identified using p-values (p<0.05) and 302 SNPs were identified as outliers using a q-value (q<0.05) approach. The comparative pcadapt analysis identified 179 outliers (p<0.05). From these four different SNP subsets, 35 SNPs were shared across all methods, where 113 pcadapt SNPs were found within the q-value SNP dataset. In all subsets, the SNPs were distributed across the genome (Figure 5, Figure S7). In a second RDA based on just the 302 q-value outlier SNPs (p=0.001, R 2 ADJ = 0.294, Figure 6), all predictors were significant in the model (p<0.001). Elevation range was the strongest predictor explaining 22.15% of the constrained variance (F = 17.05), followed by altitude (18.50%, F = 14.24) and catchment area (16.61%, F = 12.78; Figure S8). RDA1 explained 30.91% of the variance and RDA2 explained 20.75%. RDA1 was most explained by elevation range (-0.39), BIOCLIM3 (0.29) and altitude (0.27). RDA2 was explained by elevation range (0.44), BIOCLIM17 (0.34) and altitude (-0.28). In the separate absolute Pearson correlations of outliers and environmental predictors, 83 SNPs were most correlated with BIOCLIM3 (x̄=0.30), 81 SNPs with catchment area (x̄=0.30) and 56 SNPs with altitude (x̄=0.42)(Table 2). Of the 302 SNPs searched against the full NCBI nucleotide database, 32 loci had hits with an E-value under 1e-5, and of these 12 were of molluscan origin. A list of the BLAST hits per SNP is available in Table S2. These have been annotated on the Manhattan plot of RDA p-values to provide genome and RDA context of correlation (Figure S11). Most notable hits are selenium-dependent glutathione peroxidase (Se-GPx), heat shock protein (HE-90), beta-tubulin and SETMAR-like histone-lysine N-methyltransferase. Discussion Complex population structure in Scottish M. margaritifera Genomic analysis of 3,456 SNPs pruned for linkage disequilibrium across 18 populations of M. margaritifera revealed distinct and complex population structure patterns in Scotland. Overall population structure in Scotland is more pronounced compared to the genomic studies of M. margaritifera in other parts of its range (Farrington et al. 2020; Perea et al. 2022). Scottish populations have higher genomic diversity than the Massachusetts and Iberian populations and cluster into more ancestral groups (see numerical comparison in Table S3). The low genomic diversity in Iberian and Massachusetts populations have been attributed to population bottlenecks as a result of reductions in effective population size, although no evidence of significant inbreeding was found (Perea et al. 2022). Almost the opposite pattern is found in Scotland, which has both higher genomic diversity and high inferred levels of inbreeding. The higher genomic diversity in Scotland could reflect the historically large populations sizes (Cosgrove et al. 2016), which reduce the impact of genetic drift and loss of diversity. As freshwater pearl mussels have long generation times, the impact of the loss of mussels in the last 100 years may not yet be reflected in the genome. This is consistent with findings in Massachusetts where habitat fragmentation through damming in the last century has not led to significant genomic structure of M. margaritifera (Farrington et al. 2020). Neutral genomic variation among Scottish populations is not explained by Euclidian, least-cost, or environmental distances (i.e., no IBD or IBE), but could be partially explained by life-history traits. At low densities, freshwater pearl mussels display facultative hermaphroditism, self-fertilisation and high fecundity (Bauer 1987). This response could increase the chance of founder and drift effects and drive the structure observed between populations in close proximity (Bauer 1987; Geist et al. 2010). Specific adaptation to host species also likely drives the complex pattern of genomic structure and isolation (Karlsson et al. 2014; Wacker et al. 2019) and is discussed in Pritchard et al. (2025). It is possible there was human transfer of young fish or mussels in Scotland that led to these patterns, however this is unlikely due to a combination of infection rates of salmonids and post transfer survival of mussels as seen in mussel captive breeding efforts (Geist et al. 2023). Habitat and climate factors drive signals of local adaptation Although broad-scale environmental dissimilarity did not explain neutral genomic structure, specific environmental variables were associated with adaptive genetic variation when controlling for host. The RDA model showed significant evidence for environment-associated adaptation in M. margaritifera in Scotland. Freshwater pearl mussels have been observed in a range of environments and exhibit phenotypic differences in response to their environment (Eagar 1977; Curley et al. 2021; Cordero-Rivera et al. 2022; Harrison et al. 2024). Freshwater pearl mussels generally are found in more oxygenated, faster flowing water with little suspended silt/sediment although this is not always the case (see (Harrison et al. 2024). Differences in shell morphology of mussels in higher versus lower slopes has been observed, with mussels exhibiting flatter, thicker shells in the former (Eagar 1977; Cordero-Rivera et al. 2022). Freshwater pearl mussels have also shown population-specific mobility adaptations to reduce the risk of exposure to air when water levels decline (Curley et al. 2021). Therefore, the results presented here compliment previous studies, demonstrating that local adaptation is present in freshwater pearl mussels spanning a range of environments. In this study, elevation range, altitude and catchment area explained most of the genomic variation of loci identified as under selection, suggesting they contribute most to an adaptive gradient. Higher elevation range, defined here as the difference between upstream maximum and minimum elevation within a sub-catchment (Domisch et al. 2015), can increase habitat heterogeneity and amplify local adaptation (Dobrowski 2011). Higher altitude sites, often closer to the source, typically less polluted and have greater variation in water levels (Zeng et al. 2023). Several genes near the SNPs under selection may be contributing to these adaptive gradients. The gene Se-GPx is associated with protecting cells from oxidative stress, including mercury exposure in Mytilus (Chatziargyriou and Dailianis 2010), may be linked to altitude as a proxy for pollution, especially as the associated SNP is most highly correlated with altitude in the RDA (Pearson’s r=0.516). Similarly, the heat shock protein (HSP-90), associated with thermal stress and buffering phenotypic variation (Rutherford and Lindquist 1998; Queitsch et al. 2002), could also be under selection in sites of higher isothermality, particularly as the associated SNP is most correlated with BIOCLIM3 (Isothermality, Pearson’s r= 0.200). Limitations of Genomic-Environment Association As genotype-environment association analysis is a relatively new technique, specific analytical procedures are not yet standardised and will impact the results. Here, we followed the guidance of well-cited studies which test the performance of GEA methods using common garden results and simulated data (Forester et al. 2018; Capblancq and Forester 2021). They found RDA to have the lowest false positive rate (Forester et al. 2018). We also weighed outlier detection techniques against sensitivity and false positives and concluded the q-value approach was an appropriate middle ground (Capblancq and Forester 2021). Outliers detected using a pcadapt approach also had some overlap with the RDA detected outliers. However, as expected the pcadapt approach did not pick up some of the more weakly correlated SNPs identified by the RDA. This is likely to do with the higher power of the RDA to account for weak, multilocus selection (Capblancq et al. 2018; Capblancq and Forester 2021). Therefore, while no single GEA procedure is without limitations, we balanced sensitivity and specificity to provide a robust analysis of genotype associations. Additionally, while the outlier SNPs are significantly associated with local factors the evidence to confirm they are truly adaptive is very limited. Current annotation of the reference genome has inhibited detailed functional genomic investigation and common garden conditions are required to experimentally test their adaptive effects (Gomes-dos-Santos et al. 2023). Because RADSeq approaches only assay genomic variation over a small part of the genome, environmentally associated genetic variation in other parts of the genome will not be identified (Arnold et al. 2013; Lowry et al. 2017). While the predictive variables were selected to increase reproducibility of this analysis into other regions of freshwater pearl mussels, the lack of direct habitat measurements adds additional uncertainty to the GEA model and subsequent identification of adaptation. Wider sequencing coverage across M. margaritifera’s range, further genome annotation, more on-site measures of habitat parameters during sampling, and common garden experiments will increase validity of whether these SNPs are truly adaptive. However, for now these data robustly identify populations (and individuals within populations) that are most adapted along various environmental gradients that are likely to be impacted by climate change and provide a set of SNPs for identifying more such individuals for future conservation or perhaps climate-adapted translocations (Capblancq and Forester 2021). Applications Despite its limitations, recent work has shown that outlier SNPs identified in RDA are suitable to accurately predict levels of local adaptation (Lotterhos 2023). Local adaptation information can inform assisted gene flow and translocations, which are conservation techniques both proposed (Perea et al. 2022) and actively being used for pearl mussels respectively (Geist et al. 2023; Lavictoire and West 2024) . Although assisting gene flow can increase genomic diversity within the gene pool, the mixing of locally adapted populations can lead to outbreeding depression and a reduction in overall fitness (Huff et al. 2011). Previous transplant investigations into survival of native/non-native juvenile mussels in Central Europe indicated success was largely dependent on local adaptation (Denic et al. 2015). Therefore, populations identified here as similarly locally adapted could be used as transplant stocks for each other. Of the 18 populations investigated, both OH1 and WI2 show no signs of recent recruitment (Watt et al. 2015). Looking at the second, “adaptive landscape” RDA (Figure 6), the OH1 population appears adaptively distinct from the other investigated populations, indicating effort for conserving this population should focus on preserving the current mussels in their local environment. On the other hand, population WI2, while genetically isolated, shows less evidence of local adaptation in relation to the parameters in this model. Evidence presented here could support WI2 restocking from any population clustered in the centre of Figure 6b, with respect to host preference. Finally, these results can be used to aid conservation of M. margaritifera under climate change. All three climate variables are significant in the GEA model and the subsequent adaptive landscape. Though not the main drivers of selection, it is still important to consider these for climate adaptation and conservation. Traditional predictive ecological models, such as species distribution models, do not consider this intraspecific variation. Genomic offset techniques calculate the degree of maladaptation between current local current adaptation and future climate and offer an evolutionary dimension for forecasting species’ response to change (Fitzpatrick and Keller 2015; Waldvogel et al. 2020; Capblancq and Forester 2021; Láruson et al. 2022; Lind and Lotterhos 2024). These techniques have been used to inform targeted climate change management of the yellow warbler ( Setophaga petechia ) and South American conifer Araucaria araucana (Bay et al. 2018; Varas-Myrik et al. 2024). Using the framework set out here, genomic offset of these critically endangered M. margaritifera populations can aid conservation efforts of Scottish populations under climate change. Conclusions In conclusion, M. margaritifera populations in Scotland display distinct and complex genomic structure patterns not explained by distance. When accounting for glochidial host, these patterns are significantly explained by local habitat parameters, including elevation range, altitude and bioclimatic variables, to which a subset of 302 particularly correlated (outlier) SNPs are identified. While further genomic sequencing efforts or experimental analysis is needed to validate these outlier SNPs, the information provided here informs translocation efforts and climate change mitigation for the conservation of this critically endangered species. Declarations Data Availability Next-RAD sequence data are available from the authors upon reasonable request and with permission from NatureScot and UHI Inverness. All scripts and other data are available at https://github.com/victoriagillman/scotpearl_adaptivelandscape_cleaned or downloadable from cited sources. Competing Interests Statement/ Conflict of Interest The authors declare no conflict of interest. Ethical Approval Pearl mussel samples were collected under applicable NatureScot license and with landowner permissions. Work with non-cephalopod invertebrates is not regulated under the UK Animals (Scientific Procedures) Act 1986 or amendments. Acknowledgments This research was funded by NatureScot and UHI Inverness. We thank Mark Coulson, Silvia Ferreira Carvalho and Jenny O’Dell for contributions to project development, Iain Sime (NatureScot) and Chris Daphne for sample collection, and Dasha Svobodova and Lydia McGill (UHI Inverness) for laboratory assistance. Paul Etter and Eric Johnson at SNPsaurus LLC facilitated the generation and quality control of RAD-seq data. VG is supported the NERC Scottish Universities Partnership for Environmental Research (SUPER) Doctoral Training Partnership (Grant reference number NE/S007342/1 to KL and website https://superdtp.st-andrews.ac.uk/) and the University of Aberdeen. Author Contribution K.L. and V.P. secured the funding for this work, and K.L., V.P. and V.G. conceived of the study. V.P. led data collection and generation, with support from B.M., P.C. and V.G. V.G. led data analysis and manuscript writing. K.L. served as the primary PhD supervisor for this work, V.P. served as the secondary PhD supervisor and L.L. served as an additional PhD supervisor. All supervisors provided guidance on data analysis and manuscript writing. References Albaladejo-Robles G, Böhm M, Newbold T (2023) Species life-history strategies affect population responses to temperature and land-cover changes. Global Change Biology 29:97–109. https://doi.org/10.1111/gcb.16454 Arnold B, Corbett-Detig RB, Hartl D, Bomblies K (2013) RADseq underestimates diversity and introduces genealogical biases due to nonrandom haplotype sampling. Molecular Ecology 22:3179–3190. https://doi.org/10.1111/MEC.12276 Bauer G (1987) Reproductive Strategy of the Freshwater Pearl Mussel Margaritifera margaritifera . Journal of Animal Ecology 56:691–704. https://doi.org/10.2307/5077 Bauer G (1992) Variation in the Life Span and Size of the Freshwater Pearl Mussel. The Journal of Animal Ecology 61:425. https://doi.org/10.2307/5333 Baum D (2015) Glochidia surveys of three rivers in Lochaber. Scottish Natural Heritage. Commissioned Report Baum D (2018) Host salmonid specificity of selected pearl mussel populations. Scottish Natural Heritage. Research Report Bay RA, Harrigan RJ, Underwood VL, et al (2018) Genomic signals of selection predict climate-driven population declines in a migratory bird. Science 359:83–86. https://doi.org/10.1126/science.aan4380 Biological Records Centre (2019) Watsonian Vice County Boundary layers for Great Britain Bosch S, Fernandez S (2023) sdmpredictors: Species Distribution Modelling Predictor Datasets Camacho C, Coulouris G, Avagyan V, et al (2009) BLAST+: architecture and applications. BMC Bioinformatics 10:421. https://doi.org/10.1186/1471-2105-10-421 Capblancq T, Fitzpatrick MC, Bay RA, et al (2020) Genomic Prediction of (Mal)Adaptation Across Current and Future Climatic Landscapes. Annual Review of Ecology, Evolution, and Systematics 51:245–269. https://doi.org/10.1146/annurev-ecolsys-020720-042553 Capblancq T, Forester BR (2021) Redundancy analysis: A Swiss Army Knife for landscape genomics. Methods in Ecology and Evolution 12:2298–2309. https://doi.org/10.1111/2041-210X.13722 Capblancq T, Luu K, Blum MGB, Bazin E (2018) Evaluation of redundancy analysis to identify signatures of local adaptation. Molecular Ecology Resources 18:1223–1233. https://doi.org/10.1111/1755-0998.12906 Carver S, Comber L, McMorran R, et al (2008) Wildness Study in the Cairngorms National Park Final Report Cauwelier, E, Verspoor E, Tarr EC, et al (2009) Genetic diversity and differentiation of freshwater pearl mussel ( Margaritifera margaritifera ) populations in the UK. Scottish Natural Heritage Commissioned Report No.XXX (ROAME No. F05AC701). Cefas (2023) Catchment boundaries for Scotland based on Scottish Environment Protection Agency (SEPA) catchments (WGS84) 2023 Chang CC, Chow CC, Tellier LC, et al (2015) Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience 4:s13742-015-0047–8. https://doi.org/10.1186/s13742-015-0047-8 Chatziargyriou V, Dailianis S (2010) The role of selenium-dependent glutathione peroxidase (Se-GPx) against oxidative and genotoxic effects of mercury in haemocytes of mussel Mytilus galloprovincialis (Lmk.). Toxicology in Vitro 24:1363–1372. https://doi.org/10.1016/j.tiv.2010.04.008 Clements EA, Thomas R, Adams CE (2018) An investigation of salmonid host utilization by the endangered freshwater pearl mussel ( Margaritifera margaritifera ) in north‐west Scotland. Aquatic Conservation 28:764–768. https://doi.org/10.1002/aqc.2900 Cordero-Rivera A, Ondina P, Outeiro A, et al (2022) Allometry in the Freshwater Pearl Mussel ( Margaritifera margaritifera L.): Mussels Tend to Grow Flatter at Higher Water Speed. mala 64:257–267. https://doi.org/10.4002/040.064.0208 Cosgrove P, Hastie L, Watt J, et al (2012) Scotland’s Freshwater Pearl Mussels: The Challenge of Climate Change. River Conservation and Management 121–132. https://doi.org/10.1002/9781119961819.CH10 Cosgrove P, Watt J, Hastie L, et al (2016) The status of the freshwater pearl mussel Margaritifera margaritifera in Scotland: extent of change since 1990s, threats and management implications. Biodivers Conserv 25:2093–2112. https://doi.org/10.1007/s10531-016-1180-0 Cosgrove PJ, Young MR, Hastie LC, et al (2000) The status of the freshwater pearl mussel Margaritifera margaritifera Linn. in Scotland. Aquatic Conservation: Marine and Freshwater Ecosystems 10:197–208. https://doi.org/10.1002/1099-0755(200005/06)10:33.0.CO;2-S Curley EAM, Thomas R, Adams CE, Stephen A (2021) Adaptive responses of freshwater pearl mussels, Margaritifera margaritifera , to managed drawdowns. Aquatic Conservation: Marine and Freshwater Ecosystems. https://doi.org/10.1002/AQC.3759 Danecek P, Auton A, Abecasis G, et al (2011) The variant call format and VCFtools. Bioinformatics 27:2156–2158. https://doi.org/10.1093/bioinformatics/btr330 Danecek P, Bonfield JK, Liddle J, et al (2021) Twelve years of SAMtools and BCFtools. GigaScience 10:. https://doi.org/10.1093/gigascience/giab008 Degerman E, Alexanderson S, Bergengren J, et al (2009) Restoration of freshwater pearl mussel streams. WWF Sweden, Solna Denic M, Taeubert J-E, Lange M, et al (2015) Influence of stock origin and environmental conditions on the survival and growth of juvenile freshwater pearl mussels ( Margaritifera margaritifera ) in a cross-exposure experiment. Limnologica 50:67–74. https://doi.org/10.1016/j.limno.2014.07.005 Dobrowski SZ (2011) A climatic basis for microrefugia: the influence of terrain on climate. Global Change Biology 17:1022–1035. https://doi.org/10.1111/j.1365-2486.2010.02263.x Domisch S, Amatulli G, Jetz W (2015) Near-global freshwater-specific environmental variables for biodiversity analyses in 1 km resolution. Sci Data 2:150073. https://doi.org/10.1038/sdata.2015.73 Eagar R (1977) Shape of Shell in Relation to Weight of Margaritifera margaritifera (l) (bivalvia-Margaritiferidae). J Conchol 29:207–218 Etten J van (2017) R Package gdistance: Distances and Routes on Geographical Grids. Journal of Statistical Software 76:1–21. https://doi.org/10.18637/jss.v076.i13 Farrington SJ, King RW, Baker JA, et al (2020) Population genetics of freshwater pearl mussel ( Margaritifera margaritifera ) in central Massachusetts and implications for conservation. Aquatic Conservation: Marine and Freshwater Ecosystems 30:1945–1958. https://doi.org/10.1002/AQC.3439 Faske TM, Agneray AC, Jahner JP, et al (2021) Genomic and common garden approaches yield complementary results for quantifying environmental drivers of local adaptation in rubber rabbitbrush, a foundational Great Basin shrub. Evol Appl 14:2881–2900. https://doi.org/10.1111/eva.13323 Fitzpatrick MC, Keller SR (2015) Ecological genomics meets community-level modelling of biodiversity: mapping the genomic landscape of current and future environmental adaptation. Ecology Letters 18:1–16. https://doi.org/10.1111/ele.12376 Forester BR, Lasky JR, Wagner HH, Urban DL (2018) Comparing methods for detecting multilocus adaptation with multivariate genotype–environment associations. Molecular Ecology 27:2215–2233. https://doi.org/10.1111/mec.14584 Frichot E, François O (2015) LEA: An R package for landscape and ecological association studies. Methods in Ecology and Evolution 6:925–929. https://doi.org/10.1111/2041-210X.12382 Geist J (2011) Integrative freshwater ecology and biodiversity conservation. Ecol Indic 11:1507–1516. https://doi.org/10.1016/j.ecolind.2011.04.002 Geist J (2010) Strategies for the conservation of endangered freshwater pearl mussels ( Margaritifera margaritifera L.): A synthesis of conservation genetics and ecology. Hydrobiologia 644:69–88. https://doi.org/10.1007/S10750-010-0190-2 Geist J, Kuehn R (2005) Genetic diversity and differentiation of central European freshwater pearl mussel ( Margaritifera margaritifera L.) populations: implications for conservation and management. Molecular Ecology 14:425–439. https://doi.org/10.1111/J.1365-294X.2004.02420.X Geist J, Moorkens E, Killeen I, et al (2018) Genetic structure of Irish freshwater pearl mussels ( Margaritifera margaritifera and Margaritifera durrovensis ): Validity of subspecies, roles of host fish, and conservation implications. Aquatic Conservation: Marine and Freshwater Ecosystems 28:923–933. https://doi.org/10.1002/AQC.2913 Geist J, Söderberg H, Karlberg A, Kuehn R (2010) Drainage-independent genetic structure and high genetic diversity of endangered freshwater pearl mussels ( Margaritifera margaritifera ) in northern Europe. Conservation Genetics 11:1339–1350. https://doi.org/10.1007/S10592-009-9963-4/TABLES/5 Geist J, Thielen F, Lavictoire L, et al (2023) Captive breeding of European freshwater mussels as a conservation tool: A review. Aquatic Conservation: Marine and Freshwater Ecosystems 33:1321–1359. https://doi.org/10.1002/aqc.4018 Gomes-dos-Santos A, Froufe E, Amaro R, et al (2019) The male and female complete mitochondrial genomes of the threatened freshwater pearl mussel Margaritifera margaritifera (Linnaeus, 1758) (Bivalvia: Margaritiferidae). Mitochondrial DNA Part B 4:1417–1420. https://doi.org/10.1080/23802359.2019.1598794 Gomes-dos-Santos A, Lopes-Lima M, Machado AM, et al (2023) The Crown Pearl V2: an improved genome assembly of the European freshwater pearl mussel Margaritifera margaritifera (Linnaeus, 1758). Gigabyte 2023:1–14. https://doi.org/10.46471/gigabyte.81 Gosselin T, Lamoth M, Devloo-Delva F, Grewe P (2020) radiator: RADseq Data Exploration, Manipulation and Visualization using R. R package version 1.1.9 Harrison S, Cowhig K, Sullivan T (2024) Still waters and deep: Habitat characteristics, density and distribution of a lake-dwelling population of the freshwater pearl mussel Margaritifera margaritifera . Global Ecology and Conservation 54:e03130. https://doi.org/10.1016/j.gecco.2024.e03130 Hastie L, Cooksley S, Scougall F, et al (2003a) Characterization of freshwater pearl mussel ( Margaritifera margaritifera ) riverine habitat using River Habitat Survey data. Aquatic Conservation: Marine and Freshwater Ecosystems 13:213–224. https://doi.org/10.1002/aqc.560 Hastie LC, Cosgrove PJ, Ellis N, Gaywood MJ (2003b) The threat of climate change to freshwater pearl mussel populations. ambi 32:40–46. https://doi.org/10.1579/0044-7447-32.1.40 He F, Zarfl C, Bremerich V, et al (2019) The global decline of freshwater megafauna. Global Change Biology 25:3883–3892. https://doi.org/10.1111/gcb.14753 Hijmans RJ (2023) raster: Geographic Data Analysis and Modeling Hijmans RJ, Cameron SE, Parra JL, et al (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25:1965–1978. https://doi.org/10.1002/joc.1276 Huff DD, Miller LM, Chizinski CJ, Vondracek B (2011) Mixed-source reintroductions lead to outbreeding depression in second-generation descendents of a native North American fish. Molecular Ecology 20:4246–4258. https://doi.org/10.1111/j.1365-294X.2011.05271.x Karlsson S, Larsen BM, Hindar K (2014) Host-dependent genetic variation in freshwater pearl mussel ( Margaritifera margaritifera L.). Hydrobiologia 735:179–190. https://doi.org/10.1007/S10750-013-1679-2/FIGURES/3 Láruson ÁJ, Fitzpatrick MC, Keller SR, et al (2022) Seeing the forest for the trees: Assessing genetic offset predictions from gradient forest. Evolutionary Applications. https://doi.org/10.1111/EVA.13354 Lavictoire L, West C (2024) Population Reinforcement of the Endangered Freshwater Pearl Mussel (Margaritifera margaritifera): Lessons Learned. Diversity 16:187. https://doi.org/10.3390/d16030187 Lehner B, Verdin K, Jarvis A (2008) New Global Hydrography Derived From Spaceborne Elevation Data. Eos, Transactions American Geophysical Union 89:93–94. https://doi.org/10.1029/2008EO100001 Li H (2013) Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. Preprint: arXiv. https://doi.org/10.48550/arXiv.1303.3997 Lind BM, Lotterhos KE (2024) The accuracy of predicting maladaptation to new environments with genomic data. Molecular Ecology Resources e14008. https://doi.org/10.1111/1755-0998.14008 Lotterhos KE (2024) Interpretation issues with “genomic vulnerability” arise from conceptual issues in local adaptation and maladaptation. Evolution Letters 8:331–339. https://doi.org/10.1093/evlett/qrae004 Lotterhos KE (2023) The paradox of adaptive trait clines with nonclinal patterns in the underlying genes. Proceedings of the National Academy of Sciences 120:e2220313120. https://doi.org/10.1073/pnas.2220313120 Lowry DB, Hoban S, Kelley JL, et al (2017) Breaking RAD: an evaluation of the utility of restriction site-associated DNA sequencing for genome scans of adaptation. Molecular Ecology Resources 17:142–152. https://doi.org/10.1111/1755-0998.12635 Marandel F, Charrier G, Lamy J, et al (2020) Estimating effective population size using RADseq: Effects of SNP selection and sample size. Ecol Evol 10:1929–1937. https://doi.org/10.1002/ece3.6016 Mijangos JL, Berry OF, Pacioni C, Georges A (2022) dartR v2: An accessible genetic analysis platform for conservation, ecology and agriculture. Methods in Ecology and Evolution. https://doi.org/10.1111/2041-210X.13918 Moorkens E (2010) IUCN Red List of Threatened Species: Margaritifera margaritifera . IUCN Red List of Threatened Species NatureScot (2014) Scotland’s wildness - remoteness. In: SpatialData.gov.scot. https://spatialdata.gov.scot/geonetwork/srv/api/records/7220922e-5a57-450d-b8ff-6ee990d38b4f. Accessed 27 Nov 2024 Oksanen J, Simpson GL, Blanchet FG, et al (2024) vegan: Community Ecology Package Ordnance Survey (2023) Open Rivers Paradis E, Schliep K (2019) ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35:526–528. https://doi.org/10.1093/bioinformatics/bty633 Peck LS (2011) Organisms and responses to environmental change. Marine Genomics 4:237–243. https://doi.org/10.1016/j.margen.2011.07.001 Pembleton LW, Cogan NOI, Forster JW (2013) StAMPP: an R package for calculation of genetic differentiation and structure of mixed-ploidy level populations. Molecular Ecology Resources 13:946–952. https://doi.org/10.1111/1755-0998.12129 Perea S, Mendes SL, Sousa-Santos C, et al (2022) Applying genomic approaches to delineate conservation strategies using the freshwater mussel Margaritifera margaritifera in the Iberian Peninsula as a model. Sci Rep 12:16894. https://doi.org/10.1038/s41598-022-20947-5 Pritchard VL, Cosgrove P, Gillman V, et al (2025) Genetic correlates of host use in Scotland’s pearl mussels. BioRXiv Privé F, Luu K, Vilhjálmsson BJ, Blum MGB (2020) Performing highly efficient genome scans for local adaptation with R package pcadapt version 4. Molecular Biology and Evolution 37:2153–2154. https://doi.org/10.1093/molbev/msaa053 Queitsch C, Sangster TA, Lindquist S (2002) Hsp90 as a capacitor of phenotypic variation. Nature 417:618–624. https://doi.org/10.1038/nature749 Rodriguez-Sanchez F (2024) rSDM: Species distribution and niche modelling in R Rutherford SL, Lindquist S (1998) Hsp90 as a capacitor for morphological evolution. Nature 396:336–342. https://doi.org/10.1038/24550 Saitou N, Nei M (1987) The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4:406–425. https://doi.org/10.1093/oxfordjournals.molbev.a040454 Santos MA, Antunes MA, Grandela A, et al (2023) Slow and population specific evolutionary response to a warming environment. Sci Rep 13:9700. https://doi.org/10.1038/s41598-023-36273-3 Smith DG (1980) Anatomical studies on Margaritifera margaritifera and Cumberlandia monodonta (Mollusca: Pelecypoda: Margaritiferidae). Zoological Journal of the Linnean Society 69:257–270. https://doi.org/10.1111/j.1096-3642.1980.tb01125.x Steane DA, Potts BM, McLean E, et al (2014) Genome-wide scans detect adaptation to aridity in a widespread forest tree species. Molecular Ecology 23:2500–2513. https://doi.org/10.1111/mec.12751 Stoeckle BC, Araujo R, Geist J, et al (2017) Strong genetic differentiation and low genetic diversity of the freshwater pearl mussel ( Margaritifera margaritifera L.) in the southwestern European distribution range. Conservation Genetics 18:147–157. https://doi.org/10.1007/S10592-016-0889-3/FIGURES/2 Storey JD, Bass AJ, Dabney A, Robinson D (2017) qvalue: Q-value estimation for false discovery rate control Strack E (2006) Pearls. Rühle-Diebener-Verlag Tyers M (2024) riverdist: River Network Distance Computation and Applications Varas-Myrik A, Sepúlveda-Espinoza F, Toro-Núñez Ó, et al (2024) Using a genomic offset approach to guide assisted gene flow in the South American conifer Araucaria araucana . Forest Ecology and Management 553:121637. https://doi.org/10.1016/j.foreco.2023.121637 Wacker S, Larsen BM, Karlsson S, Hindar K (2019) Host specificity drives genetic structure in a freshwater mussel. Sci Rep 9:10409. https://doi.org/10.1038/s41598-019-46802-8 Waldvogel A, Feldmeyer B, Rolshausen G, et al (2020) Evolutionary genomics can improve prediction of species’ responses to climate change. Evol Lett 4:4–18. https://doi.org/10.1002/evl3.154 Watson HC (1852) Cybele Britannica : or British Plants and their geographical relations. Longman and Co., Paternoster Row, London Watt J, Cosgrove PJ, Hastie LC (2015) A national freshwater pearl mussel ( Margaritifera margaritifera , L.) survey of Scotland. Scottish Natural Heritage Commissioned Report No. 901 Wei T, Simko V (2021) R package “corrplot”: Visualization of a Correlation Matrix Weir BS, Cockerham CC (1984) Estimating F-Statistics for the Analysis of Population Structure. Evolution 38:1358–1370. https://doi.org/10.2307/2408641 Wigginton JE, Cutler DJ, Abecasis GR (2005) A Note on Exact Tests of Hardy-Weinberg Equilibrium. American Journal of Human Genetics 76:887. https://doi.org/10.1086/429864 Young M (1984) The reproductive biology of the freshwater pearl mussel Margaritifera margaritifera (Linn.) in Scotland. I, Field studies. Archiv fur Hydrobiologie 99:405–422 Young MR, Cosgrove PJ, Hastie LC (2001) The Extent of, and Causes for, the Decline of a Highly Threatened Naiad: 337–357. https://doi.org/10.1007/978-3-642-56869-5_19 Zanatta DT, Stoeckle BC, Inoue K, et al (2018) High genetic diversity and low differentiation in North American Margaritifera margaritifera (Bivalvia: Unionida: Margaritiferidae). Biological Journal of the Linnean Society 123:850–863. https://doi.org/10.1093/biolinnean/bly010 Zeng C, Xing R, Huang B, et al (2023) Phytoplankton in headwater streams: spatiotemporal patterns and underlying mechanisms. Front Plant Sci 14:1276289. https://doi.org/10.3389/fpls.2023.1276289 Tables Table 1. Population names of Margaritifera margaritifera anonymised by vice counties as follows: East Inverness-shire (EI); East Ross (ER); East Sutherland (ES); North Ebudes (NE); Outer Hebrides (OH); West Inverness-shire (WI); West Ross (WR); West Sutherland (WS). Catchment area and sample size ( N ) of individuals in each population along with inbreeding ( F IS ) , observed heterozygosity and standard deviation ( H O, H O SD) , adjusted expected heterozygosity ( H E , H E SD ), nucleotide diversity ( Π ), and number of private alleles per population. Finally, host preference for each river from Pritchard et al. (2025). Population Catchment Area (km) N F IS H O H O SD H E H E SD Π Private Alleles Host EI1 1852 6 -0.009 0.267 0.252 0.239 0.186 0.12 1012 Salmon ER1 1177 6 0.262 0.119 0.182 0.147 0.189 0.073 1967 Unknown ES1 583 6 0.178 0.171 0.203 0.19 0.189 0.095 1467 Unknown NE1 1622 2 0.046 0.248 0.324 0.194 0.21 0.097 1834 Unknown OH1 303 2 -0.047 0.05 0.186 0.035 0.117 0.017 3161 Unknown WI1 983 15 0.136 0.143 0.181 0.16 0.187 0.08 1698 Only trout available WI2 983 13 0.189 0.139 0.178 0.165 0.19 0.083 1675 Only trout available WI3 983 12 0.154 0.216 0.178 0.244 0.17 0.122 576 Salmon WI4 983 9 0.165 0.14 0.192 0.158 0.191 0.079 1825 Only trout available WR1 622 10 0.191 0.212 0.18 0.248 0.174 0.124 649 Salmon WR2 1477 9 0.152 0.161 0.177 0.179 0.173 0.09 1223 Trout WR3 1477 15 0.247 0.156 0.152 0.199 0.171 0.1 839 Trout WR4 1477 11 0.167 0.095 0.162 0.108 0.169 0.054 2229 Trout WS1 1477 8 0.044 0.078 0.173 0.076 0.154 0.038 2683 Only trout available WS2 1477 10 0.049 0.175 0.209 0.174 0.187 0.087 1529 Only trout available WS3 141 8 0.177 0.115 0.185 0.131 0.183 0.066 2099 Only trout available WS4 118 8 0.092 0.104 0.183 0.107 0.168 0.054 2271 Trout WS5 118 6 0.084 0.249 0.226 0.247 0.182 0.124 912 Unknown Table 2. Summary of top absolute correlations between putatively adaptive SNPs and environmental predictors for Margaritifera margaritifera in Scotland. SNPs were identified using q<0.05 from redundancy analysis (RDA) loadings. RDA equation as follows: rda(SNPs ~ elevation range + catchment area + wildness + altitude +BIOCLIM3 (Isothermality, calculated as BIOCLIM2/BIOCLIM7 × 100) + BIOCLIM5 (maximum temperature of the warmest month) + BIOCLIM17 (precipitation of the driest quarter) +Condition(Host). Each SNP was assigned to the environmental predictor with which it had the highest absolute correlation. For each predictor, the table reports: (1) the number of outlier SNPs most strongly associated with it (Count), and (2) the mean absolute correlation across those SNPs (Average Correlation). Elevation range BIOCLIM3 BIOCLIM5 BIOCLIM17 Altitude Catchment area Wildness Count 43 83 22 8 56 81 9 Average correlation 0.30 0.30 0.27 0.25 0.42 0.30 0.29 Additional Declarations No competing interests reported. Supplementary Files GillmanetalPearlMusselConsGenSupplementary.docx Cite Share Download PDF Status: Published Journal Publication published 04 Dec, 2025 Read the published version in Conservation Genetics → Version 1 posted Editorial decision: Revision requested 27 Oct, 2025 Reviews received at journal 17 Oct, 2025 Reviewers agreed at journal 19 Sep, 2025 Reviews received at journal 08 Sep, 2025 Reviewers agreed at journal 01 Sep, 2025 Reviewers invited by journal 27 Aug, 2025 Editor assigned by journal 19 Aug, 2025 Submission checks completed at journal 19 Aug, 2025 First submitted to journal 13 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7363533","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":508701511,"identity":"6753eb18-a0e5-4d8d-ba20-89ac0370ada0","order_by":0,"name":"Victoria H. Gillman","email":"data:image/png;base64,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","orcid":"","institution":"University of Aberdeen","correspondingAuthor":true,"prefix":"","firstName":"Victoria","middleName":"H.","lastName":"Gillman","suffix":""},{"id":508701513,"identity":"2786ec1b-6cfa-4b36-ab0a-48bcb0f87fc9","order_by":1,"name":"Peter Cosgrove","email":"","orcid":"","institution":"Alba Ecology Ltd","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"","lastName":"Cosgrove","suffix":""},{"id":508701514,"identity":"5beb7247-742e-4994-aea4-dfec15a5c8df","order_by":2,"name":"Lesley T. Lancaster","email":"","orcid":"","institution":"University of Aberdeen","correspondingAuthor":false,"prefix":"","firstName":"Lesley","middleName":"T.","lastName":"Lancaster","suffix":""},{"id":508701516,"identity":"a47fd017-7750-4aac-8a94-a1fe37eb370e","order_by":3,"name":"Barbara Morrissey","email":"","orcid":"","institution":"UHI Inverness","correspondingAuthor":false,"prefix":"","firstName":"Barbara","middleName":"","lastName":"Morrissey","suffix":""},{"id":508701517,"identity":"e8925bca-ddf5-4536-af90-7767be8d07ac","order_by":4,"name":"Victoria Pritchard","email":"","orcid":"","institution":"UHI Inverness","correspondingAuthor":false,"prefix":"","firstName":"Victoria","middleName":"","lastName":"Pritchard","suffix":""},{"id":508701518,"identity":"2530be7e-0638-454b-a4be-e6ad3b18514c","order_by":5,"name":"Kara K.S. Layton","email":"","orcid":"","institution":"University of Aberdeen","correspondingAuthor":false,"prefix":"","firstName":"Kara","middleName":"K.S.","lastName":"Layton","suffix":""}],"badges":[],"createdAt":"2025-08-13 09:53:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7363533/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7363533/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s10592-025-01739-6","type":"published","date":"2025-12-04T15:57:21+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":90596642,"identity":"78be0d65-d31f-4167-95a6-22846d2dae09","added_by":"auto","created_at":"2025-09-04 13:49:08","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":367343,"visible":true,"origin":"","legend":"\u003cp\u003e(a) Map of 18 \u003cem\u003eMargaritifera margaritifera \u003c/em\u003esampling locations for population genomic analysis (N=156, 3456 SNPs). Exact locations are not plotted to retain confidentiality of the specific site. Anonymised population names use the initials of the sampling location vice-county. Vice counties are as follows: East Inverness-shire (EI); East Ross (ER); East Sutherland (ES); North Ebudes (NE); Outer Hebrides (OH); West Inverness-shire (WI); West Ross (WR); West Sutherland (WS). A box denotes populations within the same catchment area, where an asterisk (*) denotes populations located within the same catchment across West Ross and West Sutherland. Vice counties are coloured based on corresponding total number of samples. (b) Pairwise F\u003csub\u003eST ­\u003c/sub\u003e­values for all 18 populations (N=156) based on linkage disequilibrium pruned dataset of 3456 SNPs. F\u003csub\u003eST\u003c/sub\u003e values are also explored in Pritchard et al. (2024) (c) Linearised pairwise F\u003csub\u003eST\u003c/sub\u003e values against least cost path distance between sampling points including river distance (km)(d) Linearised pairwise F\u003csub\u003eST ­\u003c/sub\u003e­values versus Euclidean distance (km). (e) Linearised pairwise F\u003csub\u003eST ­\u003c/sub\u003e­values versus PCA-derived environmental distances between populations. The PCA included wildness, altitude, elevation range, catchment area, BIOCLIM3 (Isothermality, calculated as BIOCLIM2/BIOCLIM7 × 100), BIOCLIM5 (maximum temperature of the warmest month) and BIOCLIM17 (precipitation of the driest quarter, where PC1 explained 28.50% of the variance.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7363533/v1/bf897f3bed3f8085358d3710.png"},{"id":90595417,"identity":"831f0756-36d4-492e-847f-5e98083447eb","added_by":"auto","created_at":"2025-09-04 13:33:08","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":177538,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 3. Exact location names of 18 \u003cem\u003eMargaritifera margaritifera \u003c/em\u003esampling locations\u003cem\u003e \u003c/em\u003eare not provided to retain confidentiality of the specific site. Anonymised population names use the initials of the sampling location vice-county. Vice counties are as follows: East Inverness-shire (EI); East Ross (ER); East Sutherland (ES); North Ebudes (NE); Outer Hebrides (OH); West Inverness-shire (WI); West Ross (WR); West Sutherland (WS). a) Individual ancestry proportions calculated by sparse non-negative matrix factorisation (N=156,\u003cem\u003e \u003c/em\u003e3456 SNPs), where each bar represents an individual and colour represents ancestry makeup (K=9). (b) Principal component analysis of the 3456 SNPs coloured and annotated by population. (c) Neighbour-joining tree based on pairwise matrix of pairwise F\u003csub\u003eST ­\u003c/sub\u003e­values (3456 SNPs), colours here represent vice-county to show geographic closeness. Scale corresponds to F\u003csub\u003eST\u003c/sub\u003e distance between populations. \u0026nbsp;\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7363533/v1/3eb955a32ad4aa7df5d5a7ab.png"},{"id":90596300,"identity":"bce9b825-35c1-4d25-b3bc-1ea41f87fb12","added_by":"auto","created_at":"2025-09-04 13:41:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":126213,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 4. Redundancy analysis (RDA) based on 5486 SNPs across 156 individual \u003cem\u003eMargaritifera margaritifera \u003c/em\u003ein Scotland. RDA equation as follows: rda(SNPs ~ elevation range + catchment area + wildness + altitude +BIOCLIM3 (Isothermality, calculated as BIOCLIM2/BIOCLIM7 × 100) + BIOCLIM5 (maximum temperature of the warmest month) + BIOCLIM17 (precipitation of the driest quarter) +Condition(Host). Black arrows represent environmental predictors, with their direction and proportional length indicating the strength and direction of association with the RDA axes. Grey points represent individual SNPs within the RDA space and coloured points represent individuals coloured by their population name. Population names are anonymised by vice-county to retain confidentiality of the specific site. Vice counties are as follows: East Inverness-shire (EI); East Ross (ER); East Sutherland (ES); North Ebudes (NE); Outer Hebrides (OH); West Inverness-shire (WI); West Ross (WR); West Sutherland (WS).\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7363533/v1/f8b691858651d3b52b13ff7a.png"},{"id":90597806,"identity":"fe3d9b4c-685a-496f-8452-43515be8680c","added_by":"auto","created_at":"2025-09-04 13:57:08","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":222038,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 5. \u0026nbsp;Genome-wide distribution of SNPs along the x-axis and -log10(p.values) from redundancy analysis (RDA) based on 5486 SNPs across 156 individual \u003cem\u003eMargaritifera margaritifera \u003c/em\u003ein Scotland. All panels (a-d) display the same RDA output with the x-axis representing SNP position and the y-axis representing their -log₁₀(p-values). Each panel highlights different criteria used to identify outlier SNPs across 6 RDA axes: (a) ±3 loading standard deviations from the mean;(b) bonferroni corrected p-values \u0026lt;0.05;(c) q-values\u0026lt;0.05 and d) outliers detected based on pcadapt p-value results bonferroni corrected p-values \u0026lt;0.05. RDA equation as follows: rda(SNPs ~ elevation range + catchment area + wildness + altitude +BIOCLIM3 (Isothermality, calculated as BIOCLIM2/BIOCLIM7 × 100) + BIOCLIM5 (maximum temperature of the warmest month) + BIOCLIM17 (precipitation of the driest quarter) +Condition(Host). Grey dots represent neutral SNPs whereas coloured, larger dots represent outlier SNPs. Grey dotted line is at p=0.05.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7363533/v1/c94c6c400e558e2ff42767cd.png"},{"id":90595423,"identity":"0e9aa9f0-1c61-4d56-8544-4665b7378412","added_by":"auto","created_at":"2025-09-04 13:33:08","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":149611,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 6. Redundancy analysis (RDA) based on 302 outlier SNPs across 156 individual \u003cem\u003eMargaritifera margaritifera \u003c/em\u003ein Scotland. RDA equation as follows: rda(SNPs ~ elevation range + catchment area + wildness + altitude +BIOCLIM3 (Isothermality, calculated as BIOCLIM2/BIOCLIM7 × 100) + BIOCLIM5 (maximum temperature of the warmest month) + BIOCLIM17 (precipitation of the driest quarter) +Condition(Host). Black arrows represent environmental predictors, with their direction and proportional length indicating the strength and direction of association with the RDA axes. A) Pink points represent SNPs within the RDA space. B) Coloured points represent individuals coloured by their population name. Population names are anonymised by vice-county to retain confidentiality of the specific site. Vice counties are as follows: East Inverness-shire (EI); East Ross (ER); East Sutherland (ES); North Ebudes (NE); Outer Hebrides (OH); West Inverness-shire (WI); West Ross (WR); West Sutherland (WS). Populations of interest are circled and labelled with italics.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7363533/v1/b97cfafcf8b7cc1e2255cf05.png"},{"id":97723812,"identity":"8f92eb27-fdbc-4603-bbb7-714572a52f94","added_by":"auto","created_at":"2025-12-08 16:07:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1822144,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7363533/v1/6fcae504-161b-4759-addc-4643198db388.pdf"},{"id":90595424,"identity":"3df93db0-c577-496e-986d-25082e17e3c2","added_by":"auto","created_at":"2025-09-04 13:33:08","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1306865,"visible":true,"origin":"","legend":"","description":"","filename":"GillmanetalPearlMusselConsGenSupplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-7363533/v1/6b3570a0fe9570a206c0c547.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Genomic evidence of local adaptation in Scottish Margaritifera margaritifera","fulltext":[{"header":"Introduction","content":"\u003cp\u003eClimate change is a global threat that is already impacting populations and species across diverse ecosystems. There is evidence that species with longer generation times exhibit poorer resilience and are at higher risk of extinction (Geist \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; He et al. \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Albaladejo-Robles et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Under the rapid pace of climate change, species with slower reproductive rates will have a reduced capacity for adaptation, as there will be fewer generations for selection (Peck \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). These species will be reliant on migration, phenotypic plasticity and current genomic diversity (Peck \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). Within a species, populations may have adaptive variants that confer resilience to climate change, and can therefore potentially rescue the species from extinction (Santos et al. \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Given the rapid pace of climate change, this quantification of adaptive genetic variation will aid predictions of long-lived species\u0026rsquo; and potential response to climate change and inform management decisions (Waldvogel et al. \u003cspan citationid=\"CR88\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eHigh-throughput genomics quickly captures the standing genetic variation of populations across space through single nucleotide polymorphisms (SNPs). These SNPs are often spatially structured, reflecting shared ancestry of nearer neighbours. However, SNPs associated to adaptive genetic variants also can be targets of selection and therefore may turn over strongly along environmental (and not just spatial) gradients, indicating local adaptation. Testing for habitat-specific predictors in a genotype-environment association (GEA) analysis can thus identify putatively adaptive alleles (Fitzpatrick and Keller \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Capblancq et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). GEA methods have been validated across multiple species using common garden approaches and historical records of population trends over time, for example rubber rabbitbrush (\u003cem\u003eEricameria nauseosa\u003c/em\u003e) and yellow warbler (\u003cem\u003eSetophaga petechia\u003c/em\u003e) (Bay et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Faske et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Lotterhos \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Multivariate GEA models such as redundancy analysis (RDA), a type of constrained ordination, have high power in identifying multilocus adaptation (Capblancq et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Capblancq and Forester \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Identifying alleles within a population significantly associated with habitat parameters is advantageous in guiding assisted gene-flow (i.e., aiding in assisting appropriate, adapted genotypes for translocation; Lotterhos, 2024). Defining population-specific adaptation across a landscape can also be used in further models to assess the genomic offset between current adaptation and predicted future climate, which allows quantification and evaluation of differences of future risk in individual populations across the species\u0026rsquo; range (Fitzpatrick and Keller \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Capblancq et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). As such, GEA is a valuable tool to inform management decisions for endangered species.\u003c/p\u003e\u003cp\u003eThe freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e (Linnaeus, 1758) has a widespread Holarctic distribution, occurring across northern Russia, to the Iberian Peninsula and across the Atlantic to the west coast of North America (Smith \u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e1980\u003c/span\u003e; Young et al. \u003cspan citationid=\"CR95\" class=\"CitationRef\"\u003e2001\u003c/span\u003e). Once a significant source of pearls in Europe, these mussels also improve riverbed diversity and functioning by providing habitat and water filtration services (Strack \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2006\u003c/span\u003e; Geist \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). Despite their cultural and ecological importance, \u003cem\u003eM. margaritifera\u003c/em\u003e are now critically endangered throughout Europe (Moorkens \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2010\u003c/span\u003e), with up to 90% of Central European populations having been lost in the 20th century (Bauer 1988). Reasons for these declines have been attributed to a combination of pearl fishing, pollution, river engineering, changes to their host fish and increasingly the effects of climate change, such as precipitation and temperature shifts (Cosgrove et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2000\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2012\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Hastie et al. \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2003b\u003c/span\u003e; Geist \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2010\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eScotland is a stronghold for actively recruiting \u003cem\u003eM. margaritifera\u003c/em\u003e populations, although they are in rapid decline despite conservation efforts (Cosgrove et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Pearl fishing in Scotland was banned in 1998, although evidence of continued illegal pearl fishing (i.e. piles of discarded shells) has been found at certain sites (Cosgrove et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Freshwater pearl mussels in Scotland are found in a wide variety of locations across the country, in rivers with large/small catchments, close to/far from remote areas and within a range of climate regimes (Cosgrove et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2000\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eWhile exploitation of \u003cem\u003eM. margaritifera\u003c/em\u003e for their pearls has contributed to their decline, other life history traits also increase their vulnerability. Freshwater pearl mussels are obligate parasites in the early stages of their lifecycle and depend on the presence of host salmonid fish, such as brown trout (\u003cem\u003eSalmo trutta\u003c/em\u003e, Linnaeus, 1758) and Atlantic salmon (\u003cem\u003eSalmo salar\u003c/em\u003e, Linnaeus, 1758) (Bauer \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e1987\u003c/span\u003e). \u003cem\u003eM. margaritifera\u003c/em\u003e larvae- termed glochidia\u003cem\u003e-\u003c/em\u003e are released into the water column, inhaled by these fish and encyst on the gills. After an encystment period of 10\u0026ndash;12 months (Young \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e1984\u003c/span\u003e), the juvenile mussels then drop off the gills to bury in the river sediment before emerging as adults and reaching sexual maturity at 12\u0026ndash;20 years (Bauer \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e1987\u003c/span\u003e; Young et al. \u003cspan citationid=\"CR95\" class=\"CitationRef\"\u003e2001\u003c/span\u003e). Adult mussels then display negligible senescence under cooler conditions where the oldest living freshwater pearl mussel was found to be 280 years old in a river in northern Sweden (Bauer \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e1992\u003c/span\u003e; Degerman et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Consequently, freshwater pearl mussel migration and persistence is largely dependent on the host fish, and the slow generation time reduces their adaptive capacity.\u003c/p\u003e\u003cp\u003eFreshwater pearl mussels have long been of conservation importance, so there have been multiple studies investigating genetic population structure. North, central and eastern European populations show discrete population structure and genetic diversity which are not always explained by geography (Geist and Kuehn \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Geist et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). However, southwestern Europe and Irish populations are more structured by geography, such as isolation-by-distance and river catchment (Cauwelier, et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Stoeckle et al. \u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Geist et al. \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Perea et al. \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In North America, populations are largely panmictic and display high genetic diversity (Zanatta et al. \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Farrington et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Drainage-independent structure of \u003cem\u003eM. margaritifera\u003c/em\u003e has been associated with life-history traits, such as facultative hermaphrodism and specific host use (Geist et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Karlsson et al. \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Pritchard et al. \u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). However, adaptation to other local features such as climate has not been investigated at the genetic or genomic level.\u003c/p\u003e\u003cp\u003eAs a species with limited dispersal potential, \u003cem\u003eM. margaritifera\u003c/em\u003e, persistence under climate change will depend on phenotypic plasticity and local adaptation. Previous translocation experiments in Central Europe have demonstrated that freshwater pearl mussels do show evidence of local adaptation (Denic et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). As a long-lived species, with slow reproductive rate, a GEA approach to capturing putatively adaptive alleles could be advantageous to guide conservation management (Waldvogel et al. \u003cspan citationid=\"CR88\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eGiven the ongoing decline of this critically important species, urgent work is needed to better understand the adaptive landscape in \u003cem\u003eM. margaritifera\u003c/em\u003e. First, this study will use genomic data to dissect neutral population structure of \u003cem\u003eM. margaritifera\u003c/em\u003e across Scotland. Then, using a GEA approach, it will identify putatively adaptive loci and shed light on the environmental factors influencing genomic variation in \u003cem\u003eM. margaritifera\u003c/em\u003e.\u003c/p\u003e"},{"header":"Methods","content":"\u003ch2\u003eSample collection\u003c/h2\u003e\n\u003cp\u003eSample collection, DNA extraction, and DNA quality control is described in detail in Pritchard et al. (2025). Briefly, between 2019 and 2021, 396 samples for genomic analysis were collected across 20 Scottish rivers by licenced individuals. These comprised 15 samples of glochidia larvae (each produced by a different mother) and 368 visceral swab samples of adults.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDue to the continued risk of pearl fishing in Scotland and subsequent legislation, the specific locations of pearl mussel rivers are treated as sensitive confidential information (Cosgrove et al. 2016). To provide informative locality data while maintaining confidentiality, a vice-county approach has been taken. Vice-counties for Great Britian are used often in museum collections (Watson 1852). Digitised 3-mile Watsonian Vice County boundary layers were obtained from the Biological Records Centre (2019). \u0026nbsp;This method avoids identifying specific water bodies/catchments while still offering fine scale information. Anonymised population names use the initials of the vice-county and a number. Vice counties sampled are as follows: East Inverness-shire (EI1); East Ross (ER1); East Sutherland (ES1); North Ebudes (NE1); Outer Hebrides (OH1); West Inverness-shire (WI1, WI2, WI3, WI4); West Ross (WR1, WR2, WR3, WR4); West Sutherland (WS1, WS2, WS3, WS4, WS5). All populations showed signs of recent recruitment, other than OH1 and WI2.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eDNA extraction and sequencing\u003c/h2\u003e\n\u003cp\u003eBetween November 2021 and October 2022, swabs were extracted using the Isohelix BuccalFix Plus DNA Isolation Kit while glochidia samples were extracted using Qiagen Blood and Tissue Kit. Extracts were quality checked using electrophoresis gels and quantified using a QIAxpert spectrophotometer. Those samples containing \u0026gt;10ng/ml DNA were further purified with Ampure XP beads following manufacturer protocol at a ratio of 40\u0026micro;l DNA to 16\u0026micro;l beads (1:0.4) to retain longer DNA fragments. Out of these, 175 samples across 23 rivers passed laboratory quality controls and were sent to SNPsaurus LLC, Eugene, Oregon for restriction site-associated DNA sequencing (RADseq), following the methods of Russello et al. (2015). \u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eAlignment, variant calling and filtering\u003c/h2\u003e\n\u003cp\u003eBioinformatic analysis followed Pritchard et al. (2025). Untrimmed demultiplexed RADSeq files, supplied by SNPsaurus LLC, Eugene, Oregon, were used for alignment to minimise data loss. The untrimmed files were aligned to a concatenated genome file that included the most recent reference nuclear genome (Gomes-dos-Santos et al. 2023)\u0026nbsp;and the mitochondrial genome\u0026nbsp;(Gomes-dos-Santos et al. 2019). As the PHRED scores of the raw files were consistently high, alignment was completed using a Burrows-Wheeler alignment (BWA) algorithm as opposed to the alternative Bowtie2. Specifically, \u003cem\u003eBWA MEM\u003c/em\u003e v0.7.17\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eunder default settings was used\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(Li 2013)\u003cstrong\u003e.\u003c/strong\u003e Variants were called \u0026nbsp;from the combined BAM alignment files using bcftools v1.14\u0026nbsp;(Danecek et al. 2021). The \u003cem\u003empileup\u003c/em\u003e command created genotype likelihoods which were piped to \u003cem\u003ebcftools call\u003c/em\u003e to call SNPs, with \u0026ldquo;-mv\u0026rdquo; flags to handle multiallelic variants and perform variant normalisation and \u0026ldquo;-f GQ\u0026rdquo; to provide genotype quality data for downstream filtering.\u003c/p\u003e\n\u003cp\u003eAn initial filter was performed using \u003cem\u003evcftools\u003c/em\u003e v0.1.14\u0026nbsp;(Danecek et al. 2011). Variants were filtered based on a minimum quality score of 30, a minimum depth of sequencing coverage of 20 at the site level. Genotypes below a quality threshold of 20 were excluded and finally, sites with more than 90% missing data were also excluded. Further filtering was performed using \u003cem\u003eplink\u003c/em\u003e v1.9\u0026nbsp;(Chang et al. 2015). Variants with more than 20% missing genotypes were excluded and samples with more than 40% missing data were removed. Variants aligned to the mitochondrial genome were also removed to reduce the impact of double uniparental inheritance documented in \u003cem\u003eM. margaritifera\u003c/em\u003e and because they exhibit different evolutionary dynamics compared to the nuclear genome\u003cem\u003e\u0026nbsp;\u003c/em\u003e(Gomes-dos-Santos et al. 2019). A minor allele frequency filter was set to 0.02 to balance the impact of false calls against retaining sensitivity for downstream association analysis\u0026nbsp;(Marandel et al. 2020). Plink was also used to generate a Hardy-Weinburg equilibrium (HWE) report\u0026nbsp;(Wigginton et al. 2005). Each variant was then manually inspected against HWE to identify any deviating unexpectedly from the equilibrium which could be a result of genotyping error. This was completed by hand to reduce data loss and maintain sensitivity. Finally for population genomic analysis, SNPs were pruned for linkage disequilibrium using an r2 threshold of 0.5 (within a window size of 50 SNPs and a step size of 5 SNPs) to reduce redundancy and minimise the impact of correlated variants.\u003c/p\u003e\n\u003ch2\u003eEnvironmental Predictors\u003c/h2\u003e\n\u003cp\u003eA suite of environmental data was extracted for use in GEA analysis. Catchment area, defined as the total area where surface water drains into a river, was extracted from publicly available catchment polygons (Cefas 2023). This was deemed relevant since larger catchment areas can act as a buffer for climate change, and particularly spate events (Hastie et al. 2003b). Altitude and freshwater data was downloaded using the \u003cem\u003esdmpredictors\u003c/em\u003e v0.2.15 package in R in 9.28x5.58km and 1x1km resolution respectively (Hijmans et al. 2005; Domisch et al. 2015; Bosch and Fernandez 2023). Freshwater hydroclimatic data derive from WorldClim terrestrial data (1950-2000, Hijmans et al. 2005) and the HydroSHEDs database (Lehner et al. 2008) following the framework described in Domisch et al. (2015). Elevation ranges, defined here as the difference between upstream maximum and minimum elevation within a sub-catchment, derived from HydroSHEDs (Lehner et al. 2008) were also included following the framework described in Domisch et al. (2015). For 7 small rivers for which freshwater data was unavailable, the \u003cem\u003epoints2nearestcell\u003c/em\u003e function in the rSDM v0.4.0 package was used to bump the coordinates to the nearest river raster, at a maximum of distance 10,050m (Rodriguez-Sanchez 2024). See Figure S1 for a histogram of distances moved. \u0026nbsp;These freshwater hydroclimatic data were extracted using the extract function in the \u003cem\u003eraster\u003c/em\u003e package v3.6-26 (Hijmans 2023). Finally, altitude was extracted from exact coordinates using the WorldClim terrestrial data (Hijmans et al. 2005)\u003c/p\u003e\n\u003cp\u003eAs pearl mussels have a long history of exploitation, a measure of isolation from human settlement was included within the association analysis, with the rationale that more isolated populations may be less impacted by poaching efforts. Wildness data was obtained from NatureScot with a spatial resolution of 25 x 25 metres per pixel (NatureScot 2014). This raster dataset gives \u0026ldquo;wildness\u0026rdquo; as a measure of distance from roads and ferries while accounting for the ruggedness of the terrain, slope, ground cover and barrier features such as cliffs or other rivers (Carver et al. 2008). Wildness for each sample was extracted using the original coordinates of the pearl mussel sites using the \u003cem\u003eraster\u003c/em\u003e package v3.6-26 (Hijmans 2023).\u003c/p\u003e\n\u003cp\u003eAfter the predictor variables were gathered, these were then further reduced to minimise issues of multicollinearity. This was completed by inspecting correlation plots using \u003cem\u003ecorrplot\u003c/em\u003e, initially with a correlation cutoff of 0.7, and confirmed post-hoc with variance inflation factors (tested using the \u003cem\u003evegan\u003c/em\u003e package) of less than ten (Wei and Simko 2021; Oksanen et al. 2024). The final predictor variables elevation range, catchment area, wildness, altitude, BIOCLIM3 (Isothermality), BIOCLIM5 (maximum temperature of the warmest month) and BIOCLIM17 (precipitation of the driest quarter) were included.\u003c/p\u003e\n\u003cp\u003eHost species is known to drive some of the underlying population structure in the system (Pritchard et al. 2025). \u0026nbsp;As such, host data was retrieved from Pritchard et al. (2025), who re-analysed previous glochidia count data from salmon and trout inhabiting rivers (Hastie et al. 2003a; Baum 2015, 2018; Clements et al. 2018). In some cases, populations derive from rivers where only trout were caught, and in others no host study had been performed. As such, populations were classified as \u0026ldquo;Salmon\u0026rdquo; specialist, \u0026ldquo;Trout\u0026rdquo; specialist, \u0026ldquo;Only Trout Available\u0026rdquo;, or \u0026ldquo;Unknown\u0026rdquo; as data is unavailable/inconclusive (Table 1).\u003c/p\u003e\n\u003ch2\u003eData Analysis\u003c/h2\u003e\n\u003ch3\u003eNeutral population structure\u003c/h3\u003e\n\u003cp\u003eNeutral population structure was first explored by Pritchard et al. (2025) as the backdrop for understanding host use and host specificity. Here, we examine neutral population structure as the backdrop for understanding environmental adaptation. Using the SNP dataset pruned for linkage disequilibrium, first the r package \u003cem\u003eStAMPP\u0026nbsp;\u003c/em\u003ev1.6.3 was used to generate fixation index (F\u003csub\u003eST\u003c/sub\u003e) values between each population/sample site, using 1000 bootstraps across loci (Weir and Cockerham 1984; Pembleton et al. 2013). \u0026nbsp;All samples were included to minimise data loss, despite some populations having small sample sizes. Linearized F\u003csub\u003eST\u0026nbsp;\u003c/sub\u003evalues (F\u003csub\u003eST\u003c/sub\u003e/(1-F\u003csub\u003eST\u003c/sub\u003e)) were then used to investigate isolation by distance (IBD)\u003csub\u003e\u0026nbsp;\u003c/sub\u003eusing both Euclidean distances and least cost distances between populations in Scotland, and isolation-by-environment (IBE). Least cost distances were calculated by first measuring distances from sampling point to river mouth using the \u003cem\u003eriverdist\u003c/em\u003e R package with ordnance survey river network data (Ordnance Survey 2023; Tyers 2024). Next, least cost distances at sea between river mouths around Scotland were calculated using the costDistance function from \u003cem\u003egdistance\u003c/em\u003e v1.6.4 using a 1km resolution transition raster from Scotland\u0026rsquo;s coastline (Etten 2017). Both river distances and least cost ocean distance were summed to give a total \u0026ldquo;least cost distance\u0026rdquo; path between sampling sites, or as the \u0026ldquo;fish swims\u0026rdquo; (See Figure S2 for visualisation). IBE was tested using environmental distances calculated from a principal component analysis (PCA) of site-level environmental variables including wildness, altitude, elevation range, catchment area, and bioclimatic variables BIOCLIM3, BIOCLIM5, and BIOCLIM17. Both the full dataset of linearised F\u003csub\u003eST\u003c/sub\u003e measurements and subsets of populations assigned to \u0026ldquo;Salmon\u0026rdquo; and \u0026ldquo;Trout\u0026rdquo; or \u0026ldquo;Only Trout Caught\u0026rdquo; were investigated for IBD and IBE using Mantel tests of Spearman\u0026rsquo;s rank correlation with 9999 permutations where appropriate.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIndividual-level observed and expected heterozygosity was calculated from the PLINK \u0026ndash;het flag output\u0026nbsp;(Chang et al. 2015). Population-level expected, observed heterozygosity and method-of-moments F\u003csub\u003eIS\u003c/sub\u003e (inbreeding) coefficients were also computed from the linkage-disequilibrium pruned SNPs using the \u0026ldquo;gl.report.heterozygosity()\u0026rdquo; command under default parameters in the \u003cem\u003edartR\u0026nbsp;\u003c/em\u003ev2.9.7 package\u0026nbsp;(Mijangos et al. 2022). Inbreeding in dartR is calculated using unbiased heterozygosity, which in turn is calculated using a weighted/effective sample size accounting for missing data per individual\u0026nbsp;(Mijangos et al. 2022). The pi function from the \u003cem\u003eradiator\u003c/em\u003e package\u0026nbsp;(Gosselin et al. 2020)\u0026nbsp;was used to generate nucleotide diversity (\u0026pi;) per individual and per population.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAncestry coefficients were estimated for each individual using a sparse non-negative matrix factorisation (sNMF) in the \u003cem\u003eLEA\u003c/em\u003e v3.16.0 (Landscape and Ecological Associations) package in R\u0026nbsp;(Frichot and Fran\u0026ccedil;ois 2015). Cross-entropy criterion across ten repetitions was calculated to estimate the best value of K, or number of ancestral populations. A principal component analysis (PCA) was generated to infer population structure using the \u003cem\u003epcadapt\u003c/em\u003e v4.3.5 package in R\u0026nbsp;(Priv\u0026eacute; et al. 2020). Finally, the F\u003csub\u003eST\u003c/sub\u003e values were used to create a neighbour-joining tree using the nj function in \u003cem\u003eape\u003c/em\u003e v5.8 (Paradis and Schliep 2019) to further validate the clustering patterns shown in the PCA (Saitou and Nei 1987) .\u003c/p\u003e\n\u003ch3\u003eGenotype-environment association\u003c/h3\u003e\n\u003cp\u003eTo test for genotype-environment association, a Redundancy Analysis (RDA) was employed using the \u003cem\u003evegan\u003c/em\u003e package (Oksanen et al. 2024). This method was selected because it has a low false positive rate, high true positive rate and can identify weak signatures of multilocus adaptation (Forester et al. 2018; Capblancq et al. 2020; Lind and Lotterhos 2024). As RDA cannot handle missing data, missing SNPs of individuals were imputed using the most common genotype for each SNP. The RDA was computed using the filtered predictor variables detailed above and population host specificity was used as a condition (\u003cem\u003e+Condition(Host)\u003c/em\u003e). Controlling for population structure is a point of contention in RDA protocols (Forester et al. 2018). This is because controlling for population structure can reduce the impact of covariance of non-climate driven structure, but this approach can also overcorrect and reduce the sensitivity of the analysis (Forester et al. 2018). To compromise, here we controlled for known host specificity, as we know that host use is an important driver of population structure in these populations (Pritchard et al. 2025), using the categories shown in Table 1.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGlobal R\u003csup\u003e2\u003c/sup\u003e of RDA models was adjusted using RsquareAdj() from the\u003cem\u003e\u0026nbsp;vegan\u0026nbsp;\u003c/em\u003epackage and significance of the RDA was tested using an ANOVA like permutation test (anova.cca) with 999 permutations at each the model, axis and term level (Oksanen et al. 2024).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo identify putatively adaptive loci, several methods using the RDA loadings from each selected RDA axis were used. Axes were selected considering ANOVA significance and a scree plot. First, candidate loci were selected using three standard deviations from the mean loading along each selected axes following methods described by\u0026nbsp;Forester \u003cem\u003eet al.\u003c/em\u003e (2018). Second, a more robust p-value approach, selecting loci with raw p-values below a Bonferroni-corrected threshold (0.05 divided by the number of tests), was used following\u0026nbsp;Capblancq \u003cem\u003eet al.\u003c/em\u003e (2018). Finally, a q-value approach was used, selecting loci with q-values under 0.05 to control for a \u0026nbsp;false discovery rate of 5%.\u0026nbsp;\u0026nbsp;(Storey et al. 2017; Capblancq et al. 2018). To provide an independent method of detecting local adaptation and test for robustness a separate from the RDA, \u003cem\u003epcadapt\u003c/em\u003e v4.3.5 was used to also identify outliers by selecting \u003cem\u003epcadapt\u003c/em\u003e generated p-values under a Bonferroni-corrected threshold (0.05 divided by the number of tests)(Priv\u0026eacute; et al. 2020). PCAdapt performs genome scans to identify loci under selection without the use of explanatory predictor variables\u0026nbsp;(Priv\u0026eacute; et al. 2020). It is less powerful than the RDA approach and will also pick up putatively adaptive loci unrelated to local climates \u0026nbsp;(Capblancq et al. 2018). However, SNP\u0026rsquo;s that overlap between RDA and pcadapt can improve confidence/validity of the results. The q-value outlier approach was selected going forward as an appropriate middle ground between sensitivity and an acceptable threshold for false positives.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFirst the outlier SNPs were investigated using absolute Pearson correlation coefficient against the predictor values included in the RDA to identify their top correlated predictor. The outliers were then placed into a second RDA to investigate the \u0026ldquo;adaptively enriched genetic space\u0026rdquo; (Steane et al. 2014), using the same techniques as above to test significance. Finally, the outlier SNPs and their flanking 100, 500 and 1000bp nucleotides were extracted using \u003cem\u003ebedtools\u003c/em\u003e to define the region and \u003cem\u003ebedtools getfasta\u003c/em\u003e (v2.30.0) to extract the FASTA sequences (Camacho et al. 2009). These were then queried using \u003cem\u003eblast-plus\u0026nbsp;\u003c/em\u003ev2.14.1 using the remote command line interface. Both the full nucleotide database and a molluscan level database was searched, and the .xml results were downloaded. The results were compiled in R, where an e-value threshold of 1e\u003csup\u003e-5\u003c/sup\u003e was applied and more general annotations (genome assembly, microsatellite, chromosome, sequence) were filtered out. Finally, top hits per SNP were retained, prioritising proximity to the SNP/smallest flank length and hits from the molluscan database over the full nucleotide database.\u003c/p\u003e"},{"header":"Results","content":"\u003ch2\u003eSequencing results and SNP diversity\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eAfter bioinformatic filtering, a dataset of 5,486 SNPs from 156 individuals across 18 rivers was retained with an overall genotyping rate of 95.95% (Figure 1). Population sample counts ranged in number from 2-15 (average 8.667). From the linkage-disequilibrium pruned dataset of 3,456 SNPs, inbreeding values (\u003cem\u003eF\u003csub\u003eIS\u003c/sub\u003e,\u0026nbsp;\u003c/em\u003eTable 1) ranged from -0.047 at sample site OH1 to 0.262 at site ER1 (x̄=0.0.127). Negative values, indicating outbreeding, occurred in EI1 and OH1. Observed heterozygosity (H\u003csub\u003eO\u003c/sub\u003e) averaged at 0.155 between all individuals where population-level H\u003csub\u003eO\u003c/sub\u003e ranged between 0.050 at OH1 and 0.267 at EI1. Expected heterozygosity (H\u003csub\u003eE\u003c/sub\u003e) had a range of 0.035-0.248 between OH1 and WR1 respectively, averaging at 0.230 between all individuals. Nucleotide diversity ranged from 0.017 at OH1 to 0.124 at WR1. Finally, OH1 had the highest private allele count at 3161, whereas WI3 had the lowest at 576.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eNeutral population structure\u003c/h2\u003e\n\u003cp\u003eA total of 3,456 SNPs pruned for linkage-disequilibrium from 156 individuals were used to investigate population structure of \u003cem\u003eM. margaritifera\u0026nbsp;\u003c/em\u003ein Scotland. The mean global F\u003csub\u003eST\u003c/sub\u003e across all SNPs was 0.19. Pairwise F\u003csub\u003eST\u003c/sub\u003e values between populations ranged from a maximum of 0.64 and a minimum of 0.01 where all p-values were \u0026lt;0.001 (Figure 2b). An isolation by distance analysis revealed no significant correlation between Euclidean distance and linearised F\u003csub\u003eST\u0026nbsp;\u003c/sub\u003e(Mantel r =0.21, p=0.055, Figure 2d), nor between least-cost-path distance and linearised F\u003csub\u003eST\u003c/sub\u003e (Mantel r=-0.00, p=0.478, Figure 2c). Isolation by environment (PC1 = 28.50% of environmental variance) also showed no significant correlation with linearised F\u003csub\u003eST\u0026nbsp;\u003c/sub\u003e(Mantel r = \u0026ndash;0.188, p = 0.887; Figure 2e).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe analysis of ancestry coefficients using sparse non-negative matrix factorisation (sNMF) identified K=9 as the optimum number of ancestral populations for the 18 populations sampled across Scotland (Figure 3a). This was based on the lowest cross-entropy criterion of 0.526 (range: 0.520-0.538). A plot of the cross-entropy criterion for K values ranging from one to 18 can be viewed in Figure S3, along with admixture plots ranging from K=5-12 at Figure S4. \u0026nbsp;Out of the 18 populations genotyped, 15 comprised of over 75% from a single ancestral cluster. Populations WR4, WS3, WS1, WS4, ER1, and WS2 had the most distinct genotypes, with each comprised of a single unique ancestral cluster and the least evidence of admixture. Population WR2 and WR3, which are a river/tributary pair share a dominant ancestral cluster. Populations WI1, WI2 and WI4, located in neighbouring small streams, were all dominated by the same ancestral cluster, while WI3 had a more admixed profile similar to EI1, NE1, WR1 and WS5. Populations ES1 and OH1 displayed the highest level of admixture.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn the principal component analysis (PCA, Figure 3b), PC1 accounted for 10.39% of the genomic variance and PC2, 6.56%. Subsequent axes accounted for 4.91%, 3.50%, 3.15% and 2.75% of the variance, respectively. The PCA plot of PC1 and PC2 shows that most populations in Western Inverness-shire are clearly differentiated from the remaining populations, with the exception of WI3. Within the remaining populations, EI1, NE1, WI3, WR1and WS5 are genetically similar and form a single cluster in the PCA despite being distributed across different vice-counties, as is also the case for ER1, ES1, OH1, WR2, WR3, WS2 and WS3. Neighbouring populations WR4, WS1, WS4 also cluster together and appear quite distinct from the remaining populations. A separate PCA excluding the Western Inverness-shire populations (WI1, WI2 and WI4) demonstrated that WS1 is divergent from the remaining populations (Figure S5).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe neighbour-joining tree of F\u003csub\u003eST\u003c/sub\u003e values (Figure 3c) also revealed similar patterns of population structure to the sNMF and PCA. Again, three populations from Western Inverness-shire WI1, WI2 and WI4 formed a distinct clade, whereas the population WI3 shares a branch with EI1, NE1, WR1 and WS5, all from different vice-counties. \u0026nbsp;Populations ER1, ES1, OH1and WS2 also share this same overall branch, however distances are much greater. Populations WR2 and WR3 remain similar whereas WR4, WS1, WS3 and WS4 are all have high F\u003csub\u003eST\u003c/sub\u003e distances whilst still clustering on the same overall branch.\u003c/p\u003e\n\u003ch2\u003eGenotype-environment association\u003c/h2\u003e\n\u003cp\u003eThe RDA model was performed with the full 5,486 SNP dataset as the response variable and elevation range, wildness, altitude, catchment area, BIOCLIM3 (Isothermality), BIOCLIM5 (maximum temperature of the warmest month) and BIOCLIM17 (precipitation of the driest quarter) as the predictor variables. These variables were retained since they were not strongly correlated (Figure S7). Host was used as a condition. The model was significant (F=3.711, p=0.001). The constrained part of the model (predictors) accounted for 13.700% of the variance, whereas the conditional variable (host use) accounted for 9.831%, leaving the unconstrained variance accounting for 76.470% of the variance. \u0026nbsp;The overall adjusted R\u003csup\u003e2\u003c/sup\u003e of the RDA model was 10.206%.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eA further permutation test showed each of the predictor terms were significant in the model (p=0.001), where BIOCLIM17 explained most of the variation (162.3, 21.60%) followed by elevation range (144.0, 19.16%) and wildness (109.7, 14.60%). Individual axis testing showed that each RDA axis was also significant (RDA1-7, p=0.001), with maximum variation explained at 27.74% (RDA1) to a minimum of 7.19% (RDA7) (Figure 4).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo identify putatively adaptive loci, multiple methods were applied to the RDA loadings. A subset of 278 SNPs was identified using a standard deviation approach, whereas 63 SNPs were identified using p-values (p\u0026lt;0.05) and 302 SNPs were identified as outliers using a q-value (q\u0026lt;0.05) approach. The comparative \u003cem\u003epcadapt\u003c/em\u003e analysis identified 179 outliers (p\u0026lt;0.05). From these four different SNP subsets, 35 SNPs were shared across all methods, where 113 pcadapt SNPs were found within the q-value SNP dataset. In all subsets, the SNPs were distributed across the genome (Figure 5, Figure S7).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn a second RDA based on just the 302 q-value outlier SNPs (p=0.001, R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003eADJ\u003c/sub\u003e= 0.294, Figure 6), all predictors were significant in the model (p\u0026lt;0.001). Elevation range was the strongest predictor explaining 22.15% of the constrained variance (F = 17.05), followed by altitude (18.50%, F = 14.24) and catchment area (16.61%, F = 12.78; Figure S8). RDA1 explained 30.91% of the variance and RDA2 explained 20.75%. RDA1 was most explained by elevation range (-0.39), BIOCLIM3 (0.29) and altitude (0.27). RDA2 was explained by elevation range (0.44), BIOCLIM17 (0.34) and altitude (-0.28). In the separate absolute Pearson correlations of outliers and environmental predictors, 83 SNPs were most correlated with BIOCLIM3 (x̄=0.30), 81 SNPs with catchment area (x̄=0.30) and 56 SNPs with altitude (x̄=0.42)(Table 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOf the 302 SNPs searched against the full NCBI nucleotide database, 32 loci had hits with an E-value under 1e-5, and of these 12 were of molluscan origin. A list of the BLAST hits per SNP is available in Table S2. These have been annotated on the Manhattan plot of RDA p-values to provide genome and RDA context of correlation (Figure S11). Most notable hits are selenium-dependent glutathione peroxidase (Se-GPx), heat shock protein (HE-90), beta-tubulin and SETMAR-like histone-lysine N-methyltransferase.\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003ch2\u003eComplex population structure in Scottish \u003cem\u003eM. margaritifera\u0026nbsp;\u003c/em\u003e\u003c/h2\u003e\n\u003cp\u003eGenomic analysis of 3,456 SNPs pruned for linkage disequilibrium across 18 populations of \u003cem\u003eM. margaritifera\u003c/em\u003e revealed distinct and complex\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003epopulation structure patterns in Scotland. Overall population structure in Scotland is more pronounced compared to the genomic studies of \u003cem\u003eM. margaritifera\u003c/em\u003e in other parts of its range (Farrington et al. 2020; Perea et al. 2022). Scottish populations have higher genomic diversity than the Massachusetts and Iberian populations and cluster into more ancestral groups (see numerical comparison in Table S3). \u0026nbsp;The low genomic diversity in Iberian and Massachusetts populations have been attributed to population bottlenecks as a result of reductions in effective population size, although no evidence of significant inbreeding was found (Perea et al. 2022). Almost the opposite pattern is found in Scotland, which has both higher genomic diversity and high inferred levels of inbreeding. The higher genomic diversity in Scotland could reflect the historically large populations sizes (Cosgrove et al. 2016), which reduce the impact of genetic drift and loss of diversity. As freshwater pearl mussels have long generation times, the impact of the loss of mussels in the last 100 years may not yet be reflected in the genome. This is consistent with findings in Massachusetts where habitat fragmentation through damming in the last century has not led to significant genomic structure of \u003cem\u003eM. margaritifera\u0026nbsp;\u003c/em\u003e(Farrington et al. 2020).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNeutral genomic variation among Scottish populations is not explained by Euclidian, least-cost, or environmental distances (i.e., no IBD or IBE), but could be partially explained by life-history traits. At low densities, freshwater pearl mussels display facultative hermaphroditism, self-fertilisation and high fecundity (Bauer 1987). \u0026nbsp;This response could increase the chance of founder and drift effects and drive the structure observed between populations in close proximity (Bauer 1987; Geist et al. 2010). \u0026nbsp;Specific adaptation to host species also likely drives the complex pattern of genomic structure and isolation (Karlsson et al. 2014; Wacker et al. 2019) and is discussed in Pritchard et al. (2025). It is possible there was human transfer of young fish or mussels in Scotland that led to these patterns, however this is unlikely due to a combination of infection rates of salmonids and post transfer survival of mussels as seen in mussel captive breeding efforts\u0026nbsp;(Geist et al. 2023).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eHabitat and climate factors drive signals of local adaptation\u003c/h2\u003e\n\u003cp\u003eAlthough broad-scale environmental dissimilarity did not explain neutral genomic structure, specific environmental variables were associated with adaptive genetic variation when controlling for host. The RDA model showed significant evidence for environment-associated adaptation in \u003cem\u003eM. margaritifera\u003c/em\u003e in Scotland. Freshwater pearl mussels have been observed in a range of environments and exhibit phenotypic differences in response to their environment (Eagar 1977; Curley et al. 2021; Cordero-Rivera et al. 2022; Harrison et al. 2024). Freshwater pearl mussels generally are found in more oxygenated, faster flowing water with little suspended silt/sediment although this is not always the case (see (Harrison et al. 2024). Differences in shell morphology of mussels in higher versus lower slopes has been observed, with mussels exhibiting flatter, thicker shells in the former (Eagar 1977; Cordero-Rivera et al. 2022). \u0026nbsp;Freshwater pearl mussels have also shown population-specific mobility adaptations to reduce the risk of exposure to air when water levels decline (Curley et al. 2021). Therefore, the results presented here compliment previous studies, demonstrating that local adaptation is present in freshwater pearl mussels spanning a range of environments.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn this study, elevation range, altitude and catchment area explained most of the genomic variation of loci identified as under selection, suggesting they contribute most to an adaptive gradient. Higher elevation range, defined here as the difference between upstream maximum and minimum elevation within a sub-catchment (Domisch et al. 2015), can increase habitat heterogeneity and amplify local adaptation\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(Dobrowski 2011). Higher altitude sites, often closer to the source, typically less polluted and have greater variation in water levels (Zeng et al. 2023). Several genes near the SNPs under selection may be contributing to these adaptive gradients. The gene Se-GPx is associated with protecting cells from oxidative stress, including mercury exposure in Mytilus (Chatziargyriou and Dailianis 2010), may be linked to altitude as a proxy for pollution, especially as the associated SNP is most highly correlated with altitude in the RDA (Pearson\u0026rsquo;s r=0.516). Similarly, the heat shock protein (HSP-90), associated with thermal stress and buffering phenotypic variation (Rutherford and Lindquist 1998; Queitsch et al. 2002), could also be under selection in sites of higher isothermality, particularly as the associated SNP is most correlated with BIOCLIM3 (Isothermality, Pearson\u0026rsquo;s r= 0.200).\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eLimitations of Genomic-Environment Association\u003c/h2\u003e\n\u003cp\u003eAs genotype-environment association analysis is a relatively new technique, specific analytical procedures are not yet standardised and will impact the results. Here, we followed the guidance of well-cited studies which test the performance of GEA methods using common garden results and simulated data (Forester et al. 2018; Capblancq and Forester 2021). They found RDA to have the lowest false positive rate (Forester et al. 2018). We also weighed outlier detection techniques against sensitivity and false positives and concluded the q-value approach was an appropriate middle ground (Capblancq and Forester 2021). Outliers detected using a \u003cem\u003epcadapt\u003c/em\u003e approach also had some overlap with the RDA detected outliers. However, as expected the pcadapt approach did not pick up some of the more weakly correlated SNPs identified by the RDA. \u0026nbsp; This is likely to do with the higher power of the RDA to account for weak, multilocus selection (Capblancq et al. 2018; Capblancq and Forester 2021).\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eTherefore, while no single GEA procedure is without limitations, we balanced sensitivity and specificity to provide a robust analysis of genotype associations.\u003c/p\u003e\n\u003cp\u003eAdditionally, while the outlier SNPs are significantly associated with local factors the evidence to confirm they are truly adaptive is very limited. Current annotation of the reference genome has inhibited detailed functional genomic investigation and common garden conditions are required to experimentally test their adaptive effects (Gomes-dos-Santos et al. 2023). Because RADSeq approaches only assay genomic variation over a small part of the genome, environmentally associated genetic variation in other parts of the genome will not be identified (Arnold et al. 2013; Lowry et al. 2017).\u0026nbsp;While the predictive variables were selected to increase reproducibility of this analysis into other regions of freshwater pearl mussels, the lack of direct habitat measurements adds additional uncertainty to the GEA model and subsequent identification of adaptation.\u0026nbsp;Wider sequencing coverage across \u003cem\u003eM. margaritifera\u0026rsquo;s\u003c/em\u003e range, further genome annotation, more on-site measures of habitat parameters during sampling, and common garden experiments will increase validity of whether these SNPs are truly adaptive. However, for now these data robustly identify populations (and individuals within populations) that are most adapted along various environmental gradients that are likely to be impacted by climate change and provide a set of SNPs for identifying more such individuals for future conservation or perhaps climate-adapted translocations (Capblancq and Forester 2021). \u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eApplications\u003c/h2\u003e\n\u003cp\u003eDespite its limitations, recent work has shown that outlier SNPs identified in RDA are suitable to accurately predict levels of local adaptation (Lotterhos 2023). Local adaptation information can inform assisted gene flow and translocations, which are conservation techniques both proposed (Perea et al. 2022) and actively being used for pearl mussels respectively\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(Geist et al. 2023; Lavictoire and West 2024)\u003cstrong\u003e.\u003c/strong\u003e Although assisting gene flow can increase genomic diversity within the gene pool, the mixing of locally adapted populations can lead to outbreeding depression and a reduction in overall fitness\u0026nbsp;(Huff et al. 2011). Previous transplant investigations into survival of native/non-native juvenile mussels in Central Europe indicated success was largely dependent on local adaptation\u0026nbsp;(Denic et al. 2015). \u0026nbsp;Therefore, populations identified here as similarly locally adapted could be used as transplant stocks for each other. Of the 18 populations investigated, both OH1 and WI2 show no signs of recent recruitment\u0026nbsp;(Watt et al. 2015). Looking at the second, \u0026ldquo;adaptive landscape\u0026rdquo; RDA (Figure 6), the OH1 population appears adaptively distinct from the other investigated populations, indicating effort for conserving this population should focus on preserving the current mussels in their local environment. On the other hand, population WI2, while genetically isolated, shows less evidence of local adaptation in relation to the parameters in this model. Evidence presented here could support WI2 restocking from any population clustered in the centre of Figure 6b, with respect to host preference.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFinally, these results can be used to aid conservation of \u003cem\u003eM. margaritifera\u003c/em\u003e under climate change. All three climate variables are significant in the GEA model and the subsequent adaptive landscape. Though not the main drivers of selection, it is still important to consider these for climate adaptation and conservation. Traditional predictive ecological models, such as species distribution models, do not consider this intraspecific variation. Genomic offset techniques calculate the degree of maladaptation between current local current adaptation and future climate and offer an evolutionary dimension for forecasting species\u0026rsquo; response to change\u0026nbsp;(Fitzpatrick and Keller 2015; Waldvogel et al. 2020; Capblancq and Forester 2021; L\u0026aacute;ruson et al. 2022; Lind and Lotterhos 2024). These techniques have been used to inform targeted climate change management of the yellow warbler (\u003cem\u003eSetophaga petechia\u003c/em\u003e) and South American conifer \u003cem\u003eAraucaria araucana\u0026nbsp;\u003c/em\u003e(Bay et al. 2018; Varas-Myrik et al. 2024).\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eUsing the framework set out here, genomic offset of these critically endangered \u003cem\u003eM. margaritifera\u003c/em\u003e populations can aid conservation efforts of Scottish populations under climate change.\u0026nbsp;\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn conclusion, \u003cem\u003eM. margaritifera\u003c/em\u003e populations in Scotland display distinct and complex genomic structure patterns not explained by distance. When accounting for glochidial host, these patterns are significantly explained by local habitat parameters, including elevation range, altitude and bioclimatic variables, to which a subset of 302 particularly correlated (outlier) SNPs are identified. While further genomic sequencing efforts or experimental analysis is needed to validate these outlier SNPs, the information provided here informs translocation efforts and climate change mitigation for the conservation of this critically endangered species.\u0026nbsp;\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eData Availability\u0026nbsp;\u003c/h2\u003e\n\u003cp\u003eNext-RAD sequence data are available from the authors upon reasonable request and with permission from NatureScot and UHI Inverness. All scripts and other data are available at https://github.com/victoriagillman/scotpearl_adaptivelandscape_cleaned or downloadable from cited sources.\u003c/p\u003e\n\u003ch2\u003eCompeting Interests Statement/ Conflict of Interest\u003c/h2\u003e\n\u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e\n\u003ch2\u003eEthical Approval\u003c/h2\u003e\n\u003cp\u003ePearl mussel samples were collected under applicable NatureScot license and with landowner permissions. Work with non-cephalopod invertebrates is not regulated under the UK Animals (Scientific Procedures) Act 1986 or amendments.\u003c/p\u003e\n\u003ch2\u003eAcknowledgments\u003c/h2\u003e\n\u003cp\u003eThis research was funded by NatureScot and UHI Inverness. We thank Mark Coulson, Silvia Ferreira Carvalho and Jenny O\u0026rsquo;Dell for contributions to project development, Iain Sime (NatureScot) and Chris Daphne for sample collection, and Dasha Svobodova and Lydia McGill (UHI Inverness) for laboratory assistance. Paul Etter and Eric Johnson at SNPsaurus LLC facilitated the generation and quality control of RAD-seq data. VG is supported the NERC Scottish Universities Partnership for Environmental Research (SUPER) Doctoral Training Partnership (Grant reference number NE/S007342/1 to KL and website https://superdtp.st-andrews.ac.uk/) and the University of Aberdeen.\u0026nbsp;\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eK.L. and V.P. secured the funding for this work, and K.L., V.P. and V.G. conceived of the study. V.P. led data collection and generation, with support from B.M., P.C. and V.G. V.G. led data analysis and manuscript writing. K.L. served as the primary PhD supervisor for this work, V.P. served as the secondary PhD supervisor and L.L. served as an additional PhD supervisor. All supervisors provided guidance on data analysis and manuscript writing.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAlbaladejo-Robles G, B\u0026ouml;hm M, Newbold T (2023) Species life-history strategies affect population responses to temperature and land-cover changes. Global Change Biology 29:97\u0026ndash;109. https://doi.org/10.1111/gcb.16454\u003c/li\u003e\n \u003cli\u003eArnold B, Corbett-Detig RB, Hartl D, Bomblies K (2013) RADseq underestimates diversity and introduces genealogical biases due to nonrandom haplotype sampling. Molecular Ecology 22:3179\u0026ndash;3190. https://doi.org/10.1111/MEC.12276\u003c/li\u003e\n \u003cli\u003eBauer G (1987) Reproductive Strategy of the Freshwater Pearl Mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e. Journal of Animal Ecology 56:691\u0026ndash;704. https://doi.org/10.2307/5077\u003c/li\u003e\n \u003cli\u003eBauer G (1992) Variation in the Life Span and Size of the Freshwater Pearl Mussel. The Journal of Animal Ecology 61:425. https://doi.org/10.2307/5333\u003c/li\u003e\n \u003cli\u003eBaum D (2015) Glochidia surveys of three rivers in Lochaber. Scottish Natural Heritage. Commissioned Report\u003c/li\u003e\n \u003cli\u003eBaum D (2018) Host salmonid specificity of selected pearl mussel populations. Scottish Natural Heritage. Research Report\u003c/li\u003e\n \u003cli\u003eBay RA, Harrigan RJ, Underwood VL, et al (2018) Genomic signals of selection predict climate-driven population declines in a migratory bird. Science 359:83\u0026ndash;86. https://doi.org/10.1126/science.aan4380\u003c/li\u003e\n \u003cli\u003eBiological Records Centre (2019) Watsonian Vice County Boundary layers for Great Britain\u003c/li\u003e\n \u003cli\u003eBosch S, Fernandez S (2023) sdmpredictors: Species Distribution Modelling Predictor Datasets\u003c/li\u003e\n \u003cli\u003eCamacho C, Coulouris G, Avagyan V, et al (2009) BLAST+: architecture and applications. BMC Bioinformatics 10:421. https://doi.org/10.1186/1471-2105-10-421\u003c/li\u003e\n \u003cli\u003eCapblancq T, Fitzpatrick MC, Bay RA, et al (2020) Genomic Prediction of (Mal)Adaptation Across Current and Future Climatic Landscapes. Annual Review of Ecology, Evolution, and Systematics 51:245\u0026ndash;269. https://doi.org/10.1146/annurev-ecolsys-020720-042553\u003c/li\u003e\n \u003cli\u003eCapblancq T, Forester BR (2021) Redundancy analysis: A Swiss Army Knife for landscape genomics. Methods in Ecology and Evolution 12:2298\u0026ndash;2309. https://doi.org/10.1111/2041-210X.13722\u003c/li\u003e\n \u003cli\u003eCapblancq T, Luu K, Blum MGB, Bazin E (2018) Evaluation of redundancy analysis to identify signatures of local adaptation. Molecular Ecology Resources 18:1223\u0026ndash;1233. https://doi.org/10.1111/1755-0998.12906\u003c/li\u003e\n \u003cli\u003eCarver S, Comber L, McMorran R, et al (2008) Wildness Study in the Cairngorms National Park Final Report\u003c/li\u003e\n \u003cli\u003eCauwelier, E, Verspoor E, Tarr EC, et al (2009) Genetic diversity and differentiation of freshwater pearl mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e) populations in the UK. Scottish Natural Heritage Commissioned Report No.XXX (ROAME No. F05AC701).\u003c/li\u003e\n \u003cli\u003eCefas (2023) Catchment boundaries for Scotland based on Scottish Environment Protection Agency (SEPA) catchments (WGS84) 2023\u003c/li\u003e\n \u003cli\u003eChang CC, Chow CC, Tellier LC, et al (2015) Second-generation PLINK: rising to the challenge of larger and richer datasets. GigaScience 4:s13742-015-0047\u0026ndash;8. https://doi.org/10.1186/s13742-015-0047-8\u003c/li\u003e\n \u003cli\u003eChatziargyriou V, Dailianis S (2010) The role of selenium-dependent glutathione peroxidase (Se-GPx) against oxidative and genotoxic effects of mercury in haemocytes of mussel \u003cem\u003eMytilus galloprovincialis\u003c/em\u003e (Lmk.). Toxicology in Vitro 24:1363\u0026ndash;1372. https://doi.org/10.1016/j.tiv.2010.04.008\u003c/li\u003e\n \u003cli\u003eClements EA, Thomas R, Adams CE (2018) An investigation of salmonid host utilization by the endangered freshwater pearl mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e) in north‐west Scotland. Aquatic Conservation 28:764\u0026ndash;768. https://doi.org/10.1002/aqc.2900\u003c/li\u003e\n \u003cli\u003eCordero-Rivera A, Ondina P, Outeiro A, et al (2022) Allometry in the Freshwater Pearl Mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e L.): Mussels Tend to Grow Flatter at Higher Water Speed. mala 64:257\u0026ndash;267. https://doi.org/10.4002/040.064.0208\u003c/li\u003e\n \u003cli\u003eCosgrove P, Hastie L, Watt J, et al (2012) Scotland\u0026rsquo;s Freshwater Pearl Mussels: The Challenge of Climate Change. River Conservation and Management 121\u0026ndash;132. https://doi.org/10.1002/9781119961819.CH10\u003c/li\u003e\n \u003cli\u003eCosgrove P, Watt J, Hastie L, et al (2016) The status of the freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e in Scotland: extent of change since 1990s, threats and management implications. Biodivers Conserv 25:2093\u0026ndash;2112. https://doi.org/10.1007/s10531-016-1180-0\u003c/li\u003e\n \u003cli\u003eCosgrove PJ, Young MR, Hastie LC, et al (2000) The status of the freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e Linn. in Scotland. Aquatic Conservation: Marine and Freshwater Ecosystems 10:197\u0026ndash;208. https://doi.org/10.1002/1099-0755(200005/06)10:3\u0026lt;197::AID-AQC405\u0026gt;3.0.CO;2-S\u003c/li\u003e\n \u003cli\u003eCurley EAM, Thomas R, Adams CE, Stephen A (2021) Adaptive responses of freshwater pearl mussels, \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e , to managed drawdowns. Aquatic Conservation: Marine and Freshwater Ecosystems. https://doi.org/10.1002/AQC.3759\u003c/li\u003e\n \u003cli\u003eDanecek P, Auton A, Abecasis G, et al (2011) The variant call format and VCFtools. Bioinformatics 27:2156\u0026ndash;2158. https://doi.org/10.1093/bioinformatics/btr330\u003c/li\u003e\n \u003cli\u003eDanecek P, Bonfield JK, Liddle J, et al (2021) Twelve years of SAMtools and BCFtools. GigaScience 10:. https://doi.org/10.1093/gigascience/giab008\u003c/li\u003e\n \u003cli\u003eDegerman E, Alexanderson S, Bergengren J, et al (2009) Restoration of freshwater pearl mussel streams. WWF Sweden, Solna\u003c/li\u003e\n \u003cli\u003eDenic M, Taeubert J-E, Lange M, et al (2015) Influence of stock origin and environmental conditions on the survival and growth of juvenile freshwater pearl mussels (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e) in a cross-exposure experiment. Limnologica 50:67\u0026ndash;74. https://doi.org/10.1016/j.limno.2014.07.005\u003c/li\u003e\n \u003cli\u003eDobrowski SZ (2011) A climatic basis for microrefugia: the influence of terrain on climate. Global Change Biology 17:1022\u0026ndash;1035. https://doi.org/10.1111/j.1365-2486.2010.02263.x\u003c/li\u003e\n \u003cli\u003eDomisch S, Amatulli G, Jetz W (2015) Near-global freshwater-specific environmental variables for biodiversity analyses in 1 km resolution. Sci Data 2:150073. https://doi.org/10.1038/sdata.2015.73\u003c/li\u003e\n \u003cli\u003eEagar R (1977) Shape of Shell in Relation to Weight of \u003cem\u003eMargaritifera margaritifera\u0026nbsp;\u003c/em\u003e(l) (bivalvia-Margaritiferidae). J Conchol 29:207\u0026ndash;218\u003c/li\u003e\n \u003cli\u003eEtten J van (2017) R Package gdistance: Distances and Routes on Geographical Grids. Journal of Statistical Software 76:1\u0026ndash;21. https://doi.org/10.18637/jss.v076.i13\u003c/li\u003e\n \u003cli\u003eFarrington SJ, King RW, Baker JA, et al (2020) Population genetics of freshwater pearl mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e) in central Massachusetts and implications for conservation. Aquatic Conservation: Marine and Freshwater Ecosystems 30:1945\u0026ndash;1958. https://doi.org/10.1002/AQC.3439\u003c/li\u003e\n \u003cli\u003eFaske TM, Agneray AC, Jahner JP, et al (2021) Genomic and common garden approaches yield complementary results for quantifying environmental drivers of local adaptation in rubber rabbitbrush, a foundational Great Basin shrub. Evol Appl 14:2881\u0026ndash;2900. https://doi.org/10.1111/eva.13323\u003c/li\u003e\n \u003cli\u003eFitzpatrick MC, Keller SR (2015) Ecological genomics meets community-level modelling of biodiversity: mapping the genomic landscape of current and future environmental adaptation. Ecology Letters 18:1\u0026ndash;16. https://doi.org/10.1111/ele.12376\u003c/li\u003e\n \u003cli\u003eForester BR, Lasky JR, Wagner HH, Urban DL (2018) Comparing methods for detecting multilocus adaptation with multivariate genotype\u0026ndash;environment associations. Molecular Ecology 27:2215\u0026ndash;2233. https://doi.org/10.1111/mec.14584\u003c/li\u003e\n \u003cli\u003eFrichot E, Fran\u0026ccedil;ois O (2015) LEA: An R package for landscape and ecological association studies. Methods in Ecology and Evolution 6:925\u0026ndash;929. https://doi.org/10.1111/2041-210X.12382\u003c/li\u003e\n \u003cli\u003eGeist J (2011) Integrative freshwater ecology and biodiversity conservation. Ecol Indic 11:1507\u0026ndash;1516. https://doi.org/10.1016/j.ecolind.2011.04.002\u003c/li\u003e\n \u003cli\u003eGeist J (2010) Strategies for the conservation of endangered freshwater pearl mussels (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e L.): A synthesis of conservation genetics and ecology. Hydrobiologia 644:69\u0026ndash;88. https://doi.org/10.1007/S10750-010-0190-2\u003c/li\u003e\n \u003cli\u003eGeist J, Kuehn R (2005) Genetic diversity and differentiation of central European freshwater pearl mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e L.) populations: implications for conservation and management. Molecular Ecology 14:425\u0026ndash;439. https://doi.org/10.1111/J.1365-294X.2004.02420.X\u003c/li\u003e\n \u003cli\u003eGeist J, Moorkens E, Killeen I, et al (2018) Genetic structure of Irish freshwater pearl mussels (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e and \u003cem\u003eMargaritifera durrovensis\u003c/em\u003e): Validity of subspecies, roles of host fish, and conservation implications. Aquatic Conservation: Marine and Freshwater Ecosystems 28:923\u0026ndash;933. https://doi.org/10.1002/AQC.2913\u003c/li\u003e\n \u003cli\u003eGeist J, S\u0026ouml;derberg H, Karlberg A, Kuehn R (2010) Drainage-independent genetic structure and high genetic diversity of endangered freshwater pearl mussels (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e) in northern Europe. Conservation Genetics 11:1339\u0026ndash;1350. https://doi.org/10.1007/S10592-009-9963-4/TABLES/5\u003c/li\u003e\n \u003cli\u003eGeist J, Thielen F, Lavictoire L, et al (2023) Captive breeding of European freshwater mussels as a conservation tool: A review. Aquatic Conservation: Marine and Freshwater Ecosystems 33:1321\u0026ndash;1359. https://doi.org/10.1002/aqc.4018\u003c/li\u003e\n \u003cli\u003eGomes-dos-Santos A, Froufe E, Amaro R, et al (2019) The male and female complete mitochondrial genomes of the threatened freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e (Linnaeus, 1758) (Bivalvia: Margaritiferidae). Mitochondrial DNA Part B 4:1417\u0026ndash;1420. https://doi.org/10.1080/23802359.2019.1598794\u003c/li\u003e\n \u003cli\u003eGomes-dos-Santos A, Lopes-Lima M, Machado AM, et al (2023) The Crown Pearl V2: an improved genome assembly of the European freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e (Linnaeus, 1758). Gigabyte 2023:1\u0026ndash;14. https://doi.org/10.46471/gigabyte.81\u003c/li\u003e\n \u003cli\u003eGosselin T, Lamoth M, Devloo-Delva F, Grewe P (2020) radiator: RADseq Data Exploration, Manipulation and Visualization using R. R package version 1.1.9\u003c/li\u003e\n \u003cli\u003eHarrison S, Cowhig K, Sullivan T (2024) Still waters and deep: Habitat characteristics, density and distribution of a lake-dwelling population of the freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e. Global Ecology and Conservation 54:e03130. https://doi.org/10.1016/j.gecco.2024.e03130\u003c/li\u003e\n \u003cli\u003eHastie L, Cooksley S, Scougall F, et al (2003a) Characterization of freshwater pearl mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e) riverine habitat using River Habitat Survey data. Aquatic Conservation: Marine and Freshwater Ecosystems 13:213\u0026ndash;224. https://doi.org/10.1002/aqc.560\u003c/li\u003e\n \u003cli\u003eHastie LC, Cosgrove PJ, Ellis N, Gaywood MJ (2003b) The threat of climate change to freshwater pearl mussel populations. ambi 32:40\u0026ndash;46. https://doi.org/10.1579/0044-7447-32.1.40\u003c/li\u003e\n \u003cli\u003eHe F, Zarfl C, Bremerich V, et al (2019) The global decline of freshwater megafauna. Global Change Biology 25:3883\u0026ndash;3892. https://doi.org/10.1111/gcb.14753\u003c/li\u003e\n \u003cli\u003eHijmans RJ (2023) raster: Geographic Data Analysis and Modeling\u003c/li\u003e\n \u003cli\u003eHijmans RJ, Cameron SE, Parra JL, et al (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25:1965\u0026ndash;1978. https://doi.org/10.1002/joc.1276\u003c/li\u003e\n \u003cli\u003eHuff DD, Miller LM, Chizinski CJ, Vondracek B (2011) Mixed-source reintroductions lead to outbreeding depression in second-generation descendents of a native North American fish. Molecular Ecology 20:4246\u0026ndash;4258. https://doi.org/10.1111/j.1365-294X.2011.05271.x\u003c/li\u003e\n \u003cli\u003eKarlsson S, Larsen BM, Hindar K (2014) Host-dependent genetic variation in freshwater pearl mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e L.). Hydrobiologia 735:179\u0026ndash;190. https://doi.org/10.1007/S10750-013-1679-2/FIGURES/3\u003c/li\u003e\n \u003cli\u003eL\u0026aacute;ruson \u0026Aacute;J, Fitzpatrick MC, Keller SR, et al (2022) Seeing the forest for the trees: Assessing genetic offset predictions from gradient forest. Evolutionary Applications. https://doi.org/10.1111/EVA.13354\u003c/li\u003e\n \u003cli\u003eLavictoire L, West C (2024) Population Reinforcement of the Endangered Freshwater Pearl Mussel (Margaritifera margaritifera): Lessons Learned. Diversity 16:187. https://doi.org/10.3390/d16030187\u003c/li\u003e\n \u003cli\u003eLehner B, Verdin K, Jarvis A (2008) New Global Hydrography Derived From Spaceborne Elevation Data. Eos, Transactions American Geophysical Union 89:93\u0026ndash;94. https://doi.org/10.1029/2008EO100001\u003c/li\u003e\n \u003cli\u003eLi H (2013) Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. Preprint: arXiv. https://doi.org/10.48550/arXiv.1303.3997\u003c/li\u003e\n \u003cli\u003eLind BM, Lotterhos KE (2024) The accuracy of predicting maladaptation to new environments with genomic data. Molecular Ecology Resources e14008. https://doi.org/10.1111/1755-0998.14008\u003c/li\u003e\n \u003cli\u003eLotterhos KE (2024) Interpretation issues with \u0026ldquo;genomic vulnerability\u0026rdquo; arise from conceptual issues in local adaptation and maladaptation. Evolution Letters 8:331\u0026ndash;339. https://doi.org/10.1093/evlett/qrae004\u003c/li\u003e\n \u003cli\u003eLotterhos KE (2023) The paradox of adaptive trait clines with nonclinal patterns in the underlying genes. Proceedings of the National Academy of Sciences 120:e2220313120. https://doi.org/10.1073/pnas.2220313120\u003c/li\u003e\n \u003cli\u003eLowry DB, Hoban S, Kelley JL, et al (2017) Breaking RAD: an evaluation of the utility of restriction site-associated DNA sequencing for genome scans of adaptation. Molecular Ecology Resources 17:142\u0026ndash;152. https://doi.org/10.1111/1755-0998.12635\u003c/li\u003e\n \u003cli\u003eMarandel F, Charrier G, Lamy J, et al (2020) Estimating effective population size using RADseq: Effects of SNP selection and sample size. Ecol Evol 10:1929\u0026ndash;1937. https://doi.org/10.1002/ece3.6016\u003c/li\u003e\n \u003cli\u003eMijangos JL, Berry OF, Pacioni C, Georges A (2022) dartR v2: An accessible genetic analysis platform for conservation, ecology and agriculture. Methods in Ecology and Evolution. https://doi.org/10.1111/2041-210X.13918\u003c/li\u003e\n \u003cli\u003eMoorkens E (2010) IUCN Red List of Threatened Species: \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e. IUCN Red List of Threatened Species\u003c/li\u003e\n \u003cli\u003eNatureScot (2014) Scotland\u0026rsquo;s wildness - remoteness. In: SpatialData.gov.scot. https://spatialdata.gov.scot/geonetwork/srv/api/records/7220922e-5a57-450d-b8ff-6ee990d38b4f. Accessed 27 Nov 2024\u003c/li\u003e\n \u003cli\u003eOksanen J, Simpson GL, Blanchet FG, et al (2024) vegan: Community Ecology Package\u003c/li\u003e\n \u003cli\u003eOrdnance Survey (2023) Open Rivers\u003c/li\u003e\n \u003cli\u003eParadis E, Schliep K (2019) ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35:526\u0026ndash;528. https://doi.org/10.1093/bioinformatics/bty633\u003c/li\u003e\n \u003cli\u003ePeck LS (2011) Organisms and responses to environmental change. Marine Genomics 4:237\u0026ndash;243. https://doi.org/10.1016/j.margen.2011.07.001\u003c/li\u003e\n \u003cli\u003ePembleton LW, Cogan NOI, Forster JW (2013) StAMPP: an R package for calculation of genetic differentiation and structure of mixed-ploidy level populations. Molecular Ecology Resources 13:946\u0026ndash;952. https://doi.org/10.1111/1755-0998.12129\u003c/li\u003e\n \u003cli\u003ePerea S, Mendes SL, Sousa-Santos C, et al (2022) Applying genomic approaches to delineate conservation strategies using the freshwater mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e in the Iberian Peninsula as a model. Sci Rep 12:16894. https://doi.org/10.1038/s41598-022-20947-5\u003c/li\u003e\n \u003cli\u003ePritchard VL, Cosgrove P, Gillman V, et al (2025) Genetic correlates of host use in Scotland\u0026rsquo;s pearl mussels. BioRXiv\u003c/li\u003e\n \u003cli\u003ePriv\u0026eacute; F, Luu K, Vilhj\u0026aacute;lmsson BJ, Blum MGB (2020) Performing highly efficient genome scans for local adaptation with R package pcadapt version 4. Molecular Biology and Evolution 37:2153\u0026ndash;2154. https://doi.org/10.1093/molbev/msaa053\u003c/li\u003e\n \u003cli\u003eQueitsch C, Sangster TA, Lindquist S (2002) Hsp90 as a capacitor of phenotypic variation. Nature 417:618\u0026ndash;624. https://doi.org/10.1038/nature749\u003c/li\u003e\n \u003cli\u003eRodriguez-Sanchez F (2024) rSDM: Species distribution and niche modelling in R\u003c/li\u003e\n \u003cli\u003eRutherford SL, Lindquist S (1998) Hsp90 as a capacitor for morphological evolution. Nature 396:336\u0026ndash;342. https://doi.org/10.1038/24550\u003c/li\u003e\n \u003cli\u003eSaitou N, Nei M (1987) The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4:406\u0026ndash;425. https://doi.org/10.1093/oxfordjournals.molbev.a040454\u003c/li\u003e\n \u003cli\u003eSantos MA, Antunes MA, Grandela A, et al (2023) Slow and population specific evolutionary response to a warming environment. Sci Rep 13:9700. https://doi.org/10.1038/s41598-023-36273-3\u003c/li\u003e\n \u003cli\u003eSmith DG (1980) Anatomical studies on \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e and \u003cem\u003eCumberlandia monodonta\u003c/em\u003e (Mollusca: Pelecypoda: Margaritiferidae). Zoological Journal of the Linnean Society 69:257\u0026ndash;270. https://doi.org/10.1111/j.1096-3642.1980.tb01125.x\u003c/li\u003e\n \u003cli\u003eSteane DA, Potts BM, McLean E, et al (2014) Genome-wide scans detect adaptation to aridity in a widespread forest tree species. Molecular Ecology 23:2500\u0026ndash;2513. https://doi.org/10.1111/mec.12751\u003c/li\u003e\n \u003cli\u003eStoeckle BC, Araujo R, Geist J, et al (2017) Strong genetic differentiation and low genetic diversity of the freshwater pearl mussel (\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e L.) in the southwestern European distribution range. Conservation Genetics 18:147\u0026ndash;157. https://doi.org/10.1007/S10592-016-0889-3/FIGURES/2\u003c/li\u003e\n \u003cli\u003eStorey JD, Bass AJ, Dabney A, Robinson D (2017) qvalue: Q-value estimation for false discovery rate control\u003c/li\u003e\n \u003cli\u003eStrack E (2006) Pearls. R\u0026uuml;hle-Diebener-Verlag\u003c/li\u003e\n \u003cli\u003eTyers M (2024) riverdist: River Network Distance Computation and Applications\u003c/li\u003e\n \u003cli\u003eVaras-Myrik A, Sep\u0026uacute;lveda-Espinoza F, Toro-N\u0026uacute;\u0026ntilde;ez \u0026Oacute;, et al (2024) Using a genomic offset approach to guide assisted gene flow in the South American conifer \u003cem\u003eAraucaria araucana\u003c/em\u003e. Forest Ecology and Management 553:121637. https://doi.org/10.1016/j.foreco.2023.121637\u003c/li\u003e\n \u003cli\u003eWacker S, Larsen BM, Karlsson S, Hindar K (2019) Host specificity drives genetic structure in a freshwater mussel. Sci Rep 9:10409. https://doi.org/10.1038/s41598-019-46802-8\u003c/li\u003e\n \u003cli\u003eWaldvogel A, Feldmeyer B, Rolshausen G, et al (2020) Evolutionary genomics can improve prediction of species\u0026rsquo; responses to climate change. Evol Lett 4:4\u0026ndash;18. https://doi.org/10.1002/evl3.154\u003c/li\u003e\n \u003cli\u003eWatson HC (1852) Cybele Britannica : or British Plants and their geographical relations. Longman and Co., Paternoster Row, London\u003c/li\u003e\n \u003cli\u003eWatt J, Cosgrove PJ, Hastie LC (2015) A national freshwater pearl mussel \u0026nbsp;(\u003cem\u003eMargaritifera margaritifera\u003c/em\u003e, L.) survey of Scotland. Scottish Natural Heritage Commissioned \u0026nbsp;Report No. 901\u003c/li\u003e\n \u003cli\u003eWei T, Simko V (2021) R package \u0026ldquo;corrplot\u0026rdquo;: Visualization of a Correlation Matrix\u003c/li\u003e\n \u003cli\u003eWeir BS, Cockerham CC (1984) Estimating F-Statistics for the Analysis of Population Structure. Evolution 38:1358\u0026ndash;1370. https://doi.org/10.2307/2408641\u003c/li\u003e\n \u003cli\u003eWigginton JE, Cutler DJ, Abecasis GR (2005) A Note on Exact Tests of Hardy-Weinberg Equilibrium. American Journal of Human Genetics 76:887. https://doi.org/10.1086/429864\u003c/li\u003e\n \u003cli\u003eYoung M (1984) The reproductive biology of the freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e (Linn.) in Scotland. I, Field studies. Archiv fur Hydrobiologie 99:405\u0026ndash;422\u003c/li\u003e\n \u003cli\u003eYoung MR, Cosgrove PJ, Hastie LC (2001) The Extent of, and Causes for, the Decline of a Highly Threatened Naiad: 337\u0026ndash;357. https://doi.org/10.1007/978-3-642-56869-5_19\u003c/li\u003e\n \u003cli\u003eZanatta DT, Stoeckle BC, Inoue K, et al (2018) High genetic diversity and low differentiation in North American \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e(Bivalvia: Unionida: Margaritiferidae). Biological Journal of the Linnean Society 123:850\u0026ndash;863. https://doi.org/10.1093/biolinnean/bly010\u003c/li\u003e\n \u003cli\u003eZeng C, Xing R, Huang B, et al (2023) Phytoplankton in headwater streams: spatiotemporal patterns and underlying mechanisms. Front Plant Sci 14:1276289. https://doi.org/10.3389/fpls.2023.1276289\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1. Population names of \u003cem\u003eMargaritifera margaritifera\u0026nbsp;\u003c/em\u003eanonymised by vice counties as follows: East Inverness-shire (EI); East Ross (ER); East Sutherland (ES); North Ebudes (NE); Outer Hebrides (OH); West Inverness-shire (WI); West Ross (WR); West Sutherland (WS). Catchment area and sample size (\u003cem\u003eN\u003c/em\u003e) of individuals in each population along with inbreeding (\u003cem\u003eF\u003csub\u003eIS\u003c/sub\u003e)\u003c/em\u003e, observed heterozygosity and standard deviation (\u003cem\u003eH\u003csub\u003eO,\u003c/sub\u003e H\u003csub\u003eO\u0026nbsp;\u003c/sub\u003eSD)\u003c/em\u003e, adjusted expected heterozygosity (\u003cem\u003eH\u003csub\u003eE\u003c/sub\u003e, H\u003csub\u003eE\u003c/sub\u003eSD\u003c/em\u003e), nucleotide diversity (\u003cem\u003e\u0026Pi;\u003c/em\u003e), and number of private alleles per population. Finally, host preference for each river from Pritchard et al. (2025).\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"652\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePopulation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCatchment Area (km)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 28px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eN\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eF\u003csub\u003eIS\u003c/sub\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eH\u003csub\u003eO\u003c/sub\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eH\u003csub\u003eO\u003c/sub\u003e SD\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eH\u003csub\u003eE\u003c/sub\u003e\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eH\u003csub\u003eE\u003c/sub\u003eSD\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e\u0026Pi;\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 51px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrivate Alleles\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eHost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eEI1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1852\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e-0.009\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.252\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.239\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.186\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eSalmon\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eER1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1177\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.262\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.119\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.147\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.073\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1967\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eES1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e583\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.178\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.203\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1467\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eNE1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1622\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.046\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.248\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.324\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.194\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.097\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1834\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eOH1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e303\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e-0.047\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.186\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.035\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.117\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e3161\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWI1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e983\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.136\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.143\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.181\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1698\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eOnly trout available\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWI2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e983\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.139\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.178\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.165\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.083\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1675\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eOnly trout available\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWI3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e983\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.154\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.216\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.178\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.244\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.122\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e576\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eSalmon\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWI4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e983\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.165\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.192\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.158\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.191\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.079\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1825\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eOnly trout available\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWR1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e622\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.191\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.212\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.248\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.124\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e649\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eSalmon\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWR2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.152\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.161\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.177\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.179\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.173\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.09\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1223\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eTrout\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWR3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.247\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.156\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.152\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.199\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e839\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eTrout\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWR4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.167\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.095\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.162\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.108\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.169\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.054\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e2229\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eTrout\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWS1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.044\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.078\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.173\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.076\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.154\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.038\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e2683\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eOnly trout available\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWS2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e1477\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.049\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.175\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.209\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.087\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e1529\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eOnly trout available\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWS3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e141\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.177\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.115\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.185\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.131\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.183\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.066\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e2099\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eOnly trout available\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWS4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.092\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.104\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.183\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.107\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.168\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.054\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e2271\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eTrout\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003eWS5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 76px;\"\u003e\n \u003cp\u003e118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 28px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.084\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.249\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.226\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.247\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e0.124\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 51px;\"\u003e\n \u003cp\u003e912\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 113px;\"\u003e\n \u003cp\u003eUnknown\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTable 2. Summary of top absolute correlations between putatively adaptive SNPs and environmental predictors for \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e in Scotland. SNPs were identified using q\u0026lt;0.05 from redundancy analysis (RDA) loadings. RDA equation as follows: rda(SNPs ~ elevation range + catchment area + wildness + altitude +BIOCLIM3 (Isothermality, calculated as BIOCLIM2/BIOCLIM7 \u0026times; 100) + BIOCLIM5 (maximum temperature of the warmest month) + BIOCLIM17 (precipitation of the driest quarter) +Condition(Host). Each SNP was assigned to the environmental predictor with which it had the highest absolute correlation. For each predictor, the table reports: (1) the number of outlier SNPs most strongly associated with it (Count), and (2) the mean absolute correlation across those SNPs (Average Correlation).\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"697\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003eElevation range\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eBIOCLIM3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eBIOCLIM5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003eBIOCLIM17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003eAltitude\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003eCatchment area\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003eWildness\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eCount\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003e43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 95px;\"\u003e\n \u003cp\u003eAverage correlation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 77px;\"\u003e\n \u003cp\u003e0.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 94px;\"\u003e\n \u003cp\u003e0.25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 98px;\"\u003e\n \u003cp\u003e0.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 72px;\"\u003e\n \u003cp\u003e0.29\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"conservation-genetics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"coge","sideBox":"Learn more about [Conservation Genetics](https://www.springer.com/journal/10592)","snPcode":"10592","submissionUrl":"https://submission.nature.com/new-submission/10592/3","title":"Conservation Genetics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7363533/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7363533/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eClimate change is a global threat that is already impacting populations and species across diverse ecosystems. Local adaptation of populations, identified through genotype-environment association, can be used to guide conservation management. The critically endangered freshwater pearl mussel \u003cem\u003eMargaritifera margaritifera\u003c/em\u003e has a widespread Holarctic distribution across a range of habitats. Here, we used 156 samples from 18 populations of \u003cem\u003eM. margaritifera\u003c/em\u003e across Scotland to characterize neutral population structure and identify signals of environment-associated adaptation in this endangered mussel. This study revealed a complex pattern of population structure, indicated by high genetic diversity, high genetic clustering and high inbreeding, although neutral structure was not shaped by distances or environments. Controlling for host species, freshwater pearl mussels show significant patterns of local adaptation to abiotic habitat variables, including in response to elevational range of their sub-catchment, their specific altitude, and bioclimatic factors such as isothermality. Locally adapted and generalist populations were identified which could guide genetic rescue and restocking conservation efforts, while a subset of 302 SNPs putatively under environmental selection were identified for further adaptation investigation.\u003c/p\u003e","manuscriptTitle":"Genomic evidence of local adaptation in Scottish Margaritifera margaritifera","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-04 13:33:03","doi":"10.21203/rs.3.rs-7363533/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-27T18:46:39+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-17T13:08:59+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"219096402016501440071415369215010593366","date":"2025-09-19T07:29:18+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-09-08T17:19:53+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"21195201483954237190699047099280576306","date":"2025-09-01T18:47:35+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-08-27T16:59:10+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-19T12:08:21+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-19T12:06:48+00:00","index":"","fulltext":""},{"type":"submitted","content":"Conservation Genetics","date":"2025-08-13T09:43:02+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"conservation-genetics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"coge","sideBox":"Learn more about [Conservation Genetics](https://www.springer.com/journal/10592)","snPcode":"10592","submissionUrl":"https://submission.nature.com/new-submission/10592/3","title":"Conservation Genetics","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"d38195b6-3fe0-495b-b79e-39e2f704514a","owner":[],"postedDate":"September 4th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-08T16:00:25+00:00","versionOfRecord":{"articleIdentity":"rs-7363533","link":"https://doi.org/10.1007/s10592-025-01739-6","journal":{"identity":"conservation-genetics","isVorOnly":false,"title":"Conservation Genetics"},"publishedOn":"2025-12-04 15:57:21","publishedOnDateReadable":"December 4th, 2025"},"versionCreatedAt":"2025-09-04 13:33:03","video":"","vorDoi":"10.1007/s10592-025-01739-6","vorDoiUrl":"https://doi.org/10.1007/s10592-025-01739-6","workflowStages":[]},"version":"v1","identity":"rs-7363533","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7363533","identity":"rs-7363533","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}