Abstract

The Pyrenean desman (Galemys pyrenaicus) is a small, semiaquatic mammal endemic to the Pyrenean Mountains and the northern half of the Iberian Peninsula where it lives in cold and well-oxygenated flowing mountain streams. This species is currently classified as vulnerable on the IUCN Red List and has been undergoing habitat loss and fragmentation for decades, inevitably impacting its distribution. A recent genetic study, based on mitochondrial and intronic sequences, showed that the genetic variability of the Pyrenean desman is very low in the Pyrenees. In this study, we investigated the potential existence of genetic structure and gene flow at a smaller scale using 24 polymorphic microsatellite loci. As the Pyrenean desman is a very elusive species, we supplemented our tissue sample collection with samples of feces collected in the French range of this species. We successfully identified 70 individuals based on 355 fecal samples. Bayesian analyses revealed 3 genetic and geographic clusters (1 eastern, 1 central, and 1 western, including 3 genetic subclusters), with origins tracing back only 200 years. These clusters were characterized by low levels of genetic diversity and high inbreeding coefficients. Although gene flow among clusters appeared to be limited, populations seem to have exchanged alleles recently. Therefore, connectivity between watersheds should be enhanced to maintain genetic diversity and potentially improve the long-term survival of the Pyrenean desman in France.

Conservation of endangered species is dependent on knowledge of the genetic structure and diversity of individual populations (Frankham 2003). This diversity is often spatially structured because natural habitat is not always continuous or can vary across the species range. Anthropogenic activities, resulting in fragmentation and habitat loss, also play a role in shaping the structure of wildlife populations. At the individual level, fragmentation can alter spatial and dispersal movement patterns or disrupt the social structure and increase inbreeding (Gerlach and Musolf 2000; Coster and Kovach 2012; Mona et al. 2014). At the population level, fragmentation can reduce suitable habitat, restricting gene flow and augmenting genetic drift (Frankham 2005). Knowledge of species’ population dynamics along with their genetics and basic ecology is essential to designing and implementing conservation plans and appropriate management measures, especially in the case of habitat specialists with small ranges that may be particularly sensitive to environmental change.

The Pyrenean desman (Galemys pyrenaicus, Soricomorpha, Talpidae) is a small, semiaquatic mammal endemic to the Pyrenean Mountains and to the northern half of the Iberian Peninsula (Queiroz 1999). This species lives in montane rivers with cold and well-oxygenated flowing waters and is well adapted to aquatic life. The Pyrenean desman is characterized by large webbed hindfeet, double-layered fur, a long tail, and a mobile prehensile snout, which make it a specialist in finding and feeding on larvae of benthic macroinvertebrates (Palmeirim and Hoffmann 1983; Richard 1986). The Pyrenean desman is an endangered species. It is currently classified as vulnerable on the IUCN Red List (Fernandes et al. 2008) and is legally protected in the 4 countries encompassing its range (Andorra, France, Portugal, and Spain). It has been undergoing habitat loss and fragmentation for decades, inevitably impacting its distribution (Nores et al. 2007; Némoz and Bertrand 2008).

The elusive behavior and nocturnal activity of this species make it hard to study. Some information is known about its ecology and biology (Stone 1987a; Bertrand 1994; Melero et al. 2012, 2014), but no data on the genetic structure of the Pyrenean desman were available until a recent study based on mitochondrial and nuclear markers (Igea et al. 2013). This study revealed that the Pyrenean desman was characterized by very low genetic diversity compared to other mammals. Its evolutionary history seems to have been highly influenced by Pleistocene glaciations, leading to a phylogeographic structure encompassing 4 mitochondrial lineages with parapatric distributions. More specifically, Igea et al. (2013) obtained evidence that the desman’s Pyrenean populations are genetically homogeneous and that they likely originated from a distant refuge, probably located in the Basque Mountains, after a severe bottleneck.

However, these hypotheses were based on a small number of specimens as well as on the use of genetic markers with relatively low rates of evolutionary change, so they did not provide fine-scale information concerning the studied populations. In order to gain further insight into the evolutionary history of the French Pyrenean populations, we conducted a genetic analysis of fecal samples collected throughout the region using microsatellite markers developed in our laboratory (Gillet et al. 2016). Fecal sampling is the easiest way to detect the presence of the Pyrenean desman and obtain DNA for genetic analyses, as previously demonstrated (Gillet et al. 2015). We sought to determine the spatial distribution of genetic diversity in desman populations and to quantify gene flow across the French Pyrenees. The ultimate goal of this study was to enhance our general knowledge of this endangered and elusive species to better inform its conservation.

Material and Methods

Sampling and DNA extraction.

A total of 38 tissue and 355 fecal samples derived from the entire French range of this species were used in this study. The tissue samples came from specimens found dead and collected by our research team. Samples were collected from 2011 to 2014. The license numbers from the French departments used to collect this material are available upon request. Fecal samples from Pyrenean desmans were identified by amplification of a small mitochondrial cytochrome b fragment (Gillet et al. 2015). Genomic DNA from tissue and fecal samples preserved in ethanol was extracted using the DNeasy Tissue Kit (Qiagen Inc., Hilden, Germany) and the Stool Mini Kit (Qiagen Inc.), respectively, according to the manufacturer’s instructions. To avoid cross-contamination, DNA extractions from feces were conducted in a separate room with an UV-sterilized platform where no Pyrenean desman tissue samples had been previously treated.

DNA amplification.

The 393 samples used in this study were genotyped at 24 variable microsatellite loci using the multiplex sets and PCR conditions reported in Gillet et al. (2016), with slightly modified conditions for fecal samples, where PCRs were carried out in a 10-µl volume containing 0.15 of each 20-µM primer, 7.5 µl of Multiplex PCR kit (Qiagen Inc.), and 5 µl of DNA. Amplified DNA was analyzed for length variations on an ABI 3700 sequencer using GeneScan 500LIZ size standard, and alleles were scored with GENEMAPPER 4.0 (Applied Biosystems, Foster City, California). Consensus genotypes were constructed for fecal samples to prevent genotyping errors in our data set. For this, we used a modified multitube PCR approach (Taberlet et al. 1996) and repeated each PCR 4 times. Allele scores were accepted if they appeared at least 3 times in 4 PCRs.

Statistical analyses.

Each replicate genotype was compared with the consensus genotype to quantify the error rates. The consensus genotype construction and error rate quantification—such as false alleles (FA) and allelic dropouts (ADO)—were both performed using GIMLET v1.3.3 (Valiere 2002). The probability of identity among siblings (PIDsibs), i.e., the probability that 2 related individuals have the same genotype (Waits et al. 2001), was estimated using GIMLET v1.3.3. We used MICRO-CHECKER 2.2.3 (Van Oosterhout et al. 2004) to estimate the proportion of null alleles (NA). Genetic diversity was quantified by estimating observed (HO) and expected (HE) heterozygosities with GENETIX (Belkhir et al. 1996–2004). Hardy–Weinberg equilibrium (HWE) was tested using the exact test implemented in GENEPOP 4.1.0 (Rousset 2008) for each locus separately and over all loci for each cluster (see below). Tests for linkage disequilibrium between loci for each cluster were performed using GENEPOP 4.1.0. Allelic richness (AR) was calculated using the rarefaction procedure implemented in FSTAT 2.9.3.2 (Goudet 2001). Multilocus FIS was calculated for each cluster and adjusted for multiple tests using Bonferroni’s correction with FSTAT 2.9.3.2.

Population structure.

We used STRUCTURE 2.3.1 (Pritchard et al. 2000) to detect genetic structure in our data set. We used the model-based Bayesian clustering method, with no prior identification of populations, to infer the number of genetic clusters (K) and assign individuals to these clusters according to allele frequencies at each locus. For each K-value from 1 to 10, the program was run 10 times using an admixture model with a burn-in of 105 and Markov chain Monte Carlo (MCMC) values of 106. The ΔK method (Evanno et al. 2005) was implemented with STRUCTURE HARVESTER (Earl and vonHoldt 2011) to find the most likely K-value present in the data set and a visual output of the STRUCTURE results was generated using DISTRUCT (Rosenberg 2003) and CLUMPAK (Kopelman et al. 2015).

We used ARLEQUIN (Excoffier et al. 2005) to estimate pairwise genetic differentiation among populations using FST statistics, and the online SMOGD application (http://www.ngcrawford.com/django/jost/) was used to estimate Dest statistics (Jost 2008) using 1,000 bootstrap replicates.

Isolation-by-distance (IBD) analyses among and within clusters defined with STRUCTURE 2.3.1 were performed with GENEPOP 4.1.2. The signal strength was estimated by calculating Dσ2 (i.e., product of the population density and axial mean square parent–offspring distance as defined by Rousset (1997)), according to b = 1/(4πDσ2). The value obtained (Dσ2) was inversely correlated with the IBD strength. The logarithm of the Euclidean distance on GPS coordinates was used to calculate the geographic distance and âr statistics were used to represent the genetic distance between pairs of individuals (Rousset 2000). A Mantel’s test with 10,000 permutations was used to test the significance of the correlation.

Finally, we used BOTTLENECK 1.2 (Cornuet and Luikart 1996) to perform a Wilcoxon test under a 2-phase model (TPM) to investigate recent demographic bottlenecks with estimations based on 1,000 replications.

Demographic history.

The evolutionary history of G. pyrenaicus in France was investigated using approximate Bayesian computation (ABC) as implemented in DIYABC 1.0.4.45beta software (Cornuet et al. 2010). This coalescent-based approach allows estimation of the effective population size as well as the splitting time in generations for each tested genetic cluster. A number of biogeographic scenarios were tested and compared to determine whether the observed clusters originated from fragmentation of an ancestral common population or if any cluster resulted from an admixture of the others. Specifically, the type 1 scenarios explored 2 consecutive divergences with a 1st divergence between 2 of the 3 populations, followed by divergence of the 3rd population from 1 of the 2 others (6 alternative scenarios; Fig. 1). The type 2 scenarios displayed an admixture event between 2 populations, leading to formation of the 3rd population (3 alternative scenarios; Fig. 1), while the type 3 scenario showed a radiation process, where the 3 populations would have split at the same time from a common ancestor (Fig. 1). Two runs were performed. In the 1st one, we considered all alternative scenarios, whereas only those having the highest posterior probabilities (PP) were considered in the 2nd run. The distribution and range of priors for the parameters used to describe these alternative scenarios (effective population size, time of splitting or merging events in generations, and admixture rates) are given in Table 1. A total of 107 and 2 × 106 data sets were simulated for each alternative scenario in the 1st and 2nd runs, respectively, in order to build a reference table from a set of prior parameter distributions. To check if the combination of these distributions of prior parameters and alternative scenarios could generate data sets similar to the observed ones, a principal component analysis (PCA) was performed on the first 105 simulated data sets in the 1st run and on the first 2 × 105 data sets in the 2nd run. Inspection of PCAs helped us choose the most adequate timeframe corresponding to our data (maximum of 500 generations backward in time). We used microsatellite mutation rates generally used for mammalian species (i.e., 10−3 to 10−5Dallas 1992; Weber and Wong 1993; Ellegren 1995). Software default values were chosen for admixture rates. To determine the most likely alternative scenarios, we used normalized Euclidean distances between each simulated data set and our observed data set and 1% of the closest simulated data sets were used to estimate the relative PP (with 95% confidence intervals [CIs]) of each alternative scenario with a logistic regression (Cornuet et al. 2008). The most likely alternative scenario was that with the highest PP with a nonoverlapping 95% CI.

Fig. 1.

Schematic representation of 3 scenarios designed to test the origin of the 3 populations of Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees found in our study by approximate Bayesian computation (ABC) analysis. Prior parameters are defined in Table 1. Colors correspond to the colors in Fig. 2.

Fig. 1.

Schematic representation of 3 scenarios designed to test the origin of the 3 populations of Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees found in our study by approximate Bayesian computation (ABC) analysis. Prior parameters are defined in Table 1. Colors correspond to the colors in Fig. 2.

Table 1.

Prior distribution of parameters used in our approximate Bayesian computation (ABC) analysis of the evolutionary history of the Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees based on 24 variable microsatellite loci from tissue and fecal samples from 2011 to 2014. SNI = single nucleotide insertion.

Parameter Distribution Min Max 
Effective population size 
N1, N2, N3, Na Uniform 10 500 
Time of events (in generations backward in time) 
 Time conditions: ta2 > ta1, tb2 > tb1, tc2 > tc1, td2 > td1, te2 > te1, tf2 > tf1, tg2 > tg1, th2 > th1, ti2 > ti1 Uniform 10 500 
Admixture rate (ra) Uniform 0.001 0.999 
Microsatellite mutation model parameters 
 Mean mutation rate Uniform 10−5 10−3 
 Mean coefficient p Uniform 0.1 0.3 
 Mean SNI rate Log-uniform 10−8 10−4 
Parameter Distribution Min Max 
Effective population size 
N1, N2, N3, Na Uniform 10 500 
Time of events (in generations backward in time) 
 Time conditions: ta2 > ta1, tb2 > tb1, tc2 > tc1, td2 > td1, te2 > te1, tf2 > tf1, tg2 > tg1, th2 > th1, ti2 > ti1 Uniform 10 500 
Admixture rate (ra) Uniform 0.001 0.999 
Microsatellite mutation model parameters 
 Mean mutation rate Uniform 10−5 10−3 
 Mean coefficient p Uniform 0.1 0.3 
 Mean SNI rate Log-uniform 10−8 10−4 

To assess the level of confidence of these analyses, new data sets were simulated with each alternative scenario and the same procedure to estimate their respective PPs was applied, and the proportion of times the right alternative scenario had the highest PP was measured. According to Cornuet et al. (2010), the type I error was estimated for 10,000 simulated data sets generated under the best-supported alternative scenario. The type II error was estimated by simulating 10,000 data sets generated for each alternative scenario and counting decisions in favor of the selected alternative scenario.

We also estimated historical demographic events and genetic parameters including interactions between clusters (i.e., migration rate [M]) using MIGRATE 3.4.4 (Beerli and Felsenstein 1999, 2001; Beerli 2004, 2006; Beerli and Palczewski 2010). This software is able to search through genealogies and obtain estimates of theta (Θ) and M by employing a Metropolis–Hastings MCMC algorithm and a likelihood ratio test, respectively. It assumes a constant Θ for each population but a variable Θ between them (pairwise migration rate estimates).

We first used MIGRATE 3.4.4 with default parameters, with FST-based statistics of Θ and M, 10 short chains of 10,000 sampled genealogies, and 3 long chains of 100,000 sampled genealogies. The parameter estimates of Θ and M from the previous run were used as starting values to perform a 2nd analysis. The formula xNem = M*Θ was used to calculate the headcount of immigrants per generation, with x being the inheritance scalar (set at 4 for diploid species), Ne the effective population size, and m the mutation rate per generation and per locus.

Results

Microsatellite genotyping.

A total of 70 individuals were identified out of the 355 fecal samples analyzed with a PIDsibs of 2.24e−03. The mean proportion of positive PCRs was 67%, ranging from 53% to 76% among loci. No significant allelic dropout or false allele errors could be found in our data (all loci < 0.001). MICRO-CHECKER did not detect any significant bias in our data set that could be attributed to null alleles. Therefore, the total number of individuals used in our analyses was 108 (38 tissue samples and 70 individuals identified from fecal samples). It is important to note that the results of the genotyping of fecal samples were highly dependent on the freshness and size of the feces at the time of collection as feces of Pyrenean desmans are generally small (10–15 mm long and 4–8 mm wide—Bertrand 1993) and their DNA content rapidly degrades due to contact with water and UV radiation (Lindahl 1993). In our study, only 20% of the collected feces could be attributed to distinct individuals. The percentage of detected individuals also was dependent on the threshold rule of the conservative multitube approach that we used, i.e., allele scores were accepted if they appeared at least 3 times in 4 PCRs. This step was nonetheless necessary to ensure reliable results

Population structure and genetic diversity.

After using the ΔK method on our STRUCTURE results, the highest ΔK-value was found at K = 3 (Fig. 2). The 1st cluster (C), which appeared to have the largest geographic distribution (Fig. 2), mainly included samples from the Garonne watershed in the central Pyrenees. The 2nd cluster (E) mainly included samples from the Tet–Tech–Aude watershed in the eastern Pyrenees. Finally, the 3rd cluster (W) mainly included samples from the Adour–Nive watershed in the western Pyrenees.

Fig. 2.

Population structure of the Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees estimated using STRUCTURE (K = 3). Each individual is represented by a vertical line partitioned into K color segments, with the length of each color being proportional to the estimated membership coefficient (inset, lower left). Geographic distribution of the 3 genetic clusters is shown on the map. The 3 main watersheds of the Pyrenees are, from left to right: Adour–Nive, Garonne, and Tet–Tech–Aude.

Fig. 2.

Population structure of the Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees estimated using STRUCTURE (K = 3). Each individual is represented by a vertical line partitioned into K color segments, with the length of each color being proportional to the estimated membership coefficient (inset, lower left). Geographic distribution of the 3 genetic clusters is shown on the map. The 3 main watersheds of the Pyrenees are, from left to right: Adour–Nive, Garonne, and Tet–Tech–Aude.

The mean HO ranged from 0.19 in the eastern cluster to 0.23 in the central cluster, while the mean HE ranged from 0.26 in the eastern cluster to 0.37 in the western cluster (see Supplementary Data SD1, SD2, and SD3). The mean allelic richness ranged from 1.4 to 1.8 (Table 2). Tests for HWE showed significant deviations for the eastern and western clusters. Four pairs of loci (GpyrGS22 versus GpyrGS33, GpoyrGS30 versus GpyrGS74, GpyrGS33 versus GpyrGS18, and GpyrGS11 versus GpyrGS20) showed significant linkage disequilibrium for the central cluster and 1 pair (GpyrGS33 versus GpyrGS82) for the eastern cluster after Bonferroni correction. The inbreeding coefficient (FIS) was significant for all 3 clusters (Table 2) and all pairwise FST were significant (Table 3). Moreover, the results of the Wilcoxon test under a TPM performed using the BOTTLENECK 1.2 software package indicated that both the central and eastern clusters had undergone a recent bottleneck (P < 0.001).

Table 2.

Overview of genetic parameters for each main cluster for the Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees based on 24 variable microsatellite loci from tissue and fecal samples from 2011 to 2014. N: number of samples; HO: mean observed heterozygosity; HE: mean expected heterozygosity; HWE: deviation from Hardy–Weinberg equilibrium (significance level = 0.002); AR: mean allelic richness; FIS: mean inbreeding coefficient.

Clusters N HO HE HWE AR FIS 
Central 45 0.226 0.274 0.04 1.405 0.179 
Eastern 38 0.189 0.258 < 0.002 1.803 0.271 
Western 25 0.216 0.368 < 0.002 1.703 0.434 
Clusters N HO HE HWE AR FIS 
Central 45 0.226 0.274 0.04 1.405 0.179 
Eastern 38 0.189 0.258 < 0.002 1.803 0.271 
Western 25 0.216 0.368 < 0.002 1.703 0.434 
Table 3.

FST (above diagonal—ARLEQUIN software) and Dest (below diagonal—SMOGD software) for the 3 main clusters (P < 0.05) for the Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees based on 24 variable microsatellite loci from tissue and fecal samples from 2011 to 2014.

Clusters Central Eastern Western 
Central  0.288 0.203 
Eastern 0.037  0.345 
Western 0.013 0.084  
Clusters Central Eastern Western 
Central  0.288 0.203 
Eastern 0.037  0.345 
Western 0.013 0.084  

The high observed FIS-values, especially in the western cluster (Table 2), could also indicate a Wahlund effect in the genetic clusters. To investigate this possibility, we conducted an additional clustering analysis within each cluster, under the same conditions as previously described. The analyses for clusters C and E did not give any evidence of substructure (all individuals admixed beyond K = 1), whereas that for cluster W supported the existence of 3 subclusters (Fig. 3). This substructuring could explain the higher inbreeding coefficient (0.434) in the western cluster, even though these subclusters did not seem to exhibit any clear geographical distribution (Fig. 3).

Fig. 3.

Population structure estimated in the western cluster for the Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees using STRUCTURE (K = 3). Each individual is represented by a vertical line partitioned into K color segments, with the length of each color being proportional to the estimated membership coefficient (insert, lower left). Geographic distribution of the 3 Pyrenean desman clusters is shown on the map. Each diagram represents 1 individual with its respective cluster assignations from STRUCTURE.

Fig. 3.

Population structure estimated in the western cluster for the Pyrenean desman (Galemys pyrenaicus) in the French Pyrenees using STRUCTURE (K = 3). Each individual is represented by a vertical line partitioned into K color segments, with the length of each color being proportional to the estimated membership coefficient (insert, lower left). Geographic distribution of the 3 Pyrenean desman clusters is shown on the map. Each diagram represents 1 individual with its respective cluster assignations from STRUCTURE.

Demographic history.

After using DIYABC software to investigate 10 distinct demographic alternate scenarios, a 2nd analysis was performed on the 2 most probable alternate scenarios (those with the highest PPs) obtained in the 1st run. These 2 alternate scenarios (2 and 9; Fig. 4) exhibited 2 different evolutionary demographic patterns. Alternate scenario 2 reflected the separation of cluster W from more eastern Pyrenean desmans, followed by the separation of cluster C from cluster E. In contrast, alternate scenario 9 reflected the origin of the central cluster following a merging event between the other 2 clusters. The logistic regressions performed on 1% of the closest simulated data sets revealed that the most likely of all of the tested alternate scenarios was alternate scenario 2, with a PP of 0.582 and a CI (95% CI) of 0.475–0.689. However, the CIs of alternate scenarios 2 and 9 overlapped (PP of 0.418 and 95% CI of 0.310–0.525). Analysis of confidence in alternate scenario 2 resulted in type I and type II errors of 0.266 and 0.238, respectively, and inversely for alternate scenario 9. Therefore, the 2 alternate scenarios had similar probability of being correct, with alternate scenario 2 having a slight better probability (76.2%) than alternate scenario 9 (73.4%). However, the divergence times (t2a, t2b, t9a, and t9b) and estimates of effective population size (NC, NE, and NW) were all within the same order of magnitude when comparing the posterior distribution of parameters of the 2 alternate scenarios (Fig. 4). Assuming a minimum generation time of 1 year, the median divergence times t2a and t2b of alternate scenario 2 were estimated at 80 (95% CI: 25–190) and 230 years ago (95% CI: 80–440), respectively. For alternate scenario 9, the median admixture and divergence times t9a and t9b were estimated at 60 (95% CI: 18–160) and 240 years ago (95% CI: 80–450), respectively (Fig. 4). The effective population size estimates for clusters C, E, and W had a mean of 370 (95% CI: 90–490), 100 (95% CI: 30–220), and 330 (95% CI: 140–480) individuals, respectively, for alternate scenario 2, and 290 (95% CI: 100–470), 100 (95% CI: 30–220), and 320 (95% CI: 130–480), respectively, for alternate scenario 9 (Fig. 4).

Fig. 4.

Schematic representations of the 2 most likely alternative scenarios regarding structuring of the Pyrenean desman (Galemys pyrenaicus) population in the French Pyrenees based on approximate Bayesian computation (ABC) analysis. NC, NE, and NW are the effective population sizes for the central, eastern, and western clusters, respectively. Numbers for NC, NE, and NW correspond to number of individuals included in the analysis.

Fig. 4.

Schematic representations of the 2 most likely alternative scenarios regarding structuring of the Pyrenean desman (Galemys pyrenaicus) population in the French Pyrenees based on approximate Bayesian computation (ABC) analysis. NC, NE, and NW are the effective population sizes for the central, eastern, and western clusters, respectively. Numbers for NC, NE, and NW correspond to number of individuals included in the analysis.

We also conducted another DIYABC analysis on the 3 western subclusters with the same alternate scenarios as for the 3 main clusters, but with slightly modified conditions (effective population sizes and time of events set at max. 300). After the 1st run with 10 alternate scenarios, 2 were well supported and thus a 2nd run with these 2 alternate scenarios was launched. Finally, the software unambiguously chose 1 alternate scenario from the 2, with a PP of 0.585 (95% CI of 0.565–0.605). This alternative scenario displayed the same evolutionary demographic pattern as alternate scenario 2 from Fig. 4. The median times of divergence among these clusters (populations N1, N2, and N3; Fig. 3) were quite similar to those found for the 3 main groups (100 and 200 years ago). The mean effective population sizes were estimated at 185, 160, and 120 individuals for N1, N2, and N3, respectively.

We used MIGRATE 3.4.4 on the 3 main clusters to calculate the migration rate per generation according to Nem = (Mi jj)/4, with ΘC = 2.86, ΘE = 1.14, ΘW = 1.14 and ME C = 0.83, MW C = 0.46, MC E = 1.38, MW E = 0.39, MC W = 0.84, ME W = 0.87. The effective number of migrants per generation among clusters was low, with less than 1 migrant per generation (1 year) with NmE C = 0.59, NmW C = 0.32, NmC E = 0.39, NmW E = 0.11, NmC W = 0.24, and NmE W = 0.25.

Isolation by distance.

The IBD analyses indicated an absence of IBD among clusters and within clusters E and W. In contrast, a significant signal was observed within cluster C (Mantel test P < 0.05), as indicated by the relatively low Dσ2 value (0.44). However, the slight regression slope indicated that the genetic distance between pairs of individuals was weakly correlated with the geographic distance between them (r2 = 0.05). The absence of significant signal among the 3 clusters suggests that their relationships (e.g., a closer link between the cluster E and C as compared to the cluster W as suggested by the DIYABC results) were not driven by geographic distance alone.

Discussion

The present study generated new insight into the genetic structure of Pyrenean desman populations in the French Pyrenees. The large area sampled, covering the entire current distributional area of the Pyrenean desman on the French side of the Pyrenees, and the use of 24 highly variable microsatellite markers provided a more fine-scale view of genetic structure than previously available.

Population structure and evolutionary history of the French Pyrenean desman.

The results showed evidence of 3 genetically and geographically distinct clusters (Fig. 2) situated in the eastern (E), central (C), and western (W) French Pyrenees. In addition, substructuring seemed to emerge within cluster W, where 3 subclusters were detected. These subclusters did not exhibit any clear geographical distribution but this may be the result of low sampling within each subcluster (6–10 individuals). More extensive sampling in this area is needed to better define this substructure.

The existence of eastern, central, and western genetic clusters was not evident in the study of Igea et al. (2013), which suggested the existence of a single Pyrenean population that would have recently colonized the Pyrenean region from a putative refuge situated in the Basque Mountains after a severe bottleneck event. This hypothesis was deduced from the very low levels of genetic diversity found in both mitochondrial and nuclear (intron) marker sequences in this region. Our microsatellite markers, characterized by higher mutation rates than mitochondrial or nuclear intronic sequences (Schlötterer 2000), allowed us to detect finer genetic structure in the Pyrenean desman. Therefore, according to their different evolutionary rates, both mitochondrial and nuclear markers gave complementary information concerning the evolutionary history of the desman in this region. The species probably colonized the Pyrenean region after the last ice age, and subsequent diversification led to the 3 genetic clusters, presently distributed throughout the French Pyrenean region. The only other vertebrate species having a similar reported genetic structure across the French Pyrenees is the rock ptarmigan (Lagopus mutaBech et al. 2009). However, in contrast to the Pyrenean desman, the genetic structure of this bird species was associated with a significant IBD effect, likely the result of short dispersal distances, and high natal and breeding philopatry combined with severe habitat fragmentation.

As no IBD was detected, the structure supported by our study is likely the result of concomitant environmental and anthropogenic factors. The DIYABC analysis proposed that about 80 years ago clusters C and E diverged from an ancestral population, which itself had diverged from the W cluster approximately 230 years ago, or that cluster C was a result of an admixture event between clusters E and W that occurred about 60–80 years ago, after clusters E and W diverged about 230–240 years ago. However, these estimations might be underestimated given that the software algorithm does not assume migration within scenario events. Moreover, these estimations were performed while considering a minimum generation time of 1 year even though this information is unknown for the Pyrenean desman and could be higher. Indeed, as G. pyrenaicus belongs to the family Talpidae, if we extrapolate the generation time of the Pyrenean desman from that of the European mole (Talpa europaea—1.72 years—Niethammer 1990), the estimated divergence times become 400 years ago for the separation of the E and W clusters and 100–140 years ago for the divergence of C and E (or the admixture of E and W in the 2nd alternative scenario).

Human population growth over the last century, and therefore the increased human impact on nature, inevitably led to riverine habitat loss and fragmentation of species’ populations inhabiting mountain streams. The construction and functioning of hydroelectric power plants is an example of the human impact on rivers. These can lead to physical and biotic modifications and alter both hydrologic and thermal regimes, thus impacting resources such as benthic macroinvertebrate larvae (Queiroz et al. 1992; Céréghino and Lavandier 1997). Although little information is available on the Pyrenean desman, these detrimental effects have been studied and highlighted by various authors for other mammals and birds (Nilsson and Dynesius 1994; D’Amico et al. 2000). Furthermore, the development of such infrastructures dates back to the beginning of the 20th century in the French Pyrenees, and most were built between 1930 and 1960. These dates could therefore coincide with the emergence of the 3rd population found in this study (60–80 years ago). However, more focused studies are needed to further understand the influence of this infrastructure on the distribution of Pyrenean desman and gene flow among populations.

Centennial-scale climatic change could also have played a role in the structuring of Pyrenean desman populations, notably during the last 100 years of the Little Ice Age (1750–1850). Cooler temperatures at high elevations could have induced a shift to lower elevations in the range of the Pyrenean desman. After the Little Ice Age, the Pyrenean desman populations could have shifted back to higher elevations, which in turn could have restricted dispersal. Indeed, rivers at high elevations are less connected and this species favors rivers with high water flows, which are now found at high elevations (Nores et al. 1992; Ramalhinho and Boa Vida 1993; Queiroz et al. 1996; Charbonnel et al. 2015).

Genetic diversity of the Pyrenean desman populations.

The 3 main clusters seemed to be characterized by a heterozygote deficiency, as indicated by the relatively low heterozygosity values (around 0.2) and significantly high FIS indices (Table 2). These data would be associated with recent bottleneck events, at least for clusters C and E, as confirmed by the BOTTLENECK 1.2 analysis.

The very high inbreeding coefficient in the western cluster could also be explained by a Wahlund effect as 3 subclusters were found in this population. As for the 3 main populations, the impact of both anthropogenic and climatic factors (notably during the Little Ice Age) along with watershed structure could jointly explain differentiation of the 3 subclusters. This substructure in the western population also could be due to the fact that this region is characterized by a smaller proportion of favorable habitats compared to the more eastern portion of the distribution (Charbonnel 2015). However, the effective population sizes estimated by ABC were inconsistent with this favorable habitat gradient, with a higher value in the western than in the eastern population. The substructure of the western population could have biased the effective population size estimation.

Gene flow seemed limited among the 3 main clusters, as indicated by pairwise FST values that were significantly higher than zero and ranged between 0.345 (between the W and E cluster) and 0.203 (between the C and W clusters). This trend also was revealed using the Dest index (Table 3).

Despite this apparent low gene flow between clusters, they were not geographically separated and they overlapped in some areas (Fig. 2). Moreover, individual cluster assignments from the Bayesian analysis clearly showed that some individuals shared alleles from different clusters and have admixed genomes. This pattern was observed particularly between the E and C clusters (Fig. 2), where migrations seemed to have occurred every 2 or 3 generations. Although this result has been viewed as a recent expansion process in other species such as the European otter (Lutra lutraJanssens et al. 2008), the general regression of the range of the Pyrenean desman over the last 3 decades did not allow us to retain this hypothesis. However, this overlapping of clusters could confirm that the genetics of the Pyrenean desman were not markedly impacted by the river networks (Igea et al. 2013) and that its dispersal could be complex, as pointed out by Stone (1987a, 1987b) and Melero et al. (2012, 2014). Our estimates of migration rate also tended to confirm recent but limited gene flow among clusters.

Future studies will be needed to place the patterns found here in a broader spatial framework that includes the entire Pyrenean distribution of this species. For example, it is possible that some individuals included in our study were migrants or descendants of migrants from the Spanish side of the Pyrenees, particularly the 2 green-colored individuals in the central cluster shown in Fig. 2, which could have passed through the Val d’Aran. This may suggest that Pyrenean desmans could cross the mountains from one side to the other. In addition, contact zones between both sides of the Pyrenees have been identified for Corthippus grasshoppers (Buño et al. 1994), Phylloscopus birds (Helbig et al. 2001), and viviparous lizards (Zootoca viviparaMilá et al. 2013). These contact zones, which are situated across the central high Pyrenees and across the southwestern Pyrenees, could also exist for the Pyrenean desman. Igea et al. (2013) suggested genetic homogeneity across the distribution of Pyrenean desman using relatively slowly evolving mitochondrial and nuclear markers, but further analysis of Spanish and French samples using hypervariable markers is needed to gain further insight concerning the broader-scale genetic structure of the Pyrenean desman.

Implications for conservation of the Pyrenean desman in the French Pyrenees.

Classified as “Vulnerable” on the IUCN Red List (Fernandes et al. 2008), the Pyrenean desman is legally protected in France and is the focus of a LIFE+ project (LIFE13NAT/FR/000092) under a National Action Plan (Némoz and Bertrand 2008). This species has been undergoing habitat loss and fragmentation for decades, particularly in France where its range still requires further investigation (Némoz and Bertrand 2008).

The low level of genetic diversity observed in the different French Pyrenean desman clusters as well as the heterozygote deficiency highlighted by the high inbreeding coefficient (FIS) values, and relatively low effective population sizes within clusters, could increase the risk of extinction for this species in the future (Frankham 2005).

A lack of genetic diversity within the 3 main populations would lead to increased risk of inbreeding depression. Greater connectivity throughout the Pyrenees should therefore be fostered to facilitate individual dispersal and gene flow among the 3 main populations. This would favor genetic mixing and a better response to future climatic change. Exchanges between neighboring watersheds should be promoted by improving water quality, mainly at lower elevations where rivers merge. Indeed, the Pyrenean desman favors rivers with high water flows, which are now found at high elevations (Nores et al. 1992; Ramalhinho and Boa Vida 1993; Queiroz et al. 1996; Charbonnel et al. 2015). This preference for high water flows at high elevations could contribute to the genetic structure observed in our study as this species is more inclined to live at elevations where rivers are less connected. In addition, enhanced management of hydroelectric infrastructures and of winter tourism at high elevations should be promoted throughout the mountain chain.

Another improvement could be achieved by increasing water flow in rivers with low trophic resources or by restoring suitable habitats for the Pyrenean desman, notably by placing stones and boulders in rivers to re-create adequate water flow and shelter. Connectivity between main and tributary rivers should also be favored as tributary rivers can serve as refugia in case of short and sudden hydrological events (Lake 2000; Charbonnel 2015).

Although this study generated new insight into the fine-scale genetic structure of the Pyrenean desman in France, a larger study, based on sensitive genetic markers such as microsatellite or SNP markers and encompassing the entire range of the species, particularly the Spanish side of the Pyrenees, is necessary to broaden overall knowledge on this threatened species and its worldwide conservation.

Supplementary Data

Supplementary data are available at Journal of Mammalogy online.

Supplementary data SD1.—Observed (HO) and expected heterozygosity (HE) per locus for the central cluster. Only polymorphic loci are shown.

Supplementary data SD2.—Observed (HO) and expected heterozygosity (HE) per locus for the eastern cluster. Only polymorphic loci are shown.

Supplementary data SD3.—Observed (HO) and expected heterozygosity (HE) per locus for the western cluster. Only polymorphic loci are shown.

Acknowledgments

We thank the following people who collected tissue samples: EDF agents, Pyrenees National Park agents, M. Bayon, A. Bertrand, J.-P. Besson, J.-P. Quéré, A. Charbonnel, F. Elzear, L. Fabre, P. Fantin, B. Le Roux, V. Lacaze, M. Lagardère, F. Lassèrre, B. Le Corre, M. Mas, P. Maunas, G. Nogué, F. Prud’Homme, T. Quintilla, B. Salmeron, T. Tico, S. Torreilles, and S. Vernet. We also thank representatives of the following organizations who collected feces samples: Association des Naturalistes de l’Ariège, Conservatoire d’Espaces Naturels d’Aquitaine, Conservatoire d’Espaces Naturels de Midi-Pyrénées, Fédération Aude Claire, Fédération des Réserves Naturelles Catalanes, Groupe de Recherche et d’Etude pour la Gestion de l’Environnement, Office National de la Chasse et de la Faune Sauvage, Office National des Forêts, and Parc National des Pyrénées. This study is part of the “Plan National d’Actions en faveur du Desman des Pyrénées” and the LIFE+ Desman project (LIFE13NAT/FR/000092) which are coordinated by the Conservatoire d’Espaces Naturels de Midi-Pyrénées (CEN-MP) and financially supported by the following structures: European Union Funding Network (ERDF and LIFE+), Agence de l’eau Adour-Garonne, Agence de l’eau Rhône-Méditerranée-Corse, DREAL Aquitaine, Midi-Pyrénées, and Languedoc-Roussillon, Conseil Régional Aquitaine, Midi-Pyrénées and Languedoc-Roussillon, Conseil Général des Pyrénées-Atlantiques, de l’Aude et des Pyrénées-Orientales, EDF, SHEM, Patagonia, Parc National des Pyrénées, and ANRT (Association Nationale de la Recherche et de la Technologie). FG is supported by a French research fellowship provided by ANRT (CIFRE N° 2011/1571).

Literature Cited

Bech
N.
Boissier
J.
Drovetski
S.
Novoa
C
.
2009
.
Population genetic structure of rock ptarmigan in the “sky islands” of French Pyrenees: implications for conservation
.
Animal Conservation
 
12
:
138
146
.
Beerli
P
.
2004
.
Effect of unsampled populations on the estimation of population sizes and migration rates between sampled populations
.
Molecular Ecology
 
13
:
827
836
.
Beerli
P
.
2006
.
Comparison of Bayesian and maximum-likelihood inference of population genetic parameters
.
Bioinformatics
 
22
:
341
345
.
Beerli
P.
Felsenstein
J
.
1999
.
Maximum-likelihood estimation of migration rates and effective population numbers in two populations using a coalescent approach
.
Genetics
 
152
:
763
773
.
Beerli
P.
Felsenstein
J
.
2001
.
Maximum likelihood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach
.
Proceedings of the National Academy of Sciences of the United States of America
 
98
:
4563
4568
.
Beerli
P.
Palczewski
M
.
2010
.
Unified framework to evaluate panmixia and migration direction among multiple sampling locations
.
Genetics
 
185
:
313
326
.
Belkhir
K.
Borsa
P.
Chikhi
L.
Raufaste
N.
Bonhomme
F
.
1996–2004
.
GENETIX 4.05, logiciel sous Windows TM pour la génétique des populations
 .
Laboratoire Génome, Populations, Interactions, CNRS UMR 5000, Université de Montpellier II
,
Montpellier (France)
.
Bertrand
A
.
1993
.
Répartition géographique du Desman des Pyrénées
 .
Arvicola
5
:
11
12
.
Bertrand
A
.
1994
.
Répartition géographique et écologie alimentaire de desman des pyrénées G. pyrenaicus dans les Pyrénées françaises
 .
Diplôme Universitaire de Recherche
,
Toulouse, France
.
Buño
I.
Torroja
E.
López-Fernández
C.
Butlin
R. K.
Hewitt
G. M.
Gosálvez
J
.
1994
.
A hybrid zone between two subspecies of the grasshopper Chorthippus parallelus along the Pyrenees: the west end
.
Heredity
 
73
:
625
634
.
Céréghino
R.
Lavandier
P
.
1997
.
Influence des éclusées hydroélectriques sur la distribution et le développement larvaire des Diptères Simuliidae d’une rivière de moyenne montagne
.
Comptes Rendus l’Académie des Sciences - Serie III - Sciences de la Vie
 
320
:
329
338
.
Charbonnel
A
.
2015
.
Multi-scale influence of environmental factors in the distribution of the Pyrenan desman (Galemys pyrenaicus) in France
.
Ph.D. thesis
,
University of Toulouse
,
Toulouse, France
.
Charbonnel
A.
et al
2015
.
Integrating hydrological features and genetically validated occurrence data in occupancy modelling of an endemic and endangered semi-aquatic mammal, Galemys pyrenaicus, in a Pyrenean catchment
.
Biological Conservation
 
184
:
182
192
.
Cornuet
J.-M.
Luikart
G
.
1996
.
Power analysis of two tests for detecting recent population bottlenecks from allele frequency data
.
Genetics
 
144
:
2001
2014
.
Cornuet
J.
Ravigné
V.
Estoup
A
.
2010
.
Inference on population history and model checking using DNA sequence and microsatellite data with the software DIYABC (v1. 0)
.
BMC Bioinformatics
 
11
:
401
.
Cornuet
J.-M.
et al
2008
.
Inferring population history with DIY ABC: a user-friendly approach to approximate Bayesian computation
.
Bioinformatics
 
24
:
2713
2719
.
Coster
S. S.
Kovach
A. I
.
2012
.
Anthropogenic influences on the spatial genetic structure of black bears
.
Conservation Genetics
 
13
:
1247
1257
.
D’Amico
F.
Manel
S.
Mouches
C.
Ormerod
S. J
.
2000
.
River birds in regulated rivers: cost or benefit?
Verhandlungen des Internationalen Verein Limnologie
 
27
:
167
170
.
Dallas
J. F
.
1992
.
Estimation of microsatellite mutation rates in recombinant inbred strains of mouse
.
Mammalian Genome
 
3
:
452
456
.
Earl
D. A.
vonHoldt
B. M
.
2011
.
STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method
.
Conservation Genetics Resources
 
4
:
359
361
.
Ellegren
H
.
1995
.
Mutation rates at porcine microsatellite loci
.
Mammalian Genome
 
6
:
376
377
.
Evanno
G.
Regnaut
S.
Goudet
J
.
2005
.
Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study
.
Molecular Ecology
 
14
:
2611
2620
.
Excoffier
L.
Laval
G.
Schneider
S
.
2005
.
Arlequin (version 3. 0): an integrated software package for population genetics data analysis
.
Evolutionary Bioinformatics
 
1
:
47
50
.
Fernandes
M.
Herrero
J.
Aulagnier
S.
Amori
G
.
2008
. Galemys pyrenaicus.
The IUCN Red List of Threatened Species 2008
:
e.T8826A12934876
.
Frankham
R
.
2003
.
Genetics and conservation biology
.
Comptes Rendus Biologies
 
326
:
S22
S29
.
Frankham
R
.
2005
.
Genetics and extinction
.
Biological Conservation
 
126
:
131
140
.
Gerlach
G.
Musolf
K
.
2000
.
Fragmentation of landscape as a cause for genetic subdivision in bank voles
.
Conservation Biology
 
14
:
1066
1074
.
Gillet
F.
et al
2016
.
Isolation, characterization and PCR multiplexing of polymorphic microsatellite markers in the endangered Pyrenean desman, Galemys pyrenaicus
.
Conservation Genetics Resources
 
8
:
43
81
.
Gillet
F.
et al
2015
.
PCR-RFLP identification of the endangered Pyrenean desman, Galemys pyrenaicus (Soricomorpha, Talpidae), based on faecal DNA
.
Mammalia
 
79
:
473
477
.
Goudet
J
.
2001
.
FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.3)
 . http://www.unil.ch/izea/softwares/fstat.html. Accessed
13 April 2014
Helbig
A. J.
Salomon
M.
Bensch
S.
Seibold
I
.
2001
.
Male-biased gene flow across an avian hybrid zone: evidence from mitochondrial and microsatellite DNA
.
Journal of Evolutionary Biology
 
14
:
277
287
.
Igea
J.
et al
2013
.
Phylogeography and postglacial expansion of the endangered semi-aquatic mammal Galemys pyrenaicus
.
BMC Evolutionary Biology
 
13
:
115
.
Janssens
X.
et al
2008
.
Genetic pattern of the recent recovery of European otters in southern France
.
Ecography (Cop.)
 
31
:
176
186
.
Jost
L
.
2008
.
G ST and its relatives do not measure differentiation
.
Molecular Ecology
 
17
:
4015
4026
.
Kopelman
N. M.
Mayzel
J.
Jakobsson
M.
Rosenberg
N. A.
and
Mayrose
I
.
2015
.
Clumpak: a program for identifying clustering modes and packaging population structure inferences across K
.
Molecular Ecology Resources
 
15
:
1179
1191
.
Lake
P. S
.
2000
.
Disturbance, patchiness, and diversity in streams
.
Journal of the North American Benthological Society
 
19
:
573
592
.
Lindahl
T
.
1993
.
Instability and decay of the primary structure of DNA
.
Nature
 
362
:
709
715
.
Melero
Y.
Aymerich
P.
Luque-Larena
J. J.
Gosàlbez
J
.
2012
.
New insights into social and space use behaviour of the endangered Pyrenean desman (Galemys pyrenaicus)
.
European Journal of Wildlife Research
 
58
:
185
193
.
Melero
Y.
Aymerich
P.
Santulli
G.
Gosàlbez
J
.
2014
.
Activity and space patterns of Pyrenean desman (Galemys pyrenaicus) suggest non-aggressive and non-territorial behaviour
.
European Journal of Wildlife Research
 
60
:
707
715
.
Milá
B.
Surget-Groba
Y.
Heulin
B.
Gosá
A.
Fitze
P. S
.
2013
.
Multilocus phylogeography of the common lizard Zootoca vivipara at the Ibero-Pyrenean suture zone reveals lowland barriers and high-elevation introgression
.
BMC Evolutionary Biology
 
13
:
192
.
Mona
S.
Ray
N.
Arenas
M.
Excoffier
L
.
2014
.
Genetic consequences of habitat fragmentation during a range expansion
.
Heredity (Edinb)
 
112
:
291
299
.
Némoz
M.
Bertrand
A
.
2008
.
Plan National d’Actions en faveur du Desman des Pyrénéés (Galemys pyrenaicus), 2009–2014
 .
Société Française pour l’Etude et la Protection des Mammifères/Ministère de l’Ecologie, de l’Energie, du Développement Durable et de l’Aménagement du Territoire
,
Conservatoire d’Espaces Naturels de Midi-Pyrénées
,
Toulouse, France
.
Niethammer
J
.
1990
.
Talpa europaea Linnaeus, 1758 - Maulwurf
. Pp.
99
133
in
Handbuch der Säugetiere Europas. Band 3/I: Insektenfresser - Insectivora, Herrentier - Primates
 .
AULA-Verlag
,
Wiesbaden, Germany
.
Nilsson
C.
Dynesius
M
.
1994
.
Ecological effects of river regulation on mammals and birds: a review
.
Regulated Rivers: Research and Management
 
9
:
45
53
.
Nores
C.
Ojeda
F.
Ruano
A.
Villate
I.
Gonzalez
J
.
1992
.
Aproximación a la metodología y estudio del área de distribución, estatus de población y selección de hábitat del desmán (Galemys pyrenaicus) en la Península Ibérica
 .
Ministerio de Medio Ambiente
,
Oviedo, Spain
.
Nores
C.
Queiroz
A. I.
Gisbert
J
.
2007
.
Galemys pyrenaicus (E. Geoffroy Saint-Hilaire, 1811)
. Pp.
92
98
in
Atlas Y Libro Rojo de Los Mamíferos Terrestres de España
  ( L. Palomo J. Gisbert J. Blanco , eds.).
Dirección General para la Biodiversidad-SECEM-SECEMU
,
Madrid, Spain
.
Palmeirim
J. M.
Hoffmann
S. H
.
1983
.
Galemys pyrenaicus
.
Mammalian Species
 
207
:
1
5
.
Pritchard
J. K.
Stephens
M.
Donnelly
P
.
2000
.
Inference of population structure using multilocus genotype data
.
Genetics
 
155
:
945
959
.
Queiroz
A. I
.
1999
.
Galemys pyrenaicus (E. Geoffroy, 1811)
. Pp.
78
79
in
The atlas of European mammals
  (
Mitchell-Jones
A. J
et al
, eds.).
Poyser Natural History, Academic Press
,
London, United Kingdom
.
Queiroz
A. I.
Alves
M. H.
Almada
V
.
1992
.
The small hydroplants: predicted impacts on the Pyrenean desman populations (Galemys pyrenaicus Geoffroy)
. Pp.
69
77
in
Proceeding of the Meeting on the Pyrenean desman
,
Lisboa, Portugal
.
Queiroz
A. I.
Bertrand
A.
Khakhin
G
.
1996
.
Status and conservation of Desmaninae in Europe
 .
Council of Europe
,
Strasbourg, France
.
Ramalhinho
M. G.
Boa Vida
M. J
.
1993
.
Habitat of the Pyrenean desman: assessment of running water quality. Monitoring pollution
. Pp.
63
67
in
Proceedings of the Meeting on the Pyrenean desman
,
Lisbon, Portugal
.
Richard
P. B
.
1986
.
Les Desman des Pyrénées, un mammifère inconnu à découvrir
 .
Science et Découvertes, Ed. Le Rocher
,
Monaco
.
Rosenberg
N. A
.
2003
.
Distruct: a program for the graphical display of population structure
.
Molecular Ecology Notes
 
4
:
137
138
.
Rousset
F
.
1997
.
Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance
.
Genetics
 
145
:
1219
1228
.
Rousset
F
.
2000
.
Genetic differentiation between individuals
.
Journal of Evolutionary Biology
 
13
:
58
62
.
Rousset
F
.
2008
.
GENEPOP’007: a complete re-implementation of the genepop software for Windows and Linux
.
Molecular Ecology Resources
 
8
:
103
106
.
Schlötterer
C
.
2000
.
Evolutionary dynamics of microsatellite DNA
.
Chromosoma
 
109
:
365
371
.
Stone
R. D
.
1987a
.
The social ecology of the Pyrenean desman as revealed by radiotracking
.
Journal of Zoology
 
212
:
117
129
.
Stone
R. D
.
1987b
.
The activity pattern of the Pyrenean desman (Galemys pyrenaicus) (Insectivora: Talpidae), as determined under natural condition
.
Journal of Zoology (London)
 
213
:
95
106
.
Taberlet
P.
et al
1996
.
Reliable genotyping of samples with very low DNA quantities using PCR
.
Nucleic Acids Research
 
24
:
3189
3194
.
Valiere
N
.
2002
.
GIMLET: a computer program for analysing genetic individual identification data
.
Molecular Ecology Notes
 
2
:
377
379
.
Van Oosterhout
C.
Hutchinson
W. F.
Wills
D. P. M.
Shipley
P
.
2004
.
Micro-Checker: software for identifying and correcting genotyping errors in microsatellite data
.
Molecular Ecology Notes
 
4
:
535
538
.
Waits
L. P.
Luikart
G.
Taberlet
P
.
2001
.
Estimating the probability of identity among genotypes in natural populations: cautions and guidelines
.
Molecular Ecology
 
10
:
249
256
.
Weber
J. L.
Wong
C
.
1993
.
Mutation of human short tandem repeats
.
Human Molecular Genetics
 
2
:
1123
1128
.

Author notes

* Correspondent: f.gillet@hotmail.com
Associate Editor was Marjorie Matocq.

Supplementary data