Contrasted hybridization patterns between two local populations of European wildcats in France

The European wildcat (Felis silvestris silvestris) is threatened across the totality of its area of distribution by hybridization with the domestic cat F.s. catus. The underlying ecological processes promoting hybridization, remain largely unknown. In France, wildcats are mainly present in the North-East but signs of their presence in the Pyrenees have been recently provided. However, no studies have been carried out in the French Pyrenees to assess the genetic status of wildcats. We have compared a local population of wildcats living in a continuous habitat in the French Pyrenees and a local population of wildcats living in a fragmented habitat in Northeastern France to evaluate how habitat fragmentation influence the population structure of European wildcats. Close kin were not found in the same geographic location contrary to what was observed for females in the Northeastern wildcat population. Furthermore, there was no evidence of hybridization in the Pyrenean wildcats and only one domestic cat raised suspicions while hybridization was categorically detected in northeastern France. The two wildcat populations were significantly differentiated (Fst = 0.08) and the genetic diversity of the Pyrenean wildcats was lower than that of other wildcat populations in France and in Europe. Taken together, these results suggest that habitat fragmentation, and in particular the absence of agricultural fields, may play an important role in lowering the probability of hybridization by reducing the likelihood of contact with domestic cats. Moreover, our results suggest that the French Pyrenean wildcat populations is isolated and may be threatened by a lack of genetic diversity.


INTRODUCTION
Hybridization is especially common between subspecies, due to incomplete 30 reproductive isolation and therefore a higher likelihood of successful interbreeding (1-31 3). This is the case for the European wildcat Felis silvestris silvestris, a medium-sized 32 carnivore widely spread in Europe (4), which is highly threatened over its entire 33 distribution area by its closely related domestic counterparts F. s. catus (5). 34 Interbreeding between wildcats and domestic cats may lead to introgressive 35 hybridization, followed by disruption of local genetic adaptations, and then to a loss of 36 the European wildcat genetic integrity, and even to the extinction of the sub-species (6). 37 Studies across the area of distribution of the European wildcat have shown that 38 there is a high degree of variability in the extent of admixture with domestic cats. High 39 levels (up to 45%) of hybridization have been reported in Hungary and Scotland (3,7-40 10), while low levels (between 0 and 2%) of interbreeding with domestic cats have been 41 shown in Germany, Italy, and Portugal (9)(10)(11)(12). The direction of the gene flow also varied, 42 some studies reporting a gene flow from domestic cats to wildcats (13,14) while others 43 showed the opposite with a detected flow from wildcats to domestic cats (15). Such high 44 degree of heterogeneity in hybridization modalities and subsequent introgression may 45 reflect different environmental conditions (e.g., habitat fragmentation, urban pressure). 46 Characterizing the patterns and processes of hybridization in nature is crucial to the 47 introduction of measures designed to prevent hybridization and then plan efficient 48 conservation guidelines for European wildcats. 49 The different underlying processes leading to hybridization have been under 50 investigated in the past, since this requires to focusing on the interacting populations of 51 wildcats and domestic cats at a local scale. To our knowledge, only one study combining 52 habitat with few interfaces between forests and villages in this area, we may expect 77 hybridization to be rare or inexistent in Pyrenean European wildcats. We did not expect 78 female natal philopatry when food resources are less abundant in the study area due to 79 the absence of agricultural fields. 80

Study areas and non-invasive sampling 83
In Northeastern France, the area of study covered approximately 400 km² (16). 84 The landscape is substantially fragmented (18) and consists of an alternation between 85 forests, agricultural fields and permanent grass with elevations ranging from 250-400m. 86 A total of sixteen villages (30 to 600 inhabitants per village) were in direct proximity 87 with the forest where wildcats were sampled. In the Pyrenees, the Nohèdes Nature 88 Reserve presents elevations ranging from 760-2,459m while the elevation of the Jujols 89 Nature Reserve ranges between 1,100 and 2,172m. The study area covers a total surface 90 of 325 km² of continuous forest (oak, maple, ash, pines, beech). These two nature 91 reserves are in proximity with ten villages ( was injected in each cat to aid subsequent identification. In addition, ten wildcat samples 101 were obtained on road-killed individuals. The fieldwork has been conducted by qualified 102 people according to current French legislation. Accreditation has been granted to the Pyrenean cats and Northeastern wildcats were genotyped using 31 autosomal 119 microsatellites. All cats were genotyped with a marker of sex. DNA extraction was 120 performed using a purification column kit (Nucleospin 96 Tissue kit, Macherey-Nagel) 121 following the manufacturer protocol. PCR reactions were performed step-by-step 122 following a unidirectional workflow starting in a clear room with positive air pressure 123 where sensitive reagents, enzymes and primers, were prepared. DNA and reagents were 124 then assembled in a pre-PCR room. PCR amplifications were made in 96-well 125 microplates in a post-PCR area with negative air pressure. The PCR reaction occurred in 126 a final volume of 10µl that contained 5µl of Mastermix Taq polymerase (Type-it, 127 QIAGEN), 1.35µl of primer pairs at a final concentration between 0.08 and 0.6µM, and 128 30ng of DNA. Each pair of primers was coupled with a fluorescent dye. The reaction 129 started with a denaturation step at 95°C for five minutes. This step was followed by 130 thirty PCR cycles (denaturation step = 95°C, 30s; annealing step = 55.9°C, 90s; 131 elongation step = 72°C, 30s) and a final elongation step at 60°C during 30 minutes. PCR 132 products were resolved on a capillary sequencer ABI PRISM 3130 XL (Applied 133 Biosystem) under denaturing conditions (formamide) and an internal size marker in one 134 migration for each multiplex. All these steps were performed using filtered tips. Finally repeat where the consensus genotype is obtained is rated with QI equal to 1, a repeat 145 with a homozygote genotype while the consensus genotype is heterozygote will have a 146 averaged over all repeats for each locus, and then over all loci for an individual to obtain 148 a QI per sample. Only individuals presenting a QI superior than 0.6 were included in the 149 subsequent analyses. 150

Consensus genotypes and population genetics analyses 151
Consensus genotypes were built as follows. Two genotypes were considered to 152 represent the same individual when (1) they were identical, (2) they only differed by 153 missing data and these missing data did not represent more than ten microsatellite 154 markers, (3) they only differed by missing data below the threshold of ten markers and a 155 single difference that could be explained by allelic dropout. Deviations of loci from both 156 Hardy-Weinberg equilibrium (HWE) and linkage equilibrium were both tested using 157 FSTAT v 2.9.

(20) with a Bonferroni correction and a 5% risk for all populations. Loci 158
showing a departure from HWE were discarded from the analysis. For each locus, the 159 frequency of null alleles was assessed following Brookfield's (21) method, and its impact 160 on a possible deviance from Hardy-Weinberg equilibrium tested using binomial tests 161 according to De Mêeus et al. (22). All loci exhibiting significant evidence of null alleles 162 causing HWE were discarded from further analyses. The software FSTAT v2.9.3.2. was 163 also used for estimating Weir and Cockerham's FST between wildcat and domestic cat 164 populations from the Pyrenees and from Northeastern France as well as allelic richness. 165 Expected (HE) and observed (HO) heterozygosities were calculated using GenALEx 6.501 166 (23). Finally, a discriminant analysis of principal components (DAPC, 24) was used in 167 order to visualize the differentiation between domestic and European wildcat 168 populations from northeastern France and the Pyrenees. 169

Spatial structure and relatedness 170
The 52 fresh fecal samples for which the sampling location was recorded were 171 displayed on a map using QGIS v2.8.1., together with the other indices of presence of 172 European wildcats (feces and photo trapping). The program ML-Relate (25) was used to 173 calculate pairwise relatedness between all wildcat individuals. Using a linear model, we 174 tested whether sex or relatedness were a significant predictor of the pairwise 175 geographical distance between individuals. Geographical distances between individuals 176 were calculated with QGIS v2.8.1. We considered the mean pairwise distance between 177 samplings for individuals sampled several times. Statistical analysis was performed in R 178

Admixture analysis 180
The Bayesian clustering method implemented in STRUCTURE v2.3.4 (26,27) was 181 used to identify wildcats, domestic cats and possible hybrids by applying the admixture 182 model with correlated allele frequencies. The optimal number of clusters K was 183 determined using the method described by Evanno et al. (28). STRUCTURE was used to 184 assess membership proportions (qi) to the inferred K clusters, which correspond to the 185 proportion of each individual's multilocus genotype belonging to each of the inferred K 186 clusters. The threshold level to differentiate wildcats and domestic cats from hybrids 187 was determined by selecting individuals which were believed to be representative of the 188 parental populations using the iterative algorithm described in Beugin

Genetic diversity and kinship pattern 257
We did not detect any linkage disequilibrium in any of the populations. On the 258 contrary, the northeastern population of domestic cats showed signs of Hardy-Weinberg 259 disequilibrium for six loci (Fca8, Fca45, Fca96, Fca229, Fca453) (Figure 2) confirmed the sharp distinction between the two European wildcat populations.  Figure S3). On the contrary, in Northeastern France, related females were captured significantly closer together than unrelated females (F = 6.88, df = 1, p = 0.0095; Supplementary Material, Figure S3).

Admixture analysis
Evanno's method showed that K=2 best described our data in both populations, one cluster corresponding to the domestic cats, the other corresponding to the European wildcats ( Figure 3). The iterative algorithm allowed us to define hybrids as individuals presenting a mean probability of assignment (conservative approach) or a lower bound of credibility interval (relaxed approach) below 0.79 for domestic cats, and below 0.83 for European wildcats in the Pyrenees; below 0.82 for domestic cats and below 0.89 for European wildcats in Northeastern France. . Each cat genotype is represented by a vertical bar split into K=2 colored sections, according to its relative assignment to the genetic cluster. The proportion of the bar in a given color represents the assignment probability of the individual for the corresponding cluster.
Using the conservative approach, we did not detect any hybrid and snapclust confirmed the absence of hybrids in the Pyrenees, while the relaxed approach detected one of the domestic cats sampled in Nohèdes as being hybrid and this individual detected as being hybrid by the relaxed approach was substantially assigned to the first-generation backcross category by snapclust (Supplementary material, Figure S4). The absence of gene flow from domestic cats to wildcats (m = 0.0148 -CI95: 0-0.042) as well as from wildcats to domestic cats (m = 0.0160 -CI95: 0-0.046) was consistent with the absence of hybridization between the two sub-species. On the contrary, in Northeastern France, hybrids were systematically detected, at least in the domestic cat population. Thus, using the conservative approach, two hybrids were detected in domestic cats and none in the European wildcats while sixteen and six hybrids were detected in the domestic cats and European wildcats respectively using the relaxed approach. All the individuals detected as hybrids with the relaxed approach were substantially assigned to the first-generation The primary interesting result is that males as well as females in close proximity are not kin related suggesting that both males and females disperse in this continuous forest landscape, i.e., related females did not tend to remain in the same area contrary to the wildcat population of northeastern France. The dispersal pattern may directly reflect the level of food resource availability. In fragmented environments as observed in northeastern France, with forest alternating with field crops, large areas rich in resources are available for wildcats (34,35). With carnivores, food distribution has been suggested to be the major determinant of species spatial distribution (36). The importance of resource distribution on the spacing pattern of wildcat females has already been proposed (37,38) and was supported by the study in northeastern France (16). Thus, although the European wildcat is acknowledged to live solitarily (39,40), its dispersal pattern may show more variability than has been described up to now.
The second important result is that no hybrid was categorically found neither in the Compared with estimations in the same area, the rate of hybridization we found in northeastern France is surprisingly low (13.8% against 25% on average, 17,42-44). A possible explanation for this difference may be linked to different sampling strategies.
While our sampling is local, focused on one population of living wildcats and on neighboring populations of living domestic cats, previous studies in France relied upon opportunistic sampling of road-killed animals, over a much larger area. Germain et al. (42,43) showed that hybrids tend to live in intermediary environments, between forests and villages. This would expose them to road mortality more often than wildcats living in the forest or domestic cats in the villages and thus, sampling schemes based on road-killed animals may be biased towards a higher proportion of hybrids.
The absence of hybrids in the continuous Pyrenean forest landscape may imply that the absence of crops frequented by both wild and domestic cats for hunting purposes may impede encounters between the two subspecies as domestic cats, even feral domestic cats, do not enter the forest environment (during the last seven years no pictures of domestic cats have been taken by camera trapping). Unfortunately, the sample size of this study was limited and more extensive studies will be required to confirm the absence of hybridization in this environmental context. Finally, the Pyrenean wildcats showed values of genetic diversity lower than other wildcat or domestic cat populations in France and in Europe (4.84 alleles per locus on average in the Pyrenean wildcat population while between 3 and 11.8 can be found in the literature with rare populations below 6; ref 13,17,32,44,45), suggesting that wildcats in the Pyrenees may be threatened by a lack of genetic diversity. Not surprisingly, the genetic differentiation between Pyreneans domestic cats and their northeastern counterparts was moderate (FST =0.04). In contrast, the Pyrenean wildcat population was significantly differentiated from the northeastern wildcats with a higher FST value (0.08). This genetic divergence could result from a classic genetic process of isolation by distance (IBD, 46), which supposes continuity in the distribution of the European wildcat on French territory; such a cline has been described in the northeastern wildcat population (17,47). Conservation strategies of wildcats should take into account the local habitat such as the existence of a fragmented or continuous forest environment, and the presence of agricultural fields (41). Further investigation should also focus on the spatial distribution of French wildcats -in particular we need to confirm whether the French Pyrenean population is isolated from the main area of wildcats in France but connected to the Spanish Pyrenean wildcat population. Depending upon the answer, the usefulness of wildlife corridors to enhance connectivity between the different wildcat populations should be addressed to ensure the long-term viability of the French Pyrenean wildcat population.
observations. We also thank Robin Buckland for his careful check of English language. MPB, DP are supported by the LabEx ECOFECT (ANR-11-LABX-0048). Figure S1: Distribution of the quality indexes of the samples included in the analyses.