How reliable are morphological and anatomical characters to distinguish European wildcats,domestic cats and their hybrids in France?

Phenotypic variation in hybridizing species or subspecies is a prerequisite for allowing conservation ecologists and wildlife managers to identify parental populations and their hybrids in the field. We assessed the reliability of a set of eight morphological (body size and pelage characters) and four anatomical criteria (skull and intestine morphometric measurements) to distinguish between 302 French specimens classified as wildcat, domestic cat or hybrid on the basis of a Bayesian analysis (STRUCTURE) of their multilocus microsatellite genotypes. This aim was achieved by performing a set of multivariate analyses on morphological, anatomical and genetic data sets (Hill and Smith's analysis, co‐inertia analysis and discriminant analysis of principal components). Wildcats and domestic cats were very satisfactorily distinguished, even when using simple non‐invasive morphological criteria easily usable in the field like the morphology of the tail, dorsal line or flank stripes. Using anatomical instead of morphological characters slightly increased the discriminating power. Many more difficulties arose when we tried to distinguish hybrid specimens from both wildcat and domestic ones. Anatomical characters performed better than morphological ones in recognizing hybrids, but the assignment success rate remained very low, about 31.6% and 1.5%, respectively. Overall, the most discriminating characters were two continuous, derived anatomical characters: the cranial index followed by the intestinal index. Classification of specimens in three classes based on their microsatellite genotypes appeared to be inadequate for identifying hybrid specimens, as hybrid specimens seemed to be distributed along an anatomical continuum. With this observation in mind, we assessed the linear relationships between a proxy of the individual level of hybridization (q_ik) and the cranial and intestinal indices, respectively. Both relationships were highly significant. The greatest correlation was found with the cranial index (R² = 60.4%). Altogether, our results suggest that future work should be geared towards enhancing the measure of hybridization using more discriminating molecular markers and improving morphometric skull measurements through the use of modern geometric morphometric methods, using landmarks rather than skull dimension.