One of the most important issues in evolutionary biology is the genetic basis of evolutionary novelties [1, 2], and it is now clear that hybridization phenomena could induce significant contribution . Indeed, hybridization is a common phenomenon in animals in general [4, 5] and in fish in particular [6–10] but the part played by such phenomenon in the emergence of new genomic architectures and phenotypes in admixed individuals remains unclear .
Hybridization between divergent gene pools generates recombinant individuals. Analysis of such individuals provides information about genomic architecture associated with reproductive isolation between populations or species . Both genetic factors contributing to genomic compatibility (endogenous selection) and environmental factors sorting specimens based on their fitness (exogenous selection) apply in zones where hybridization occurs naturally (called hybrid zones hereafter). The remoulding of genomic regions in hybrid zones may serve as a source of evolutionary novelties and these zones behave as biodiversity reactors generating waves of new genotypes. Selection may thus act on a vast number of genomic combinations from parental species to different hybrid genotypes; such hybrid genotypes may also serve as intermediaries for the transfer of adaptive genetic variation between parental populations [13–15]. Moreover, some authors have demonstrated that this phenomenon is possible in the early stages of the hybridization process [16–19].
Mathematical models are increasingly used to study the complexity of such biological phenomena. As reviewed in Payseur , most models used to analyse hybrid zones describe the relationship between allele frequency and geography. The clines in hybrid zones take a sigmoid shape to model the transition between species gradients. Several variables are used in these models (reviewed in ): the migration of the two species along the zone where clinal variation is observed (i.e. diffusion flattens the cline over space), the strength of selection (that maintains the sigmoid form) and the geographical transect of the observed alleles. Nolte et al. pointed out that the analyses of such phenomenon need diagnostic loci to allow the two parental populations to be distinguished, and that the exclusion of loci that are not diagnostic can remove a part of the species polymorphism from the analysis.
Another approach is to use hybrid ancestries across the genome to predict introgression at individual loci. Pritchard et al. developed a model-based clustering method for using multilocus data to infer population structure and assign individuals to populations. An alternative approach is based on multinomial logistic modelling [23, 24]. The aim is to estimate the genomic cline as the genotype frequency at individual loci along a genomic admixture gradient (hybrid index = h ). A statistical test has been developed to identify markers that deviate from expectation (neutrality) based on a genome-wide admixture, and that can handle potential locus-specific selection .
Currently, it is not clear which evolutionary forces constrain the genomic architecture generated by hybridization process such that it “stabilizes” a new lineage. One of the best known examples in fish is that of the invasive hybrid lineages of Cottus in Europe. Nolte et al. studied two hybrid zones involving Cottus perifretum and Cottus rhenanus using the genomic cline approach. Although the two species share similar overall genomic compositions, the observed patterns at individual loci differed substantially between zones indicating differences in external selection pressures or cryptic genetic differentiation of distinct parental populations. Recently, Stemshorn et al. identified three distinct hybrid lineages, which have emerged out of a situation of secondary contact between C. rhenanus and C. perifretum. The examination of partially isolated lineages, such as invasive hybrid sculpins, may allow early adaptive genetic changes to be identified before they become confounded by differences arising due to speciation process.
Another particularity in hybrid zones concerns phenotypes of recombinant individuals. In the Cottus sp. complex for example, Nolte and Sheets  found a specific hybrid shape that was intermediate along the axes separating their parental groups, but that also displayed additional differentiation. In the Chondrostoma species complex, convergence of body shape and coefficient condition between the two species have been described for this model despite heterogeneous genetic patterns (i.e. different hybrid combinations and parental individuals). Corse et al. reported that the mouths of the F1 hybrids display an extreme phenotype resulting from the lower lip widening slower than in the two parent species. Phenotypic diversity of F1 specimens is of particular interest, because the phenotypes of these individuals in some cases exceed the range of phenotypes in the corresponding parental lineages demonstrating transgressive segregation for the first time in backcrossed individuals [27, 28]. This new character range constitutes a source of evolutionary novelties (i.e. transgressive segregation) on which selection could act. Thus, hybridization may contribute to evolutionary novelties in animal through the emergence of novel phenotypes due to transgressive segregation [3, 26].
It would be valuable to be able to identify adaptive peaks or sets of pathways that spread in the adaptive landscape, and to describe the interactions between new genomic architectures and observed phenotypes in hybrid zones. To decipher the complex evolutionary history linking genotype and phenotypes in hybrid zones, two major conditions need to be respected. First, the genetic diversity of source populations has to be estimated as this is an important factor of morphological variation among admixed populations . Second, it is important to model phenotypic variability among admixed individuals; such variability can change significantly within a few generations in a particularly dynamic process . The change in variability depends on the distribution of admixed individuals in the hybrid zone, the environmental parameters, time and duration of hybridization, and the level of gene flow . It would therefore be of interest to evaluate transgressive segregation in natural systems including in particular a recent hybrid zone. Indeed, an ancestral hybrid zone or a lineage of hybrid origin may be subject to secondary evolutionary processes that reshape the hybrid morphology and may have hidden the transgressive traits.
The Chondrostoma species complex appears to be a useful model for investigating how admixture between two divergent lineages (Chondrostoma species and another species belonging to a different genus) shapes morphological variation. Chondrostoma hybrid zones have been previously described in a fragmented habitat (the Durance River), assuming that dams favour hybridization between Parachondrostoma toxostoma and Chondrostoma nasus. With this model, we can identify source populations for each species, recent hybrid zones and it is possible to genotype large numbers of new hybrid specimens from this habitat.
The aim of this study was to describe genetic-phenotypic interactions and to assess the transgressive traits in body shape and thereby to decipher the evolutionary dynamic of this “natural” Chondrostoma hybrid zone. Recently, intermediate individuals were found in the Ardèche River (a non-fragmented river) raising questions about the capacity of such zone to generate large numbers (and diversity) of hybrids. We report a study of 41 microsatellites, partial cytochrome b gene sequences and body shape morphology in 970 specimens from (i) a non-fragmented hybrid zone and (ii) source populations including allopatry and parapatry for both species (Parachondrostoma toxostoma and Chondrostoma nasus). To study interactions between morphological changes and genetic architecture in hybrid zones, we combined clustering methods  and the genomic cline approach  to explain genotype as a function of hybrid index and test potential effects of selection. The findings were then used to describe body shape transformation (identified by a morphometric analysis) along the h-index gradient. Finally, we interpret morphological variations linked to hybridization, taking into account ontogenetic effects on shape.