A genetic linkage map of grapevine was constructed using a pseudo-testcross strategy based upon 138 individuals derived from a cross of Vitis vinifera Cabernet Sauvignon x Vitis riparia Gloire de Montpellier. A total of 212 DNA markers including 199 single sequence repeats (SSRs), 11 single strand conformation polymorphisms (SSCPs) and two morphological markers were mapped onto 19 linkage groups (LG) which covered 1,249 cM with an average of 6.7 cM between markers. The position of SSR loci in the maps presented here is consistent with the genome sequence. Quantitative traits loci (QTLs) for several traits of inflorescence and flower morphology, and downy mildew resistance were investigated. Two novel QTLs for downy mildew resistance were mapped on linkage groups 9 and 12, they explain 26.0-34.4 and 28.9-31.5% of total variance, respectively. QTLs for inflorescence morphology with a large effect (14-70% of total variance explained) were detected close to the Sex locus on LG 2. The gene of the enzyme 1-aminocyclopropane-1-carboxylic acid synthase, involved in melon male organ development and located in the confidence interval of all QTLs detected on the LG 2, could be considered as a putative candidate gene for the control of sexual traits in grapevine. Co-localisations were found between four QTLs, detected on linkage groups 1, 14, 17 and 18, and the position of the floral organ development genes GIBBERELLIN INSENSITIVE1, FRUITFULL, LEAFY and AGAMOUS. Our results demonstrate that the sex determinism locus also determines both flower and inflorescence morphological traits.

We explored single nucleotide polymorphism (SNP) variation in candidate genes for bud burst from Quercus petraea populations sampled along gradients of latitude and altitude in Western Europe. SNP diversity was monitored for 106 candidate genes, in 758 individuals from 32 natural populations. We investigated whether SNP variation reflected the clinal pattern of bud burst observed in common garden experiments. We used different methods to detect imprints of natural selection (F ST outlier, clinal variation at allelic frequencies, association tests) and compared the results obtained for the two gradients. F ST outlier SNPs were found in 15 genes, 5 of which were common to both gradients. The type of selection differed between the two gradients (directional or balancing) for 3 of these 5. Clinal variations were observed for six SNPs, and one cline was conserved across both gradients. Association tests between the phenotypic or breeding values of trees and SNP genotypes identified 14 significant associations, involving 12 genes. The results of outlier detection on the basis of population differentiation or clinal variation were not very consistent with the results of association tests. The discrepancies between these approaches may reflect the different hierarchical levels of selection considered (inter-and intrapopulation selection). Finally, we obtained evidence for convergent selection (similar for gradients) and clinal variation for a few genes, suggesting that comparisons between parallel gradients could be used to screen for major candidate genes responding to natural selection in trees.L OCAL adaptation of temperate forest trees to environmental variation has been widely investigated in common garden experiments or provenance tests (Langlet 1971;Matyas 1996; König 2005 for reviews). Comparisons of populations sampled over wide ranges have revealed continuous patterns of differentiation along environmental gradients (Wright 1976;Morgenstern 1996). Clinal variation along geographic gradients (latitude, altitude, longitude) is frequently reported for phenological traits (see Alberto et al. 2013 for a review). Congruent clines have been identified across species and across experiments within species, suggesting the occurrence of directional selection in response to identical environmental gradients. For example, in northern temperate spruces and pines, populations from northern latitudes flush earlier than southern populations (Wright 1976;Morgenstern 1996) and steep clines have been observed in species extending over large ranges of latitude (Pinus sylvestris in Scandinavia, Giertych 1991; Picea sitchensis in North America, Mimura and Aitken 2007). Bud burst clines have also been observed in northern temperate broadleaved trees, but following the opposite trend. In beech (von Wuehlisch et al. 1995), birch (Worrell et al. 2000), and sessile oak (Liepe 1993;Deans and Harvey 1995;Ducousso et al. 1996), southern populations flush earlier than northern populations. Similar clinal variation has also been repo...

Loci considered to be under selection are generally avoided in attempts to infer past demographic processes as they do not fit neutral model assumptions. However, opportunities to better reconstruct some aspects of past demography might thus be missed. Here we examined genetic differentiation between two sympatric European oak species with contrasting ecological dynamics (Quercus robur and Quercus petraea) with both outlier (i.e. loci possibly affected by divergent selection between species or by hitchhiking effects with genomic regions under selection) and nonoutlier loci. We sampled 855 individuals in six mixed forests in France and genotyped them with a set of 262 SNPs enriched with markers showing high interspecific differentiation, resulting in accurate species delimitation. We identified between 13 and 74 interspecific outlier loci, depending on the coalescent simulation models and parameters used. Greater genetic diversity was predicted in Q. petraea (a late-successional species) than in Q. robur (an early successional species) as introgression should theoretically occur predominantly from the resident species to the invading species. Remarkably, this prediction was verified with outlier loci but not with nonoutlier loci. We suggest that the lower effective interspecific gene flow at loci showing high interspecific divergence has better preserved the signal of past asymmetric introgression towards Q. petraea caused by the species' contrasting dynamics. Using markers under selection to reconstruct past demographic processes could therefore have broader potential than generally recognized.