A multivariate Bayesian modeling strategy coupled with QTL analysis reveals genetic loci linking important sensory...

Wine quality assessment is the most time-consuming and labor-intensive task in grapevine breeding. Breeding programs worldwide would benefit by accelerating this process, regardless of their primary objectives, such as climate adaptation or disease resistance, because wine quality ultimately determines consumer acceptance. In this study, an F1 mapping population derived from 'Calardis Musqué' × 'Villard Blanc', segregating for perceived wine quality, was comprehensively characterized for key aroma-relevant metabolites and sensory attributes. A total of 25,284 sensory evaluations were conducted on 1,635 micro-vinified wine samples, which were obtained from 147 genotypes grown at two locations over six vintages. A multivariate Bayesian multilevel modeling approach was applied to disentangle the effects of genotype, environment, and individual taster subjectivity on eight sensory attributes and the total quality score (TQS). Quantitative trait locus (QTL) analysis based on a dense genetic map was performed using the modeled sensory phenotypes as well as quantification data of volatile monoterpenes acquired with a targeted SIDA-SPE-GC-MS. This integrative approach identified overlapping genetic loci linking aroma-active metabolites with their sensory quality attributes. Two major pleiotropic QTL clusters associated with wine quality were identified. The first, located on chromosome 2, simultaneously controlled all major quantified monoterpenes (e.g., linalool: LOD 10.1, PVE 19.7%) and is co-localized with QTLs for the most discriminating sensory attributes, including "floral" (LOD 14.8, PVE 14.3%), "fruity" (LOD 14.4, PVE 17.2%), and "tropical fruit" (LOD 13.1, PVE 16.3%). A second major QTL cluster on chromosome 11 harbored strong QTLs for terpene concentrations as well as quality-related aroma attributes (e.g., floral: LOD 11.1, PVE 10.9%). In addition, QTLs for TQS and the attribute "off-flavor" were identified on several chromosomes, indicating further genomic targets relevant for breeding on wine quality. This integrated workflow combines sensory phenotyping with genotype × environment modeling and estimation of taster bias for subsequent genomic analysis. The approach represents a significant advancement in grapevine quality assessment and provides a framework to accelerate the breeding of new grapevine cultivars that are agronomically valuable, consumer-accepted, and both ecologically and economically sustainable.