Sensory and chemical phenotyping of wines from a F1 grapevine population

Abstract

The European Green Deal, a concept of the European Commission, aims at the reduction of pesticides in EU agriculture for 2030 by 50%. Viticulture uses the largest amounts of fungicides in the EU, compared to other crops such as grains. In order to achieve the ambitious target of 50% pesticide reduction in viticulture, the increased cultivation of new pathogen-resistant grape varieties is indispensable. New pathogen-resistant grape varieties, which have been selected for their high quality potential, allow up to 80% less fungicide use. These varieties are therefore an important building block in the transformation process to more sustainable viticulture. The project Predictive Breeding for Wine Quality »SelWineQ« (Select Wine Quality) focuses on the development of robust predictive models for the genetic quality potential (GQP) of grapevine varieties during the breeding process based on sensory, metabolomic, and genomic data. Predictive models for wine quality traits will considerably increase the efficiency of grapevine breeding. The centerpiece of the “SelWineQ” project is an F1 breeding population of Calardis Musqué and Villiard blanc consisting of 150 genotypes (8 vines each). Over three vintages experimental wines of each genotype were made. Every year a professional trained panel evaluated the wines of all genotypes. This sensory evaluation forms a broad data basis for modeling sensory quality traits from genetic and metabolic data. One of the most important results from the sensory evaluation is the “Total Quality Score”, a sum parameter for the olfactory and gustatory total quality of the wines. This quality parameter was found to be constant for the best and worst wines of the breeding population over several years. Thus, the best and worst wines could be reproducibly identified. This result shows, besides an excellent panel performance, that the quality potential is mainly determined by the genetic properties of the plants and that environmental influences (different vintages) are less important. The combination of analytical data and data from the sensory evaluation facilitated the identification of linalool and cis-rose oxide (among other terpenoids) as molecular quality markers. These aroma-active compounds were present in the best evaluated wines far above their olfactory threshold and showed a high correlation (r > 0.7 Pearson) with the attribute “floral”. Moreover, metabolomic data from non-targeted LC-HRMS and GC-MS analysis allowed predictions of the best and worst genotypes from one to the other vintage (model building on one vintage, validation on another vintage). These findings form a solid base for the development, improvement and validation of predictive models based on genetic data. A novel genotyping by sequencing approach lead to a full informative genetic map of the breeding population based on SNP markers.