Predicting provenance and grapevine cultivar implementing machine learning on vineyard soil microbiome data: implications in grapevine breeding

Antagonistic yeasts applied to wine grapes must be compatible with the thereafter winemaking process, avoiding competition with the fermentative Saccharomyces cerevisiae or affecting wine flavour. Therefore, fifteen epiphytic yeasts (6 Metschnikowia sp., 6 Hanseniaspora uvarum, 3 Starmerella bacillaris) previously selected for its biocontrol ability against Alternaria on wine grapes were evaluate for possible competition with S. cerevisiae by the Niche Overlap Index (NOI) employing YNB agar media with 10 mM of 17 different carbonate sources present in wine grapes (proline, asparagine, alanine, glutamic acid, tirosine, arginine, lisine, methionine, glicine, malic acid, tartaric acid, fructose, melibiose, raffinose, rhamnose, sucrose, glucose).

In the Bordeaux vineyards, press wine represents approximately 15% of the total volume of wine produced[1]. Valuing this large volume of wine is necessary from an economic point of view, but also because of their organoleptic contribution to the blend, and their contribution to the construction of wines for laying down. Therefore, this study was developed considering the lack of recent scientific knowledge on the composition of red press wines. The aim of this study is to establish an initial assessment of their chemical composition including aromatic compounds and a phenolic part.

By dissecting the root system architecture (RSA) into its underpinning components (e.g. root emission, axial growth, radial growth, branching, root direction or tropism) and identifying the relationships between them, functional-structural 3D root models are promising tools for analyzing the diversity and complexity of root system phenotypes with Genotype × Environment interactions. The model parameters are assumed to be synthetic traits, less influenced by the environment, and consequently with less polygenic architectures than the integrative RSA traits they drive. Root models can serve as a basis for in silico development of root system ideotypes by highlighting the developmental processes and parameters that most likely influence RSA fitness.

Targeted analytical methods can overlook compounds that are a priori unknown to play a role in the mouthfeel sensations. This limitation can be overcome with the information provided by untargeted metabolomic analysis using UPLC‐QTOF-MS. To this end, an untargeted metabolomic approach applied to 42 red wines has allowed development of a model with predictive capacity by cross-validation for the “dry”, “oily” and “unctuous” sensations perceived by a sensory panel. The optimal PLS model for “dry” retained compounds with positive regression coefficients (≥ 0.17) including a trimer procyanidin, a peptide, and four anthocyanins.

During red wine fermentation, the extraction of phenolics compounds and sufficient oxygen provision are critical for wine quality [1,2]. In this trial, we aimed at evaluating the kinetics of phenolic extraction and dissolved oxygen during red wine fermentations using the airmixing system. Twenty lots of red grape musts were fermented in 300.000 L tanks, equipped with airmixing, using two injection regimes (i.e., high and low intensity, and high and low daily frequency). An oxygen analyzer was introduced into the tanks in order to record the concentration of dissolved oxygen over time.