Hyperspectral imaging and machine learning for monitoring grapevine physiology

Developing a tractable genetic engineering and gene editing system is an essential tool for grapevine. We initiated a plant transformation and biotechnology program at Oregon State University using the grape microvine system (V. vinifera) in 2018 to interrogate gene-to-trait relationships using traditional genetic engineering and gene editing. The microvine model is also used for nanomaterial-assisted RNP, DNA, and RNA delivery. Most reference genomes and annotations for grapevine are collapsed assemblies of homologous chromosomes and do not represent the specific microvine cultivar ‘043023V004’ under study at our institution.

Berry shriveling (BS) in vineyards are caused by numerous factors such as sunburn, dehydration, stem necrosis. Climate change results in an increase in day and night temperatures, rainfall throughout the year, changes in the timing and quantities, long dry summers and a combination of climatic variability such as floods, droughts and heatwaves). Grape development and its composition at harvest is influenced by the latter as grape metabolites are sensitive to the environmental conditions. The grape berry experiences water loss and an increase in flavour development as a result of the BS. An increased sugar content in grapes will result in higher alcohol wines and concentration of grape aromas which may be detrimental to the final wine quality.

During red wine vinification, maceration allows the must, and consequently the wine, to be enriched with several compounds that contribute to the creation of the typical organoleptic characteristics of red wines. Among these, extraction of polyphenols (PPs) during maceration is a major process of enological interest.
The purpose of this study was the evaluate the suitability of a rapid analytical approach based in linear sweep voltammetry to monitor PPs extraction during vinification.

Many farming operations aim at controlling the leaf area of the vine according to its load. There are several techniques, direct and indirect, of estimate of this leaf area in a specific way, but impossible to implement at great scales. These last years, research in airborne and satellite remote sensing made it possible to show that a multispectral index of vegetation, computed from measurements of reflectances (red and near infrared), the « Normalised Difference Vegetation Index » (NDVI), is well correlated to the « Leaf Area Index » (leaf area per unit of ground) of the vine. Nevertheless these methods of acquisition and processing data are rather constraining and complex.

The effect of cover crop on the water relations, yield and grape composition of Pinot noir vines was investigated during two seasons (2003 and 2004) in a gravely soil located in Tarragona (Spain). Seventeen-year-old vines, grafted onto R110 and trained onto a Ballerina training system, were used.