Less effort, improved results: the potential of sensor fusion to optimize grape cluster phenotyping

Abstract

Sustainability and the consequences of climate change challenging today’s viticulture resulted in a high demand for new disease resistant varieties (DRVs). However, breeding of DRVs with multiple resistances to biotic and abiotic stresses and high wine quality is a time-intensive task. An expanded availability of molecular markers especially for complex traits like grape cluster architecture or berry quality might accelerate the breeding of DRVs significantly and improve breeding efficiency. The application of user-friendly, preferably non-invasive, sensor techniques has proven them to be promising tools for facilitating high-throughput phenotyping of large biparental populations. This, in turn, enabled the detection of new Quantitative Trait Loci (QTLs), the base for the development of trait-linked molecular markers. However, most of the established user-friendly methods focus on one specific trait like grape cluster architecture or sugar content by using one sensor modality like 3D, Red-Green-Blue (RGB) or Near-Infrared (NIR) imaging. We aim to overcome the need for individual sensor-based methods by fusing optical and spectral sensors for combined phenotyping of geometric and spectral grape cluster traits using a single handheld tool for both laboratory and field applications. The major challenges from a biological point of view are: (i) the reliable linkage of normalized spectral reflectance at specific wavelength bands with analytical and chemical ground truth, such as phenolic compounds, acidity, or cuticle waxes; (ii) reproducibility of results in different years and material; as well as (iii) data analysis and prediction model development. Finally, the connection of data from up to twelve spectral bands fused with geometry will support fast assessments of berry ripening and berry skin traits as well as the detection of stress indicators for phenotyping of DRVs, mapping populations or genetic resources with less effort and improved outcomes.