A novel dataset and deep learning object detection benchmark for grapevine pest surveillance

In Portugal, and particularly in the Demarcated Region of Douro (DDR), wine production has a great tradition, producing the unique and worldwide famous Port wine as well as other remarkably good table wines. In this study the impact of projected climate change to wine production is analysed for the DDR. A statistical grapevine yield model (GYM) is developed using climate parameters as predictors.

Downy mildew, caused by Plasmopara viticola, is the most economically impactful disease affecting grapevines. This polycyclic pathogen triggers both primary and secondary infection cycles, resulting in significant yield losses when effective disease control measures are lacking. Over the winter, the pathogen survives by forming resting structures, the oospores, derived from sexual reproduction, which produce the inoculum for primary infections. To optimize grapevine downy mildew control and obtain the desired levels of production while minimizing chemical inputs, it is crucial to optimize the timeframe for fungicide application. Disease forecasting models are useful to identify the infection risk.

The ongoing climate change results in increasing mean air temperature, which is manifested by weather extremes or sudden changes between drought and local heavy rainfalls. These changing conditions are especially challenging for the established grapevine varieties growing under cool climate conditions due to an increased biotic infection pressure. Thus, the scope of most grapevine breeding programs is the selection of mildew fungus-resistant and climatic adapted grapevines with balanced, healthy yield and outstanding wine quality. Since no resistances or candidate genes have yet been described for Botrytis bunch rot (BBR), physical-mechanical traits like berry size and thick, impermeable berry cuticles phenotyped with high-throughput sensors represent novel effective parameters to predict BBR.

The use of pectinolytic enzymes in winemaking is state of the art. These enzymes catalyse the degradation of pectic substances through depolymerization (hydrolases and lyases) and de-esterification. As a result, it supports the extraction of juice and facilitates filtration. It has also been shown in winemaking that the presence of pectinolytic enzymes improves the stability, taste, texture, colour and aroma of products. With regard to enzymes currently applied in winemaking, enzymes derived from filamentous fungi dominate the enzyme industry. Fungal-based pectinolytic enzymes specifically require purification from the culture medium to eliminate unwanted side reactions, which is poorly sustainable. Some non-traditional yeast strains have been reported to exhibit pectinolytic activities. Therefore, the direct use of pectinolytic yeast during wine fermentation process can be an attractive and alternative source for the use of enzymes as input.

Press wines play a crucial role in red winemaking, representing up to 15% of the final blend [1]. Optimizing their value is essential both economically and for maintaining wine identity, especially given evolving climatic and societal challenges. However, little recent research exists on their composition.