Institut für Rebenzüchtung
One of the tasks of grapevine breeding is to contribute to the preservation of viticultural cultural landscapes by developing robust and climate-adapted varieties. New grape varieties must have a strong resistance to pathogens, pests, and abiotic factors such as heat, while maintaining consistently high wine quality. During the breeding process, molecular markers can already be applied for early selection of resistance to diseases such as powdery and downy mildew. Additionally, further morphological and physiological traits, particularly viticultural characteristics and resilience to new biotic and climate-induced abiotic stressors, must be assessed in the field. The institute's extensive breeding material is currently evaluated manually concerning key breeding traits such as resistance, yield, vitality, and wood maturity. To support selection in the breeding process and to screen genetic resources, high-throughput phenotyping pipelines are being developed, validated, and implemented. These pipelines enable automated and objective trait assessments using field-based sensor technology. Custom AI models are being developed and applied for image data analysis. A field phenotyping platform is available at ZR for sensor data acquisition. The PHENOquad is equipped with a 5-channel prism camera (RGB, 2x NIR), GPS, and illumination (Engler et al. 2023). The DigiVine, KI-iREPro, and PhytoMo projects have already developed a prototype image analysis pipeline. Individual AI-supported image analysis algorithms for the traits of vigor, phytoplasma infection, and yield were developed and validated on a small test set of images. The platform proved superior to its predecessors PHENObot (Kicherer et al. 2015) and PHENOliner (Kicherer et al. 2017), particularly with regard to its acquisition speed of 4 km/h. Another advantage is the 5-prism camera, which delivers a 100% superimposed image in all 5 color channels, thus significantly increasing the amount of information available. At the end of the project, the Vineyard Viewer database will be available for storing and visualizing the sensor data. The PHENOquad is a platform established at JKI-ZR and can be used for image data acquisition in this project. In addition to implementing new traits into the PHENOquad pipeline, we also need to develop how these traits can be integrated into the selection process for breeding. The project aims to enhance breeding efficiency by enabling faster, non-destructive, and more objective evaluation of breeding material.
Federal Ministry of Agriculture, Food and Regional Identity
