Prediction-based approaches for grapevine improvement

Abstract

Grapevine breeding is fundamentally constrained by long generation times, perennial growth and clonal propagation. These characteristics limit the rate of genetic gain and make selection decisions both critical and uncertain.

Advances in genomics, phenomics and quantitative genetics are creating new opportunities to accelerate breeding progress via prediction-based approaches. The increasing availability of large-scale and multi-environment datasets is offer the possibility of shifting grapevine breeding from descriptive data generation towards prediction-driven decision making.

This presentation will discuss how predictive approaches may reshape grapevine improvement across different stages of breeding programs. Examples from genomics, phenomics, clonal evaluation and genotype-by-environment modelling will illustrate how increasingly integrated datasets enable new forms of biological inference and predictive modelling in grapevine.

The talk will further explore concepts and emerging modelling frameworks designed to improve information transfer across populations, environments and breeding cycles. Perspectives on scalable prediction systems and new computational approaches will also be discussed. The central message is that future progress in grapevine breeding will depend not only on generating increasingly complex datasets but on integrating and transforming these data into robust and continuously improving prediction systems.