A Computational Approach for Balancing Competing Objectives in Winegrape Production

Abstract

Operational field targets, such as leaf layer numbers or leaf-area-to-fruit-weight ratios, are often used to guide cultural practices to achieve particular production objectives, including reduced disease pressure, increased yield, and improved fruit composition. However, these targets are not customized to individual producers nor do they account for the need to balance potentially conflicting objectives, such as simultaneously increasing fruit-zone sunlight exposure to reduce disease pressure while minimizing fruit exposure to avoid sunburn or the development of undesirable aroma compounds. The approach here balances multiple production objectives by adapting a multiobjective optimization method known as nondominated sorting genetic algorithm. The method was demonstrated in a hypothetical situation where a grower has competing objectives of reducing 1,1,6-trimethyl-1,2-dihydronaphthalene (TDN) precursors in Riesling while concurrently minimizing the amount of active ingredient used in an existing spray program. The demonstration model used two previously quantified responses for cluster exposure flux availability (CEFA) versus TDN precursor concentrations, and one previously quantified response for CEFA versus fruit spray residue. Optimal targets for CEFA and fruit spray inputs were computed that balance minimal TDN potential versus minimal spray active ingredient. A second analysis demonstrated how evolving knowledge of field responses can be used to update vineyard targets to maintain an optimal balance among production objectives.