The use of models at field and farm levels for the ex ante assessment of new pea genotypes

Abstract The use of models for genotype assessment has increased greatly in the last 10–15 years, either to assist the choice of the best cultivar for growing conditions, or to identify virtual genotypes as targets for breeding. Most studies are at the plant and field scales. However, restrictions due to work organisation at the farm level can affect genotype performance. The objective of our study was to propose a method to analyse the effects of farm constraints on crop production in the context of genotype × management × environment (G × M × E) interactions. Three types of pea genotypes, with their optimal sowing times, were compared: classic winter types sown at the end of autumn, spring sown types, and newly developed Hr types sown at the beginning of autumn. Their grain yields were simulated combining three models: a pea crop model sensitive to the main stresses affecting pea yield, a model simulating the change in soil structure according to the operations performed on the field, and a work organisation model simulating the dates of the main operations (soil tillage, sowing, harvest) at the farm scale. Simulations considering the constraints of farm work organisation showed that soil compaction was more frequent and sowing date more variable in the case of winter and spring peas than in the case of the new Hr pea genotype. This increased the risk of yield loss and the year-to-year variability of grain yield. Variety performances were thus modified taking into account farm constraints. The consideration of farm work organisation in combination with soil modelling allowed a wider variety of growing conditions to be explored than is usually the case for breeding trials on experimental stations. Generalizing the proposed method based on coupling models at farm level could thus help to define the traits favourable for innovative genotypes adapted to farmers’ conditions at no extra experimental cost.