A MODELLING APPROACH TO GENOTYPE × ENVIRONMENT INTERACTION Genetic analysis of the response of maize growth to environmental conditions

Expansive growth of organs has a very large genotype × environment (G×E) interaction. Maximum leaf expansion rate observed in the absence of stress and of evaporative demand has a genetic variability which is usually smaller than environmental effects. The mechanisms driving the reduction in leaf growth rate under stress, namely changes in cell division rate, in cell-wall mechanical properties and in turgor, and their signalling pathways, interact in such a way that a bottom-up approach from genes to the G×E interaction cannot be envisaged. We propose an approach combining modelling and genetic dissection of model parameters. Three genotype-dependent parameters are considered for analysing the G×E interaction for leaf elongation rate of maize. The maximum leaf elongation rate per unit thermal time is stable during the night and over several nights, and it is repeatable for each genotype over several experiments. The responses of leaf elongation rate to evaporative demand and soil water status are linear and their slopes are reproducible over several experiments. Maximum elongation rate and slopes of the responses to evaporative demand and to soil water potential have been analysed genetically in three mapping populations. QTLs of maximum leaf elongation rate tended to co-localize with QTLs of leaf length under well-watered conditions, but also under water deficit. They also co-localized with QTLs of the Anthesis Silking Interval (ASI). In contrast, QTLs of response parameters did not co-localize with QTLs of length under water deficit. They are therefore ‘adaptive’ traits which cannot be identified otherwise. Each parameter of the ecophysiological model was computed as the sum of QTL effects, allowing calculation of parameters of new RILs known by their allelic values only. Leaf elongation rates of these new RILs were simulated and were similar to measurements in a growth-chamber experiment. This opens the way to the simulation of virtual genotypes, known only by their alleles, in any climatic scenario.