Acquisition de données sur terrain en vue de tester la capacité des modèles de végétation à prédire les différences de productivité entre variétés d'espèces cultivées

Abstract

The context of climate change poses many questions on the fate of plant species and particularly on crop species. Mathematical models applied to cultures are one way of studying these problems. Among these models, dynamic vegetation models (MDV) are the tools of choice because they simulate the biological, chemical or physical processes that take place in ecosystems. However, their capacities are linked to the possibilities of parameterization: they must be correctly parameterized. Functional traits allow specifying the peculiarities of species. However, these traits vary in species depending on environmental conditions. It is also established that traits can vary between cultivars and varieties of the same species. Our objective was to compare the performance of a MDV and another specific crop model by applying them to experimental cereal crops carried out in the Grand Duchy of Luxembourg. We created a database of 4 traits (specific leaf area (SLA), nitrogen concentration of stems (% N stem) and leaves (% N leaf), minimum stomatal conductance (g0)) in 5 varieties of 3 cereals. winter (barley, wheat and rye). We measured the primary production of the different varieties and the concentration of nitric and ammoniacal nitrogen in the soil of each of the culture samples. We tested the relationship between soil nitrogen and functional traits and investigated whether there were differences in trait values and primary production between varieties of species and between species. Our results show non-significant variations in functional traits (specific leaf area, nitrogen concentration in stems and leaves) between varieties of the same species. Only stomatal conductance showed significant variations within the species Hordeum vulgare and Triticum aestivum for cultivars KWS-Higgins, Paradas and Safari-Himalaya respectively. Unlike varieties, species show significant differences in their traits but not for g0. Analysis of variance and Tukey's post hoc test allowed us to highlight highly significant variances for SLA traits, leaf% N, stem% N and primary production. Due to the unfavourable weather conditions, the crops matured late, which greatly delayed our sampling and measurements. Therefore, we were unable to perform the simulations with the two models.