How does STICS crop model simulate crop growth and productivity under shade conditions?

Abstract

Most crop models have been developed with crops growing under full sunlight conditions and they commonly use daily cumulated global radiation as part of the climatic input data. This approach neglects the spatio-temporal dimension of the light reduction experienced by the crop in agroforestry systems. In this study, we evaluate the ability of the crop model STICS to predict winter wheat (Triticum aestivum L.) growth and yield under three distinct light conditions using field observations from a two year artificial shade experiment. The shade structure induced a continuous shade (CS) treatment characterized by a reduction of the proportion of light during the entire day and a periodic shade (PS) treatment defined by an intermittent shade varying on the plot throughout the day. These two shade conditions were compared to a no shade treatment (NS) receiving 100% of the available light. The model accurately predicted the timing of the grain maturity stage under the PS treatment by reducing the daily global radiation only. A correct prediction of this growth stage in the CS treatment required a decrease of the daily maximum air temperature in addition to the reduction of global radiation. Overall, the model accurately reproduces the total aboveground dry matter dynamics under the CS and NS treatments, but did not simulate the reduction observed under the PS treatment correctly. Three parameters (nbjgrain, cgrain and cgrainvo) involved in the determination of the number of grains have been calibrated with the NS treatment data and were then used to predict the crop behavior under the shaded treatments. Using this adjusted parameter set, the STICS model gave a good prediction of the grain number under all treatments. Nevertheless, the simulation of final grain yield under the shade treatments was not satisfactory yet, presumably due to an overestimation of the reallocation of the biomass between shoots and grains. Improving the prediction of these reallocation processes is challenging and critical to improve the simulation of crop behavior under fluctuating light environments such as encountered in agroforestry systems.