Modelling varietal differences in response to phosphorus in West African sorghum

International audienceIn West Africa’s highly weathered soils, plant-available soil-P levels determine sorghum performance and yield to a far greater extent than projected variability in climate. Despite local landrace varieties having excellent adaptation to the environment and a relatively stable yield, sorghum grain yield remains quite low, averaging less than 1 t ha−1. Low P availability in West African soils has significant effects on crop development and growth with potential grain yield losses of more than 50%. Use of mechanistic models, which integrate physiological processes, could assist with understanding the differences in P-uptake among varieties and guide effective P management. Yet only few crop models include a soil-plant P model for simulating crop yield response to P management. A generic soil-plant P module was developed for crop models in the Cropping System Model (CSM) of the Decision Support System for Agrotechnology Transfer (DSSAT) but the module was adapted and tested only on two crops, groundnut and maize. The aim of the study was to adapt the soil-plant P module for sorghum and perform initial testing on highly weathered soils in West Africa. Data used in adapting and testing the soil-plant P model for sorghum consisted of in-season P concentrations and dry weights of stems, leaves and grain from four sorghum varieties covering a range of maturities and photoperiod sensitivities and grown in high-P and P-deficient soils at ICRISAT-Mali. Results showed that the coupled CERES-Sorghum − P module reasonably reproduced the vegetative and grain yield reductions experienced in the field experiments with an average RMSE of 1561 and 909 kg ha−1 under high P conditions and 1168 and 466 kg ha−1 under low P conditions, respectively. The simulations are in most cases within the observation error. We also confirmed that contrasting variety types differ in their P-uptake dynamics relative to above-ground growth change over time, and hence respond differently to available P