Assessment of the Simple Model for Durum Wheat Biomass and Yield Prediction in Semi-arid Algeria

Production of durum wheat (Triticum durum Desf.) in semi-arid Mediterranean regions is often limited by water and heat stress. Parsimonious crop models are valuable tools for supporting management decisions in such environments. This study evaluated the performance of the SIMPLE model in simulating biomass accumulation and grain yield of durum wheat under diverse agro-environmental conditions in Algeria. The model was calibrated using data from a controlled trial (ITGC site, 2012-2013 season) and independently validated with data from three additional sites across two seasons (2012-2013 and 2023-2024), including contrasting management practices (rainfed vs. supplementary irrigation). During validation, model performance was high, with a coefficient of determination (R²) > 0.74, modeling efficiency (EF) > 0.90, and normalized root mean square error (nRMSE) < 25% for both biomass and yield. The model accurately predicted grain yields, ranging from 4.91 t ha-1 under rainfed conditions to 6.37 t ha-¹ at the irrigated site. Deviations between simulated and observed values were mainly associated with errors in total biomass simulation (source-driven) rather than biomass partitioning to the grain (sink-driven). Overall, these results demonstrate that, despite its minimal parameter requirements, the SIMPLE model reliably reproduces wheat growth dynamics and final yield across different seasons, sites, and water regimes. Therefore, it is a robust tool for in-season yield forecasting and for evaluating the potential benefits of management strategies, such as supplementary irrigation, in semi-arid agricultural systems.