Time-dependent sensitivity analysis of WOFOST-Potato for temperate and subtropical environments

Potato, the fourth most important food crop globally, is increasingly grown in subtropical regions, and India is now the second largest producer. Subtropical climates will be affected by the intensified occurrence of heatwaves and droughts due to climate change. Crop growth models (CGMs) are critical tools to predict and build strategies for crop climate adaptation, with the World Food Studies (WOFOST) being among the most widely used in this regard. However, WOFOST-Potato was developed based mainly on data acquired under mild or optimal weather conditions, which impedes its predictive ability in relatively hotter subtropical climates such as India. Experiments from subtropical regions can be used to broaden WOFOST applicability. Comparing the WOFOST processes that dominate crop growth during the growing season in temperate versus subtropical climates may provide insights to improve WOFOST and identify key climate adaptive traits. Systematic sensitivity analysis (SA) is useful for this purpose. Here, we applied three SA methods (local SA, Sobol and PAWN) to WOFOST-Potato to identify the key temperature-related processes responsible for simulating potato phenology and growth under hot conditions. We selected two locations with contrasting climates, i.e. the Netherlands (temperate) and India (subtropical). By computing time-dependent parameter sensitivities, we illustrate that the dominant processes change throughout the growing season. Sensitivity patterns differ fundamentally between the two locations. In the Netherlands, the leaf area index (LAI) and tuber yield (TWSO) were sensitive to development-related parameters, whereas in India, LAI and yield were predominantly sensitive to growth-related parameters. We find that time-dependent SA is able to capture the seasonal dynamics of WOFOST-Potato, and it is more insightful than the end-of-season SA to identify key parameters in subtropical environments. Building upon the global sensitivity analysis findings, we find that local sensitivity analysis still has added value to capture model behaviours driven by critical parameters. Based on our results, we argue that a combination of various time-dependent SA methods provides a robust approach to identifying the key processes of WOFOST-Potato in different environments. Our results indicate that the relationship between photosynthesis and temperatures in subtropical environments may need extra attention for modelling potato production. • Time-dependent sensitivity analysis accounts for temporal dynamics of CGMs • We applied three time-dependent sensitivity analysis methods to WOFOST-Potato. • Phenology-related parameters showed higher sensitivity in temperate conditions. • Growth-related parameters showed higher sensitivity in high-temperature conditions. • Experimentation is needed to model leaf senescence appropriately in warm environments