Improving potato drought simulations: Assessing water stress factors using a coupled model

•Importance of including four water stress factors evaluated in potato modeling.•Shorter-term stress factors were most important for improving accuracy.•Stomatal regulation via hourly leaf water potential was most impactful.•Model performance was shown to be independent of varietal calibration. Effective water management is important for drought sensitive crops like potato (Solanum tuberosum L.). Crop simulation models are well suited for evaluating water limited responses in order to provide management and phenotypic-trait recommendations for more efficient production practice. There is considerable variation in how water stress components are implemented with regards to simulating soil, plant, and atmospheric relationships, thereby influencing the utility of model recommendations. Four water-stress factors were developed and implemented in the potato model SPUDIM in order to assess the contribution each factor provided for improving modeling accuracy. Iterative comparisons versus experimental data consisting of six irrigation treatments were used. Factors included F1, shifts in carbon allocation among shoot and root organs based on soil moisture availability, F2, coupled response of leaf water potential on leaf expansion, F3, and on stomatal conductance, and F4, increased tuber sink strength. F2 and F3 accounted for up to 88% of the improvement in root mean square error for total biomass. However, F1 and F4 were necessary to accurately simulate yield. F4 was also required to reproduce trends of increasing water-use efficiency and harvest index with declines in water availability. When the full 4-factor model was considered, simulated responses for total, leaf, stem, and tuber dry weights were within 11% of observed values. Daily comparisons for whole plant net photosynthesis and evapotranspiration indicated F3 was required to accurately simulate water use, but was too sensitive to very low leaf water potentials. The calibration coefficients used in the model remained stable over all six irrigation treatments, suggesting the full model can be utilized to evaluate water management strategies for potato.