Preliminary evaluation of the SPOTCOMS model for simulation of Sweetpotato growth and development in the southern United States

Abstract

Sweetpotato [Ipomoea batatas (L.) Lam.] is an economically and nutritionally important root crop, yet the application of process-based crop models for its simulation remains limited, particularly when compared with major cereal crops. Process-based models provide powerful tools for evaluating crop responses to environmental variability, management strategies, and climate- and policy-driven scenarios. This study evaluated the applicability and performance of the SPOTCOMS (Sweet Potato Computer Simulation) model in simulating sweetpotato growth, development, and yield formation. Field experiments were conducted over two consecutive growing seasons (2022 and 2023) using the widely grown ‘Covington’ cultivar under conventional production practices in Alabama. Comprehensive datasets were collected, including phenological and growth parameters (vine length, leaf area, branch number, and storage root development at multiple growth stages), soil physicochemical properties, and daily weather variables. These data were used to derive cultivar-specific genetic coefficients and to calibrate and validate the SPOTCOMS model. The SPOTCOMS model successfully captured key phenological stages of sweetpotato development, demonstrating high predictive accuracy for storage root number and total yield, with greater than 90% agreement between simulated and observed values. Vine length was also well simulated (R² = 0.90). In contrast, predictions of leaf number and branching showed moderate agreement with observations (R² = 0.50), indicating model limitations in simulating vegetative architecture. Sensitivity analysis identified early-season growth and mid-season leaf expansion as critical drivers of final storage root yield. Overall, these findings highlight the potential of SPOTCOMS as a decision-support tool for predicting sweetpotato performance and optimizing production in the southeastern United States. Further calibration and validation across multiple cultivars, environments, and management systems are recommended to improve model robustness and broader applicability.