Simulating potato growth and nitrogen uptake in eastern Canada with the STICS model

Abstract

The ability of process-based soil–crop models to simulate potato (Solanum tuberosum L.) yield and N uptake for a range of N fertilization under the conditions of eastern Canada has never been tested. Our objectives were (i) to calibrate and evaluate the performance of the STICS model for the Shepody and Russet Burbank cultivars with cultivar-specific critical N concentration dilution curves, and (ii) to quantify the gain in model performance with cultivar-specific N concentration curves rather than a generic curve. Data sets including measurements of leaf area index (LAI), total and tuber biomass, and total and tuber N uptake for several N rates (0–280 kg N ha–1) collected at Charlottetown, PE; Fredericton, NB; and Québec, QC, Canada were used. Calibration was done with one data set from Charlottetown for Shepody and one data set from Québec for Russet Burbank, while all the other data sets were used to evaluate model performance. Following calibration, the STICS model generally performed well, with a normalized root mean square error (NRMSE) < 30% and a normalized mean error (NME) ranging from –8% to 23%, for LAI and biomass. Model performance was slightly worse for total and tuber N uptake, although cultivar-specific N concentration curves for Shepody and Russet Burbank improved model performance compared with a generic curve with a lower NRMSE (18–50% vs. 21–63%) and NME (–9 to 23% vs. –14 to 23%). Cultivar-specific critical N concentration curves should therefore be used for assessing the N budget of potato cropping systems.