Simulating Cassava Growth and Yield under Potential Conditions in Southern Brazil

Abstract

Cassava [ Manihot esculenta (L.) Crantz] plays an important role as staple food in the tropics. The GUMCAS model is a process‐based dynamic simulation model for cassava that has been adapted to the Cropping System Model (CSM) framework of DSSAT (DSSAT‐CSM). The objective of this study was to calibrate and evaluate the original GUMCAS model, a modified version of the GUMCAS model, and the current cassava model in DSSAT under potential conditions in the subtropical environment of Rio Grande do Sul State, Brazil. The modified original GUMCAS model consisted of three modifications in the code: we included a third independent “clock” in the cassava development for the onset of starch accumulation, we replaced the rate of leaf appearance submodel with the Wang and Engel model, and we modified the leaf senescence submodel. Model calibration was with a field experiment for cultivar Fepagro–RS 13 during the 2011/2012 growing season in Santa Maria, Brazil. Independent data from 16 experiments conducted at four sites in Rio Grande do Sul State were used for testing the performance of the three versions of the cassava model. The original GUMCAS model had the poorest performance, and the modified GUMCAS model slightly improved the predictions of stem and storage root yield compared with the current DSSAT cassava model. The modified GUMCAS model greatly improved the predictions of developmental stages, leaf development, and leaf area growth dynamics compared with the current DSSAT cassava model. Results from this study contribute to our understanding on how a cassava system functions in the subtropics.