Detailed trait characterization is needed for simulation of cultivar responses to water stress

The potential of crop modelling to aid farm management decisions has been demonstrated in the sugar industry. Models such as the Agricultural Production Systems Simulator (APSIM) have been used in scheduling irrigation and fertilisation and for forecasting yields. APSIM models were developed based on cultivar parameters of a limited number of traits such as leaf area and how biomass is partitioned. Many of these trait parameters use the same value for all cultivars because they have not been measured or defined for different genotypes. The cultivars included in the model are no longer widely used for commercial applications and may not represent the genetic and phenological variability found in currently grown cultivars. This could reduce the accuracy of simulated results for new cultivars. To improve APSIM as a decision support system, deficiencies in APSIM's ability to simulate cultivar differences should be identified. The biomass yields of four commercially released sugarcane cultivars that are currently defined within APSIM were simulated in this paper. These cultivars were also grown under water stressed and fully irrigated conditions at Home Hill over three years. An exploratory data analysis procedure compared simulated and observed differences between cultivars. Further, a descriptive sensitivity analysis that varied the transpiration efficiency parameter (k) in APSIM demonstrated how additional cultivar-specific parameters can affect yield simulations. Simulated yields closely resembled observed yields if cultivar differences were ignored. However, APSIM did not contain many trait parameter differences between cultivars and this limited the ability of APSIM to reproduce observed cultivar responses to water stress. To simulate these interactions effectively, cultivars need to be accurately defined in the model. Incrementally adjusting k as part of the sensitivity analysis had a large effect on simulated yields. Under stressed conditions, k values as low as 6.1 g kPa/kg and as high as 12.1 g kPa/kg were required to closely simulate mean observed yields for different cultivars. This varied from the default APSIM value of 8 g kPa/kg. This paper illustrated the need for accurate characterisation of cultivar-specific traits to improve simulation of cultivar performance in different environments. Future research must consider the effect of all cultivar-specific parameters to identify which parameters have the greatest effect on simulated yields. Cultivar-specific modelling can potentially be used to identify ideal cultivars for specific environments and to tailor management strategies to specific combinations of cultivars and environments