Modelling Genotype×Enivironment×Management Interactions To Increase Canola Yield Across China

Understanding the interactions of Genotype(G)x Environment(E)x Management(M)and their impact on crop yield enables prediction of the performance of genotypes across environment.It can help select the most suitable cultivars in target regions and identify contributions of novel traits in crop breeding.However experimental approaches to investigate the GxExM interactions are expensive and time-consuming,crop modelling has become an effective means for evaluation of likely impact of crop traits as they interact with environments and management practices.While process-based crop models are increasingly applied in multiple research areas,the potential uncertainties in simulated phenology,biomass growth and seed yield caused by parameter estimation have not been properly addressed.In addition,the role of model calibration and its data requirement have been rarely addressed.In this study,we firstly constrained the APSIM-Canola model to experimental data to derive the parameters controlling canola phenology and biomass growth using a Bayesian optimisation method.The dataset covered observations for 13 cultivars,86 site-years,with maximum of seven sowing dates and five plant densities each year from four canola growing regions in China.The derived parameters were used to evaluate the ability of the model to simulate canola growth and seed yield,and to investigate how data from similar or contrasting environments impacted on the efficacy of model calibration with the process-based APSIM-Canola model.Secondly,the calibrated APSIM-Canola model was applied to study the GxExM interactions and their impact on canola seed yield,aiming to identify traits that have potential to increase canola yield across canola growing regions in China.Two representative genotypes were used as base cultivars,together with seven putative traits,to simulate canola potential yield and evaluate trait contributions to yield changes at three representative sites with contrasting environments.Main results of the study are as follows:(1)Our results demonstrated that multiple combinations of parameters could lead to the same or similar simulation accuracy of canola phenology(a phenomena called equifinality)due to insufficient information and understanding to separate vernalisation and photoperiod sensitive phases.This could potentially lead to incorrect cultivar characterisation and wrong yield simulations.Our results further showed that the critical photoperiod below which canola phenological development slows down is likely to be 20 hours instead of the 16.3 hours currently used in the APSIM model.With this correction,the model was able to accurately simulate canola phenology across environments,and the impact of equifinality on simulated yield was small.Cultivar differences in terms of phenology could be accurately described by only three parameters in APSIM,i.e.,vernalisation sensitivity,photoperiod sensitivity,and thermal time required for grain-filling period.(2)Model calibration using data from a single or multiple similar seasons(or environments)would likely result in significant equifinality and simulation uncertainty.The most effective approach for calibration was to use data from contrasting environments(e.g.,at least two contrasting seasons),and better with in-season growth measurements.Such approach was demonstrated to minimise parameter equifinality and simulation uncertainty.Due to the inherent error in experimental measurements(typically 13.5%),any parameter combinations that result in simulation error(NRMSE)less than the inherent measurement error should be treated as equally good,because model calibration cannot reduce the simulation error to less than the error in the data used for calibration.(3)For the current cultivars free of nutrient stresses,potential yield of winter canola in Upper and Middle Yangtze River Basin(with sufficient rainfall or irrigation)were simulated to be 6.4 and 4.8 t ha-1 respectively.In northern dryland region of China where irrigation was not possible,the rainfed potential yield of current spring canola was simulated to be 1.1 ha-1.Yield gap between the simulated potential yield of winter and spring canola and the average county-level seed yield were 4.3,3.0 and 0.4t ha-1,respectively,in these three regions.Based on our simulation results,trait that extended the grain filling duration by 20%increased potential yield of winter canola by nearly 10%.A 20%increase in radiation use efficiency(RUE)led to 10-20%yield increase in Yangtze River Basin,but had no impact on yield where canola growth is limited by available rainfall,such as in the Northern Region.Trait that increased transpiration efficiency(TE)had the opposite impact,with 20%increase in TE leading to 30%increase in yield in the northern dryland region,but with no significant yield increase at wet sites in the Yangtze River Basin.A 20%increase in harvest index(HI)increased potential yield by nearly 20%across all the environments.The much larger yield gap for winter canola in the Yangtze River Basin indicates a much greater potential to increase canola seed yield than in the Northern Region.