Evaluation of WARM for different establishment techniques in Jiangsu (China)

• We developed a new algorithm for post-transplanting shock and growth dynamics.
• The algorithm was tested within the WARM rice model using data from 24 experiments.
• Experiments were available for direct sowing, manual and mechanical transplanting.
• Same values for plant traits parameters were used for three establishment methods.
• Good results were achieved in simulation of biomass for all establishment methods.

WARM is a model for rice simulation accounting for key biotic and a-biotic factors affecting quantitative and qualitative (e.g., amylose content, chalkiness) aspects of production. Although the model is used in different international contexts for yield forecasts (e.g., the EC monitoring and forecasting system) and climate change studies, it was never explicitly evaluated for transplanting, the most widespread rice establishment method especially in tropical and sub-tropical Asia. In this study, WARM was tested for its ability to reproduce nursery growth and transplanting shock, using data on direct sown and transplanted (both manual and mechanical) rice collected in 24 dedicated field experiments performed at eight sites in Jiangsu in 2011, 2012 and 2013. The agreement between measured and simulated aboveground biomass data was satisfactory for both direct sowing and transplanting: average R2 of the linear regression between observed and simulated values was 0.97 for mechanical transplanting and direct sowing, and 0.99 for manual transplanting. RRMSE values ranged from 5.26% to 30.89%, with Nash and Sutcliffe modelling efficiency always higher than 0.78; no notable differences in the performance achieved for calibration and validation datasets were observed. The new transplanting algorithm – derived by extending the Oryza2000 one – allowed WARM to reproduce rice growth and development for direct sown and transplanted datasets (i) with comparable accuracy and (ii) using the same values for the parameters describing morphological and physiological plant traits. This demonstrates the reliability of the proposed transplanting simulation approach and the suitability of the WARM model for simulating rice biomass production even for production contexts where rice is mainly transplanted.