Does drought stress eliminate the benefit of elevated CO2 on soybean yield? Using an improved model to link crop and soil water relations

Crop simulation models are indispensable tools that facilitate studies to assess climate impacts and adaptation responses, but have not been adequately tested in terms of accurately predicting crop growth and water stress responses to high carbon dioxide concentrations [CO2]. The soybean model, GLYCIM, previously modified with a coupled leaf-level gas exchange - energy balance model, was integrated with a two-dimensional convective diffusive root growth module which linked soil and leaf water potentials with the regulation of stomatal conductance. We evaluated the accuracy of this modified GLYCIM using experimental data from a Free-Air CO2 Enrichment (FACE) site, SoyFACE, and used the calibrated model to quantify current and projected elevated CO2 effects on soybean yields in the United States (US) Midwest. The model accurately matched SoyFACE observed data including canopy temperature (RMSE, 2.26-2.42 degrees C; IA, 0.89-0.90), photosynthesis rate (RMSE, 6.33-12.88 mu mol m- 2 s- 1; IA, 0.82-0.92), seasonal LAI (RMSE, 0.96-1.08 cm2 cm-2; IA, 0.91-0.93), biomass (RMSE, 2.06-2.34 Mg ha 1; IA, 0.87-0.90), soil water contents (RMSE, 5.42-7.82 %; IA, 0.80-0.85) and root length (RMSE, 2.12-5.64 m plant 1; IA, 0.70-0.86). GLYCIM also reproduced the observed declining response of yield to CO2 fertilization with increasing drought. Using an ensemble of five climate model projections, GLYCIM showed that the CO2 fertilization effect on yield was highest in northern counties and declined in southern and the western counties by late 21st century in response to increased drought characterized by higher daily maximum temperature and decreased rainfall. This study highlights the need to continue to test and improve crop models and use geospatially referenced adaptive strategies to capitalize the benefits from elevated [CO2] to better counteract increasing drought.