Agroecosystems cover nearly 40% of the global land surface and are the largest continuous ecosystem in the United States. Therefore future effects of climate change on ecosystem goods and services in the Midwest United States will be dominated by the response of crops. Previous modeling studies predicted how temperature changes might affect yield, but projections of the effects of climate change and increasing atmospheric carbon dioxide concentration ([CO₂]) on ecosystem-level processes of croplands has been limited by our scientific understanding and available validation data. We adapted the Agro-IBIS dynamic global vegetation model to simulate the response of soybean to elevated [CO₂] using data from the Soybean Free Air Concentration Enrichment (SoyFACE) facility. We modified the physiological and developmental algorithms to simulate short-term responses (e.g., assimilation and stomatal conductance) as well as long-term responses (e.g., timing of growth stages and leaf development). We added a response to photoperiod within the soybean development module; previously development depended only on thermal time. Our new algorithms allow us, for the first time, to quantify the potential effects of increasing [CO₂] on regional energy and water budgets across the major soybean region of the United States.  
Results/Conclusions

We ran the model with atmospheric CO₂ concentrations of 370 ppm and 550 ppm with a 50-year climate record over a domain that encompasses land east of the Rocky Mountains. The new development module produces more realistic simulated values of leaf area index over the growing season. Because the photoperiod response accounts for geography, the new algorithm will facilitate the incorporation of genetic variation among soybean maturity groups across latitudes as we simulate the effects on regional carbon and water fluxes. On average, evapotranspiration (ET) is decreased everywhere soybean is grown at [CO₂] of 550 ppm. In contrast, the sensible heat flux term is increased, corresponding to elevated canopy temperatures. For a given amount of precipitation, the decreased ET is associated with increased sub-surface drainage. While the model predicts the correct response to elevated [CO₂], the magnitude is much lower than observed at SoyFACE. In order to capture the observed response to ET and canopy temperature and retain the correct simulation of other variables (e.g., leaf area index), it is necessary to adjust several model parameters. Ongoing work involves deciding how to adjust these parameters with transient increases in [CO₂]. Future experiments will quantify the interactions between the response to elevated [CO₂] and projected climate change within this region.