Abstract
Conversion of native vegetation to cropland and intensification of agriculture typically result in increased greenhouse gas (GHG) emissions (mainly N2O and CH4) and more NO3 leached below the root zone and into waterways. Agricultural soils are often a source but can also be a sink of CO2. Regional and larger scale estimates of GHG emissions are usually obtained using IPCC emission factor methodology, which is associated with high uncertainty. To more realistically represent GHG emissions we used the DAYCENT biogeochemical model for non-rice major crop types (corn, wheat, soybean). IPCC methodology estimates N losses from croplands based solely on N inputs. In contrast, DAYCENT accounts for soil class, daily weather, historical vegetation cover, and land management practices such as crop type, fertilizer additions, and cultivation events. Global datasets of weather, soils, native vegetation, and cropping fractions were mapped to a 1.9°×1.9° resolution. Non-spatial data (e.g., rates and dates of fertilizer applications) were assumed to be identical within crop types across regions. We compared model generated baseline GHG emissions and N losses for irrigated and rainfed cropping with land management alternatives intended to mitigate GHG emissions. Reduced fertilizer resulted in lower N losses, but crop yields were reduced by a similar proportion. Use of nitrification inhibitors and split fertilizer applications both led to increased (~6) crop yields but the inhibitor led to a larger reduction in N losses (~10). No-till cultivation, which led to C storage, combined with nitrification inhibitors, resulted in reduced GHG emissions of ~50 and increased crop yields of ~7.
