An improved version of an ecosystem nitrogen cycling model (NLOSS) is described, tested, and used to analyze nitrogen cycling in the Yaqui Valley, Sonora, Mexico. In addition to previously described modules in NLOSS that simulate soil water and solute fluxes, soil evaporation, soil energy balance, and denitrification, modules were added to estimate crop growth, soil carbon cycling, urea hydrolysis, and nitrification. We first tested the model against season-long measurements of soil NO 3 -, NO 2 -, and NH 4 + aqueous concentrations; NO and N 2 O soil effluxes; and crop biomass accumulation in three fertilizer treatments. We used NLOSS to test the sensitivity of wheat production, NO 3 -losses, and trace-gas emissions to fertilizer application rate. With the model, we compared the typical farmer's fertilization of 250 kg N ha -1 with five other fertilization scenarios, ranging from 110 to 220 kg N ha -1 . The typical farmer's practice produced higher wheat yield than the lower fertilization treatments. However, the increase in yield per increase in kg N applied decreased with increasing fertilizer addition as a result of higher leaching losses, higher residual N, and higher trace-gas emissions. In addition, with respect to the lowest fertilization treatment, the highest fertilization treatment resulted in an 11% decrease, a 10% increase, and a 157% increase in N 2 , N 2 O, and NO emissions, respectively, and a 41% increase in leached NO 3 -+ NO 2 -. These results demonstrate that a small decrease in fertilizer application rate can increase N-use efficiency for wheat growth, while reducing leaching losses and emissions of harmful trace gas fluxes.