Performance of the DNDC in Estimating CO 2 and N 2 O emissions of Integrated Crop-Livestock Systems

Integrated crop-livestock (ICL) systems can have a complex of effects on soil properties that can influence greenhouse gas emissions (GHG). The ICL aim to capture atmospheric CO2 and sequester it in the soil, holding promise for reducing GHG emission intensity from livestock products. Moreover, modeling N2O emissions can help assess the potential impact of N management on the ICL system to optimize the sustainability of agriculture production. Field data were obtained from an ICL experiment of EMBRAPA-Rice and Beans, located on Capivara farm, Santo Antnio de Gois/GO, Brazil (1628S; 4917W; 823 m alt.). The ICL experiment was evaluated for four years (2013-2016) with the following crop rotation sequence: pasture-fallow-maize, fallow-soybean, maize-fallow-maize, and beans-fallow. The N2O data was obtained from the 2013-14 season, which was measured in a static chamber during maize cultivation. The experiment consisted of 9 treatments (N sources and rates) with 5 replicates. The N2O was measured in 30 sampling events over almost 100 days. The daily N2O fluxes from the treatments control (No N), urea (UR), calcium ammonium nitrate (CAN), and ammonium sulfate (AS) at an N rate of 150 kg/ha were used to parametrize the DNDC. Model crop and soil parameters were adjusted to better simulate maize production and N2O emission according to observed data. DNDC simulated CO2 emissions, quantified as Net Ecosystem Exchange (NEE), were validated against CO2 emissions derived from eddy-covariance data, using statistical parameters such as R2, RMSE, MAE, and Bias. While data refinement is ongoing, preliminary findings indicate that DNDC shows promise for estimating CO2 emissions IPS under tropical conditions The DNDC had a satisfactory performance in predicting N2O emission in the ICL system, resulting in a significant correlation with the observed data (r = 0.63, p < 0.001), MAE of 0.024, and RMSE of 0.036. The average daily N2O-N emission observed was 0.026 kg ha-1 day-1 and simulated was 0.025 kg ha-1 day-1. The UR, CAN and AS applications showed a peak of N2O emission on 31th day after sowing (2 days after fertilization) corresponding to 0.175, 0.217, and 0.163 kg ha-1 day-1, respectively, where the model simulated N2O peaks of 0.151, 0.123, and 0.173 kg ha-1 day-1. The accumulated N2O emissions were 0.513, 1.148 1.738, and 0.890 kg ha-1 for control, UR, CAN, and AS respectively, in which the simulated by DNDC were 0. 778, 1.612, 1.391, and 1.755 kg ha-1. In general, the model had a good fit with daily N2O emissions, but it tended to overestimate the N2O emission from UR and AS, and underestimate from CAN. Further model parametrization and calibration may be necessary to better predict N2O and CO2 emissions. The DNDC satisfactory simulated the N2O emissions from different N sources applied to ICL system, which can be used to evaluate the potential emissions and mitigation according to N management in ICL.