Simulating CO2 and N2O emissions from sub-Saharan African croplands under conservation agriculture

Agricultural ecosystems are significant contributors to greenhouse gas (GHG) emissions, yet they also offer mitigation potential through soil carbon sequestration and improved nutrient management. However, field-based assessments of major GHG emissions (e.g., CO₂ and N₂O) remain scarce in croplands in sub-Saharan African (SSA), limiting the development of region-specific mitigation strategies. Process-based crop-soil models can complement experimental studies by explicitly representing the biogeochemical processes controlling gas fluxes and by assessing the impacts of management practices.

In this study, we applied the STICS (Simulateur mulTIdisciplinaire pour les Cultures Standard, (Brisson et al., 2003)) soil-crop model to simulate soil CO₂ and N₂O emissions at two experimental sites in Zimbabwe: the Domboshava Training Centre (DTC; abruptic lixisols), and the University of Zimbabwe Farm (UZF; xanthic ferralsols). The model represents key processes governing CO₂ and N₂O production from soil, including decomposition, nitrification, and denitrification, as well as their main environmental drivers (soil temperature, water-filled pore space, ammonium and nitrate availability). Model outputs were evaluated against field GHG measurements done between 2019 and 2021 at both sites across six treatments, each replicated four times: conventional tillage, conventional tillage with rotation, no-tillage, no-tillage with mulch, no-tillage with rotation, no-tillage with mulch and rotation. Soil CO₂ emissions were simulated by combining STICS-simulated heterotrophic respiration with an independent autotrophic respiration module accounting for root respiration. After calibration, the model reproduced the main environmental drivers of soil CO₂ and N₂O emissions reasonably well. The simulated and measured soil CO₂ emissions showed moderate agreement at the daily scale (R² = 0.40, RMSE = 18.1 kg C ha^-1 d^-1, EF = 0.28) and strong agreement for cumulative emissions (R² = 0.87, RMSE = 800.74 kg C ha^-1, EF = 0.84). Simulated N₂O emissions were of the same order of magnitude as the observations across all treatments (observed range: 0-0.0126 kg N ha^-1 d^-1; simulated range: 0-0.0132 kg N ha^-1 d^-1). However, both daily and cumulative emissions were overestimated across treatments, particularly during the 2020-2021 season at UZF, potentially reflecting missed short-lived emission pulses due to non-continuous measurements. Across all treatments, simulated and observed mean seasonal N₂O emissions ranged from 0.155 to 0.580 kg N ha^-1 and 0.154 to 0.285 kg N ha^-1, respectively (R² = 0.15, RMSE = 0.18 kg N ha^-1, EF = -3.26). Overall, this modelling framework provides a useful tool to further explore the effects of crop management practices on GHG emissions in cropping systems in SSA.