Machine learning reveals dynamic controls of soil nitrous oxide (N2O) emissions from diverse long-term cropping systems

Soil nitrous oxide (N2O) emissions exhibit high variability in intensively managed cropping systems, which challenges our ability to understand their complex interactions with controlling factors. We leveraged 17-years (2003-2019) of measurements at the Kellogg Biological Station LTER/LTAR site to better understand controls of N2O emissions in four corn–soybean–winter wheat rotations employing Conventional, No-till, Reduced input, and Biologically-based/organic inputs. We used a Random Forest machine learning model to predict daily N2O fluxes, trained separately for each system with 70% of observations, using variables such as crop species, daily air temperature, cumulative 2-day precipitation, water-filled pore space, and soil nitrate and ammonium concentrations. The model explained 29 to 42% of daily N2O flux variability in test data, with greater predictability for the corn phase in each system. The long-term rotations showed different controlling factors and threshold conditions influencing N2O emissions. In the Conventional system, the model identified ammonium (>15 kg N ha-1) and daily temperature (>23 °C) as the most influential variables; in the No-till system, climate variables, precipitation, and temperature were important variables. In low input and organic systems, where red clover (Trifolium repens L.; before corn) and cereal rye (Secale cereale L.; before soybean) cover crops were integrated, nitrate was the predominant variable, followed by precipitation and temperature. In low input and biologically-based systems, red clover residues increased soil nitrogen availability to influence N2O emissions. Long-term data facilitated machine learning for predicting N2O emissions in response to differential controls and threshold responses to management, environmental, and biogeochemical drivers.