OOS 19-1 - Constraining reactive and non-reactive N losses from sugarcane systems – a synthesis of high-frequency N2O, in-situ 15N2O and 15N2 measurements and fertiliser 15N recoveries and statistical models.

Nitrous oxide (N2O) emissions from agricultural soils are a major contributor to global climate change and to the depletion of stratospheric ozone. The reduction of reactive N2O to non-reactive, and thus environmentally benign dinitrogen (N2) via denitrification is the main sink for N2O in agricultural soils. Despite their agronomic and environmental significance, magnitude and relative balance of N2O and N2 losses remain major uncertainties across scales. This limits our quantitative process understanding and accurate forecasting of N cycling, which is urgently needed to develop improved farming practices driving environmental and agronomic benefits.

To constrain reactive and non-reactive N losses, we used a novel approach combining a) high-frequency N2O measurements b) in-situ N2 and N2O quantification via the 15N gas flux method and c) fertiliser 15N recoveries at the end of the season with d) statistical models in two subtropical sugarcane systems. The use of generalised additive models (GAMs) allowed defining dynamic changes in the in-situ daily N2O/(N2O+N2) ratio as a function of environmental variables. This enabled us to upscale high-frequency N2O data to daily field-scale N2 emissions over the season.

The daily N2O/(N2O+N2) ratio varied over time from < 0.01 to 0.768 and the fitted GAM explained 52% of the deviance. Seasonal N2O+N2 losses increased non-linearly from 12 to 87 kg N ha-1 in response to increasing N rates. The seasonal N2O/(N2O+N2) ratio increased with increasing N rates, with N2O accounting for 2.4–5.8% of N2O+N2 losses. Denitrification losses accounted for 31–78% of overall fertiliser 15N loss and were dominated by N2 emissions. The relative contribution of N2 to N fertiliser loss decreased with increasing N rates, suggesting other N loss pathways including leaching and runoff become increasingly dominant.

Our findings demonstrate large amounts of applied N fertiliser lost via denitrification. Importantly, the fraction of reactive N removed via N2 production declines, showing a shift from non-reactive to reactive N loss pathways with increasing N rates. This shift, together with the overall amount of N lost highlights both agronomic and environmental inefficiencies of excessive fertiliser N inputs in sugarcane systems. The unique combination of stable isotope and statistical methods introduced here can serve as a blueprint to estimate denitrification losses for a wide range of agroecosystems. Future implementation of refined estimates of the N2O/(N2O+N2) ratio in biogeochemical models will help to advance our quantitative understanding of N cycling across temporal and spatial scales.