Uncertainty in modelled soil organic carbon changes under various cropping systems in Australian cropland

Abstract

Agricultural soils have been suggested to sequester soil organic carbon (SOC) through adopting conservation agricultural practices such as residue retention, no-till and increasing cropping complexity. At regional or continental scales, however, it is a significant challenge to quantify the changes in SOC stock and the corresponding uncertainty associated with variation of cropping systems and environmental conditions at high spatiotemporal resolution. Here, our aim was to predict changes in SOC stock after adopting different cropping systems under optimal management for sequestering soil carbon (i.e., no nutrient deficiency and 100% residue retention) and quantify the relevant uncertainties at regional scale. Using the farming systems model APSIM, we simulated changes in SOC stocks for a 20-year period from 1990 to 2010 under a total of 59 cropping systems at 613 references sites across the Australian cereal-growing regions. These cropping systems were identified based on GRDC agro-ecological zones through expert consultation. To further understand the effects of cropping system in terms of carbon input on SOC dynamics, those cropping systems were divided into three categories to represent low-, medium- and high-input cropping systems in terms of carbon input. The simulation results indicated that, on average, the Australian agricultural soils could gain 0.19 t C ha -1 yr -1 under optimal management. However, the predicted change in SOC stocks had high variability among the three carbon-input categories. Generally, cropping systems with higher carbon input had higher efficiency for reducing SOC losses or enhancing SOC gains compared to lower carbon-input systems. For the same category of cropping system, its ability to reduce SOC losses or enhance SOC gains varied across difference GRDC zones. For example, in Qld Central zone where has higher temperature, the SOC experienced loss regardless of cropping system. In SA Vic High Rainfall zone where has lower temperature and higher rainfall, the SOC showed increase. This result indicated the importance of local soil and climate conditions to regulate the SOC dynamics under different cropping systems. A Monte Carlo approach was applied to assess the uncertainty induced by cropping system and scaling of point results to GRDC zone. Averaged across all representative cropping systems, the predicted mean SOC change was -0.1 t C ha -1 yr -1 with the 95% confidence interval ranging from -0.22 to +0.007 t C ha -1 yr -1 in Qld Central. In NSW NW/Qld SW, the predicted SOC change was zero with the 95% confidence interval ranging from -0.05 to +0.05 t C ha -1 yr -1 . In other zones, the predicted SOC change was positive. In Vic High Rainfall zones, the SOC change reached the greatest increase of +0.44 t C ha -1 yr -1 with the 95% confidence interval ranging from +0.22 to +0.66 t C ha -1 yr -1 . In Western Australia, the predicted SOC change was generally positive across all representative cropping systems. There was significant difference between different cropping system categories. In general, cropping systems with higher carbon input could reduce the SOC losses or enhance SOC gains compared to cropping systems with lower carbon input. In three zones of Western Australia (WA Northern, WA Eastern and WA Central), however, the predicted average SOC change under high-input cropping systems was lower than that under medium-input cropping systems. We further calculated the contribution of cropping systems and scaling to overall uncertainty. The simulation indicated that the variability of cropping system accounted for ~30% of the overall uncertainty. The greatest contribution of cropping system change to uncertainty in simulated SOC (>60%) was observed in three GRDC zones, i.e., NSW NW/Qld SW, NSW NW/Qld SE, and WA Northern. Our results suggested that the uncertainty in scaling of point results to regional scale is dominant in the overall uncertainty. More detailed soil databases and information on cropping systems are needed for reliable prediction of SOC dynamics in agricultural soils at regional scale.