Effects of long-term fertiliser application on cropland soil carbon dynamics mediated by potential shifts in microbial carbon use efficiency

Fertiliser application in agricultural ecosystems affects not only the potential carbon input (via crop biomass and/or manure) into the soil, but also a series of soil processes regulating soil organic carbon (SOC) decomposition. A detailed mechanistic understanding of how fertiliser regimes affect SOC dynamics is still needed. Here we constrained the Agricultural Production Systems sIMulator (APSIM model) to long-term (> 20 years) crop and SOC measurement data collected from four trials under contrasting climatic and edaphic conditions in China. By optimizing the three most influential model parameters for SOC dynamics, i.e., the decomposition rate constant of the slow humic pool (rd_hum), the fraction of stable (non-decomposable) pool (Finert), and microbial carbon use efficiency (CUE), we analysed their responses to fertiliser application regimes to infer potential mechanisms underpinning SOC changes. Our results revealed strong effects of fertiliser regimes and sites on CUE. Sites, fertiliser regimes and their interactions explained 67 % and 1.4 % of the variation in the derived CUE and rd_hum values, respectively. Linear mixed-effects modelling showed that soil C:N ratio together with carbon input amount as a random effect explained 90 % of the variation in optimised CUE values across sites and treatments. Such impact on CUE could partly explain the impact of fertiliser and carbon input on the priming effect. Fertilisers with more carbon input (i.e., straw or manure) increased CUE by 27 % - 57 % compared with chemical fertilisers in three of four sites. However, their impacts on rd_hum was divergent when decomposition of carbon pools was simulated with first-order processes. Our results demonstrate the significant effects of fertiliser regimes on CUE and thus SOC dynamics, highlighting the importance of site-specific calibration of the current SOC models and the need to quantify uncertainty bounds of any model simulated further SOC sequestration. This study also calls for developing a clear understanding to quantify the relationship between carbon input and CUE under different environment.