| 英文摘要: | Soil organic carbon (SOC) cycling schemes used in land surface models (LSMs) typically account only for the effects of net primary production and heterotrophic respiration1. To demonstrate the significance of omitting soil redistribution in SOC accounting, sequestration and emissions, we modified the SOC cycling scheme RothC (ref. 2) to include soil erosion. Net SOC fluxes with and without soil erosion for Australian long-term trial sites were established and estimates made across Australia and other global regions based on a validated relation with catchment-scale soil erosion. Assuming that soil erosion is omitted from previous estimates of net C flux, we found that SOC erosion is incorrectly attributed to respiration. On this basis, the Australian National Greenhouse Gas inventory overestimated the net C flux from cropland by up to 40% and the potential (100 year) C sink is overestimated by up to 17%. We estimated global terrestrial SOC erosion to be 0.3–1.0 Pg C yr−1 indicating an uncertainty of −18 to −27% globally and +35 to −82% regionally relative to the long-term (2000–2010) terrestrial C flux of several LSMs. Including soil erosion in LSMs should reduce uncertainty in SOC flux estimates3, 4 with implications for CO2 emissions, mitigation and adaptation strategies and interpretations of trends and variability in global ecosystems5. Soils are estimated to store up to 80% of the organic carbon in the terrestrial biosphere (2,376–2,450 Pg C to a depth of 2 m) and contain more than three times the organic carbon in the atmosphere6. The amount of carbon dioxide (CO2) captured and converted to SOC annually through terrestrial net primary production or released as CO2 by soil microbial respiration is about an order of magnitude greater than the annual increase in atmospheric CO2 (ref. 7). Soil therefore represents a substantial component within the global carbon cycle and small changes in the SOC stock may result in large changes of atmospheric CO2 particularly over tens to hundreds of years8. Similarly, global climate change could influence fixation and respiration with implications for terrestrial ecosystems and feedbacks to global biogeochemistry and radiative forcing9. Soil erosion is a global issue10 that occurs more intensively on cultivated land than on rangeland11. Since agricultural expansion, many global regions have subsequently introduced conservation agriculture to reduce soil erosion and consequently there is considerable spatial and temporal variation of soil erosion associated with the history of land use and management (for example, conservation agriculture in Australia12, 13, 14). Soil erosion has removed considerable quantities of topsoil10, which may continue and perhaps be exacerbated by projected extremes, for example, Australian climate change15. There is renewed awareness of the global significance of soil erosion probably due in part to the debate about whether soil erosion is a sink or source16 of CO2 and its impact on soil nutrient redistribution and global biogeochemistry17. Carbon cycling schemes, for example, RothC (ref. 2), Century18 and crop models such as APSIM (ref. 19) are used to predict C change for different environmental and management conditions. These types of model underpin regional assessments of the terrestrial carbon budget that consider the processes of net primary production and respiration. One of the most widely used SOC turnover models is RothC. For example, SOC cycling in RothC underpins the Australian Government’s spatial modelling framework Full Carbon Accounting Model (FullCAM) and the modified Carnegie–Ames–Stanford approach (CASA) within the LSM CABLE20. Here we demonstrate that SOC flux should also include losses (and gains) due to soil erosion (and deposition). We modified the C cycling scheme RothC (version 26.3) to include an erosion component (RothCE; Supplementary Section 1.1). For several long-term experimental plots in Australia (Supplementary Section 1.3) we measured 137Cs and estimated net (1973–1993) soil erosion using three approaches (Table 1 and Supplementary Sections 1.4 and 1.5). These results demonstrated that plots had been exposed to soil erosion involving the loss of SOC (SOC erosion). The different estimates of soil erosion were used with RothCE, established decomposition rates for Australia21 and the measured SOC fractions, to estimate SOC loss with and without erosion (Fig. 1). Depending on the magnitude of erosion there may be a considerable difference between the model predictions with and without erosion. The net C flux of optimized RothCE was compared with that of the model without erosion and showed a consistent under-estimate of net C flux in the presence of soil erosion that we interpreted as SOC erosion (Table 1 and Fig. 1).
|