Abstract:Unravelling the dynamics of soil organic carbon loss induced by soil erosion in subtropical forest is the key to formulate specific measures to maintain and improve forest carbon sink. Prior studies tend to measure the erosion-induced carbon loss of one specific forest type with a very low frequency over a short observing period, which fails us to comprehensively understand the erosion induced soil organic carbon loss in the forest. In this study, we measured runoff depth and sediment, and assayed dissolved organic carbon (DOC) concentration in the runoff and particle organic carbon (POC) concentration in the sediment in the evergreen broad-leaved forest and Cunninghamia lanceolate plantation. We tried to compare the differences of DOC and POC losses in the evergreen broad-leaved forest and Cunninghamia lanceolate plantation, and to analyze the relation of the erosion-induced DOC and POC losses with rainfall depth, rainfall intensity, the maximum 5-minutes intensity, and rainfall erosivity? Our study tried to answer the following two questions:(1) are erosion-induced DOC and POC losses in the Cunninghamia lanceolate plantation higher than those losses in the evergreen broad-leaved forest? (2) is rainfall depth a better variable than rainfall intensity, the maximum 5-minutes intensity, and rainfall erosivity in explaining the variation of erosion-induced DOC and POC losses. Our results showed that the runoff depth, sediment, runoff DOC and sediment POC concentration, DOC and POC flux in the evergreen broad-leaved forest were all significantly higher than those in the Cunninghamia lanceolate plantation. Regression analysis showed that the DOC and POC flux in both evergreen broad-leaved forest and Cunninghamia lanceolate plantation were either linearly or exponentially associated with rainfall depth, rainfall intensity, and rainfall erosivity. The goodness-of-fit of those regressions in both the evergreen broad-leaved forest and Cunninghamia lanceolate plantation was the highest for rainfall depth. The higher runoff depth and sediment in the evergreen broad-leaved forest than Cunninghamia lanceolate plantation may be ascribed to its lower understory vegetation biomass. The higher runoff DOC and sediment POC concentration in the evergreen broad-leaved forest may be a result of higher overall aboveground biomass and soil organic carbon concentration. Our results highlight that future forest management should manage understory vegetation via avoiding or reducing slashing so that the erosion-induced carbon loss in the forest will decrease. Under the scenario of climate change-induced precipitation change, rainfall depth can be used as a predictive variable to evaluate and forecast future erosion-induced carbon loss in subtropical forests.