Modelling weed management strategies to control erosion in rubber plantations
The role of weeds in soil conservation in agroforestry systems has been largely ignored. We used the Land Use Change Impact Assessment (LUCIA) model to simulate the effects of weed management on erosion in rubber plantations (Hevea brasiliensis Muell. Arg). In order to quantify the impact of a dynamic, spatially explicit multi-layer plantation structure on erosion processes in agroforestry systems, we updated LUCIA's erosion module. Its new version simulates soil detachment due to rainfall and runoff, considering the separate effects of the tree canopy and surface cover on soil erosion. The updated LUCIA model was calibrated and validated based on an established rubber plantation experiment in Xishuangbanna, Southwest China, to evaluate the impact of different weeding strategies on soil loss. The model successfully represented the impact of the dynamic multi-layer structureon erosion and was able to predict well the effects of weed management on soil loss and runoff at the test site over 1 year, with a modelling efficiency (EF) of 0.5–0.96 and R2 of 0.64–0.92. Subsequently, we validated the ability of the model to simulate surface cover changes under rubber plantations of different age (up to 40 years). Simulation outputs for 4-, 12- and 18-year-old rubber plantations revealed satisfying to good results. However, the predicted change in surface cover for old rubber plantations (25- and 36-year) failed to meet the field trends. The model predicted the greatest erosion in the year when the rubber canopy started to close. During this period, weed growth was limited by light, while litter input from rubber was insufficient to provide good soil cover. Four weeding strategies (“clean-weeding”, “twice-weeding”, “once-weeding” and “no-weeding”) were designed for scenario simulations. Based on the results of 20-year runs, we concluded that “once-weeding” and “no-weeding” both efficiently minimized soil loss during one rotation length. A high degree of surface and weed cover (over 95% and 60%) under “no-weeding” makes this management strategy with dense undergrowth hardly acceptable by local farmers due to reduced tree accessibility for tapping and increased potential danger through poisonous caterpillars. “Once-weeding”, on the other hand, controlled overgrowth of understoryvegetation by keeping weed cover below 50%. We therefore suggest “once-weeding” as an improved herbicide management strategy in rubber plantations, to meet ecological system service maintenance and to facilitate adoption in practice.