A SD-MaxEnt-CA model for simulating the landscape dynamic of natural ecosystem by considering socio-economic and natural impacts

Background climate and socio-economy signi?cantly affect the landscape dynamics of natural ecosystem. And the simulated method to understand the landscape dynamics of natural ecosystems in responding to human and natural effects is increasingly important for ecosystem management. In this study, we integrated system dynamics, MaxEnt and cellular automata models (called as SD-MaxEnt-CA model) to consider climatic effects and human activities during the simulation of natural ecosystems landscape dynamics. In the integrated model, the MaxEnt process was used for the interactive coupling of the top-down SD and bottom-up CA, which improve the models ability to accurately acquire the weights of each layer and transition rules during the simulation with the effects of climate changes. This proposed model was applied to simulate the landscape degradation of natural forest caused by the expanding of farmland, tea gardens and rubber plantations in Xishuangbanna from 2000 to 2010, one of the worlds 34 biodiversity hotspots. By the comparison with other algorithms, the MaxEnt achieve a good performance in acquiring the weights of each layer. The spatial consistent between simulated map and actual result reached to 84.2%, indicating the promising grid-to-grid agreement. Proposed model considered the driving of climate changes and socio-economy, which improved the accuracy of simulating the landscape dynamics of natural ecosystems. The method was further used in simulating the landscape dynamics of natural forest in Xishuangbanna from 2010 to 2030 under four different climate scenarios. The variation of conversion area among different climate scenarios is consistent with their response to climate changes. The simulation results demonstrate that the proposed model is effective for the landscape simulation of natural ecosystems.