Research on Forest Farmland Ecosystems Based on ARIMA-LSTM and Lotka-Volterra Model

Abstract

This study integrates the ARIMA-LSTM and Lotka-Volterra model to construct a farmland evolution model simulating the process of forest-to-farmland conversion. By incorporating biological factors and human decision-making, the study further develops a forest-farmland ecosystem simulation model. The research first conducts a comparative analysis to determine that spatially differentiated management strategies are the optimal land reclamation method, and employs the ARIMA-LSTM hybrid model to enhance the accuracy of forest-to-farmland trend predictions. Second, the study constructs a forest-farmland ecosystem simulation model based on the Lotka-Volterra model, incorporating factors such as interspecific predation and competition, seasonal fluctuations, and pesticide interference. By introducing native species to expand the analysis to the community level, the study reveals the mechanisms maintaining ecosystem stability, achieving a leap from single land-use simulation to complex ecosystem modelling. This study integrates time series prediction with ecological dynamics mechanisms to establish a comprehensive analytical framework encompassing ‘reclamation strategy screening - evolutionary trend prediction - ecosystem simulation.’ This framework provides quantifiable and verifiable scientific tools for the sustainable management of agroforestry systems, aiding in the resolution of challenges related to harmonious human-land development.