A landscape ecology approach: Modeling forest fragmentation with artificial neural networks and cellular‐automata Markov‐chain for improved environmental policy in the southwestern Brazilian Amazon

Abstract

Land degradation and forest fragmentation have been prominent issues in the Amazon since the 1970s, primarily driven by the suppression of primary forests due to land use changes. In this study, we propose an innovative approach by integrating artificial neural networks (ANN) and cellular automata Markov chain (CA‐MA) models to predict future land use and assess the associated forest fragmentation using landscape ecology metrics in the Jamari River Basin. The analysis reveals a significant increase in fragmentation between 1985 and 2018, as evidenced by a rise in the number of fragments from 7162 to 28,170. Moreover, we observed a decrease in the core area from approximately 2.5 million hectares to less than 1 million hectares, accompanied by an increase in edge density from 6 to 18 m.ha−1. Additionally, the average distance between fragments expanded from 66 to 95 m. Highlighting the urgency of addressing this issue, our study emphasizes the necessity of exploring effective strategies within the Brazilian public environmental management system. By utilizing various rural policy planning tools, we aim to tackle unsustainable development and mitigate the adverse impacts of land use changes in the region. In conclusion, this research offers an innovative approach that combines ANN and CA‐MA models to predict land use and assess forest fragmentation, shedding light on the alarming trend of land degradation and providing valuable insights for informed decision‐making and sustainable land management practices.