Modelling future landscape fragmentation through integrated simulation and metrics-based analysis in Bila Watershed, Indonesia

Abstract

Land use and land cover (LULC) changes are significant factors driving ecological shifts and landscape fragmentation. However, most predictive frameworks rarely integrate spatial simulations with structural landscape interpretations, particularly in tropical watersheds. This study presents an integrated cellular automata–artificial neural network (CA–ANN) framework coupled with landscape metrics to predict future fragmentation in the Bila Watershed, South Sulawesi, Indonesia. Multi-temporal Landsat data (2012, 2018, 2024) were classified, and future scenarios for 2030 and 2036 simulated. Five landscape metrics were computed, including number of patches (NP), largest patch index (LPI), edge density (ED), proportion of landscape (PLAND), and Shannon’s diversity index (SHDI). The CA–ANN model achieved high predictive performance (overall kappa (koverall) = 0.94, overall accuracy (OA) = 96.19%). From 2012 to 2036, forest cover (PLAND) declined (from 56 to 51%), the landscape’s LPI decreased (from 54.93 to 50.48), and SHDI increased (from 1.241 to 1.323), indicating growing heterogeneity and declining core forest dominance. Concurrently, NP (from 295 to 552) and ED (from 0.0019 to 0.0033 m∙ha–1) for dryland agriculture surged, indicating intensified patch proliferation in agricultural zones. The cumulative conversion of 5,690 ha of forest and shrubland into agriculture, mainly in the upper and accessible terrains, emerged as the dominant structural driver. By linking predictive accuracy with ecological structure, this framework enhances LULC modelling for landscape planning, risk assessment, and sustainable watershed management. Fully reproducible with open-source Geographic Information Systems (GIS), the approach supports Sustainable Development Goal 15 (“Life on land”).

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint