Assessing spatio-temporal changes in landscape structure: An index-based approach in a forested watershed

Abstract

Land use composition and structure changes in an ecosystem originates from driving forces of landscape changes. Quantifying the landscape using different metrics helps to better understanding of processes and determining the driving forces of ecosystem changes. This study investigates land use changes in the Ramian watershed in Golestan province of Iran using a landscape metrics approach to assess spatial pattern shifts over 30 years. Landsat images from 1991, 2006, and 2021 were classified using the Support Vector Machine (SVM) algorithm, and land use maps were evaluated for accuracy using the Kappa coefficient and Overall Accuracy. Seventeen landscape metrics were analyzed at both landscape and class levels using FRAGSTATS and the eight-cell neighbourhood method. Forest cover declined from 61.06% (1991) to 58.31% (2021) but remained the dominant land use, while rainfed agriculture (8.16% to 20.24%) and residential areas (0.63% to 1.3%) increased, indicating land use expansion. Over 30 years, the landscape structure became more complex. While the total number of patches (NP) slightly decreased, the reduction in the Largest Patch Index (LPI) indicated the fragmentation of the dominant forest cover. Concurrently, the Interspersion and Juxtaposition Index (IJI) increased, indicating greater intermixing of land-use types. Results revealed a clear decline in natural land covers and an expansion of human-modified uses, leading to a more complex, dispersed, and subdivided landscape structure. These findings highlight the impact of human activities on landscape degradation, emphasizing fragmentation and the growing complexity of anthropogenic patches, which signal the need for targeted ecological restoration and sustainable management practices.