Spatial-temporal urban sprawl and landscape fragmentation in secondary towns in western Kenya: Multi-Scale landscape metrics from Kakamega central region

Urban sprawl has become one of the principal modes of urban growth globally, with secondary towns continuing to absorb a rising number of urban dwellers in Sub-Saharan Africa (SSA). Despite this development, the surrounding lands in such towns remain poorly understood in terms of structure and clustering of development pressure. This study employed a multi-scale landscape metric approach to quantify Land use/Land cover (LULC) changes precisely between 2003 and 2023; to analyze the spatial-temporal patterns of urbanization using landscape metrics, and to identify urbanization hotspots associated with structural landscape transformations in Western Kenya. Classified LULC maps from Landsat images were used to quantify class sizes and extract landscape metrics. The classification achieved high overall accuracies of 95.8%, 92.0%, and 86.7% for 2003, 2013, and 2023, respectively. Results show that built-up area increased from 0.39% in 2003 to 5.02% in 2023, representing a twelvefold increase. Additionally, cropland expanded steadily while grassland sharply declined, reflecting the influence of built-up expansion. Further, landscape metrics at the sub-county scale showed an increasing fragmentation of urban form, characterized by disconnected built-up patches. On the contrary, analysis of grid-based spatial autocorrelation and hotspot revealed that fragmentation was spatially organized. Moreover, organization, stability, and consistent growth from the peripheries of Navakholo and Malava sub-counties indicated pressure on the natural ecosystem and predictability for targeted management. The findings demonstrate that peri-urban sprawl in secondary towns can simultaneously exhibit fragmented form and spatially concentrated growth pressure. Therefore, the study provides actionable spatial evidence for targeted and prioritized planning in resource-constrained secondary towns.