The combined effects of multiple factors on farmland and built-up land landscape patterns-A case study of Chengdu, China

Extensive farmland loss and fragmentation due to urbanization pose a significant challenge to sustainable urban development and food security. However, existing research often separately examines the drivers of area or fragmentation-related metrics of built-up land and farmland, potentially undermining effectiveness of policy since the same factors influence the two land types differently. This study addresses the gap by exploring the spatiotemporal dynamics of and the combined effects of various factors on both farmland and built-up land. By utilizing the Multiscale Geographically Weighted Regression model (MGWR), this study investigates the dynamics and combined effects of two socioeconomic factors and three location factors on the landscape patterns of built-up land and farmland in Chengdu from 2000 to 2015. To examine the performance of MGWR, the results are compared with Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR). The results indicate that MGWR exhibits stronger explanatory power than OLS and GWR. All five factors showed influence at the county scale on the area of both farmland and built-up land, with only the distance to rural settlements and population density affecting landscape metrics at the city scale. The five selected factors showed opposite effects on the aggregation index and the area of built-up land and farmland. Specifically, gross domestic product and population density were similarly associated with the landscape patterns of both land types, but these relationships shifted to contradictions with urban development. Distance to urban and rural settlements inversely affected the size of built-up land and farmland yet had consistent effects on their fragmentation. This study presents an integrated perspective and novel methodology for analyzing landscape pattern evolution, providing policymakers with insights for balancing urban expansion and farmland protection by considering area and landscape patterns of two land types. The findings advance the application of ecological indicators in sustainable land-use strategies and provide a novel perspective for exploring sustainable urbanization-associated farmland evolution.