Qinhua Ke, Shan Xu, Shanshan Zong, Xinyao Jiang, Shan Li
Abstract
Agricultural expansion and forest loss (AEFL) continues to threaten tropical ecosystems. While previous studies have examined AEFL at national scales or over short timeframes, the long-term, province-level dynamics and driving forces of AEFL remain poorly characterized. This study employs interpretable machine learning to explore AEFL dynamics in Southeast Asia across three decades (1992–2002, 2002–2012, 2012–2022) at the provincial scale. As a global hotspot for agro-deforestation, Southeast Asia's forest-to-agriculture transitions were analyzed using 12 factors spanning biophysical conditions, socioeconomic development, and agricultural land status. Key findings include: (1) AEFL was widespread, with total converted areas of 6.28, 7.73, and 5.95 million ha respectively. Indonesia exhibited the highest conversion rates, followed by Malaysia. Spatiotemporal trends diverged: slowdowns dominated the Indochina Peninsula, whereas acceleration occurred in the Malay Archipelago. (2) Agricultural land scale emerged as the dominant driver, with contribution rates increasing from 24.80 % to 44.13 %. (3) Land fragmentation, climate, soil quality, and agricultural intensification significantly influenced AEFL. High rainfall zones and fertile soils were more vulnerable to conversion, while fragmented landscapes were significantly associated with AEFL. Conversely, agricultural intensification mitigated forest loss. These results enable policymakers to design spatially targeted strategies balancing agricultural productivity and forest conservation, advancing the sustainable development goals.
Spatiotemporal dynamics and driving mechanisms of agricultural expansion into forests in Southeast Asia.pdf