Vulnerability assessment of Social-Ecological systems in arid Regions: A Cross-Efficiency modified DEA model with entropy weight aggregation

Abstract

Arid social-ecological systems (SES) face escalating vulnerability from climate change and anthropogenic pressures, thus necessitating robust assessment frameworks for targeted governance. To address this need, this study extracted a typical hyper-arid area of the Tarim Basin, the Hotan Prefecture, as the research area and developed an integrated methodology combining an enhanced Data Envelopment Analysis (DEA) model and a Pressure-State-Response (PSR) indicator framework. Specifically, the DEA model was modified through cross-efficiency evaluation and entropy weight aggregation, thereby resolving traditional DEA ranking paradoxes while enabling objective, data-driven assessments. Crucially, the results revealed profound spatial-structural imbalances within the SES, as the social system (SS) exhibited significantly higher vulnerability than the ecological system (ES), contributing 68.6% of total variance. Furthermore, spatiotemporal analysis (2005–2023) indicated fluctuating but marginally declining regional vulnerability, albeit with heightened instability in the east. Regarding key drivers, pressures propelling SS vulnerability included low income, unemployment, and population crowding. ES pressures stemmed primarily from industrial emissions, inadequate wastewater treatment, and fertilizer overuse, with concentrations in northern industrial/agricultural belts. Significant SS deficiencies involved urbanization lag and income deficits, whereas agricultural resilience was observed. For ES, deficiencies centered on severe vegetation cover shortages in northern deserts, where afforestation was constrained by low precipitation and soil organic matter. Collectively, these findings demonstrate the dominance of socioeconomic drivers in arid SES vulnerability and underscore the necessity for spatially targeted governance. Key priorities therefore include: livelihood diversification, equitable economic development, ecological restoration in fragile northern corridors, and enhanced institutional capacity. Finally, future research should advance dynamic simulation, multi-scale integration, and policy coordination to effectively bridge vulnerability diagnosis with comprehensive governance.