Study on the Spatiotemporal Evolution of Habitat Quality in Highly Urbanized Areas Based on Bayesian Networks: A Case Study from Shenzhen, China

Rapid urbanization presents significant challenges to biodiversity through habitat degradation, fragmentation, and loss. This study focuses on Shenzhen, China, a highly urbanized region experiencing substantial land use changes and facing a considerable risk of biodiversity decline, to investigate the dynamics of habitat quality over two critical periods: 2010–2015 and 2015–2020. Using the InVEST (Integrated Valuation of Ecosystem Services and Trade-offs) model for habitat quality assessment and Bayesian networks to analyze causal relationships, this research offers an innovative comparison between habitat recovery and degradation across these two phases. Results indicate that from 2010 to 2015, localized habitat recovery was achieved on 0.53% of the land area due to restoration policies, yet the overall trend remained negative. During the 2015–2020 period, habitat degradation intensified (7.19%) compared to recovery (5.7%); notably, 70.6% of areas that had been previously restored are now experiencing degradation once again. This re-degradation highlights the instability of earlier restoration efforts under ongoing urban pressure. By integrating spatial analysis with Bayesian network modeling, this study provides offers a nuanced understanding of where and why initial recovery efforts were unsuccessful, identifying areas susceptible to persistent degradation. The research emphasizes that urban expansion—particularly the development of construction land, was the primary driver of habitat degradation, while ecological sensitivity played a crucial role in determining the long-term success of recovery efforts. This approach provides valuable insights for designing more effective, sustainable conservation strategies in rapidly urbanizing regions.