Abstract:

Aims: Hainan Tropical Rainforest National Park is a key biodiversity hotspot in China, yet it has experienced increasing pressure in recent decades from land-use changes including deforestation, agricultural expansion, and urbanization. However, a systematic understanding of the structural characteristics of stream fish communities in this region and their specific responses to these changes remains limited. 

Methods: From 2024 to 2025, we collected data of fish assemblages and environmental variables across 40 stream sites in and around the Limushan and Jianfengling. We compared spatial patterns in fish community composition and diversity between the two regions using multiple statistical approaches. Similarity percentage (SIMPER) analysis was applied to identify key species contributing most to community dissimilarities. We used random forest models to identify the key environmental factors of stream fish diversity under land-use change and to reveal their nonlinear response patterns. Results: Fish community composition differed significantly between Limushan and Jianfengling, with Onychostoma lepturum, Aphyocypris normalis, Opsariichthys hainanensis, and Microphysogobio kachekensis identified as the key species driving these distinctions. Species richness was significantly higher in Limushan than in Jianfengling, while fish diversity in Jianfengling also exhibited lower inter-site variability, indicating a more homogeneous spatial distribution. The major environmental drivers of fish diversity and their response patterns showed clear regional specificity and pronounced significant nonlinear relationships with land use and associated environmental variables. In Limushan, elevation and the percentage of rainfed cropland were the dominant factors, whereas in Jianfengling, the percentage of herbaceous cover cropland and impervious surfaces, pH, and conductivity played more prominent roles. 

Conclusion: This study demonstrates significant spatial variation in stream fish community composition and diversity within Hainan Tropical Rainforest National Park, with diversity indices exhibiting nonlinear responses to land-use changes and environmental variables. These findings advance our understanding of how tropical rainforest stream fish communities are influenced by land-use change and provide a scientific basis for biodiversity conservation and sustainable landscape planning in tropical rainforest regions.

Key words: species composition, biodiversity, spatial distribution, nonlinear response, random forest model