Identifying the combined impact of human activities and natural factors on China's avian species richness using interpretable machine learning methods.
Journal:
Journal of environmental management
PMID:
39938289
Abstract
With human activities-derived escalating climate change and rapid urbanization, avian species face significant survival challenges. Understanding the impact of human activities and environmental drivers on avian species richness is critical for effective biodiversity conservation. Unlike prior studies on avian biodiversity modeling that often rely on traditional statistical or single-model approaches, this study introduces a novel combination of Extreme Gradient Boosting (XGBoost), SHapley Additive exPlanations (SHAP) and Structural Equation Modeling (SEM) to both predict and interpret avian species richness in China. Using data from the most comprehensive citizen science bird database available in the country, we explore how both natural and human-driven factors affect avian species richness. Our results showed that: (1) The number of bird species categorized as rare or endangered underscores the urgent need for bird conservation. (2) The XGBoost model outperformed the other four models in predicting avian species richness, achieving a testing R of 0.84 and a testing MSE of 2379. (3) Interpretable machine learning indicated that key factors influencing avian species richness include GDP, wind speed, longitude, farmland area, sewage treatment ratio, GDP growth, forest area, and secondary industry ratio. (4) The SEM revealed the interactive effects of natural and human factors on bird species richness, supporting the results obtained from interpretable machine learning. Based on these findings, we recommend leveraging GDP for stronger biodiversity policies, implementing afforestation to mitigate wind and enhance bird habitats, optimizing agriculture to reduce habitat destruction, and improving sewage treatment to enhance water quality for avian ecosystems. This study offers new insights for biodiversity conservation strategies by integrating economic development and environmental management, marking a pioneering effort in avian biodiversity modeling in China through the application of this novel combination of methods, which distinguishes it from earlier research approaches.