遥感数据在描述解释预测物种分布模式和物种丰富度时起着至关重要的 作用，因为它可帮助确定物种适宜的栖息地以及对物种丰富度起积极作用的 生态因子。但在研究中一个关键问题是如何总结遥感数据提取关键参数特征 和指标，使其与物种丰富度分析最为相关。动态生境指数（Dynamic Habitat Indices，DHIs）便是为此设计的三个指标。另外影响物种分布的生境条件作 为区域化变量，往往存在空间自相关，问题是哪种模型能更好地减少空间自相 关效应。本文研究目标是将全球动态生境指数与物种丰富度相关分析进行空 间上的拓展，并且使用地理加权随机森林对空间关系进行建模，同时与经典的 全局模型普通最小二乘回归和局部模型地理加权回归进行对比，在空间关系 上评估有关全球物种丰富度的假设。本文根据中分辨率成像光谱仪（MODIS） 日反射率产品计算了全球 0.05°空间分辨率的增强型植被指数（EVI），使用 EVI 表征植被生产力计算动态生境指数。以世界自然保护联盟濒危物种红色名 录（IUCN Red List of Threatened Species）为鸟类、哺乳动物和两栖动物 物种丰富度的数据来源。研究发现在全球范围内动态生境指数与物种丰富度 具有显著相关关系，为所有三个假设提供了强有力的支持。并且地理加权随机 森林在所有模型中表现最优，动态生境指数分别平均解释了鸟类、哺乳动物和 两栖动物物种丰富度的 90.3%、90.3%和 89.7%的空间变异。动态生境指数与 生物多样性生态假说密切相关，可以很好地预测物种丰富度，有望在生物多样 性科学和保护中得到应用。

Remote sensing data play a crucial role in describing and interpreting predicted species distribution patterns and species richness, as it can help identify suitable habitats for species and ecological factors that play a positive role in species richness. However, a key issue in the study is how to summarize the remote sensing data to extract key parameter features and indicators that are most relevant to species richness analysis. Dynamic Habitat Indices (DHIs) are three indicators designed for this purpose. In addition, habitat conditions that influence species distribution are often spatially autocorrelated as regionalized variables, and the question is which model can better reduce the spatial autocorrelation effect. The goal of this paper is to spatially extend the analysis of global dynamic habitat indices in relation to species richness and to model the spatial relationships using a geographically weighted random forest, and to assess the hypothesis about global species richness in terms of spatial relationships in comparison with the classical global model ordinary least squares regression and the local model geographically weighted regression. In this paper, a global enhanced vegetation index (EVI) at 0.05° spatial resolution was calculated based on the Moderate Resolution Imaging Spectroradiometer (MODIS) daily reflectance product, and a dynamic habitat index was calculated using the EVI to characterize vegetation productivity. The IUCN Red List of Threatened Species (IUCN Red List) was used as the data source for bird, mammal and amphibian species richness. Dynamic habitat indices were found to be significantly correlated with species richness at the global scale, providing strong support for all three hypotheses. And geographically weighted random forests performed best among all models, with dynamic habitat indices explaining on average 90.3%, 90.3%, and 89.7% of the spatial variation in bird, mammal, and amphibian species richness, respectively. Dynamic habitat indices are closely related to the ecological hypothesis of biodiversity, can predict species richness well, and are expected to be applied in biodiversity science and conservation.