The loss of biodiversity has become one of the major global environmental problems, with strengthening of biodiversity conservation and protection urgently required. The first step in solving this problem is the identification and understanding of the spatial variation of biodiversity on a regional scale, with the formulation and implementation of biodiversity protection schemes imperative. This has become the focus of biodiversity assessments over both large and medium scales. The Bailong River Watershed of the Gansu Province (BRWGP), one of the most abundant biodiversity regions in China, is located in the transitional ecotone along the Tibet Plateau, Loess Plateau, and Qinba Mountains. The BRWGP was selected in the present study to analyze the spatial heterogeneity of biodiversity. Inclusion of three assessment indicators (habitat quality, vegetation net primary productivity, and landscape state index) enabled the construction of a comprehensive index system of biodiversity spatial patterns and application of spatial differentiation analysis under the landscape geography perspective. Our goal was to revise the biodiversity evaluation module of the InVEST model and demonstrate biodiversity spatial patterns on a grid cell. The results indicated that biodiversity was high and had obvious spatial pattern variations in the BRWGP. The land area containing higher biodiversity accounted for 39.8% of the total watershed area, which was mainly distributed in the national nature reserve and forest. The relatively poor biodiversity areas were mainly distributed in the valleys of the BRWGP between Zhouqu-Wudu-Wenxian, the valley of Minjiang in Tanchang County, alpine mountain snow regions, and bare rock zones.