Field sampling
The Chishui River Basin in China is located on the border between the Yunnan and Guizhou Plateau and the Sichuan Basin. Within its distinct geographical context, this area has diverse topography, soil types, various vegetation types, and rich biodiversity. A survey was conducted in September 2021 covering an altitude range from 179–1433 m. A total of 52 sampling points were selected based on the geographical characteristics of the Chishui River Basin and the composition of fish fauna, with the goals of accessibility and maximal coverage of different habitats. The sampling points were distributed throughout five main tributaries along the Erdao, Tongzi, Guling, Datong, and Xishui rivers. The sampling points were divided across three reaches: 14 points in the upstream reach (nos. 1–14), 20 in the middle (nos. 15–34), and 18 in the downstream (nos. 35–52). The sampling point locations are shown in Fig 1.
To obtain accurate data, GPS was utilized to determine altitude, a handheld velocity radio (HZBP, Beijing, China) was used to measure flow velocity, and an infrared rangefinder (Trueyard SP1500H, Nevada, USA) was used to measure river width. A portable chlorophyll sensor (YSI6920V2-2, Ohio, USA) was used to record the chlorophyll a concentration in the water, and a multiparameter water quality analyzer (YSI6920V2-2, Ohio, USA) was used to measure dissolved oxygen (DO), pH, electrical conductivity (EC), and water temperature. The Chishui River Basin survey encompassed a wide range of altitudes and sampling points, providing valuable data on the diverse topography, soil types, vegetation types, and biodiversity within the area.
Chishui River human Footprint (FPT) data were obtained from the human footprint map of the area (https://datadryad.org/stash/dataset/doi:10.5061/dryad.052q5) (Venter et al., 2016). Additionally, Chishui River land-cover data from high-resolution (10 m) land-cover maps were obtained from http://data.ess.tsinghua.edu.cn/fromglc10_2017v01.html. Based on the land use types used in a previous fish study (Zhang Q et al., 2022) and considering the characteristics of this study area, we selected three land types; farmland, forest, and impervious surface data were acquired using ArcGIS 10.6. Values were extracted from an area of 1 km radius around each sampling location, and each value was calculated as the proportion of the relevant land cover type to the total terrestrial area, excluding water areas. The land cover values at each sampling site were averaged over three locations. In subsequent analyses, we used generalized additive models (GAMs) to analyze the relationships between land use types and fish diversity metrics.