Data for: Aquatic communities and water quality encode urban environmental gradients at contrasting spatial scales

This dataset supports the study “Aquatic communities and water quality encode urban environmental gradients at contrasting spatial scales”. It brings together a regional aquatic-community survey, multiscale environmental predictors, regional public water-quality records, and global biological and water-quality indicators used in the analysis. The regional biological analyses comprise 564 site-events at 130 sites and 13 spatial ranges from 200 to 6,000 m. The files include standardised taxon-level and fish individual records, analytical response matrices, taxon annotations, feature definitions, statistical results and figure source data. The inferential regional water-quality input comprises 2,643 source-unit/month records from December 2020 to June 2026. The 101 source units with class observations are mapped to 54 physical analysis units. The separate descriptive series in Supplementary Data 4 covers 2021–2025 and contains 1,755 waterbody/month records. Files comprise four Supplementary Data workbooks, seven figure Source Data workbooks and Regional_Input_Data.zip, together with README.md, THIRD_PARTY_DATA.md and a SHA-256 file manifest. Workbook indices and dictionaries describe variables, units and missing values. The regional ZIP contains source-record and analytical-input provenance, a 564-event crosswalk, the monthly water-quality table and the source-to-analysis-unit mapping. Standardised records are provided; scanned field sheets and copies of complete public monitoring reports are not included. Associated analysis and plotting code: https://github.com/zylonglong/aquatic-communities-urban-gradients . Its documentation distinguishes the tested reproduction examples from the configuration requirements of the archived analysis scripts. Licences apply by component. Author-owned survey observations, documentation and results are CC BY 4.0. GLORICH-derived components retain CC BY-NC-SA 4.0, and OpenStreetMap-derived database components retain ODbL 1.0. FishBase-derived trait values retain their source-specific non-commercial attribution conditions. Other reused inputs retain the terms and attribution identified in THIRD_PARTY_DATA.md; no blanket CC BY grant is made for third-party components. Cite the original input datasets as well as this data record when reusing the relevant material.