Fig. 1

Sankey diagram illustrating the data filtering and quality-control steps. To obtain the final marine, terrestrial and mixed-habitat animal dataset, we downloaded data from GBIF/OBIS data portals and from additional occurrence data sources (Table S2), we removed: records earlier than 1950 (20,967 occurrences), absence data (< 1000 occurrences), occurrences based on invalid recording methods (< 1000 occurrences), invalid taxonomical information (58,490 occurrences), and true duplicates (77,651 occurrences). Finally, data with unavailable habitats were excluded from subsequent analyses (< 1000 occurrences). The cleaned dataset is available in SEANOE (https://www.seanoe.org/data/00878/99018/).