A dynamic, depth-resolved Area-of-Habitat dataset for global marine species

This dataset provides global, multi-decadal, depth-resolved raster layers of Summed Area of Habitat (Summed AOH) for seven target years: 1995, 2000, 2005, 2010, 2015, 2020 and 2025. The dataset represents changes in modelled marine habitat suitability across geographic space, time and ocean depth. Species-specific MaxEnt models were fitted for 5,619 marine species using occurrence records originally obtained from the Ocean Biodiversity Information System (OBIS) and the Global Biodiversity Information Facility (GBIF). Each occurrence record was matched to the corresponding annual and depth-resolved environmental raster layers from the Copernicus Marine Service (CMEMS) according to its geographic coordinates, collection year and assigned depth. Values of five environmental predictors—seawater temperature, salinity, chlorophyll-a concentration, nitrate concentration and dissolved oxygen concentration—were then extracted for model calibration. Species range maps and depth information from the IUCN Red List were used to screen occurrence records, supplement missing depth information and constrain the spatial extent of the final predictions. Species-level predictions were generated separately for each target year and relevant depth layer using the complementary log–log output from MaxEnt, producing continuous environmental suitability values ranging from 0 to 1. Predictions were spatially constrained to each species’ original, unbuffered IUCN range. Marine AOH is therefore operationally defined here as continuous modelled environmental suitability within a species’ mapped IUCN range, rather than as a binary estimate of occupied habitat. For each year and depth layer, the range-constrained suitability values of all represented species were summed cell by cell to generate the Summed AOH raster layers. Higher values indicate locations where more modelled species, higher species-level suitability values, or both contribute to the cumulative suitability surface. Because continuous predictions were retained without applying a binary suitability threshold, Summed AOH should not be interpreted directly as observed species richness, occupancy, abundance or population density. The consistent spatial, temporal and vertical structure of the dataset supports analyses of long-term changes in marine environmental suitability, horizontal and vertical habitat redistribution, persistent suitability hotspots, potential marine climate refugia and the coverage of suitable environmental conditions by marine protected areas. Users should consider uncertainties associated with uneven occurrence sampling, static IUCN range boundaries, annual averaging of environmental variables, estimated depths for records lacking observed depth information and the omission of ecological factors such as substrate, prey availability and biotic interactions.