Statistically classified marine epifaunal bioregions on the Southern Benguela margin, South Africa

Data-driven bioregional maps have advantages over expert-derived maps, since they are able to predict the most probable bioregion at unsampled locations, are more ecologically meaningful and can incorporate measures of uncertainty. This dataset contains the site x species covariate matrix and stacked raster layers required to conduct Regions of Common Profile (RCP) analyses from Wozniak et al. (in review) to classify and map marine epifaunal, benthic invertebrate bioregions for the southern Benguela margin, South Africa. Applying an RCP approach using the ‘ecomix’ R package (Woolley, 2022), this dataset derived five marine epifaunal bioregions, including their associated spatial uncertainty, species profiles and environmental gradients. Data were collected by the Department of Forestry, Fisheries and the Environment (DFFE) and the South African Environmental Observation Network (SAEON) during annual demersal research trawl surveys between 2017 and 2020, and cover 325 trawl stations across the southern Benguela margin of South Africa ranging from 70 – 800m. Biological and environmental data associated with each trawl station are stored in a site x species counts + covariates matrix. Each observation consists of one trawl station with geospatial information provided, including the name of the survey, trawl number, a unique station number, date, trawl duration in minutes and coordinates of the trawl transect (start and end). Abundances (counts) for the 46 most commonly occurring epifaunal species are recorded for each trawl station. Zero values represent a species’ absence from that station. Values of nine environmental variables are included for each station. Temperature (˚C), pressure (db), depth (m), oxygen (ml/l), turbidity (NTU) and salinity (PSU) were recorded in-situ via a CTD attached to the headline of the trawl net and represent average values of the bottom trawl profile. Slope (˚) was derived from bathymetry compiled from digital single-beam echo-sounding data (de Wet & Compton, 2021). Substratum and sediment type, which were qualitative, were assigned to each trawl based on the national marine ecosystem classification and map (Sink et al. 2019; SANBI, 2022). Gridded multidimensional stacked raster layers of environmental variables (0.004˚ resolution) are used for spatial prediction of bioregions. After excluding highly correlated variables (Pearson correlation coefficient of r > 0.7), the environmental covariates included as predictors for modelling epifaunal bioregions were bottom temperature (˚C), bottom dissolved oxygen (ml/l) and slope (˚).