Biogeographical partitioning of ecological communities has been renewed in recent decades to illustrate broad distributional patterns. In the oceans, observational datasets have grown substantially and open new access to test bioregional patterns beyond the classical fixed thresholds of endemism to differentiate regions. This work combines a recently collated dataset of 29 different scientific bottom trawl surveys spanning 21 years with network-based clustering to illustrate biogeographical partitions of vast tracts of the northern hemisphere’s continental shelf seas. Our work contributes to testing bioregionalization patterns in demersal fishes using observational data, totaling >2.5 million species records and >2000 species, with bipartite network clustering weighted by species occurrence frequencies. We propose eight major bioregions across shelf seas which fall along the longest geographical axis in each shelf region and against continua of species richness gradients, endemicity, and phylogenetic turnover rates. These patterns capture known biogeographical boundaries (e.g., North Sea–Baltic Sea, Cape Hatteras) alongside potential transition areas deduced from uncertainty estimates based on shared network nodes between bioregions. The most species-rich areas include the Southeast US Shelf, Temperate Pacific, Northeast Atlantic Shelf, and the Outer European Shelf— corresponding to relatively high endemicity. However, the relatively species-poor partitions including the Baltic Sea and the North & Celtic Seas display comparatively low endemicity (<8%), illustrating apparent statistical differences in partitions captured by bipartite networks and occurrence frequencies that would otherwise be missed using a fixed endemic criterion. Our proposed bioregionalization can be compared against the growing availability of species occurrence data, dispersal limitations, or other quantitative observations of ecological communities.
