The southern North Sea constitutes one of the best studied and data rich, but also most exploited marine areas globally. However, still we lack sufficient knowledge of core ecological features and processes, e.g., species distribution dynamics, environmental drivers, and their links. Although meaningful management requires knowledge of the spatial structure and variability of the systems, the traditional approach sees species handled together according to the concept of ecological communities. Consequently, the scientific “handling” of the linkage between environmental drivers and biota does neither consider nor test whether the targeted communities actually exist or are just artificial artifacts created by the classification process. Crisp classifications are used to identify and define “communities” adding the further limitation of correctly setting a community border that possibly does not exist. But how valid is this approach? This is the question we tackle here. We analyse a large data set of about 1150 grab samples of benthic macrofauna collected in the German Bight. We applied fuzzy logic to provide an unsupervised classification of any degree of species association. Random Forest aided in mapping all degrees of species association and to shed light on their potential environmental drivers. Our approach overcomes the problem of crisp borders between communities. It classifies faunal associations in a continuous range from areas where one community can be well defined to areas where no community is distinguishable. One endpoint of this range is characterized by associations with highly structured interactions and dependency between species. The other endpoint is characterized by associations assembled by random processes. The German Bight benthos displays the full range of association types. Regions where random association of species occurs show higher small-scale spatial variability, which indicates higher turnover rates than areas characterized by communities. These findings raise important questions for conservation strategies. Are these dynamic areas of higher “value”? How can conservation management account for a more complex spatial pattern as well as for the different turnover rates?