The concept of the marine ecological community has recently experienced renewed attention, mainly owing to a shift in conservation policies from targeting single and specific objectives (e.g. species) towards more integrated approaches. Despite the value of communities as distinct entities, e.g. for conservation purposes, there is still an ongoing debate on the nature of species associations. They are seen either as communities, cohesive units of non-randomly associated and interacting members, or as assemblages, groups of species that are randomly associated. We investigated such dualism using fuzzy logic applied to a large dataset in the German Bight (southeastern North Sea). Fuzzy logic provides the flexibility needed to describe complex patterns of natural systems. Assigning objects to more than one class, it enables the depiction of transitions, avoiding the rigid division into communities or assemblages. Therefore we identified areas with either structured or random species associations and mapped boundaries between communities or assemblages in this more natural way. We then described the impact of the chosen sampling design on the community identification. Four communities, their core areas and probability of occurrence were identified in the German Bight: AMPHIURA-FILIFORMIS, BATHYPOREIA-TELLINA, GONIADELLA-SPISULA, and PHORONIS. They were assessed by estimating overlap and compactness and supported by analysis of beta-diversity. Overall, 62% of the study area was characterized by high species turnover and instability. These areas are very relevant for conservation issues, but become undetectable when studies choose sampling designs with little information or at small spatial scales.