The extended Weddell Sea (WS) shelf region, including the Antarctic Peninsula, is increasingly threatened by effects of climate change. In order to establish conservation strategies and forecast the benthic diversity under changing environmental conditions, it is crucial to understand the benthic community composition, distribution and their relationships to abiotic drivers. However, the limited accessibility in such remote areas results in scarce biological data, forcing the use of environmental surrogates, used for defining habitats (Jerosch et al. 2018), to identify infauna distribution. Polychaetes are a dominant faunal group in the WS soft-bottom ecosystems, contributing up to 50% to the total macrobenthic abundance (Säring et al. 2022), with a high functional diversity. However, their distribution patterns, particularly in relation to ecological drivers in the WS are poorly understood. Here, we describe polychaete communities including their taxonomy and functional identity at the Antarctic Peninsula and the Filchner Trough region related to sea-ice cover and benthic food regimes. We further present the attempt to fit their distribution to bioregionalization based on environmental parameters. We used point data of fauna, sediment (grain size, TOC, TN, pigment content) and of the water column (temperature, salinity, chlorophyll a) from three expeditions (PS81, PS96, PS118) with the RV Polarstern, ice-cover data (2010–2019) extracted from remote sensing imagery, as well as nine environmental raster data sets (e.g. sea-ice cover, TOC, current speed). We observed 34 polychaete families that were grouped into 14 functional groups based on categories. Using cluster analysis we identified 6 taxonomic and 5 functional community types. Ice-cover variation and TOC were identified as the best suitable environmental parameters explaining the variation of both taxonomic (39%) and functional (45%) community compositions. Although the four bioregions defined by the k-means cluster algorithm could not explain the complex distribution patterns of the taxonomic or of the functional communities, we could highlight potentially vulnerable areas across the WS, e.g. the Filchner Trough region with heterogenous community compositions. We assume that the different resolution of input data, and insufficient fauna data density compared to vast survey areas were limiting factors to run reliable models combining biological and physical information. Our findings underscore the relevance of filling spatial gaps of infauna sampling and environmental data to apply advanced models, in order to specify reliable conservation strategies for vulnerable areas.