Analysing the structure and dynamics of biotic interaction networks and the processes shaping them is currently one of the key fields in ecology. In this paper, we develop a novel approach to gut content analysis, thereby deriving a new perspective on community interactions and their responses to environment. For this, we use an elevational gradient in the High Arctic, asking how the environment and species traits interact in shaping predator-prey interactions involving the wolf spider Pardosa glacialis. To characterize the community of potential prey available to this predator, we used pitfall trapping and vacuum sampling. To characterize the prey actually consumed, we applied molecular gut content analysis. Using joint species distribution models, we found elevation and vegetation mass to explain the most variance in the composition of the prey community locally available. However, such environmental variables had only a small effect on the prey community found in the spider's gut. These observations indicate that Pardosa exerts selective feeding on particular taxa irrespective of environmental constraints. By directly modelling the probability of predation based on gut content data, we found that neither trait matching in terms of predator and prey body size nor phylogenetic or environmental constraints modified interaction probability. Our results indicate that taxonomic identity may be more important for predator-prey interactions than environmental constraints or prey traits. The impact of environmental change on predator-prey interactions thus appears to be indirect and mediated by its imprint on the community of available prey.