The pervasive computing vision encompasses scenarios where services are offered and delivered to users as a result of opportunistic encounters between their personal devices and computational resources embedded in their surrounding environment. The decentralised and dynamic nature of such environments complicates service provision, providing no setting for a conventional orchestrator to manage the resource discovery process. This paper proposes a novel approach to resource discovery in this setting, employing nature-inspired patterns to manage the search for and retrieval of information across a dynamic arrangement of devices. We show how the results of fuzzy matching based on semantic resource descriptions can be can incorporated at the pattern level to route only the best matched resources to a requestor, and how application context extrinsic to the matching algorithm may augment this process.