The National Oceanic and Atmospheric Administration Mussel Watch Program (MWP) has assessed hundreds of contaminants using bivalves since the early 1980s. Polycyclic aromatic hydrocarbons (PAHs), the focus of this study, represent a ubiquitous suite of contaminants derived from sources such as forest fires, vehicle exhaust, oil, and roadway runoff that have been consistently measured by MWP since its inception. Many of the aforementioned sources of PAHs contain a suite of compounds that have a unique signature defined by their relative concentration. In the environment, multiple sources/signatures combine to make mixtures that can be used to distinguish between watersheds (i.e. developed, coastal, and reference). We used unsupervised artificial intelligence and combinatorial clustering to successfully identify different mixtures in PAH bivalve body burden results. Relationships between PAH mixtures, total PAH concentration, and developed land use were found. Mixtures are an integral part of the MWP contaminant framework, used to support prioritization, characterization, and forecasting of contaminants nationwide.