Small-scale distributions of polycyclic aromatic hydrocarbons in urban areas using geospatial modeling: a case study using the moss Orthotrichum lyellii in Portland, Oregon, U.S.A.

Abstract Polycyclic aromatic hydrocarbons (PAHs) are widespread pollutants that are costly to measure with air quality instruments. We used a spatially balanced sample of an epiphytic moss, Orthotrichum lyellii, to develop fine-scale maps of atmospheric PAHs across residential areas of Portland, Oregon. The unusual abundance of this stress-tolerant species enabled us to systematically collect a large sample (n = 350) on a 1x1km grid across the city. This provided the unique opportunity to investigate potential effects of more than 20 factors on PAH concentrations in moss, including emissions sources, tree canopy cover, relative elevation, short-term weather, and nuance of sample collection. We measured 20 PAHs in moss samples and assigned them to 4 groups for analysis based on molecular weight (MW): the Napthalene, Low, Medium, and High MW groups. Spatial regression models explained a modest amount of variation in PAH concentrations in moss, with spatial structure increasing as MW increased. Motor vehicle emissions were the only PAH source we detected and was highly significant in all models. Deciduous tree cover correlated with lower PAH concentrations in moss for all except the high MW group. Short-term weather and nuance of sample collection, like height on the sampled tree, tree taxonomic family, and whether sample was primarily collected from a branch or bole, had surprisingly large effects on all groups. Overall, results aligned with expectations based on each group’s dispersal potential due to MW. All PAH groups were highest over downtown Portland, along major highways, and lowest in parks and outer neighborhoods, although the details varied. Our results supported two common themes in studies directly measuring PAHs: the importance of roads as an emissions source in residential areas, and the potential for trees to reduce atmospheric concentrations. Moreover, the large effects of weather and sampling nuance have critical implications for the design of future studies.