Adverse respiratory effects have been observed in relation to livestock exposure. Airborne microbial agents from farms are a likely culprit. Accurate assessment of residential exposure to microbial air pollution is of great importance in epidemiological studies.

This study aimed to estimate residential exposure to four markers of livestock-related microbial emissions through random forest land use models.

Two-week averaged PM₁₀ air samples were repeatedly collected at 61 residential sites in a livestock dense region in the Netherlands. These underwent DNA extraction and qPCR analyses to quantify concentrations in ambient air of common bacterial livestock commensals (Escherichia coli (E. coli) and Staphylococcus species (spp.)), and two antimicrobial resistance genes that confer resistance to antibiotic classes widely used in the industry (tetW and mecA). Microbial concentrations were used alongside Geographic Information System-derived livestock-related characteristics of the surroundings to develop random forest-based land use models to predict year-averaged exposure.

Models showed moderate predictions of spatial variation of the four microbial agents. The models could respectively explain 38.0%, 14.4%, 30.7% and 30.2% of the spatial variation of E. coli, Staphylococcus spp., tetW and mecA at the holdout sites from 10-fold cross-validation.

Land use models could be harnessed to refine estimates of residential exposure to livestock-related emissions of microbial origin, allowing future epidemiological studies to gain greater insight into the public health relevance of exposure to microbial agents and resistant bacteria in livestock dense areas.