INNOVATIVE SAMPLING AND SURVEILLANCE METHODS IN THE AIR MICROBIOME

Abstract

The air microbiome is a growing community of increasing interest for both public health and One Health. Various types of particles, both organic and inorganic, can be transported in the air. Subsequently, this exposes humans, animals, and environments to these particles. This dissertation investigates the air microbiome, innovating novel sampling methods to sample in a vertical profile to investigate how altitude influences microbial distribution, as well as what different types of microbes are present. Additionally, this dissertation also implements a One Health framework to predict the location of pathogens of One Health interest that may either exist in or use the air microbiome for transport. This predictive work could help guide future sampling efforts for difficult to environmentally detect pathogens. In Chapter 2, we used drones and a novel air sampler to improve both flight time and sample collection to aid future research efforts. Drone platforms are an emerging tool for scientific research, including air sampling. However, a challenge arises when the weight of air samplers impairs drone performance by approaching or exceeding payload capacity. This chapter addresses this limitation by comparing the SASS 3100 with a lightweight prototype designed for drone use across three experimental scenarios. The full-sized SASS 3100 and Light SASS prototype were compared across three environments: in a controlled chamber, outdoor ground level conditions, and elevated outdoors positions (boom lift for Full SASS, and quadcopter and hexacopter drone for the Light SASS). Samples were processed by extracting and quantifying the total DNA using a Qubit 4.0 fluorometer, and assessment of microbial diversity via next-generation sequencing. Across most conditions, no significant differences were observed between samplers. An exception was observed in the elevated hexacopter trials, where differences in total DNA collected were detected, and in chamber experiments when comparing gene copies of Pseudomonas syringae recovered from aerosol. Diversity analyses showed there were no significant differences in the microbial diversity recovered from the filters on the different samplers used. Overall, these results indicate that the Full SASS and Light SASS samplers perform similarly under chamber, ground, and quadcopter-mounted conditions, but may differ when deployed on a hexacopter. Additional work is needed to better understand the influence the hexacopter operational field has on sampler bioaerosol recovery. In Chapter 3, we expanded on the use of drones as an air surveillance platform. This chapter addresses the need for active microbial surveillance in the Planetary Boundary Layer (PBL) to mitigate risks to human, animal, and environmental health. While airborne transport of pathogens is well-documented, current research often lacks practical surveillance applications for real-time public health action. We demonstrate the utility of a drone-mounted air sampling system for conducting vertical microbial profiles within the PBL to bridge this gap. Conducted in Big Spring, Texas, we utilized a DJI Matrice 350 drone equipped with a SASS 3100 dry air sampler at altitudes of 100, 200, and 400 ft AGL, alongside ground-level controls. Microbial communities were characterized through 16S rRNA and ITS amplicon sequencing. Results from PERMANOVA and diversity analyses indicate that altitude and site location are primary drivers of microbial community variance. Significant differences in alpha and beta diversity were observed between ground-level and samples at altitude. Notably, several genera of One Health interest, including the human pathogen Legionella, showed significant enrichment at higher altitudes compared to ground level. Fungal communities also exhibited distinct stratification, with specific taxa such as Fusarium showing enrichment with altitude. The findings validate the use of readily attainable drone technology to provide high-resolution data on aerosolized microbial communities. This approach equips public health officials with a scalable tool for monitoring pathogen transport, informing exposure assessments, and implementing timely communication strategies to protect vulnerable populations. In Chapter 4, we used ecological niche modeling to predict where to deploy additional surveillance. Coccidioidomycosis (Valley Fever), caused by the soil-borne fungi Coccidioides immitis and C. posadasii, is an escalating public health threat in the Southwestern United States. Traditional environmental surveillance is hindered by the low reliability of soil sampling, while human case data are often confounded by long incubation times, resulting in spatial and temporal misalignment due to case reporting clustering in urban centers with exposure most likely occurring elsewhere. Other mammals, such as canines, also contract Valley Fever at comparable rates and symptoms to humans, making them ideal sentinels as they also live in close proximity with humans. Furthermore, canines do not travel as much as humans and are assumed to be more likely to have been exposed within their local areas. This chapter proposes a One Health framework to model Coccidioides spp. habitat suitability in Arizona (2004–2023) by utilizing both human surveillance data and simulated canine cases as proxy presence points for the pathogen. We integrated confirmed Valley Fever human case data from the Arizona Department of Health Services with canine cases simulated based on decennial census and veterinary survey data. Using an ensemble modeling approach incorporating Random Forest, Boosted Regression Trees, and Maximum Entropy, we combined both reported human case data and simulated canine case data with environmental and meteorological predictors, including soil moisture and temperature, air temperature and humidity, soil pH, sand fraction, and land cover, to model relative habitat suitability for Coccidioides spp. Bias weighting and spatial thinning were applied to mitigate the influence of human population density. Human-only models demonstrated strong performance (AUC 0.7–0.9) but exhibited significant spatial bias toward urban centers (Phoenix and Tucson). Conversely, canine models, while yielding lower AUC scores (0.6–0.7), provided superior resolution in non-urban areas and identified suitable habitats away from human population hubs. Combined One Health models identified expanded suitability in Northern and Northeastern Arizona, aligning with recent increases in Valley Fever cases and environmental DNA detections. Integrating simulated canine Valley Fever sentinel data into fungal pathogen modeling addresses some limitations of human case reporting. This approach provides a more accurate representation of Coccidioides spp. distribution, offering a scalable framework to improve surveillance and public health risk communication in both endemic and emerging regions.