UC Davis

Soils contain diverse and dynamic communities of microbes that interact with, and influence, microbiota in and on plants, animals, and humans. These exchanges, such as between soil and human microbiota, are part of a vast, integrated microbial ecosystem that underpins much of terrestrial life on Earth. This dissertation addresses relationships between soil microbiota and human microbiota, and their connections with human health. I approached these investigations by collecting and compiling a range of data spanning soil microbiota, child gut microbiota, health and disease factors, household characteristics, and biophysical landscape characteristics among 144 households across three rural counties in western Kenya. Using these data I first investigated the variation in gut microbial communities between children living in different rural environments, including agriculture, rural settlements, and uncultivated forest and shrubland landscapes. Although we did not find differences in child gut microbial diversity between children from these different landscapes, we found (1) significant differences in abundances of specific gut microbial taxa that are associated with health and disease, and (2) landscape-level differences in the taxa that discriminate children with a recent diarrheal episode from those without. These findings are initial evidence of a potential landscape association with the child gut microbiota, with implications for understanding childhood diarrheal disease burden. Next I investigated the relationship between the child gut microbial community and the soil microbiota in children’s households using a microbial source tracking approach. I found that the proportion of the gut microbiota attributed to soils (S2F proportion) varied among households, and in some was greater than a third, indicating a significant relationship between soil and gut microbiota at the household level. Furthermore I observed that S2F proportions were notably higher in households where I collected data later in the region’s seasonal rainy-dry cycle. Using three complementary regression modeling approaches, I examined potential correlates of S2F proportion from among the household and landscape biophysical data. I found stronger associations among soil and climate characteristics than among household and social factors, and soil pH and seasonal precipitation had the greatest association with S2F proportion. These results demonstrate the potential involvement of seasonal climate and soil properties in the relationship between the child gut microbiota and soil microbiota. Finally, I evaluated the capability of the soil and fecal microbiota to predict child and household characteristics using a series of machine-learning based classifier models. Utilizing only the fecal microbial data, the classifier successfully identified household-level landscape classes. With soil microbial data, the model successfully classified households by their landscape classes, soil classes, drinking water source, and season of sampling. A relatively small number of soil taxa were more important in the model’s performance, including taxa commonly found in the human gut. These findings indicate that a seasonal change in soil microbiota may influence the composition of the child gut microbiota in rural households. Altogether, this research presents novel evidence that household- and landscape-level exposures, including to soil microbiota, are strongly associated with the human gut microbial community, and that these associations hold potential implications for health and disease.