The results from this study suggest that the gut mycobiome of the South African population is structured by geography and lifestyle. This finding is supported by the clustering of a large proportion of the fungal OTU’s into discrete rural and urban groups within the Venn diagram. Only a small percentage of OTUs were shared between the two populations, which may suggest that factors such as the environment, age and diet may play a role in shaping the differences in OTU clustering. These results were further corroborated by NMDS and Permanova analyses, which not only show that both populations cluster distinctly and share just a few taxa, but that they have diverse fungal community composition consistent with rural and urban locality. Redundancy analysis (RDA) also predicted that the measured variables account for only a small proportion of the variability (R2 = 0.2) in fungal community composition. Indicating that other undetermined factors may be driving differences fungal community composition between the two groups.
Several studies have investigated the healthy human mycobiome [6, 7, 19, 25, 30, 33]. In these studies, geography was not considered as a potential factor structuring the gut mycobiota. For instance, previous studies found no association between host phenotypic characteristics with mycobiome profile . Nash et al. (2017) also suggests that diet, the environment, diurnal cycles, and host genetics may substantially influence the human gut mycobiome. However, the finding that the majority of the variation could not be explained by their metadata does suggest that other environmental factors, such as geography, may contribute to structuring the human mycobiome .
Our study provides the first results showing the importance of geography in African populations. Geographic locality may be associated with different environmental factors, such as different climatic regimes, which may effect structural changes in the mycobiota. For example, climate significantly influences vegetation and farming practices and leads to region specific diets. These region-specific diets may ultimately influence the gut mycobiota. This is a reasonable prediction given previous findings showing that fungi have climate dependent biogeographic patterns [34, 35]. These patterns are likely to determine the type of fungi individuals may be exposed to, which may in turn impact the colonization of fungi in the human gut. The most abundant rural-associated biomarker species found in this study, Dipodascus geotrichum, is ubiquitous in nature  whereas, Hypopichia burtonii is commonly isolated from corn, wheat, and rice . The urban-associated biomarkers were dominated by the species Dekkera bruxellensis, which are commonly isolated from fermented food such as wine, beer, feta cheese and sour dough [38-40]. In contrast, Hannaella sinensis is commonly isolated from plants and soil [41, 42]. The staple diet of the rural South African population primarily consists of a corn-based porridge (called ‘pap’). It is therefore not uncommon for a fungal species commonly isolated from corn to be a dominant biomarker for the rural population. Conversely, the urban population diet was more diverse and included fermented foods such as wine, sour dough bread and feta cheese, which are commonly available in supermarkets. Thus, the species Dekkera bruxellensis was identified as a dominant biomarker in the urban population.
In addition to geographic location, we found that smoking, mode of birth and breastfeeding significantly influenced gut fungal communities. Several studies have previously reported that these factors may significantly influence the initial colonization, subsequent composition and structure of bacterial members of the human gut microbiome [28, 43-45]. Suhr et al. (2016) and Hallen-Adams et al. (2017) investigated the gut mycobiome of two cohorts that were exclusively on a vegetarian or a western diet. These studies found that the distribution of fungi differed considerably between the two cohorts [7, 46]. Plant-associated Fusarium, Malassezia, Penicillium and Aspergillus species were detected at higher abundances within the vegetarian cohort, compared to the cohort on a conventional diet. The finding that smoking affected fungal community composition and structure is supported by several studies [47-49]. The approximately 4000 chemical compounds produced by cigarettes have been shown to alter the composition of the gut microbiome [47, 49-52]. The reported increase of Clostridia induced by smoking in murine models has also been indirectly confirmed in humans where an increased rate of C. difficile infection was greater in former and current smokers compared to never smokers . Moreover, the abundance of the fungus Candida tropicalis has also been reported to be significantly higher in C. difficile infection patients compared to healthy individuals . The abundance of C. tropicalis has also been detected to be positively correlated with levels of anti-Saccharomyces cerevisiae antibodies (ASCA) . In our study C. tropicalis was detected to be higher in individuals who smoke compared to non-smokers whereas the inverse was true for S. cerevisiae. These findings may confirm the antagonistic association between the species C. tropicalis and S. cerevisiae, as previously reported by Hoarau et al. (2016) .
Most studies have identified the genera Candida, Saccharomyces, Malassezia and Aspergillus as the three most abundant in the gut of healthy individuals [6, 25, 54]. To the best of our knowledge, our study is the first to report Pichia as one of the top four (Pichia, Candida, Cladosporium and Paraconiothyrium) most abundant genera in the human gut mycobiome. This may be due to several factors including differences in cohort characteristics (e.g., geographical location, diet, genetic predisposition and climate). Pichia have been identified as both constituent members of the human oral [55, 56] and gut microbiome . Mukherjee detected a 1:1 abundance ratio in the oral mycobiome of individuals when Candida and Pichia were present together . Pichia was also observed to have an antagonistic effect against Candida, Fusarium and Aspergillus .
The yeast genera, Pichia, Candida and Cladosporium, dominated the South African gut mycobiome. Our findings agree with previous studies which show that members of the Aspergillus, Candida, Debaryomyces, Malassezia, Penicillium, Pichia, and Saccharomyces genera were the most recurrent and/or dominant fungal genera [33, 46, 57]. In contrast to previous findings, our data indicate higher relative abundances of Cladosporium, detection of Mucor and the absence or low abundance of genera such as Cyberlindnera, and Galactomyces [6, 19, 58]. Previous studies found that the gut mycobiome of a cohort from Houston, Texas, was dominated by Saccharomyces, Malassezia and Candida . By contrast, the genus Malassezia was not detected in the gut mycobiome of a Pennsylvania cohort, which was instead dominated by the genera Saccharomyces and Candida . Differences in study methodologies may be a source of these conflicting findings . One study amplified the Internal Transcribed Spacer 2 (ITS2) region of the fungal rRNA gene , and the second amplified the ITS1 region . Studies similar to the work by Gardes et al. (1993) and White et al. (1990), where ITS1F and ITS2 primer sets were used to amplify the ITS2 region, did not detect Malassezia [59, 60]. The second reason for the observed differences has been attributed to differences in cohort characteristics, such as diet and/or geographical location. Strati (2016) and Raimondi’s (2019) investigating cohorts in Italy, detected same dominant fungal genera [58, 61], and the investigation of cohorts in two different states in the USA observed different results [6, 45]. We used ITS1 and ITS4 in this study and found that the genera Pichia, Candida and Cladosporium dominated the urban cohort, whereas genera Pichia, Candida and Aspergillus dominated the rural cohort. The dominant taxa identified in urban and rural locations further support our assertion that geographic location plays a major role in the observed differences.
Candida albicans was the most dominant taxon in our cohort and is frequently reported as the most abundant Candida species in both diseased  and healthy individuals . Candida spp. not only colonize the gut [19, 33] but several other body sites, including the oral cavity [55, 64], vagina , and skin [66, 67]. However, Candida are autochthonous to the mammalian digestive tract and species including Candida albicans, C. tropicalis, C. parapsilosis, and C. glabrata may grow and colonize at 37˚C . A review of the literature suggest that C. albicans carriage in healthy individuals ranges from 30–60%  and that living mammals are considered a niche for these species as they are not found in significant concentrations in soil, food or air [69, 70]. Raimondi et al., (2019) reported that C. albicans was frequently detected and dominated the cultivable mycobiota of different faecal samples .