Variations in abundance, diversity and community composition of airborne fungi in swine houses across seasons

We examined the abundance, diversity and community composition of airborne fungi in swine houses during winter and summer seasons by using quantitative PCR and Illumina HiSeq sequencing of ITS1 region. The abundance of airborne fungi varied significantly only between seasons, while fungal diversity varied significantly both within and between seasons, with both abundance and diversity peaked in winter. The fungal OTU composition was largely structured by the swine house unit and season as well as by their interactions. Of the measured microclimate variables, relative humidity, particulate matters (PMs), ammonia, and stocking density were significantly correlated with fungal OTU composition. The variation in beta diversity was higher within swine houses during summer, which indicates that the airborne fungal community composition was more heterogeneous in summer compared to winter. We also identified several potential allergen/pathogen related fungal genera in swine houses. The total relative abundance of potential allergen/pathogen related fungal genera varied between swine houses in both seasons, and showed positive correlation with PM2.5. Overall, our findings show that the abundance, diversity and composition of airborne fungi are highly variable in swine houses and to a large extent structured by indoor microclimate variables of swine houses.

(1) How does the abundance, diversity and composition of airborne fungi vary in swine houses both within and between seasons? (2) What are the major microclimate variables linked to the variations in abundance, diversity and composition of airborne fungi in swine houses between seasons? (3) What are the potential allergen/pathogen related fungal genera present in swine houses, and how does their overall relative abundance vary in swine houses both within and between seasons?

Results and Discussion
In this study, we used quantitative PCR (qPCR) and a high-throughput Illumina sequencing to extensively examine seasonal variations in abundance, community composition and diversity of airborne fungi in seven commercial swine houses in South Korea (Fig. 1). While several studies have comprehensively investigated the airborne bacterial community composition and diversity in swine houses using next-generation sequencing (NGS) methods [13][14][15] , relatively little is known about the diversity and community composition of airborne fungi in swine houses.
Microclimate variables. All the measured microclimate variables significantly varied between swine houses in winter (Table 1), and except PMs the other microclimate variables (temperature, relative humidity and air speed) also varied significantly between swine houses in summer (Table 1). We also found that except relative humidity, H 2 S, and stocking density, the other microclimate variables varied significantly in swine houses between seasons (Table 1).
Total airborne fungal abundance in swine houses. The total airborne fungal abundance (universal fungal ITS primer) measured using qPCR was significantly higher in winter compared to summer (Fig. 2). However, fungal abundance did not vary between swine houses within both seasons. The bacterial abundance was also significantly higher during winter in same swine houses 14 . One of the possible explanations for this finding could be related to reduced ventilation in swine houses during winter to avoid heat loss resulted in increased concentrations of airborne particulates 16,17 , which could lead to increase in fungal abundance during winter. Indoor air fungal concentration was shown to increase with PMs concentrations 18 . This explanation is also supported by the results of correlation analysis, which indicate a negative correlation between fungal abundance and temperature, and airspeed (Table 2), whereas CO 2 , NH 3 , and PM2.5 showed positive correlation with fungal abundance (Table 2). However, earlier studies in swine houses based on culture-depended methods reported either higher concentrations of airborne fungi in summer 19 or no significant difference in abundance of airborne fungi between winter and summer 20 . The discrepancy in our results from those of the previous studies could be attributed to the difference in the techniques used to measure the airborne fungal concentrations. Culture-independent qPCR method used in our study is more sensitive and accurate than conventional culture based methods for determining concentrations of total airborne microbes 21 .
Diversity of airborne fungi in swine houses. From the 42 samples, we observed 22,399 OTUs with an average 1,983 OTUs (range 1,524 to 2,345). The Shannon diversity index of airborne fungi was significantly different in swine houses both within and between seasons (Fig. 3). The Shannon diversity index was more variable between swine houses during summer compared to winter, however overall the Shannon index was significantly higher in winter compared to summer (Fig. 3). The bacterial diversity indices were also reported higher in winter  season in our earlier study in same swine houses 14 . Due to the lack of studies on seasonal variation of airborne fungi in swine houses, we could not compare directly these results with others studies. However, our results are similar to the results obtained in indoor environment of residential buildings by Adams et al. 22 , who also observed higher fungal diversity in winter season. We found significant positive correlation between fungal diversity and PM2.5 and PM10 concentration ( Table 2), and concentration of these particles were higher in winter season (Table 1). These results together indicate one possible explanations of the high diversity in winter, which could be that to maintain the indoor air temperature during winter, all of the openings are closed and the ventilation rates are reduced to minimal, which in turn increases the indoor bioaerosol particles and thus increases microbial diversity. The indoor fungal diversity information might be useful in terms of health and exposure evaluations of animals and farmers working in swine houses, as indoor fungal diversity have been shown to be associated with asthma development 23,24 .
Dominant airborne fungal taxa in swine houses. The most abundant fungal phyla across all of samples were Ascomycota, representing 75.4% of all sequences, followed by Basidiomycota (15.3%), Zygomycota (4.2%), and Glomeromycota (1.5%) (Fig. 4). Similar to the diversity, the relative abundance of the dominant fungal phyla also varied significantly in swine houses both within and between seasons ( Table 3). The relative abundance of Ascomycota was significantly higher in summer (P < 0.01; Table 3), whereas the relative abundances of Basidiomycota and Zygomycota were significantly higher in winter (P = 0.01; Table 3). The relative abundance of Ascomycota and Basidiomycota showed negative and positive correlation, respectively with PM2.5 and PM10 ( Table 2). The growth forms of Ascomycota are small enough to become easily aerosolized compared to the large growth forms of Basidiomycota 25 , this might be the explanation of dominance of Ascomycota in indoor air of swine houses across both seasons. However, the relative abundance of Ascomycota had declined in winter at  Table 2. Spearman rank correlations between microclimate variables and diversity indices, the relative abundance of fungal phyla and pathogen/allergen related genera. * P < 0.05; * * P < 0.01; * * * P < 0.001. expense of increase in relative abundance of Basidiomycota, which might have enriched due to high concentration of indoor airborne particles during winter. Comparisons between swine houses were also conducted at class and genus levels. The predominant fungal classes detected in this study were Dothideomycetes and Sordariomycetes of phylum Ascomycota, which has been shown to dominate aerosol samples in several previous studies [26][27][28] . Dothideomycetes class is known to contain several allergenic fungal taxa [29][30][31] . Agaricomycetes was the most abundant class of phylum Basidiomycota, which does not generally contain described human allergenic/pathogenic fungal taxa. Although the relative abundance of all dominant fungal classes varied significantly in swine houses within seasons (Table S1), the relative abundance of only four dominant classes (Agaricomycetes, Pezizomycetes, Eurotiomycetes, and Tremellomycetes) varied significantly between seasons (Table S1). Of these, Agaricomycetes, Pezizomycetes, and Tremellomycetes were more abundant in winter, whereas Eurotiomycetes was more abundant in summer. The predominant fungal genera detected were Clavaria and Fusarium. Most of the Clavaria species are believed to be saprotrophic and probably originated from the litter bedding material, whereas genus Fusarium have been shown to dominate indoor environment of swine houses 6,10,32 . Similar to dominant fungal phyla and families, the relative abundance of 30 most   dominant fungal genera also varied in swine houses both within and between seasons (Fig. 5). The relative abundance of most of the dominant fungal classes and genera were significantly correlated with the concentration of PMs and stocking density (Table S2). These results indicate the potential role of PMs concentration in structuring the airborne fungal community even at lower taxonomic levels.
Fungal OTU composition and effect of microclimate variables. The PerMANOVA results showed that the airborne fungal OTU composition was influenced by both swine house unit (F = 4.25, P < 0.0001) and season (F = 6.25, P < 0.0001). The interaction between swine house unit and season also impacted significantly the fungal OTU composition (F = 3.5, P < 0.0001). Seasonal difference in bacterial OTU composition was also observed in our previous study in same swine houses 14 . Furthermore, similarly to our findings in the present study, Adams et al. 33 demonstrated that both residential unit and season largely influenced the indoor airborne fungal communities. The beta diversity of indoor airborne fungi varied significantly in swine houses between seasons, with summer having significantly higher beta diversity than winter (P < 0.05) (Fig. 6). The variation in beta

Figure 6. Community variance (beta diversity) of airborne fungal communities of swine houses both within and between seasons.
An asterisk indicates statistical significant at P < 0.05, and different letters represent means that are significantly different at P < 0.05 (Tukey's HSD test).
diversity among replicates within swine houses was also high during summer, which suggests that the airborne fungal community composition is more heterogeneous in summer than in winter. The RDA analysis showed that relative humidity, PMs (PM2.5 and PM10), NH 3 , and stocking density best explained the variation in community composition of airborne fungi in swine houses across seasons (Fig. 7). Relative humidity and stocking density is shown to affect the aerosolization of fungal particulates [34][35][36] , which in turn could affect the composition of airborne fungi. Whereas it has been shown that NH 3 is one of the main precursors of secondary PMs 37 , which could then affect the community structure of airborne fungi.
Potential Allergen/pathogen related fungal genera in swine houses. Potential allergen/pathogen related fungal genera were also examined in swine houses. A total of 80 allergen/pathogen related fungal genera are listed by Simon-Nobbe et al. 38 and of these, we identified a total of 29 potential allergen/pathogen related fungal genera. The relative abundance of total potential allergen/pathogen related fungal genera varied from 13.0-24.9%. Overall, the relative abundance of potential allergen/pathogen related fungal genera varied significantly between swine houses during each season (Fig. 8), however their relative abundance did not vary significantly between seasons (Fig. 8). The most abundant potential human allergen/pathogen related fungal genus detected was Fusarium (10.8%). Fusarium is an emerging fungal pathogen and can cause infections in humans, especially  in immunocompromised hosts 39,40 . The relative abundance of potential allergen/pathogen related fungal genera was positively correlated with PM2.5 concentrations ( Table 2). The potential allergen/pathogen related genera deposited on PM2.5 could cause respiratory problems in both farm workers and animals, as PM2.5 can penetrate and deposit deeper in the tracheobronchial and alveolar regions 41 . These results, however, should be taken with the caveat that the taxonomy of an organism does not necessarily provide the information about the allergenic/ pathogenic level of that organism, so the information reported here on allergenicity/ pathogenicity of fungal genera is solely of 'potential' character.

Conclusions
In conclusion, our results suggest that the swine house unit and season and their interaction influenced the diversity and community composition of airborne fungi. The indoor air fungal abundance and diversity were significantly higher in winter than in summer. The abundance, diversity and composition of airborne fungi were significantly correlated with microclimate variables, mainly by the relative humidity, PMs, and stocking density. Several potential allergen/pathogen related fungal genera were also observed, and their total relative abundance varied between swine houses in both seasons. These potential allergen/pathogen related fungal genera present in the indoor air of swine houses and their association with PM2.5 could impact the health of farm workers and animals. Overall, this study provides a better understanding of abundance, diversity and community composition of airborne fungi in swine houses across seasons. Microclimate variables. The microclimate variables were measured several times during the sampling period from the same three aerosol collection points, and the average values were reported corresponding to each sampling point. A hygrothermograph (SK-110TRH, SATO, Tokyo, Japan) was used to measure the air temperature and relative humidity. We used an anemometer (model 6112, KANOMAX, Osaka, Japan) to measure the air speed. An aerosol mass monitor (GT-331, SIBATA, Soca-city, Japan) was used to measure the concentrations of PM2.5 (mean aerodynamic diameter ≤ 2.5 μ m) and PM10 (average aerodynamic diameter ≤ 10 μ m). Gas detector tubes (Gastec Co., Ltd., Kanagawa, Japan) were used to measure the concentrations of ammonia (NH 3 ), hydrogen sulfide (H 2 S) and carbon dioxide (CO 2 ). DNA extraction and universal fungal qPCR. The PowerSoil DNA isolation kit (MoBio Laboratories, Carlsbad, CA) was used to extract DNA from filters by following the initial processing methods as described in Kumari et al. 14 . The purified DNA samples were used for quantitative PCR (qPCR) analysis to quantify the copy numbers of fungal ITS. The ITS1 region was targeted with the universal fungal primers ITS1F and ITS2 42,43 . Each of 20 μ L reaction mixtures contained 1× Fast SYBR Green Master mix reagent (Clontech Laboratories, Inc., Mountain View, CA, USA), 10 μ M forward and reverse primers, and 1 μ L of template DNA. Quantitative PCR was performed on the 7300 Real-Time PCR System (Applied Biosystems, Inc., Foster City, CA, USA). Amplification was initiated with a 15 min denaturation at 95 °C, followed by 45 cycles of a 15 sec dissociation at 95 °C and a 1 min annealing and extension at 60 °C. Standard curves were generated based on a standard ITS1 amplicon with a known copy number concentration. The standard amplicon was prepared with conventional PCR using the universal primers ITS1F/ITS2 and DNA extract from Aspergillus fumigatus ATCC MYA4609 as a template. The concentration of the standard amplicon was quantitated by the Quant-iT PicoGreen dsDNA reagent kit (Life Technologies, Carlsbad, CA, USA). The amplicon was serially diluted from 1 to 10 6 copy number μ L −1 to calibrate qPCR. Each qPCR measurement was in triplicate. PCR inhibition was checked with the method reported elsewhere 44 . No inhibition was found. In this study, 10% of DNA extraction efficiency was assumed to calculate copy number per 1 m 3 of air 44 , though the efficiencies might be different across the studies since the different extraction methods were used. However, the comparison was possible for the samples collected within this study since the same extraction method was used for all the samples collected within this study.

Methods
PCR amplification and sequencing. We amplified the ITS1 region using primer pairs ITS1FI2, 5′ -GAACCWGCGGARGGATCA-3′ 45 and ITS2, 5′ -GAACCWGCGGARGGATCA-3′ 43 . Amplifications were carried out in a total volume of 30 μ L composed of 1 or 2 μ L of template DNA, 1× PCR Master Mix (Takara Bio, Shiga, Japan), and 0.3 μ M of each primer. To account for the stochasticity of PCR reactions amplification was performed at two annealing temperatures (52 °C and 55 °C) in three replicates for each annealing temperature 45 . Schmidt et al. 45 suggested that using multiple annealing temperatures could decrease the primer binding bias, and enable the recovery of more complete fungal communities. PCR was carried out under the following conditions: initial denaturation step for 15 min at 95 °C, followed by 30 cycles of 30 sec at 95 °C, 30 sec at either 52 °C or 55 °C, and 30 sec at 72 °C, and a final elongation step for 5 min at 72 °C. The DNA extracted from sterilized filters kept in − 20 °C was served as negative controls. The negative controls did not show any amplification. The amplicons were Scientific RepoRts | 6:37929 | DOI: 10.1038/srep37929 purified using the QIAquick PCR purification kit (Qiagen, CA, USA), and sent to the Beijing Genome Institute (BGI) (Hong Kong, China) for sequencing using 2 × 150 bp Hiseq2000 platform (Illumina, San Diego, CA, USA).
Sequence processing. The paired-end ITS1 sequences were assembled using Pandaseq software 46 , followed by removal of the flanking rRNA gene fragments from ITS1 region using ITSx version 1.0.9 47 . The ITS1 sequences were further processed in mother version 1.36.1 48 . The chimeric ITS1 sequences were detected and removed with mothur's implementation of uchime in de novo mode 49 . Taxonomic assignments of ITS1 sequences were performed using BLASTn ver. 2.2.19 50 against a named fungal ITS sequences database 51 , and further classified using a fungal taxonomic identification tool FHiTINGS 52 . The sequences were not clustered into operational taxonomic units (OTUs) before taxonomic assessments as clustering process may reduce the taxonomic coverage 53 . The taxonomic identifications of ITS sequences are uncertain at the species level 54 , therefore the taxonomic identifications were performed only down to the genus level. The potential allergen/pathogen related fungal genera were identified using a list of known fungal allergen/pathogen related fungal genera 27 . The ITS1 sequences were clustered into OTUs with a threshold of 97% sequence similarity using the QIIME implementation of UCLUST 55 . All sequence data are deposited in the MG-RAST server 56 under MG-RAST IDs 4633146.3-4633187.3.

Statistical analysis.
Prior to the calculation of diversity indices, all the samples were rarified to 24,000 reads per sample by random subsampling using the 'sub.sample' command in mothur. The Bray-Curtis distance was used to calculate the OTU-based community dissimilarity 57 . The differences in microclimate variables, fungal abundance, OTU richness, Shannon index and the relative abundance of the most abundant phyla between swine houses within seasons were analyzed using ANOVA followed by Tukey's HSD test with Benjamini and Hochberg correction to adjust the p-values 58 . Adjusted p-values of less than 0.05 were considered statistically significant. Furthermore, the seasonal differences in these parameters were evaluated using t-test. We performed Spearman rank correlation to test the relationship between microclimate variables (temperature, relative humidity, air speed, NH 3 , H 2 S, CO 2 and stocking density) and microbial measurements (fungal abundance, diversity index and relative abundance of fungal taxa). A heatmap of the 30 most abundant fungal genera was constructed using the 'pheatmap' package in R.
To test how the fungal community composition is influenced by the swine house unit and season we performed a permutational multivariate analysis of variance test (PerMANOVA, 'adonis' function in 'vegan' R package 59 . We used redundancy analysis (RDA) to test which microclimate variables best explain the variation in fungal community composition using Canoco 5.0 (Biometrics, Wageningen, The Netherlands), applying forward selection and the Monte Carlo permutation test with 999 random permutations. We also performed a permutational dispersion analysis to test whether beta diversity is significantly different in swine houses both within and between seasons by using the betadisper function in the 'vegan' R package 60 . All statistical analysis, graphs, and ordinations were produced using R version 3.0.2 61 .