The mean concentrations of bacteria and fungi for all ambient stations (ST2-ST5) were computed by season. The mean concentrations ± SDs of bacteria were 869 ± 1003 CFU m⁻³, 1263 ±1955 CFU m⁻³, and 1193 ± 1497 CFU m⁻³ for winter, spring and summer, respectively. An ANOVA test indicates that there was no significant difference in ambient bacterial concentrations among three seasons (p = 0.33). The mean concentrations ± SDs of fungi were 280 ± 373 CFU m⁻³, 1403 ± 1461 CFU m⁻³, and 2558 ± 2276 CFU m⁻³ for winter, spring and summer, respectively. Mean ambient fungal concentrations were the highest in summer and the lowest in winter (p < 0.0001).

The lowest fungal concentration in winter may be due to the inactivation of fungi at low temperature and RH levels. Fungal concentrations were higher in summer than in spring although solar radiations were significantly higher in summer than in spring (p < 0.0001). Bacterial concentrations did not vary with season although atmospheric conditions changed significantly (Table 2).

Bacterial and fungal concentrations at each ambient station were averaged by season to illustrate the seasonal effects at different stations. Table 3 and Table 4 show the mean values of bacterial and fungal concentrations at four ambient stations (ST2-ST5) in three seasons (winter, spring and summer). As shown in these tables, seasonal variations of bacterial concentrations varied at ambient stations. At ST2 and ST5, bacterial concentrations were the highest in spring and the lowest in winter, while at ST3 and ST4, bacterial concentrations were the lowest in spring and higher in winter and summer. This observation may suggest that wind played an important role in affecting ambient airborne bacterial concentration because there was significant difference (p < 0.0001) between concentrations at downwind locations and upwind locations. Unlike bacterial, fungal concentrations showed coincident pattern of seasonal variations at all four ambient stations. Fungal concentrations were the lowest in winter and increased from winter to summer.

Ambient temperature and RH changed not only with season, but also with time of day on each sampling day. In

^*Mean of 166 measurement points.

^*From the two-way ANOVA test, it is shown that the weather and location were not significant (p = 0.46 and 0.25, respectively), but the interaction between them was significant (p = 1.83E−8).

^*From the two-way ANOVA test, it is shown that the weather was significant (p = 3.28E−13), while the location and the interaction between them were not significant (p = 0.43 and 0.10, respectively).

order to estimate the time effect on ambient bioaerosol concentrations in different seasons, mean concentrations of bacteria and fungi at different times of day were computed by season. Table 5 and Table 6 list the mean concentrations of bacteria and fungi at ambient locations (ST2-5) in three seasons at different times of day. In winter, due to the low temperature and RH, concentrations of bacteria and fungi were low and changed only

^*Mean was calculated using measured concentrations at all four ambient stations. ^**From the two-way ANOVA test, it is shown that the time, weather, and the interaction were not significant (p = 0.99, 0.30, and 0.66, respectively).

^*Mean was calculated using measured concentrations at all four ambient stations. ^**From the two-way ANOVA test, it is shown that the weather was significant (p = 2.44E−12), while the time and the interaction between them were not significant (p = 0.14 and 0.57, respectively).

slightly during the day. In spring, there was no consistent pattern observed for both bacteria and fungi. Bacteria concentrations were the lowest in afternoon, while fungi concentrations were the lowest in early morning and at noon. In summer, bacteria and fungi variations showed a similar pattern.

To investigate diurnal time effects on bacterial and fungal concentrations at different locations, mean concentrations of bacteria and fungi at different times of day were calculated by station. Table 7 and Table 8 show the time of day effects at different ambient stations. Bacterial and fungal concentrations over time varied greatly at different stations. At ST2 and ST3, bacterial concentrations were higher in the afternoons than mornings. At ST4 and ST5, bacterial concentrations were higher in the mornings than afternoons. At ST2 and ST4, fungal concentration decreased from morning to afternoon. At ST3, fungal concentrations were the highest at noon. At ST5, fungal concentrations were the lowest at noon. Bacterial and fungal concentrations displayed variations with time at different stations, which was mainly because of the spatial effects.

To investigate spatial effects, data were grouped into the upwind and downwind classes. The overall mean concentrations of bacteria and fungi at downwind locations were 1856 ± 1688 CFU m⁻³ and 1155 ± 1179 CFU m⁻³, respectively. The overall mean concentrations of bacteria and fungi at upwind locations were 291 ± 606 CFU m⁻³ and 1690 ± 2321 CFU m⁻³, respectively. Mean concentration of bacteria was significantly higher downwind than upwind (p < 0.0001). In contrast, mean concentration of fungi at downwind was not significantly different from upwind (p = 0.1206). This observation indicated that wind vector (i.e., combined effects of wind direction

^*Mean was calculated using measured concentrations over three seasons at any given station stations. ^**From the two-way ANOVA test, it is shown that the time, weather, and the interaction between them were not significant (p = 0.94, 0.53, and 0.72, respectively).

^*Mean was calculated using measured concentrations over three seasons at any given station stations. ^**From the two-way ANOVA test, it is shown that the time, weather, and the interaction between them were not significant (p = 0.94, 0.53, and 0.72, respectively).

and wind speed) played an important role in affecting bacterial and fungal concentrations in the vicinity of the farm.

The overall mean concentrations of bacteria and fungi at downwind locations were computed by season. Table 9 shows concentrations of bacteria and fungi at downwind location in three seasons. Downwind concentrations of bacteria were not significantly different in three seasons (p = 0.51). Fungi concentration at downwind were the highest in summer and the lowest in winter (p < 0.0001). The variations in downwind bacterial and fungal concentrations followed the same patterns as the overall mean bacterial and fungal concentrations in the vicinity of the farm did.

The average downwind and upwind concentrations of bacteria and fungi at each ambient station were calculated to illustrate the spatial variation of their concentrations at different locations (Table 10). Mean concentrations of bacteria at each station were higher downwind than upwind (ST2: p = 0.0005, ST3: p = 0.0004, ST4: p = 0.0002, ST5: p = 0.0016), which was in agreement with the variations of overall mean ambient bacteria concentration. The mean concentrations of downwind bacteria at four ambient stations were not significantly different (p = 0.70). Similarly, there was no significant difference among the mean concentrations of upwind bacteria at four ambient stations (p = 0.36). In contrast, there were no significant differences between upwind and downwind fungi concentration at any station but ST3 (ST2: p = 0.64, ST3: p = 0.0019, ST4: p = 0.15, ST5: p = 0.92). This observation suggests that wind direction and speed seemed to have lower impact on fungal concentration variations than bacteria. The mean concentrations of downwind fungi at ST5 and ST2 were higher than those at ST3 and ST4 (p = 0.03). The mean concentrations of upwind fungi at four ambient stations were

^*Means with the same letter are not significantly different.

^*Means with the same letter are not significantly different.

not significantly different (p = 0.28). The variations might be mainly because of the location of each station. As shown in Figure 1, ST2 was the furthest away from the production houses and close to a local road and a waste water spray field. There was also a line of trees between ST2 and the houses that might block transport of bioaerosol from the production houses to ST2. However, ST2 had a higher mean value of fungi downwind concentration. It was probably that the spray field, instead of the farm, was the largest source of fungi for ST2. ST3 was very close to the cross-ventilated house (House 6), and the fans of house 6 could blow bioaerosols directly toward ST3. ST4 was close to a large waste treatment lagoon. ST5 was further away from the production houses than ST3 and ST4. In addition, there was an egg packing plant located between the production houses and ST5. However, upwind bacteria and upwind fungi concentrations at ST5 were not lower than other ambient stations, and downwind fungi concentrations were even higher than ST3 and ST4. There might be other source of airborne fungi that had impact on bioaerosol concentrations at ST5. The upwind concentrations should be considered the concentrations that were impacted by the sources other than the house emissions.

In comparison to bacterial and fungal concentrations at source (ST1), there were more factors affecting the concentrations at ambient stations in the vicinity of the farm. While various meteorological variables and emission rate were selected to examine their impact on downwind concentrations, distance from the source was not due to lack of resource and accessibility to the neighboring properties. By all means this fact should be included in future studies of fate and transport of bioaerosols.

Figure 2 and Figure 3 illustrate the impacts of meteorological factors and emission rate on ambient bacterial and fungal concentrations at downwind locations.

As shown in Figure 2, no obvious linear relationships between downwind bacterial concentrations and air temperature (Figure 2(a)), RH (Figure 2(b)), or solar radiation (Figure 2(d)) were observed from the plots. High concentrations of bacteria were more likely to be detected at temperatures from 20˚C to 40˚C; medium to high RH levels (40% to 80%) resulted in higher bacterial concentrations. Higher concentrations of bacteria were detected with moderate to strong solar radiation (>300 W/m²). Bacterial concentrations were observed to be positively related to wind vector (W) (Figure 2(c)) and emission rate (Figure 2(e)). On the other hand, downwind

fungal concentrations were positively related to T, RH, solar radiation, and emission rate (Figure 3(a), Figure 3(b), Figure 3(d)), and were negatively related to wind vector (W) (Figure 3(c)). Both bacterial and fungal concentrations were found to be higher at stronger solar radiation (Figure 2(d), Figure 3(d)), although solar radiation was reported to cause bioaerosol damage [15] . This might be because that strong solar radiation with low wind speed (<5 m/s) resulted in unstable to very unstable atmosphere, and the consequent unstable atmosphere caused high bacterial and fungal concentrations at local downwind locations (<2 km away from the sources).

For bacterial concentration analysis, the linear regression analysis suggests that downwind bacterial concentration was only linearly dependent on wind vector, W (p = 0.0046) at a significant level of 0.05. However, it did not mean that other influencing factors (i.e., T, RH, solar radiation, and emission rate) were not important in impacting downwind bacterial concentration, because there were strong linear dependencies between them. Tests of the linear dependencies between the six variables indicate that T had significant linear dependencies with RH (p = 0.0022), solar radiation (p < 0.0001), and emission rate (p < 0.0001); RH had significant linear dependencies with T, wind vector W (p = 0.0022), and emission rate (p = 0.0169); wind vector W had significant linear dependencies with RH; solar radiation had significant linear dependencies with T and emission rate (p = 0.0002); and emission rate had significant linear dependencies with T, RH, and solar radiation.

For fungal concentration analysis, Pearson correlation coefficients suggested that downwind fungal concentration was significantly linearly dependent on all five influencing factors at a 0.05 level. Similarly, there were strong linear dependencies between the influencing factors. Temperature had significant linear dependencies

with RH (p = 0.0062), solar radiation (p < 0.0001), and emission rate (p < 0.0001); RH had significant linear dependencies with T and wind vector, W (p = 0.0488); wind vector W had significant linear dependencies with RH; solar radiation had significant linear dependencies with T and emission rate (p < 0.0001); and emission rate had significant linear dependencies with T and solar radiation.

The linear correlation between variables might cause some predictors to be not important when fitted in a linear regression model. To better study the impacts of influencing factors on downwind bacterial and fungal concentrations, log transformed bacterial and fungal concentrations were used as new responses. As suggested by residual plots (not shown), emission rate was also log transformed. Quadratic forms of wind vector (W²) and emission rate (emission²) were introduced as two new predictors for predicting downwind bacterial concentration. Among all subset models, the model included T, W, W², solar radiation, emission rate, and emission² was preferred by R-square and Mallow’s CP selections. For predicting downwind fungal concentration, residual plots suggested that the relationships between fungal downwind concentration and emission or wind vector (W) were not quadratic. Among all subset models, the model included every influencing factor but solar radiation was preferred by R-square and Mallow’s CP selections. Table 11 lists the results of the selective regression models. As shown in this table, wind vector, W had different effects on bacterial and fungal concentrations. On the one hand, high speed wind might bring more bacteria and fungi to a sampling station more quickly; on the other hand, strong wind might also cause damage to bacteria and fungi. Therefore, increasing wind vector resulted in increasing in bacterial concentrations and decreasing in fungal concentrations. For the influencing factors that were not included in the regression models, it does not mean that they were not important. The results

^*Model DF = 6, total DF = 85, R² = 0.5221; ^**Model DF = 4, total DF = 85, R² = 0.7481.

of the models only indicated that when other factors were controlled, RH was no longer an important predictor of bacterial concentration downwind while solar radiation was no longer an important predictor of fungal concentration downwind. Due to the differences in size and specie characteristics, bacteria and fungi responded differently to the influencing factors. Temperature, wind vector, and emission rate were important predictors for both bacterial and fungal variations.