Chapter 5 Influence of Combined Sewage Overflow (CSO) on Microbial Risk at a Korean Coastal
5.3 Results and Discussion
5.3.4 Influence of environmental variables on ARGs
The abundance of bacterial 16s rRNA, intI1, and 10 ARGs at G4 from sub-daily monitoring water samples (3-hr and 5-hr intervals) was surveyed in 2018. PCA was employed to identify related ARGs, wherein two PCs were obtained (eigen values > 1) that explained 92.54% of the total variance in the dataset (Table 5.3). The first PC accounted for 64.33 % of the total variation and was participated by 16s rRNA gene, intI1, and 6 ARGs, while the second PC explained 28.21 % of the total variation which 4 ARGs (Table 5.3).
Table 5.3 Varimax rotated factor matrix from principal component analysis (PCA) for 16S rRNA gene, intI1, and 10 ARGs monitored at G4 in 2018.
Variables PC1 PC2
All 10 ARGs were detected during both sampling periods in 2018, revealing the relatively low frequency of tetD detection (63%) (Figure 5.6). Most ARGs were consistent on June 20, whereas the ARGs were elevated after the first rainfall event (6 mm) (June 27, 10:00) (Figure 5.6). The ARGs in PC1 were the most prevalent during the second rainfall event (121 mm). Tetracycline resistance genes, such as tetM (ranging from 5.40 to 8.02 log10 copies/mL) and tetQ (ranging from 3.96 to 8.10 log10
copies/mL), were the most abundant genes in PC1. These results were analogous with previous studies that reported the increased abundance of ARGs after rainfall (Di Cesare et al. 2017; Garner et al. 2017;
Yu et al. 2018b; Zhang et al. 2016). The previous study observed that tetracycline resistance genes ranged from 3 to 6 log10 copies/mL in Stroubles Creek, Virginia, USA (Garner et al. 2017), and the concentrations of 47 ARGs were between 1 and 5 log10 copies/mL in Tampa Bay, Florida, USA (Ahmed et al. 2018) in wet conditions. In comparison, the resulting peak values of tetM and tetQ in this study were much greater than previous findings.
Figure 5.6 The sub-daily absolute abundance (Log copies/mL) of 16S rRNA gene, intI1 gene, and 10 ARGs at G4 from June 10 – 29, 2018 with rainfall (mm), tide level (m), and air
The ARGs in PC2 increased at the ebb tide (14:00 on June 29), suggesting the contribution of CSO on the prevalence of ARGs (Figure 5.6). Al Aukidy and Verlicchi (2017) demonstrated the high microbiological load of CSO having relatively low volume, especially during wet period. Higher levels of antibiotic resistant bacteria were detected at CSO-impacted coastal regions (Overbey et al. 2015b).
The sul1, blaTEM, and aac(6’)-lb-cr were the most prevalent among the ARGs in PC2. Similarly, relatively high levels were observed in South Korea (Choi et al. 2008; Kim et al. 2009). In this study, the detected ranges were 4.21 to 9.32 log10 copies/mL, 3.75 to 8.57 log10 copies/mL, and 2.07 to 9.15 log10 copies/mL for sul1, blaTEM,and aac(6’)-lb-cr, respectively.
We used PCs rather than individual ARGs for further SOM analysis to decrease the number of inputs but still have a clear interpretation of the influence of environmental variables on ARGs. The relationship among the mean copies of grouped ARGs (i.e., ARGs in PC1 and ARGs in PC2), E. coli concentration, and environmental variables were analyzed using SOM. Figure 5.7 shows the SOM maps of 9 variables including 16SrRNA genes, intI1, mean ARGs in PC1, mean ARGs in PC2, E. coli numbers, tide level, 5-hour rainfall, 10-hour rainfall, and temperature. In terms of rainfall, cluster I and
II were characterized as dry days, while cluster IV depicted the highest 5-hour and 10-hour rainfalls, followed by cluster III (Figure 5.7). The SOM map of temperature showed the opposite distribution to maps of rainfall. With regards to the tide level, cluster III and II represented the lowest and highest tide levels, respectively.
Figure 5.7 SOM maps for E. coli numbers (MPN/100 mL), ARGs (Copies/mL), and environmental variables including tide level (m), rainfall (mm), and temperature (°C).
The SOM maps illustrated a high correlation between ARGs in PC2 and E. coli; highest concentrations were in cluster IV, followed by cluster III. This implies that E. coli has a high probability of carrying ARGs in PC2. On the other hand, the components in PC1 were dominant in cluster III.
Based on the map difference between ARGs in PC1 and E. coli, other host bacteria can possibly carry ARGs in PC1.
High concentrations of ARGs in PC1 were related to low tide levels with little rainfall in cluster III. Comparing the SOM maps in Figure 5.7 and the monitoring data from Figure 5.6, we can assume that the abundance of ARGs in PC1 in cluster III was influenced by delayed CSO after rainfall events.
Figure 5.7 also presents that high concentrations of ARGs in PC2 were associated with intensive rainfall in cluster IV, resulting to storm runoff and the immediate discharge of CSO. We can conclude that bacteria carrying ARGs in PC2 were majorly transported via immediate CSO (cluster IV), while other bacteria that have ARGs in PC1 existing farther from the beach were released by delayed CSO (cluster III). As a result, E. coli can be frequently found in nearby sources to the beach, resulting in the rise of ARGs in PC2 with immediate CSO during rainfall events. The highest concentrations of mean ARGs in PC1 and PC2 were about 100-fold as compared to their base concentrations in cluster III and IV, respectively (Figure 5.7).