3.3. Trace Metals
Trace Metals in Deposited Soil Samples. Figure 7 (A through D) shows the measured metal concentrations in the deposited soil collected from the banks of Buffalo and Brays Bayous (arsenic and cadmium were plotted as one group (Figure 7B and D) and the other four metals as another group to account for the different ranges in concentrations). No significant spatial trend in trace metal concentrations was found for the deposited soil in either of the two bayous. Measured metal concentrations were below both EPA’s RSLs and TCEQ’s PCLs except for arsenic. In four out of six Brays Bayou and two out of five Buffalo Bayou stations, arsenic levels were above EPA’s RSLs and below TCEQ’s PCL. One station with commercial land use (Brays Bayou at Broadway Street (B6), see Figure 1T), located at the confluence of Brays Bayou with the HSC, showed the highest concentrations of arsenic, cadmium, copper, and lead; such relatively high concentrations, however, might be attributed to construction activities that were observed during sampling. It is noteworthy that the most downstream sampled site (Buffalo Bayou at Sabine Rd (BB5), see Figure 1T), showed a significantly lower concentration of metals compared to upstream station BB4. This may be attributed to the unusally high and unexpected deposited amounts of sediment at BB5 (Figure S6 in the SI).
Trace Metals in Bed Sediment Samples. Figures 7E and 7F show trace metal concentrations for the bed sediment samples; all were below PCLs. By comparing values shown in Figure 7 with historical data (Table S4 in the SI), it was found that cadmium levels in sediment measured after Hurricane Harvey were higher than historical levels in all stations although they were still below TCEQ’s PCL. The highest levels of metals were observed in samples collected from the most western part of the HSC (station S1, Figure 1B). The observed values of cadmium, chromium, and nickel in station S1 were significantly higher than the values observed in deposited soils. This difference was attributed to the presence of Superfund sites and industrial activities near the western part of the channel. As shown in Figure S7 in the SI, between stations S1 and BB5, there are seven Superfund sites, five of which are still active (TCEQ, 2018); historically, metals contamination has been reported for these sites (U.S. EPA, 2018). During Harvey, leaks were reported from the U.S. Oil Recovery Superfund site located very close to the sampled station (NRC, 2017). Thus, the U.S. Oil Recovery Superfund site and local runoff from industrial areas surrounding the HSC might potentially be the source of high trace metals found in sediments collected from station S1.
3.4. Microbial Communities in Bed Sediment Samples
Alpha Diversity of Bacterial Communities. Alpha diversities defined by Shannon-Wiener and Simpson indices are shown in Figure 8. In general, an increasing trend in alpha diversity from station S1 to S4, all located in the western part of the HSC was observed followed by a decreasing trend from S5 to S10. Station S1 showed the lowest alpha diversity among all samples followed by stations S10, S9, and S6. As shown in Figure 1B, there are three active Superfund sites near station S9, three close to S10, one close to S1 (U.S. Oil Recovery), and one close to S6 (San Jacinto Waste Pits). The low alpha diversity of microbial communities observed at sampled sites close to the discharge points of Superfund sites might be attributed to the potential leaks and spills from the Superfund sites over the years, especially during severe events. Contrary to what might be expected, Station S5 that is very close to the San Jacinto Waste Pit Superfund site that exhibited failure in its cap during Hurricane Harvey (United States Coast Guard, 2020), showed higher diversity and number of OTUs compared to S6. This could be attributed to the specifics of the hydrodynamic regime at S5 that is located in the deeper part of the channel; the relatively high stream flow of the SJR (up to 10,200 m3/s) during Harvey most likely washed out much of the sediment at this location.
Beta Diversity of Bacterial Communities and Clustering. The unweighted UniFrac measure, coupled with PCoA displayed a pattern in the samples that seems to reflect the specificity of the sample physical location (Figure 9T). Axis 1 from the PCoA plot, accounted for 28.5% of the variation, grouped the S1/S2, S5/S6, S3/S4, S7/S8, and S9/S10/S11 stations together. As noted before, unweighted UniFrac mostly emphasizes the rare taxa, which could be affected by anthropogenic constituents found at these sites due to industrial and agricultural runoff or the presence of Superfund sites that could favor the proliferation of specific bacterial species and inhibit others. Stations S3, S4, S7, and S8 that are more distant from the Superfund site did not group with the ones that are near Superfund sites.
Stations S1 and S2, based on the diversity data, are influenced by the Superfund site and industrial activities adjacent to S1. Thermotogae, an anaerobic hyperthermophilic phylum commonly found in hydrocarbon-impacted sites (Nesbø et al. , 2010; Gupta and Bhandari, 2011) , such as U.S. Oil Recovery Superfund site, was only found in station S1 (5.13%), Latescibacteria, a saccharolytic and proteolytic phylum (Farag et al. , 2017), showed the highest richness in S1 (7.99%), and Atribacteria, an anaerobic phylum usually found in marine sediments and petroleum reactors and reservoirs (Nobuet al. , 2016), was the most abundant in S1 (9.37%) and S2 (0.36%). At the genus level, Nitrosomonas were most abundant in the western part of the HSC system (1.27% in S1) with high level of anthropogenic activities compared to Nitrospira that were most abundant in stations with the minimum level of anthropogenic activities (S5=2.55%, S7=1.29%) as observed in other studies (Cao et al. , 2006). Stations S5 and S6 were affected by potential leaks and spills from the San Jacinto Waste Pits, and the presence of grasslands and farms within the drainage area of the SJR that was the source of significant amounts of sediment transported into the HSC-GB system during Hurricane Harvey. Verrucomicrobia, a facultatively or obligately anaerobic phylum commonly found in grasslands (Janssen, 2006; Bergmannet al. , 2011) and in trace metals and nitrate-contaminated sites (Vishnivetskaya et al. , 2011), was most abundant in S6 (3.71%) and S5 (2.17%). In addition, Cyanobacteria, a photosynthetic and toxic phylum (Stanier et al. , 1979; O’Neil et al. , 2012) commonly found in freshwater lakes (Jia et al. , 2017), were found in S5 (11.05%) and S6 (7.72%) at considerably higher levels than other stations. The source of Cyanobacteria could be from the freshwater that was released from Lake Houston, a constructed lake behind the Lake Houston dam, into the SJR at a rate of 10,000 m3/s during Hurricane Harvey. Five Superfund sites, spills and leaks from wastewater treatment plants as well as industrial runoff from various petrochemical industries located on the southwestern part of Galveston Bay potentially influence microbial diversity in S9 through S11. Ammonia-oxidizing phyla, commonly found in wastewater treatment plans (WWTPs) were the most abundant in stations S9 through S11. The highest abundance of Thaumarchaeota (Hatzenpichler, 2012) was found in S9 (1.94%) and S10 (1.10%) as well as the highest level of Phycisphaerae (Fuerst, 2017), which was found in abundance in S10 (2.85%) and S9 (1.39%). In addition, Chloroflexi (Mohamed et al. , 2010), found often in WWTPs (Kindaichi et al. , 2012), showed the highest sequences in S10 (5.67%), S9 (5.26%), and S11 (5.20%).
The PCoA plot of the weighted Unifrac shows a different diversity pattern compared to the unweighted UniFrac (Figure 9A and B). Axis 1 from weighted Unifrac PCoA accounted for the highest variation of 65.6%. As shown in Figure 9B, weighted Unifrac grouped samples into three groups. Group 1 consisted of sample S1, group 2 comprised S11 and S10, and group 3 included the rest of the samples. Proteobacteria followed by Bacteroidetes were the most abundant phyla of bacteria in all samples and together accounted for more than 70% of OTUs in all samples except for station S1 (63.45%). The observed range of abundance for Proteobacteria in bed sediment samples (35.10%-67.34%) was within the reported range in the literature (22.95%-64%) (Spain et al. , 2009; Kormas et al. , 2010; Vishnivetskaya et al. , 2011; Bowen et al. , 2012; Sun et al. , 2013; Zhu et al. , 2013; Hieke et al. , 2016). However, Bacteroidetes showed higher abundance in bed sediment samples (11.21%-32.70%) compared to the reported values in the literature (1.9%-14.42%) (Spain et al. , 2009; Vishnivetskaya et al. , 2011; Bowen et al. , 2012; Sun et al. , 2013; Zhu et al. , 2013; Hieke et al. , 2016). Higher levels of Bacteroidetes, up to 50%, were only observed in anaerobic digesters (Chen et al. , 2016) and the high measured range of this phylum in almost all bed sediment samples may be indicative of the effect of spills and leaks from wastewater treatment plants during Hurricane Harvey. This is an important finding since Bacteroidetes have been linked to human metabolic diseases (Johnsonet al. , 2017) that could potentially increase the risk of exposure to both water and sediment.
The results of the clustering analysis for all classical methods and K-mean clustering test (K=3) showed similar outputs. The output clusters were 1) stations S1, S5, and S6, all located close to Superfund sites with reported leaks during Hurricane Harvey, 2) S2, S3, S4, and S8, all located in the deeper part of the HSC, and 3) all other stations located in the open and shallow bays (S7, S9, S10, and S11). Here again, the effects of geometry (the specificity of the sample physical location) and the presence of hazardous sites were confirmed. PERMANOVA results using the three clusters showed P-values smaller than 0.01 that indicates the rejection of null hypothesis; the phylum distribution and abundances among the clusters are not equal.