Biosensor Applications in Galveston Bay: Implications for Disaster Research

Graphical Abstract

1. Introduction

Given the recent record-breaking frequency and magnitude of natural disasters within North America (National Centers for Environmental Information, 2021), the ability to conduct research efficiently following disasters continues to grow in importance. Within the state of Texas, an area of interest in natural disaster research has been the intersection between public health and the impacts of post-disaster, especially for vulnerable human and ecological populations (Aly et al., 2020; Bera et al., 2019; Horney et al., 2019; Karaye et al., 2019; Knap & Rusyn, 2016; Sansom et al., 2018). Galveston Bay and the Houston Ship Channel (GB/HSC) serve as a unique economic and industrial marine navigation channel for the city of Houston, TX, which is the world’s second-largest petrochemical complex and home to over 2 million people (Dellapenna et al., 2020; Houston, 2021; US Army Corps of Engineers, 2017). The region is an urban estuary, which serves as both a diverse ecological resource and a natural filter for nearby industrial and urban outputs (e.g. atmospheric deposition, agricultural runoff, roadway runoff, wastewater spills, oil spills, etc.) (Al Mukaimi et al., 2018b; Dellapenna et al., 2020; HARC & Galveston Bay Foundation, 2020; Oziolor et al., 2014, 2018; Park et al., 2001a, 2001b).

Hurricane Harvey made landfall along the Texas coast in August 2017, bringing between 26–47 inches of rainfall to the Houston area within 4 days (Harris County Flood Control District, 2018). Due to this extreme rainfall event, an estimated 1.31 × 10⁸ metric tons of sediment were deposited into Galveston Bay, with an average of 14 cm of flood layer sediment deposited (Dellapenna et al. in review, Du et al., 2019b, 2019a). Additionally, the slow movement of Hurricane Harvey contributed to severe flooding that inundated both the City of Houston as well as its nearby bayous and waterways (Dellapenna et al., 2020; Harris County Flood Control District, 2018; Kiaghadi & Rifai, 2019). Some of the damages related to this flooding event included, but were not limited to, overflow from wastewater treatment plants and spills from local industrial facilities (Kiaghadi & Rifai, 2019), which coupled with the sediment flood layer (Dellapenna et al., 2020; Du et al., 2019b, 2019a) implicated the feasibility of chemical, biological and physiological contaminant redistribution. An additional factor that is influential in contaminant redistribution is the local subsidence rates (Al Mukaimi et al. 2018a, 2018b) since deeper sediments with historical anthropogenic inputs could be uncovered and thereby become available for redispersion under flood conditions.

Given GB/HSC is a highly industrial and urban estuary, chemical contaminants such as polychlorinated biphenyls (PCBs), polycyclic aromatic hydrocarbons (PAHs), organochlorine pesticides, heavy metals, and dioxin/furans have been historically documented within the region (Al Mukaimi et al., 2018b; Bera et al., 2019; Camargo et al., 2020; HARC & Galveston Bay Foundation, 2020; Hieke et al., 2016; Howell et al., 2011; Lakshmanan et al., 2010; Louchouarn et al., 2018; Qian et al., 2001; Santschi et al., 2001; Yeager et al., 2007). Of these contaminants, PAHs are of interest due to their ubiquitous prevalence within environmental matrices (e.g., sediments, water), their occurrence as complex mixtures and the risk that exposure may pose to both ecological species and public health in urban environments (Agency for Toxic Substances and Disease Registry, 1995; Brown et al., 2017; Hussar et al., 2012; Hwang & Foster, 2006; Kim et al., 2019; Oziolor et al., 2014; Vane et al., 2014).

Due to the severe flooding event resulting from Hurricane Harvey, PAH redistribution was a concern due to particulate matter availability in the floodwaters to which PAHs can sorb to as well as ample PAH inputs from the urban environment (e.g. crude oil, roofing tar, asphalt, creosote, industrial combustion, roadway dust, atmospheric deposition, etc.). Since PAHs occur as complex mixtures, a subset of sixteen PAH compounds has been established as the United States Environmental Protection Agency’s (USEPA) Priority 16 PAHs due to their suspected carcinogenicity and toxicity as well as prevalence at National Priority List (NPL) sites (Agency for Toxic Substances and Disease Registry, 1995; Appendix A to 40 CFR, Part 423–126 Priority Pollutants, 2014). Traditional and highly sensitive detection methods for PAHs include high-performance liquid chromatography (HPLC) with either ultraviolet (UV) absorption or UV fluorescent detection; gas chromatography (GC) with either mass spectrometry (MS) or flame ionization detection (FID) (Li et al., 2016). However, these methods are time and resource-intensive when used to detect PAHs.

To supplement and fast-track these traditional PAH detection methods, several immunoassays paired with biosensing platforms have been developed with each ranging in their PAH detection sensitivity (Behera et al., 2018; United States Environmental Protection Agency, 1996a, 1996b). One biosensor technology, The KinExA Inline Biosensor (Sapidyne Instruments), employs a mouse-derived anti-pyrene-butyric acid monoclonal antibody (mAb), 2G8, capable of detecting all 3–5 ring PAHs (Li et al., 2016; Spier et al., 2011). With this level of specificity, the KinExA Inline Biosensor (biosensor) employs the 2G8 mAb to obtain real-time quantification of PAHs in environmental samples; in particular sediment porewater (Hartzell et al., 2017; Hartzell, et al., 2018). This particular biosensor also offers a cost-effective and rapid analysis of environmental samples with results per sample obtained within 10 minutes (Li et al., 2016). Consequently, this tool offers both expanded PAH coverage for monitoring programs and a method capable of providing results the same day samples are collected in the field. This would allow for additional sampling the next day around areas with high concentrations to determine the extend of the contamination.

For sediment and risk management purposes, understanding both bulk-sediment PAH concentrations and freely available PAH (C_free) concentrations in sediment porewater aid in the estimation of PAH bioavailability and bioaccessibility (Ghosh et al., 2011; Hartzell et al., 2018; McGrath et al., 2019; Muz et al., 2020). As a result, passive sampling methods have emerged as common practice to rapidly and accurately predict and measure the bioavailable and bioaccessible chemical fractions within soils and sediments (Cui et al., 2013; Riding et al., 2013). When the biosensor was tested against a passive sampler, the C_free values measured were agreeable despite differences between each approach methodology for C_free (Conder et al., 2021). The measurement of sediment porewater PAH concentrations when compared to surface water PAH concentrations provides relevant PAH exposure estimations for ecotoxicology and disaster research response (DR2).

The goal of this study was to determine the field application of the biosensor within GB/HSC as well as to characterize the PAH profile in soil and sediment porewater within this region. This project is also the first, to our knowledge, to apply the biosensor technology to soils obtained within a neighborhood setting. Previous field-experiments utilizing the biosensor documented its utility in monitoring and detecting the toxic fraction of PAHs in field collected sediments (Hartzell et al., 2017, 2018). Our study is novel as it provides unique insights for the biosensor’s near real-time application for DR2 and its viability in diverse environments. Several Elizabeth River sediments served our known PAH contaminated samples, because our GB/HSC sediments and soils contained unknown PAH contamination levels. This project’s applied-research approach supports the utility of a flexible and cost-effective technology capable of supporting both DR2 and longer-term sediment management of PAHs.

2. Materials & Methods

2.1. Sediment Collection and Study Sites

2.1.1. Galveston Bay/Houston Ship Channel/Clear Lake, TX

Two sediment cores (SB1 and SB2) were collected in 2016, where each core was sectioned at 5cm intervals for the first 45cm (Figure 1a), while surface sediments were collected during May 2019 (n=13) and December 2019 (n=30) (Figure 1c). Both surface sediment and sediment core collection methodologies are previously described (Camargo et al., 2020). The GB/HSC transect was selected to deploy and employ the biosensor for rapid characterization of the area with Clear Lake, Texas serving as a comparator site. Clear Lake (CL) is a recreational and sheltered system compared to the commercial shipping lane the GB/HSC transect serves. The geocoordinates for the sediment samples are reported in Supplementary Table 3.

Figure 1:

Study site base maps for all sediment and soil data collected in 2016 (pre-Harvey sediment cores), 2017 (post-Harvey soils), 2019 (post-Harvey surface sediments), and 2018 (Elizabeth River surface sediments). The 2016 data comprised of 2 sediment cores (a), while the 2017 data comprised of 44 soil samples taken within the Manchester Neighborhood (b). In 2019 43 surface sediment samples were collected from both Galveston Bay/Houston Ship Channel and Clear Lake (c), while 43 surface sediments were collected from the Elizabeth River (d). Each map demonstrates the four unique sites assessed in this project.

2.1.2. Elizabeth River, VA

Surface sediments from the Elizabeth River (ER) were collected in 2018 as a part of a long-term killifish study seeking to understand tumor prevalence in local fish populations (Elizabeth River Project & Virginia Institute of Marine Science, 2020; Vogelbein & Unger, 2006). Sediments from all Elizabeth River locations were collected by a boat-mounted Ponar 1 grab sampler, according to standard protocols (ASTM International 2014; USEPA 1995). In brief, the top 2cm of approximately 3–6 grabs at each location were composited in a stainless steel bucket and stirred until homogeneous in texture and color and then place in pre-cleaned 1L glass containers that were capped with Teflon® lined lids. All samples were held on ice during collection and transport; they were then stored frozen until further processing (n=30) (Figure 1d). All sample geocoordinates are reported in Supplementary Table 3.

2.2. Soil Collection and Study Sites

The soil sample collection methodology has been previously described (Sansom et al., 2020 In Press) and their geocoordinates were also recorded (n=42) (Figure 1b).

2.3. Sediment Chemical Analysis for PAHs

All Galveston Bay/Houston Ship Channel sediment and Manchester soil PAH extraction and analysis methods have been previously described (Camargo et al., 2020; Sansom et al., 2020 In Press). Elizabeth River sediment samples were analyzed for PAH concentrations with gas chromatography-mass spectrometry-selective ion monitoring (GC-MS-SIM) methods used previously (Spier et al., 2011; Unger et al., 2008). Briefly, the samples were lyophilized, spiked with deuterated surrogate standards, and extracted with dichloromethane in a Dionex accelerated solvent extractor. The extracts were reduced under dry nitrogen and separated by size exclusion chromatography and open column chromatography to isolate the compounds of interest. The internal standard p-terphenyl was added before analysis on a Varian Saturn GC/MS/MS ion trap mass spectrometer operated in electron ionization mode (EI). Six-point calibration curves were generated for PAH analytes and identifications were based on retention time and matches to library spectra.

2.4. Total Organic Carbon (TOC) Analyses

The soil and sediment TOC analysis for the GB/HSC sediments and Manchester soils were analyzed by TDI-Brooks International, Inc. using the Lloyd Kahn procedure and the LECO corporation model 632 carbon analyzer with direct combustion/infrared detection (TDI-Brooks International Inc., 2019). The Elizabeth River sediment TOC analysis was completed using the Exeter CHN Nodel 440 CE analyzer. The samples (between 8–25 mg) were packed in a silver cup or nickel sleeve and were dropped into the combustion chamber (at 975 °C), which was purged with helium to removed atmospheric nitrogen. In the mixing volume, the sample gas was thoroughly homogenized at a precise volume, pressure, and temperature. The sample gas was passed between three thermal conductivity cells measuring, first the differential between the gas before and after the first trap measures (H), then the second trap removing CO₂ measures (C), and the third trap, which removed helium measures (N). The role of TOC data in this project is to quantify the presence of organic content per sampling site. As each site will be unique, the TOC data also serve as a marker for potential material PAHs can sorb to that can then be detected by GC/MS in the whole sediment/soil analyses.

2.5. Porewater PAH Analysis

Porewater was extracted through centrifugation (3500g × 15 min), where the porewater extracts were then filtered using 0.45 μm Teflon Millipore filters to exclude fine particulate matter in the extracts. However, some PAHs bound to finer particulates (e.g. colloids) may have not been sufficiently removed (Conder et al., 2021). A 2G8 monoclonal antibody (mAb) with a fluorescent tag, AlexaFluor 647 (Invitrogen), was used to analyze each sample (Li et al., 2016). This mAb has highest affinity for 3–5 ring PAHs, but there is some limited binding to 2-ring PAHs especially those that are alkyl substituted (Conder et al., 2021; Li et al., 2016; Spier et al., 2011). As methylated PAHs will also be detected, the total PAH value quantified by the biosensor encompasses both substituted and unsubstituted PAHs; both of which can then be further identified by GC-MS.

The KinExA Inline instrument detected the excitation level of the fluorescent tag’s signal, where the final porewater concentrations were quantified based on the detector response to phenanthrene standards from 0.5 to 2.5 μg/L (Hartzell et al., 2018; Spier et al., 2011). Each phenanthrene standard was made daily before sample analysis through a serial dilution of a stock solution. If a sample exceeded the calibration curve, it was diluted with deionized water to make the detection response within the linearly range of the calibrated concentration curve. The minimum detection limit (MDL) for the biosensor method is 0.5 μg/L. For all porewater samples that were less than the MDL, these values were standardized to half the MDL (e.g., 0.25 μg/L).

The detected signal is a function of competition between antibody bound to PAH in the environmental samples versus antibody bound to a competing antigen stationary in the detector so an inverse relationship is observed for the fluorescence, where high fluorescence response corresponds to lower PAH concentration and low fluorescence response corresponds to higher PAH concentration. The biosensor’s automated sample handling procedure first allows mixing of the fluorescent tagged antibody with each sample as previously described (Spier et al., 2011). The automated sampling also incorporates instrument line rinsing between each sample analyzed to avoid cross-contamination (Spier et al., 2011).

2.6. Regional Screening Level (RSL) Calculations

To aid in human health risk characterization, the United States Environmental Protection Agency (USEPA) Regional Screening Level (RSL) Calculator was used to estimate potential oral, dermal, and inhalation hazard quotients (HQ) and sub-chronic risks of these three exposure routes. Each parent PAH concentration as detected by GC-MS was inputted to the RSL Calculator using the following settings: Screening Level Type: Regional Screening Levels (RSLs), Hazard Quotient: 0.1; Target Risk: 10⁻⁶, Scenario: Recreator (for soils and sediments), Media selected: “Soil” or “Soil/Sediment”, Screening Level Choice: Site-Specific/User-Provided, Select Chemical Info Type: “Database hierarchy defaults,” Risk Output: Yes; RfD/RfC Choice: Sub-chronic, and all of the USEPA 16 PAH (parent compounds) were selected and retrieved. The exposure values were downloaded as Excel files that were then compiled into a single excel sheet for further data analysis.

2.7. Region 4 Ecological Screening Values

To compare detected PAH levels with relevant environmental screening values, the USEPA’s Region 4 Ecological Risk Assessment Supplemental Guidance was referenced (United States Environmental Protection Agency, 2018). The Refinement Screening Values (RSVs) for freshwater sediments were reported in this guidance and were juxtaposed with our results in the Supplementary Tables 3–5. The RSVs are reported in μg/kg 1% OC and to have comparable values each sample was normalized for organic carbon (OC) by dividing by that sample’s fraction of organic carbon (f_oc) value.

2.8. Prediction Model

We aimed to assess the predictivity of Biosensor-derived C_free with respect to PAH concentrations measured by GC-MS and corresponding risks. Due to the expected association between PAHs and both organic carbon and pore water, in addition to expected equilibrium partitioning (Burkhard et al., 2017), combining TOC and pore water measurements may provide a better prediction for overall PAH concentrations as compared to either measurement alone. Specifically, we hypothesized that the product of the pore water C_free and the fraction organic carbon (C_free*f_oc) would correlate with GS-MS measured PAH concentrations as well as PAH-associated risks.

2.9. Statistical Analyses

All statistical analyses were carried out in GraphPad Prism 9.0.0. and the raw data were processed using Microsoft 365 Excel. Tableau 2020.2.9 was used to construct the sampling maps. Original GC-MS data are reported in μg/kg Table 1, while these concentrations are log-transformed comparative analysis in Figure 2 and Figure S2. Double-plot ratios for BaA/(BaA+CHR) vs Fl/(Fl+PY) (Figure 4) and An/(An+PHE) vs. Fl/(Fl+PY) (Figure S2) were used to determine PAH sourcing. To assess for additional trends, the RSL Calculator hazard quotient (HQ) results were log-transformed. The predictivity analyses for the biosensor for PAH concentrations and risk utilized GraphPad Prism to calculate a straight-line fit with location-specific intercepts (due to differences in partitioning expected between Elizabeth River and GB/HSC) but shared slopes. Additionally, correlation coefficients (Spearman r and Pearson r) were calculated along with p-values.

Table 1:

Summarized are detected soil and sediment Total Organic Carbon (TOC) measurements, total PAH concentrations in porewater (C_free), PAH totals (Total EPA 16; Totals w/alkylated), a pyrogenic index, and a perylene index. TOC measurements are reported in milligrams per gram Carbon (mg C), percentage TOC (%TOC), or as the fractional organic carbon (f_OC). The C_free values are reported in μg/L, while both the Total Priority 16 EPA PAHs (Total EPA 16) and Total PAHs with alkylated PAHs (Totals w/alkylated) sums are reported in μg/kg. A detailed list of the PAHs included for the two PAH totals listed is in Supplementary Table 1. The pyrogenic index is unitless while the perylene index is reported as a percentage.

	SAMP	mg C	% TOC	foc	Cfree	Total EPA 16	Totals w/alkylated	Pyrogenic Index	Perylene Index (%)
Min	2016 Cores	0.96	0.38	0.00	0.00	115	176	0.21	15.80
	2017 Soils	0.54	0.21	0.00	0.25	87	121	0.06	4.89
	2019 GB/HSC Sed	0.28	0.11	0.00	0.00	41	51	0.15	5.19
	2018 ER Sediments	0.00	0.19	0.00	0.25	714	1029	0.00	4.68
Median	2016 Cores	2.33	0.93	0.01	0.42	414	607	0.56	63.75
	2017 Soils	4.86	1.94	0.02	0.60	1481	1883	2.28	8.28
	2019 GB/HSC Sed	2.40	0.97	0.01	0.87	763	1019	1.48	26.98
	2018 ER Sediments	-	0.97	0.01	0.69	11498	15636	-	8.22
Max	2016 Cores	3.56	1.41	0.01	1.94	794	1270	1.29	92.02
	2017 Soils	64.14	25.82	0.26	27.37	14953	17809	3.72	25.89
	2019 GB/HSC Sed	5.81	2.37	0.02	5.65	18086	21320	5.38	98.34
	2018 ER Sediments	-	11.38	0.11	129.06	6792632	7019023	-	19.19
IQR	2016 Cores	1.58, 2.69	0.63, 1.06	0.01, 0.01	0.00, 0.69	297, 499	386, 845	0.38, 0.68	34.93, 78.12
	2017 Soils	3.59, 7.00	1.49, 2.77	0.01, 0.03	0.25, 1.25	759, 2684	1013, 3565	1.69, 2.68	7.25, 12.11
	2019 GB/HSC Sed	1.14, 3.44	0.44, 1.36	0.00, 0.01	0.50, 1.46	307, 1709	475, 2053	1.13, 2.00	19.80, 38.36
	2018 ER Sediments	-	0.43, 2.03	0.00, 0.03	0.25, 1.31	4048, 38583	9292, 46628	-, -	6.92, 9.18

Figure 2:

Comparative PAH distributions between the EPA 16 PAHs (a) and Total PAHs with alkylated PAHs (b) in log(μg/kg). Both (a) and (b) boxplots illustrate sample type differences: the 2016 sediment cores (black), GB/HSC surface sediments (pink), Manchester soils (green), and Elizabeth River surface sediments (purple) with a one-way ANOVA confirming each sample type is unique and different from the others (EPA 16 PAHs: p<0.0001; R²=0.4543 and Totals with alkyl PAHs: p<0.0001; 0.4861). Panel (c) demonstrates the ranges of each sample type porewater values in log-normalized μg/L, while panel (d) compares the percent perylene (% PER) ranges between sample types.

Figure 4:

The double-ratio plot of BaA/(BaA+CHR) vs Fl/(Fl+PY) are illustrated for the 2016 sediment cores (a), the 2017 soils (b - pink), the 2019 sediments from GB/HSC (c – green), and the 2020 sediments from the Elizabeth River (d - purple). Both cores indicate petrogenic/pyrogenic sourcing (a) while both the 2017 soils (b) and 2019 GB/HSC sediments have mixed sourcing.

3. Results

3.1. Soil and Sediment Chemistry

3.1.1. Total Organic Carbon (TOC)

In the 2016 sediment cores, the %TOC ranged from 0.38–1.41%, with differing depth profiles for SB1 compared to SB2. For instance, SB1 has a similar %TOC at the surface as at 40–45cm depth, while the %TOC was higher at depths of 25–30cm and 30–35cm (Figure S1). The 2019 surface sediments TOC range was 0.11–2.37%, with the HSC having higher %TOC values compared to the Clear Lake samples. In contrast to the GB/HSC/CL sediments, the soils contain the highest %TOC, with the range of 0.21–25.82%. The highest %TOC within the soil samples was at site 98 (25.82%), which is in a recreational park with several other sites in this area also having elevated %TOC (sites: 85, 97, 99) that may be attributed to leafy substances and/or added peat moss. Compared to the GB/HSC sediments, the soils in most sites exceed both the highest 2016 and 2019 %TOC values while the ER sites are comparable with the GB/HSC sites (%TOC range: 0.19–11.38%). The highest Elizabeth River %TOC was at site CS-A, which is an industrialized site adjacent to an active shipyard and is the location of a historical wood treatment facility.

3.1.2. PAH Distributions

Four distinct sample groups were analyzed by GC-MS in this study: sediment cores (2016 Cores), soils (2017 soils), GB/HSC surface sediments (2019 GB/HSC Sed), and ER surface sediments (2018 ER Sediments). To verify the variability between each group’s PAH distribution, a one-way ANOVA analysis was run between all four sample groups’ EPA Priority PAHs (Figure 2a) and Total PAHs with alkylated PAHs (Figure 2b). The Total PAHs with alkylated PAHs included a short list of alkylated PAHs, which are summarized in Supplementary Table 1. Both the EPA Priority PAHs and the Total PAHs with alkylated PAHs groups were found to be statistically distinct (EPA PAHs: p<0.0001; R²= 0.4543 and Total PAHs with alkylated PAHs: (p<0.0001; R²=0.4816).

The Total 16 EPA Priority PAHs (Total EPA 16) for Galveston Bay sediments ranged between 115–794 μg/kg (2016 Cores) and 41–18,086 μg/kg (2019 GB/HSC surface sediments), while the Elizabeth River sediments ranged from of 714–6,792,632 μg/kg. These range differences indicate both regions have unique PAH sediment source loading with distinct PAH concentrations. In contrast to these sediment samples, the Manchester soils have relatively low Total EPA 16 concentrations (range of 87–14,953 μg/kg) compared to the Elizabeth River sediments, but a comparable range to the Galveston Bay sediments. Similar ranges are observed for the Total PAHs with alkylated PAHs (Figure 2b; Table 1) as this sum included several alkylated PAHs also determined in the soils by GC-MS.

Given the biosensor’s sensitivity for 3–5 ring PAHs (Li et al., 2016; Spier et al., 2009), these ring structures along with a portion of 2-ring PAHS were observed in all four sample types (Figure S2). A one-way ANOVA confirmed variability between each ring structure group (Figure S2: (a): p<0.0001, R²=0.8065; (b): p<0.0001, R²=0.5688; (c): p<0.0001, R²=0.4137; (d): p<0.0001, R²=0.1880). In each of the sample types, the median values for 4 and 5-ring PAHs were 210 and 203 μg/kg for the 2016 sediment cores, 836 and 448 μg/kg for the 2017 soils, 365 and 340 μg/kg for the 2019 GB/HSC sediments, and 6,483 and 3230 μg/kg for the Elizabeth River sediments. Since there are detectable 4–5 ring PAHs in whole sediments, these compounds are also likely to be detected in the porewater samples.

Figure 3 illustrates the screening ability of the biosensor for PAHs in comparison to traditional whole soil and whole sediment samples. A comparison between the ER and GB/HSC sediments found several trendlines with good predictivity for PAH concentrations in whole sediment or soil as measured by GC-MS. All C_free values were multiplied by the fraction of organic carbon (f_OC) to adjust for organic matter content in each sample. As described in Methods, this calculation was used as an adjusted calculation of the equilibrium partitioning equation (Burkhard et al., 2017). Between the Total EPA 16 PAH concentrations (Figure 3a, 3b) and the Total PAHs with alkylated PAHs (Figure 3c, 3d) similar trendlines are observed, though with different R² values.

Figure 3:

A correlation analysis between the GC-MS total PAH concentrations (mg/kg) versus porewater Cfree (μg/L) for the predictivity of PAH detection by the KinExA Inline Biosensor where y = 39.04x − 41.6 (Goodness of Fit: R2 = 0.766 and the standard deviation of the residuals (Sy.x) = 384.0).

There were no cases of low porewater measurements and high total PAH measurements by GC-MS. The soil sample at site 61 flagged high on the biosensor with a PAH value of 27.37 μg/L. In contrast, the GC-MS whole soil PAH concentration was 2,953 μg/kg. Upon reviewing this sample’s chromatogram and PAH data, the fingerprint indicates this sample is primarily composed of alkylated PAHs suggesting a potential petrogenic source and the possibility of liquid petroleum present in the sample that may have contributed to the high porewater measurement via dissolution from the liquid phase. Additionally, this sample may not have reached equilibrium to allow for partitioning of the PAHs to remain in both solid and aqueous phases in the soil.

Correlation analysis revealed additional support for the biosensor in combination with TOC data to screen and predict PAH concentrations in whole sediment and soil analyses (Table 2). For example, the ER Total EPA 16 PAH Pearson r was 0.8061, while the ER Total PAHs with alkylated PAHs Pearson r was 0.8117. Both of these r values were statistically significant and support the trendlines observed in Figure 3. In contrast, the HSC Total EPA 16 PAH Pearson r was 0.5808 and the Total PAHs with alkylated PAHs Pearson r was 0.5987. While the HSC r values are lower than the ER samples, this is likely due to the smaller range of concentrations from the HSC, as the residual error estimates were similar. To verify and check the trendline analyses a multiple linear regression (considering Cfree and foc separately) was conducted on these sample sets with similar results observed.

Table 2:

Each category’s correlation results of Spearman r and Pearson r within the Elizabeth River (ER) samples and Houston Ship Channel (HSC) samples. The five categories considered are: Total EPA 16 PAHs, Total PAHs with alkylated PAHs, Carcinogenic Risk, Noncancer Child HI, and Noncancer Adult HI.

	ER	HSC
Total EPA 16 PAHs
Spearman r	0.6859****	0.5346****
Pearson r	0.8061****	0.5808****
Total PAHs with alkylated PAHs
Spearman r	0.6899****	0.5345****
Pearson r	0.8117****	0.5987****
Carcinogenic Risk
Spearman r	0.6303****	0.4999****
Pearson r	0.7333****	0.5217****
Noncancer Child HI
Spearman r	0.6165****	0.5035****
Pearson r	0.7311****	0.5253****
Noncancer Adult HI
Spearman r	0.6423****	0.5124****
Pearson r	0.7416****	0.5355****

^****p < 0.0001.

3.2. PAH Sourcing

To understand potential PAH inputs, double-plot ratios were analyzed for all four sample groups using BaA/(BaA + CHR) vs Fl/(Fl + PY) (Figure 4). Both pyrogenic indices and perylene indices were calculated for the GB/HSC sediments and soils; however, only the perylene index was calculated for the Elizabeth River sediments as the alkylated PAHs detected by VIMS were not the same as GERG (Supplemental Table 1). The pyrogenic index suggests the predominance of petrogenic or pyrogenic origins in a sample where a range of 0–0.05 indicates petroleum sources and an excess of this range indicates pyrogenic sources (Wang et al., 2014). In contrast, the perylene index is the percentage of perylene detected within the total of the parent 5-ring PAH isomers (Wang et al., 2014). This latter index is useful for indicating whether detected PAHs occurred due to anoxic conditions transforming organic matter, which is also known as biogenic or diagenetic origins (Boehm, 2005; Stogiannidis & Laane, 2015; Wang et al., 2014).

However, a unique pyrogenic PAH source is creosote, a distillate of coal tar, both of which are predominated by 5–6 ring PAHs and there is a distinct prevalence of these PAHs parent compounds compared to their alkylated homologs (Boehm, 2005; Merrill & Wade, 1985). Due in part to the historical wood treatment facilities along the Elizabeth River, such as Atlantic Wood, the Elizabeth River sediments are known for predominantly creosote contamination (Brown et al., 2017; Di Giulio & Clark, 2015; Merrill & Wade, 1985). Consequently, the sediments collected within this tidal estuary served as a known creosote contaminated area to compare with the GB/HSC sediments and soils.

In Figure 4a, the two 2016 sediment cores predominantly originate from a mixture of petrogenic (0.15–0.35 Fl/(Fl+PY)) and pyrogenic (0.3–0.6 BaA/(BaA+PY)) sources. The pyrogenic index solidifies the pyrogenic inputs since all values exceed crude oil/heavy oil and fuel pyrogenic index range (Wang et al., 2014). The perylene index, on the other hand, indicates all core samples have a biogenic origin, which is sensible due to the sediments being under anoxic conditions at depth. The soils in comparison, consist of both petroleum and biomass combustion (0.45–0.60 Fl/(Fl+PY)) as well as coal combustion (0.25–0.35 BaA/(BaA+PY)) (Figure 4b). Two unique sites from this general trend are sites 59 and 61 both of which are comprised of more alkylated PAHs than the parent PAHs. However, like the sediment cores, all samples except site 61 have pyrogenic inputs. At site 61, the pyrogenic index of 0.06 indicates this site may have a heavy fuel or oil input. Unlike the sediment cores, the soil perylene index varied with a few sites exceeding 10%, which indicated the biogenic origins may be associated with the organic matter in the floodwaters.

The 2019 GB/HSC sediments follow a similar trend as the soils for relative PAH sources, except for several sites with a petrogenic/pyrogenic source (HSC 7; HSC 13; HSC 15; HSC 17, HSC 18) and one with a petrogenic/coal combustion source (HSC 2) (Figure 4c). The five sites with petrogenic/pyrogenic sourcing are located within the upper HSC and near the more industrialized portions of the channel. Site HSC 2 is located near Morgan’s Point and Atkinson Island; both of which are secondary entry points to the Upper HSC from the Bolivar Roads/Gulf of Mexico HSC entry. Of the sites with petroleum combustion or biomass combustion and pyrogenic origins, there is a near split between these groupings from 0.4–0.46 Fl/(Fl+PY) and 0.49–0.54 (Fl/(Fl+PY) (Figure 4c). The former Fl/(Fl+PY) range could be attributed to roadway dust (Yunker et al., 2002), while the latter is bordering the transition range of petroleum sources to combustion sources (Yunker et al., 2002). The pyrogenic index for all the GB/HSC samples indicated pyrogenic origins, while the perylene index indicated biogenic origins influenced several sites. Compared to the 2016 sediment cores, most of the GB/HSC sediments did not have biogenic inputs as a key PAH source, except for site HSC 12, which is located across from the San Jacinto Park in Buffalo Bayou.

The Elizabeth River sediments are uniquely clustered within the range of 0.51–0.62 Fl/(Fl+PY) with five sites separate from this range (PC-A; SC1-A; SP-A; SP-B; RF-A) (Figure 4d). The clustered sites create a boundary within the range of coal tar’s Fl/(Fl+PY) ratio value (0.58) (Yunker et al., 2002) implicating creosote, a distillate of this PAH. The perylene index for the Elizabeth River is mostly below 10% indicating few sites have biogenic inputs; however, a few sites such as LF-A, MP2-B, SC1-A, and WB-A exceed 10% indicating these sites have some biogenic inputs.

3.3. PAH Risk in Soils and Sediments

The GC-MS PAH results for the soils were inputted into the RSL Calculator using the settings discussed in the methods section. Of the outputs, risks to both human children (Figure 5a and 5d) and adults (Figure 5b and 5e) were considered for three routes of exposure: ingestion, dermal, and inhalation. Both the ingestion and dermal exposure routes are of concern, because of documented sediment deposits being reported near residences after Hurricane Harvey (Karaye et al., 2019). However, each route of exposure for PAHs had a hazard quotient (HQ) value less than 1 (Figure 5). Should these HQs exceed 1, additional soil studies for exposure assessment and soil management would have been required to determine specific site characteristics and potential health impacts from these soils.

Figure 5:

Given the GC/MS whole-soil analysis collected for PAHs, these values were inputted to the USEPA Regional Screening Level (RSL) Calculator (United States Environmental Protection Agency, 2017) to calculate the ingestion risk (black circles), dermal risk (pink squares), and inhalation risk (turquoise triangles) for both children (a) and adults (b). Should values exceed 1.00 (dotted line at y = 0) further investigation is required to determine the extent of the exposure risk. For all the soils analyzed in the Manchester Neighborhood, their risk values all fell well below 1.00 indicating low exposure risk in this particular media.

Figure 5 illustrates the TOC+biosensor porewater predictions for screening level risk estimates of Carcinogenic Risk, Noncarcinogenic Child Hazard Index (HI), and Noncarcinogenic Adult Hazard Index (HI). These results show that the C_free PAH concentration when adjusted with the f_oc can provide prediction of relative risk. The correlation results for all risk results are tabulated in Table 2. The correlation analyses consistently range between 0.61–0.64 for the ER samples, while the HSC sample correlations range between 0.49–0.51. Multiple linear regression analyses run on both the ER and HSC sample sets resulted in similar goodness of fit results as the results presented in Figure 5.

Additional risk to benthic organisms was considered by comparing the USEPA’s Region 4’s Refinement Screening Value (R4- RSV) in Supplemental Table 5. All PAH concentrations were normalized to organic carbon and are reported in μg/kg OC. Of the sediments analyzed, the 2016 sediment cores were below the R4-RSV, thus indicating these sediments were within bounds and did not require additional lines of evidence for ecological risk assessment (Supplemental Table 5). However, of the 2019 GB/HSC sediments and the Elizabeth River sediments, all maximum values of the individual PAHs listed exceeded the R4-RSV values (Supplemental Table 5). These sediment results indicate a need for additional investigations that consider background concentrations, mixture effects, toxicity mechanism, exposure relevance, and detection prevalence among other lines of evidence (United States Environmental Protection Agency, 2018).

4. Discussion

4.1. PAH Predictivity Using KinExA Biosensor

Environmental sampling is often extensive and the use of traditional analytical methods of HPLC or GC to characterize chemical compounds requires both time and money. When a natural disaster occurs, comparisons to pre-disaster baseline are sought to improve understanding of where contaminated environmental media may have traveled and the impact these shifted contaminants may pose for the environment and the public’s health (Bera et al., 2019; Birch & Lee, 2018; Camargo et al., 2020; Dellapenna et al., 2020; J.A. Horney et al., 2018; Kiaghadi & Rifai, 2019).

The biosensor technology applied in this project demonstrates a rapid, flexible, and cost-effective method to assess for PAHs within both soils and sediments; two matrices often implicated with natural disasters. Additionally, our samples show how the aqueous fraction of PAH (C_free), as quantified by the biosensor, is capable of flagging samples for additional analyses, particularly when combined with data on organic carbon. Of course, the prediction is not perfect, and the porewater C_free concentration is also influenced by organic content, sample composition, and partitioning properties (Arp et al., 2009; Burkhard et al., 2017; Ghosh et al., 2001; Ghosh & Hawthorne, 2010). For instance, if a sample contained black carbon or coal material, then there are different PAH desorption rates and consequently variable PAH partitioning between the aqueous and solid phases (Ghosh & Hawthorne, 2010). Given our results, the biosensor is capable of prioritizing samples requiring follow-up sediment or soil kinetic and toxicity studies in terms of both expected PAH concentrations as well as expected risk levels. This prioritization can improve the efficiency of sampling need to perform initial site characterization and exposure assessment.

Since traditional guidance related to sediment toxicity utilizes dry-weight concentrations or mass-based concentrations, a knowledge gap remains between how aqueous PAH concentrations (C_free) impact sediment toxicity and bioavailability (Burkhard et al., 2017; United States Environmental Protection Agency, 2018). Therefore, research continues to uncover the relationship between sediment toxicity and bulk-sediment compared to chemical porewater partitioning. Efforts using passive sampling devices (PSDs) also continue to expand our understanding of bioavailable chemicals (Allan et al., 2011, 2012; Muz et al., 2020). The terms “bioavailable” and “bioaccessible” are also key measures to include for toxicology and exposure assessments (Ruby et al., 2016). Our findings consequently implicate the utility of porewater PAH concentrations for monitoring programs as well as DR2 research. However, there is still a need to investigate single PAH compounds in conjunction with the biosensor (Conder et al., 2021) in addition to expanding the PAH analytical suite to include emerging PAHs (Gao et al., 2019).

When comparing traditional targeted analyses (e.g., GC-MS) with the biosensor, the traditional analyses are expensive, with samples costing hundreds of dollars and taking extensive time to analyze. The biosensor mitigates these costs by reducing the need for GC-MS analysis for each sample collected. Instead, the biosensor provides a screening method to prioritize samples and can do it within hours after collection. If an adjustment for TOC is included, the detected C_free in porewater provides initial predictions for whole-soil/sediment GC-MS analysis. Organic carbon and other organic matter present in sediments and soils often influence PAH partitioning and warrant additional investigation (Arp et al., 2009; Ruby et al., 2016; United States Environmental Protection Agency, 2018; Xia et al., 2016). By using the biosensor to characterize the aqueous phase PAH concentrations and adjusting for f_oc, traditional GC-MS and additional passive sampling can be implemented for follow-up studies. When porewater analyses overpredict GC-MS whole sediment or soil results, additional work may be warranted to better understand the drivers of bioavailability at certain locations.

As analytical equipment continues to improve sensitivity analysis for emerging PAHs, the biosensor offers the opportunity to screen for total PAHs and provides preliminary results to quantify sites that may be of concern. Compared to other surface plasmon resonance (SPR) biosensors, the main limitation of the biosensor is the need for GC-MS to quantify individual PAHs (Behera et al., 2018). However, our results show that our biosensor is capable of screening both soils and sediments as well as the ability to relatively predict PAH concentration in whole sediment/soil using TOC adjusted C_free values.

4.2. Implications for Disaster Research and Exposure Assessment

The value of the biosensor in disaster research and exposure assessment is to serve as an adaptable tool both in the lab and in the field. The resulting PAH concentrations can serve as a guide to prioritizing environmental matrices for environmental health exposures after a disaster. The biosensor also provides preliminary data that can be readily shared in the field for risk communications. Due to the ubiquitous nature of PAHs found throughout the world, and especially within urban and industrialized environments (Bacosa et al., 2020; Kanzari et al., 2014; Kim et al., 2019; Soliman et al., 2019; Vane et al., 2014), the biosensor offers the opportunity to determine follow-up experimental needs within a wide range of environments. The biosensor’s comparable results to passive samplers also support its implementation for future fieldwork (Conder et al., 2021).

While PSDs readily characterize chemical patterns and measure potential bioavailable fractions (Allan et al., 2012; Muz et al., 2020), the biosensor helps streamline this process by rapidly detecting samples of potential concern prior to extensive analyses. As a result, the biosensor serves as a screening tool to lessen the number of analytical steps associated with a given passive sampling field experiment. If used in conjunction with multiple analytical methods (e.g., PSDs, GC-MS, HPLC, other biosensors), the biosensor used in this study can continue to add to our understanding of “bioavailable” and “bioaccessible” factors lending to PAH toxicity and exposure assessments.

4.3. Limitations

This project has several important limitations. The quantities of porewater samples were low, despite centrifugation. This was in part due to there not being enough samples as well as the lack of porewater in the samples. For future experiments involving soil analysis, in particular, we recommend testing a wide range of soil types to understand how the porewater and soil chemistry impact porewater yield as well as the PAH portioning properties. Regarding the sediment analysis, pre-disaster pollution sources are important to document, as many of the GB/HSC sediments were well below any of the Elizabeth River sediment PAH concentrations. Comparisons with known polluted samples can help gauge levels of potential contamination of sites with unknown PAH inputs.

5.0. Conclusions

This project demonstrates the screening ability of the KinExA Inline Biosensor within two tidal estuarine environments and for the first time in soils. The biosensor is an asset for DR2, especially when vulnerable human and ecological populations are located near a potential site of concern. The biosensor is capable of prioritizing GC/MS analysis as well as passive sampler analysis (Conder et al., 2021). Data are provided in real-time and the biosensor is compatible with field-work analysis. Despite a lack of guidelines for freely dissolved PAHs, the biosensor, especially when combined with TOC data, offers insights into which fraction of PAHs are bioavailable or bioaccessible in the freely dissolved phase compared to the bulk sediment on which many sediment quality guidelines are based (McGrath et al., 2019; United States Environmental Protection Agency, 2018). These results offer preliminary PAH exposure risk considerations and ultimately will help streamline additional analytical analyses and lines of evidence required for environmental health.