Potential Hazards of Polycyclic Aromatic Hydrocarbons in Great Lakes Tributaries Using Water Column and Porewater Passive Samplers and Sediment Equilibrium Partitioning

The potential for polycyclic aromatic hydrocarbon (PAH)‐related effects in benthic organisms is commonly estimated from organic carbon‐normalized sediment concentrations based on equilibrium partitioning (EqP). Although this approach is useful for screening purposes, it may overestimate PAH bioavailability by orders of magnitude in some sediments, leading to inflated exposure estimates and potentially unnecessary remediation costs. Recently, passive samplers have been shown to provide an accurate assessment of the freely dissolved concentrations of PAHs, and thus their bioavailability and possible biological effects, in sediment porewater and overlying surface water. We used polyethylene passive sampling devices (PEDs) to measure freely dissolved porewater and water column PAH concentrations at 55 Great Lakes (USA/Canada) tributary locations. The potential for PAH‐related biological effects using PED concentrations were estimated with multiple approaches by applying EqP, water quality guidelines, and pathway‐based biological activity based on in vitro bioassay results from ToxCast. Results based on the PED‐based exposure estimates were compared with EqP‐derived exposure estimates for concurrently collected sediment samples. The results indicate a potential overestimation of bioavailable PAH concentrations by up to 960‐fold using the EqP‐based method compared with measurements using PEDs. Even so, PED‐based exposure estimates indicate a high potential for PAH‐related biological effects at 14 locations. Our findings provide an updated, weight‐of‐evidence–based site prioritization to help guide possible future monitoring and mitigation efforts. Environ Toxicol Chem 2024;43:1509–1523. © 2024 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.


INTRODUCTION
Polycyclic aromatic hydrocarbons (PAHs) are common contaminants in urban and suburban environments, typically occurring as complex mixtures originating from a wide variety of legacy and contemporary anthropogenic and natural sources (Meador, 2008;Neff et al., 2005;Van Metre & Mahler, 2010).Different PAHs have been shown to cause lethal effects in exposed organisms through narcosis, as well as sublethal effects via carcinogenic, mutagenic, teratogenic, and endocrine pathways (Honda & Suzuki, 2020; US Environmental Protection Agency [USEPA], 2003;Zhang et al., 2016).Narcotic toxicity of PAHs to aquatic organisms is regarded as particularly important from a monitoring/regulatory perspective (USEPA, 2003(USEPA, , 2012)).Within specific modes of action, PAH effects can be additive (e.g., USEPA, 2003;Villeneuve et al., 2002), a critical observation from an environmental perspective because exposures almost always involve complex PAH mixtures (see Van Metre & Mahler, 2010).
In aquatic environments, PAHs tend to bind to sediment organic carbon (OC), and have an especially strong affinity for the partially combusted carbon known as black carbon or soot (hereafter collectively referred to as black carbon [BC]; Buckley et al., 2004;Cornelissen, Gustafsson, Bucheli, et al., 2005;Jonker & Koelmans, 2002), as well as coal tar particles (Khalil et al., 2006).However, it is the unbound, freely dissolved concentration (C free ) of PAHs in overlying and/or sediment porewater that primarily drives exposure and toxicity to aquatic organisms (Hawthorne et al., 2007;Mayer et al., 2014).Accurate estimation or measurement of the PAH C free , rather than the bulk sediment PAH concentration or even the whole porewater or water column concentration, is therefore preferable for accurate risk assessment (Hawthorne et al., 2007;McGrath et al., 2019).
Assessments of PAH-related toxicity in contaminated sediments commonly use equilibrium partitioning (EqP) methods to estimate porewater C free using the sediment PAH concentration, the OC content of the sediment, and chemical-specific OC-water partitioning coefficients (K OC ; Di Toro et al., 1991;McGrath et al., 2019).The porewater C free can then be used to estimate exposure to and effects of PAH mixtures using methods such as the summed EqP benchmark toxicity units (ΣESBTU; USEPA, 2003).However, a "one-carbon" EqP-based approach does not account for the fraction of OC present as BC or coal tar in the sediment (Cornelissen & Gustafsson, 2005).Black carbon commonly constitutes 4% to 18% of OC and can exhibit orders of magnitude stronger sorption of PAHs compared with natural OC (Cornelissen et al., 2005;Jonker & Koelmans, 2002).Sorption to coal tar particles, when present, can be even greater (Khalil et al., 2006).As a result, in sediments high in BC or coal tar, EqPbased toxicity assessments such as the ΣESBTU can overestimate C free , and thus PAH bioavailability and biological effects, by orders of magnitude (Arp et al., 2009;Burgess et al., 2021;Endo et al., 2020;McGrath et al., 2019).In recent years, improvements in passive sampling technology have facilitated the direct measurement of porewater PAH C free in sediments, potentially enabling more accurate estimates of biological effects compared with EqP-based methods (Apell & Gschwend, 2014;Burgess et al., 2021;Endo et al., 2020;Fernandez et al., 2009;Mayer et al., 2014).
In the Laurentian Great Lakes (USA and Canada), PAHs have been identified as critical pollutants by the International Joint Commission (Agency for Toxic Substances and Disease Registry, 2008) and have contributed to the listings of 61% of the Great Lakes Areas of Concern (USEPA, 2013).Previous studies have reported exceedances of PAH sediment quality guidelines and the potential for associated biological effects in sediments from some Great Lakes tributaries (Baldwin et al., 2017(Baldwin et al., , 2022;;Hull et al., 2015).A study of 71 locations across 26 Great Lakes watersheds reported potential toxicity at 41% of the locations (Baldwin et al., 2020).These screening-level assessments were based on bulk sediment PAH concentrations (i.e., probable effects concentrations) or EqPbased estimates (ΣESBTU), rather than direct measurement of porewater PAH C free , and therefore they may have overestimated PAH bioavailability and toxicity at some locations.This potential is underscored by the relatively high proportion of OC as BC in some areas of the Great Lakes Basin: >20% in lakebed surface sediments in Lake Erie, and up to approximately 30% to 35% in Lakes Michigan and Huron (Buckley et al., 2004).The proportion of OC as coal tar is not known, but coal tar has been identified as a primary source of PAHs to Great Lakes tributaries (Baldwin et al., 2017(Baldwin et al., , 2020;;Valentyne et al., 2018).
The present study provides a comparison of EqP-based toxicity estimates from Baldwin et al. (2020) with toxicity estimates based on measurement of C free using passive samplers.We used polyethylene passive sampling devices (PEDs) to measure PAH C free in porewater and the water column at 55 Great Lakes tributary locations.The PED-measured porewater PAH C free values were compared with sediment EqP-based estimates of porewater C free , and also used to estimate PAH-related biological effects.Estimates of possible biological effects were made using multiple approaches: 1) toxicity quotients derived from water quality guidelines; 2) interstitial water and water column toxic units; and 3) exposure-activity ratios based on ToxCast highthroughput bioactivity screening values (USEPA, 2022; Figure 1), the latter for relative prioritization purposes only.Biological effect estimates based on PED-based exposure estimates were then compared with those employing EqP-based ΣESBTU values determined from concurrently collected sediment samples.The results highlight differences between bioavailability and biological effects estimates derived from sediment EqP-and PED-based methods, and provide an updated, weight-ofevidence-based site prioritization that may be used to guide future monitoring and mitigation efforts.

Site selection
Passive sampler and sediment samples were collected from locations in 24 Great Lakes tributary watersheds across six states (Minnesota, Wisconsin, Indiana, Michigan, Ohio, and New York, USA; Figure 2 and Table 1).Each of the five Great Lakes was represented by at least one watershed.Within each watershed, samples were collected from up to six subwatershed locations, for a total of 55 sampled locations.Watersheds and subwatersheds were selected to represent a range of drainage areas, human population densities, land uses, and the percentage of the watershed covered by impervious surfaces (percent imperviousness; Supporting Information, Table S1; Baldwin et al., 2020).

Sample collection and analysis
Passive sampler deployment and analysis.The PEDs were prepared by the Battelle Memorial Institute (Norwell, MA, USA) FIGURE 2: Map of the Great Lakes Basin and the sampled watersheds (red) and locations (black dots).Watershed identifiers (A, B, C,…) correspond with map identifiers in Table 1.Modified from Baldwin et al. (2022).
following internal standard operating procedure (SOP) 5-366, Preparation of Polyethylene Devices for Deployment and Extraction.The PEDs consisted of 25-µm-thick low-density polyethylene (LDPE) sheets that were cut to approximately 40-× 15-cm size, cleaned using a mixture of organic solvents, and then spiked with performance reference compounds (PRCs; anthracene-d 10 , fluoranthene-d 10 , and pyrene-d 10 ; consistent with Burgess et al., 2021) using an 80% methanol to 20% water solution to determine equilibrium status and facilitate quantification of PAH concentrations (Ghosh et al., 2014).After 7 days of agitation in the spiking solution, the solution was drained and the PEDs were rinsed and soaked overnight in Milli-Q water to remove any remaining spiking solution.The PE sheets were then dried, wrapped in clean aluminum foil, sealed in plastic bags, and stored frozen.In the field, the PED samplers were assembled by mounting a PE sheet in a stainless-steel frame constructed from piano hinges (Supporting Information, Figure S1).The PED was driven into the streambed sediment so that the lower half of the PE sheet was below the sediment surface to measure porewater concentrations, and the upper half remained in the water column (Figure 1).The PED deployments occurred during the summers of 2016 and 2017, with an average deployment duration of 32 days (range 27-45 days).In 2016, PEDs were deployed in triplicate at seven locations as a pilot study to examine heterogeneity and define variability at each location, and concentrations were averaged for each location.In 2017, two PEDs were deployed at each location.At three locations, PEDs were deployed in both 2016 and 2017, and the results were averaged.After retrieval, the PE sheet from each PED was cut to separate the porewater and water column portions (discarding the portions of the sheets covered by the frame), each of which were then placed into 125-mL glass amber jars and stored on ice until shipment to the laboratory.Porewater PEDs were not analyzed at 7 of the 55 locations because after retrieval they were no longer buried in the sediment or were damaged.One water column PED was not analyzed because the water level dropped and the sampler was dry on retrieval.Chemical residues were extracted from the PEDs using a twostep dialysis with hexane (Alvarez et al., 2008).Extracts were concentrated and underwent copper treatment to remove residual sulfur, and then passed through gravity-flow columns  , 2003, 2008, 2012); µg/g OC = micrograms/gram organic carbon.
containing layers of acidic, basic, and neutral silica gel to remove potential interferences prior to analysis (Alvarez et al., 2008).Chemical analysis of PAHs (listed in Table 2) in the PEDs was performed using a Thermo Fisher Trace Ultra gas chromatograph coupled to an ISQ mass spectrometer (Alvarez et al., 2021) running in selected ion mode.Quantitation was performed using internal standard calibration and an 8-point calibration curve ranging from 1 to 4000 ng/mL of each PAH.The separations were conducted using an Agilent HP-5MS capillary column (30 m, 0.25-mm inner diameter, 0.25-µm film thickness).

Calculation of porewater passive sampler concentrations
The C free values in the porewater PEDs were calculated using the following relationship (Equation 1) based on EqP models where C free is the estimated water concentration in units of ng/L, C LDPE is the PAH concentration measured in the PED in units of µg/kg, f eq is the fractional equilibria (expressed as a decimal) used for chemicals that have not reached equilibrium, and K LDPE is the LDPE-water partition coefficient.
Under nonequilibrium conditions, the f eq for the amount of each PRC lost from the PED was calculated by Equation ( 2): where f eq-PRC is the fractional equilibria for the PRC; C PRCi is the concentration of PRC in the PED at construction, and C PRCf is the concentration of PRC remaining in the PED after deployment (see the calculated f eq-PRC values in the Supporting Information, Tables S2 and S3).Our study was consistent with published guidance and other studies by using three representative PRCs covering a range of fugacities (Burgess et al., 2021;Choi et al., 2013;Fernandez et al., 2009;Ghosh et al., 2014;Reitsma et al., 2013;USEPA, 2017).Fernandez et al. (2009) showed that the accuracy of using a few PRCs for assessing multiple target compounds was in good agreement with C free estimates when one was using target analyte-matched PRCs, thereby eliminating the need to have a full one-to-one PRC to target analyte match.
To calculate the f eq values for each targeted PAH, a graphical user interface (GUI; PRC_Correction_Sediment_win64) was downloaded from the USEPA website (2022).Calculated f eq values for each porewater sample are given in the Supporting Information, Table S4.There was not an exact match for all the PAHs measured in our study in the GUI, so f eq values for PAHs with similar log octanol/water partition coefficients (K OW s) were used in the subsequent calculation of C free (Supporting Information, Table S5).Additionally, because the K OW range of the PRCs was narrower than that of the target PAHs, the C free data for target PAHs between anthracene and fluoranthene are considered definitive whereas the C free values for target PAHs below anthracene and above fluoranthene are considered provisional.
The resultant C free concentrations for the porewater PEDs are shown in the Supporting Information, Table S5 along with the log K LDPE values used in the calculations.

Calculation of water column passive sampler concentrations
The 2016 water column PEDs exhibited a complete loss of PRCs during the field deployment indicating equilibrium had been achieved for target PAHs with molecular weights up to that of pyrene.Concerns that the water column PEDs may not have been spiked with PRCs are not valid as the lower halves of each PED (buried in the sediment) all had measurable PRCs remaining after deployment.Since the water column PEDs were at equilibrium, the C free was calculated using a rearrangement of the basic EqP model (Equation 3): The 2017 water column PEDs all exhibited nearly complete loss of PRCs during the field deployment.Calculated f eq-PRC values (Equation 2) were outside of the acceptance range of 0.15 to 0.85 used in the GUI Excel macro (PRC_GUI_v2.xlsm,downloaded from the USEPA website (2022) for calculating f eq values for targeted PAHs in each sample.This acceptance range is similar to the suggested range of ±20% due to analytical variability (Ghosh et al., 2014;Huckins et al., 2002).Because the 2017 water column PEDs had f eq-PRC values outside of the acceptable range, C free values were calculated using the basic EqP model (Equation 3).Due to the complete, or nearly complete, loss of PRCs in the water column PEDs, it was assumed that the PEDs were subjected to highly turbulent conditions resulting in the thinning of the aqueous boundary layer at the PED surface, greatly increasing the elimination rate constants for the PRCs.According to predicted equilibrium estimates by Lohmann (2012), under turbulent conditions, the largest PRC used in our study, pyrene (molecular weight 202g/mol), potentially reached equilibrium in as little as 2 days, whereas the targeted PAH with the largest molecular weight, benzo[ghi]perylene (276 g/mol), would not approach equilibrium for over 128 days.At the mean deployment time of 33 days, benzo[ghi]perylene potentially was at 60% equilibrium.It should be noted that using the equilibrium approach will potentially underestimate C free if nonequilibrium conditions actually existed (USEPA, 2017).The resultant C free concentrations for the water column PEDs are shown in the Supporting Information, Table S6 along with the log K LDPE values used in the calculations.

Sediment samples
Sediment sample collection, analysis methods, and concentrations have been described previously (Baldwin et al., 2020).Briefly, sediments were collected to a depth of 15 cm using a push core sampler (WaterMark ® Universal Core Head Sediment Sampler) with polycarbonate tubing, emptied into a stainless steel pan, then transferred into baked amber glass jars, and stored on ice.Depositional areas with fine sediments (silts) were targeted, and a new tube was used at each location.Samples were shipped within 48 h to the Battelle Memorial Institute for analysis of PAHs using gas chromatography mass-spectrometry operated in selected ion monitoring mode and following Battelle procedures 5-157-17 and 5-191-07 (detailed in Baldwin et al., 2020).A sample split was shipped to ALS Environmental (Kelso, WA, USA) for analysis of OC using a modified ASTM International Method D 4129-05 (ASTM International, 2012).

Quality assurance/quality control
A total of 12 PED duplicate samples were collected in 2017.Among the resulting 252 matched pairs (12 duplicate samples x 21 chemicals), 21% of pairs had one value < the reporting limit (RL) and one value > the RL.Among the other 79% of pairs (those in which both samples were <RL or both samples were >RL), median relative percent differences varied by chemical from 0% to 148% (Supporting Information, Table S7 and Figure S2A and B).However, large relative percent differences between matched pairs were often associated with low sample concentrations; absolute differences in concentrations between matched pairs were generally <10 ng/L (Supporting Information, Figure S2C and D).A total of 10 PED field blanks were collected by exposing a PED to the atmosphere during the active deployment and retrieval of a regular sample.The mean concentration of each PAH in the field blanks was subtracted from the regular PED concentrations for background correction.If background correction resulted in a value less than the method detection limit, then the value was reported as <RL.Detections in the field blanks were not infrequent, but concentrations were typically orders of magnitude lower than those in the environmentally deployed samples (Supporting Information, Figure S3).However, for some low-molecularweight PAHs (e.g., naphthalene, 1-and 2-methylnaphthalene, biphenyl), field blank concentrations were similar to or even greater than those in environmental samples, potentially reflecting initial concentrations in the virgin polyethylene that dissipated in environmental samples (Durell et al., 2006).
Sediment sample quality assurance/quality control has been described previously (Baldwin et al., 2020).Briefly, duplicate sediment samples were collected at eight locations.Median relative percent differences varied by chemical, from 10% to 38%.Sediment field blanks were collected at five locations by rinsing the sediment collection equipment with organic-free water (OmniSolv ® ), with 67% of all blank results being <RL.The maximum blank concentration of any chemical was 5.2 ng/L (naphthalene).Field OC blanks (n = 5) were all <RL.

Data analysis
Data analysis focused on the 21 PAHs that were common to the PED and sediment sample analyses (Table 2).Concentrations below the reporting limit were substituted at one-half the reporting limit.Porewater PAH C free was estimated from sediment PAH concentrations and an EqP model for comparison against PED-measured porewater PAH C free .
where C Sed is the dry weight sediment PAH concentration, K OC is the OC-water partitioning coefficient for each PAH from the USEPA's CompTox Chemicals Dashboard (Table 2; USEPA, 2021), and f OC is the sediment OC content expressed as a fraction (%OC/100).CompTox K OC values were averages of predicted values from the USEPA's EPI Suite (modeled using log K OW and the Sabljic molecular connectivity method with improved correction factors; USEPA, 2015) and from OPERA 2.6 quantitative structure-activity/property relationships (Mansouri et al., 2018).
A derivation of Equation ( 4) was used to calculate field-based K OC values from sediment and passive samplers for comparison against literature-based K OC values from CompTox: where C free is the PED-measured freely dissolved concentration in porewater.

Potential for biological effects
Multiple methods were used to estimate potential PAHrelated biological effects based on concentrations in sediment and PED samples, enabling comparison of method results and providing a weight of evidence for site prioritization.Two of the approaches, ΣESBTUs (USEPA, 2003) and toxicity quotients, are based on toxicity benchmarks from evaluation of in vivo toxicity tests with multiple aquatic taxa (e.g., representative fish, invertebrates, plants/algae), focused on apical impacts on survival, growth, or reproduction, and subjected to evaluation of study quality.Such values are generally viewed as high quality for ecological risk assessment applications, but they are available for only a relative handful of contaminants among the hundreds to thousands of xenobiotic contaminants that may be detected in sediment, water, or porewater.Fortunately, PAHs are well represented among contaminants for which National Water Quality Criteria and/or water quality benchmarks have been established by government agencies (Buchman, 2008;Canadian Council of Ministers of the Environment, 2023;USEPA, 2003).
In contrast, exposure activity ratios (EARs; Blackwell et al., 2017) are based on ToxCast (USEPA, 2022) highthroughput bioactivity screening of chemicals, primarily using cell-based or cell-free in vitro assays.Because the assays employed in ToxCast are run in standardized formats, results for different chemicals are highly comparable in terms of relative potencies and efficacies in each assay.However, in vitro bioactivity does not translate directly to in vivo effects or effect thresholds.Even though translational frameworks like adverse outcome pathways (Ankley & Edwards, 2018) can help link bioactivity to biologically plausible apical hazards, direct inference of risk from EARs should still be avoided.Instead, EARs are best employed in the context of relative prioritization of chemicals with regard to potential to produce biological effects (not all of which may be adverse) and/or as a flag that more specific mode(s) of action, other than so-called baseline or narcotic toxicity, readily predicted from log K OW (Escher & Schwarzenbach, 2002), may need to be considered.That said, the EAR approach employed in the present study has some advantage over previous applications of the EAR based on concentrations detected in water or sediment extracts.Most notably, EARs calculated based on C free from passive samplers consider the probable bioavailable fraction.Conceptually, this may lead to stronger inference and more accurate prioritization based on EARs, given that unavailable fractions, such as chemicals bound to dissolved OC, are not considered.An additional advantage of the EAR approach compared with the water quality benchmark or water quality criteria-based approaches is that ToxCast bioactivity screening is available for thousands of contaminants, as opposed to just dozens.

∑ESBTUs
The ∑ESBTU is a common EqP-based screening approach for potential toxicity of PAH mixtures in sediments (Burgess et al., 2021;Endo et al., 2020;Kemble et al., 2013).The ΣESBTU accounts for the OC content of the sediment and its effect on contaminant bioavailability, and is therefore considered to be applicable across a range of sediment types (USEPA, 2003).For sediment PAHs, the ESBTU was calculated as where ESBFCV Sed is the EqP sediment benchmark final chronic value for PAHs in sediment (USEPA, 2003; Table 2).Values of ESBFCV Sed are PAH-specific, derived from National Water Quality Criteria Guidelines (Stephen et al., 1985) and K OC values.The ESBTUs for each of the 21 PAHs were summed to get the ΣESBTU.For site prioritization purposes in the present study, an ΣESBTU of ≥1.0 was considered to indicate high potential for biological effects, and an ΣESBTU < 0.1 was considered to indicate low potential for biological effects.

Toxicity quotients
Toxicity quotients (TQs) were calculated for PED porewater and water column samples as where WQB is the whole organism water quality benchmark associated with acute or chronic aquatic life, compiled from US and Canadian government agencies (Supporting Information, Table S8).Benchmarks that account for phototoxicity were excluded because of the uncertainty of ultraviolet exposure to organisms in the environment (Ankley et al., 2003).Benchmarks were found for 8 of the 21 PAHs in our study.For each sample, the minimum benchmark for each PAH was used to calculate PAH-specific TQs, and then the maximum of the PAH-specific TQs in each sample was used as the sample TQ.Therefore, the TQ for one sample may be associated with benzo[a]pyrene, for example, whereas the TQ for another sample may be associated with fluorene.Unlike the other toxicity prediction methods used in our study, TQs do not account for the possible additive effects of PAH mixtures.The PAH-specific TQs were not summed for a total sample TQ because the underlying benchmarks were associated with different modes of action, which were not necessarily additive.Like the ΣESBTU, a TQ of ≥ 1.0 was used to indicate high potential for biological effects, and TQ < 0.1 was considered indicative of low potential for biological effects.In all cases, the minimum benchmarks used to calculate TQs were different from the EqP sediment benchmark final chronic value for freely dissolved PAHs (ESBFCV Dissolved ) used to calculate summed interstitial water and water column toxicity units described in the next subsection.

Sum interstitial water and water column toxicity units
The summed interstitial water and water column toxicity units (ΣIWTUs and ΣWCTUs) were calculated as where C free is the PED-measured PAH in the interstitial water (porewater) or water column, respectively, and ESBFCV Dissolved is the EqP sediment benchmark final chronic value for freely dissolved PAHs (Table 2; USEPA, 2012).The IWTU and WCTU for each of the 21 PAHs in a sample were summed to obtain the ΣIWTU and ΣWCTU.The ΣIWTU and ΣWCTU are similar to the ΣESBTU, except that the ΣIWTU and ΣWCTU use measured rather than estimated C free .Like the ΣESBTU and TQ, ΣIWTU and ΣWCTU values of ≥ 1.0 were considered to have high potential for biological effects, and ΣIWTU and ΣWCTU values < 0.1 were considered to have low potential for biological effects.

Exposure-activity ratios
Exposure-activity ratios based on high-throughput in vitro effects values from the USEPA ToxCast database (USEPA, 2022) have been used in recent studies for screening level assessments and prioritization (Blackwell et al., 2019;Pronschinske et al., 2022).The EARs were calculated using the toxEval R package (DeCicco et al., 2018) as where ACC is the assay-and chemical-specific concentration at which biological activity exceeds the baseline.Biological activity may include, for example, cellular stress responses, impacts on metabolic pathways, or endocrine disruption, and may or may not be adverse.Previous studies have used an EAR threshold of 0.001 as a level of potential concern (Alvarez et al., 2021;Bradley et al., 2023;Corsi et al., 2019 Mar 27;Oliver et al., 2023).In the present study, EAR values for individual PAHs were summed to get ΣEAR for each sample; ΣEAR values <0.001 were considered indicative of low potential for biological effects, and values of ≥0.1 were considered indicative of high potential for biological effects.

PAH concentrations
The PED sample results are available in the Supporting Information, Tables S5 and S6, and online (US Geological Survey, 2021).Bulk sediment PAH concentrations varied by orders of magnitude across sites and individual chemicals (Baldwin et al., 2020), with the low-molecular-weight chemicals having the fewest detections and lowest concentrations (Figure 3A).The PED porewater PAH C free exhibited similar variability and generally lower detections and concentrations for the lowmolecular-weight chemicals (Figure 3B).Field-based K OC values calculated from bulk sediment PAH concentrations and OC content and PED porewater C free (Equation 5) were up to three orders of magnitude greater than the average predicted K OC values from CompTox (USEPA, 2021), with low-molecularweight PAHs showing the greatest differences (Figure 3C).

Estimates of PAH toxicity
Summary of toxicity values from different methods.Three of the four toxicity estimation methods (ΣESBTU, TQ, and ΣIWTU/ΣWCTU) use a common value of 1.0 as a threshold of potential toxicity and therefore were compared directly.Estimates of potential toxicity to benthic organisms were similar between sediment ΣESBTUs and PED porewater TQs, with respective medians of 0.21 and 0.20, and means of 1.0 and 4.3 (Figure 4B and Supporting Information, Table S10).The sediment ΣESBTU threshold of 1.0 was exceeded at 13 of the 55 locations, and the PED porewater TQ threshold of 1.0 was exceeded at 14 locations, but these exceedances overlapped at only two sites (WI-NRL and WI-KKL).The PED porewater ΣIWTU values were generally lower than the sediment ΣESBTUs and PED porewater TQs by nearly an order of magnitude.However, a subset of sites had notably higher porewater ΣIWTU values (comparable to ΣESBTUs), and nine of these exceeded the ΣIWTU threshold of 1.0 (Figure 4B).The ΣESBTUs were significantly correlated with porewater TQs (r = 0.49, p < 0.001) and ΣIWTUs (r = 0.49, p < 0.001; Supporting Information, Figure S5A  and C).The PED Porewater ΣEAR values were also significantly correlated with ΣESBTUs (r = 0.49, p < 0.001; Supporting Information, Figure S5B) but were approximately two orders of magnitude lower (Figure 4B), which is expected given the lower threshold value for EAR evaluation.As discussed, ΣEAR values are best used for relative prioritization of chemicals with regard to the potential to produce biological effects that may or may not translate to risk.At six sites, PED porewater exceeded an ΣEAR value of 0.1.
The PED water column TQ, ΣEAR, and ΣWCTU values were generally lower than those for PED porewater samples, reflecting the lower PAH concentrations in the water column (Figure 4A).As observed with PED porewater samples, PED water column TQ values were approximately an order of magnitude greater than ΣWCTUs.The water column TQ and ΣWCTU thresholds of 1.0 were not exceeded at any PED water column location.

Site prioritization
Nine locations were identified as having high potential for toxicity using multiple methods (Figure 6).An additional 16 locations were identified as having high potential for toxicity based on a single method (sediment ΣESBTU or porewater ΣIWTU; Figure 6).Thirteen locations were identified as having low potential for PAH-related toxicity using all estimation methods.

DISCUSSION
Our study evaluated potential PAH-related biological effects in water and porewater from 55 Great Lakes tributary locations using measurements from bulk sediment samples and passive samplers.Estimations of porewater PAH C free values based on sediment PAH concentrations and EqP were 19-fold greater than PAH C free values measured using PEDs, on average (median 185-fold, max.960-fold; sum of 21 chemicals; Figure 4A).PEDs have been shown to accurately measure C free concentrations (Lotufo et al., 2022;Yan et al., 2022), indicating that the large discrepancy is likely attributable to overestimation of the EqP-derived concentrations.This conclusion is supported by comparison of field-based K OC values with literature-based CompTox K OC values (USEPA, 2021), which shows that CompTox K OC values underestimated the actual partitioning of PAHs from field sediments into water by 1 to 3 orders of magnitude for most of the PAHs studied (Figure 3C).Previous studies have reported similar differences between field-based and literature-based PAH K OC values (Burgess et al., 2021;Endo et al., 2020;Fernandez et al., 2009;Hawthorne et al., 2006).In Great Lakes tributaries, this may be the result of unusually high proportions of OC as BC in sediments.Black carbon was not measured in the present study, but a previous study reported the proportion of OC as BC to be >20% in lakebed surface sediments in Lake Erie, and up to approximately 30% to 35% in Lakes Michigan and Huron (Buckley et al., 2004).Coal tar particles may also contribute to the elevated K OC s in Great Lakes tributaries.Coal tar associated with pavement sealant has been identified as a primary source of PAHs to Great Lakes tributaries (Baldwin et al., 2017;Baldwin et al., 2020;Valentyne et al., 2018), andKhalil et al. (2006) showed that coal tar particles can dominate PAH partitioning even in the presence of relatively high proportions of BC.Fieldbased K OC s in the present study generally decreased with decreasing molecular weight PAHs, as others have observed (Endo et al., 2020;Hawthorne et al., 2006).However, PAHs with the lowest molecular weights (e.g., naphthalene, 1-and 2methylnaphthalene, acenaphthalene, acenaphthene) diverged from this pattern in the present study, with higher than expected K OC s (1-2 orders of magnitude greater than those reported by Hawthorne et al., 2006;Figure 3C).This divergence between fieldand literature-based K OC s among lowmolecular-weight PAHs (e.g., naphthalenes) may be associated with volatile loss during passive sampler retrieval and processing (Lotufo et al., 2022;Thomas & Reible, 2015), and (or) overcorrection of environmental concentrations (artificially lowering) through the subtraction of field blank concentrations.On the whole, the broad range of K OC s derived from samples in the present study-spanning multiple orders of magnitude (Figure 3C)-highlights the uncertainty associated with using a single K OC value to represent different sediments (Arp et al., 2009).Additional uncertainty in C free estimates comes from the K PED values, which have been shown to vary by up to 30% depending on the source of the polyethylene (Jonker, 2022).The K PED values used in the present study, obtained from USEPA (2012), were generally lower than those in Jonker (2022), potentially resulting in a low bias in C free .
The high ΣESBTU values relative to the ΣIWTU values (Figure 4B; median ΣESBTU:ΣIWTU ratio = 20) suggest potential overestimation of previous ΣESBTU-based toxicity estimates for Great Lakes tributaries (Baldwin et al., 2020).The ΣESBTU and the ΣIWTU and ΣWCTU are derived from the same National Water Quality Criteria Guidelines (Stephen et al., 1985) and are designed to be equally protective of aquatic life (USEPA, 2003(USEPA, , 2012)); the fundamental difference between the two methods is that the ΣESBTU relies on EqP-based estimation of porewater C free rather than direct measurement.Indeed, previous studies have observed similar overestimations of PAH C free and bioavailability determined through EqP-based methods such as the ΣESBTU (Burgess et al., 2021;Endo et al., 2020;Fernandez et al., 2009;Hawthorne et al., 2006).Endo et al. (2020) reported that the ΣESBTU overestimated PAH bioavailability by 18 to 930-fold, and Burgess et al. (2021) reported a <2 to 54-fold overestimation using ΣESBTU versus the ΣIWTU.However, a previous study of PAHs in sediments in the Milwaukee area (WI, USA) showed decreased mobility and survival in Hyalella azteca at ΣESBTU values <2.0, indicating that the ΣESBTU did not overestimate toxicity (Baldwin et al., 2017).These conflicting results may reflect the complexity of site-specific conditions controlling the partitioning of PAHs (Arp et al., 2009).Given the uncertainties in the ΣESBTU model (e.g., variability of K OC values), the potential overestimation of PAH bioavailability may provide a degree of environmental protection.
Although the use of passive samplers for porewater C free provides a more accurate assessment of bioavailable PAHs in Great Lakes tributaries than EqP-based estimation, the passive sampler-based toxicity (or bioactivity) predictions reported in the present study-specifically the ΣIWTU and ΣWCTU-are likely underestimated for several reasons.First, the ΣIWTU and ΣWCTU are meant to be the summed toxic units of 34 PAHs, but only 21 PAHs were included in the present study.Second, alkylated PAHs, which can contribute a disproportionate amount of the toxicity, were among those not analyzed in  passive samples.A study of 34 parent and alkylated PAHs spanning a wide range of concentrations reported that alkylated PAHs may account for up to 80% of total PAH toxicity in porewater (Hawthorne et al., 2007).Alkylated PAHs made up 29% (median) of total PAHs measured in the sediment samples in the present study (Baldwin et al., 2020).Based on these findings, more of the locations in the present study would likely approach or exceed the ΣIWTU and ΣWCTU threshold of 1.0 had alkylated PAHs been evaluated in passive samplers and included in this analysis.In future studies, inclusion of alkylated PAHs with available K PED values (USEPA, 2012) will provide a more accurate toxicity assessment.Lastly, passive sampler concentrations represent averages over the approximately 1-month duration of deployment and, in 2016, the three subsampling locations where passive samplers were deployed at each site.The sampled locations likely do not represent the greatest PAH concentrations in the area, given the spatial heterogeneity of PAHs in sediment, nor do the sampled concentrations represent maximum concentrations associated with short-duration events like urban storms, which may have acute biological effects.Among the biological effects assessment methods that used PED-measured C free , the ΣIWTU and ΣWCTU methods were the least sensitive in that, compared with other methods, they more frequently estimated a low potential for biological effects (74% and 100% of locations, respectively; Figure 6).In comparison, the TQ and ΣEAR methods estimated a low potential for biological effects at only 48% to 50% of locations based on porewater C free , and 86% to 96% of locations based on water column C free .
The current hazard assessment is focused on PAHs, but it is important to recognize that other contaminants may exert additional stresses on aquatic organisms.Many of the locations we identified as having high potential for PAH-related biological effects have been prioritized previously for potential effects associated with other organic chemicals in the sediment (e.g., alkylphenols, carbazole, and bis(2-ethylhexyl) phthalate; Baldwin et al., 2022) or water column (e.g., neonicotinoids and other pesticides, pharmaceuticals, and other wastewater-and stormwater-related chemicals; Alvarez et al., 2021;Corsi et al., 2019;Hladik et al., 2018;Pronschinske et al., 2022).
The use of PRCs in passive sampler studies is important for understanding the equilibrium behavior of the target contaminants.Ideally, the PRCs should reflect a range of K OW values representative of the target contaminants.However, studies are often limited in the selection of PRCs by considerations of availability, cost, and potential for overlap with analytical standards that may also be the stable isotope versions of the target contaminants.The PRCs in the present study (anthracene-d 10 , fluoranthene-d 10 , and pyrene-d 10 ) were selected based on these considerations but did not span the full range of K OW values of the target PAHs.As a result, there is greater uncertainty in the C free values for target PAHs above pyrene because equilibrium status had to be estimated.In future studies, the use of PRCs representing a wider range of K OW s would reduce this uncertainty, especially on the higher end of the range; for example, along with anthracene-d 10 and fluoranthene-d 10 or pyrene-d 10 , inclusion of the stable isotopic version of a higher K OW PAH such as chrysene.
Polycyclic aromatic hydrocarbons are common contaminants in aquatic environments, often occurring as complex mixtures.Although some PAHs exhibit lethal or sublethal effects, their bioavailability to aquatic organisms is uncertain due to location-specific sediment characteristics.We have shown how some of these challenges can be addressed through the use of passive samplers to measure the bioavailable PAH concentrations (C free ), and enhanced effects data sources to improve our understanding of both apical and nonapical effects.The findings of our study lend support to the increasing acceptance and use of passive samplers in screening level risk assessments of PAH-contaminated surface waters and sediments (Burgess et al., 2021;Greenberg et al., 2014;Mayer et al., 2014;McGrath et al., 2019), and highlight the potential overestimation of PAH bioavailability using a common sediment-based EqP method.The results provide an updated, weight-of-evidence-based site prioritization that may help guide future monitoring and mitigation efforts.Future assessments may benefit from measurement of, and toxicological data for, a greater number of PAHs.In addition, assessment of organic particle types and concentrations, including BC and coal tar particles, could improve our understanding of PAH partitioning in Great Lakes tributaries.
Supporting Information-The Supporting Information is available on the Wiley Online Library at https://doi.org/10.1002/etc.5896.

FIGURE 3 :
FIGURE 3: Individual polycyclic aromatic hydrocarbon (PAH) concentrations in bulk sediment (A), freely dissolved concentrations (C free ) in porewater measured using polyethylene passive sampling devices (PEDs; B), and field-based (blue) and literature-based (red) organic carbon-water partitioning coefficients (K OC ; C) at 55 Great Lakes tributary locations, 2016-2017.Field-based K OC values were calculated from sediment and PED porewater concentrations in (A) and (B) and organic carbon content using Equation (5).Literature-based K OC values are average predicted values from the CompTox Chemicals Dashboard (USEPA, 2021).

FIGURE 4 :
FIGURE 4: Summary of (A) freely dissolved concentrations of total polycyclic aromatic hydrocarbons (PAH C free ; sum of 21 chemicals) in porewater estimated from sediment concentrations using equilibrium partitioning (EqP) methods, and porewater and the water column measured using polyethylene passive sampling devices (PEDs); and (B) toxicity estimates derived from PAH C free in sediment and PED samples from 55 Great Lakes tributaries, 2016-2017.Colors correspond to the sample type.Horizontal dashed lines in (B) mark thresholds for high potential for PAH-related biological effects.ΣESBTU = sum equilibrium partitioning sediment benchmark toxic unit; TQ = toxicity quotient; ΣIWTU = sum interstitial water toxic unit; ΣWCTU = sum water column toxic unit; ΣEAR = sum exposure-activity ratio.

FIGURE 5 :
FIGURE 5: Relationships between concentrations of total polycyclic aromatic hydrocarbons (ΣPAH; sum of 21 chemicals) in bulk sediments and freely dissolved concentrations (C free ) in the water column measured using polyethylene passive sampling devices (PEDs) (A); in PED porewater C free and PED water column C free (B); in bulk sediment and in PED porewater C free (C); and estimated using sediment concentrations and equilibrium partitioning (EqP; Equation4) and in PED porewater C free (D) at 55 Great Lakes tributary locations, 2016-2017.Spearman correlation coefficient (r) and p values shown in blue; linear regression line shown in orange.

FIGURE 6 :
FIGURE 6: Site prioritization summary of potential PAH-related biological effects in sediment, porewater, and water column samples based on different toxicity estimation methods.Watersheds are ordered west to east, top to bottom, and locations are ordered upstream to downstream within each watershed (PAH = polycyclic aromatic hydrocarbon; ΣESBTU = sum equilibrium partitioning sediment benchmark toxic unit; TQ = toxicity quotient; ΣIWTU = sum interstitial water toxic unit; ΣEAR = sum exposure-activity ratio; ΣWCTU = sum water column toxic unit).

TABLE 1 :
Locations of passive sampler and sediment samples collected in 24 Great Lakes tributary watersheds, 2016-2017 a

TABLE 2 :
Polycyclic aromatic hydrocarbons (PAHs) analyzed in porewater and water column passive samplers and sediment samples from 55 GreatLakes tributaries, 2016Lakes tributaries,  -2017 a U.S. Environmental Protection Agency Priority PAH.CAS = Chemical Abstract Service; Log K OC = base 10 log of the organic carbon/water partitioning coefficient (USEPA, 2021); ESB FCV = equilibrium partitioning sediment benchmark final chronic value (USEPA