Predicting Pore Water EPA-34 PAH Concentrations and Toxicity in

May 19, 2011 - Traditional impacted site assessments comprise measuring sediment concentrations, Csed, of the U.S. Environmental Protection Agency (EP...
0 downloads 0 Views 5MB Size
ARTICLE pubs.acs.org/est

Predicting Pore Water EPA-34 PAH Concentrations and Toxicity in Pyrogenic-Impacted Sediments Using Pyrene Content Hans Peter H. Arp,*,† Nicholas A. Azzolina,*,‡ Gerard Cornelissen,†,§,|| and Steven B. Hawthorne^ †

)

 Department of Environmental Engineering, Norwegian Geotechnical Institute (NGI), P.O. Box 3930 Ulleval Stadion, N-0806 Oslo, Norway ‡ Department of Civil and Environmental Engineering, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, United States § Department of Applied Environmental Sciences (ITM), Stockholm University, 10691 Stockholm, Sweden Department of Plant and Environmental Sciences, University of Life Sciences (UMB), 1432 Ås, Norway ^ Energy and Environmental Research Center, University of North Dakota, Grand Forks, North Dakota 58201, United States

bS Supporting Information ABSTRACT:

Sediment and freely dissolved pore water concentrations of the U.S. Environmental Protection Agency’s list of 34 alkyl and parent PAHs (EPA-34) were measured in 335 sediment samples from 19 different sites impacted by manufactured gas plants, aluminum smelters and other pyrogenic sources. The total EPA-34 freely dissolved pore water concentration, Cpw,EPA-34, expressed as toxic units (TU) is currently considered one of the most accurate measures to assess risk at such sites; however, it is very seldom measured. With this data set, we address how accurately Cpw,EPA-34 can be estimated using limited 16 parent PAH data (EPA-16) commonly available for such sites. An exhaustive statistical analysis of the obtained data validated earlier observations that PAHs with more than 3 rings are present in similar relative abundances and their partitioning behavior typically follows Raoult’s law and models developed for coal tar. As a result, sediment and freely dissolved pore water concentrations of pyrene and other 3- and 4-ring PAHs exhibit good loglog correlations (r2 > 0.8) to most individual EPA-34 PAHs and also to Cpw,EPA-34. Correlations improve further by including the ratio of high to low molecular weight PAHs, as 2-ring PAHs exhibit the most variability in terms of their relative abundance. The most practical result of the current work is that log Cpw,EPA-34 estimated by the recommended pyrene-based estimation techniques was similarly well correlated to % survival of the benthic amphipods Hyalella azteca and Leptocheirus plumulosus as directly measured log Cpw, EPA-34 values (n = 211). Incorporation of the presented Cpw,EPA-34 estimation techniques could substantially improve risk assessments and guidelines for sediments impacted by pyrogenic residues, especially when limited data are available, without requiring any extra data or measurement costs.

’ INTRODUCTION Polycyclic aromatic hydrocarbons (PAHs) are formed from a variety of natural and anthropogenic processes and are common contaminants in sediments. The main sources of PAHs can be classified as either pyrogenic or petrogenic. Pyrogenic derived PAHs originate from oxygen-depleted, high-temperature processes such as incomplete combustion, pyrolysis, cracking, and destructive distillation. Two common pyrogenic sources include manufactured gas plants (MGPs) and metal smelters. Petrogenic r 2011 American Chemical Society

derived PAHs originate from petroleum, including crude oil, fuels, lubricants, and derivatives of those materials. The composition of PAHs from pyrogenic and petrogenic sources differ; for

Received: March 8, 2011 Accepted: May 6, 2011 Revised: April 25, 2011 Published: May 19, 2011 5139

dx.doi.org/10.1021/es2007935 | Environ. Sci. Technol. 2011, 45, 5139–5146

Environmental Science & Technology instance, alkyl PAHs contribute approximately 60% and 99% of their total PAH compositions, respectively.1 Traditional impacted site assessments comprise measuring sediment concentrations, Csed, of the U.S. Environmental Protection Agency (EPA) previously established list of 16 priority pollutant PAHs (EPA-16), which consists of only parent PAHs. However, the EPA now recommends the measurement of 18 parent PAHs and 16 groups of prominent C1- to C4-alkyl PAH derivatives (EPA34),2 where “C#” refers to the number of alkyl carbons attached to the PAH. These 16 alkyl PAH groups altogether comprise several hundred compounds, as many isomers of the alkyl PAH groups can exist. Due to their greater lipophilicity and abundance in impacted sediments, alkyl PAHs contribute a greater proportion of net toxicity than corresponding parent PAHs.3 Recent research has shown that the prevalent mechanism of PAH toxicity to benthic invertebrates is uptake from freely dissolved pore water fractions, Cpw, accumulation in the cell lipid, and narcosis, which results in the alteration of cell membrane function.2 Note that “freely-dissolved” refers to the concentration in the interstitial pore water not associated with colloids or dissolved organic matter, and which thereby is readily available to benthic invertebrates (below use of the term pore water will imply freely dissolved). Approaches to delineate Cpw and ecotoxicological risks using chemical methods are continuously evolving. These methods currently include empirically derived sediment quality guidelines (SQGs),47 mechanistic approaches using equilibrium partitioning (EqP) models to convert from measured bulk Csed to predicted Cpw,2,8 and direct measurements of Cpw.3,9 Currently, the best predictor of toxicity for PAH-impacted sediments is considered the directly measured, total Cpw of all EPA-34 PAHs, Cpw,EPA-34, expressed in toxicity units (TUs). This provides a more accurate prediction of biological effects than either SQGs or available EqP models based on Csed.3,810 However, current sediment quality indicators used by most regulatory agencies2,8,11 are not based on Cpw,EPA-34 but rather on benchmark Csed values of either some or a group of parent PAHs,11 as unfortunately few laboratories have the capacity to measure Cpw, particularly for alkyl-PAHs. There is, therefore, a disconnection between thousands of available site assessments performed worldwide based on Csed and their relevance to the state-of-the-art understanding of sediment PAH bioavailability and ecological risk, that is, Cpw,EPA-34. Since 2003, the Energy and Environmental Research Center (EERC) has measured both Csed and Cpw for EPA-34 PAHs on sediment samples collected from pyrogenic-residue impacted sites using internally consistent methods.1,9 In addition, these sediment samples were also evaluated for chronic toxicity to the benthic invertebrates Hyalella azteca (freshwater) and Leptocheirus plumulosus (seawater). To date Csed and Cpw are available for 335 samples from 19 sites, and toxicity tests are available for 211 of these samples. This is by far the largest compilation of such data. Subsets of less than half of these data have been used earlier to demonstrate that (i) measured Cpw,EPA-34 is a better predictor of H. azteca toxicity than those estimated from EqP models based on Csed3,10 and (ii) alkylPAHs dominate toxicity in pyrogenic impacted sediments.1 Instead of using this large data set to illustrate the shortcomings of current approaches used in risk assessments, in this paper we demonstrate how accurately log Cpw,EPA-34 can be estimated using limited data commonly attainable and available, and how well these estimations correlate with laboratory toxicity tests. As demonstrated, the recommended estimation method can readily facilitate more accurate assessments and sustainable regulations using limited,

ARTICLE

commonly available resources and Csed PAH data sets. We emphasize that this study is based on—and applies exclusively to—sites that are impacted mainly by pyrogenic residues.

’ THEORETICAL BASIS The basis for the Cpw,EPA-34 estimation techniques introduced here originates from two independent observations from the literature and our earlier data. The first observation is that the relative abundance of the EPA-34 compounds having three or more rings in pyrogenic-impacted sediments is largely similar regardless of the site.1,1214 The relative abundance ratio of two given PAHs is referred to here as A1,2: Csed, PAH1 =Csed, PAH2 ¼ A1, 2

ð1Þ

Previous studies have noted that for some pairs of PAHs, A1,2 is more consistent across diverse pyrogenic-impacted sediments than others, especially isomers like pyrene-fluoranthene, phenanthrene-anthracence and benz(a)anthracene-benzo(a)pyrene.1214 For 2-ring PAHs, however, their relative abundance can vary as a result of their higher volatility, solubility, metabolization and depletion rates. To account for this in our analysis, we include ”weathering ratios”, WEPA-34 and WEPA-16, which are defined as the ratio of the sum of 4-, 5-, and 6-ring PAHs to the sum of 2- and 3-ring PAHs in the sediment for the EPA-34 and EPA-16 data sets, respectively. The second observation employed for the presented Cpw,EPA-34 estimation technique is that, as recently elaborated in a critical review,15 the total organic carbon (TOC) normalized partition constant, KTOC, for PAHs at such sites typically follows Raoult’s law partitioning behavior: 

K TOC, PAH ¼ ðSL, PAH MW TOC Þ1

ð2Þ

Where S*L is the subcooled saturated molar water solubility (mol/ Lwater), MWTOC is the molar weight of the sediment’s organic phase. A possible explanation for this is that PAHTOC partitioning in such sites is also similar to coal tar,15,16 which closely follows Raoult’s Law,17 and is a plausable surrogate for carbonaceous material at pyrogenic-residue impacted sites. Note further that this partitioning model requires less assumptions and provides better precision than the increasingly utilized black carbon (BC) inclusive models18 (see also below). KTOC is related to overall sedimentpore water distribution ratio, KD, and the dry weight fraction of TOC, fTOC, by K TOC ¼ Csed =ðCpw 3 f TOC Þ ¼ K D =f TOC

ð3Þ

Combining the two observations implies that the Csed based A1,2 ratio for any two given PAHs should result in similar S*L,PAHCpw ratios, by combining eqs 13: 



Cpw, PAH1 =Cpw, PAH2 ¼ A1, 2 SL, PAH1 =SL, PAH2

ð4Þ

From here Cpw,EPA-34 can plausibly be related to individual PAH Csed by summing A1,2 for all EPA-34 PAHs, ∑34 i = 1Ai,j, providing that accumulative variability in all A1,2 is not too large: 34

∑ Csed, PAH i¼1 Csed, PAH j

i

¼

34

∑ A i, j i¼1

ð5Þ

On this basis correlations between Csed or Cpw values of individual PAHs with each other and Cpw,EPA-34 are investigated. 5140

dx.doi.org/10.1021/es2007935 |Environ. Sci. Technol. 2011, 45, 5139–5146

Environmental Science & Technology Frequently reported Csed or Cpw values for PAHs that exhibit the best correlations with Cpw,EPA-34 values would be favored for establishing regulatory benchmarks, for the practical reasons of method and data availability.

’ MATERIALS AND METHODS Sediment Collection. 335 sediment samples were collected (Ponar grab; sieved over a 4 mm mesh) from 18 freshwater and one marine site from 2003 through 2009. Fourteen of the sites were impacted by former MGP activities (n = 256), three by aluminum smelters (n = 59), and two by urban background (no identified source area, n = 20). Chemicals. The EPA-34 compounds are (with abbreviations in parentheses): 2-ring  naphthalene* (NAP), 1-methyl naphthalene (1MN), 2-methyl naphthalene (2MN), C2-naphthalenes (C2N), C3-naphthalenes (C3N), C4-naphthalenes (C4N), acenaphthene* (ACE), acenaphthylene* (ACEY); 3-ring  fluorene* (FLU), C1-fluorenes (C1F), C2-fluorenes (C2F), C3-fluorenes (C3F), anthracene* (ANT), phenanthrene* (PHE), C1-phenanthrenes/anthracenes (C1P), C2-phenanthrenes/anthracenes (C2P), C3-phenanthrenes/anthracenes (C3P), C4-phenanthrenes/ anthracenes (C4P); 4-ring  fluoranthene* (FLUA), pyrene* (PYR), C1-fluoranthenes/pyrenes (C1FP), benz[a]anthracene* (BaA), chrysene* (CHR), C1-chrysenes/benz[a]anthracenes (C1C), C2-chrysenes/benz[a]anthracenes (C2C), C3-chrysenes/ benz[a]anthracenes (C3C), C4-chrysenes/benz[a]anthracenes (C4C); 5-ring  perylene (PER), benzo[a]pyrene* (BaP), benzo[b]fluoranthene* and benzo[k]fluoranthene* (BBKF), benzo[e] pyrene (BeP), dibenz[a,h]anthracene*; 6-ring  indeno[1,2,3-cd]pyrene* (IND),benzo[ghi]perylene* (BGP). Parent PAHs indicated with an (*) belong to the EPA-16 list. Analytical Procedures. Measurement of PAHs was determined in all bulk sediment and pore water samples by the EERC using gas chromatography/mass spectrometry (GC/MS) using methods previously described.1,9 Detection limits and QA/QC from this approach are presented in the Supporting Information (SI). TOC fractions, fTOC (kgTOC/kgsediment), and BC fractions, fBC (kgBC/kgsediment), were determined using established methods.19 Toxicity Testing. Chronic toxicity to the freshwater amphipod, H. azteca, and marine amphipod, L. plumulosus, was conducted in accordance with the procedures outlined in U.S. Environmental Protection Agency (EPA) Test Method 100.420 and EPA 600/R01/020,21 respectively, on 211 of the 335 sediment samples (186 freshwater and 25 marine). The test design entailed the exposure (28 days) of four replicates of 10 amphipods to control, field reference, and test samples. All statistical comparisons were made at a 95% confidence level (p < 0.05) to the field reference sample(s). Samples determined to have significantly reduced survival in comparison to the field reference sample(s) were classified as “toxic”, whereas samples that were not significantly different were classified as “nontoxic”. Toxic Unit Calculations. As discussed in detail elsewhere,2,22,23 a toxic unit is a hazard quotient defined as the measured or modeled Cpw compared to the Cpw that is expected to result in toxic effects to a benthic organism. The aqueous PAH-specific final chronic value (FCV) (μg/L) was used to calculate the TU concentration associated with each individual PAH in the pore water (i.e., Cpw(TU) = Cpw(μg/L)/FCV) (see Table S1.1 in the SI). The TUs for each of the 34 PAHs analyzed were then summed to estimate the narcotic potential, Cpw,EPA-34, in each sediment

ARTICLE

sample.2 The benchmark value for the protection of sensitive benthic organisms is considered 1.0 TU.2 Data Analysis. All data were electronically transferred into Minitab Release 14.20 (Minitab, Inc.) and Microsoft Excel for data summary and analysis. Prior to evaluating the data in any statistical analyses, chemical concentrations were log-transformed (log10[X]) to reduce assumptions of normality and constant variance.24 The data reduction and analyses used the following process for both Csed and Cpw: (1) Identify sediment samples which had at least 12 detectable PAHs (n = 332); (2) Index the samples according to their WEPA-34 ratio; (3) Quantify the degree of correlation among each of the EPA-34 PAHs; (4) Isolate individual PAHs that have a high correlation with most of the other EPA-34 PAHs; (5) Use the identified PAHs to build predictive models of log Cpw,EPA-34 from log Csed and log Cpw values, based on the theory section above; (6) Assess the predicted log Cpw,EPA-34 for their ability to predict toxicity.

’ RESULTS Quantifiable PAHs. Sediment fTOC values ranged from 0.12 to 42.4% and fBC ranged from 0.03 to 39.7%. Total sediment concentrations, Csed,EPA-34, ranged from 0.23 to 30,784 mg/kg. Each of the EPA-34 PAHs could be quantified in the majority of sediment samples, with the exception of the C3-fluorenes (n = 4), C3-chrysenes/benz[a]anthracenes (n = 161), and C4-chrysenes/benz[a]anthracenes (n = 3). The remaining PAHs were quantifiable in 268 to all 335 of the sediment samples. Cpw,EPA-34 ranged from nondetect (ND) to 10,866 μg/L, and expressed as TU from ND to 310.8. In the pore water, there were several compounds that were quantifiable in less than 30% of the samples, including the C3-fluorenes (n = 2), C4-phenanthrenes/anthracenes (n = 43), C2-chrysenes/benz[a]anthracenes (n = 3), C3-chrysenes/ benz[a]anthracenes (n = 1), C4-chrysenes/benz[a]anthracenes (n = 0), perylene (n = 80), dibenz[a,h]anthracene (n = 22), indeno [1,2,3-cd]pyrene (n = 69), and benzo[ghi]perylene (n = 75). Cpw of the remaining PAHs could be quantified within 116 to 293 of the sediments (with pyrene, fluoranthene, phenanthrene, and anthracene being the most abundant). All data and additional summary statistics can be found in SI Section S1. Weathering Index. The sediments WEPA-34 index was < 0.5 for 19%, < 1 for 41%, < 2 for 76%, < 3 for 84%, and < 15.6 for 100% of sediments. Sediments with a WEPA-34 index > 3 all belonged to aluminum smelter sites with the exception of 2 MGP samples. No samples from aluminum smelter sites had a WEPA-34 index < 3, except for the reference background samples.25 PAH Ratios. Variability in the observed PAH ratios, A1,2, for Csed and Cpw were evaluated by calculating the correlation coefficients (r2) of the following loglog least-squares regression models:

log Csed, PAH1 ¼ msed log Csed, PAH2 þ bsed

ð6Þ

log Cpw, PAH1 ¼ mpw log Cpw, PAH2 þ bpw

ð7Þ

where msed and mpw are slopes, and bsed and bpw are intercepts, respectively. The resulting r2 noticeably decreased when increasing the range of WEPA-34 ratios considered. Because the samples with WEPA-34 < 2 (76% of data) are substantially more toxic than the remainder (see below), only this subset was used in the model calibration data sets, though all sediments are included in the validation data sets. 5141

dx.doi.org/10.1021/es2007935 |Environ. Sci. Technol. 2011, 45, 5139–5146

Environmental Science & Technology

ARTICLE

Figure 1. Loglog correlation coefficients of Csed (μg/kg) (a) and Cpw (TU) (b) for the EPA-34 PAHs among sediments (WEPA-34 < 2) following eq 6 and 7, respectively, with r2 > 0.8 in green and r2 > 0.6 in pink. The blue triangle indicates the generally good correlation between the 2-, 3-, 4-ring PAHs and the red triangle between the 4-, 5-, 6-ring PAHs. Anthracene, phenanthrene, fluoranthene, and pyrene appear to be suitable PAHs for estimating other PAH concentrations in general. Chart inlays show the difference between measured and modeled values for (a) log Csed,EPA-34 and (b) log Cpw,EPA34 for all 335 sediments based on pyrene concentrations using eq 8 and eq 9, respectively.

Figure 1a presents an r2 correlation matrix generated for Csed from eq 6 for the WEPA-34 < 2 subset. All msed and bsed coefficients can be found in SI Section S2, along with an r2 correlation matrix for the entire data set. Within a given ring size cluster, parent isomers and C1-alkyl-PAH isomers are generally highly

correlated with each other (r2 > 0.8). By contrast, the C2-, C3-, and C4-alkyl-PAHs do not correlate well with other PAHs of the same ring size (r2 ranging from 0.2 to 0.8). The exception to this is the 2-ring PAHs, where all but the C4-alkyls exhibit good correlations to each other (r2 > 0.6). Between adjacent ring sizes, 5142

dx.doi.org/10.1021/es2007935 |Environ. Sci. Technol. 2011, 45, 5139–5146

Environmental Science & Technology there are also good correlations between many parent and C1PAH isomers. The blue triangle in Figure 1a surrounding the 2-, 3,- and 4-ring PAHs shows that many of these PAHs are correlated with r2 > 0.6. The red triangle shows many of the 3-, 4-, 5-, and 6-ring PAHs are also correlated with r2 > 0.6. The worst correlations are between 2-ring and 5-, 6-ring PAHs (mostly r2 < 0.6), likely since their ratios change the most as a result of weathering or the PAH source (e.g., a newly exposed coal tar vs coal tar pitch). Figure 1b presents a Cpw r2 correlation matrix from eq 7. All mpw and bpw coefficients are in SI Secion S2. The correlations for Cpw (average r2 = 0.62) are not as good as those for Csed (average r2 = 0.66). However, as with Csed, the general patterns hold: the best correlations can be found among parent isomers and their C1-derivatives within a given ring size; the correlations tend to get worse with increasing alkylation and with increasing difference in ring sizes; Csed, 2-, 3-, and 4-ring PAHs usually exhibit correlations with r2 < 0.6 (blue triangle), as do the 4-, 5-, and 6-ring PAHs (red triangle); and, the worst correlations (r2 < 0.6) are for 2-ring vs 5- and 6- ring PAHs. Pyrene, fluoranthene, phenanthrene, and anthracene are the parent PAHs that exhibit the best correlations with other EPA-34 PAHs, and are thus highlighted in Figure 1. Overall, pyrene exhibits the highest average r2 value for log Csed correlations (0.77 ( 0.22) and was among the highest for log Cpw correlations (0.69 ( 0.16). Additionally, pyrene was the most commonly quantified compound in pore water (88% of samples), and among the most commonly quantified in sediments (98% of samples). Previous assessments noted that 3-ring alkyl PAHs comprise nearly 40% of the toxicity,3 which are shown here to be well-correlated with pyrene (Figure 1). Thus, pyrene is considered the best candidate for estimating total EPA-34 concentrations, and for the purpose of brevity, will be the focus of presenting Cpw,PAH-34 estimation methods herein. Use of anthracene, phenanthrene, and fluoranthene for such estimations is presented in SI Section S3. Least-square correlations of total log Csed,EPA-34 and log Cpw,EPA-34 concentrations with corresponding log pyrene concentrations are presented in eq 8 and 9, respectively. log Csed, EPA-34 ðμg=kgÞ ¼ 0:943ð ( 0:014Þlog Csed, pyrene þ 1:401ð ( 0:018Þðr 2 ¼ 0:95, rmse ¼ 0:27, n ¼ 245Þ ð8Þ log Cpw, EPA-34 ðTUÞ ¼ 1:177ð ( 0:030Þlog Cpw, pyrene þ 1:818ð ( 0:054Þðr 2 ¼ 0:88, rmse ¼ 0:45, n ¼ 221Þ ð9Þ From eq 8, log Csed,EPA-34 can be estimated within a factor of 3 for 97% of the 335 sediments, and within a factor of 10 for all sediments. Using eq 9, log Cpw,EPA-34 can be estimated within a factor of 3 for 70% of the total sediments and within a factor of 10 for 97% of sediments. Histograms plotting the difference between measured and modeled results of eqs 8 and 9 are presented in the inlays of Figure 1. To test the robustness of these correlations, a random subset of 50% of the data (n = 168) was removed and reassessed using eqs 8 and 9. The resulting models gave similar values, with slopes and intercepts differening by