The issue of additivity
One of the key assumptions in the use of mSQGq is that the contributions of each chemical to toxicity are additive. Because sets of commonly used SQGs include a wide variety of chemicals, including those with differing modes of toxicity, it is possible that the assumption of additivity does not apply in all situations (Batley et al. 2005). For example, antagonism between different chemicals in clean spiked sediments has been reported in laboratory experiments (Oakden et al. 1984).
However, there is evidence of additivity in bioassays of clean sediments spiked with individual chemicals and combinations of these chemicals (McLeese and Metcalfe 1980; McLeese et al. 1982; Sundelin 1984; Plesha et al. 1988; Swartz et al. 1988). Additivity among priority chemicals has been reported in whole effluent toxicity tests (Sarakinos et al. 2000). Models have been developed for estimating risks of chemical mixtures in tissue residues and water (Dyer et al. 2010) and in sediments (Field et al. 2002). Some organic compounds indicated synergistic toxicity (i.e., more than additive) in various combinations (Pape-Lindstrom and Lydy 1997). Van Gestel et al. (2011) recently summarized the state of the science in evaluations of chemical mixtures and predicting their effects.
Evidence of Ecotoxicological effects in Puget Sound
Evidence from both laboratory toxicity tests and various benthic metrics indicate that biological conditions changed as mSQSq increased incrementally between the 4 chemical categories (Table 8). Clearly some biological metrics did not change with increasing contamination, for example, amphipod survival in laboratory toxicity tests. Conflicting results among the biological metrics underscores the need for multiple metrics. The weight of evidence from the multiple analyses suggests that there are 3 cut points in the mSQSq (i.e., 0.1, 0.3, and 0.5) where most biological measures changed. Different kinds of adverse effects occurred within the 3 elevated ranges in mSQSq.
Table 8. Summary of statistically significant increases (+) and decreases (−) in biological measures from either the Minimum exposure or both the Minimum and Low exposure categories
|Median biological measure||1. Minimum exposure (mSQSq <0.10)||2. Low exposure (mSQSq 0.10– < 0.30)||3. Moderate exposure (mSQSq 0.30– < 0.50)||4. Maximum exposure (mSQSq >0.50)|
| Amphipod survival (%)||ns||ns||ns||ns|
| Sea urchin fertilization (%)||ns||ns||ns||b|
| Microtox EC50||ns||−a||ns||ns|
| Cytochrome P-450 induction||ns||+a||+b||+b|
|Calculated benthic indices|
| Total abundance||ns||ns||+a||+a|
| Taxa richness||ns||ns||+b||+a|
| Pielou's evenness index||ns||−a||ns||ns|
| Swartz's dominance index||ns||ns||ns||ns|
|Abundance of benthic phyla|
|Abundance of selected species or taxa groups|
| Cirratulid polychaetes||ns||+a||+a||+a|
| Spionid polychaetes||ns||ns||+a||+a|
| Axinopsida serricata||ns||ns||+b||+a|
| Other mollusks||ns||ns||+a||ns|
| All amphipods||ns||−a||−a||−a|
| Phoxocephalid amphipods||ns||ns||ns||ns|
| Other amphipods||ns||ns||−a||ns|
Samples in the Minimum Exposure category 1 were the least toxic in terms of both the percent incidence and degree of toxicity. Mean and median amphipod percent survival were relatively high. Median sea urchin percent fertilization was highest and the incidence of toxicity was the lowest. Microtox EC50s were relatively high. Both the incidence and degree of response in the HRGS assays were the lowest in this category. Samples in this category had the lowest abundance of stress-tolerant benthic phyla and taxa, and the highest abundance of stress-sensitive phyla and taxa. The median abundance of benthic mollusks, including the stress-tolerant bivalve A. serricata, was lowest. Median abundance of the annelids, including the stress-tolerant cirratulids and spionids, was lowest. Median abundance of amphipods, including the stress-sensitive phoxocephalids, was the highest. Therefore, there is evidence that mSQSq values less than 0.1 are highly protective of marine benthic resources in Puget Sound. About 71% of the 664 samples were in this category.
Relative to the samples in category 1, samples in the Low Exposure category 2 had significantly higher cytochrome P-450 induction, significantly reduced Microtox EC50s, and increased incidence of toxicity in all 4 laboratory tests. The benthic evenness index was significantly reduced. There were significant increases in abundance of stress-tolerant annelids and cirratulid worms. There were significant decreases in the abundance of all amphipods and a nonsignificant trend of decreased abundance of phoxocephalid amphipods. The median abundance of the phoxocephalids was zero.
In Moderate Exposure category 3 samples, median and mean cytochrome P-450 induction increased 19-fold relative to category 1, and the incidence of toxicity in the HRGS assays increased to 71.4%. Total benthic abundance and total benthic taxa richness increased as a result of increased abundance of stress-tolerant annelids, cirratulids, spionids, A. serricata, and other mollusks. In contrast, the abundance of stress-sensitive amphipods decreased significantly relative to category 1. The mean abundance of echinoderms was about one-third of that in category 1. The median abundance of the phoxocephalids was zero.
In the Maximum Exposure category 4 samples, median sea urchin fertilization decreased significantly, median cytochrome P-450 induction increased significantly (a 25-fold increase relative to category 1), all of the samples were classified as toxic in the HRGS assays, and the incidence of toxicity to the sea urchins was the highest. The incidence of toxicity in the Microtox tests was the highest. Median total benthic abundance and taxa richness were significantly elevated as a result of the increased abundance of the stress-tolerant annelids, mollusks, cirratulids, spionids, A. serricata, and other mollusks. The mean abundance of stress-sensitive arthropods was about one-half of that in category 1. Median abundance of the amphipods was reduced significantly. Median and mean abundance of benthic echinoderms were reduced, but not significantly. The median abundance of the phoxocephalids was zero.
We speculate that some relatively tolerant animals are attracted to the organic matter as food in the soft, depositional sediments that tend to also accumulate higher chemical concentrations, but that are not lethal to them. This relationship has been reported previously for areas of Santa Monica Bay, California (Bergen et al. 2001) and Puget Sound (Nichols 2003). Their changes in abundance generally correspond to the nonlinear empirical model developed along stress gradients in Scandinavian fjords (Pearson and Rosenberg 1978). The Spearman rank correlation coefficient (ρ) for mSQSq and percent total organic carbon (TOC) was 0.31, indicating a weak trend of covariance.
Comparisons with previous evaluations of toxicity
This study is the first of its kind performed with the Puget Sound SQS (Ingersoll et al. 2005; Word et al. 2005). It was conducted using methods previously reported with other kinds of sediment quality guidelines (SQG) and field validation data sets (Long et al. 2000; Fairey et al. 2001; Ingersoll et al. 2005). All have been carried out with either Effects Range Median (ERM) values, Probable Effects Level (PEL) values, or consensus freshwater Probably Effect Concentrations (PEC). None were previously performed with the Washington SQS.
The SQS and ERM values were derived with different statistical methods carried out with different kinds of data. Concentrations of organic compounds were treated on a dry weight basis in the ERMs, whereas they were normalized to organic C in the SQS. There are 47 SQS (Ecology 1995), whereas there are 28 ERM values (Long et al. 1995). The SQS were derived for many more organic compounds (e.g., phenols, phthalates, chlorinated benzenes) than the ERM values.
Despite these differences, the distributions of mSQSq and mERMq calculated for the same 664 Puget Sound samples were similar or the same within each of the 4 exposure categories (Table 9). For example, the medians of the calculated mERMq and mSQSq were 0.12 and 0.16, respectively, in the Low Exposure category and they were 0.51 and 0.69 in the Maximum Exposure category. Also, the numbers of ERMs and SQSs exceeded in the samples within each category were similar or exactly the same. Therefore, because the mERMq and mSQSq track well with each other, it appears reasonable to compare the predictive abilities of the mSQSq in Puget Sound with previous evaluations elsewhere of the predictive abilities of mERMq.
Table 9. Distribution of mERMq and mSQSq values within Puget Sound sediment samples over 4 exposure categories
|Summary statistics||mERMq||mSQSq||Nr ERMs exceeded||Nr SQSs exceeded|
|1. Minimum exposure (mSQSq <0.10)|
| 95% CI||0.04–0.45||0.05–0.05||na||na|
|2. Low exposure (mSQSq 0.10– < 0.30)|
| 95% CI||0.11–0.15||0.15–0.17||na||na|
|3. Moderate exposure (mSQSq 0.30– < 0.50)|
| 95% CI||0.19–0.39||0.37–0.43||0.00–1.00||1.00–2.00|
|4. Maximum exposure (mSQSq ≥0.50)|
| 95% CI||0.33–1.05||0.59–0.78||1.00–9.20||3.00–11.00|
Ingersoll et al. (2005) compiled matching chemistry, toxicity, and benthic data from numerous studies to evaluate the predictive abilities of the various sets of SQGs. They assembled data from more than 8000 samples in peer-reviewed published reports for both saltwater and freshwater to evaluate how well SQGs accurately predicted either the absence or the presence of toxicity and/or adverse benthic effects. Most of the data compiled to evaluate predictive abilities of SQGs focused on tests of amphipod survival as the biological measure (Long et al. 2000; Fairey et al. 2001; Ingersoll et al. 2005). Note that Long et al. (2000) evaluated both the PELs and ERMs with the same database. As is typical of data encountered from ambient monitoring programs, the numbers of samples invariably decreased as the degree of contamination increased. These data are compared with those from Puget Sound (Table 10).
Table 10. Numbers of samples, average percent survival, and percent incidence of toxic samples in amphipod survival tests in the Puget Sound database and in 3 national databases
|Study||Nr samples||Totals (%)||Average % survival||Incidence of toxicity (%)|
|Present study (ranges in mSQSq, test species were A. abdita and E. estuarius)|
| < 0.10||399||70.7||96.4||3.8|
| 0.10– < 0.30||131||23.2||95.6||6.1|
| 0.30– < 0.50||23||4.1||97.2||0.0|
| Total||564|| || || |
| (ranges in mERMq, test species unknown)|
| < 0.10||1392||50.3||n/a||9.0|
| > 1.00||155||5.6||n/a||61.0|
| Total||2767|| || || |
| (ranges in mERMq, test species were A. abdita and R. abronius)|
| > 1.50||28||1.9||41.0||76.0|
| Total||1513|| || || |
| (ranges in mPELq, test species were A. abdita and R. abronius)|
| > 2.30||45||3.0||46.0||73.0|
| Total||1513|| || || |
| (ranges in mSQGQ1, test species were A. abdita, R. abronius, E. estuarius)|
| > 3.50||10||0.6||13||100|
| Total||1692|| || || |
The incidence of toxicity to amphipods in samples with mSQSq less than 0.1 invariably was less than 10% in Puget Sound and in 3 national databases (Table 10). In the national databases, the incidence of toxicity invariably increased to about 20% to 30% in mSQGq ranges of 0.1 to 0.5 or 0.1 to 1.5, whereas in Puget Sound the incidence remained low (i.e., 6.1%). The incidence of toxicity increased again to about 40% to 50% in mSQGq ranges of 0.5 to 1.0, 0.51 to 1.5, and 1.51 to 2.3, whereas in Puget Sound it dropped to zero. The incidence of toxicity peaked at 61% to 100% in the highest ranges in the national databases, whereas it remained at zero in Puget Sound.
These comparisons also demonstrate that the degree of toxicity, expressed as average percent survival, decreased incrementally in the national databases, but not in Puget Sound (Table 10). In both Puget Sound and in the national databases, average percent survival was greater than 90% in the lowest ranges in mSQSq, usually less than 0.1. However, in the national databases the average percent survival decreased immediately in the next highest range in mSQGq and continued to decrease in the subsequent higher ranges, whereas in Puget Sound it remained at greater than 90%.
These data provide evidence that the initial cut point of less than 0.1 is highly protective against acute toxicity and nearly a universal safe level. These comparisons also demonstrated that for unknown reasons the amphipod survival tests did not respond to higher levels of contamination in Puget Sound, whereas they consistently did so elsewhere. In contrast, the HRGS assay and sea urchin test responded to the higher levels of contamination in the Sound.
All of these national analyses were performed with data gathered from ambient monitoring programs in U.S. waters. None of them involved data collected specifically to test the predictive abilities of SQGs. Invariably, sample sizes decreased as the degree of chemical contamination increased. In contrast, the results of a study designed specifically to test predictive abilities of SQGs was reported for Sydney Harbor, Australia (McCready et al. 2006). Analyses of tests of 106 samples showed that the incidence of toxicity in laboratory tests was very low in samples with mERMq less than 0.1 (i.e., 0% with amphipods alone and 11% in any 1 of 4 tests). The incidence of toxicity to amphipods peaked at 40% with amphipods alone and 73% in any test where the mERMq values were greater than 1.5. Therefore, the Australian data were similar to the national US data but not the Puget Sound data.
Comparisons with previous evaluations of adverse benthic effects
The 3 cut points identified for mSQSq and the resulting ranges in values identified with a combination of toxicity and benthic data for Puget Sound agree remarkably well with cut points identified elsewhere with other kinds of SQGs and biological data (Table 11). For example, the cut points for Minimum Exposure categories ranged from 0.013 to 0.15, thereby bracketing the Puget Sound mSQSq value of 0.10. The cut points for the Maximum Exposure category ranged from 0.062 to 1.362, also bracketing the Puget Sound mSQSq value of 0.50.
Table 11. Comparisons among US saltwater regions in ranges in mSQGq within narrative categories of sediment contamination as established with adverse benthic effects
|Study area||Publication||Unit of measure||Exposure categories|
|Puget Sound||Present study||mSQSq||<0.1||0.1– < 0.3||0.3– < 0.5||≥0.5|
|Southern California||Ritter et al. 2011||mERMq||<0.06||0.06– < 0.13||0.13– < 0.40||≥0.40|
| || ||Consensus-based SQGs||<0.15||0.15– < 0.30||0.30– < 0.68||≥0.68|
|Virginia province||Hyland et al. 2003||mERMq||≤0.022||>0.022–0.098||>0.098–0.473||>0.473|
|Carolina province||Hyland et al. 2003||mERMq||≤0.018||>0.018–0.057||>0.057–0.196||>0.196|
|Louisiana province||Hyland et al. 2003||mERMq||≤0.013||>0.013–0.036||>0.036–0.062||>0.062|
|San Francisco Bay, CA||Thompson and Lowe 2004||mERMq||<0.051||0.051–0.146||0.147–0.635||>0.742|
|San Diego Bay, CA||Thompson et al. 2009||mERMq||<0.054||n/a||n/a||>1.362|
|Biscayne Bay and Miami River, FL||Long et al. 2002||mERMq||<0.03||0.03–0.2||n/a||>0.2–2.0|
Matching chemistry, toxicity, and benthic data from 441 samples collected in marine bays of southern California were analyzed to develop an effects-based sediment chemistry score index (Ritter et al. 2011). Five toxicity-based sets of SQGs and a newly developed benthic index were evaluated as candidate methods. One of the many outcomes of this study was to identify thresholds in the 6 candidate index methods, using the results of toxicity tests with marine amphipods and a benthic health index. Thresholds of low, moderate, and high were identified as mERMq of 0.06, 0.12, and 0.38 with the toxicity data and 0.06, 0.13, and 0.40 with the benthic data. With consensus-based SQGs, the cut points were 0.15, 0.30, and 0.68 with the benthic data (Table 11).
Evaluations of matching mERMq and benthic assemblage data from the Virginia, Carolina, and Louisiana estuarine provinces showed a pattern of increasing incidence of adverse effects with increasing contamination in samples that were not toxic in amphipod survival tests (Hyland et al. 2003). In the 3 provinces in order, the incidence of degraded benthos was lowest (i.e., 9%, 7%, 30%) in samples with mERMq of less than or equal to 0.022, less than or equal to 0.018, and less than or equal to 0.013, respectively. The incidence increased to 31%, 43%, and 52% with mERMq of greater than 0.022 to 0.098, greater than 0.018 to 0.057, and greater than 0.013 to 0.036, respectively. The incidence increased incrementally in 2 subsequent ranges calculated for each province, reaching a maximum at 85% to 100% among the 3 provinces. The cut points for the maximum exposure category (i.e., 0.473, 0.196, and 0.062) differed among the 3 regions (Table 11).
In San Francisco Bay (CA), the incidence of benthic impacts numerically corresponded with increasing mERMq in a regional monitoring data set (Thompson and Lowe 2004). No benthic impacts were observed in 13 samples with mERMq less than 0.051. Impacts were recorded in 9.4% of 106 samples with values between 0.051 and 0.146, in 63.2% of 19 samples with values between 0.147 and 0.635, and in all 3 samples with values greater than 0.742 (Table 11). Whereas only 8.4% of samples with mERMq values below 0.146 were impacted, 68.2% were impacted with values above 0.146. The authors concluded that a mERMq value of 0.146 was reasonably predictive and protective.
In a database composed of 161 samples assembled from 4 previous surveys of San Diego Bay (CA), adverse benthic impacts were correlated with indices of sediment contamination, including mERMq (Thompson et al. 2009). Samples with mERMq less than 0.054 were never impacted, whereas samples with mERMq greater than 1.362 were always impacted (Table 11).
In data from Biscayne Bay and the lower Miami River (FL) (Long et al. 2002), average amphipod survival decreased from 96% to 93% and 41% in mERMq ranges of less than 0.03, 0.03 to 0.2 and greater than 0.2 to 2.0, respectively (Table 11). Over these same 3 ranges, average numbers of species in the benthos decreased from 91 to 58 and 6 per sample and average numbers of arthropods decreased from 179 to 96 and 10 per sample, respectively.
Meta-analyses of ranges in amphipod survival in laboratory toxicity tests and matching benthic assemblage data from surveys conducted along all 3 coastlines of the United States have empirically demonstrated a trend of losses of or absence of relatively sensitive amphipods, other crustaceans, and echinoderms with decreasing percent survival (Long et al. 2001). Similar outcomes, albeit with differing benthic metrics, were reported for studies conducted in estuarine Rhode Island (Kuhn et al. 2002), Elliott Bay in Puget Sound (Ferraro and Cole 2002), and Baltimore Harbor and vicinity in Chesapeake Bay (McGee et al. 2004).
In our present study of samples from Puget Sound, amphipods were present in the minimum exposure category samples, but median abundance of all amphipods decreased significantly and the phoxocephalids were absent in the low, moderate, and maximum exposure categories. However, mortality to amphipods in the laboratory tests did not change significantly across these 4 ranges. We speculate that, along with the increasing degree of chemical contamination, other natural factors, such as low dissolved O2 concentrations, changes in depth, changes in sediment particle sizes, changes in organic C content among the sediment samples had an influence on the benthos, but not in the toxicity tests.
Implementing SCI-based recovery targets for Puget Sound
The development of mSQSq and their associated SCI values, biologically and toxicologically-relevant cut points, and the 4 associated ranges in sediment quality on the 0 to 100 scale, have enabled the PSP leadership to adopt 2 of 3 recovery targets for the Marine Sediment Quality Dashboard Indicator. These 2 target values were based on the SCI range associated with the minimal exposure category. They include the following (PSP 2011):
“By 2020, all Puget Sound regions and bays should:
Have sediment chemistry measures reflecting ‘minimum exposure’, as defined by having a Sediment Chemistry Index (SCI) score of >93.3.
Have no measurements exceeding the Sediment Quality Standard values set in Washington State.”
These 2 targets are derived directly from the analysis of the data described in this report. A third is derived from the derivation of the revised SQTI, and will be described in a later article (in preparation).
Environmental managers implementing the Action Agenda will make use of the Dashboard Indicators and their associated targets as they identify, design, prioritize, and fund strategies and activities that contribute to ecosystem recovery. The indicators and targets will then be used to evaluate the effects of these investments and activities toward the recovery of Puget Sound by 2020.