Sediments are a repository for many contaminants released into surface waters. Because of this, organisms inhabiting sediments may be exposed to a wide range of contaminants (United States Environmental Protection Agency (USEPA) United States Environmental Protection Agency 2000; American Society for Testing and Materials [ASTM] American Society for Testing and Materials International 2012). Contaminants of potential concern in sediments typically include trace elements (metals), organochlorine (OC) pesticides, polychlorinated biphenyls (PCBs), and polycyclic aromatic hydrocarbons (PAHs; Ingersoll et al. 2001). In 2000, the USEPA began to restrict the use of organophosphate pesticides, such as diazinon and chlorpyrifos (Spurlock and Lee 2008). These restrictions have led to increased use of pyrethroid pesticides, which have widespread applications in both agricultural and urban environments (Kuivila et al. 2012).

Pyrethroids are hydrophobic compounds that have been observed to accumulate in sediments (Laskowski 2002). Toxicity of pyrethroids in field-collected sediment from small urban streams (Weston et al. 2005; Holmes et al. 2008; Ding et al. 2010; Domagalski et al. 2010) or with pyrethroids spiked into sediment (Amweg et al. 2006; Hintzen et al. 2009) have been evaluated primarily in 10 day lethality tests conducted with the amphipod Hyalella azteca. However, the sublethal effects in long-term exposures to pyrethroids in sediment have not been evaluated, and the distribution of pyrethroids sediments has not typically been evaluated in wadeable streams (Gilliom et al. 2006).

This article is the second in a series that describe the results of a study of the distribution and toxicity of pyrethroids and other co-occurring trace elements and organic contaminants (PCBs, PAHs, OC pesticides) in stream sediments from 7 metropolitan areas across the United States (Moran et al. 2012). The study evaluated 98 sediment samples collected from streams ranging from undeveloped to highly urban and differs from previous studies by sampling larger wadeable streams and avoiding point sources (such as storm drains) and other inflows (Gilliom et al. 2006). Part 1 of the series characterizes sediment contaminants in relation to urbanization and other factors in the 7 metropolitan study areas (Nowell et al. 2012). Part 2 (this article) evaluates relationships between sediment chemistry and sediment toxicity in 28 day whole-sediment exposures conducted with the amphipod H. azteca and in 10 day whole-sediment exposure conducted with the midge Chironomus dilutus (USEPA United States Environmental Protection Agency 2000; ASTM American Society for Testing and Materials International 2012). Toxicity end points evaluated in the amphipod and midge exposures included the effects of these field-collected sediments on survival, weight, or biomass of the test organisms.

Materials and Methods

Sediment Collection and Characterization

Surficial bed sediment samples were collected in 2007 from 98 sites in 7 metropolitan study areas: Atlanta (GA); Boston (MA, NH); Milwaukee-Green Bay (WI [Milwaukee]); Dallas-Fort Worth (TX [Dallas]); Denver (CO, WY); Salt Lake City (UT); and Seattle-Tacoma (WA [Seattle]; [see Table 1 and see Fig. 1 in Nowell et al. 2012]). Sites within each study area ranged in urban intensity from undeveloped (<5 % urban land) to highly urbanized (Nowell et al. 2012). Most sampling sites were in wadeable streams in small- to moderate-sized basins (see Figs. 2–8 in Moran et al. 2012). One sediment sample was collected from each of the 98 sites, which ranged from approximately 150–300 m stream length depending on stream width (length sampled was approximately 20 times the average wetted width). Each sample was a composite of multiple grab samples of recently deposited material collected by skimming approximately the top 2 cm of sediment with Teflon sheeting or a clean small scoop from multiple depositional zones within the stream reach during low-flow conditions.

Table 1 Whole-sediment toxicity to the amphipod H. azteca (28 day exposures) or to the midge C. dilutus (10 day exposures) by different end points compared with reference sediments within and across the study areas
Full size table
Fig. 1
figure 1

Relationship between sediment chemistry and survival of H. azteca (HA) (Figs. 1a–e) or total biomass of C. dilutus (CD) (Fig. 1f). Study areas are represented by the following symbols: pink square = Dallas; red diamond = Denver; green hexagon = Salt Lake City; blue circle = Milwaukee; grey triangle = Boston; turquoise triangle = Atlanta; orange star = Seattle. Symbols outlined in black represent toxic samples with an end point response below the lower limit of a reference envelope value within a study area. The vertical line represents the appropriate sediment toxicity threshold (from Table 2 or 3) for each measure of sediment chemistry

Full size image

Sediment samples were analyzed for trace elements, OC pesticides, PAHs, PCBs, pyrethroids, total organic carbon (TOC), and grain size (Moran et al. 2012). Sediments were sieved in the field to <2 mm for organic contaminant analyses and toxicity testing and to <63 μm for trace-element analyses. Porewater was isolated from the sediments at the start of the toxicity tests by centrifugation at 4 °C for 15 min at 5,200 revolutions/min; Kemble et al. 1994) and was analyzed for ammonia, pH, hardness, alkalinity, and conductivity when sample volume was sufficient (n = 32 of 98 samples).

Sediment-Toxicity Tests

Sediment-toxicity tests were conducted with amphipods (28 day exposures) and midge (10 day exposures) according to methods outlined in USEPA (United States Environmental Protection Agency 2000) and ASTM (American Society for Testing and Materials International 2012) and listed in Table S1. Sediments were homogenized in a stainless steel bowl using a plastic spoon and added to exposure beakers 1 day before test organisms were added (day-1). Ten amphipods or midge were exposed to 100 ml sediment with 175 ml overlying water in 300 ml beakers with a total of four replicates per treatments per species with two volume additions/d of overlying water (Ingersoll et al. 2002). The source of overlying water was well water diluted with deionized water to a hardness of approximately 100 mg/L (as CaCO3), alkalinity of approximately 95 mg/L (as CaCO3), and pH of approximately 8.2. Amphipods in each beaker were fed 1.0 ml yeast–cerophyll–trout chow (1.7–1.9 g/L) in a water suspension daily. Midge in each beaker were fed 1.5 ml Tetrafin goldfish food (6.0 mg of dry solids [Tetra, Blacksburg, VA]) in a water suspension daily (USEPA United States Environmental Protection Agency 2000; ASTM American Society for Testing and Materials International 2012). Toxicity end points for midge included 10 day survival, ash-free dry weight (AFDW), and total biomass (total biomass was calculated as the sum AFDW for all surviving organisms in each replicate chamber). Toxicity end points for amphipods included 28 day survival, length, weight, and total biomass. Surviving amphipods on a sampling date were preserved in 8 % sugar formalin for subsequent 28 day length measurement. The biomass of surviving amphipods from each replicate was estimated as the sum of individual amphipod weights calculated from the following empirical relationship: weight (mg) = ((0.177* length (mm)) − 0.0292)3 (Ingersoll et al. 2008; Moran et al. 2012).

Toxicity tests were conducted in four batches and were started within 1 month of sediment collection. Batch 1 testing started in May 2007 and included 21 samples from Seattle. Batch 2 and batch 3 testing started in June 2007 and included 13 samples from Atlanta and 13 samples from Dallas. Batch 4 testing started in October 2007 and included 51 samples from four study areas: Salt Lake City, Boston, Denver, and Milwaukee (Table 1). A control sediment collected from West Bearskin Lake, MN (approximately 2 %TOC; Ingersoll et al. 1998) was tested with each batch of sediments. Two changes were made to the USEPA (United States Environmental Protection Agency 2000) and ASTM (American Society for Testing and Materials International 2012) methods in exposures conducted with batch 3 and 4 sediments to improve control response of midge (Ingersoll et al. 2008) as follows: (1) exposures were started with approximately 7 day old midge larvae (rather than 10 day old larvae) to decrease the possibility of adult midge emerging by the end of a 10 day sediment exposure; and (2) exposures were started with midge larvae isolated from cultures still in their surrounding culture tubes rather than with larvae that have left (or have been removed from) their culture tubes. Larvae outside of their culture tubes may not be as healthy as larvae still inside their culture tubes (Dave Mount, USEPA, Duluth MN, personal communication, April 2007). Using these modifications to the midge exposure methods, our laboratory has seen improvement in sediment-toxicity tests conducted with midge during the past 5 years (e.g., control survival typically >90 % in 10 day exposures).

Data Analysis

Toxicity was established using a reference envelope approach by comparing responses of test organisms in test sediments to those of test organisms in reference sediments (Hunt et al. 2001; Ingersoll et al. 2009; Besser et al. 2009). The range of mean responses of test organisms in reference sediment samples with minimal levels of contamination was assumed to represent the normal range of responses of test organisms to uncontaminated sediments within a study area. The requirements for establishing a sample as a reference sediment for each toxicity test within a study area included the following: (1) mean PECQ for five classes of compounds (mean PECQ-5B) <0.1 (see later text for a description of the PECQ calculation), (2) a significant difference between sediments within a study area for a given toxicity end point, and (3) mean survival or weight of test organisms in a sample that met control test acceptability requirements (>80 % survival of amphipods, >70 % survival of midge, and >0.48 mg AFDW/individual midge; USEPA United States Environmental Protection Agency 2000, ASTM American Society for Testing and Materials International 2012; Tables S1 and S2). A minimum of three reference samples were needed to establish a reference envelope within a study area for that test organism (Table S3). A test sediment within a study area was classified as toxic to a test organism if the mean response of one or more toxicity end points (survival, weight, or total biomass) in that sample was less than the lowest mean for reference samples from the same study area. The mean responses of test organisms in each sediment sample were expressed as a percentage of the median responses of test organisms in reference sediments within a study area (Tables S2 and S3); this helps to account for the potential influence of physical characteristics of the sediment samples within a study area on the response of test organisms (e.g., TOC, grain size).

Statistical analyses were performed using SAS statistical software (SAS/STAT version 9.2; SAS Cary, NC). Differences in a toxicity end point among sites within a study area and within batch of sediments tested were determined by analysis of variance (ANOVA). Toxicity end point data were transformed before ANOVA to improve normality as indicated by Shapiro–Wilk test (USEPA United States Environmental Protection Agency 2000; ASTM American Society for Testing and Materials International 2012). If transformations (arcsine square root for survival; square root or log for weight or total biomass) did not improve normality, data were rank-transformed before analysis (Conover and Iman 1981). Spearman rank correlation analysis was used to evaluate relationships between responses in the toxicity tests and the physical or chemical characteristics of sediments (Tables 2, 3). Statistical significance for the rank correlations was established at 0.05.

Table 2 Spearman rank correlations among H. azteca 28 day toxicity end points and sediment characteristics and frequencies of observed toxicity for samples below and exceeding toxicity thresholds
Full size table
Table 3 Spearman rank correlations among C. dilutus 10 day toxicity end points and sediment characteristics as well as frequencies of observed toxicity for samples below and above toxicity thresholds
Full size table

Probable effect concentrations (PECs) were used to evaluate relationships between sediment contamination and sediment toxicity (MacDonald et al. 2000). A PEC is a contaminant concentration above which adverse effects on sediment-dwelling organisms are expected to occur more often than not. The PEC quotient (PECQ) for a contaminant is the measured concentration divided by the PEC. The mean PECQ for a sample is the average of PECQ values for all contaminants in that sample. A higher likelihood of toxicity is predicted for samples with PECQ values >1 for individual contaminants or mean PECQ values of >0.1 to >0.2 for sample mixtures (Ingersoll et al. 2001, 2005, 2009).

The current study deviates from the originally published mean PECQ procedure as follows:

1. The PEC values from MacDonald et al. (2000) were assumed to apply to 1 %TOC in sediment (reflecting sediment levels typical of the data set used to derive PECs; Ingersoll et al. 2009). Hence, in the current study, contaminant concentrations were normalized by the measured concentration of TOC in the samples (express as μg/g [or μg/kg] dry weight at 1 %TOC before comparison with PECs).

2. The mean PECQ was expanded to include additional contaminants. Dieldrin and chlordane were added because their PECs come close to meeting published reliability criteria (MacDonald et al. 2000), and these compounds were frequently detected in the current study. This approach is consistent that of with other studies (Tao et al. 2010) in which chlordane and dieldrin were identified as contaminants of concern in field-collected sediments.

3. For pyrethroids, 28 day sediment toxicity thresholds for H. azteca were estimated based on the results of published H. azteca 10 day lethality tests conducted with spiked sediments (Table 4) by (a) dividing the 10 day lethal concentrations for pyrethroids by a factor of 2.5 to estimate sublethal effects (Amweg et al. 2005) and (b) further dividing by a factor of 2.0 to estimate 28 day effect thresholds from 10 day effect thresholds (Ingersoll et al. 2001). Although derived using the results of 10 day spiked-sediment toxicity tests conducted with H. azteca, these pyrethroid toxicity thresholds have a similar intent as PECs (i.e., as estimates of concentrations above which toxicity is expected to sediment-dwelling organisms), so a pyrethroid toxicity quotient is analogous to a PECQ value.

Table 4 Estimated pyrethroid whole-sediment toxicity thresholds for H. azteca in 28 day exposures
Full size table

4. For all contaminants except pyrethroids, nondetections were assumed equal to half of the reporting level (detection limits ranged from 0.2 to 0.5 ng/g [dry weight; Table 4, Moran et al. 2012]), which is consistent with the original PECQ methodology (Ingersoll et al. 2001). For pyrethroids, nondetections were assumed equal to zero because pyrethroid reporting levels were relatively close to their toxicity thresholds (otherwise even samples with no detectable pyrethroids would have substantial pyrethroid toxicity quotient values; Moran et al. 2012).

5. The mean PECQ value equally weights three to five contaminant classes being studied (Ingersoll et al. 2001). For example, the PECQ-3 equally weights the contributions of: (a) trace elements (mean trace element PECQ of arsenic (As), cadmium (Cd), chromium (Cr), copper (Cu), lead (Pb), nickel (Ni), and zinc (Zn), (b) PAHs (PECQ for total PAHs), and (c) PCBs (PECQ for total PCBs). The PECQ-4 equally weights the contributions of the following: (a) mean trace element PECQ, (b) PECQ for total PAHs, (c) PECQ for total PCBs, and (d) OC pesticides (mean PECQ for DDE, dieldrin, and total chlordane). The PECQ-5P equally weights the contributions of the: (a) mean trace-element PECQ, (b) PECQ for total PAHs, (c) PECQ for total PCBs, (d) mean PECQ for OC pesticides, and (e) pyrethroids (the sum of the toxicity quotients for bifenthrin, cypermethrin, cyhalothrin, and permethrin). The toxicity quotients for pyrethroids were summed because the pyrethroids share a common mode of action, so toxicity is expected to be additive (Trimble et al. 2009). However, results were similar when pyrethroids were represented by the mean toxicity quotient of these four pyrethroids (PECQ-5M) or by the toxicity quotient for bifenthrin alone (PECQ-5B).

Relationships between toxicity and concentrations of PAHs in sediment samples were also evaluated using mechanistically based chronic sum equilibrium-partitioning sediment benchmark toxic units (ΣESBTUs). The ΣESBTU approach integrates benchmarks for total PAHs (34 parent and alkylated PAHs) into a toxic unit model that accounts for the joint toxicity of various PAHs with the same mode of toxicity (i.e., nonpolar narcosis; USEPA United States Environment Protection Agency 2003). The ΣESBTU for the 34 PAHs was estimated from 18 parent PAHs measured in the present study using methods described in USEPA (United States Environment Protection Agency 2003), including the application of a 50 % uncertainty factor (USEPA United States Environment Protection Agency 2003). The ΣESBTU approach was developed to account for the bioavailability of nonionic organic compounds in different sediments and incorporates select toxic effects concentrations in porewater (i.e., final chronic values [FCVs]). A ΣESBTU threshold of 1.0 was established based on an estimated critical body burden of approximately 2 μmol total PAHs/g lipid for 28 day weight of H. azteca in water-only exposures and an FCV of approximately 2.2 μmol total PAHs normalized to the fraction OC in sediment (Dave Mount, USEPA, Duluth, MN, personal communication). Using this approach, sediment samples with ΣESBTU values >1.0 were predicted to be toxic to H. azteca in 28 day sediment exposures to PAHs.

Results and Discussion

Sediment-Toxicity Tests

Mean survival and weight of amphipods or midge in the control sediment met test acceptability requirements (USEPA United States Environmental Protection Agency 2000; ASTM American Society for Testing and Materials International 2012) for all of the batches of sediment samples (Table S1). Fifty of 98 samples across the study areas met the requirement of a reference sediment (low chemistry and low toxicity; Table 1 and Table S3).

A total of 37 % of the 98 sediment samples from the 7 study areas were identified as toxic to amphipods or midge (i.e., survival, weight, or total biomass were below the lower limit of a reference envelope value for the study area; Table 1 and Tables S2 and S3). More sediment samples were identified as toxic to amphipods (33 %) compared with midge (12 %). Of the 36 samples that were toxic to either amphipods or midge, 67 % were toxic to only amphipods, 11 % were toxic to only midge, and 22 % were toxic to amphipods and midge, suggesting that concentrations of contaminants in the sediments were more frequently toxic to amphipods compared with midge.

Survival was the most sensitive end point in the amphipod test (32 % toxic), whereas total biomass was the most sensitive end point in the midge test (11 % toxic). Samples from Dallas had the highest toxicity to amphipods (69 % toxic) and Boston samples the highest toxicity to midge (36 % toxic), whereas samples from Denver (23 % toxic) and Seattle (24 % toxic) had the lowest toxicity to amphipods or midge.

Comparisons of Sediment Characteristics and Sediment Toxicity

Significant rank correlations were infrequently observed between the water-quality characteristics measured in the overlying water or in the porewater relative to the survival, weight, or total biomass of amphipods or midge (data not shown). However, a significant negative correlation was observed between concentrations of unionized ammonia in porewater measured at the start of the exposures and midge weight (Table 3). None of the concentrations of unionized ammonia in porewater exceeded a 10 day median lethal effect concentration (LC50) of 0.94 mg N/L established for C. dilutus in water-only exposures (Table 3). Hence, ammonia likely did not substantially contribute to the decreased weight of midge.

The grain size and TOC exhibited a broad range in sediment samples (e.g., sand-size particles ranged from 2 to 88 %, and TOC in the <2 mm size fraction ranged from 0.16 to 17 %; Moran et al. 2012). However, responses of amphipods or midge were not significantly correlated to grain size or to TOC measured in the <2 mm size fraction of sediment, and only a weak significant correlation (r = 0.293) was observed between amphipod biomass and TOC measured in the <63 μm fraction of sediment (Tables 2, 3). Therefore, the physical characteristics of the sediment samples likely did not substantially contribute to the responses of either test organism across the study areas. These results are consistent with reports by others that H. azteca and C. dilutus are relatively tolerant of wide ranges in these physical characteristics in sediment (USEPA United States Environmental Protection Agency 2000; ASTM American Society for Testing and Materials International 2012).

Concentrations of metal, PAHs, OC pesticides, and PCBs in sediment infrequently exceeded PECs (Tables 2 and 3; Moran et al. 2012). Concentrations of pyrethroids were detected in 44 of 98 sediment samples. Pyrethroids were detected most often in the Dallas study area followed by the Salt Lake City and Atlanta study areas. The Seattle and Denver study areas had the fewest samples with pyrethroids detected. Of the 14 pyrethroids analyzed for, only 5 were detected; of these, bifenthrin was most commonly detected pyrethroid across all 7 of the study areas (at 40 of 98 sites) and within each of the 7 study areas (Moran et al. 2012). Bifenthrin was detected most often in samples from the Dallas and Salt Lake City study areas. Of the 14 sediment samples exceeding the calculated threshold for bifenthrin (Table 4), 6 were from the Dallas study area, 4 were from Salt Lake City study area, and 1 each was from the Atlanta, Denver, Boston and Seattle study areas.

Relationships to toxicity end points were initially evaluated relative to sediment-chemistry data expressed as (1) dry weight concentrations, (2) TOC-normalized concentrations in the <2 mm size fraction of sediment, or (3) TOC-normalized concentrations in the <63 μm size fraction of sediment (Moran et al. 2012). Preliminary correlations between chemistry and toxicity end points were generally strongest when chemical concentrations were normalized to TOC measured in the appropriate sediment fraction (<2 mm fraction for organic contaminants and <63 μm fraction for trace elements) rather than on a dry-weight basis. Hence, these TOC-normalized values were the primary ones used in correlation analysis between sediment-chemistry and -toxicity end points (Tables 2 and 3), although the tables also include examples of correlations based on dry-weight concentrations.

Only a limited number of significant correlations were observed between toxicity end points for amphipods or midge and the concentrations of individual metals, individual or total PAHs, total PCBs, or mean PECQ for OC pesticides (normalized to 1 %OC or to dry weight), and these correlations were relatively weak. Moreover, concentrations of these contaminants infrequently exceeded individual PEC sediment-toxicity thresholds (Tables 2, 3). Weaker correlations were generally observed between sediment toxicity and sediment chemistry for midge compared with amphipods (Tables 2, 3).

Responses of amphipods or midge were not significantly correlated to concentrations of PAHs normalized to sediment OC using ΣESBTUs (Tables 2, 3, Fig. 1B). However, of the 20 samples that exceeded a ΣESBTU of 1.0, 50 % of them were toxic to amphipods, suggesting that PAH concentrations in some sediment samples may have contributed to the toxicity to amphipods based on exceedances of the ΣESBTU toxicity threshold of 1.0.

Responses of amphipods, but not midge, were weakly correlated to the mean PECQ-3 (calculated by averaging the quotients for mean trace elements, total PAHs, and total PCBs) and were weakly correlated to the mean PECQ-4 (which also included the mean PECQ for OC pesticides), and only 18–19 % of the samples exceeded a PECQ-3 or PECQ-4 toxicity threshold of 0.1 (Tables 2, 3). Of the samples that exceeded a PECQ-3 or PECQ-4 toxicity threshold of 0.1, a higher incidence of toxicity was observed to amphipods (50–53 % toxicity) compared with midge (16–22 % toxicity; Tables 2, 3; Figs. 1D, F). However, the frequency of toxicity below PECQ-3 or PECQ-4 thresholds (28–29 % for amphipods and 10–11 % for midge) was similar to the frequency of toxicity in samples above these thresholds. Hence, while PECQ-3 and PECQ-4 were correlated to toxicity, and component contaminants may have contributed to the toxicity, concentrations of these contaminants or groups of contaminants were only infrequently high enough to have solely caused the toxicity to amphipods or midge observed across the study areas.

Concentrations of bifenthrin, permethrin, and ΣPyrethroid toxic quotients were significantly negatively correlated to survival, weight, or total biomass of amphipods (Table 2, Fig. 1C) but not to midge (Table 3). A total of 14 % of the sediment samples exceeded the OC-normalized sediment toxicity threshold for bifenthrin; however, only one sediment sample exceeded the corresponding threshold for permethrin (Tables 2 through 4). Of the samples exceeding the OC-normalized bifenthrin toxicity threshold, 79 % were toxic to amphipods; however, only 25 % of the samples below this bifenthrin toxicity threshold were toxic to amphipods (Table 2; Fig. 1C). Similarly, of the samples exceeding the dry-weight bifenthrin toxicity threshold, 75 % were toxic to amphipods, whereas only 15 % of the samples below this bifenthrin toxicity threshold were toxic to amphipods (Table 2). Overall, correct classification of sediment samples as not toxic below, or as toxic above, the bifenthrin threshold was 76 % based on concentrations normalized to TOC or 82 % based on dry-weight concentrations (Table 2). Sediments from the Dallas study area exhibited the highest incidence of toxicity to amphipods (69 %; Table 1) and had some of the highest sediment concentrations of bifenthrin (Fig. 1C). Of the samples exceeding a ΣPyrethroid toxicity quotient threshold of 1.0, 80 % were toxic to amphipods compared with 24 % toxicity for samples below this toxicity threshold, and the overall correct classification of samples as not toxic to amphipods below, or as toxic above, the threshold was 77 % based on the ΣPyrethroid toxicity quotient (Table 2).

Including bifenthrin as a fifth contaminant class in the calculation of the mean PECQ-5B (or ΣPyrethroids as the fifth class in the mean PECQ-5P or mean pyrethroid toxicity quotients in the mean PECQ-5M) resulted in stronger negative correlations to responses of amphipods compared with the mean PECQ-3 or PECQ-4 (Table 2; Fig. 1E). However, correlations were not as strong for the mean PECQ-5M compared with the PECQ-5B or PECQ-5P. A total of 36–37 % of the samples exceeded a mean PECQ-5B or PECQ-5P toxicity threshold of 0.1. Above this threshold, 67–68 % of the samples were toxic to amphipods, whereas only 11–13 % of the samples below this threshold were toxic to amphipods (Table 2; Fig. 1C). In addition, more samples exceeded the mean PECQ-5B or mean PECQ-5P threshold of 0.1 (36–37 %) compared with the thresholds based solely on bifenthrin. Only 14 % of the samples exceeded the OC-normalized bifenthrin toxicity threshold, and only 29 % of the samples exceeded the dry-weight bifenthrin toxicity threshold (Table 3). Hence, including combined measures of trace elements, total PAHs, total PCBs, OC pesticides, and pyrethroid pesticides (bifenthrin in particular) in the calculation of the mean PECQs improved the correct classification of samples as toxic or not toxic to amphipods (76–83 %) compared with the measures of individual classes of contaminants (66–69 %) or to mean PECQ-3 (68 %) or mean PECQ-4 (67 %; Table 3).

The toxicity of samples to midge was poorly predicted by the sediment-toxicity thresholds based on the trace metals and organic contaminants measured, including pyrethroids. The low incidence of toxicity to midge across the study areas (12 %, Table 1) may have resulted from other unmeasured contaminants in the sediment samples or from random chance. A minimum incidence of toxicity of approximately 20 % was observed in midge sediment-toxicity tests even at low sediment chemistry (Ingersoll et al. 2001). The finding that sediments were frequently more toxic to amphipods compared with midge is consistent with the pyrethroids being a primary cause of the toxicity. Spiked-sediment toxicity studies indicate that the 10 day weight of H. azteca was decreased at lower concentrations of bifenthrin (by a factor of 5–10) or permethrin (factor of 2) compared with concentrations that decreased the 10 day weight of C. dilutus (Amweg et al. 2005; Maul et al. 2008).

Weston et al. (2005), Amweg et al. (2006), and Hintzen et al. (2009) reported field-collected sediment samples to be lethal to H. azteca in 10 day laboratory exposures at sum pyrethroid toxic units (TUs) >1.0. The pyrethroid TUs in these previous studies were based on 10 day pyrethroid toxicity thresholds derived from spiked-sediment toxicity tests (Amweg et al. 2005). In contrast, lethality to H. azteca in 10 day exposures to field-collected sediments was observed at 10 day pyrethroid TUs greater than approximately 0.3 by Holmes et al. (2008), Ding et al. (2010), and Domagalski et al. (2010) In these previous studies, bifenthrin was a primary pyrethroid contributing to the 10 day H. azteca pyrethroid TUs in these field-collected sediment samples. Similarly, decreased populations of H. azteca in the field have been reported at 10 day pyrethroid TUs >0.6 (Weston et al. 2005).

In the present study, the estimated 28 day H. azteca ΣPyrethroid toxicity quotient threshold of 1.0 (Tables 2, 3) is numerically equivalent to a 10 day pyrethroid TU of 0.2 because the 28 day pyrethroid toxicity thresholds in the current study were derived by dividing 10 day H. azteca lethality data from spiked sediment-toxicity tests by a factor of 5 to account for longer-term exposure and sublethal effects (Table 4). Hence, the toxicity associated with pyrethroids to H. azteca in 28 day exposures based on survival or growth at 10 day pyrethroid TUs >0.2 (numerically equivalent to the 28 day pyrethroid TU of 1.0 in Table 2) were similar to the toxicity reported at 10 day pyrethroid TUs >0.3 for H. azteca in 10 day exposures based on survival (Holmes et al. 2008; Ding et al. 2010; Domagalski et al. 2010). However, these previous studies did not as extensively characterize concentrations of other contaminants of interest in the sediments (e.g., metals, PAHs, PCBs, other pesticides). Hence, other contaminants may have contributed to the toxicity observed in these previous 10 day lethality sediment-toxicity tests with H. azteca, which would decrease the apparent pyrethroid toxic threshold from exposure to pyrethroids alone.

Additional studies are needed to better establish chronic toxicity thresholds for pyrethroids in sediment based on longer-term exposures and including sublethal end points. The current study estimated 28 day pyrethroid toxicity thresholds for H. azteca on the basis of H. azteca 10 day lethality tests conducted with spiked sediments (Table 4). Given the uncertainties associated with these estimated pyrethroid toxicity thresholds, studies should be conducted using 28–42 day spiked sediment toxicity tests with H. azteca measuring effects on survival, weight, biomass, or reproduction to improve these estimates of chronic toxicity thresholds for pyrethroids in sediment (USEPA United States Environmental Protection Agency 2000; ASTM American Society for Testing and Materials International 2012). Moreover, there is some uncertainty from deriving a toxicity threshold using a combination of empirically based PECs (for contaminants other than pyrethroids) and pyrethroid toxicity thresholds estimated from mechanistically based spiked-sediment toxicity tests (10 day lethality tests with H. azteca; Table 4). Hence, it would be useful to develop empirically derived PECs for pyrethroids using matching chemistry and toxicity data for field-collected samples according to procedures outlined in MacDonald et al. (2000). Future studies should also include toxicity-identification evaluations and should further evaluate factors controlling the bioavailability of pyrethroids in sediment to better determine the contribution of pyrethroid to the overall toxicity observed in field-collected samples (e.g., Weston et al. 2008, You et al. 2008; Phillips et al. 2010).