Acute sensitivity of a broad range of freshwater mussels to chemicals with different modes of toxic action

Abstract

Freshwater mussels, one of the most imperiled groups of animals in the world, are generally underrepresented in toxicity databases used for the development of ambient water quality criteria and other environmental guidance values. Acute 96-h toxicity tests were conducted to evaluate the sensitivity of 5 species of juvenile mussels from 2 families and 4 tribes to 10 chemicals (ammonia, metals, major ions, and organic compounds) and to screen 10 additional chemicals (mainly organic compounds) with a commonly tested mussel species, fatmucket (Lampsilis siliquoidea). In the multi-species study, median effect concentrations (EC50s) among the 5 species differed by a factor of ≤2 for chloride, potassium, sulfate, and zinc; a factor of ≤5 for ammonia, chromium, copper, and nickel; and factors of 6 and 12 for metolachlor and alachlor, respectively, indicating that mussels representing different families or tribes had similar sensitivity to most of the tested chemicals, regardless of modes of action. There was a strong linear relationship between EC50s for fatmucket and the other 4 mussel species across the 10 chemicals (r² = 0.97, slope close to 1.0), indicating that fatmucket was similar to other mussel species; thus, this commonly tested species can be a good surrogate for protecting other mussels in acute exposures. The sensitivity of juvenile fatmucket among different populations or cultured from larvae of wild adults and captive-cultured adults was also similar in acute exposures to copper or chloride, indicating captive-cultured adult mussels can reliably be used to reproduce juveniles for toxicity testing. In compiled databases for all freshwater species, 1 or more mussel species were among the 4 most sensitive species for alachlor, ammonia, chloride, potassium, sulfate, copper, nickel, and zinc; therefore, the development of water quality criteria and other environmental guidance values for these chemicals should reflect the sensitivity of mussels. In contrast, the EC50s of fatmucket tested in the single-species study were in the high percentiles (>75th) of species sensitivity distributions for 6 of 7 organic chemicals, indicating mussels might be relatively insensitive to organic chemicals in acute exposures.

INTRODUCTION

Freshwater mussels (order Unionoida) are one of the most imperiled groups of animals, and environmental contamination has been identified as a causal or contributing factor to the decline of mussel populations ^1–4. Previous studies indicate that freshwater mussels are more sensitive than commonly tested organisms to some chemicals (e.g., copper, ammonia, and chloride ^5–11). However, mussels are generally under-represented in toxicity databases used for the development of water quality criteria, standards, and other environmental guidance values ^{6, 7, 12, 13}. In addition, the limited mussel data available have been generated mostly from toxicity tests with a few species of the taxonomic tribe Lampsilini ^5–16. Studies with broader phylogenetic representation are needed to understand the range of sensitivity among freshwater mussels, compare the overall sensitivity of mussels to other freshwater species, and evaluate the degree to which existing or proposed environmental guidance values are protective of mussels.

Freshwater mussels are a taxonomically diverse group of bivalve mollusks with a complex reproductive cycle. Approximately 300 species in the order Unionoida historically occur in North America. These species are classified in 2 families, with 5 in the family Margaritiferidae and the rest in the family Unionidae. Unionidae include 2 subfamilies, the Unioninae and the Ambleminae. Subfamily Unioninae includes the tribe Anodontini, and subfamily Ambleminae includes tribes Lampsilini, Amblemini, Pleurobemini, Quadrulini, and Gonideini ¹⁷. Ambleminae includes 250 North American species and 37 genera, which represent 85% of North American species and 75% of North American genera of Unionoida ^{17, 18}. Most freshwater mussels have a reproductive cycle involving a parasitic stage on fish. Sperm released by a male enters a female through the incurrent siphon, and fertilized eggs develop to larvae called glochidia, which mature in specialized chambers (marsupia) of the mussel gills. Glochidia are released into the water and must attach to the gills or fins of a suitable host fish. After 1 wk to several weeks of the parasitic stage, glochidia transform to juvenile mussels, detach from the fish, and drop to the stream or lake bottom to begin the free-living juvenile stage.

The primary objectives of the present study were, first, to compare the sensitivity of 5 phylogenetically diverse species of mussels in acute exposures to 10 chemicals with different modes of toxic action (multi-species study) and, second, to screen 10 additional chemicals with a commonly tested mussel species (single-species study). For the multi-species study, the family Margaritiferidae was represented by western pearlshell (Margaritifera falcata), which is native to the western United States, and the family Unionidae was represented by threeridge (Amblema plicata; Amblemini), paper pondshell (Utterbackia imbecillis; Anodontini), fatmucket (Lampsilis siliquoidea; Lampsilini), and washboard (Megalonaias nervosa; Quadrulini), which are distributed widely in the Midwest and the Southeast of the United States. These mussels include 2 life-history strategies 4: 2 short-term brooders and 3 long-term brooders (Table 1). Fatmucket was selected for the single-species study with 10 additional chemicals because it has been used widely in previous toxicity tests in the United States and Canada ^{5–7, 11, 13, 14}. In addition, the sensitivity of the commonly tested fatmucket among different populations was evaluated in acute exposures to 2 commonly used reference toxicants (copper and sodium chloride).

Table 1.

Gravid female mussels and host fish for juvenile mussel culture

Family (Tribe)	Species	Reproduction	Collection location	No. of mussels for a batch culture	Host fish
Unionidae (Amblemini)	Threeridge (Amblema plicata)	Short-term brooder	Mississippi River, Prairie du Chien, WI, USA	4 or 8	Largemouth bass (Micropterus salmoides)
Unionidae (Anodontini)	Paper pondshell (Utterbackia imbecillis)	Long-term brooder	Kansas City Zoo, Kansas City, MO, USA	3	Bluegill (Lepomis macrochirus)
Unionidae (Lampsilini)	Fatmucket (Lampsilis siliquoidea)	Long-term brooder	Perche Creek, Boone County, MO, USAa	3 or 5	Largemouth bass (Micropterus salmoides)
Unionidae (Quadrulini)	Washboard (Megalonaias nervosa)	Long-term brooder	Mississippi River, Crawford County, WI, USA	15	Channel catfish (Ictalurus punctatus)
Margaritiferidae	Western pearlshell (Margaritifera falcata)	Short-term brooder	South Fork Eel River, Mendocino County, CA, USA	7 or 8	Rainbow trout (Oncorhynchus mykiss)

^aTwo other populations of fatmucket were collected from Bourbeuse River (Gasconade County, MO, USA) and Kansas City Zoo (Kansas City, MO, USA) for reference toxicant (sodium chloride and copper) tests to determine the sensitivity among three populations (details in text).

The 10 chemicals selected for the multi-species study were 2 organic compounds used as herbicides (alachlor and metolachlor), ammonia, 3 major ions (potassium, chloride, and sulfate), and 4 metals (chromium [VI], copper, nickel, and zinc; Table 2). These chemicals were selected based on interest of the US Environmental Protection Agency (USEPA) in developing or updating ambient water quality criteria, the availability of toxicity data for non-mollusk species (e.g., at least 10 species) in the USEPA Interspecies Correlation Estimation (ICE) database ^{19, 20}, the high sensitivity of mollusks that are not mussels, different chemical classes and modes of toxicity action ²¹, and environmental relevance. The 10 chemicals for the single-species study (Table 2) were selected based on the same approach for chemical selection in the multi-species study except that no or few toxicity data from mussels were available for these chemicals. The chemicals used in the single-species study included 7 organic compounds used as pesticides (2,4-dichlorophenoxyacetic acid, 4-nonylphenol, azoxystrobin, bifenthrin, carbaryl, malathion, and molinate), arsenic (V), calcium chloride, and aluminum.

Table 2.

Chemicals and modes of toxic action in two studies^a

Toxicant	Chemical formula	Purity (%)	Broad MOAb	Specific MOAb
Multi-species study
Alachlor	C14H20ClNO2	99.2	Narcosis	Nonpolar
Metolachlor	C15H22ClNO2	97.6	Narcosis	Nonpolar
Ammonia	NH4Cl	99.5	Lono/Osmoregulatory/Circulatory impairment	Other osmoregulatory impairment
Potassium	KCl	99.7	Lono/Osmoregulatory/Circulatory impairment	Other osmoregulatory impairment
Chloride	NaCl	100	Lono/Osmoregulatory/Circulatory impairment	Other osmoregulatory impairment
Sulfate	Na2SO4	100	Lono/Osmoregulatory/Circulatory impairment	Other osmoregulatory impairment
Chromium (VI)	CrO3	98	Lono/Osmoregulatory/Circulatory impairment	Metallic iono/osmoregulatory impairment
Copper	CuSO4 · 5H2O	99	Lono/Osmoregulatory/Circulatory impairment	Metallic iono/osmoregulatory impairment
Nickel	NiCl2	100	Lono/Osmoregulatory/Circulatory impairment	Metallic iono/osmoregulatory impairment
Zinc	ZnCl2	98	Lono/Osmoregulatory/Circulatory impairment	Metallic iono/osmoregulatory impairment
Single-species study
2,4-dichlorophenoxyacetic acid	C8H6Cl2O3	97	Narcosis	Nonpolar
4-nonylphenol	C15H24O	100	Narcosis	Polar
Azoxystrobin	C22H17N3O5	100	Narcosis	Ester
Bifenthrin	C23H22ClF3O2	100	Neurotoxicity	Pyrethroid sodium channel modulation
Carbaryl	C12H11NO2	97	Acetylcholinesterase inhibition	Carbamate
Malathion	C10H19O6PS2	100	Acetylcholinesterase inhibition	Organophosphate
Molinate	C9H17NOS	100	Narcosis	Nonpolar
Arsenic (V)	Na2HAsO4 · 7H2O	98	Electron transport inhibition	Arsenical respiratory inhibition
Calcium chloride	CaCl2	98	Uncertain	Uncertain
Aluminum	AlCl3	99	Lono/Osmoregulatory/Circulatory impairment	Metallic iono/osmoregulatory impairment

^aAll chemicals were obtained from Sigma-Aldrich, except ammonia chloride and potassium chloride (obtained from Fisher Scientific).

^bMode of action (MOA) determined from Barron et al. 21.

Two air-breathing freshwater snail species (Physa gyrina and Lymnaea stagnalis) and 3 commonly tested invertebrates (amphipod Hyalella azteca, and cladocerans Ceriodaphnia dubia and Daphnia magna) were also tested in acute exposures with the 10 chemicals used in the multi-species study following ASTM International standard methods ²². The methods and results for the acute exposures with the snails and other invertebrates are reported in a companion paper Ivey et al. ²³ in this issue. However, the toxicity data from these non-mussel species were included in toxicity databases compiled in the present study to compare the mussel sensitivity with that of other freshwater species.

MATERIALS AND METHODS

Acute 96-h toxicity tests with newly transformed (6 d to 10 d old) juvenile mussels were conducted following ASTM International standard methods ²⁴. Conditions for conducting the toxicity tests are summarized in Supplemental Data, Table S1.

Test organisms

Gravid female mussels brooding mature glochidia (larval mussels) were collected during early season of glochidia release between 2012 and 2015 from Missouri, USA (fatmucket, paper pondshell); Wisconsin, USA (threeridge, washboard); and California, USA (western pearlshell; Table 1). Mussels at all collection sites were abundant and apparently healthy, and reproduction and recent recruitment were evident. For toxicity tests to evaluate the sensitivity among different populations of fatmucket, adult mussels were obtained from 2 wild populations and 1 captive population in May 2014: a population from Silver Fork of Perche Creek, Boone County, Missouri (this population has been used routinely in previous toxicity studies ^6–10); a population from the Bourbeuse River, Gasconade County, Missouri; and a captive population at the Kansas City Zoo (Kansas City, MO, USA). Mussels in the captive population were 4 yr old and were derived from glochidia obtained from 6 females collected from the Silver Fork of Perche Creek.

The collected female paper pondshell, fatmucket, and western pearlshell were transferred to culture facilities at Missouri State University (Springfield, MO, USA), and female threeridge and washboard were transferred to Genoa National Fish Hatchery (Genoa, WI, USA). The adult mussels were held at 10 °C to 12 °C and fed commercially available cultured algae (Nannochloropsis and Shellfish Diet; Reed Mariculture). The culture water used at Missouri State University was filtered (10 μm) river water collected from James River, Greene County, Missouri, with a hardness 160 mg/L as CaCO₃ and pH 8.1; the culture water used in the Genoa National Fish Hatchery was well water with a hardness 200 mg/L as CaCO₃ and pH 7.5. To culture juvenile mussels, roughly equal numbers of glochidia were removed from each of at least 3 adult mussels (Table 1). The viability of glochidia isolated from each adult mussel was tested with a subsample using the closing response to sodium chloride 24. The viability of glochidia from all samples exceeded 90%. The glochidia isolated from the adult mussels were pooled and placed on hatchery-reared fish for metamorphosis. The number of adult mussels used for glochidia sampling and the viability of glochidia met the recommendations in ASTM International methods ²⁴. The host fish (Table 1) were maintained at 20 °C to 22 °C (except for western pearlshell at 13 °C to 15 °C) in a recirculating system equipped for recovery of juveniles. Juvenile mussels were recovered 2 wk (5 wk for western pearlshell) after fish infestation. Juvenile mussels recovered from the host fish during the 2-d peak of recovery were shipped overnight to the US Geological Survey, Columbia Environmental Research Center (Columbia, MO) for toxicity testing.

The newly transformed juvenile mussels (generally <5 d old) were acclimated to test water and temperature in 1-L glass beakers with gentle aeration through a glass pipette or in a recirculating bucket ²⁵ for at least 2 d before the start of a test. Approximately 25% of the water in the containers was gradually replaced with test water twice daily. The juvenile mussels were fed an algal mixture (Nannochloropsis concentrate and Shellfish Diet; Reed Mariculture 7) during the acclimation period twice daily in the morning and afternoon, with an algal density of 5 nL cell volume/mL to 10 nL cell volume/mL after each feeding.

Test water and exposure concentrations

Test water (control water) was prepared by diluting Columbia Environmental Research Center well water of hardness 300 mg/L as CaCO₃ with deionized water to a hardness of 100 mg/L as CaCO₃. The formula, purity, source, and modes of toxic action of test chemicals are listed in Table 2. For each chemical, 5 concentrations plus a control were tested. Test concentrations were chosen based on results of previous acute toxicity tests with juvenile mussels. When a prediction of toxicity for a particular chemical was not available for mussels, an initial 96-h range-finding test was first conducted with a limited number of mussels (5 to 10) and replicates (1 or 2) in a control and 5 concentrations of the test chemical that differed by a dilution factor of up to 10.

Toxicity tests were conducted under static-renewal conditions, with the exception that ammonia toxicity tests were conducted under flow-through conditions to maintain constant concentrations over 96-h exposures 8. For static-renewal toxicity tests, a solution of the highest exposure concentration was prepared 4 h to 24 h before the start of a test by adding either a certain amount of chemical (e.g., NaCl, KCl, and Na₂SO₄) or a certain volume of stock solution (metals and organic chemicals) into 1000 mL or 2000 mL of control water in a glass jar. One-half of the solution was then used for 50% dilutions to create other solutions of lower concentrations. The control water and solutions were held in the dark at 4 °C and warmed to test temperature in water baths for use at the beginning of a test and for water renewal. Triethylene glycol (Sigma-Aldrich) was used as solvent to prepare stock solutions of organic chemicals, with the exceptions of carbaryl and bifenthrin, for which acetone (Sigma-Aldrich) was used. The solvent concentration in test solutions was kept to a minimum and did not exceed 0.5 mL/L ²⁴. A dilution-water control and a solvent control containing the highest concentration of the solvent in test solutions were included in tests with organic chemicals. Ammonia toxicity tests were conducted in an intermittent flow-through proportional diluter system ⁸. The diluter delivered 5 ammonia concentrations with a 50% dilution series plus a control. Ammonium chloride (NH₄Cl; Table 2) was used to prepare an ammonia stock solution in a 2000-mL volumetric flask. The pH in the stock solution was adjusted to a pH of 8.0 (close to the pH in the control water) by adding <1 mL of a solution of ammonium hydroxide (NH₄OH; 30% purity; Fisher Scientific). The stock solution was delivered with each cycle of the diluter system by Hamilton syringe pump (Hamilton).

Toxicity testing

At the beginning of each static-renewal test, 5 juvenile mussels were impartially transferred into each of 4 separate 50-mL replicate glass beakers containing 30 mL of water. Test beakers were held in a plastic container (30 cm × 18 cm × 10 cm) with a cover to reduce evaporation. The containers were held in a water bath at 23 ± 1 °C (or 20 ± 1 °C for western pearlshell, a western US resident species that inhabits cooler water). Water temperature was monitored daily. Ambient laboratory light was 400 lux to 500 lux with 16:8-h light:dark photoperiod. Test organisms were not fed during 96-h exposures. Approximately 75% of the water in each replicate beaker was removed and renewed after 48 h. At the end of exposures, mussels in each beaker were examined under a dissecting microscope. The endpoint was mortality (empty shell with little or no tissue) plus immobility (no foot or shell movement within a 5-min observation period). The test acceptability criterion was ≥90% control survival ²⁴.

For the flow-through ammonia toxicity tests conducted in the diluter, 10 mussels were impartially transferred into each of 4 replicate exposure units at the beginning of exposures. The exposure unit consisted of an inner chamber and an outer beaker 10. The inner chamber was a 160-mL glass tube with stainless-steel screen (254-μm opening) at the bottom, used to facilitate retrieval of small juvenile mussels at the end of the acute exposure. Two glass rods were attached to the top of the inner chamber to keep the chamber suspended in a 300-mL glass outer beaker. The outer beaker had a 2.5-cm hole in the side covered with a stainless-steel screen (279-μm opening) and contained 200 mL of water. The diluter system delivered 125 mL of test solution into each inner chamber once every 4 h. The excess water flowed out through the side screen of the outer beaker. At the end of exposures, mussels in each inner chamber were rinsed into a glass tray with control water, and recovered and transferred into a 50-mL glass beaker containing 20 mL of the test solution; survival was determined under a dissecting microscope. Other test conditions and test acceptability criteria were the same as those described previously for the static-renewal tests.

Water quality and chemical analysis

Dissolved oxygen, pH, conductivity, hardness, and alkalinity were measured using standard methods ²⁶ on composite water samples collected from the replicates in the control, medium, and high exposure concentrations at the beginning and the end of each test. Water samples for major cations (calcium, potassium, magnesium, and sodium), major anions (chloride and sulfate), and dissolved organic carbon (DOC) were periodically collected from control water. Major cations were analyzed by inductively coupled plasma–mass spectrometry (ELAN DRC-e; PerkinElmer) according to USEPA method 6020B ²⁷, and major anions were analyzed by ion chromatography (ICS-1100; Dionex) according to USEPA method 9056A ²⁸. Samples for the analysis of DOC were filtered through 0.45 μm polyethersulfone membranes and measured by a total organic carbon analyzer (Model TOC-L CSH; Shimadzu Scientific Instruments) according to USEPA method 415.3 ²⁹.

Water samples for the analyses of toxicants were collected in all exposure concentrations generally at the beginning of metal exposures, and at the beginning and the end of exposures to other chemicals for which the stability of concentration over 96-h exposure was uncertain (Supplemental Data, Tables S4 and S5). Water samples for metal analysis were drawn from mid-depth of each exposure chamber with an all-polypropylene syringe fitted with a tetra-fluoroethylene sipper straw. The sample was then dispensed through a 0.45-μm pore size polyethesulfone membrane filter into an acid-cleaned polyethylene bottle, except the sample for analysis of total aluminum, which was not filtered. Each 20-mL sample was stabilized within 24 h by adding 0.2 mL of concentrated nitric acid. Concentrations of the metals were determined by inductively coupled plasma–mass spectrometry according to USEPA method 6020B 27. Chloride and sulfate were analyzed by the ion chromatography described previously or by a Hach HQ440d benchtop dual input, multi-parameter meter (Hach). In 2 chloride toxicity tests conducted earlier with threeridge and washboard, chloride concentrations were not measured. Rather, salinity and conductivity were measured in all exposure concentrations at the beginning and the end of each test to confirm the chloride concentrations (Supplemental Data, Table S4–4). Total ammonia nitrogen (N) was determined by a Hach HQ440d benchtop meter with a PHC301 Re-fillable pH Electrode. The meter was calibrated before measuring samples with 1.0 mg N/L and 10 mg N/L calibration standards. The method detection limit was 0.03 mg N/L.

Water samples for analysis of organic chemicals were extracted using Oasis HLB solid phase extraction cartridges (60 mg, 3 cc; Waters Corp). The cartridges were conditioned using ethyl acetate and methanol followed by deionized water prior to the drop-wise addition of the 10-mL water samples. The cartridges were then dried under nitrogen or by pulling laboratory air through for 5 min, placed in plastic bags and stored at −20 °C until extraction. Chemicals were eluted from the cartridges using 80:20 (v/v) dichloromethane:methyl-tert-butyl ether (alachlor, carbaryl, malathion, and metolachlor), methanol (2,4-dichlorophenoxyacetic acid and 4-nonylphenol), and ethyl acetate (azoxystrobin, bifenthrin, and molinate) prior to analysis. Analyses for alachlor, carbaryl, malathion, and metolachlor were conducted using an Agilent 6890 gas chromatograph with a 5973N mass selective detector. Analyses for 2,4-dichlorophenoxyacetic acid and 4-nonylphenol were conducted using liquid chromatography with photodiode array detection (Surveyor system; Thermo-Finnigan). Analyses for azoxystrobin, bifenthrin, and molinate were conducted using an Agilent 7890A gas chromatograph with a 5975C mass selective detector. Additional details of the extraction and analysis methods can be found in Alvarez et al. ³⁰ and Hladik et al. ³¹.

Analyses of tested chemicals were performed by chemistry laboratories at the Columbia Environmental Research Center and US Geological Survey (Sacramento, CA), following internal standard operating procedures and quality assurance/quality control protocols. Established laboratory quality assurance/quality control procedures and sample types (second source calibration verification, laboratory spikes, duplicates, reference/laboratory control materials) were used to verify instrument performance, accuracy, and precision throughout the analyses. These established procedures were in place to ensure method performance and instrumental suitability. Results from each laboratory underwent data quality review prior to use in the present study.

Data analysis

Median effect concentrations (EC50s) based on mortality plus immobility of juvenile mussels were estimated with the Toxicity Response Relationship Analysis Program (TRAP; Ver 1.30a ³²), using the tolerance distribution analysis with the Gaussian (normal) distribution model and with log-transformed exposure concentrations. When the data did not meet the requirements of the TRAP (least 2 partial responses), either a Spearman-Karber or a trimmed Spearman-Karber method was used following the flowchart for EC50 determination recommended by the USEPA ³³ using TOXSTAT® software (Ver 3.5; Western EcoSystems Technology).

To compare the relative sensitivity of mussels with other freshwater species, toxicity databases were compiled by adding the mussel toxicity data from the present study to additional databases, including the USEPA ICE database ^{19, 20} for alachlor, metolachlor, chloride (as NaCl), potassium (as KCl), calcium chloride, 2,4-dichlorophenoxyacetic acid, 4-nonylpgenol, azoxystrobin, bifenthrin, malathion, molinate, chromium (VI), and nickel; the databases used to derive USEPA water quality criteria for arsenic (V) ³⁴, ammonia ³⁵, copper ³⁶, and carbaryl ³⁷; the database used to update water quality criteria for aluminum (D. Eignor, USEPA, Washington, DC, unpublished data); a zinc database ³⁸; a sulfate database representing a hardness range of 80 mg/L to 120 mg/L 16; and the toxicity data from the companion study with the 2 snails and 3 other invertebrates (Ivey et al. ²³ in this issue) and unpublished toxicity data from acute toxicity tests conducted at the Columbia Environmental Research Center that met ASTM International test acceptability criteria, such as >90% control survival. Toxicity data for nickel and zinc were normalized to a hardness of 50 mg/L as CaCO₃ using the equations in the USEPA water quality criteria for nickel or zinc ³⁴, and copper data were normalized using a biotic ligand model ³⁶. Ammonia data were normalized to total ammonia-nitrogen at pH 7 and 20°C ³⁵. Aluminum data in a pH range of 6.5 to 9.0 were normalized to a hardness of 100 mg/L as CaCO₃ (D. Eignor, USEPA, Washington, DC, draft update water quality criteria for aluminum). Because sodium chloride toxicity typically decreases with increasing water hardness ^{11, 39, 40}, the chloride data were compiled in a hardness range of 80 mg/L to 180 mg/L.

RESULTS AND DISCUSSION

Water quality

Water quality characteristics are summarized in Supplemental Data, Tables S2 and S3. The concentrations of dissolved oxygen ranged from 7 mg/L to 9 mg/L. Mean measured water quality characteristics were similar to nominal values of the diluted well water with hardness of 100 mg/L as CaCO₃, alkalinity of 90 mg/L as CaCO₃, conductivity of 250 μS/cm, and pH of 8.2 (Supplemental Data, Tables S2 and S3), with a few exceptions. Higher alkalinity of 131 mg/L to 168 mg/L was observed in high exposure concentrations in the toxicity tests with 2,4-dichlorophenoxyacetic acid, arsenic (V), and aluminum, probably because of the addition of a large amount of the tested chemical or the counterion of chemicals (i.e., up to 800 mg/L 2,4-dichlorophenoxyacetic acid, 1020 mg/L sodium arsenate dibasic heptahydrate, or 360 mg/L aluminum chloride). In addition, as expected, the conductivity increased with increasing exposure concentrations of ammonium chloride, potassium chloride, sodium chloride, and sodium sulfate. Measured conductivity values at different exposure concentrations for these chemicals are presented in Supplemental Data, Tables S4–3 to S4–6. Mean major ions and DOC measured periodically in the control water were 26 mg Ca/L, 9.0 mg mg/L, 9.0 mg Na/L, 0.9 mg K/L, 11 mg Cl/L, 20 mg SO₄/L, and 0.5 mg/L DOC, which were similar to those reported in previous studies with the diluted well water ^{16, 41}.

Chemical analysis

Mean measured concentrations of inorganic chemicals typically differed less than 20% from nominal concentrations, whereas measured concentrations of most organic chemicals were 20% to 50% lower than nominal concentrations (Supplemental Data, Tables S4 and S5). Measured concentrations of organic chemicals were generally consistent at the beginning and end of a test, except that the concentrations of carbaryl and malathion decreased substantially at the end of testing (Supplemental Data, Table S5). The EC50s were calculated based on measured concentrations or on nominal concentrations adjusted by measured concentrations in some of organic chemical tests where not all concentrations were measured (Supplemental Data, Tables S4 and S5). Nominal concentrations were used for the EC50 calculation in 2 chloride tests with threeridge and washboard where chloride concentrations were not measured. Mean measured salinity and conductivity values were relatively consistent with the nominal NaCl concentrations (Supplemental Data, Table S4–4), confirming that the chloride concentrations were close to the nominal concentrations. Nominal concentrations were also used to calculate EC50s in 2 organic chemical tests (threeridge–alachlor test and fatmucket–carbaryl test) because the measured concentrations showed high, unexplained variability (1 measured concentration was similar to the nominal but another measured concentration was much different from the nominal; Supplemental Data, Tables S4–1 and S5) and the nominal concentrations represented our best estimate of actual exposure concentrations. Detailed information on the concentrations used for EC50 calculation is provided in Supplemental Data, Tables S4 and S5.

Toxicity tests with the 5 mussel species and 10 chemicals in the multi-species study

Toxicity data

In the multi-species study, 60 acute toxicity tests were completed with 5 mussels and 10 chemicals (Table 3). A test was repeated (if a new batch of mussels was available) when an EC50 could not be estimated because more than 50% mortality was observed in all exposure concentrations (fatmucket–potassium test and western pearlshell–potassium test; Table 3) or because there were no partial kills (fatmucket–nickel test). The EC50s for chloride and copper from the tests with fatmucket from different populations are also reported in Table 3 (test number 1 for the Silver Fork population, test number 2 for the Bourbeuse River population, and test number 3 for the Kansas City Zoo population). Ammonia toxicity was not tested with fatmucket in the present study because fatmucket has been intensively tested at the Columbia Environmental Research Center in previous ammonia studies ^{6, 8, 10}. Two ammonia EC50 values for fatmucket, obtained previously with similar test conditions (e.g., flow-through testing) and with the same life stage (newly transformed mussels) are included in Table 3 for the purpose of comparison among the 5 species.

Table 3.

Acute median effect concentrations (EC50s) and 95% confidence limits (CLs; in parenthesis) for 10 chemicals in the multi-species study with 5 species of mussels representing different tribes or families of freshwater mussels

		EC50 (95% CL)
Toxicant (Test no.)	Unit	Threeridge	Paper pondshell	Fatmucket	Washboard	Western pearlshell
Alachlor	mg/L	1.2 (0.8–1.8)a	6.7 (5.6–7.9)	2.7 (2.5–2.8)	5.7 (5.0–6.4)b	15 (12–20)
Metolachlor	mg/L	4.6 (3.0–7.1)a	15 (12–18)	20 (16–26)	6.8 (6.6–7.1)	29 (27–31)
Ammonia (total)	mg/L	1.5 (1.1–2.0)	2.1 (1.4–3.0)	5.2c/11c	5.1 (4.7–5.5)	8.0 (7.7–8.4)
Chloride (1)d	mg/L	1038 (808–1333)	1657 (1346–2039)	1897 (1850–1945)	1398 (1169–1670)	1576 (1391–1786)
Chloride (2)	mg/L	—e	—	1944 (1773–2131)	—	—
Chloride (3)	mg/L	—	—	2246 (2211–2281)	—	—
Chloride (4)	mg/L	—	—	1893 (1751–2045)f	—	—
Chloride (5)	mg/L	—	—	2092 (1910–2290)f	—	—
Potassium (1)d	mg/L	31 (17–57)	38 (35–42)	<56	48 (41–56)	<56
Potassium (2)	mg/L	—	—	46 (38–54)	—	38 (37–39)
Sulfated	mg/L	1338 (no CL)a, g	2709 (2548–2880)	2325 (2176–2485)	2279 (2206–2355)	1378 (1335–1422)
Chromium (VI)	μg/L	233 (179–303)a	213 (187–242)	266 (224–316)	138 (132–144)	624 (603–647)
Copper (1)	μg/L	11 (10–12)	13 (11–15)	48 (40–57)	25 (24–26)	36 (35–38)
Copper (2)	μg/L	10 (8.2–13)	—	35 (30–42)	—	—
Copper (3)	μg/L	—	—	55 (47–64)	—	—
Nickel (1)	μg/L	234 (207–264)	676 (648–705)	506 [350–731]h	173 (165–181)	269 (259–280)
Nickel (2)	μg/L	—	—	350 (298–411)	—	—
Nickel (3)	μg/L	—	—	445 (429–461)	—	—
Zinc	μg/L	299 (205–438)	520 (446–606)	576 (507–654)	566 (532–602)	447 (97–502)

^aThe EC50 should be used with caution due to low (<90%) control survival. See text for details.

^bOnly 45% mortality was observed in the highest test concentration (Supplemental Data, Table S4–1).

^cValues for fatmucket were from a previous study at pH 8.1 and 20 °C 8 (see details in text).

^dChloride was tested as sodium chloride, potassium as potassium chloride, and sulfate as sodium sulfate (details in Table 2).

^eA dash (—) indicates that the chemical was not tested with that species.

^fExposure solutions were prepared with a 30% series dilution (rather than 50% dilution in other tests).

^gNo CL could be calculated due to inadequate partial effects (Supplemental Data, Table S4–6).

^hAn EC50 could not be calculated due to no partial mortality (Supplemental Data, Table S4–9). The geometric mean of the bracketing concentrations with 0% and 100% mortality was calculated to obtain an estimated EC50. The 0% and 100% effect concentrations are provided in bracket as [0–100% effect concentration].

Survival was ≥90% in the dilution water controls in 56 of the 60 tests, and survival in the solvent controls was ≥90% in all organic chemical tests (Supplemental Data, Tables S4 and S5) and met the test acceptability criterion of ≥90% control survival 24. The control survival in 4 tests with threeridge (alachlor, metolachlor, sulfate, and chromium tests) ranged from 75% to 88% (Supplemental Data, Table S4–1,2,6,7) and did not meet the test acceptability criterion of ≥90% control survival. However, the control survival in the other 6 chemical tests, which were conducted concurrently with the alachlor, metolachlor, sulfate, and chromium tests using same batch of threeridge juveniles, was above 90%, ranging from 92% to 100% (Supplemental Data, Table S4–3,4,5,8,9,10), and an overall control survival for all 10 chemical tests was 90% (total of 44 control replicates, including solvent control). Furthermore, another copper toxicity test was conducted with a different batch of threeridge (Supplemental Data, Table S4–8), and the copper EC50s between the 2 threeridge tests were almost identical (Table 3). Therefore, the EC50s from the 4 tests with lower control survival were included in Table 3 with a footnote denoting that the control survival was less than 90%. When using the overall control survival data to recalculate the EC50s for the 4 tests, the EC50s would be slightly lower than those reported in Table 3: 1.1 mg/L (0.7–1.8 mg/L) alachlor, 4.4 mg/L (2.9–6.6 mg/L) metolachlor, 1017 mg/L (834–1241 mg/L) sulfate, and 206 μg/L (156–273 μg/L) chromium (VI).

Sensitivity comparisons among the 5 mussel species

The EC50s of the 5 mussel species ranged from 1.2 mg/L to 15 mg/L alachlor, 4.6 mg/L to 29 mg/L metolachlor, 1.5 mg/L to 8.0 mg/L total ammonia, 1038 mg/L to 2246 mg/L chloride, 31 mg/L to 48 mg/L potassium, 1338 mg/L to 2709 mg/L sulfate, 138 μg/L to 624 μg/L chromium, 10 μg/L to 55 μg/L copper, 173 μg/L to 676 μg/L nickel, and 299 μg/L to 576 μg/L zinc (Table 3). No single tested species had lower or higher EC50s than other species across all 10 chemicals. However, threeridge had the lowest EC50s for 8 of the 10 chemicals, and western pearlshell had the highest EC50s for 4 of the 10 chemicals. The EC50s for each chemical among the 5 species differed by a factor of ≤2 for chloride, potassium, sulfate, and zinc; a factor of ≤5 for ammonia, chromium, copper, and nickel; and factors of 6 and 12 for metolachlor and alachlor, respectively. In most cases, the 95% confidence limits for each chemical overlapped among the 5 mussel species. The results indicate that the mussels across different families or tribes had similar sensitivity to most of the tested chemicals, regardless of modes of action.

Sensitivity comparisons within a mussel species from the same population or from different populations

The EC50s for a single species in the repeated tests were similar (Table 3). Specifically, the EC50s for a species differed within a factor of 1.1 to 1.4 in the tests with 2 or 3 different batches of juveniles from the same population (i.e., fatmucket–chloride tests 1, 4 and 5; fatmucket–potassium test; threeridge–copper test; fatmucket–nickel test; Table 3). The low variability in EC50 among tests in a single laboratory was consistent with a previous study, in which the differences in EC50s in repeated copper tests with juvenile fatmucket from 1 population were within a factor of 1.4 ⁴². The EC50s for chloride or copper among different populations of fatmucket were also similar, within a factor of 1.2 for chloride and 1.6 for copper (Table 3; test number 1 for fatmucket from the Silver Fork, test number 2 for fatmucket from the Bourbeuse River, and test number 3 for fatmucket from the Kansas City Zoo). The results indicate that the sensitivity of juvenile fatmucket among different populations was similar in acute exposures to copper or chloride. Importantly, the sensitivity of juvenile fatmucket cultured from larvae of wild adults was similar to the sensitivity of juveniles cultured from larvae of captive-cultured adults (Kansas City Zoo). This result shows that captive-cultured adult mussels can reliably be used to reproduce juveniles for toxicity testing.

Sensitivity comparisons between fatmucket and other mussels

The sensitivity of the commonly tested fatmucket was compared to the other 4 mussels across the 10 chemicals with a regression plot. A strong linear relationship was obtained between EC50s for fatmucket and the other mussel species (r² = 0.97), and the slope of the regression was close to 1.0 (Figure 1). Over 73% of EC50s for the other 4 species were within 2-fold of EC50s for fatmucket. The results indicate that the toxicity of these chemicals to different species of mussels can be predicted with fatmucket toxicity data. Raimondo et al. 20 developed a toxicity database, including preliminary data from the present study, to examine the variability in sensitivity among mussels, cladocerans, and fish to chemicals with different modes of action. Their study demonstrated that 71% to 79% of EC50s for the most commonly tested mussel species (fatmucket, paper pondshell, or rainbow mussel Villosa iris) were within 2-fold of EC50s for up to 10 other mussel species; however, only 34% to 37% of EC50s for commonly tested cladocerans (D. magna and C. dubia) and 16% to 23% of commonly tested fish (Oncorhynchus mykiss, Pimephales promelas, and Lepomis macrochirus) were within 2-fold of EC50s for mussels ²⁰.

Fig 1.

Relationship between the median effect concentrations (EC50s) from a commonly tested mussel (fatmucket) and the EC50s from 4 other mussels in acute 96-h exposures with the 10 chemicals for the multi-species study. The solid line indicates the 1:1 line of perfect agreement, and dashed lines indicate 1:2 and 2:1 lines (i.e., bounds for EC50s from the 4 mussel species being within 2-fold of EC50s from fatmucket). For repeat tests with a chemical (Table 3), a geometric mean of EC50s was used. The results of linear regression analysis (SigmaPlot, Ver 13.0; Systat Software) are also shown.

Species sensitivity comparisons among mussels and other freshwater organisms

In the companion study (Ivey et al. 23 in this issue), the freshwater snail species (P. gyrina and L. stagnalis) and commonly tested amphipod (H. azteca) and cladocerans (C. dubia and D. magna) were tested in acute exposures to the 10 chemicals used in the multi-species study with the 5 mussels. Compared with the 5 mussels tested in the present study, the 2 snails had greater EC50s for ammonia, potassium, sulfate, and chromium but had similar EC50s for the other 6 chemicals. For the 10 chemicals, the EC50s for C. dubia were similar to or less than the EC50s for D. magna, H. azteca, and the 2 snails. However, C. dubia had substantially greater EC50s for ammonia and potassium compared with the 5 mussels. The results indicate that C. dubia may sufficiently represent sensitivity of the snails to the 10 chemicals but may not adequately represent the sensitivity of mussels to some chemicals. In the expanded database with more chemicals and mussel species, Raimondo et al. ²⁰ found that, in general, C. dubia was less sensitive than mussels.

Species mean acute values (calculated as geometric mean of EC50s for a test species) for all freshwater species in the compiled toxicity databases were ranked and plotted in cumulative distribution for the 10 chemicals tested in the present study (Figure 2). The ranges of ranked SMAVs for mussels in different families or tribes were relatively narrow, generally within the lower 50th percentile of the species sensitivity distributions for the tested chemicals, except for the 2 organic chemicals (alachlor and metolachlor; Figure 2). One or more mussel species were among the 4 most sensitive species in the databases for 8 of the 10 chemicals tested in the multi-species study. Of the 6 tested chemicals for which the USEPA has water quality criteria ⁴³, most species mean acute values for chloride or nickel from mussels and some species mean acute values for ammonia, copper, or zinc from mussels were similar to or less than the final acute value used to derive the USEPA acute criterion (Figure 2). These results indicate that mussels representing different tribes or families had similar sensitivity to inorganic chemicals across different modes of toxic action whereas the sensitivity to organic chemicals was relatively variable among mussels, and the water quality criteria development for alachlor, chloride, potassium, sulfate, copper, nickel, and zinc should reflect the sensitivity of mussels to these chemicals. The USEPA uses the genus level sensitivity distribution approach to develop acute water quality criterion, which is typically based on laboratory toxicity data from a suite of aquatic organisms that are assumed to represent the sensitivity of untested species (i.e., minimum data requirement of 8 different families 44). Because of the high sensitivity of mussels to numerous chemicals across different modes of action, water quality criteria that better protect freshwater organisms may be obtained if the minimum data requirement for deriving water quality criteria were updated to include native mussels as a required family.

Fig 2.

Ranked freshwater species mean acute values in compiled databases, with addition of species mean acute values for freshwater mussels of different tribes or families tested in the present study. Dashed line indicates the final acute value (FAV) in the US Environmental Protection Agency ambient water quality criteria (WQC)

The USEPA updated the ammonia water quality criteria in 2013, and the acute criterion is 1.4-fold lower than the previous acute criterion, primarily because of the inclusion of mussel toxicity data ³⁵. In the 2013 water quality criteria, the 10 lowest species mean acute values, 7 lowest genus mean acute values, and 2 lowest species mean chronic values are for mussels. It is reasonable to expect some of the approximately 270 US native species of mussels that were not included in that dataset may be equally or more sensitive. A few mussel species, including threeridge from the present study, had ammonia EC50s below the 2013 water quality criteria final acute value (Figure 2). This and other information on mussel sensitivity to ammonia may be utilized to develop site-specific environmental guidance values and to provide enhanced conservation of especially vulnerable or important populations and/or communities of mussels. Another approach may be the development of taxon-specific criteria. Conceptually, a taxon-specific criterion could be derived to protect a species, genus, or family that is not adequately protected by general national aquatic life water quality criteria. Taxon-specific criteria would provide probabilistic estimates of hazard based on the subset of data most relevant to individual taxa of conservation concern. A mussel-specific ammonia criterion would complement the general national water quality criteria, which are derived to be protective of a large number of taxa but not meant to protect all species, by providing a technically sound risk management option for water quality managers’ consideration when developing state or tribal water quality standards.

Toxicity tests with fatmucket and 10 additional chemicals in the single-species study

Ten additional chemicals were tested with fatmucket in the single-species study. Mean survival was >90% in the controls, including the solvent controls in organic chemical tests, and met the test acceptability criterion of ≥90% control survival ²⁴. Survival was >80% in all treatments at the end of exposures to 2,4-dichlorophenoxyacetic acid, bifenthrin, carbaryl, and aluminum (Supplemental Data, Table S5), even though the high nominal concentrations of the 3 organic chemicals were close to solubility and a large amount of aluminum floc was observed on the bottom of test beakers in the high and medium-high concentrations.

The EC50s for the 10 chemicals tested in the single-species study with fatmucket and the percentiles of the species mean acute values for fatmucket in the species sensitivity distribution for all freshwater species are presented in Table 4. The EC50s for fatmucket were in the higher percentiles (≥45th) of the species sensitivity distribution, except for 4-nonylphenol in the 27th percentile. There were limited mussel data for chemicals tested in the single-species study. The EC50 of 23 mg/L malathion for fatmucket tested in the present study was close to the low range of 96-h EC50s from 24 mg/L to 219 mg/L for juveniles of 6 other mussel species tested in a previous study ⁴⁵. Milam et al. 46 conducted acute toxicity tests with glochidia of 6 mussel species in 24-h exposures that included 3 organic chemicals tested in the present study with juvenile fatmucket. The EC50s of >311 mg/L 2,4- dichlorophenoxyacetic acid, 0.099 mg/L 4-nonylphenol, and >8.0 mg/L carbaryl for juvenile fatmucket tested in the present study were within the 24-h EC50 ranges of 82 mg/L to 437 mg/L 2,4- dichlorophenoxyacetic acid, 0.057 mg/L to 1.19 mg/L 4-nonylphenol, and 3.1 mg/L to 43 mg/L carbaryl for glochidia of the 6 mussel species 46, respectively. Of the 3 tested chemicals (carbaryl, arsenic [V], and aluminum) for which the USEPA has water quality criteria, the species mean acute values for fatmucket tested in the present study and other mussel species tested in previous studies ^{45, 46} were far above the final acute values (Table 4). The results indicate that mussels tested were not sensitive to the chemicals tested in the single-species study, except 4-nonylphenol. Notably, 7 of the 10 chemicals tested in the single-species study were organic chemicals (Table 4). The mussels appeared to be less sensitive to organic chemicals. The relative (to other taxa) insensitivity of mussels to other organic contaminants has been reported previously ⁴⁷.

Table 4.

Acute median effect concentrations (EC50s) and 95% confidence limits (CLs; in parentheses) for 10 chemicals in the single-species study with fatmucket (Lampsilis siliquoidea)^a

Toxicant	EC50 (95% CL) (mg/L)	FAV (mg/L)	SSD percentile
2,4-dichlorophenoxyacetic acid	>311	NA	76 (n = 24)
4-nonylphenol	0.099 (0.096–0.102)	0.056	27 (n = 29)
Azoxystrobin	0.725 (0.697–0.754)	NA	83 (n = 5)
Bifenthrin	>0.0267	NA	86 (n = 6)
Carbaryl	>8.0	0.004	77 (n = 61)
Malathion	23 [16–33]b	NA	82 (n = 75)
Molinate	53 (51–55)	NA	94 (n = 17)
Arsenic (V)	117 (113–122)c	0.68	62 (n = 12)
Calcium chloride	5383 [3862–7502]b	NA	45 (n = 10)
Aluminum (total)	>54	1.5d	72 (n = 17)

^aFinal acute values (FAVs) in the US national ambient water quality criteria and a percentile of EC50 for fatmucket in the species sensitivity distribution (SSD) for all freshwater species are presented.

^bAn EC50 could not be calculated because of no partial mortality (Supplemental Data, Table S5). The geometric mean of the bracketing concentrations with 0% and 100% mortality was calculated to obtain an estimated EC50. The 0% and 100% effect concentrations are provided in bracket as [0–100% effect concentration].

^cThe effect concentration was calculated in terms of assayed arsenic concentration.

^dThe FAV in draft updated water quality criteria for aluminum at hardness 100 mg/L and pH range of 6.5 to 9.0 (D. Eignor, US Environmental Protection Agency, Washington, DC, unpublished data).

NA = not applicable.

The demonstrated sensitivity of mussels to a diversity of inorganic toxicants makes ensuring mussels are represented in water quality criteria development or other hazard assessments critical for the protection of mussels. As mentioned previously, direct testing of mussels is recommended when deriving estimates of protective concentrations for inorganic constituents because mussels have been demonstrated to be among the most sensitive forms of aquatic life to some metals and common ions. Where direct testing of the taxa is not feasible, extrapolation approaches such as the USEPA ICE models can provide estimated values that represent inherent taxa sensitivity. The ICE models are log-linear relationships of acute sensitivity between a surrogate species and predicted taxa of interest and can be used to estimate toxicity to the predicted taxa (species, genus, family) from measured toxicity of the surrogate ¹⁹. The ICE models rely on an existing database of diverse species and chemicals, such as that used to develop species sensitivity distributions in the present study. Previous versions of the USEPA web-based ICE application had limited predictions to freshwater mussels because of limited measured data. A recent study expanded the database for freshwater mussels and provided additional guidance on inclusion of the mussel data into ICE models ²⁰.

CONCLUSIONS

The present study confirms findings from previous studies that mussels are generally sensitive to ammonia, metals, and ion constituents, but are not generally among the most sensitive organisms for organic chemicals (including various pesticides) in acute exposures. Furthermore, mussels representing different families or tribes had similar sensitivity to most tested chemicals regardless of modes of toxic action. The sensitivity of the commonly tested fatmucket was similar to other mussel species tested. In addition, the sensitivity of juvenile fatmucket among different populations or cultured from larvae of wild adults and captive-cultured adults was also similar in acute exposures to copper or chloride. Use of toxicity data from fatmucket, in conjunction with the available USEPA ICE models, should provide good estimates of risk to mussels regardless of their taxonomic classification for the purpose of deriving water quality criteria or other environmental guidance values and conducting risk assessments.

In compiled toxicity databases for freshwater organisms, mussels were among the more sensitive species to alachlor, ammonia, chloride, potassium, sulfate, copper, nickel, and zinc. Therefore, the development of water quality criteria and other environmental guidance values for these chemicals should reflect the sensitivity of mussels. Including a native mussel as a required family in the minimum data requirement for deriving water quality criteria ⁴⁴ should be considered in any water quality criteria developments or updates. Further studies are warranted to evaluate chronic sensitivity of mussels across phylogenetically diverse species in longer-term (e.g., 28 d to 90 d) exposures to chemicals with different modes of toxic action and to derive or update water quality criteria and other guidance to protect these long-lived mussels from long-term exposures.