Methods

Water samples were collected and composited from vertical transits throughout the stream depth at multiple locations along the channel cross section using depth- and width-integration methods described by Ward and Harr (1990), Edwards and Glysson (1999), and the USGS (U.S. Geological Survey, variously dated). These methods provided vertically and laterally discharge-weighted composite samples that represented the flow passing through the cross section of the stream. Samples were collected with isokinetic depth-integrating water-quality samplers (Davis and the Federal Interagency Sedimentation Project, 2005) constructed of plastic or coated with a nonmetallic rubber-coating paint and equipped with polytetrafluoroethylene nozzles.

Instantaneous streamflow was determined at the time of water sampling either by direct measurement or from stage-discharge rating tables (Rantz and others, 1982). Daily mean streamflow values during ice periods were labeled as estimated because backwater affected the stage-discharge relation. Onsite measurements of pH, specific conductance, and water temperature were measured during water sample collection. Sample filtration and preservation were completed according to procedures described by Ward and Harr (1990), Horowitz and others (1994), and the USGS (U.S. Geological Survey, variously dated).

Concentrations of arsenic, cadmium, copper, lead, manganese, and zinc in filtered samples (0.45-micrometer [µm] pore size) were measured using inductively coupled plasma-mass spectrometry (ICP–MS; Garbarino and others, 2006). Concentrations of calcium, magnesium, and iron in filtered samples were measured using inductively coupled plasma-atomic emission spectrometry (Fishman and Friedman, 1989). Calcium and magnesium concentrations were used to calculate water hardness. Concentrations of potassium, sodium, alkalinity, chloride, fluoride, silica, sulfate, nitrogen (nitrate plus nitrite), and dissolved organic carbon were measured in filtered samples collected at selected sites in the upper Clark Fork Basin. Potassium was measured by Standard Method 3120 (American Public Health Association, 1998); sodium and silica were analyzed by inductively coupled plasma-atomic emission spectrometry (Fishman, 1993); alkalinity was measured by electrometric titration (Fishman and Friedman, 1989); chloride, fluoride, and sulfate were measured by ion chromatography (Fishman and Friedman, 1989); nitrogen (nitrate plus nitrite) was measured by colorimetric enzymatic reduction (Patton and Kryskalla, 2011); and dissolved organic carbon was measured by persulfate oxidation and infrared spectrometry (Brenton and Arnett, 1993). Recoverable concentrations of trace elements were measured in unfiltered samples that were first digested with dilute hydrochloric acid (Hoffman and others, 1996). For cadmium, iron, lead, and manganese, the digested samples were analyzed by ICP–MS as described by Garbarino and Struzeski (1998). For arsenic, copper, and zinc, the digested samples were analyzed by ICP–MS as described by Garbarino and others (2006). Turbidity was measured using Standard Method 2130 (American Public Health Association, 1998) in selected unfiltered samples. All samples were analyzed at the USGS National Water Quality Laboratory (NWQL) in Denver, Colorado.

Water samples for analysis of suspended sediment were collected from multiple vertical transits when periodic water samples were collected. Water samples were analyzed for suspended-sediment concentration and the percentage of suspended-sediment mass finer than a 0.062-millimeter (mm) diameter (silt size and smaller) by the USGS Wyoming-Montana Water Science Center Sediment Laboratory (hereinafter referred to as the “Wyoming-Montana Sediment Laboratory”) in Helena, Mont., according to methods described by Guy (1969) and Dodge and Lambing (2006).

Continuous turbidity was measured from early spring (after ice breakup) to early fall (before stream freezeup) using model 6136 turbidity sensors (YSI, Inc, Yellow Springs, Ohio) at four tributary sites in the upper Clark Fork Basin near Anaconda (table 1). Turbidity values were recorded every 15 minutes and are stored in the USGS NWIS database (U.S. Geological Survey, 2022) at https://waterdata.usgs.gov/mt/nwis/current?type=quality. The in-field values differ from the results of laboratory-measured turbidity in discrete water-quality samples because of differences in the instrumentation and sampling procedures. Continuous recordings enable determination of the minimum and maximum turbidity values for each day and a daily mean turbidity, which is based on the average of all values in a 24-hour period. Procedures for the operation of continuous turbidity monitors and for daily record computations are described by Wagner and others (2006).

Results

Water-quality data from samples collected periodically during water year 2020 (October 1, 2019, through September 30, 2020) are listed in the accompanying data release (“table_4_water_quality_data_Clark_Fork_Basin.xlsx” in Clark and others [2022]). Daily maximum, minimum, and mean turbidity at four sites are listed in the accompanying data release along with monthly summary statistics (“table_5_daily_turbidity_at_Mill_Creek_at_Opportunity.xlsx,” “table_6_daily_turbidity_at_Willow_Creek_at_Opportunity.xlsx,” “table_7_daily_turbidity_at_Warm_Springs_Creek_at_Warm_Springs.xlsx,” and “table_8_daily_turbidity_at_Lost_Creek_near_Anaconda.xlsx” in Clark and others [2022]). Water-quality data are available in the USGS NWIS database (U.S. Geological Survey, 2022).

Quality Assurance

Quality-assurance procedures used for the collection and field processing of water samples are described by Ward and Harr (1990), Horowitz and others (1994), Edwards and Glysson (1999), Lambing (2006), and the USGS (U.S. Geological Survey, variously dated). Standard procedures used by the NWQL for internal sample handling and quality assurance are described by Friedman and Erdmann (1982), Jones (1987), and Pritt and Raese (1995). Quality-assurance procedures used by the Wyoming-Montana Sediment Laboratory are described by Dodge and Lambing (2006). Standard procedures used for the calibration, measurement, and quality assurance of turbidity monitors are described by Anderson (2005).

The quality of analytical results reported for water samples was evaluated using quality-control samples that were sampled and analyzed concurrently with primary environmental samples. These quality-control samples consisted of replicates, spikes, and blanks that provided quantitative information on the precision and bias of the overall field and laboratory processes. The number of quality-control samples represented about 15 percent of the total number of water samples.

Replicate data provided an assessment of the precision (reproducibility) of analytical results and collection variability (U.S. Geological Survey, variously dated). Replicate samples were collected in the field (field replicate) either by repeating the collection process (sequential or concurrent replicate) to obtain two or more independent composite samples or by splitting a single composite sample into two or more subsamples (split replicate). The individual replicate samples were analyzed separately. Likewise, a single sample can be analyzed two or more times in the laboratory to obtain a measure of analytical precision (laboratory replicate).

Precision of analytical results for field replicates can be affected by numerous sources of variability within the field and laboratory environments, including sample collection, processing, and analysis. Overall precision for samples exposed to field and laboratory sources of variability were provided by obtaining replicate stream samples for chemical analysis by splitting a composite stream sample. Replicate stream samples for suspended-sediment analysis were obtained in the field by collecting two independent cross-sectional samples. Analyses of field replicate samples indicated the reproducibility of environmental data that were affected by the combined potential variability introduced by field and laboratory processes.

In addition to analyzing quality-control samples submitted from the field, internal quality-assurance practices are completed systematically by the NWQL to provide quality control of analytical procedures (D.L. Stevenson, U.S. Geological Survey, written commun., 2012). These internal practices include analyses of quality-control samples such as calibration standard samples, standard reference water samples, replicate samples, deionized-water blank samples, or spiked samples at a proportion equivalent to at least 10 percent of the samples loaded. The NWQL participates in a blind-sample program in which standard reference water samples prepared by the USGS Quality Systems Branch are routinely inserted into the sample line for each analytical method at a frequency proportional to the sample load. The laboratory also participates in external evaluation studies and audits with the National Environmental Laboratory Accreditation Program, the U.S. Environmental Protection Agency, Environment and Climate Change Canada, and the USGS Quality Systems Branch to assess analytical performance.

Precision of analytical results for laboratory replicates, which exclude field sources of variability, was determined using two independent chemical analyses of aliquots from a single sample selected from the group of samples constituting each analytical run. A separate analysis of the sample was completed at the beginning and end of each analytical run to provide information on the reproducibility of laboratory analytical results independent of variability caused by field sample collection and processing. Laboratory replicates of suspended-sediment samples were not obtainable because the samples were consumed during the analysis.

Spiked samples were used to evaluate bias, which measures the ability of an analytical method to accurately quantify a known amount of analyte added to a sample. In the laboratory, deionized-water blank samples and aliquots of stream samples were spiked with known amounts of the same trace elements for which water samples were being analyzed. Analyses of spiked blanks indicated if the spiking procedure and analytical method are within control for water that is presumably free of chemical interferences. Analyses of spiked aliquots of stream samples indicated if the chemical matrix of the stream water interferes with the analytical measurement and if these interferences could contribute substantial bias to reported trace-element concentrations for stream samples.

Field blank samples were submitted for every field trip and analyzed to identify the presence and magnitude of contamination that could bias analytical results. Field blanks consisted of deionized water that is certified as constituent free and is processed in the field through clean sampling equipment used to collect stream samples. These blanks were subjected to the same processing (sample splitting, filtration, preservation, transportation, and laboratory handling) as stream samples. Blank samples were analyzed for the same constituents as stream samples to detect contamination.

All water samples were handled in accordance with chain-of-custody procedures that provide documentation of sample identity, shipment, receipt, and laboratory handling (Driscoll and Hatcher, 2010). All environmental and quality-control samples submitted from a sampling episode were stored in a secure area of the NWQL and analyzed as a discrete sample group, independent of other samples submitted to the NWQL; therefore, the quality-control data apply solely to the analytical results for stream samples reported herein and provide a direct measure of data quality for this study.

Data-quality objectives (table 3) were established by the USGS and U.S. Environmental Protection Agency as part of the study plan for the expanded monitoring program started in 1993. The objectives identify the analytical requirements of detectability and serve as a guide for identifying questionable data by establishing acceptable limits for precision and bias of laboratory results. Comparisons of quality-control data to data-quality objectives were used to evaluate if sampling and analytical procedures produced environmentally representative data in a consistent manner. Data that did not meet the objectives were evaluated for acceptability; if necessary, additional quality-control samples were submitted and corrective action was taken.

The precision of analytical results was determined by calculating the standard deviation of the differences in concentrations between replicate analyses. These replicate analyses consisted of pairs of field replicate samples and laboratory replicates. Standard deviations were calculated according to the following equation (Taylor, 1987):

Precision also was expressed as a relative standard deviation (RSD), in percent, which was computed from the standard deviation and the mean concentration for all the replicate analyses. Expressing precision relative to a mean concentration standardized the comparison of precision among individual constituents. The RSD was calculated according to the following equation (Taylor, 1987):

Sample results and the corresponding field replicate data are listed in the accompanying data release (“table_9_results_field_replicates_Clark_Fork_Basin.xlsx” in Clark and others [2022]). The precision for each constituent estimated from field replicate analyses and from laboratory replicate analyses is listed in the accompanying data release (“table_10_precision_field_replicates_Clark_Fork_Basin.xlsx” and “table_11_precision_laboratory_replicates_Clark_Fork_Basin.xlsx” in Clark and others [2022]). Statistics summarizing the precision of analytical results for field and laboratory replicates were calculated using unrounded values. Precision estimates for field and laboratory replicates were within the acceptable 20-percent RSD limit for all constituents except for suspended-sediment concentration, which resulted from one sample pair having concentrations exceeding 350 milligrams per liter (mg/L; “table_10_precision_field_replicates_Clark_Fork_Basin.xlsx” in Clark and others [2022]). No adjustments were made to analytical data based on replicate analysis precision.

Recovery efficiencies for constituent analyses were determined by comparing a sample and a spiked aliquot of the same sample. The data-quality objective for acceptable spike recoveries of trace elements in water samples was a maximum deviation of 25 percent from a theoretical 100-percent recovery of an added constituent. At the NWQL, a spiked deionized-water blank sample and a spiked aliquot of a stream sample were prepared and analyzed along with the original unspiked sample. The differences between the spiked and unspiked sample concentrations were determined and used to compute recovery, in percent, according to equation 3:

If the spike recovery of a trace element was outside a range of 75 to 125 percent, the instrument was recalibrated, and the sample set and all spiked samples were reanalyzed for that element until recoveries were improved to the extent possible. Recovery efficiency for individual trace elements in spiked blank samples and in spiked stream samples is listed in the accompanying data release (“table_12_recovery_efficiency_deionized_water_blank_samples.xlsx” and “table_13_recovery_efficiency_stream_samples.xlsx” in Clark and others [2022]). The mean recovery for spiked blank samples ranged from 97.6 to 104 percent (“table_12_recovery_efficiency_deionized_water_blank_samples.xlsx” in Clark and others [2022]). The 95-percent confidence intervals (Taylor, 1987) for the mean recovery for each constituent for which spiked blank samples were analyzed did not exceed a 25-percent deviation from an expected 100-percent recovery. The mean recovery for spiked stream samples (“table_13_recovery_efficiency_stream_samples.xlsx” in Clark and others [2022]) ranged from 80.4 to 111 percent. The 95-percent confidence intervals for the mean spike recovery for each constituent for which stream water samples were analyzed did not exceed a 25-percent deviation from an expected 100-percent recovery except for copper, unfiltered recoverable (33.3 and 128 percent); iron, unfiltered recoverable (66.7 percent); lead, unfiltered recoverable (37.4 and 126 percent); manganese, unfiltered recoverable (29.6 and 132 percent); zinc, unfiltered recoverable (35.1 and 138 percent); arsenic, filtered (60.1 and 125 percent); and arsenic, unfiltered recoverable (129 percent). No adjustments were made to analytical data based on the mean spike recovery.

High or low bias is indicated if the 95-percent confidence interval does not include 100-percent recovery, thereby indicating a consistent deviation or bias, either high or low. Confidence intervals for percentage recovery include 100 percent for all laboratory-spiked blank samples (“table_12_recovery_efficiency_deionized_water_blank_samples.xlsx” in Clark and others [2022]) except for cadmium, unfiltered recoverable (102–105 percent); lead, filtered (96.5–99.5 percent); and manganese, unfiltered recoverable (96.2–99.0). Confidence intervals for percentage recovery include 100 percent for all laboratory-spiked stream samples except for iron, filtered (86.8–99.1 percent; “table_13_recovery_efficiency_stream_samples.xlsx” in Clark and others [2022]). No adjustments were made to analytical results for stream samples based on spike recoveries.

Analytical results for field blanks are listed in the accompanying data release (“table_14_analyses_of_field_blanks_for_water_samples.xlsx” in Clark and others [2022]). Field blanks with constituent concentrations less than or equal to the laboratory reporting level (LRL) indicated that the process of sample collection, field processing, and laboratory analysis is presumably free of contamination. If detectable concentrations of trace elements in field blanks were greater than or equal to twice the LRL, the concentrations were noted during data review. Analytical results from the field blank were evaluated for evidence of a consistent trend that could indicate systematic contamination. Sporadic, infrequent, nonconsecutive exceedances of twice the LRL most likely represented random contamination or laboratory-instrument calibration error that was not persistent in the process and was not likely to cause positive bias in a long-term record of analytical results; however, if concentrations for a particular constituent exceeded twice the LRL in field blanks from two consecutive field trips, additional blank samples were collected and submitted for analysis to identify the source of contamination.

Constituent concentrations in field blanks almost always were less than the LRL. Two sample concentrations of chloride, filtered (0.04 microgram per liter [µg/L] and 0.06 µg/L), exceeded the LRL of 0.02 µg/L. Two sample concentrations of silica, filtered (0.10 mg/L and 0.08 mg/L), exceeded the LRL of 0.05 mg/L. One sample concentration of copper, filtered (1.4 µg/L), exceeded the LRL of 0.4 µg/L. Two sample concentrations of copper, unfiltered recoverable (1.1 µg/L, and 2.6 µg/L), exceeded the LRL of 0.4 µg/L. One sample concentration of iron, unfiltered recoverable (5.2 µg/L), exceeded the LRL of 5.0 µg/L. Two sample concentrations of organic carbon, filtered (0.26 mg/L and 0.30 mg/L), exceeded the LRL of 0.23 mg/L. No adjustments were made to water-quality sample data based on a review of these results.

Methods

Fine-grained bed-sediment samples were collected in July 2020 using protocols described by Axtmann and Luoma (1991). Samples were collected from the surfaces of streambed deposits in areas near the edge of the stream using an acid-washed polypropylene scoop. Whenever possible, samples were collected from both sides of the stream.

Individual samples of bed sediment were collected by scooping material from the surfaces of three to five randomly selected deposits along pools or low-velocity areas. The three to five individual samples were combined to form a single composite sample. This collection process was repeated three times to obtain three composite samples. Each composite sample was wet sieved onsite through a 0.063-mm polyester-mesh sieve using ambient stream water. The fraction of bed sediment in each composite sample that was finer than 0.063 mm was collected in an acid-washed 500-milliliter (mL) polyethylene bottle and transported to the laboratory on ice.

Bed-sediment samples were processed and analyzed at the USGS Metal Bioavailability Laboratory in Menlo Park, California. Bed-sediment samples were oven dried at 60 degrees Celsius and ground into smaller particle sizes using an acid-washed, ceramic mortar and pestle. Single aliquots of about 0.5–0.6 gram of sediment from each of the three composite bed-sediment samples were digested using a hot, concentrated nitric acid reflux according to methods described by Luoma and Bryan (1981). Laboratory replicates were analyzed by taking an aliquot from one of the three sieved replicate samples at each site. After a 2-week digestion period, the aliquots were evaporated to dryness on a hot plate. The dry residue was reconstituted in 10 mL of 0.6 N (normal) hydrochloric acid. The reconstituted aliquots were then filtered through a 0.45-µm pore-size filter using a syringe and an in-line disposable filter cartridge. The filtrate was diluted to a 1 to 10 ratio with 0.6 N hydrochloric acid. These final solutions were analyzed for arsenic, cadmium, chromium, copper, iron, lead, manganese, nickel, and zinc using inductively coupled plasma-optical emission spectrometry (Hornberger and others, 1997). The smallest concentration of a constituent that can be reliably reported for analyses of bed sediment is termed the minimum reporting level (MRL).

Results

Solid-phase concentrations of trace elements measured in samples of fine-grained bed sediment collected during July 2020 are listed in the accompanying data release (“table_15_bed_sediment_data_Clark_Fork_Basin.xlsx” in Clark and others [2022]). Liquid-phase concentrations, measured in micrograms per milliliter, were analyzed in the reconstituted aliquots of digested bed sediment. Solid-phase concentrations, measured in micrograms per gram, were calculated using the following equation:

The reported solid-phase concentrations (Clark and others, 2022) are the means of all analyses for replicate aliquots from each composite bed-sediment sample collected at each site. Because the conversion from liquid-phase to solid-phase concentration is dependent on the dilution ratio and the dry weight of the sample, MRLs for some trace elements might differ among sites and among years.

Quality Assurance

The USGS protocols for field collection and processing of bed-sediment samples are designed to prevent contamination from metal sources. Nonmetallic sampling and processing equipment (white plastic scoop, funnel-frame apparatus, and 500-mL sample bottles) was acid washed and rinsed with deionized water before the collection of the first sample. Polyester-mesh sieves were washed in laboratory-grade detergent and rinsed with deionized water. All equipment received a field rinse onsite with native water. Sampling equipment used at more than one site was rinsed thoroughly between sites with site-specific stream water. Separate sieves were used at each site and, therefore, did not require decontamination between sites.

Quality assurance of analytical results for bed-sediment samples included laboratory-instrument calibration with standard solutions and analysis of quality-control samples designed to identify the presence and magnitude of bias (Ellen V. Axtmann, U.S. Geological Survey, written commun., 1994). Quality-control samples consisted of standard reference materials (SRMs) issued by the National Institute of Standards and Technology (NIST) and procedural blanks. In total, 12 low-concentration SRMs, 12 high-concentration SRMs, and 13 procedural blanks were analyzed.

SRMs are commercially prepared materials that have certified concentrations of trace elements. Analyses of SRMs were used to indicate the ability of the method to accurately measure a known quantity of a constituent. Multiple analyses of SRMs were made to derive a mean and 95-percent confidence interval for recovery. Recovery efficiency for trace-element analyses of SRMs for bed sediment is listed in the accompanying data release (“table_16_recovery_efficiency_bed_sediment_samples.xlsx” in Clark and others [2022]). Two SRMs, consisting of agricultural soils and representing low (SRM sample 2709, NIST San Joaquin soils; hereinafter referred to as “SRM sample 2709”) and high concentrations (SRM sample 2711, NIST Montana soil II; hereinafter referred to as “SRM sample 2711”) of trace elements, were analyzed to test recovery efficiency for a range of concentrations that are likely detected in the bed sediment in streams in the upper Clark Fork Basin. The digestion used for the bed-sediment samples was not a “total” digestion (does not liberate elements associated with crystalline lattices) but rather a “near-total” digestion. As a result, 100-percent recoveries were unlikely for elements strongly associated with crystalline phases. Recoveries of trace elements from SRMs processed by “near-total” digestion are useful to identify trace elements weakly or strongly associated with sediment particles and to evaluate the consistency of recoveries among multiple sets of analyses.

Mean recoveries of arsenic, cadmium, chromium, copper, iron, lead, manganese, nickel, and zinc from SRM sample 2709 ranged from 49.0 to 92.1 percent (“table_16_recovery_efficiency_bed_sediment_samples.xlsx” in Clark and others [2022]). Elements with low certified concentrations (cadmium, 0.371 microgram per gram) or with a strong association with crystalline lattices (arsenic, chromium, lead) had lower percentage recoveries (mean recoveries between 49.0 and 76.2 percent). Mean recoveries in SRM sample 2711 ranged from 70.6 to 97.0 percent (“table_16_recovery_efficiency_bed_sediment_samples.xlsx” in Clark and others [2022]). Because of the strong binding nature of chromium’s crystalline lattice, chromium had the lowest mean recovery at 70.6 percent. No adjustments were made to trace-element concentrations in bed-sediment samples based on recovery efficiencies.

Procedural blanks for bed-sediment samples consisted of the analysis of the same reagents used for sample digestion and reconstitution. Concentrated nitric acid used for sample digestion was heated and evaporated to dryness. After evaporation, 0.6 N hydrochloric acid was added to reconstitute the dry residue. Analytical results of procedural blanks for bed sediment were reported as a liquid-phase concentration, in micrograms per milliliter. A procedural blank was prepared and analyzed concurrently with bed-sediment samples for each site. Concentrations of trace elements in all procedural blanks were less than the MRL for all elements (“table_17_analyses_of_procedural_blanks_for_bed_sediment_samples.xlsx” in Clark and others [2022]). No adjustments were made to analytical data based on procedural blanks.

Methods

Insect samples were collected using protocols described in Hornberger and others (1997). Benthic insects at immature stages were collected with a large nylon-mesh kick net. A single riffle at each site was sampled repeatedly until an adequate number of individual insects was collected to provide enough mass for analysis (for example, about 10 stoneflies, about 100–200 caddisflies). Two caddisfly species of the genus Hydropsyche (Hydropsyche cockerelli and Hydropsyche occidentalis) were targeted for collection in this study because they were detected at most sites. Hydropsyche species (spp.) that could not be positively identified were categorized as Hydropsyche spp. or Hydropsyche morosa group. On the few occasions when Hydropsyche were not present, other caddisflies, including Brachycentrus spp. and Rhyacophila spp., were collected. The caddisfly Arctopsyche grandis and the stoneflies Claassenia sabulosa and Hesperoperla spp. were collected where available to represent additional insect taxa commonly distributed in the Clark Fork Basin but with different feeding traits.

Samples of each taxon were sorted by genus in the field and placed in acid-washed plastic containers. Samples were frozen in a small amount of ambient stream water on dry ice within 30 minutes of collection. Between 1986 and 1998, macroinvertebrate containers were kept on ice to allow the insects to evacuate their gut contents (depurate) for 6 to 8 hours. Excess water was drained, and insects were frozen for transport to the laboratory. Since 1999, samples were immediately frozen on dry ice in the field to reduce the possibility of metal loss through intracellular breakdown during depuration. A comparison of immediately frozen to depurated samples indicated that although no substantial difference was detected for most metals, concentrations of copper were about 20 percent lower in the depurated samples than in the samples that were immediately frozen. The data were not adjusted for this difference.

Invertebrate samples were processed and analyzed at the USGS Metal Bioavailability Laboratory in Menlo Park, Calif. Insects were thawed, rinsed with ultrapure deionized water to remove particulate matter, and then sorted to their lowest possible taxonomic level. If large numbers of specimens were collected at a site, similar-sized individuals were composited into replicate subsamples. Subsamples were placed in tared scintillation vials and oven dried at 70 degrees Celsius. Subsamples were weighed to obtain a final dry weight and digested by reflux using concentrated nitric acid (Cain and others, 1992). After digestion, insect samples were evaporated to dryness on a hot plate. The dry residue was reconstituted in 0.6 N hydrochloric acid, filtered through a 0.45-µm pore-size filter, and analyzed undiluted using inductively coupled plasma-optical emission spectrometry (Hornberger and others, 1997) for arsenic, cadmium, chromium, copper, iron, lead, manganese, nickel, and zinc. The smallest concentration of a constituent that can be reliably reported for analyses of tissue samples is termed the MRL.

Results

Concentrations of trace elements in whole-body tissue of aquatic invertebrates collected during July 2020 are listed in the accompanying data release (“table_18_biological_data_Clark_Fork_Basin.xlsx” in Clark and others [2022]). The variability in the number of composite samples among species and among sites reflects differences in invertebrate abundance (the number of composite samples increases with the relative abundance of insects). Liquid-phase concentrations, in micrograms per milliliter, analyzed in the reconstituted samples were converted to solid-phase concentrations, in micrograms per gram, using equation 4 (used earlier in this report to calculate solid-phase concentrations of trace elements in bed sediment). All tissue samples were analyzed undiluted (the dilution ratio was 1 to 1).

Quality Assurance

The USGS protocols for field collection and processing of tissue samples were designed to prevent contamination from metal sources. Nonmetallic nets, sampling equipment, and processing equipment were used in all sample collection. Equipment was acid washed and rinsed in ultrapure deionized water before the first sample collection. Nets and equipment were thoroughly rinsed in stream water at each main-stem site. Clean nets were used at each tributary site.

Quality control of analytical results for tissue samples included laboratory-instrument calibration with standard solutions and analyses of quality-control samples designed to quantify precision and to identify the presence and magnitude of bias. Quality-control samples consisted of 12 replicates of the certified reference material (CRM) TORT–3 (lobster hepatopancreas), which was purchased from the National Research Council Canada. Quality-control samples were analyzed in a proportion equivalent to about 20 percent of the total number of biota samples. Recovery efficiencies for trace-element analyses of the TORT–3 CRM are listed in the accompanying data release (“table_19_recovery_efficiency_biota_samples.xlsx” in Clark and others [2022]). Data-quality objectives have not been established for analytical recovery in tissue, but percentage recoveries indicate analytical performance.

Mean CRM recoveries for TORT–3 ranged from 85.3 to 112 percent for all constituents except lead (“table_19_recovery_efficiency_biota_samples.xlsx” in Clark and others [2022]). The mean recoveries were within 15 percent of theoretical 100-percent recovery (based on the 95-percent confidence interval) for arsenic, cadmium, chromium, copper, iron, manganese, nickel, and zinc. Recoveries for lead were less than the analytical detection limits for the CRM and are not reported here; however, lead concentrations in the tissue samples are much greater than detection limits and are included in the accompanying USGS data release (Clark and others, 2022). No adjustments were made to trace-element concentrations in biota samples based on recovery efficiencies.

Procedural blanks for biota consisted of undiluted aliquots of the same reagents used to digest and reconstitute tissue of aquatic insects. Analytical results of procedural blanks for biota (“table_20_analyses_of_procedural_blanks_for_tissue_samples.xlsx” in Clark and others [2022]) were reported as a liquid-phase concentration, in micrograms per milliliter. A procedural blank was prepared and analyzed concurrently with tissue samples for each site. Concentrations of trace elements in all procedural blanks were less than the MRL; therefore, no adjustments to the data were necessary.