Assessment of Columbia River Sediment Toxicity to White Sturgeon: Concentrations of Metals in Sediment, Pore water and Overlying Water

Copyright: © 2015 Vardy DW, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Assessment of Columbia River Sediment Toxicity to White Sturgeon: Concentrations of Metals in Sediment, Pore water and Overlying Water


Introduction
Alteration of habitat, including pollution, is hypothesized as a major contributing factor to the global decline of populations of sturgeons [1][2][3][4][5][6]. Given their epi-benthic nature, sturgeon are potentially at risk of exposure to contaminants associated with sediments. In the Upper Columbia River (UCR), between Grand Coulee Dam in the USA and Hugh L. Keenleyside Dam in southern British Columbia, Canada, resides a population of fewer than 2500 white sturgeon (Acipenser transmontanus) that have been experiencing poor recruitment for over forty years [4,7,8]. Although specific reasons for their decline are not fully understood, pollution has been hypothesized as a potential contributing factor to the observed recruitment failure [8]. Specifically, there are concerns that contaminated sediments in the UCR may be bioavailable to sturgeon and that early life stages, including the early hiding stage where fry are in proximity to sediments, may be at risk.
The UCR is subject to multiple sources of pollution, including discharges from pulp and paper mills, wastewater treatment plants, and diverse mining and smelting operations [8]. In particular, a metallurgical facility in Trail, BC, Canada, historically released slag into the river, and historically and presently releases liquid effluents. Slag is a partially vitreous by-product of the metal refining process and there are concerns about the leaching of metals into water. Elevated concentrations of trace-elements, such as copper (Cu), lead (Pb), cadmium (Cd), and zinc (Zn), relative to reference sites, have been found in sediments downstream of the metallurgical facility [9][10][11][12]. Sediments are sinks for pollutants and can contain elevated concentrations of metals, which can be released back into the water column following remobilization [13,14]. Therefore, in addition to exposure to pollutants in the water column, sturgeon might be exposed to contaminants associated with sediments or contaminants released into the sediment-water interface.
In order to assess the risk posed by exposure to metals bound in sediments of demersal fishes, such as sturgeon, bioavailability and concentrations of metals in pore water, overlying water, and at the sediment-water interface, need to be characterized. exposure to metals bound in sediments to demersal fishes, such as sturgeon, bioavailability and concentrations of metals in pore water, overlying water, and at the sediment-water interface, need to be characterized. Total concentrations of metals in sediments are poor indicators of potential toxicity and risk because a significant proportion of metal might be sequestered and biologically unavailable [15]. The bioavailable fraction of a metal is largely dependent on modifying factors of the environment, both abiotic and biotic, that influence the amount of a metal that can interact with biological processes in an organism, and thus, result in toxicity [16]. Redox potential and pH can greatly affect the chemistry of a sediment-bound metal, governing its distribution between the solid and dissolved phases, and in turn its movement between the various matrices, with dissolved metals in pore water considered to be the most bioavailable [15,16]. To assess the bioavailability and toxicity of metals associated with sediments various techniques have been employed, several of which include chemical analysis of whole digested sediment samples, or active or passive measurements of concentrations of metals in pore water using centrifugation or sampling devices such as peepers or diffusive gradients in thin films (DGTs).
Previous studies have investigated releases of elements from contaminated sediments in the Columbia River by use of several methods, such as quantification of metals in pore water (interstitial water), overlying water, and supernatants of aggressively tumbled slurries [17,18]. From these studies Paulson and Cox [18] concluded that under certain conditions releases of elements from sediment could result in concentrations of metals in various matrices that might be toxic to aquatic organisms. However, the Paulson and Cox [18] study employed techniques to simulate the potential release of metals from sediments under laboratory conditions that are unlikely to be directly applicable to conditions found in the above-mentioned region of concern of the UCR. Therefore, the purpose of the present study was to assess toxicity of chemicals of potential concern (COPC) associated with sediments to early life stage white sturgeon in the UCR by use of a controlled, laboratory, fluvial, exposure that simulated water quality characteristics found within the UCR in the vicinity downstream of Trail, BC, Canada. The present study was conducted under the oversight of the US EPA (www.ucr-rifs.com), and data obtained from this work will be used to supplement information in a baseline ecological risk assessment (BERA) and as part of a remedial investigation and feasibility study (RI/FS).
In order to accurately assess toxicity of UCR sediments to white sturgeon, a laboratory-based experimental design was needed that captured potential exposure routes of early life stages of white sturgeon to COPC, especially metals in the river, while assessing bioavailability by allowing quantification of chemical parameters in several matrices. To assess bioavailability of metals in UCR sediments and overlying waters, the present study employed a variety of sampling techniques. Peepers [19], DGTs [20], and active sampling/suction techniques were employed as a multiple lines of evidence approach to investigate concentrations of metals and chemical parameters in various matrices and to compare results of the various methods. Typically, analysis of pore water is achieved through either active or passive sampling, such as centrifuged and filtered core samples or use of membranes and dialysis chambers (peepers). Because disturbances of sediments during sampling of pore waters have been found to alter chemistries of sediments and affect bioavailability [21], in the present study peepers were employed to minimize disruption of sediment during sampling and extraction. In addition, peepers enable direct comparisons of concentrations of metals in pore water and overlying water [22], which is where early life stages of white sturgeon are likely to occur.
As a secondary measure and to compare concentrations of metals in pore water and at the sediment-water interface, DGT's were employed. DGTs utilize an ion-exchange resin and an ion-permeable gel membrane to quantitatively measure concentrations of metals in situ [20], and are a relatively non-intrusive method of sampling. Lastly, active sampling methods through direct collection of overlying water, sediment-water interface water, and pore water, by use of suction techniques with syringes, pipettes, and sediment-embedded air stones, were employed. These sampling techniques allowed for a greater volume of water to be collected from the relevant locations, which facilitated a larger suite of chemical analyses.
To identify specific data needs in addressing the aforementioned concerns and to establish decision rules for the collection of data, the EPA data quality objective (DQO) process [23] was used for this study. Specific DQOs addressed included survival and growth of white sturgeon reared on sediments from the UCR relative to reference sediments. The present article reports concentrations of metals to which early life stages of white sturgeon could be exposed and bioavailability of metals associated with sediments from the UCR. The responses of white sturgeon are presented separately [24].

Site selection and collection of sediments
Locations in the UCR from which sediments were collected were in areas known to encompass confirmed white sturgeon spawningand/or nursing-grounds [25,26], as well as to represent a range of exposure conditions [11,12]. Sampling focused on the reach of the UCR between Kettle Falls (river mile [RM] 703) to the U.S.-Canada border (RM 745), and was intended to represent a gradient of COPC concentrations in sediments associated with granulated slag [11,12]. Specifically, the primary COPCs were postulated to include Cd, Cu, Pb, and Zn [11,12]. Areas from which samples of sediments were collected included Deadman's Eddy (DME; RM 737), Northport ( Map 2). In addition to site-specific sediments and reference sediments, artificial substratum sediment (Aquarium Substratum Item No. 12648, Rolf C. Hagen Inc., Baie d'Urfe, QC, Canada) was also used as a negative control (CTRL), as evaluated and selected through method development work (see study design section in methods and Supplemental Material).
Surface sediments, defined as the upper 10 to 15 cm (4 to 6 in.) of the sediment column, were collected using a custom-built stainless steel power VanVeen grab sampler that was specifically designed to operate in hard bottom substrata. Depending on sampling success, as much as ten 20-L (5-gal) polyethylene buckets per sampling location (30 buckets per sampling area) were collected to attain the target sediment volume of approximately 200-L (~50 gal) per location. Immediately after collection, sediments were transferred into 20-L decontaminated polyethylene buckets, sealed, and transported in a refrigerated truck (4°C) to the University of Saskatchewan (UofS), Saskatoon, Saskatchewan, Canada where they were held at 4°C until initiation of experiments.
Prior to use, composites of each type of sediment were made by thoroughly homogenizing individual samples within an area for site sediments collected from the potentially affected areas of the UCR, reference sediments, and control sediments. This was achieved by use of a Teflon ® -lined, cement mixer retrofitted with a high density polyethylene drum and stainless steel paddles, as deemed an appropriate and effective method of mixing through method development work (Supplemental Material).

Study design
Prior to initiation of the definitive study, extensive method development work was conducted in order to evaluate and inform critical design components and considerations of specifically designed flow-through, fluvial simulation system for use in sediment toxicity tests with early life stage white sturgeon at the UofS ATRF. Specifically, an experimental exposure system was needed to allow for adjustment of flow velocity, water replacement time, and recirculation frequency, and provide versatility in sampling techniques while maintaining a practical and reproducible fluvial exposure. A full description of method development work, results, and final test design is provided in Supplemental Material.
Exposure chambers were continuous flow-through systems designed and operated at a rate of flow of approximately 20 L/min, with an illumination cycle of 16-light:8-dark (16:8) hrs., and target water temperature of 16 ± 1°C. Water was both renewed and re-circulated within each system. Flow-through conditions were set such that, on average, one complete water replacement in each exposure system occurred every 6 h. Test water used during the study had a target water hardness of 65 to 70 mg/L as CaCO 3 to simulate conditions found in the UCR, and consisted of a 1:1 mixture of de-chlorinated City of Saskatoon water and ATRF reverse osmosis water. The overall study design elements were in accordance with standard American Society for Testing Materials (ASTM) guidelines for testing early life stages of fish [27], with minor modifications for white sturgeon.
Homogenized sediments were evenly layered at the bottom of dedicated continuous flow-through exposure chambers at a thickness of approximately 2 inches. Replicate exposure chambers were established based on available sediment volume, with up to a maximum of six replicates per sample location. In addition to exposure chambers containing site sediments, reference sediments, or control sediment, a second negative control (water-only [H 2 O] control) was also established and monitored throughout the duration of the study. Six replicates were established for sediments collected from UMF-01, LD-01, and LALL, four replicates from LMF-02 and GE, and two replicates from NP-03. In addition, three replicates from substrata collected above the water line from the gravel bar at Deadman's Eddy (hereafter referred to as "DE") were also included, as there was difficulty in collecting sufficient volumes of site sediments (see results section).
In order to create a pseudo-hyporheic zone, large pebbles (Aquarium Substratum Item No. 12422, Rolf C. Hagen Inc., Baie d'Urfe, QC, Canada) were systematically placed in each exposure chamber at approximately 4 stones per 100 cm 2 to fulfill early life stage white sturgeon habitat requirements (Supplemental Material).

Collection of water and pore water
Concentrations of metals were quantified in overlying water, sediment-water interface water, and pore water to characterize exposure through the various possible aqueous exposure routes. Of the 42 exposure chambers, 11 were designated as chemistry-only (Supplemental Material), in which passive sampling devices, such as peepers and DGT probes, were installed and used to obtain additional water quality information within the top ± 1 cm of the sediment-water interface. Given that both DGT probes and peepers require a distinct period of equilibration (2 and 7 days, respectively), and necessitate disturbing the sediment during deployment and retrieval, dedicated chemistry-only exposure chambers were used for these measurements. These exposure chambers were seeded with the same number of white sturgeon and treated in the same manner as the regular exposure chambers except for the incorporation of the additional analytical devices. Exposure chambers designated as chemistry-only were to ensure that potential stress, if any, resulting from perturbations in deploying and retrieving DGT probes and peepers were not erroneously considered when interpreting effects on white sturgeon.
Direct sampling of pore water and overlying water: During placement of sediments/substrata into test chambers, up to eight ceramic air-stones (RENA Micro Bubbler 6-in. ceramic air-stones, Mars Inc., Hackettstown, NJ, USA) were distributed along the length of each exposure chamber for non-intrusive collection of pore water at a depth of approximately 2.5 cm (1 in.) below the sediment surface (Supplemental Material). Each air-stone was connected to a 15-ml syringe through a port in the side of the exposure chamber that would allow for extraction of pore water (Supplemental Material). Samples of water were also collected at the sediment-water interface and overlying water via suction by use of high density polyethylene (HDPE) pipettes and syringes, respectively. For the purpose of the present study, sediment-water interface water is defined as that water located within the boundary between sediment and the overlying water column, within 1 cm above the sediment surface, within respective exposure chambers. Samples of overlying water within exposure chambers were collected within the top 15 cm (6 in.) of the water column.
removing any residual particles from the gel surface. Upon removal and rinsing, DGT probes were sliced along the sediment-surface water line using a Teflon ® coated blade. Respective top and bottom gel portions were further sliced into three equal stripes, transferred into dedicated 15-mL high-density polypropylene centrifuge tubes, preserved with 5-mL of 1-M HNO 3 , and transported to CAS for chemical analysis.

Sampling of sediments
Following homogenization and prior to placement in the exposure chambers at the initiation of the study, sub-samples of sediment were collected for each sampled site and submitted to CAS for chemical analyses (see chemical analysis and water quality section in methods). Furthermore, at the end of the study, samples of sediments were also collected from each exposure chamber and submitted for further analytical testing (see chemical analysis and water quality section in methods). In addition, confirmatory analytical testing of reference and control sediments was also completed prior to study initiation.

Chemical analysis and water quality
As required by the design (see Supplemental Materials), temperature, pH, dissolved oxygen (DO), and conductivity were monitored daily using appropriate YSI electrodes (YSI Inc., Yellow Springs, OH, USA), while alkalinity, inorganic nitrogen, such as ammonia and nitrate, and hardness were monitored weekly using LaMotte Company colorimetric and titration test kits (Chestertown, MD, USA). In addition, pore water (at 2.5 cm below the sediment surface), sediment-water interface, and overlying water samples were collected weekly and submitted to CAS for chemical analyses.
All water samples were analyzed for target analyte list (TAL) metals, major cations/anions, alkalinity, hardness, and organic carbon (dissolved and total fractions). TAL metals (dissolved and total fractions) included: aluminum (Al), antimony (Sb), arsenic (As), barium (Ba), beryllium (Be), Cd, chromium (Cr), cobalt (Co), Cu, iron (Fe), Pb, manganese (Mn), mercury (Hg), molybdenum (Mo), nickel (Ni), selenium (Se), silver (Ag), thallium (Tl), vanadium (V), and Zn. Major cations/anions as defined for this study include: calcium (Ca), magnesium (Mg), potassium (K), sodium (Na), sulfate (SO 4 ), chloride (Cl), and fluoride (F). All water samples, including pore water, sedimentwater interface, and overlying water, were extracted using acid-cleaned and nanopure water rinsed HDPE syringes. With the exception of dissolved fractions, extracted samples were directly discharged into pre-preserved sampling containers (see Supplemental Materials for list of preservatives), and transported at 4°C to CAS for chemical analysis. Samples in which analyses required the dissolved fraction were filtered through 0.45-μm polyethersulfone filters (Whatman, Sigma-Aldrich, Oakville, ON, Canada) before being transferred into pre-preserved sampling containers. A summary of analytical methods and associated detection limits employed by CAS for water samples in the present study is provided in Supplemental Materials.
Total concentrations of TAL metals, Acid Volatile Sulfide (AVS) and simultaneously extracted metals (SEM), total organic carbon (TOC), polychlorinated biphenyls (PCBs), organochlorine pesticides, polycyclic aromatic hydrocarbons (PAHs), pH, and grain size were determined for all samples of sediment. Organochlorine pesticides included Dichloro Diphenyl Trichloroethane DDT), dichlorodiphenyldichloroethylene (DDE), and dichlorodiphenyldichloroethane (DDD). A summary of analytical methods and associated detection limits for sediment/ substratum samples is provided in Supplemental Materials. Environmental Standards Inc. (ESI; Valley Forge, PA, USA) performed an independent quality assurance and data validation review of the results produced by CAS. The review was performed in accordance with requirements specified by US EPA guidance documents [29][30][31][32]. Data were examined to determine usability of the analytical results and compliance relative to requirements specified above and the analytical methods. In addition, deliverables were evaluated for completeness and accuracy. Most analytical data were useable (> 88 %), with qualifications presented in data validation reports and summarized in Supplemental Materials. Only useable data were included herein. There were instances where analytes were considered "not-detected" because they were detected at concentrations equivalent to that in the associated field blanks. For these samples, concentrations were reported as the Limit Of Quantification [29][30][31][32]. In cases where measured values were 10x ≥ the associated blank, measured values were reported directly without correction. An EPA Quality Assurance/ Quality Control (QA/QC) chemist reviewed the draft data and data validation reports, and EPA approved the data for public use.

Validation assessment-overall data quality
During routine cleaning operations of exposure chambers UMF-D (Day 22) and CTRL-D (Day 23), a significant number of sturgeon fry were lost. Given that these events occurred well beyond the 48hr permissible re-seeding window, these exposure chambers could no longer be used for biological measurements, such as survival and growth. Nevertheless, in an attempt to salvage information from these exposure chambers, they were converted into and designated as chemistry-only replicates (see collection of water and pore water section in methods).

Statistics
Analytical data were categorized based upon sample type, including overlying water, sediment-water interface, and pore water, treatment type, such as sediment source at LALL, GE, DE, NP, LD, UMF, or LMF, tank replicate, and measurement type, including acid-extractable and dissolved metal. These categories allowed for explicit and systematic processing of data to quantify and evaluate a wide-range of exposure conditions throughout the duration of the study.
Concentrations were summarized on the basis of sources of sediments, replicate tank, and depth within exposure chambers. In addition, techniques for collecting samples of water were factored into the analysis. As noted in sections 2.3, as many as three distinct sampling techniques, such as suction, peepers, and DGT probes, were employed to characterize sediment-water interface and pore water samples. Data for each of these sampling techniques were used to quantify concentrations of various analytes within exposure chambers.
In some instances, concentrations of metals were less than the Limit of Quantification (LOQ) and thus, resulted in non-detectable concentrations. Given the relatively large number of non-detectable concentrations observed during the present study, summary statistics and statistical comparisons and correlations were determined using maximum likelihood estimate (MLE) procedures. MLE procedures consider the presence of non-detectable concentrations when estimating parameters such as the mean, median, and variance for a given dataset. Procedures such as MLE provide better estimates of statistics for censored data, for samples for which the concentration is less than the limit of detection than simple "blind" calculations that treat BDLs as detected measurements or 'fabricating' values with the use of substitution methods, such as one-half the value of the detection limit [33]. A detailed description and reasoning for use of the MLE procedure employed in the present study is provided in Supplemental Materials. Conformation of distributions of data to approximate the normal probability function and equal variances was assessed by use of box and probability plots. If parametric assumptions were met, statistical comparisons of concentrations of metals within the different matrices of exposure chambers containing site sediments versus reference sediments were calculated using analysis of variance (ANOVA) and Dunnett's equivalents following MLE procedures. If parametric assumptions were violated, a Wilcoxon Score Test was performed following MLE procedures. When appropriate, statistical significance was adjusted with a Bonferroni correction factor. In addition, to investigate relationships between sampling techniques within a matrix, and concentrations of metal in sediment to concentrations in pore water and overlying water, linear regression and correlation coefficients were used following MLE procedures. Linear regression analyses were performed on measurements obtained from active sampling techniques, such as acid-extractable metals in whole sediments and sampling of matrices through suction. All data analysis procedures were conducted with R version 2.9. 1 [34].

Characterization of sediments
Despite the relatively large sampling area, the presence of coarsegrained substrata, such as gravels, cobbles, and boulders, and an armored riverbed made collecting sufficiently large volumes of sediment difficult in some locations. Sediments were collected from NP, LD, UMF, LMF, LALL and GE, with volumes at LD, NP, and LMF being less than the targeted 600 L (Supplemental Material). Insufficient volumes of sediment were retrieved from CB, DE, and BBE. Given the difficulty in collecting sufficient volumes of site sediments, substrata collected above the water line from the gravel bar at DE were incorporated into this study, as directed by EPA.
With locations where sediment was collected targeted in areas north of RM-703 that contain white sturgeon spawning-and/or nursing-grounds, it is reasonable that predominant sediment grain sizes collected and evaluated for the present study were sand-sized particles, having diameters ranging from very coarse (1 < 2.0 mm) to very fine (62.5 < 125 μm) sands. The mean grain size distribution of site sediments was approximately 0.5 % gravels, 97.3 % sands, and 1.9 % silts/clays (Supplemental Material). Reference sediments were slightly coarser with a mean grain size distribution of 20.9 % gravels, 76.5 % sands, and 1.4 % silts/clays.
Since the primary focus of this study was to assess toxicity of early life stages of white sturgeon, under laboratory controlled conditions, to a gradient of COPC associated with sediments, with a primary focus on those COPC commonly associated with granulated slag, the target metals were Cd, Cu, Pb and Zn. The analyses presented here focus on these four metals. However, data for all COPC in whole sediment are presented in Supplemental Materials. Sediment concentrations of Cd, Cu, Pb and Zn were significantly greater (p < 0.01) in all site sediment samples than in reference sediments (Supplemental Materials).
Concentrations of metals spanned the spectrum of concentrations observed to date within the site, and often exceeded the 90 th centile of previously reported data ( Figure 1) [11,12]. Similarly, concentrations within reference sediments were lesser, often less than the 10 th centile of site sediments. Based on these results, sediments evaluated for this study appear representative and consistent of the range of concentrations observed with site sediments.

Characterization of water samples
Major cation/anion water quality conditions: Concentrations for major cations/anions, including calcium (Ca ++ ), potassium (K + ), sodium (Na + ), and sulfate  ) within overlying water and pore water (at 2.5 cm depth) were consistent between treatments and for the duration of the study (Supplemental Material). Unlike major cations/ anions, concentrations of DOC were more variable among treatments (e.g., LALL and DE) and greater within pore waters than overlying waters. A significant number of measured concentrations of DOC were qualified as estimated due to field duplicate imprecisions. As a result, summary statistics for reported DOC concentrations were corrected for these imprecisions by factoring in any reported blank contamination (see application of biotic ligand model section in methods in [24]) Concentrations of major cations/anions in water collected from the sediment-water interface measured by use of suction (pipette) and peeper techniques were comparable, with concentrations calculated by use of DGT probes being different (Supplemental Material). Given that DGT probes used in the present study were specifically designed and deployed to measure the flux of the four primary metals of interest (Cu, Cd, Pb, and Zn), calculated concentrations of major cations for the DGTs was likely due to saturation of the resin in the DGTs. Because diffusion coefficients are comparable in magnitude, fluxes of the cations comprising the hardness are on the order of 2,000 to 20,000 times greater than typical trace metal fluxes, the potential existed for the resin to become saturated over a time scale of hours rather than days [35,36]. Data for pore water quality at the 1 cm depth was limited because samples were collected via peepers and DGT probes only.

Dissolved concentrations of target metals
Copper: There were significant differences (p < 0.01) in concentrations of dissolved Cu between exposure chambers containing site sediments versus reference sediments ( Figure 2; Supplemental Materials). Dissolved concentrations of Cu in negative controls (H 2 O and CTRL) and reference sediment (LALL and GE) exposure chambers were consistently lesser for the duration of the study with estimated median concentrations ≤ 1 μg/L for all matrices, including overlying water, sediment-water interface, and pore water. In comparison, concentrations of Cu measured in samples containing UCR site sediments were significantly greater (p ≤ 0.001) in all matrices compared to those of reference sediments, with the exception of a few measurements from passive sampling devices in DE and LMF sediment, and active sampling at the sediment-water interface in NP sediments (Supplemental Materials). The greatest concentrations of Cu were observed in pore water collected at a depth of 2.5 cm. In exposure chambers containing DE, LD, and UMF sediments median concentrations of 30, 20, and 10 μg Cu/L, respectively were measured in pore waters ( Figure 2). In contrast, estimated median concentrations of Cu in pore water at 2.5 cm for exposure chambers containing sediments from NP and LMF approached 3 and 1 μg/L, respectively. Concentrations of Cu in pore water collected at a depth of 1 cm were lesser than those samples collected at 2.5 cm for exposure chambers containing DE, LD, and UMF sediments, with estimated median values < 4 μg/L. Concentrations of Cu were comparable between pore water depths for exposure chambers containing NP and LMF sediments. In overlying water and at the sediment-water interface, estimated median concentrations of Cu in samples from exposure chambers containing site sediments (0.8-2 μg/L) were equal to or slightly greater than the estimated median concentrations of Cu in control and reference sediments (<1 μg/L).
All sampling devices used to extract samples from the sediment-water interface or from pore water were in agreement for concentrations of Cu ( Figure 2). Concentrations of Cu measured in pore water at 1 cm below the sediment surface using peepers and DGT probes were significantly correlated (p<0.001) and not statistically different (p > 0.05; Supplemental Materials). Similarly, concentrations of Cu measured at the sediment-water interface were significantly correlated (p < 0.05) among the three sampling techniques, including pipette suction, peeper, and DGT probes, and not statistically different (p > 0.05) between peeper and DGT techniques, and DGT and suction techniques (Supplemental Materials), with measurements from DGT probes having the greatest variability. Differences between concentrations, achieved with any of the aforementioned sampling devices, appeared to be small and random, which suggests that any of these sampling methods could be used to characterize concentrations of Cu in pore water and water at the sediment-water interface.
Results of linear regression indicated that in all site sediment samples there was a significant positive relationship between concentrations of Cu in sediment to concentrations in pore water and overlying water.
This indicated that acid-extractable concentrations of Cu in sediment were a reasonable predictor of concentrations of Cu in pore water and could prove to be useful in assessing bioavailability and risk (DE r 2 = 0.85; NP r 2 = 0.82; LD r 2 = 0.92; UMF r 2 = 0.91; LMF r 2 = 0.77). Zinc: Differences in median concentrations of Zn among all exposure chambers, regardless of sediment source, were relatively small ( Figure 3). Estimated concentrations of Zn in overlying water in negative control (H 2 O and CTRL) and reference sediment (LALL and GE) exposure chambers were < 6 μg/L. Concentrations of Zn in overlying water measured in exposure chambers containing site sediments had slightly greater concentrations with estimates ranging between 6 and 15 μg/L and were considered statistically different (p <0.001) from GE reference sediment but not LALL reference sediment, with the exception of LMF and LALL comparison (p ≤ 0.001; Supplemental Materials). Differences in concentrations of Zn at the sediment-water interface were lesser between negative controls, references, and site sediments. However, concentrations of Zn measured in water at the sediment-water interface with passive sampling devices in all site sediments were statistically greater (p ≤ 0.001) from LALL reference      Figure 2: Concentrations of dissolved copper (Cu) as a function of treatment and sample type. Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2nd panel), pore water at 1 cm (3rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles and color represents replicates, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25th and 75th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75th or below the 25th centiles. Treatments included negative controls with water only (H2O) and artificial sediment (CTRL), reference sediments from Lower Arrow Lake (LALL) and Genelle (GE), and site sediments from Deadman's Eddy sediments. Median concentrations of Zn in pore water (at 1 cm) from exposure chambers containing site sediments ranged between 3 and < 70 μg/L (Figure 3), and concentrations varied within sites depending on sampling technique, with the greatest concentrations measured in DGT probes. Regardless, median concentrations of Zn in pore water (at 1 cm) were slightly greater than those of the negative control and/ or reference pore water (at 1 cm) concentrations (3 -15 μg/L). All concentrations of Zn in pore water (at 1 cm) in site sediments were statistically greater (p ≤ 0.001) than concentrations of Zn in pore water (at 1 cm) of one or more reference sediments (Supplemental Materials). Concentrations of Zn for pore waters collected at 2.5 cm fluctuated within sites and across all exposure chambers and ranged from 1 to 1000 μg/L. However, values were largely qualified, primarily due to contamination in blanks (Figure 3), and results should be interpreted with caution. Zn is ubiquitous in the environment and in the laboratory and the exact sources of contamination in the present study are unknown.
The different sampling methods used to quantify Zn in pore water or at the sediment-water interface often resulted in similar median measurements among techniques ( Figure 3). Concentrations of Zn measured in pore water at 1 cm below the sediment surface using peepers and DGT probes were significantly correlated (p < 0.001) and not statistically different (p > 0.05; Supplemental Materials). At the sediment-water interface, only concentrations of Zn measured in peepers and DGT probes were significantly correlated (p ≤ 0.05), but were not statistically similar (p < 0.05) with a tendency for measurements from peepers to be slightly less than from DGTs and suction devices. Overall, the sampling methods used to characterize concentrations of Zn in pore water appear to be more interchangeable than methods used to characterize concentrations of Zn at the sediment-water interface.
Results of linear regression indicated that in all site sediments there was a significant positive relationship between concentrations of Zn in sediment to concentrations in pore water and overlying water. However, r 2 values were ≤ 0.50 in all site sediments, indicating that concentrations of Zn in sediment are a poor predictor of concentrations of Zn in pore water (DE r 2 = 0.49; NP r 2 = 0.33; LD r 2 = 0.50; UMF r 2 = 0.40; LMF r 2 = 0.27).
Cadmium: Median concentrations of Cd within all exposure chambers, including site sediments, for all sample types, including overlying water, sediment-water interface, and pore water, and all sampling techniques were generally ≤ 0.1 μg/L ( Figure 4). Concentrations of Cd in Pore water at 2.5 cm depth in exposure chambers with site sediments were significantly greater (p ≤ 0.001) than exposure chambers with a reference sediment (Supplemental Materials). Specifically, median concentrations of Cd (0.2, 0.07, and 0.15 μg/L) in exposure chambers with DE, NP, and UMF sediments, respectively, were significantly greater than those of the negative control and reference sediment exposure chambers (0.02 -0.04 μg Cd/L). Based on measurements from active sampling devices, concentrations of Cd in overlying water, pore water, and at the sediment-water interface in exposure chambers with DE and UMF sediments were significantly greater than exposure chambers with reference sediment. In contrast, there were no statistical differences in concentrations of Cd between all site sediments and reference sediments in pore water or at the sediment-water interface based on peeper measurements.
Of the three sampling devices used to measure sediment-water interface samples, measurements with DGTs and suction devices demonstrated relatively good agreement for concentrations of Cd and were significantly correlated (p ≤ 0.05), although statistically different, while measurements made with peepers were consistently greater ( Figure 4, Supplemental Material). Likewise, measurements in pore water made with peepers at 1 cm depth demonstrated poor correlation and were consistently greater for concentrations of Cd than measurements of pore water with DGTs at the same depth. Based on these results, DGTs, peepers, and suction devices do not appear to be interchangeable methods for measuring concentrations of Cd.
Results of linear regression indicated that in all site sediment samples there was a significant positive relationship between concentrations of Cd in sediment to concentrations in pore water and overlying water. Only in certain site sediment samples, however, were concentrations of Cd in sediment moderate predictors of concentrations of Cd in pore water (DE r 2 = 0.65; NP r 2 = 0.71; LD r 2 = 0.68; UMF r 2 = 0.63; LMF r 2 = 0.50).
Lead: Median concentrations of Pb within all exposure chambers, including site sediments, for all sample types, such as overlying water, sediment-water interface, and pore water, and all sampling techniques were generally < 1.0 μg/L ( Figure 5). Differences in estimated median concentrations of Pb within the different matrices were relatively small between negative controls, reference sediments, and site sediments. However, concentrations of Pb in all site sediments were significantly greater (p ≤ 0.001) in overlying water and at the sedimentwater interface than in reference sediment, based on measurements from active sampling techniques (Supplemental Materials). Also, concentrations of Pb in pore water measured in samples collected by active sampling (suction at 2.5 cm depth) in exposure chambers containing DE, UMF, and LMF sediments were significantly greater than those of reference sediments. In contrast, there were no significant differences in concentrations of Pb in pore water or at the sediment-water interface in DE, LD, or UMF sediments compared to reference sediments, based on passive sampling techniques. There were no significant correlations among concentrations of Pb among different sampling techniques utilized in pore water or sediment-water interface measurements. The different sampling techniques within individual exposure chambers measured different concentrations of Pb at the sediment-water interface, with a tendency towards elevated concentrations of Pb in samples from peepers in comparison to DGTs and suction devices. A similar tendency towards greater concentrations of Pb measured in pore water at 1 cm with peepers versus DGTs was also observed. Based on these results, DGTs, peepers, and suction devices do not appear to be interchangeable methods for measuring concentrations of Pb. Results of linear regression indicated that in all site sediment samples there was a significant positive relationship between concentrations of Pb in sediment to concentrations in pore water and overlying water. Concentrations of Pb in sediment were moderate predictors of concentrations of Pb in pore water (DE r 2 = 0.65; NP r 2 = 0.71; LD r 2 = 0.58; UMF r 2 = 0.70; LMF r 2 = 0.64).
Other metals: There were no major differences in concentrations of the additional metals analyzed in pore water, overlying water, or at the sediment-water interface, between exposure chambers containing site sediments and reference sediments, except for antimony (Supplemental Materials). Estimated median concentrations of antimony in site sediment treatment groups were slightly elevated in overlying water and at the sediment-water interface, whereas concentrations in pore water differed significantly (p ≤ 0.001) in exposure chambers containing site sediments (0.1 -100 μg/L) compared to reference sediments (< 0.5 μg/L).    Figure 3: Concentrations of dissolved zinc (Zn) as a function of treatment and sample type. Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2nd panel), pore water at 1 cm (3rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles and color represents replicates, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25th and 75th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75th or below the 25th centiles. Treatments included negative controls with water only (H2O) and artificial sediment (CTRL), reference sediments from Lower Arrow Lake (LALL) and Genelle (GE), and site sediments from Deadman's Eddy ( Figure 4: Concentrations of dissolved cadmium (Cd) as a function of treatment and sample type. Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2nd panel), pore water at 1 cm (3rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles and color represents replicates, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25th and 75th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75th or below the 25th centiles. Treatments included negative controls with water only (H2O) and artificial sediment (CTRL), reference sediments from Lower Arrow Lake (LALL) and Genelle (GE), and site sediments from Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2nd panel), pore water at 1 cm (3rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles and color represents replicates, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25th and 75th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75th or below the 25th centiles. Treatments included negative controls with water only (H2O) and artificial sediment (CTRL), reference sediments from Lower Arrow Lake (LALL) and Genelle (GE), and site sediments from Deadman's Eddy (DE), Northport (NP), Little Dallas (LD), Upper Marcus Flats (UMF) and Lower Marcus Flats (LMF).

Comparisons of sampling techniques
There were differences in measured concentrations of metal among sampling techniques within a matrix for most of the metals analyzed. Considering all the TAL metals, concentrations of metals from DGTs and suction devices tended to be similar at the sediment-water interface whereas measurements from peepers were often greater in comparison. Similarly, in pore water at 1 cm depth, concentrations of metals in water in peepers were often greater than those estimated by DGTs. This trend could, in part, be due to the fact that concentrations in peepers represent metals averaged over time, once equilibrium has been established with the surrounding matrix, whereas suction devices measure point source concentrations of metals. Alternatively, assuming that the metal that is removed from a matrix is rapidly resupplied from the solid phase in sediment, DGTs characterize fluxes of metals and provide a measurement of average concentrations of metals during deployment time within a matrix [37]. Previous studies have investigated issues with re-supply and have found that in cases where it is insufficient, DGT measurements can result in calculated concentrations of metals that are significantly less than concentrations of metals measured in pore water using other, more traditional techniques, such as core sampling and centrifugation of sediments [38][39][40]. For certain metals in the present study, however, such as Cu, Zn, and Al, concentrations of metals in water at the sediment-water interface and pore water at 1 cm depth were relatively similar between these two methods of sampling.
Use of air stones for extraction of pore water has several advantages over peepers and DGTs, including the ability to extract relatively large sample volumes that in turn allow for a greater suite of chemical analyses. In addition, air stones can be used for more than one sampling event without having to disturb the sediment. Peepers and DGTs, however, can sample more than one matrix at a time and can be used in the laboratory or in the field.

Conclusions
Sediments investigated during the present study covered a range of concentrations of targeted metals that were representative and consistent of the extent of concentrations observed with UCR site sediments, and captured the upper concentration range of previously reported data. In the present study, acid-extractable concentrations of Cu, Cd, Zn, and Pb in whole samples of site sediments were significantly greater than those in reference sediments. Of the four primary metals of concern, Cu, Cd, Pb, and Zn, concentrations of Cu, primarily in pore water, were greatest in exposure chambers containing site sediments compared to those in reference sediments. Concentrations of Cu and Zn were significantly greater in matrices of site sediments compared to reference sediments more often than Cd and Pb. Active and passive sampling techniques resulted in similar estimates of concentrations of Cu in pore water and at the sediment-water interface, and comparable measurements for concentrations of Zn. In contrast, sampling techniques resulted in dissimilar measurements for concentrations of Pb and Cd, with a tendency towards greater concentrations in peepers compared to DGTs in both pore water and at the sediment-water interface. Discrepancies in measurements of concentrations of metals between peepers and DGTs were a common trend for a number of the non-target metals analyzed in the present study. At the sediment-water interface, however, DGTs and active sampling through suction often resulted in relatively similar measurements for many of the metals.
Bioavailability of metals is a key factor to assessing risk of exposure to metals associated with sediment. In general, concentrations of dissolved metals in pore water are considered to be the most representative of the fraction of metal bioavailable to sediment-dwelling organisms. With contaminated sediments, there is a concern that metals might leach out of sediments into pore water at concentrations sufficient to have adverse effects on organisms associated with the sediments. When considering effects of metals associated with sediments on benthic dwelling fish such as sturgeon, matrices other than pore water might be a more applicable route of exposure. Therefore, the present study characterized concentrations of metals in various matrices associated with sediment. Concentrations of metals in sediment often resulted in significant concentrations of metals in pore water and overlying water; although results varied depending on the sampling technique employed. When active sampling through suction was considered, linear regression was a relatively effective means of characterizing the movement of metal in sediment to pore water and overlying water for certain metals such as Cu. However, the present study demonstrated that measured concentrations of dissolved metal in matrices associated with sediment can differ depending on the choice of sampling method, matrix, and analyte. Differences in estimated concentrations of metals among the methods applied highlights the difficulty in assessing the true risk of exposure to metals associated with sediments. To minimize uncertainty, concentrations of metals could be measured in bodies of organisms, but this is difficult under field conditions. Regardless of these differences among methods of sampling, concentrations of Cu, Cd, Pb, and Zn calculated from any of the techniques employed to sample water at the sediment-water interface or in overlying water in the present study were less than the EPA national criteria of 6.0 µg/L, 0.174 µg/L, 1.46 µg/L, and 78 µg/L, respectively, and criteria for the state of Washington for the protection of aquatic life of 7.4 µg/L, 1.51 µg/L, 1.46 µg/L, and 69 µg/L, respectively. These criteria do not take into account the bioavailable fraction of metals, except for EPA criteria for Cu that utilizes the Biotic Ligand Model. Therefore, when considering pore water, interpretations are more difficult since water quality characteristics, such as hardness and DOC, were more variable among sediments and affect site-specific water quality standards. In addition, questions as to whether or not benthic fish such sturgeon are exposed to metals associated with pore water become important. A more appropriate comparison would be concentrations corrected for chemical activity, and binding to inorganic and organic ligands by use of the Biotic Ligand Model. The analytical data reported herein are utilized in a parallel article to characterize risk, and in turn, compare predictions to the biological results from exposure of early life stage white sturgeon to metals in sediments of the Upper Columbia River [24].

SUPPLEMENTAL MATERIALS
Assessment of Columbia River sediment toxicity to white sturgeon: concentrations of metals in sediment, pore water, and overlying water

Method Development Work
Prior to the assessment of toxicity of metals associated with sediment in the Columbia River to early life stages of white sturgeon, methods were developed and evaluated of a flowthrough fluvial simulation system at the University of Saskatchewan's (UofS) Aquatic Toxicology Research Facility (ATRF) that was specifically designed for use in studies of toxicity of sediments to early life stages of white sturgeon. The primary goals of the method development were to evaluate and confirm the performance of the flow-through fluvial exposure system and to establish suitable control sediments. Specifically, an experimental exposure system was needed to allow for adjustment of flow velocity, water replacement time, recirculation frequency, sediment thickness, and pore water sampling volume and depth. The method development work was aimed at establishing uniform flow conditions to minimize "dead spaces" at the inflow and outflow of the exposure chambers, to establish which hydrological operating conditions (i.e. flow) would result in the smallest gradient between pore water and overlying water, and what was the time to steady-state. In addition, the effects of different volumes and distributions of substratum on hydrological conditions in the chambers were examined, as well as the influence of sediment depth, depth of pore water sampling within the sediment layer, and sampled volume on exchange between overlying water and pore water. Optimum techniques for homogenization of sediment, sampling of pore water and system cleaning techniques were also investigated.
Results from methods development work and associated final study design are presented herein and summarized (Table A1). Objective: Confirm the effectiveness and reliability of sediment homogenization procedures

Experimental Design
Samples collected from the gravel bar at Deadman's Eddy were mixed and homogenized in a specially designed 'concrete mixer'. The large rotating drum of the mixer contained a plastic liner that had been tested to confirm lack of leaching of metals into a water rinsate. Composited sediment was tumbled for a period long enough (e.g., hours) to create a visual appearance of complete mixing. Two sediment samples each were taken from the top, middle, and bottom layers of the drum and analyzed for Cu, Cd, Pd, and Zn to verify the visual determination of homogenized sediment. If analyses had greater than ± 20 percent maximum calculated difference in concentration, then the sample was tumbled for another period and the analysis repeated.

Analysis of metal concentrations was conducted by Inductively Coupled Plasma-Mass
Spectrometry (ICP-MS) at the University of Saskatchewan, Saskatoon, SK, Canada, following methods outlined by Creed et al. (1994).

Decision Criteria
Sediments were determined to be completely homogenized when all six samples collected were within ± 20 percent the maximum calculated difference. Photographs of homogenized sediment were taken to document the visual appearance of samples at homogeneity.

Homogenization:
Two sets of six samples were analyzed for homogeneity, for a total of 12 data points for each metal of concern. The decision criteria for sediment homogenization stated that the 6 samples, two from the top, middle and bottom of the mixer, were to be analyzed for Cu, Zn, Cd, and Pb and that concentrations ≤ 20% of the mean of the six samples would be considered acceptable. Cu, Zn and Cd results all were well below the 20% criterion, whereas Pb results were not. When comparing all 12 samples, Pb exceeds 20% 6 times (50%, 48%, 26%, 33%, 33% and 21%).

Conclusions
Results for Cu, Cd and Zn indicated that the sediment achieved homogeneity after three hours mixing time. The cement mixer appeared to be an effective method of mixing sediment and it was used for mixing sediments for the definitive study. Samples were tumbled for 3 hours, stopping every half an hour to rotate the drum and scrap any sediment from the sides before repeating the process.
The probable cause for the greater variation in concentrations of Pb among subsamples than was observed for Cu, Zn or Cd, is that Pb was likely present in part as larger solid particles as opposed to associated with fines as surface oxides and bound to organic ligands. Since concentrations of Cu, Zn and Cd all indicated that the bulk sediment was homogenized, the greater variation in concentrations of Pb was likely due to the size of the sub-subsample collected for quantification. A sub-subsample of 0.1 g dry weight (dw) was taken for digestion. If a larger sub-subsample had been collected and digested, and the digestate diluted prior to analysis, it is likely that the variation among sub samples would be less. Since Pb is generally occluded in the solid matrix of particles it is less likely to contribute directly to toxicity in pore water. This conclusion is supported by the results reported by Besser et al. (2008) for location L7 in the Columbia River, which is in the vicinity of Deadman's Eddy. In that study of the toxic potential of metals in sediments from the UCR, the concentration of Pb in bulk sediment from L7 was 590 μg Pb/g dw while the concentration in the pore water was 7.1 μg Pb/L. At other locations such as L6, which is downstream from L7, the concentration in the bulk sediment was 200 μg Pb/g, dw but the concentration in the pore water was 250 μg Pb/L. A fractionation of the sediment at L7 showed that more than 50% of the Pb was in the residual unextractable fraction and an additional 40% was present as a sulfide (or organic), but most likely an insoluble sulfide material. These results taken together indicate that Pb in the vicinity of Deadman's Eddy is more likely to be bound in the matrix of the sediments and less likely to contribute to lead concentrations in pore water. The volume of bulk sediment added to the experimental systems was such that the variation observed in very small sub-subsamples will not be observed among exposure systems. It can be concluded based on the results for Cu, Cd and Zn that the sediments were sufficiently homogenized to be used in further experimentation. Following mixing, sediments were placed into 5 gallon high density polyethylene (HDPE) buckets, overlaid with water, and stored at 4°C under a nitrogen atmosphere until placement into the fluvial system chambers.

Order 2: Experimental exposure systems
Objective: Establish an appropriate exposure system to test potential toxicity of sediments to early life stage white sturgeon

Experimental Design
Test systems comparable to those used in the present study had been previously employed at the UofS ATRF in chronic experiments with white sturgeon exposed to aqueous metals (Vardy et al. 2011). However, the test systems had not been used or specifically tested for the purpose of conducting flow-through sediment toxicity tests. Specifically, parameters associated with the design, such as fluvial chamber dimensions and layout, location of sampling devices, delivery, re-circulation, and chilling of water, and other operational conditions were established and tested to inform the definitive study design.

Results
Different designs and various prototypes of exposure chambers were tested. For the definitive study, artificial flow-through exposure chambers were constructed from high density polyethylene (HDPE) and screens were fabricated from plexi-glass with fiberglass mesh ( Figure   A1). A head tank contained the reverse osmosis and dechlorinated lab water mixture supplied to the experimental exposure systems ( Figure A2). The mixture was delivered from the head tank to the 85-L exposure system reservoir via a metering pump. The mixture delivery rate from the reservoir to the exposure chambers was regulated by a delivery manifold attached to a recirculating march pump. The mixture flowed from the delivery manifold at the exposure chamber inflow, through the exposure chamber, and exited through outflow drain holes, which in turn were connected back to the 85-L reservoir. There was an overflow drain at the back of each reservoir that discarded wastewater and a baffle within each reservoir that prevented shortcircuiting of the inflow to the overflow drain. The water was cooled to the desired temperature by placing a chiller unit inside the 85-L reservoir. The delivery manifold at the inflow of the exposure chamber and four drains with ball valves at the outflow of the exposure chamber allowed for adjustments to flow regime. Ports were built into the side of the exposure chambers to connect airstones for pore water sampling.

Conclusions
An experimental exposure system was established that allowed for adjustment of flow velocity, water replacement time, recirculation frequency, sediment thickness, and pore water sampling volume and depth.

Experimental Design
A fluorescent dye (Fluorescein) was used to measure water flow; such dyes are cost effective and easily and accurately measured with a fluorometer and observed with an ultraviolet (UV) lamp. Flow-rates ranging from 5 to 25 L/min were tested in duplicate. After the dye was introduced into the test chamber (t = 0), it was made visible by UV lighting, and dispersal of dye and associated water flows were recorded by means of a digital video camera across the entire chamber. Additionally, at flow rates that appeared acceptable as gauged against the goal of this experiment (i.e., ≥17 L/min), water samples were taken at t = 10 sec (intake), t = 20 sec (middle) and t = 30 sec (outflow) at 3 locations equally distributed over the cross-section of the chamber.
This was repeated for cross-sections close to the inflow, centre, and outflow of the test chamber, resulting in a total of 3 x 9 = 27 water samples. The first and last sampled cross-sections were located at the inflow and outflow screens of the sediment exposure chamber, respectively, to identify potential dead spaces (see Figure A3). Samples obtained during the second experiment (flow-rates ≥17 L/min) were analyzed for dye concentrations using a microtiter plate fluorescent reader (Polastar Optima, BMG Labtech, Offenburg, Germany), and dye concentrations were mapped throughout the chamber. Sampling was conducted using 10 mL pipettes modified such that samples could be taken at different depths throughout the chambers. Dye concentration experiments during which water samples for fluorometer quantification were collected were run either in two (17 and 25 L/min) or three (20 L/min) replicates.

Data Presentation
Descriptive statistics (mean ± SD) of relative dye intensities. Fluorescence measurements are expressed relative to the maximum fluorescence intensity (FLU) measured during each experiment (FLU/maximum FLU). Assessment and evaluation of water flow conditions within the exposure chamber was also based on visual observations and records (e.g., video).

Results
The dye experiment revealed a significant impact of flow rate on the uniformity of flow throughout the test chamber (Table A2)

Conclusions
Stable and homogenous flows were achieved at flow rates greater than 17 L/min. It was therefore recommended to initiate the study with a flow rate of 20 L/min to accommodate low flow requirement for yolksac larvae, and then increase flow rates to 25 L/min when larvae initiated exogenous feeding, and were large enough to easily withstand the increased flows. The proposed flow-rates resulted in ground velocities that were less than those occurring in the UCR, and did not impact sediments layered into the chambers (e.g. causing re-suspension).   Cross sectional position in exposure system (viewed from inflow end of test chamber): 1=left; 2=centre; 3=right. Values are expressed relative to the maximum dye concentration measured in the same experiment (maximum dye concentration = 1).

Order 4: Determination of optimum stone volume and distribution in fluvial test chambers
Objective: Indentify the effect of different volumes and distribution of stones on hydrological conditions in fluvial chambers.

Stone volume selection
Stones (Geosubstrate # 12422, Hagen) (10 to 20 mm diameter) were placed in exposure chambers without sediment to visually assess optimal spatial densities. Spatial densities of stones tested included: 0, 3, 5 and 7 stones per 100 cm 2 . Tests with densities greater than 7 stones per 100 cm 2 were not included because of crowding issues.

Water quality evaluation
Conductivity measurements were taken at 0, 24 and 48 hrs in the 1 cm water layer overlying the stones to assess variation in water quality. Stone densities of 0, 3, 5 and 7 stones per 100 cm 2 were tested in exposure chambers without sediment while stone densities of 0, 4 and 7 stones per 100 cm 2 were tested in exposure chambers with sediment. Sediments were layered at a thickness of 2 and 3 inches into the exposure portion of the test chambers and stones were placed on top. Sediment depths greater than 3 inches were not included based on the evaluation of optimum sediment depths; refer to summary Table A1.

Flow condition evaluation
Flow condition was evaluated visually by use of fluorescent dye to determine if the stones altered the flow of water.

Decision Criteria
Conductivity measurements to evaluate variation in water quality: target value of <30% variation among samples taken. Visual assessment of flow conditions using fluorescent dye.

Stone volume selection
Visual assessment of stone densities of 3 and 7 stones per 100 cm 2 appeared too sparse and too crowded, respectively, whereas stone densities of 4 and 5 per 100 cm 2 appeared appropriate leaving sufficient room to enable observations of sediments while providing appropriate refuge for white sturgeon early life stages ( Figure A6).

Water quality evaluation
Conductivity analysis revealed less than 15% variation among samples taken from all stone densities at 0, 24 and 48 hrs in exposure systems with and without sediment ( Figure A7 and 8). Variation in conductivity was less than 10% in exposure systems with sediment and stone densities of 4 and 7 stones per 100 cm 2 ( Figure A8).

Flow condition evaluation
Visual assessments of flow condition revealed no impact on water flow with the incorporation of stones into the exposure chamber or with increased stone densities in chambers with or without sediment.

Conclusions
Addition of stones to the fluvial exposure chambers had no impact on flow conditions and little to no impact on water quality. In our experience early life stages of white sturgeon appear less stressed when provided a refuge under experimental conditions. A density of 4 stones per 100 cm 2 was recommended as the optimal loading density. Figure A6. Photograph of exposure system with stone density of 4 per 100 cm 2 with sediment for methods development of Upper Columbia River sediment toxicity tests. Figure A7. Variation in conductivity in water-only exposure systems over 48 hrs at stone densities of 0, 3, 5 and 7 stones per 100 cm 2 for methods development of Upper Columbia River sediment toxicity tests. Figure A8. Variation in conductivity at 2" sediment depth over 48hrs with 0 and 4 stones per 100cm 2 (A) and variation in conductivity at 3" sediment depth over 48hrs with 0 and 7 stones per 100cm 2 (B) for methods development of Upper Columbia River sediment toxicity tests.

Order 5 & 6:
Assessment of pore water sampling techniques and sediment depth using suction devices Objective: Establish optimum methods and conditions for sampling pore water. cat. #BW10) were tested regarding their utility to sample water by means of suction through a 10 Deadspace in airstones were minimized by inserting an acrylic rod into the hollow core. A selection of experiments were video-documented.

Static water tests with sediment at various depths using ceramic airstones:
This test was conducted in 10 L aquaria with two laboratory control sediments. Loose gravel (2.0 -5.0 mm, geosystems substrate, #12418, hagen) was placed into test chambers at depths of 1" (inch), 2", and 3", while finer sediment (1.0 -2.0 mm, geosystems substrate, #12648, hagen) was placed into test chambers at depths of 1", 2", 3", and 4". Airstones were placed at the bottom of the aquarium. Dye was introduced to the surface water and the amount of water that could be removed by the airstone before the introduction of dye was recorded. The presence of dye was measured visually by use of a black light.

Flow through tests with sediment at various depths using ceramic airstones:
Sediments were layered at a thickness of 1, 1.5, 2, and 3 inches into the exposure portion of the fluvial test chambers. Up to sixteen pore water sampling ports were equally distributed throughout the exposure chambers at a depth of 1 and 1.5 inches below the sediment surface ( Figure A9). This represents a slight deviation from the originally proposed design that asked for testing sediment depths of 2, 3, and 4 inches, and for sampling depths of 0.5 and 1 inch. It was not possible to install sampling devices at a sampling depth of 0.5 inches due to exposure of portions of the sampling device to the overlying water. Also, it was determined as part of the experiment "Static water tests with sediment at various depths using ceramic airstones" that sediment depth greater than 3 inches did not proportionally improve overall recoverable sample volumes and resulted in sufficient volumes (>350 mL), and thus, it was decided to focus on the characterization of lesser sediment depth. Furthermore, the experiment using 2 inches of sediment depth was conducted both in absence and presence of gravel in accordance with the density determined in order #4 (4 stones per 100 cm 2 ). All other initial experiments were conducted without addition of gravel. After introduction of dye into the test system, pore water was sampled until dye was visible in the sampling device (10 mL syringe), and the volume sampled was recorded for each airstone. These initial experiments were followed by a second series of tests where 2 and 3 inches of sediment were layered into the fluvial test chambers, and which were supplemented with gravel as described above and in order #4. In these experiments airstones were installed at a depth of 1.5 inches, and after the introduction of dye pore water was sampled as described for the initial experiment with the difference that in addition to the visual assessment a sub-sample of the extracted pore water was subjected to fluorescence determination by use of a microtiter plate fluorescent reader (Polastar Optima, BMG Labtech, Offenburg, Germany). Samples for dye concentration determination were collected at 10 mL intervals up to 110 mL of total volume sampled (= 11 samples). All experiments described within this section were conducted at flow rates of 25 L/min.

Descriptive stats (means +/-SD). Comparison of mean values using parametric statistics (Student's t-test). Visual assessment of the presence of fluorescence based on video records and
other visual documentation, as appropriate.

Pore water sampling device selection
Static water only test: Of the three airstones tested the Plax Aqua-Mist 6" (sandstone) did not allow for pulling significant amounts of water (≥ 30 mL). While it was possible to retrieve the required amounts of water (by suction from the Penn-Plax Bubble Wall 10" (paper) device, this airstone had a strong preference for pulling water from the front ½ of the airstone's length as demonstrated by the dye test. The ceramic airstone (Rena Micro Bubble) appeared to pull water evenly along the whole surface of the airstone, and therefore, was selected for further evaluation. All assessments were made by means of visual inspection of the presence or absence of dye.

Static water tests with sediment at various depths using ceramic airstones:
Volumes of pore water that could be sampled without incorporation of overlying water increased significantly (p=0.001) with sediment depths greater than 1 inch. 20 mL of pore water was retrieved in 1.0 inch of loose gravel sediment depth compared to 130 mL in 2.0 inch of loose gravel sediment depth. 230 mL of pore water was retrieved in 3.0 inch of loose gravel sediment depth. Greater pore water volumes were sampled in equivalent depths of the fine sediment and averages ranged from 30 mL pore water retrieved in 1.0 inch of fine sediment depth compared to 193 mL in 2.0 inch of fine sediment depth. Average volumes of 320 mL and 870 mL of pore water were retrieved from 3.0 inch and 4.0 inch depths of fine sediment, respectively.

Flow through tests with sediment at various depths using ceramic airstones:
Volumes of pore water that could be retrieved without incorporation of overlying water were dependent on the depth of airstone installment rather than overall sediment depth ( Figure   A10). Minimum retrievable volumes per sampling port ranged from 15 mL at a sediment depth and sampling depth of 1 inch; to 100 mL at a sediment depth of 3 inches and a sampling depth of 1.5 inch ( Figure A11). The minimum and average sample amounts that could be retrieved at a sampling depth of 1.5 inch were 73 and 107 mL/port, respectively. Furthermore, there was an influence of the number of ports sampled on the retrievable pore water volume ( Figure A12).
When only 8 ports were sampled the average retrievable pore water amount was 125 mL/port, while the retrievable amount was almost proportionally and significantly reduced when 16 ports were sampled (79 mL/port; p=0.003). However, total number of ports had no influence on the total retrievable volume from all ports (p=0.133).
To verify the above visually determined retrievable volumes an objective assessment of dye concentrations was conducted by means of fluorescence quantification using the Polastar Optima microtiter plate fluorescent reader. This experiment indicated that volumes that could be retrieved without incorporation of dye from the overlying water were 50 and 60 mL/port for sediment depths of 2 and 3 inches, respectively ( Figure A13). Exceptions were ports 1 or 16, which are located directly at the inflow and outflow of the test chambers. If ports 1 and 16 were included, these volumes were reduced to 30 and 40 mL/port for the 2 and 3 inch sediment depths, respectively.

Conclusions:
Based on the data it was recommended to utilize a sampling depth of 1.5 inches due to the enhanced sampling properties as defined by significantly large retrievable volumes. Because depth of sediments did not significantly affect pore water sampling, and considering volume restrictions for certain sediments, it was proposed to use a seeding depth of 2 inches in the definitive exposure study. This study did not permit assessment of the influence of the removed sediment volumes on re-equilibration of pore water but it was assumed that the here described maximum volumes will significantly deplete pore water. Therefore, it was recommended to reduce the least amounts of pore water to a third of the minimum amount reported during the objective assessment of dye concentrations (fluorometer measurement) that could be removed without detection of dye from overlying water, namely 15 and 8 mL per port using a design of 8 or 16 ports, respectively. Also, given some of the variation observed at the ports located closest to the in-and outflow of the test chambers, it was proposed not to sample within the first and last 4 inches of the fluvial test systems. To insure that during a sampling event there was equal drawing of pore water throughout the sediment layer, a minimum of 8 ports was sampled, at approximately 8-10 mL each, resulting in a total sample volume of approximately 80 mL, which was sufficient for the water quality and metals analyses in the white sturgeon definitive sediment toxicity study. Figure A9. Distribution of airstones throughout the test chamber (shown without sediment) for methods development of Upper Columbia River sediment toxicity tests.
The example presented here has 8 airstones installed. The external view (B) shows sampling ports through which pore water would be removed.
A B Figure A10. Mean maximum recoverable volumes of pore water per test system (A&B) and average recoverable volumes per individual sampling port (C&D) under different sampling regimes and sediment depths for methods development of Upper Columbia River sediment toxicity tests.
Data was sorted by sediment depth (A&C) and depth of sampling device (B&D). Bars represent mean values ± 1xSD (error bars). X-axis numbers represent sediment (first number) and pore water sampling (second number) depths in inches. The numbers in brackets after the sediment and pore water sampling depth indicate the number of ports that were sampled during each experiment. Experiments were conducted at sediment seeding depth of 2 (A) and 3 (B) inches in the test chamber, and pore water was sampled at a depth of 1.5 inches beneath the sediment surface. Red dotted line: Average dye concentration in overlying water.

Order 7: Gradients between pore-and overlying water
Objective: Evaluate potential gradients between pore water and overlying water under different hydrological conditions (e.g., flow velocity). Gradients in conductivity, pH and dissolved organic carbon (DOC) between pore water and overlying water were measured by means of suction devices (airstones) that were installed at 10.2 cm intervals along the entire length of the centre of the sediment exposure chamber. Suction devices were buried at different depths to enable sampling of pore water in the top 2.5 cm of sediment and in the sediment-surface water transitional zone (pseudo hyporheic area with 4 rocks per 100 cm 2 for habitat enrichment as described in Order #4). Experiments without rocks for habitat enrichment were excluded as it was decided based on the findings from Order 2 that rocks were to be used in subsequent sturgeon experiments. Parameters that were used to assess potential gradients between pore water and overlying water were conductivity, DOC and pH.

Experimental Design
Measurements were made 24, 48, and 96 hr after initiation of the experiment. Dye was not used in these experiments since it was discovered during studies for Orders 5 and 6 that dye would not readily seep into the sediment without being physically pulled through. Flows that were tested ranged from greater (25 L/min) to lesser flow rates (17 L/min). These flow-rates were deemed appropriate for maintaining conditions appropriate for white sturgeon early life stage culture.

Decision Criteria
The goal of this experiment was to establish conditions under which the gradient between pore water and overlying water in the pseudo-hyporheic area is minimal while maintaining conditions appropriate for sturgeon early life stage culture.

Results
There were significant differences between overlying water and pore water for a number of parameters ( Figures D14-16). These differences were most prominent for conductivity where significantly greater values were recorded in pore water samples regardless of depth and time of sampling ( Figure A14). There were differences in conductivity between pore water at different greater depths in the sediment. However, the differences were less than those between overlying water and pore water. Also, in the reference sediment treatment group (Genelle sediment) statistically significant increases in conductivity in pore water were observed between the 24 and 96 hr at the greatest sampling depth. No such differences occurred in the experiment with Deadman's Eddy (DE) sediment. Flow rate did not have an effect on conductivity in any of the matrices analyzed.
In the reference sediment experiment there was a statistically significant decrease in pH of pore water from both depths relative to that in overlying water. No such difference was observed in the DE sediment test group. In fact, pH was not different among sampling times or depths in this group. In general, DOC concentrations were highly variable in pore water when measured 24h after initiation of the experiment. It is assumed that these differences between the early (24h) and later measurements are due to the fact that the DE substrate tested was of a dry nature prior to submersion in the test systems, and therefore, at 24h there were still significant dissolution processes ongoing. Similarly, the reference sediment (saturated with water during storage) used was mixed and introduced into the test system just prior to t=0, likely resulting in a very different initial pore water composition that slowly mixed with overlying water until a certain degree of steady-state was reached. It is assumed that after 48h this dissolution or exchange between overlying and pore water was mostly completed or had stabilized. While the reference sediment group showed no further change in pore water DOC concentrations after 48h, there was still an apparent decrease in DOC concentrations in the DE sediments after 96h. It is not possible, however, to extrapolate from this observation to riverine sediments because the DE substrate was dry (collected above the water line from a beach/gravel bar). It can be assumed that water saturated sediments will behave very differently due to the lack of initial dissolution processes. This may also explain the differences observed between the DE and reference substrate. Depth of pore water sampling did not have a marked effect on DOC patterns.

Conclusions
There were gradients among measurement parameters between overlying water and pore water. These appeared, however, to not be influenced by flow-rate or duration with the exception of a small difference in the reference sediment at the greatest pore water sampling depth. This also indicates -with the exception of some sediments at greater sampling depth -that gradients were relatively stable and do not change over time under constant flow-conditions. Also, the time-depended increase in conductivity at greater sampling depth as observed for the reference sediment could be indicative of shallower sediment horizons reaching steady-state more quickly.
The reason why this only occurred for the reference substratum could be due to the fact that the DE sediment was collected from the gravel bar above the water line, and thus, may contain lesser amounts of fines. This could result in a lesser porous structure of the reference sediment causing more resistance in the flow of pore water between sediment horizons.
Furthermore, sediment-sampling depth had a significant influence on conductivity but not pH or DOC. Based on this result and the findings from Orders 5 and 6, to enable sampling of sufficient volumes while reducing differences between overlying water and pore water sampling ports at shallower sampling depths between 2.5 and 3.4 cm was recommended. Also, considering the lack of time-dependency of gradients between overlying water and pore water, indicating rapid establishment of steady-state after initiation of the experiments, it was favoured to reduce the equilibration time prior to introduction of test organisms in the definite exposure studies. Figure A14. Mean conductivity in overlying water and pore water at flow-rates of 17 (A&B) and 25 (C&D) L/min for methods development of Upper Columbia River sediment toxicity tests.
Overlying and pore waters were sampled at depths of 1 and 2 inches at 24, 48 and 96 h after initiation of experiment. Sediment types tested were reference sediment and sand bar substrate collected at Genelle (A&C) and Deadman's Eddy (B&C). Asterisks indicate significant difference from mean response in overlying water measured at the same time (p<0.05; Student's t-test). Figure A15. Mean pH in overlying water and pore water at flow-rates of 17 (A&B) and 25 (C&D) L/min for methods development of Upper Columbia River sediment toxicity tests.
Overlying and pore waters were sampled at depths 1 and 2 inches at 24, 48 and 96 h after initiation of experiment. Sediment types tested were reference sediment and sand bar substrate collected at Genelle (A&C) and Deadman's Eddy (B&C). Asterisks indicate significant difference from mean response in overlying water measured at the same time (p<0.05; Student's t-test). Overlying and pore waters were sampled at depths of 1 and 2 inches at 24, 48 and 96 h after initiation of experiment. Sediment types tested were reference sediment and sand bar substrate collected at Genelle (A&C) and Deadman's Eddy (B&C). Asterisks indicate significant difference from mean response in overlying water measured at the same time (p<0.05; Student's t-test).

Order 8: Time to 'steady-state' after introduction of test sediments into the test chambers
Objective: The objective was to identify the minimum period of time necessary for the exposure chamber to attain a steady-state based on basic water quality parameters. The objective of this work was not to attain steady-state conditions for chemicals of potential concern (COPCs); but rather, to ensure that non-COPCs do not adversely affect test results (i.e., introduce uncertainty) when organisms are introduced.

Experimental Design
Basic water quality parameters were monitored in test chambers containing river sediment at 0, 12, 24, and 48 h, and every 48 h thereafter until steady state. Measurements included conductivity, dissolved oxygen (DO), ammonia, nitrate, colour, total dissolved solids (TDS), and pH at the inflow and outflow of the exposure chamber.

Decision Criteria
'Steady-state' is attained when measured water quality parameters do not vary more than 10 percent from one measurement event to the next.

Results
Over the course of the experiment values for temperature, dissolved oxygen, pH, and colour had no significant variability (greater than ± 10 percent between measurement events) at

Conclusions
Steady-state is achieved for temperature, conductivity, dissolved oxygen, pH, colour, and total dissolved solids within 48 hours. Ammonia and nitrate did not reach steady-state, but this is likely due to the natural variability of these measurements. DOC did not reach steady-state and after further experimentation it is suspected that DOC may not reach steady-state.
Order 9: Determination of most efficient cleaning method minimizing resuspension of sediments.
Objective: Identify optimum cleaning techniques without utilization of invasive suction devices while employing large particle filters-with and without addition of diet (bloodworms, oligochaetes, semi-moist diet, and Artemia).

Experimental Design
Food was introduced simulating three feeding event per day using Artemia, worms and semi-moist diet. At days 2, 3, 4, and 5 chambers were manually cleaned (daily) to remove as much biofilm as possible without significant re-suspension of sediments. At 5, 10, 20, and 30 minutes after each cleaning event, near bottom water samples, approximately 1 cm above the sediment surface, were sampled. Turbidity of samples as a measure of re-suspended matter was determined using light scattering methods as described in EPA Method 180.1 or Standard Method 2130B (Standard Method 1995. Three different cleaning techniques were initially investigated at the beginning of the experiment: 1.) Siphoning the sediment surface with the use of a 3/8" ID hose.
2.) Scraping the sediment surface with the use of a plastic spatula.
3.) Pipetting debris with the use of a modified pipette.
After initial attempts it was decided that siphoning and scraping with a spatula were not appropriate methods of cleaning sediment surfaces as they were too invasive and ineffective, respectively.

Decision Criteria
Optimum cleaning techniques were determined as a function of minimizing re-suspension of sediment and efficiency of cleaning as determined by measurements of turbidity. It is acknowledged that any type of physical removal of bio-growth will cause re-suspension to a certain degree, and the final method to be established will be a compromise between efficiency of cleaning and amount of sediment re-suspended during the cleaning event.

Siphoning
Siphoning debris from the sediment surface was effective but was deemed too invasive as it removed and disturbed sediment in the process. Grains of sediment were removed from the chamber in the cleaning process. Past experience demonstrated an increased risk of fish injury as some organisms would be sucked into the cleaning tube. Turbidity analyses did not reveal any significant differences between pre-cleaning conditions or at any time period of up to 30 minutes post cleaning (Figure A17 A). Investigation of siphoning as a cleaning method was discontinued after day 2.

Scrapping with a spatula
Scrapping with a spatula was ineffective. It was time consuming and once the debris was dislodged from the sediment surface it was difficult to remove from exposure. Turbidity analyses revealed an increase in turbidity 5 and 10 minutes post cleaning and decreasing there after (Figure A17 B). Investigation of scrapping with a spatula as a cleaning method was discontinued after day 2.

Modified pipette
Pipetting the sediment surface to remove debris proved to be the most efficient and less invasive method. Biofilm and food could be easily dislodged from the sediment surface with minimal disturbance to sediment and effectively removed from the exposure chamber. Turbidity analysis revealed no significant differences between pre-cleaning conditions or at any time period of up to 30 minutes post cleaning for the entirety of the experiment (day 2 -5), with the one exception at day 3, ten minutes post cleaning ( Figure A17 C). Turbidity analyses at this time point revealed 0.6 NTU. This is considered to be a condition of other factors than cleaning techniques as all other turbidity results were within normal ranges throughout the 30 minute test.
The possible introduction of foreign material during sampling could explain elevated turbidity levels at this time point.

Conclusions
Cleaning by use of a modified pipette allowed the technician to select unwanted debris and remove it from the exposure chamber with minimal disturbance to the sediment. In addition, the risk of injury to fish was minimized. Siphoning proved effective when cleaning reservoirs or screens, but only when direct contact with fish and sediment was not involved. The modified pipette was used as the primary cleaning method for sediment within the exposure chamber. Figure A17. Turbidity in surface water prior (-5 [minutes]) and after (5, 10, 20 and 30 [minutes]) addition of food to fluvial test chamber at days 2 (A, B, C), 3 (C), 4(C) and 5(C) after initiation of feeding routine as determined by nephelomety using different cleaning techniques (A: Siphoning; B: Scraping; C: Modified Pipette) for methods development of Upper Columbia River sediment toxicity tests. NTU = Nephelometric Turbidity Units (NTU).

Order 10: Artificial laboratory control sediment
Objective: The objective was to select laboratory control sediment that had physical characteristics suitable for early life stages of sturgeon, was similar to sediments found in the UCR, and comparable to sediments used historically in standard early life stage tests with fishes.

Experimental Design
Several different silica sand and crushed/ground granites available from commercial vendors were considered. Sediments were layered into the test chamber and water quality parameters were assessed in comparison to average reference riverine sediment water quality parameters.

Decision Criteria
Criteria used to select suitable laboratory reference sediment were: • Grain sizes between 0.5 and 2 mm in diameter • Color similar to UCR sediments (preference for dark coloration) • pH, dissolved oxygen, hardness, and alkalinity do not differ by more than 50 percent from average values determined for riverine sediments • Certificate of analysis for contamination Gravel did not change water quality parameters in any significant (± 50%) way compared to average water quality parameters found in test chambers with reference riverine sediments. Upon analysis for percent gravel and percent sand, it was found that the two reference sediments (Genelle and Lower Arrow Lake) were gravelly sand with approximately 25% gravel content.

Several
Analysis by CAS showed that none of the chemicals/metals analyzed exceeded SEVs in the artificial control sediment.

Conclusions
The Hagen Geosystem Black Fine Gravel (ART #12648) was suitable laboratory reference control sediment. This sediment was within the 0.5 to 2 mm grain size with an average grain size of 1.11 mm and a least and greatest grain size of 0.85 mm and 1.68 mm, respectively.
Statistically significant correlations are defined as p ≤ 0.05 = *; p ≤ 0.01 = **; p ≤ 0.001 = ***. If correlations were observed between sampling techniques the Mann-Whitney U test was performed to assess whether concentrations of metal were statistically different; p values are presented in brackets.
N/A Not applicable because sampling technique was not performed in matrix  Table A8. Mean (± standard deviation) of general water quality data during the study. Table A9. Required sample containers, preservation, and holding times for overlying water, sediment-water interface water, and pore water samples. Table A10. Number of replicate exposure chambers per treatment group evaluated during the course of the study. Chemistry only chambers represent replicates in which peepers and diffusive gradient thin-film [DGT] probes were installed and used to obtain additional water quality information. Table A11. Volumes of sediment collected for determining toxicity of sediments to white sturgeon. Table A12. Distributions of sizes of particles in sediments evaluated in exposure chambers.

APPENDIX A1
Estimating summary statistics for datasets that include below detection limit values In the analysis of water quality data that accompanies the 2010 chronic sturgeon sediment exposure, reported concentrations as measured on a parts per billion (ppb) basis were frequently below analytical detection limits (BDLs). As a result, the true concentrations for chemicals of potential concern (e.g., dissolved metals) lay somewhere between zero and the analytical method detection limit (MDL) or method reporting limit (MRL). To enable the use of this data (i.e., censored data) in evaluating summary statistics such as arithmetic-and geometric means, and standard deviations, maximum likelihood estimation (MLE) procedures were used. Procedures such as MLE provide better estimates of summary statistics for censored data (e.g., BDLs) than simple "blind" calculations that treat BDLs as detected measurements or 'fabricating' values with the use of archaic substitution (e.g., one-half the value of the detection limit) methods To evaluate these methods, a Monte Carlo procedure was used to generate a sample dataset from a known distribution. Estimates of the geometric mean and standard deviation from each of the methods could then be compared to the known answer to evaluate the accuracy and precision of each method. Sample datasets generated by Monte Carlo were intended to resemble the types of metal concentration data encountered in the study. Values for geometric mean, standard deviation, number of data points, and fraction of data that were BDL were all chosen to resemble the actual metal data. For a given distribution with specified geometric mean and standard deviation, individual data points were randomly generated, some noise representing plus or minus 10 percent of the value was introduced to represent analytical variability, and a detection limit was then chosen so that a specified fraction of the available data were BDL.
Different values of the fraction BDL were used ranging from 20 percent to 80 percent of the total number of data to test these methods over a range of conditions representative of metal concentration data in the 2010 sturgeon database. An example dataset is shown in Figure A1.1.
For these example data, there are 10 data points and 70 percent of them are BDL. The "true" lognormal distribution used to generate the data is shown as the black diagonal line, and represents a dataset with a geometric mean of 0.017, and standard deviation of 0.75.
For this example, the 10 data points were then supplied to four different estimation procedures to evaluate how well these methods could estimate summary statistics. Values that were BDL were replaced by the detection limit (as shown in Figure A1.1). Figure A1.1. An example test dataset generated for a Monte Carlo evaluation of numerical procedures to estimate statistical distributions for data sets that include values below an analytical detection limit.

Note:
The true distribution is a lognormal distribution with a geometric mean of 0.017, and standard deviation of 0.75 and is shown with the solid black line. Ten random sample points were generated from this distribution. To each data some random noise was added. A detection limit was selected and any points with values below the detection limit were replaced with the detection limit. Observations that remain above the detection limit are shown as circles and those below the detection limit are shown as less-than (i.e. "< ") symbols plotted at the value of the detection limit. The geometric mean, standard deviation, number of points, and proportion of data below the detection limit were all selected to be similar to actual metal datasets produced as part of the chronic sturgeon study.
The MLE procedures are based on optimization of a likelihood function (Shumway et al. 2002). For a given dataset with n observations, the likelihood function is based on the following equations: Where, for the likelihood function L across n observations P(x) is the probability density function for a normal distribution used for non-BDL values of x, and C(x) is the cumulative density function used for BDL values of x. For a given mean (μ) and standard deviation (σ), the probability density function is: For the same distribution, a cumulative density function is defined as: Where erf is the Gauss error function (Andrews 1997). For detected observations, the censored flag δ is 0, so only the term for P(x) is used in the likelihood function, and the C(x) term will drop out. For BDL observations, the censored flag δ is 1 and the P(x) term will drop out. The goal of the MLE procedure is to find a mean (μ) and standard deviation (σ) that maximizes "L" (Equation E-1) for a given dataset, that includes both detected and BDL observations. For a lognormally distributed dataset, a geometric mean can be found by applying the MLE to logtransformed values of x. The "blind" calculation was included to allow comparison of MLE methods that consider BDL values against a simple alternative to demonstrate the benefit of incorporating these methods into the overall analysis. As shown in the pink line in Figure A1.2 if the BDL data are treated the same as other measurements, calculation of the geometric mean tends to produce a value that is higher than the true value, and the estimate of the standard deviation is lower than the true value. For this example, the resulting "blind" estimate of the geometric mean is 0.047, compared to an actual value of 0.017; while the estimate of the standard deviation is 0.36, compared to an actual value of 0.75.  Note: Summary statistics for datasets that include values below analytical detection limits can be estimated using MLE techniques. An estimate of the distribution using MLE is shown for the sample data described in Figure A1.1 (brown line). The MLE estimates of the geometric mean (0.011) and standard deviation (0.48) are closer to the true distribution (shown in black) than a blind calculation that does not consider BDL values (shown in Figure A1.2). Figure A1.4. Frequency histograms for 4000 estimates of the geometric mean from 4000 different synthetic datasets generated as part of the Monte Carlo evaluation.
Note: Results for four estimation methods are shown, including the CENMLE and ROS procedures in the R-statistical package (Panels A and B;Helsel 2005), the MLE procedure built into the BLM (Panel C) and a blind calculation that treats BDL as normal measurements (Panel D). Estimated values are shown as a ratio to the actual geomean. For this comparison 40 percent of the synthetic data were replaced by a detection limit value.
Similar conclusions are reached from comparisons of the estimates of the standard deviation (see Figure A1.5). Both MLE methods and ROS produce histograms centered around a value of 1, indicating that there is no systematic bias in these methods. However, estimates from the ROS method tend to deviate from the true values more frequently (and hence a broader histogram in Figure A1.5 Panel B). The blind calculation shows a systematic bias with estimates of standard deviation consistently lower than the true values. Figure A1.5. Frequency histograms for 4000 estimates of the standard deviation from 4000 different synthetic datasets generated as part of the Monte Carlo evaluation. to be biased to values greater than the actual value (as in Figure A1.4, Panel D).
The MLE procedure was also used to develop box-and-whisker plots for data that included BDL values. An example of this application is shown in Figure A1. If MLE methods are used to consider BDL values, a much better estimate of the true distribution results such that the red box in Figure A1.6 Panel C is nearly identical to the gray box that results from the true distribution. It is important to note, however, that the resulting box and whisker plot that results from the MLE analysis appears to produce a geometric mean that is lower than all of the observed data (i.e., measured values and detection limits for samples that are BDL). This apparent discrepancy results from the fact that a high proportion of the observed data are actually BDL values plotted at the detection limit. The real values that correspond to these BDL values are, by definition, lower than the detection. The MLE procedure considers this fact, and produces an estimate of the geometric mean accordingly. Similar box and whisker plots that correspond to metal concentrations in the chronic sturgeon exposure chambers frequently exhibit similar behavior, and the comparably low geometric means evident in those figures are likewise an understandable and expected consequence that results from a high proportion of BDL values in the metals datasets. is developed from the true distribution. In Panel B, the measurements are used to develop the box plot (pink) without considering that some values are below detection limit and show the typical overestimation of the mean, and underestimation of the standard deviation. In Panel C, the summary statistics for the box plot (red) were derived from the MLE estimate of the distribution, as characterized by the geometric mean and standard deviation.

SUMMARY
 Metals concentration datasets in the chronic sturgeon exposures frequently include values that are below BDLs.
 A Monte Carlo analysis showed that ignoring the presence of BDLs resulted in systematic errors in estimates of the mean and standard deviation and should be avoided.
 Consideration of BDLs using either MLE or ROS produced unbiased estimates of the mean and standard deviation, and of these two methods the MLE procedure produced an accurate result.
 The MLE procedure, and specifically the CENMLE procedure in the R-statistical software package, was chosen for the analysis of metal concentrations in chronic sturgeon exposures to produce summary statistics, and summary graphics (e.g., box and whisker plots). Data were examined to determine usability of the analytical results and compliance relative to requirements specified above and the analytical methods. In addition, deliverables were evaluated for completeness and accuracy. Qualifier codes have been placed next to results on the data tables to enable the data user to quickly assess the qualitative and/or quantitative reliability of any result based on the criteria evaluated. EPA's Quality Assurance/Quality Control (QA/QC) chemist reviewed the draft data and data validation reports. Issues were resolved and EPA approved the data for public. The following sections summarize results of the validation. It should be noted that general in-house water quality parameters monitored (e.g., temperature, pH, DO, conductivity, and inorganic nitrogen) during the tests were not validated by ESI. These routine water quality parameters were monitored to ensure that toxicity tests were within design constraints at the time of testing.

Overall data quality
Most analytical data were useable, with qualifications presented in data validation reports and included in the project database. Only useable data were included in this report, although all rejected data are in the project database. Data qualifiers were assigned to data by the laboratory and validators to signify when data were out of calibration range (i.e., below or above levels of quantification), where contaminated blanks compromised data interpretability, or if matrix spikes, internal standards, or other quality control metrics were exceeded. Tables summarizing the number of samples with each type of data qualifier by analyte and information regarding how data qualifiers were used, along with full data validation reports, are available upon request from the corresponding author.
Of the 8,482 analytical data points collected during this Study, a total of 1,014 data points (<12 percent) were rejected as follows: · Results for all organochlorine pesticide compounds, PCB congeners, and PAH compounds in 910 sediment data points were qualified as unusable due to exceeded sample preservation temperatures.
· Results for molybdenum 20 data points were qualified as unusable due to very low reporting limit (RL) standard recoveries.
· Results for calcium, iron, and/or molybdenum in 84 aqueous samples were qualified as unusable due to very low RL standard recoveries.

Sample transport and holding times
Validity of analytical data was evaluated with regard to sample preservation conditions and sample holding times from the date of collection to the date of analysis. Aqueous and sediment samples were to be preserved at 4 ± 2°C. Most samples received by the laboratory were within the required preservation temperature range. In cases where the sample receipt temperature was outside of the study-required range, and the associated analytical method did not specify preservation criteria, data were not qualified. Results for PCB congeners and PAH compounds in several sediment samples were qualified as estimated due to receipt temperatures above the method-specified range. In a limited number of instances, aqueous samples submitted for metals analyses could not be analyzed for pH by the laboratory due to limited sample volume. Aqueous samples for metals analysis were to be preserved to a pH < 2 su. As a result, there was insufficient information for ESI to verify that the affected samples were properly preserved; and data were not qualified due to this issue. No transport holding times were exceeded.

Equipment Rinse Blanks
Results for calcium, iron, magnesium, sodium, aluminum, barium, copper, lead, manganese, molybdenum, nickel, zinc, cadmium, chromium, cobalt, silver, and/or antimony in several sediment samples were qualified as "not-detected" due to the presence of these analytes at similar concentrations in the associated rinse blanks.
Results for aluminum, antimony, barium, cadmium, calcium, chromium, cobalt, copper, iron, lead, magnesium, manganese, molybdenum, nickel, potassium, silicon, silver, sodium, vanadium, zinc, mercury, chloride, fluoride, dissolved organic carbon, and/or total dissolved solids in several aqueous samples were qualified as "not-detected" due to the presence of these analytes at similar concentrations in the associated rinse blanks.

Laboratory Holding Times
Results for organochlorine pesticide compounds, PCB congeners, AVS/SEM, TOC, pH, and/or grain size in several sediment samples were qualified as estimated due to holding time exceedance.

Inorganics
All inorganic analyses were conducted by CAS in Kelso, WA. Overall, the data reviewed are usable with the qualifications detailed in the validation reports and database. Figure A4.1. Concentrations of dissolved aluminum as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.

Calibration
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.3. Concentrations of dissolved arsenic as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.4. Concentrations of dissolved barium as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.5. Concentrations of dissolved beryllium as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.6. Concentrations of dissolved chromium as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.7. Concentrations of dissolved cobalt as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.8. Concentrations of dissolved iron as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.9. Concentrations of dissolved manganese as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.10. Concentrations of dissolved mercury as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.11. Concentrations of dissolved molybdenum as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.12. Concentrations of dissolved nickel as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.13. Concentrations of dissolved selenium as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.14. Concentrations of dissolved silicon as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.15. Concentrations of dissolved silver as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.16. Concentrations of dissolved thalium as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles. Figure A4.17. Concentrations of dissolved vanadium as a function of treatment and sample type evaluated within the white sturgeon sediment toxicity tests.
Samples types are presented in the following order: overlying water (top panel), sediment water interface water (2 nd panel), pore water at 1 cm (3 rd panel), and pore water at 2.5 cm (bottom panel). Where appropriate and applicable, concentrations as determined by use of different sampling techniques are identified within the top right-hand corner of respective panels. Individual measurements are shown as circles, measurements below detection are illustrated with a less than symbol ("<") plotted at the detection limit, qualified samples due to blank contamination are illustrated with an asterisks (*), and date qualified as "estimated" are represented with the symbol "E" at the estimated value. The 25 th and 75 th centiles from the maximum likelihood estimate (MLE; see Supplemental Materials) calculated distribution is used for the edges of the box, the median is shown as a horizontal line drawn through the middle of the box. Whiskers show maximum and minimum values exclusive of extreme values. Extreme values were identified as values that were more than 1.5 times the inter-quartile range above the 75 th or below the 25 th centiles.