Laboratory studies of toxicity to aquatic organisms in single metal solutions provide the fundamental data for model validation. Laboratory data include the composition of water samples [temperature, pH, concentrations of major ions (calcium [Ca], magnesium [Mg], sodium [Na], potassium [K], chlorine [Cl], sulfate [SO₄], carbonate species), dissolved organic carbon (DOC), and metals of interest] as well as responses of biota upon exposure to metals. Laboratory studies provide an opportunity to validate models because they consider well-controlled and less-complex systems and generally examine a wider range of metal concentrations than those observed in natural, pristine environments. Previously collected laboratory data for six organisms are used in this study and summarized in sections, “Juvenile White Sturgeon (Acipenser transmontanus),” “Hyalella azteca,” and “Natural Benthic Macroinvertebrate Community.”

Two sets of in situ porewater were collected in the UCR and analyzed for their composition (that is, temperature, pH, dissolved concentrations of major ions, DOC, and metals of interest). The first set of 78 samples included surface water and porewater and was collected in 2015 across the sediment-water interface at 7.5 centimeters (cm) above the interface (surface water), at the interface (0 cm; sediment-water interface), 4.5 cm below the interface (shallow porewater), and 14.5 cm below the interface (deep porewater) at locations between the United States-Canadian border and China Bend (fig. 1). These data are summarized in Cox and others (2016) and more fully discussed in Balistrieri and others (2018). The porewater had an average pH of 7.91±0.21, a hardness of 94±14 milligrams per liter (mg /L) of calcium carbonate (CaCO₃), and 0.56±0.22 mg/L of DOC. The second set of samples was 122 composite porewater samples collected in 2019 from 0 to 15 cm below the sediment-water interface at areas of interest (AOIs) in the UCR. These areas include reference sites above the United States-Canadian border and sites downstream from the border at Deadmans Eddy, China Bend, and Evans (fig. 1; Environmental Resources Management, 2022). Samples at Deadmans Eddy were collected at an eddy site in the main stem of the river and within the backwater of a sand bar. The data are summarized and more fully discussed in Balistrieri (2024). These porewater samples had average pH of 7.82±0.78, hardness of 88±43 mg/L of CaCO₃, and 1.1±1.2 mg/L of DOC. Although metal concentrations were reported for all porewater samples in this study, many were qualified as non-detectable or less than detection limits. In these cases, dissolved metal concentrations were assigned to be 50 percent of reported values.

The WHAM-F_TOXβ model, which was developed by Tipping and others (2023), is used to evaluate hydrogen and metal toxicity to biota in laboratory and field studies. An overview of the model using Cd and Zn as example metals and mayfly and fish as example organisms is presented in figure 2. Responses for mayfly to metals are less sensitive than responses for fish in the example. Laboratory and field metal-toxicity studies characterize the responses of organisms that are exposed to ranges of dissolved metal concentrations. Typically, responses are related to metal concentrations for single metal solutions or toxic units or other multiple metal metrics for metal mixtures (Schmidt and others, 2012; Mebane and others, 2015). In the WHAM-F_TOXβ model, biological responses are related to toxicity functions. These toxicity functions, F_TOXβ, include accumulations of hydrogen and metals on biological receptors, intrinsic potency coefficients for hydrogen and metals, sensitivity values for biota, and times of exposure to metals. The accumulations of hydrogen and metals are determined by treating biological receptors as humic acid. This approach uses a comprehensive database of metal-organic matter interactions that is incorporated into the computer program Windermere Humic Aqueous Model 7 (WHAM 7) (Tipping and others, 2011; Lofts, 2012). Thus, accumulations of elements on all biological receptors are represented by accumulations of elements on humic acid. This assumption means that the amount of hydrogen and metals bound to biological receptors is the same for all organisms in each water sample. Toxicity functions, F_TOXβ, are compared to positive responses (R) of biota exposed to hydrogen and metals and the data are fit using three piece-wise linear sections and lower (F_TOXβ_{, LT}) and upper (F_TOXβ_{, UT}) thresholds of F_TOXβ. For F_TOXβ<F_TOXβ_{, LT}; R=maximum response (for example, 100 percent) and for F_TOXβ>F_TOXβ_{, UT}; R=minimum response (for example, 0 percent). Responses decrease linearly between the lower and upper thresholds. Therefore, as F_TOXβ increases, responses become less positive (fig. 2). This model is applicable to single and multiple metal toxicity studies.

The toxicity function in equation form is:FTOXβ=αH θH+∑αMX θMX, with(1)αMX=β αMX*×kt/kt+1,(2)where

Fractional accumulations of hydrogen and metals on biological receptors are calculated from the compositions of water in laboratory and field studies using WHAM 7. The water composition of each sample is required for the calculations and includes temperature, pH, and total concentrations of major cations (Ca, Mg, Na, K), major anions (Cl, SO₄, inorganic carbonate species or alkalinity), DOC, and metals of interest (Cd, Co, Cu, Ni, Pb, Zn). Dissolved organic matter is assumed to be 100 percent fulvic acid (FA) and 50 percent DOC. Sixty five percent of DOC is considered to actively complex with metals (Bryan and others, 2002). Thus, FA (in grams per liter [g/L]) = 1.3×DOC (g/L). The analog for biological receptors [that is, humic acid (HA)] is included for each sample at concentrations of 1×10^-9 g/L. This concentration does not affect the dissolved chemical speciation of the solution. Solutions are assumed to be in equilibrium with amorphous iron and aluminum hydroxides (Tipping and others, 2002). Fractional accumulations are moles of hydrogen or metal per gram HA normalized to total binding sites on humic acid (HA–hydrogen [H]/HA_total or HA–MX/HA_total, where HA_total=5.1×10^-3 moles/g HA; Tipping and others, 2023).

The meta-analysis of toxicity studies done by Tipping and others (2023) resulted in values for intrinsic potency coefficients for metals relative to the potency coefficient for hydrogen (that is, α_H=1), sensitivity parameters for many biological species including juvenile white sturgeon and Hyalella azteca, mean values for the lower and upper thresholds of F_Toxβ, and the time dependent rate constant (k). The lower threshold of the toxicity function (F_TOXβ, _LT) is 0.503 and the upper threshold (F_TOXβ, _UT) is 1.137 by assuming that the toxicity function equals 0.820 at 50-percent response (Tipping and others, 2023). A summary of all parameters that are needed to model metal toxicity using the WHAM-F_TOXβ model is in table 1.

Porewater samples in the UCR contain mixtures of metals and the toxicity function for each mixture is calculated using the WHAM-F_TOXβ model. The toxicity of these metal mixtures to various benthic biota is assessed by comparing the calculated toxicity function for a porewater sample (F_TOXβ) to a benchmark toxicity function. The benchmark chosen for this work is the toxicity function at 20 percent negative response (TF20). Thus, this ratio, which is called a Toxicity Quotient (TQ), is defined as:

TQs greater than or equal to (≥) 1 for porewater samples indicate adverse impacts to benthic biota.

The WHAM-F_TOXβ model was developed for individual organisms. All parameters that are needed for modeling toxicity are available only for juvenile white sturgeon and Hyalella azteca. Sensitivity parameters for mayfly, caddisfly, or a community of BMI were not determined in the meta-analysis of Tipping and others (2023). One test of the model is to evaluate whether responses of juvenile white sturgeon and Hyalella azteca are predicted in single metal solutions using the parameters developed in the meta-analysis. The first step is to calculate accumulations of hydrogen and metals in the single metal toxicity studies for juvenile white sturgeon and Hyalella azteca. The toxicity functions then are calculated using the accumulations and meta-analysis parameters (eqs. 1 and 2; table 1). The results indicate that responses to zinc are reasonably predicted for the two organisms (fig. 3). In contrast, responses to Cd, Co, Cu, Ni, and Pb for these organisms are poorly predicted. The WHAM-F_TOXβ model was developed on the premise that each metal has intrinsic potency (that is, does not vary among organisms) and that the sensitivities of organisms to metals vary (Tipping and others, 2023). If the lower and upper thresholds for F_TOXβ and minimum and maximum responses are set at the meta-analysis values, then it is not possible for Cd, Cu, or Pb to have unique intrinsic potency coefficients that fit both sets of data.

The approach taken in this work is to use the meta-analysis values for α_H (1), α_Zn* (12.5), lower threshold (0.503) and upper threshold (1.137) of the toxicity function, and minimum positive response (0 percent) and maximum positive response (100 percent). The intrinsic potency coefficient for Zn (α_Zn*) determined in the meta-analysis was used because laboratory Zn data for juvenile white sturgeon and Hyalella azteca were reasonably predicted using this value (fig. 3). Additionally, fixing some parameters minimizes the number of adjustable parameters in the model. Values of the sensitivity coefficients (β) for all studied organisms are determined by fitting Zn data. Intrinsic potency coefficients (α_MX*) for Cd, Co, Cu, Pb, and Ni then are determined by fitting the single metal data for each organism (that is, juvenile white sturgeon, Hyalella azteca, BMI community, Ephemerellidae family, Heptageniidae family). Intrinsic potency coefficients for all metals were kept at the meta-analysis values for the insensitive Brachycentridae family. Data were fit using SOLVER in Excel and by minimizing the absolute difference between measured and predicted responses. The fitting parameters are in table 1 and the model fits for single metal exposures are in figure 4. It is clear from table 1 that a single intrinsic potency coefficient for each metal does not fit all datasets. For example, intrinsic potency coefficients for Cd range from 316.4 to 5,080.5 for the organisms.

The four mesocosm studies also evaluated responses of families of mayfly and caddisfly and a community of BMI to metal mixtures (Schmidt and others, 2019). Because porewater solutions contain metal mixtures, the predicted responses of these organisms in metal mixtures based on parameters developed in single metal solutions are of interest. Like the single metal exposures, the toxicity function was calculated for each multi-metal mesocosm sample and compared to the normalized responses of the mayfly and BMI community. The responses of mayfly and the BMI community are well predicted in multiple metal solutions using the optimized WHAM-F_TOXβ model based on fits in single metal solutions (fig. 5) and bodes favorably for predicting toxicity functions and responses in porewater with metal mixtures.

Because there are no organism responses for the field-collected porewater samples, the toxicity functions, responses, and TQs are predicted for multiple metal porewater in the UCR using the optimized WHAM-F_TOXβ model and parameterizations from laboratory experiments. The first step is to run WHAM 7 to determine accumulations of hydrogen and metals on HA for each porewater sample. The accumulations of hydrogen and metals are the same for all organisms in each field porewater sample as the biological receptors are all treated as HA. Boxplots of hydrogen and metal accumulations by HA in 2019 UCR porewater indicate that hydrogen accumulation is largest followed by Cu accumulations (fig. 6). The smallest accumulations are for Cd. The toxicity functions are calculated (eqs. 2 and 3; table 1) and used to predict responses for each porewater sample and each organism. Because intrinsic potency coefficients of metals and sensitivity coefficients are different among the organisms (table 1), the magnitude of the toxicity function for a given porewater varies for the organisms. Predictions of response for benthic biota in 2019 UCR porewater indicate that their sensitivity decreases as follows: juvenile white sturgeon > Ephemerellidae family > Hyalella azteca > Heptageniidae family ~ BMI community > Brachycentridae family (fig. 7). The relative sensitivities of the studied organisms to metal mixtures are reflected in decreasing toxicity functions for a given porewater sample. Biological responses become more positive as calculated toxicity functions decrease.

The fraction of porewater samples that predict adverse impacts to organisms is examined using TQs. The three piece-wise linear sections of the model that depict positive responses compared to toxicity functions are the same for all organisms (that is, same threshold values and minimum and maximum responses); thus, the value of the toxicity function at 20-percent negative response (TF20) is calculated at 0.63. Hence, a TQ can be calculated for each porewater sample (eq. 3). The fraction of porewater samples collected during 2015 and 2019 with a TQ≥1 varies with the metal toxicities and sensitivities of the studied organisms (fig. 8). Adverse impacts are predicted to decrease as the sensitivity of the organisms decreases. From 44 to 48 percent of porewater samples are predicted to have adverse impacts on the most sensitive organism (that is, juvenile white sturgeon), whereas no samples are predicted to have adverse impacts on the least sensitive Brachycentridae family.

The two porewater datasets from the UCR were collected for different reasons. The 2015 study examined water compositions across the sediment-water interface and the 2019 study considered porewater compositions at AOIs in the basin (that is, reference, Deadmans Eddy at eddy and bar sites, China Bend, and Evans; fig. 1). The fraction of shallow and deep porewater samples that are predicted to have a TQ≥1 for juvenile white sturgeon, the Ephemerellidae family, and Hyalella azteca is greater than 0.2 whereas the fraction of porewater samples that have a TQ≥1 for the Heptageniidae family and BMI community is less than 0.1 (fig. 9). All surface water and other sampled depths for the Brachycentridae family have no samples with a TQ≥1. The bar samples at Deadmans Eddy have the most adverse impacts (that is, fractions range from 0.28 to 0.89) for all organisms except the Brachycentridae family. The fraction of China Bend samples that have a TQ≥1 for the most sensitive organisms (that is, juvenile white sturgeon, Ephemerellidae family, and Hyalella azteca) ranges from 0.33 to 0.64.

The composition of metal mixtures in UCR porewater varies and each organism has unique responses to the mixtures depending on metal potency and their sensitivity. Organism sensitivity, hydrogen and metal potency, accumulations of H and metals on the biological receptors, and exposure time contribute to toxicity functions (eqs. 1 and 2). Identification of the dominant ion or ions that result in adverse toxic conditions is important because of differences in the chemical behavior and toxicity of each metal to biota. Knowledge of the dominant metal that causes toxicity is critical for assessing risk and developing appropriate remediation strategies. TQs for each organism compared to contributions of H and metals to the toxicity function indicate that Cu is the major contributor to adverse conditions (that is, TQ≥1) in the UCR for most benthic biota (fig. 10). The exceptions are the insensitive Brachycentridae family and Hyalella azteca, where Cu and Pb contribute to the most toxic conditions.