Metal Bioavailability Models: Current Status, Lessons Learned, Considerations for Regulatory Use, and the Path Forward.

Since the early 2000s, biotic ligand models and related constructs have been a dominant paradigm for risk assessment of aqueous metals in the environment. We critically review 1) the evidence for the mechanistic approach underlying metal bioavailability models; 2) considerations for the use and refinement of bioavailability-based toxicity models; 3) considerations for the incorporation of metal bioavailability models into environmental quality standards; and 4) some consensus recommendations for developing or applying metal bioavailability models. We note that models developed to date have been particularly challenged to accurately incorporate pH effects because they are unique with multiple possible mechanisms. As such, we doubt it is ever appropriate to lump algae/plant and animal bioavailability models; however, it is often reasonable to lump bioavailability models for animals, although aquatic insects may be an exception. Other recommendations include that data generated for model development should consider equilibrium conditions in exposure designs, including food items in combined waterborne-dietary matched chronic exposures. Some potentially important toxicity-modifying factors are currently not represented in bioavailability models and have received insufficient attention in toxicity testing. Temperature is probably of foremost importance; phosphate is likely important in plant and algae models. Acclimation may result in predictions that err on the side of protection. Striking a balance between comprehensive, mechanistically sound models and simplified approaches is a challenge. If empirical bioavailability tools such as multiple-linear regression models and look-up tables are employed in criteria, they should always be informed qualitatively and quantitatively by mechanistic models. If bioavailability models are to be used in environmental regulation, ongoing support and availability for use of the models in the public domain are essential. Environ Toxicol Chem 2019;39:60-84. © 2019 SETAC.

The aquatic hazard of hydrocarbon liquids and gases and the modulating role of pressure on dissolved gas and oil toxicity.

Hydrostatic pressure enhances gas solubility and potentially alters toxicity and risks of oil and gas releases to deep-sea organisms. This study has two primary objectives. First, the aquatic hazard of dissolved hydrocarbon gases is characterized using results of previously published laboratory and field studies and modeling. The target lipid model (TLM) is used to predict effects at ambient pressure, and results are compared to effect concentrations derived from extrapolation of liquid alkane hazard data. Second, existing literature data are used to quantify and predict pressure effects on toxicity using an extension of the TLM framework. Results indicate elevated pressure mitigates narcosis, particularly for sensitive species. A simple adjustment is proposed to allow TLM-based estimates of acute effect and TLM-derived HC5 values (concentrations intended to provide 95% species protection) for oil or gas constituents to be calculated at depth. Future applications, and opportunities and challenges for providing validation, are discussed.

Influence of Varying Water Quality Parameters on the Acute Toxicity of Silver to the Freshwater Cladoceran, Ceriodaphnia dubia.

The acute toxicity of silver to Ceriodaphnia dubia was investigated in laboratory reconstituted waters as well as in natural waters and reconstituted waters with natural organic matter. The water quality characteristics of the laboratory reconstituted waters were systematically varied. The parameters that demonstrated an ability to mitigate the acute toxic effects of silver were chloride, sodium, organic carbon, and chromium reducible sulfide. Factors that did not have a consistent effect on the acute toxicity of silver to C. dubia, at least over the range of conditions tested, included hardness, alkalinity, and pH. The biotic ligand model was calibrated to the observed test results and found to be of use in quantifying the effect of changing water quality characteristics on silver bioavailability and toxicity. The model generally predicted silver toxicity within a factor of two and should be useful in modifying water quality criteria.

The effect of food on the acute toxicity of silver nitrate to four freshwater test species and acute-to-chronic ratios.

Acute silver toxicity studies were conducted with and without food for four common freshwater test species: Daphnia magna, Ceriodaphnia dubia, Pimephales promelas (fathead minnow-FHM), and Oncorhynchus mykiss (rainbow trout-RBT) in order to generate acute-to-chronic ratios (ACR). The studies were conducted similarly (i.e., static-renewal or flow-through) to chronic/early-life stage studies that were previously performed in this laboratory. The acute toxicity (EC/LC50 values) of silver without food ranged from 0.57 µg dissolved Ag/l for C.dubia to 9.15 µg dissolved Ag/l for RBT. The presence of food resulted in an increase in EC/LC50 values from 1.25× for RBT to 22.4× for C. dubia. Invertebrate food type was also shown to effect acute silver toxicity. Food did not affect EC/LC50s or ACRs as greatly in fish studies as in invertebrate studies. ACRs for both invertebrate species were <1.0 when using acute studies without food but were 1.22 and 1.33 when using acute studies with food. ACRs for FHMs ranged from 4.06 to 7.19, while RBT ACRs ranged from 28.6 to 35.8 depending on whether food was present in acute studies. The data generated from this research program should be useful in re-determining a final ACR for silver in freshwater as well as in risk assessments.

Utility of tissue residues for predicting effects of metals on aquatic organisms.

As part of a SETAC Pellston Workshop, we evaluated the potential use of metal tissue residues for predicting effects in aquatic organisms. This evaluation included consideration of different conceptual models and then development of several case studies on how tissue residues might be applied for metals, assessing the strengths and weaknesses of these different approaches. We further developed a new conceptual model in which metal tissue concentrations from metal-accumulating organisms (principally invertebrates) that are relatively insensitive to metal toxicity could be used as predictors of effects in metal-sensitive taxa that typically do not accumulate metals to a significant degree. Overall, we conclude that the use of tissue residue assessment for metals other than organometals has not led to the development of a generalized approach as in the case of organic substances. Species-specific and site-specific approaches have been developed for one or more metals (e.g., Ni). The use of gill tissue residues within the biotic ligand model is another successful application. Aquatic organisms contain a diverse array of homeostatic mechanisms that are both metal- and species-specific. As a result, use of whole-body measurements (and often specific organs) for metals does not lead to a defensible position regarding risk to the organism. Rather, we suggest that in the short term, with sufficient validation, species- and site-specific approaches for metals can be developed. In the longer term it may be possible to use metal-accumulating species to predict toxicity to metal-sensitive species with appropriate field validation.

Critical tissue copper residues for marine bivalve (Mytilus galloprovincialis) and echinoderm (Strongylocentrotus purpuratus) embryonic development: conceptual, regulatory and environmental implications.

Critical tissue copper (Cu) residues associated with adverse effects on embryo-larval development were determined for the Mediterranean mussel (Mytilus galloprovincialis) and purple sea urchin (Strongylocentrotus purpuratus) following laboratory exposure to Cu-spiked seawater collected from San Diego Bay, California, USA. Whole body no-observed-effect-residues (NOER) were similar, with means of 21 and 23 microg g(-1) dw, for M. galloprovincialis and S. purpuratus, respectively. Mean whole body median effect residues (ER50) were 49 and 142 microg g(-1) dw for M. galloprovincialis and S. purpuratus, respectively. The difference in ER50s between species was reduced to a factor of <2 when expressed as soft tissue residues. Coefficients of variation among whole body-ER50s were 3-fold lower than median waterborne effect concentrations (EC50) for both species exposed to samples varying in water quality characteristics. This suggests that tissue concentrations were a better predictor of toxicity than water concentrations. The CBRs described herein do not differentiate between the internal Cu concentrations that are metabolically available and those that are accumulated and then detoxified. They do appear, however, to be well enough related to the level of accumulation at the site of action of toxicity that they serve as useful surrogates for the copper concentration that affects embryonic development of the species tested. Results presented have potentially important implications for a variety of monitoring and assessment strategies. These include regulatory approaches for deriving saltwater ambient water quality criteria for Cu, contributions towards the development of a saltwater biotic ligand model, the conceptual approach of using CBRs, and ecological risk assessment.

Validation study of the acute biotic ligand model for silver.

An important final step in development of an acute biotic ligand model for silver is to validate predictive capabilities of the biotic ligand model developed for fish and invertebrates. To accomplish this, eight natural waters, collected from across North America, were characterized with respect to ionic composition, pH, dissolved organic carbon, and sulfide. Tests were conducted with the cladoceran Ceriodaphnia dubia (48-h static) and the fish Pimephales promelas (96-h static renewal) to determine the concentrations causing lethality to 50% of the organisms (LC50s) for silver in each of these waters. Overall, the biotic ligand model adequately predicted silver toxicity to C. dubia; however, in some cases, predicted LC50 values exceeded measured values. The accuracy of the biotic ligand model predictions was less convincing for silver toxicity to P. promelas with pronounced problems in low-ionic strength waters. Another issue was the use of acclimated organisms in toxicity studies because the biotic ligand model has been developed with the use of a mix of studies with acclimated and nonacclimated test organisms of varying ages and sizes. To evaluate whether effects of acclimation to test waters influence biotic ligand model predictions, a subset of the natural waters were also tested with P. promelas that had been acclimated to the natural water for 7 d before testing. These experiments revealed no differences in toxicity between acclimated and nonacclimated P. promelas. To determine the influence of organism size, which has been previously correlated to Na(+) turnover and acute silver toxicity across multiple species, Na(+) and Cl(-) influx rates were measured in P. promelas of different sizes. Our results show that Na(+) and Cl(-) influx rates were inversely related to fish mass and positively correlated with silver sensitivity.

Comparison of short-term chronic and chronic silver toxicity to fathead minnows in unamended and sodium chloride-amended waters.

The chronic (early life stage [ELS]) and short-term chronic (STC) toxicity of silver (as silver nitrate) to fathead minnows (FHM) was determined concurrently in flow-through exposures (33 volume additions/d). Paired ELS (approximately 30 d) and STC (7 d) studies were conducted with and without the addition of 60 mg/L Cl (as NaCl). The paired studies in unamended water were later repeated using standard flow conditions (9 volume additions/d). The purpose of the paired studies was to determine if short-term chronic endpoints can be used to predict effects in ELS studies. For each experiment, a "split-chamber" design (organisms were held in a common exposure chamber) allowed the direct comparison between short-term and chronic exposures. It appeared that the chronic toxicity of silver was mitigated to some extent by NaCl addition. The maximum acceptable toxicant concentration for growth in the ELS study was 0.53 microg dissolved Ag/L under standard flow conditions. Early life stage and STC endpoints in all three studies typically agreed within a factor of two. Whole-body sodium and silver concentrations measured in individual fathead minnows during these studies showed an increase in silver body burdens and a decrease in sodium concentration. These results indicate that the STC study could be used as a surrogate test to estimate chronic toxicity and that the mechanism of chronic silver toxicity may be the same as for acute toxicity.

Predicting sediment metal toxicity using a sediment biotic ligand model: methodology and initial application.

An extension of the simultaneously extracted metals/acid-volatile sulfide (SEM/AVS) procedure is presented that predicts the acute and chronic sediment metals effects concentrations. A biotic ligand model (BLM) and a pore water-sediment partitioning model are used to predict the sediment concentration that is in equilibrium with the biotic ligand effects concentration. This initial application considers only partitioning to sediment particulate organic carbon. This procedure bypasses the need to compute the details of the pore-water chemistry. Remarkably, the median lethal concentration on a sediment organic carbon (OC)-normalized basis, SEM*(x,OC), is essentially unchanged over a wide range of concentrations of pore-water hardness, salinity, dissolved organic carbon, and any other complexing or competing ligands. Only the pore-water pH is important. Both acute and chronic exposures in fresh- and saltwater sediments are compared to predictions for cadmium (Cd), copper (Cu), nickel (Ni), lead (Pb), and zinc (Zn) based on the Daphnia magna BLM. The SEM*(x,OC) concentrations are similar for all the metals except cadmium. For pH = 8, the approximate values (micromol/gOC) are Cd-SEM*(xOC) approximately equal to 100, Cu-SEM*(x,OC) approximately equal to 900, Ni-SEMoc approximately equal to 1,100, Zn-SEM*(x,OC) approximately equal to 1,400, and Pb-SEM*(x,OC) approximately equal to 2,700. This similarity is the explanation for an empirically observed dose-response relationship between SEM and acute and chronic effects concentrations that had been observed previously. This initial application clearly demonstrates that BLMs can be used to predict toxic sediment concentrations without modeling the pore-water chemistry.

The biotic ligand model: a historical overview.

During recent years, the biotic ligand model (BLM) has been proposed as a tool to evaluate quantitatively the manner in which water chemistry affects the speciation and biological availability of metals in aquatic systems. This is an important consideration because it is the bioavailability and bioreactivity of metals that control their potential to cause adverse effects. The BLM approach has gained widespread interest amongst the scientific, regulated and regulatory communities because of its potential for use in developing water quality criteria (WQC) and in performing aquatic risk assessments for metals. Specifically, the BLM does this in a way that considers the important influences of site-specific water quality. This journal issue includes papers that describe recent advances with regard to the development of the BLM approach. Here, the current status of the BLM development effort is described in the context of the longer-term history of advances in the understanding of metal interactions in the environment upon which the BLM is based. Early developments in the aquatic chemistry of metals, the physiology of aquatic organisms and aquatic toxicology are reviewed first, and the degree to which each of these disciplines influenced the development of water quality regulations is discussed. The early scientific advances that took place in each of these fields were not well coordinated, making it difficult for regulatory authorities to take full advantage of the potential utility of what had been learned. However, this has now changed, with the BLM serving as a useful interface amongst these scientific disciplines, and within the regulatory arena as well. The more recent events that have led to the present situation are reviewed, and consideration is given to some of the future needs and developments related to the BLM that are envisioned. The research results that are described in the papers found in this journal issue represent a distinct milestone in the ongoing evolution of the BLM approach and, more generally, of approaches to performing ecological assessments for metals in aquatic systems. These papers also establish a benchmark to which future scientific and regulatory developments can be compared. Finally, they demonstrate the importance and usefulness of the concept of bioavailability and of evaluative tools such as the BLM.

Application of the biotic ligand model to predicting zinc toxicity to rainbow trout, fathead minnow, and Daphnia magna.

The Biotic Ligand Model has been previously developed to explain and predict the effects of water chemistry on the toxicity of copper, silver, and cadmium. In this paper, we describe the development and application of a biotic ligand model for zinc (Zn BLM). The data used in the development of the Zn BLM includes acute zinc LC50 data for several aquatic organisms including rainbow trout, fathead minnow, and Daphnia magna. Important chemical effects were observed that influenced the measured zinc toxicity for these organisms including the effects of hardness and pH. A significant amount of the historical toxicity data for zinc includes concentrations that exceeded zinc solubility. These data exhibited very different responses to chemical adjustment than data that were within solubility limits. Toxicity data that were within solubility limits showed evidence of both zinc complexation, and zinc-proton competition and could be well described by a chemical equilibrium approach such as that used by the Zn BLM.

Extension of the biotic ligand model of acute toxicity to a physiologically-based model of the survival time of rainbow trout (Oncorhynchus mykiss) exposed to silver.

Chemical speciation controls the bioavailability and toxicity of metals in aquatic systems and regulatory agencies are recognizing this as they develop updated water quality criteria (WQC) for metals. The factors that affect bioavailability may be quantitatively evaluated with the biotic ligand model (BLM). Within the context of the BLM framework, the 'biotic ligand' is the site where metal binding results in the manifestation of a toxic effect. While the BLM does account for the speciation and complexation of dissolved metal in solution, and competition among the free metal ion and other cations for binding sites at the biotic ligand, it does not explicitly consider either the physiological effects of metals on aquatic organisms, or the direct effect of water chemistry parameters such as pH, Ca(2+)and Na(+) on the physiological state of the organism. Here, a physiologically-based model of survival time is described. In addition to incorporating the effects of water chemistry on metal availability to the organism, via the BLM, it also considers the interaction of water chemistry on the physiological condition of the organism, independent of its effect on metal availability. At the same time it explicitly considers the degree of interaction of these factors with the organism and how this affects the rate at which cumulative damage occurs. An example application of the model to toxicity data for rainbow trout exposed to silver is presented to illustrate how this framework may be used to predict survival time for alternative exposure durations. The sodium balance model (SBM) that is described herein, a specific application of a more generic ion balance model (IBM) framework, adds a new physiological dimension to the previously developed BLM. As such it also necessarily adds another layer of complexity to this already useful predictive framework. While the demonstrated capability of the SBM to predict effects in relation to exposure duration is a useful feature of this mechanistically-based framework, it is envisioned that, with suitable refinements, it may also have utility in other areas of toxicological and regulatory interest. Such areas include the analysis of time variable exposure conditions, residual after-effects of exposure to metals, acclimation, chronic toxicity and species and genus sensitivity. Each of these is of potential utility to longer-term ongoing efforts to develop and refine WQC for metals.