Environmental Toxicology and Chemistry, Vol. 34, No. 4, pp. 761–776, 2015 # 2015 SETAC Printed in the USA
Metal Mixture Modeling Evaluation EXPANDING METAL MIXTURE TOXICITY MODELS TO NATURAL STREAM AND LAKE INVERTEBRATE COMMUNITIES LAURIE S. BALISTRIERI,y CHRISTOPHER A. MEBANE,*z TRAVIS S. SCHMIDT,x and WENDEL (BILL) KELLERk yUS Geological Survey, and University of Washington, School of Oceanography, Seattle, Washington zUS Geological Survey, Idaho Water Science Center, Boise, Idaho xUS Geological Survey, Fort Collins Science Center, Fort Collins, Colorado kCooperative Freshwater Ecology Unit, Laurentian University, Sudbury, Ontario, Canada (Submitted 24 April 2014; Returned for Revision 9 July 2014; Accepted 23 November 2014) Abstract: A modeling approach that was used to predict the toxicity of dissolved single and multiple metals to trout is extended to stream benthic macroinvertebrates, freshwater zooplankton, and Daphnia magna. The approach predicts the accumulation of toxicants (H, Al, Cd, Cu, Ni, Pb, and Zn) in organisms using 3 equilibrium accumulation models that define interactions between dissolved cations and biological receptors (biotic ligands). These models differ in the structure of the receptors and include a 2-site biotic ligand model, a bidentate biotic ligand or 2-pKa model, and a humic acid model. The predicted accumulation of toxicants is weighted using toxicantspecific coefficients and incorporated into a toxicity function called Tox, which is then related to observed mortality or invertebrate community richness using a logistic equation. All accumulation models provide reasonable fits to metal concentrations in tissue samples of stream invertebrates. Despite the good fits, distinct differences in the magnitude of toxicant accumulation and biotic ligand speciation exist among the models for a given solution composition. However, predicted biological responses are similar among the models because there are interdependencies among model parameters in the accumulation–Tox models. To illustrate potential applications of the approaches, the 3 accumulation–Tox models for natural stream invertebrates are used in Monte Carlo simulations to predict the probability of adverse impacts in catchments of differing geology in central Colorado (USA); to link geology, water chemistry, and biological response; and to demonstrate how this approach can be used to screen for potential risks associated with resource development. Environ Toxicol Chem 2015;34:761–776. # 2015 SETAC Keywords: Metal mixture toxicity
Biotic ligand model
Stream benthic macroinvertebrate
Zooplankton
Geology
biological response using toxicant-specific weighting coefficients and a logistic equation. To provide a stronger foundation between water chemistry and toxicant accumulation, we next used measured accumulation of metal toxicants (Cd, Cu, Pb, and Zn) on trout gills during short-term waterborne exposures (0.75–24 h) and 96-h responses to constant-concentration, individual metal exposures and responses, rather than LC50 estimates, to develop a 2-site biotic ligand–multiple toxicant model. Predictions of toxicant accumulation from this model were combined with weighting coefficients and a logistic equation to evaluate the survival of trout exposed to individual metals and metal mixtures in a variety of synthetic and natural waters [10]. Although the model results with trout were encouraging, several additional questions were raised during that study, some of which are the subject of the present study. Specifically, the present study examines 1) whether metal mixture toxicity models can be applied to natural communities of aquatic organisms where biota likely experience longer-term and temporally variable water and dietary exposures to metals, and 2) how the conceptual and mathematical formulations of such models influence predictions of biological response. Biotic ligand models that predict effects of metals to aquatic organisms through interactions of cations with biological receptors typically are based on experimental evidence of short-term accumulation of metals by fish and their associated mortality (e.g., MacRae et al. [11] and Playle et al. [12]). If disruption of ionoregulation is the primary cause of mortality and if biochemical pathways for this process function similarly across all non–air-breathing, freshwater animals [13–15], then, conceptually, a biotic ligand model approach for predicting
INTRODUCTION
Significant advances in understanding metal toxicity to aquatic organisms have occurred in the last 2 decades, including the development of models that predict toxicity to various organisms during exposure to individual metals and mixtures of metals in diverse waters [1–4]. In addition to metal toxicants, hydrogen ions have a role both as direct toxicants and as controls on dissolved metal speciation and metal accumulation on organisms [4–7]. To assess the state of the art of modeling approaches, a Metal Mixture Modeling Evaluation (MMME) project workshop was conducted in Brussels, Belgium, in May 2012. The results of that workshop and related research are summarized in a collection of articles in the present issue [8]. Our approach for modeling metal mixture toxicity to aquatic organisms has evolved, in large part because of lessons learned at the MMME workshop. As discussed in Farley et al. [9], our initial approach was to develop a common set of equilibrium constants that defined interactions between a single type of biological receptor (i.e., biotic ligand) and cations (H, Na, Ca, Mg, Cd, Cu, Ni, Pb, and Zn) using water chemistry and 50% lethal concentration (LC50) estimates from single metal toxicity studies on rainbow and cutthroat trout. These constants were combined to predict accumulation of toxicants (Cd, Cu, Ni, Pb, and Zn) on the biotic ligand from metal mixtures. Then, predicted accumulation of toxicants in single and multiple metal solutions were related to observed * Address correspondence to [email protected]. Published online 5 December 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2824 761
762
Environ Toxicol Chem 34, 2015
L.S. Balistrieri et al.
Table 1. Summary ranges of selected dissolved data used in the model calculationsa
Daphnia magna Minimum Maximum Stream invertebrates Minimum Maximum Zooplankton Minimum Maximum
pH
Hardness (mg CaCO3/L)
DOC (mg/L)
Al (mg/L)
Cd (mg/L)
Cu (mg/L)
Ni (mg/L)
Pb (mg/L)
Zn (mg/L)
7.02 8.37
72 103
0.24 4.21
— —
Metal Mixture Modeling Evaluation EXPANDING METAL MIXTURE TOXICITY MODELS TO NATURAL STREAM AND LAKE INVERTEBRATE COMMUNITIES LAURIE S. BALISTRIERI,y CHRISTOPHER A. MEBANE,*z TRAVIS S. SCHMIDT,x and WENDEL (BILL) KELLERk yUS Geological Survey, and University of Washington, School of Oceanography, Seattle, Washington zUS Geological Survey, Idaho Water Science Center, Boise, Idaho xUS Geological Survey, Fort Collins Science Center, Fort Collins, Colorado kCooperative Freshwater Ecology Unit, Laurentian University, Sudbury, Ontario, Canada (Submitted 24 April 2014; Returned for Revision 9 July 2014; Accepted 23 November 2014) Abstract: A modeling approach that was used to predict the toxicity of dissolved single and multiple metals to trout is extended to stream benthic macroinvertebrates, freshwater zooplankton, and Daphnia magna. The approach predicts the accumulation of toxicants (H, Al, Cd, Cu, Ni, Pb, and Zn) in organisms using 3 equilibrium accumulation models that define interactions between dissolved cations and biological receptors (biotic ligands). These models differ in the structure of the receptors and include a 2-site biotic ligand model, a bidentate biotic ligand or 2-pKa model, and a humic acid model. The predicted accumulation of toxicants is weighted using toxicantspecific coefficients and incorporated into a toxicity function called Tox, which is then related to observed mortality or invertebrate community richness using a logistic equation. All accumulation models provide reasonable fits to metal concentrations in tissue samples of stream invertebrates. Despite the good fits, distinct differences in the magnitude of toxicant accumulation and biotic ligand speciation exist among the models for a given solution composition. However, predicted biological responses are similar among the models because there are interdependencies among model parameters in the accumulation–Tox models. To illustrate potential applications of the approaches, the 3 accumulation–Tox models for natural stream invertebrates are used in Monte Carlo simulations to predict the probability of adverse impacts in catchments of differing geology in central Colorado (USA); to link geology, water chemistry, and biological response; and to demonstrate how this approach can be used to screen for potential risks associated with resource development. Environ Toxicol Chem 2015;34:761–776. # 2015 SETAC Keywords: Metal mixture toxicity
Biotic ligand model
Stream benthic macroinvertebrate
Zooplankton
Geology
biological response using toxicant-specific weighting coefficients and a logistic equation. To provide a stronger foundation between water chemistry and toxicant accumulation, we next used measured accumulation of metal toxicants (Cd, Cu, Pb, and Zn) on trout gills during short-term waterborne exposures (0.75–24 h) and 96-h responses to constant-concentration, individual metal exposures and responses, rather than LC50 estimates, to develop a 2-site biotic ligand–multiple toxicant model. Predictions of toxicant accumulation from this model were combined with weighting coefficients and a logistic equation to evaluate the survival of trout exposed to individual metals and metal mixtures in a variety of synthetic and natural waters [10]. Although the model results with trout were encouraging, several additional questions were raised during that study, some of which are the subject of the present study. Specifically, the present study examines 1) whether metal mixture toxicity models can be applied to natural communities of aquatic organisms where biota likely experience longer-term and temporally variable water and dietary exposures to metals, and 2) how the conceptual and mathematical formulations of such models influence predictions of biological response. Biotic ligand models that predict effects of metals to aquatic organisms through interactions of cations with biological receptors typically are based on experimental evidence of short-term accumulation of metals by fish and their associated mortality (e.g., MacRae et al. [11] and Playle et al. [12]). If disruption of ionoregulation is the primary cause of mortality and if biochemical pathways for this process function similarly across all non–air-breathing, freshwater animals [13–15], then, conceptually, a biotic ligand model approach for predicting
INTRODUCTION
Significant advances in understanding metal toxicity to aquatic organisms have occurred in the last 2 decades, including the development of models that predict toxicity to various organisms during exposure to individual metals and mixtures of metals in diverse waters [1–4]. In addition to metal toxicants, hydrogen ions have a role both as direct toxicants and as controls on dissolved metal speciation and metal accumulation on organisms [4–7]. To assess the state of the art of modeling approaches, a Metal Mixture Modeling Evaluation (MMME) project workshop was conducted in Brussels, Belgium, in May 2012. The results of that workshop and related research are summarized in a collection of articles in the present issue [8]. Our approach for modeling metal mixture toxicity to aquatic organisms has evolved, in large part because of lessons learned at the MMME workshop. As discussed in Farley et al. [9], our initial approach was to develop a common set of equilibrium constants that defined interactions between a single type of biological receptor (i.e., biotic ligand) and cations (H, Na, Ca, Mg, Cd, Cu, Ni, Pb, and Zn) using water chemistry and 50% lethal concentration (LC50) estimates from single metal toxicity studies on rainbow and cutthroat trout. These constants were combined to predict accumulation of toxicants (Cd, Cu, Ni, Pb, and Zn) on the biotic ligand from metal mixtures. Then, predicted accumulation of toxicants in single and multiple metal solutions were related to observed * Address correspondence to [email protected]. Published online 5 December 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2824 761
762
Environ Toxicol Chem 34, 2015
L.S. Balistrieri et al.
Table 1. Summary ranges of selected dissolved data used in the model calculationsa
Daphnia magna Minimum Maximum Stream invertebrates Minimum Maximum Zooplankton Minimum Maximum
pH
Hardness (mg CaCO3/L)
DOC (mg/L)
Al (mg/L)
Cd (mg/L)
Cu (mg/L)
Ni (mg/L)
Pb (mg/L)
Zn (mg/L)
7.02 8.37
72 103
0.24 4.21
— —
