SEARCH

SEARCH BY CITATION

Keywords:

  • Bioaccumulation;
  • Laboratory;
  • Field;
  • Model;
  • Guideline

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

In the regulatory context, bioaccumulation assessment is often hampered by substantial data uncertainty as well as by the poorly understood differences often observed between results from laboratory and field bioaccumulation studies. Bioaccumulation is a complex, multifaceted process, which calls for accurate error analysis. Yet, attempts to quantify and compare propagation of error in bioaccumulation metrics across species and chemicals are rare. Here, we quantitatively assessed the combined influence of physicochemical, physiological, ecological, and environmental parameters known to affect bioaccumulation for 4 species and 2 chemicals, to assess whether uncertainty in these factors can explain the observed differences among laboratory and field studies. The organisms evaluated in simulations including mayfly larvae, deposit-feeding polychaetes, yellow perch, and little owl represented a range of ecological conditions and biotransformation capacity. The chemicals, pyrene and the polychlorinated biphenyl congener PCB-153, represented medium and highly hydrophobic chemicals with different susceptibilities to biotransformation. An existing state of the art probabilistic bioaccumulation model was improved by accounting for bioavailability and absorption efficiency limitations, due to the presence of black carbon in sediment, and was used for probabilistic modeling of variability and propagation of error. Results showed that at lower trophic levels (mayfly and polychaete), variability in bioaccumulation was mainly driven by sediment exposure, sediment composition and chemical partitioning to sediment components, which was in turn dominated by the influence of black carbon. At higher trophic levels (yellow perch and the little owl), food web structure (i.e., diet composition and abundance) and chemical concentration in the diet became more important particularly for the most persistent compound, PCB-153. These results suggest that variation in bioaccumulation assessment is reduced most by improved identification of food sources as well as by accounting for the chemical bioavailability in food components. Improvements in the accuracy of aqueous exposure appear to be less relevant when applied to moderate to highly hydrophobic compounds, because this route contributes only marginally to total uptake. The determination of chemical bioavailability and the increase in understanding and qualifying the role of sediment components (black carbon, labile organic matter, and the like) on chemical absorption efficiencies has been identified as a key next steps. Integr Environ Assess Manag 2012;8:42–63. © 2011 SETAC


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

Potential hazards and risks of new and existing substances are assessed by regulatory agencies worldwide according to their persistence (P), bioaccumulation (B), and toxicity (T) criteria (e.g., USEPA 1976; Environment Canada 2003; Council of the European Union 2006). Substances that meet these criteria are referred to as PBTs and are potentially subject to regulatory controls or further characterization to determine their harmfulness. Field studies can play a key role in the regulatory assessment of bioaccumulation criteria for specific chemicals. Some regulatory processes do not define how to include field data in risk assessments, as regulatory criteria are set only for bioconcentration factors (BCFs) derived from laboratory tests. Field data can effectively be used to supplement and validate laboratory observations, or, based on the more environmentally realistic conditions that field data represent, refute these observations. Despite these advantages, there are challenges in using field data in regulatory assessments due to the complex variables encountered in field studies that make them difficult to compare to simpler and often more precise laboratory studies. It is therefore important that the causes of variability surrounding data collected from the field be understood so that field sampling strategies can be better optimized to assess B criteria, and the appropriateness of a particular field study for regulatory assessments can be evaluated in a quantitative manner. In addition, as the regulatory community is increasing its reliance on the use of model approaches toward B assessment and as screening tools, it is important to understand uncertainty of model output. The models vary widely from simple correlative models (e.g., BCF and octanol–water partitioning coefficient [KOW] regressions) to more complicated, multicompartment physiologically based toxicokinetic models (Nichols et al. 2004) all having error associated with them which may be propagated as a result of error in parameter measurement and variation associated with model inputs (Jørgensen 1990). Model predictions can deviate substantially from what is measured either in the laboratory or the field which may be a result of a conceptual flaw in the model structure or it may be due to inappropriate model parameterization relative to the empirical situation being simulated. It is therefore important to place such deviations in model performance in context by evaluating model output uncertainty, understanding which parameters contribute most toward model uncertainty (sensitivity analysis) and whether or not the magnitude of model parameters and inputs used are consistent with the true range of these terms under actual laboratory or field conditions.

Recent collaborative international efforts have summarized the state of the science with respect to PBT assessment of organic chemicals and model approaches to estimate chemical bioaccumulation of these substances (e.g., Gobas et al. 2009; Nichols et al. 2009; Weisbrod et al. 2009). Several areas were identified as knowledge gaps, including the extrapolation of laboratory-generated bioaccumulation data in light of its relevance to field measurements, the ability of bioaccumulation models to predict high variation in bioaccumulation metrics commonly measured in field studies, and the interpretation of field measurements relative to regulatory numeric and descriptive criteria. A follow-up workshop involving scientists from government, academia, and industry was organized to address these remaining issues. This article presents findings from 1 of 3 working groups of this workshop.

Bioaccumulation of contaminants occurs through multiple exposure routes including dietary assimilation, transport across respiratory surfaces and dermal absorption. When uptake from all exposure routes are high relative to elimination, an increased chemical concentration in the organism compared to the surrounding environment occurs (Gobas and Morrison 2000). Bioaccumulation may be expressed in terms of several endpoints such as the bioconcentration factor (BCF: L/kg wet weight [wet wt]), bioaccumulation factor (BAF: L/kg wet wt), biota–sediment accumulation factor (BSAF: kg lipid/kg organic matter), biota–suspended solids accumulation factor (BSSAF: kg lipid/kg organic matter), biomagnification factor (BMF: laboratory-based: kg dry weight [dry wt]/kg dry wt; field-based: kg wet wt/kg wet wt) and the trophic magnification factor (TMF: unitless) (Burkhard et al., this issue2012). In a regulatory context, a substance with a field measured BAF or a laboratory-derived BCF greater than 5000 L/kg wet wt or a log KOW above 5 is, in principle, considered bioaccumulative (Stockholm Convention 2001). These endpoints have been shown to vary among field and laboratory studies (Arnot and Gobas 2006), depending on experimental and empirical conditions under which bioaccumulation assessments were completed. Understanding differences between laboratory and field studies and developing bioaccumulation models that are capable of describing field variability in bioaccumulation metrics begins with an understanding of the constraints of experimental settings relative to field conditions and whether or not such constraints are incorporated into the bioaccumulation model itself. Laboratory experiments often include optimal conditions like a constant environment, constant dose conditions or exposure to homogeneous contaminated media (water or food), food in abundance, a single exposure route, and study of one contaminant at a time to simplify and to assign cause and consequence. On the simplified end of the scale, standardized test protocols are used to assess PBT of contaminants using BCF as an endpoint, even for hydrophobic contaminants with low water solubility (e.g., fish tests). However, because BCF quantifies bioaccumulation from water and neglects accumulation from diet, using BCF as an endpoint for hydrophobic compounds may not be appropriate (e.g., Gobas et al. 2009) and may underestimate bioaccumulation potential for the chemical, thus adding to the discrepancy between laboratory and field measures. In the field, environmental (e.g., wind, current, temperature, light, turbidity, substrate, salinity, alterations of habitat characteristics) and biological conditions (e.g., physiology including metabolic rate, reproduction and growth rate, food quality and availability, foraging behavior), and ecological factors (e.g., organism migrations and movements, feeding ecology and avoidance behavior) may vary both spatially and temporally. These factors may in turn both directly and indirectly influence parameters that control bioaccumulation including exposure route and concentrations, potential for multiple contaminant exposure, bioavailability, rates of uptake and loss, and potential for steady state. Clearly, the degree of complexity increases dramatically from simple, fully standardized laboratory experiments (e.g., OECD 305 guideline experiments) over laboratory mesocosm studies to studies analyzing data from field observations. Consequently, a large divergence can often be observed for specific substances between BCF/BAF determined using in vivo laboratory tests and in situ data especially with increasing KOW (Arnot and Gobas 2006; Weisbrod et al. 2009). For example, Fisk et al. (1998) found that calculated BMFs for organic contaminants (OCs) in juvenile rainbow trout were lower in laboratory experiments compared to field measures and suggested that this was related to the content of lipid in the diet. Alterations in dietary lipid can impact both diet digestibility and chemical assimilation efficiency (Drouillard and Norstrom 2003). In addition, it was found that food composition and fish size affected bioaccumulation of polychlorinated biphenyl (PCB) congeners (Liu et al. 2010). Conversely, a study of tadpoles showed a higher uptake of cypermethrin in the laboratory compared to the field due to adsorption of cypermethrin onto suspended solids, sediment, and aquatic plants in the field (Greulich and Pflugmacher 2004). Thus, both environment and biology are important in explaining why model simulations and laboratory and field measures of bioaccumulation often do not align, and why one measure (model, field, or laboratory) is not always greater than the others.

This article evaluates how variation in physicochemical, physiological, ecological, and environmental parameters influences the variation of bioaccumulation measures generated by laboratory and field studies. The overall goal is to identify the relative importance of the different underlying processes in driving this variation in bioaccumulation and to develop guidelines for minimizing variation of bioaccumulation in field measurements, so that assessments of chemical bioaccumulation potential against bioaccumulation criteria are conservative and account for normal variation likely to be encountered in the field. To achieve this goal, we first reviewed the current state of the science of bioaccumulation (Supplemental Data 1) to select the most important contributors to the observed variance in bioaccumulation measures in the field. Second, we used mechanistic bioaccumulation models, uncertainty and sensitivity analyses to evaluate and rank variables with respect to their contribution to bioaccumulation variability. This approach provided concentration estimates for a given set of predefined conditions and served as a tool to identify those model parameters contributing most to the variation in bioaccumulation. Third, we compared the simulation output to relevant empirical data in order to gauge the degree to which predicted variation in simulation output matched empirically measured field variation for selected bioaccumulation metrics, discussed model uncertainty and identified knowledge gaps. Finally, we assessed the applicability of models in bioaccumulation assessment and identified key parameters that, when measured and reported in field studies, will provide a means by which field data from different systems can be more appropriately compared with each other and with laboratory studies.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

Selection of chemicals and model organisms

The model chemicals were selected to present a few key differences with respect to environmental partitioning and bioaccumulation behavior. In terms of hydrophobic partitioning to organic matter and lipid reservoirs, pyrene and PCB-153 represent a range of moderate to highly hydrophobic chemicals with log KOW values of 5.2 and 6.9, respectively. Black carbon (BC), formed through the incomplete combustion of fossil fuels, biofuel, and biomass, and emitted in both anthropogenic and naturally occurring soot, has a strong binding capacity for organic chemicals. As for bioavailability, the selected chemicals differ in their relative association with BC in that the planar aromatic structure of pyrene results in stronger association with environmental BC compared to lipid than is the case for the more bulky PCB-153 molecule (Koelmans et al. 2006). The 3rd difference relates to persistence, i.e., degradation and susceptibility to biotransformation within the animal tissues. Biotransformation of PCB-153 is found to be negligible in our test organism groups (Gobas et al. 1989; Drouillard et al. 2001; Goerke and Weber 2001; Paterson et al. 2007). In contrast, biotransformation of pyrene and other polycyclic aromatic hydrocarbons (PAHs) have been reported to be at higher rates than PCB-153 in polychaetes (Kane-Driscoll et al. 1996; Selck et al. 2003b; Jorgensen et al. 2008), fish (Namdari and Law 1996) and birds (Ronis and Walker 1989) but is unlikely to occur in mayflies (Landrum and Poore 1989).

We selected 4 groups of organisms representing different habitats: 3 aquatic (infaunal, epifaunal and/or pelagic, and pelagic) and 1 terrestrial. The aquatic organisms included deposit-feeding polychaetes, mayfly larvae (Hexagenia sp.), and fish (Perca flavescens, yellow perch). Aside from habitat, these organisms also differ with regard to biotransformation capability of organic contaminants (see above). Whereas the aquatic invertebrates have a relatively uniform diet of ingested sediment and/or seston, the yellow perch generally represents aquatic vertebrates that possess a complex diet. The aquatic species are exposed to chemicals in water and via ingested food and/or sediment. The terrestrial group includes a top avian predator represented by the little owl (Athene noctua), which is exposed to organic compounds through food web accumulation and complex diet components that include both vertebrates and invertebrates. In addition, chemical exposures via respired air may also contribute to bioaccumulation in terrestrial species (little owl). The 4 selected organism groups differ in routes of uptake (terrestrial versus aquatic, vertebrate versus invertebrate) and their ability to biotransform organic compounds (PCB-153 versus pyrene for owl, polychaete, and fish).

Model approach

Probabilistic scenario studies were performed to quantify propagation of variance in the traditional bioaccumulation metrics. Using a modeling approach, our goal was to prescribe values for different chemical/physical factors (e.g., chemical KOW, chemical partitioning to proximate components of the animal, its diet, and sediment), physiological factors (e.g., organism respiratory ventilation rate, potential for biotransformation, feeding rate, digestive physiology, and so forth), environmental conditions (e.g., contaminant concentration in food items and respired media) and ecological factors (e.g., foraging range, dietary proportions, and so forth) and their associated measures of variation to develop a range of bioaccumulation scenarios encountered in nature. Indeed, modeling approaches have provided a framework for assessing and better understanding the sources of variation in bioaccumulation among various pollutants and taxonomic species (Thomann and Connolly 1984; Gobas and Mackay 1987; Luoma et al. 1992; Wang and Fisher 1999; Arnot and Gobas 2004; Luoma and Rainbow 2005; Hauck et al. 2007; Moermond et al. 2007; Wang and Rainbow 2008; Ciavatta et al. 2009; De Laender et al. 2010).

Model simulations for the 2 chemicals (PCB-153 and pyrene) and the 4 organism groups were performed using a general 1-compartment bioaccumulation model designed for organic compounds. Further details about the bioaccumulation model and its predictive algorithms can be found in Arnot and Gobas (2004). The Arnot and Gobas (2004) model was selected because of the generality of its framework, which allowed ready adaptation to the different animal species and chemicals being considered for simulation. This model has a long history, originating in its general framework from Thomann and Connolly (1984) and undergoing several iterations with advances in parameter information and predictive algorithms (e.g., Clark et al. 1990; Gobas 1993; Morrison et al. 1997). The general equation for the model is given as:

  • equation image(1)

Where Corg (ng · g−1 wet wt), Cr (ng · mL−1) and Cfood (ng · g−1 wet wt) refer to chemical concentrations in the animal, respired media and ingested food, respectively. The rate coefficients are denoted as k values and for simplicity in model, simulations were assumed to follow first-order kinetics. The rate coefficients are individually defined as follows: kv is the uptake rate coefficient from respired media (mL · g−1 wet wt · d−1), kfood is the uptake rate coefficient from ingested food (g food · g−1 wet wt · d−1), k2 is the elimination rate coefficient (d−1) across respiration surfaces, kex is the fecal elimination rate coefficient (d−1), km is the metabolic biotransformation rate coefficient (d−1) and kg is the growth dilution coefficient (d−1) (Table 1). Other processes such as dermal absorption and elimination, molting losses, or reproductive losses (maternal deposition to eggs) were not included in model simulations because of lack of species-specific data and/or confidence about the degree of uncertainty associated with these parameters. Species- and chemical-specific rate coefficients were estimated on the basis of equations summarized in Supplemental Data 2.

Table 1. Rate coefficient calculation methods for a general toxicokinetic model
Rate termCalculationCalculation Parameters
  1. BW = body weight.

kvQv·EwQv = air or water ventilation rate (mL·g−1 wet BW·d−1)
  Ew = chemical transfer efficiency term (unitless)
kfoodQfood·EfoodQfood = animal feeding rate (g·g−1 wet BW d−1)
  Efood  = chemical transfer efficiency term (unitless)
k2Qv·Ew·Kmath imageQv = air or water ventilation rate (mL·g−1 wet BW d−1)
  Ew = chemical transfer efficiency term (unitless)
  Korg,w = organism/water partition coefficient (mL·g−1 wet BW)
kexQex·Eex·Kmath imageQex = fecal production rate (g·g−1 wet BW·d−1)
  Eex = chemical transfer efficiency term, set = Efood (unitless)
  Kmath image = organism/feces partition coefficient (g·g−1 wet BW)
kmAssigned value 
kgAssigned value 

For some organism simulations, multiple diet items (up to 3 diet items) were included in the model and/or separate contributions of respired sediment porewater and overlying water to chemical uptake. The expanded model, solved for steady state, is thus given as:

  • equation image(2)

Where the terms p(o,w), p(p,w) refer to the proportion of overlying water respired and proportion of porewater respired by the animal such that p(o,w) + p(p,w) = 1. This term was substituted to a single air ventilation and air concentration term for the little owl. Qv express the ventilation rate, flow of water across gills and integument (mL/g body wt/d). Similarly for organisms having multiple diets, the proportion of a given diet to the total food ingestion is provided by p(i). Each diet has its own chemical assimilation efficiency term (Efood(i)) and the same value was assigned to the fecal egestion efficiency term (Eex(i)) for the equivalent diet item. The term Ew refers to the chemical exchange efficiency across the gills and/or integument. The organism–water partition coefficient (Korg,w), or bioconcentration factor, was estimated according to:

  • equation image(3)

Where pw(org), plip(org) and pNLOM(org) are the proportions of water, lipid, and nonlipid organic matter in the animal, respectively. KOW is the n-octanol–water partition coefficient and φNLOM is the partition capacity of nonlipid organic matter in the organism relative to octanol (Debruyn and Gobas 2006).

For fecal elimination, each dietary food item contributes to a diet-specific fecal production rate based on the digestibility of proximate components in the diet as follows:

  • equation image(4)

Where AEw, AElip, and AENLOM represent diet digestibility (unitless) of water, lipid, and nonlipid organic matter components, respectively, in the ingested food item. Similarly, the proximate composition of feces produced from each digested diet type results in a diet-specific organism–feces partition coefficient (Korg,ex) described below in Equation 5:

  • equation image(5)

Additional modifications were performed for simulations with benthic organisms (mayfly larvae and polychaetes) to consider the effect of chemical distribution between individual sediment components: labile organic matter (LOM), black carbon (BC), inorganic matter (IM) and porewater (PW). Any chemical associated with porewater was considered bioavailable via respiratory surfaces as well as from the small amount of porewater ingested as part of sediment feeding activity. Chemicals associated with LOM, BC, and IM were treated as separate diet components with different chemical assimilation and digestibility terms. Among the latter components, only LOM was considered to be partially digested and assimilated such that it provided nutritional value to the animal. It was further assumed that animals did not engage in selective feeding of sediment components. Chemical partitioning to BC is nonlinearly related to porewater concentration and often estimated using a Freundlich equation (Koelmans et al. 2006). Assuming the chemical distribution among sediment components is in equilibrium, individual sediment component concentrations were estimated by mass balance and the difference in their relative partitioning capacities according to:

  • equation image(6)

Where pLOM, pIM, and pBC represent the proportions of labile organic matter, inorganic matter (pIM and φIM), and BC, respectively. The term φIM is the partition capacity of inorganic matter in sediments relative to octanol. The term KBC is the Freundlich constant and nBC the Freundlich coefficient (Accardi-Dey and Gschwend 2002; Koelmans et al. 2006). In the uncertainty analysis for mayflies and polychaetes, uncertainty in bulk sediment concentration (i.e., LOM, BC, IM, PW) was covered by varying Cw(pw) (see discussion in Supplemental Data 2).

We used equations from 3 common bioaccumulation endpoints: BAF (L/kg wet wt; Equation 7), BMF (kg wet wt/kg wet wt; Equation 8) and BSAF (kg l.w./kg org. w, Equation 9) to assess the contribution of the various factors to the observed differences between laboratory and field measures of bioaccumulation.

  • equation image(7)
  • equation image(8)

In Equations 7 and 8, chemical concentrations in the organism (Corg), water (Cw), and food (Cfood) are expressed on a wet weight basis, although lipid-normalized BAF and BMFs can be calculated by dividing the BAF by the proportion of lipid (plipid(org)) in the organism or by multiplying the BMF by the ratio of the proporption of lipid in the food (plipid(food)) over plipid(org).

A 3rd commonly used bioaccumulation endpoint is the BSAF, which by convention is always reported on a lipid (organism concentration) and organic carbon (sediment concentration; Csed) normalized basis. The BSAF can be related to rate constants via the following expression:

  • equation image(9)

Where pOC(sed) is the proportion of organic carbon in bulk sediments, pfood is the proportion of diet composed of nondetrital material and psed is the proportion of diet consisting of ingested bulk sediments.

Additional information on modeling BC sediment water partitioning uncertainty, modified equations for benthic feeding species are documented in Supplemental Data 2 for mayflies and polychaetes. Supplemental Data 2 further provides documentation of model parameters, parameter values, and literature sources for selected values and ranges for each of the organism–chemical simulation trials.

Uncertainty analysis

For each of the parameters, we examined the uncertainty in 2 steps. First, a literature search was performed to identify parameters considered as main drivers of variation. The results and motivations for this selection are provided as Supplemental Data 1. In the second step, model uncertainty and sensitivity was quantified using Monte Carlo simulations by assigning probability distributions to the most influential parameters as identified in step 1 (Supplemental Data 2). Among the different models, between 29 and 38 model inputs and parameters of the total 56 had variability associated with them. An overview of motivation, parameters and assumed parameter distributions used in probabilistic model simulations is provided in Supplemental Data 2. Monte Carlo simulations used 10 000 iterations for each organism–chemical simulation trial using Crystal Ball software (Goldman 2002) interfaced with a Microsoft Excel spreadsheet. Overall model uncertainty was evaluated by examining the distribution (mean, range, or standard deviation) and percentiles (1%, 5%, 25%, 50%, 75%, 95%, and 99%) of model output trials across simulation iterations. Model sensitivity analysis was performed to determine which parameters contribute the greatest degree of variation in model output. Although some model parameters are likely to be correlated to one another, i.e., they covary, they were treated as independent when running the model uncertainty and sensitivity analysis. Examples of potentially covarying model parameters include respiratory ventilation rates and animal feeding rates which are both scaled to animal metabolic rate (Drouillard et al. 2009), potential interactions between chemical assimilation efficiencies across respiratory surfaces and air or water ventilation rate (Drouillard et al. 2009) or chemical assimilation efficiencies from food with animal feeding rate (Drouillard and Norstrom 2003). These potential limitations of the model uncertainty analysis are expected to inflate variation in model output relative to more complex models that explicitly consider covarying model parameters as part of the probabilistic assessment. However, such influences are only likely to become important if multiple covarying parameters are found to be strong contributors to overall model variation. All results in the text are presented as mean ± standard deviation (SD).

Organism-specific simulation properties

Mayfly

Model simulations focused on Hexagenia sp., an insect common in the Laurentian Great Lakes and frequently monitored for organic contaminant bioaccumulation (Gobas et al. 1989; Corkum et al. 1997; Gewurtz et al. 2000). In the present study, we focused on the aquatic larvae stage of mayflies and included 2 food items: organic matter associated with sediment and seston. Upon reaching maturity, mayfly nymph undergo incomplete metamorphosis into a flying adult stage (Merritt and Cummins 1996). Because mayflies are extensively fed upon by fish during emergence, we used mayfly as diet for the fish simulations (see below). Mayflies were selected because this species has been well investigated for its feeding ecology and organic contaminant toxicokinetics (Landrum and Poore 1988; Drouillard et al. 1996). Previous studies suggest times to steady state for PCBs on the order of 30 d (Drouillard et al. 1996), and as such, the simulations focused on the larval naiad stage prior to metamorphosis, which lasts from 3 months to 2 y depending on species (Merritt and Cummins 1996). Because it is one of the benthic-feeding species used for model simulations, additional equations describing chemical distribution to sediment components were applied as described above and in the supplemental information on mayfly simulations (Supplemental Data 2). In addition, unlike the other 3 species, there is little evidence for Hexagenia being capable of biotransforming PAHs. Model simulations were set up using contaminant concentrations in sediment and water from the Detroit River and Lake Erie (see below).

Polychaete

Deposit-feeding polychaetes were chosen as an indicator species for the variability in contaminant accumulation from the sediment-associated pool, and as such included only one food item: sediment organic matter. Deposit-feeding polychaetes are stationary, living in tubes, and burrow through sediment, processing several times their own body weight in wet sediment (i.e., containing both porewater and sediment particles) per day obtaining food from the organic fraction of ingested sediment particles (Lopez and Levinton 1987; Kofoed et al. 1989). Both respiratory (overlying water and porewater) and digestive (from different sediment components) chemical accumulation were considered in the simulations. The range in ventilation used in the present study reflects that ventilation of burrows vary considerably among polychaetes and greatly depend on tolerance to anoxic conditions (Supplemental Data 2). Both sediment organic matter (OM) quantity and quality vary seasonally. We included variation in OM quantity in the simulations based on field studies for pyrene (El Nemr et al. 2007) and PCB-153 (Gevao et al. 2005), respectively, where the contaminant and OM concentration were determined in surface sediment. The sediment OM quality was kept constant in the simulations. Because sediment OM quality and chemical assimilation efficiency is linked in polychaetes, this needs to be considered when interpreting the results. Sediment concentrations are as a general rule presented on a dry weight basis to avoid the influence of varying water contents in sediments that may range from approximately 10% in gravel and/or sand up to as much as 85% in clay. Because of the model structure, the porewater concentrations of PCB-153 and pyrene were estimated from a dry weight sediment concentration assuming a water content of 35%, which is seen in most estuarine environments (Supplemental Data 2). Feeding rate is usually also presented as amount dry weight sediment processed per day. Again, the mean ± SD was transformed to wet weight sediment, assuming a water content of 35%. Chemical assimilation from sediment labile organic matter (LOM), and sediment BC was estimated from literature values of PCB-153 and pyrene when available as well as from other contaminants (e.g., PAHs, benzo[a]pyrene, hexachlorobenzene), due to lack of information (see reasoning for polychaetes in Supplemental Data 2). Assimilation from inorganic particles (IM) and from porewater (PW) associated with ingested sediment were both set to zero. Because assimilation efficiency of organic matter depends on organic matter type (bulk sediment OM versus phytoplankton and/or bacteria) in polychaetes, we included a high variation around OM assimilation efficiency (AE) for the simulations.

Yellow perch

Fish were included in the present study because of their demonstrated high contaminant bioaccumulation and biomagnification potentials and because organic chemical bioaccumulation in fish may represent a substantive exposure source of contaminants to humans via the consumption of fish (e.g., Weisbrod et al. 2009 and references therein). Model simulation trials focused on adult yellow perch (Perca flavescens) because the bioenergetics, seasonal variation in tissue lipids, and chemical toxicokinetics for PCB-153 have been well described for this species (Paterson et al. 2007; Drouillard et al. 2009). For PCB-153, which requires multiple years to reach steady state, variation in fish proximate contents and bioenergetic parameters were considered over an annual cycle. Also, because this species feeds heavily on benthic invertebrates, we could use the mayfly model simulation output to gauge likely variation in diet concentrations for one of its major diet components (mayfly: diet 2). The other 2 diets included young of the year fish (diet 1) and chironomids (diet 3). Pyrene is strongly subject to rapid biotransformation kinetics with half-lives <1 d in fish (Namdari and Law 1996). As such, model simulations for this chemical were restricted to a short time period (10 d) associated with fish optimal temperatures. Variation in fish proximate composition, feeding, and ventilation rates are expected to be minimized under these conditions. As for mayfly, the model simulations were set up using data from the Detroit River (see below). Parameter values are presented in Supplemental Data 2.

Little owl

In the present study, the little owl (Athene noctua) was used as an indicator species for the variability of accumulation in birds of prey in general, in particular because of the abundance of information on its food choice and ecology. The little owl simulations were set up such that chemical exposures occurred via respired air and ingested food (Supplemental Data 2). The diet of the little owl varies strongly between the different regions of its distribution. To reflect the diet composition in Northwest Europe, which consists of earthworms, beetles, other insects, small mammals, birds, and sometimes even amphibians and bats (Van Zoest and Fuchs 1988), we selected 3 diet items for the simulations: earthworms (diet 1), beetles (diet 2), and field mice (diet 3). The diet proximate compositions were well established (Hounsome et al. 2004), as were representative PCB-153 and pyrene concentrations in the different diet items (Van den Brink et al. 2001, 2003). Dietary proportions are considered an ecological factor in model simulations. Recognizing the very large differences in time to steady state between the 2 chemicals (measured in years for PCB-153 in contrast to 1–3 d for pyrene; Drouillard et al. 2001) the uncertainty in diet proportions was assigned differently for each compound. For PCB-153, a very large range of diet 1 proportions (0%–60%, mean of 45%) and diet 2 proportions (0%–40%, mean of 19%) was assigned in the simulations, with diet 3 proportion being calculated as the remainder of feeding activity after subtracting diet 1 and diet 2. The large range in diet composition for this chemical was developed to consider both individual differences in diet choices as well as seasonal differences in diet availability. For pyrene, the same mean diet proportions were used but the error ranges were constrained (diet 1 ranged from 23–52% and diet 2 ranged from 10–29%). In this case, seasonal variation in diet availability was not considered and the truncated error in diet proportions refers mainly to individual variation in diet choice. Birds exhibit negligible biotransformation of PCB-153 (Drouillard et al. 2001) but rapidly biotransform PAHs (Ronis and Walker 1989). A pyrene biotransformation rate for birds could not be derived from the literature. Instead, a value at the upper range of PAH biotransformation rate constants described for mammals (3 d−1 for chrysene; Hendriks 1995) was used as the most likely value in model simulations. The range of this parameter in the uncertainty analysis had a lower limit equivalent to the most likely value of km used for fish (1 d−1) and an upper limit of 10-fold higher than fish in keeping with the metabolic capacity differences commonly demonstrated between birds and lower vertebrates (Ronis and Walker 1989).

Model validation using empirical data

Model simulation output was compared to relevant field data in order to gauge the degree to which predicted variation in simulation output matched empirically measured field variation for selected bioaccumulation metrics. Empirical data used for comparisons included published data as well as unpublished raw data from the authors or reported summary statistics of bioaccumulation metrics. Applicable empirical data sets could not be found for all organism–chemical combinations. The model trials from Monte Carlo simulations and empirical data were used to generate box and whisker plots to present 1st, 5th, 25th, 50th (median and mean), 75th, 95th, and 99th percentiles of each metric for comparisons.

For mayflies and yellow perch, empirical field data were obtained from the Great Lakes Institute for Environmental Research (GLIER), University of Windsor, Canada (Drouillard 2010; Kashian et al. 2010). This data included samples of organism concentrations or benthic invertebrate BSAFs collected from the Detroit River, Michigan, USA, and were considered consistent with model simulations for these organism–chemical trials because the simulations used Detroit River water (river-wide averages from 1998–2008) and sediment (1998 river-wide survey) (Drouillard et al. 2006, 2010). The yellow perch data consisted of 24 fish samples (total lengths ranging from 15–26 cm) collected from 4 locations in the Detroit River from 2000 to 2003. The data included lipid contents and PCB-153 concentrations in skinless fillet samples. Matched data were not available on yellow perch diets to compute field BMFs in this species and therefore comparisons were made between empirical data and model output of steady state fish concentration estimates. Yellow perch fillet sample data were converted to whole-body residue estimates by lipid normalizing the PCB-153 concentration in fillet samples and then multiplying by the assumed whole-body lipid fraction of 0.065 (Drouillard et al. 2009). The mayfly data consisted of a GLIER data set of 12 BSAFs calculated from pooled samples of Hexagenia spp. and matched sediments obtained from 9 sampling locations in the Detroit River collected during July 2008 (Drouillard 2010).

The polychaete data were obtained from Nesto et al. (2010) and Ruus et al. (2005). Nesto and colleagues reported 8 matched field BSAF estimates for PCB-153 and pyrene in the polychaete species Perinereis rullieri from 2 sampling stations in the lagoon of Venice, Porto Marghera, Italy. In the above case, BSAF metrics were reported at the 2 stations during 4 different sampling months. Ruus et al. (2005) reported laboratory bioassay-derived BSAFs (pyrene, n = 6 estimates; PCB-153, n = 3 estimates) in the polychaete species Nereis diversicolor exposed to 2 harbor sediments from Kristiansand Harbor, Norway, following 28-d laboratory exposures. Because of the limited sample numbers for PCB-153 in the laboratory bioassay data set, box and whisker plots were only generated for pyrene using both field and bioassay BSAF data.

For the little owl, an appropriate data set on BMFs in adults of this species could not be located. A surrogate data set on avian PCB BMFs for guillemots (Uria aalge) was used for comparison purposes. This is a marine species, mainly feeding on small fish and marine pelagic invertebrates (Mehlum 2001). The difference in ecology between these 2 species may hamper the comparison. The main reason for choosing the little owl as a model species is that relatively much is known about its ecology in Northwestern Europe, enabling the compilation of a case study for modeling purposes, e.g., detailed information on diet composition, prey concentrations, and so forth. For the guillemot, much empirical information is available on the derivation of the BMFs from a specific study by Lundstedt-Enkel et al. (2005), making it an excellent study to compare the modeling results, whereas key parameters for the model are relatively less well known. Despite the differences between the species, comparison of their specific accumulation is feasible, because the most important route of uptake for both species is through the diet. Hence, this comparison is based on the same process, namely, food web accumulation, which in our opinion justifies the comparison between the 2 species even though they may seem to be very different. The guillemot BMF data was obtained from Lundstedt-Enkel et al. (2005), based on measured concentrations of PCB-153 in 29 guillemot samples and 72 herring (Clupea harengus) samples. These authors used a statistical bootstrapping technique to compute randomly sampled BMF ratios from individual guillemot and prey samples, to generate a data set of 50 000 BMF combinations on which to report summary statistics. From this data set, Lundstedt-Enkel et al. (2005) reported a mean guillemot BMF of 31.2 ± 21.4, geometric mean of 24.7 and BMF range of 3.04 to 158. These summary statistics were used in conjunction with a new set of Monte Carlo simulations, assuming a lognormal distribution, to generate a surrogate empirical data set on which to compute quartiles for box and whisker plots and to compare with the little owl model simulation output.

RESULTS AND DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

Below, we present and discuss results from the mechanistic bioaccumulation models and uncertainty analyses by ranking variables with respect to their contribution to bioaccumulation variability for each of the 4 organisms, and compare variation in simulation output with relevant empirical data. This is followed by a discussion on model uncertainty and future research needs. Subsequently, we discuss the applicability of models in bioaccumulation assessment and identify key parameters that, when measured and reported in field studies, will provide a means by which field data from different systems can be compared with each other and with laboratory studies.

Organism-specific simulations and relation to empirical data

Mayfly

Sediment and porewater were the dominant sources (ng/g wet wt/d) of pyrene (99%) to mayfly larvae compared to seston and overlying water. For PCB-153, sediment and seston dominated uptake (99%) compared to porewater and overlying water (Table 2).

Table 2. Uptake of pyrene and PCB-153 from different exposure routes for mayfly, polychaetes, yellow perch, and little owla
Organism PorewaterOverlying waterSedimentSestonTotal uptake flux
  • a

    The most contributing uptake route for each simulation is marked in bold. Numbers in parentheses are percent of total flux (ng/g wet wt/d). PCB = polychlorinated biphenyl.

  • b

    Young of the year fish.

MayflyPCB5.7E–3 (0.4)1.8E–3 (0.1)1.1 (84.6)0.19 (14.6)1.30
 Pyrene0.1 (8.3)1.8E–3 (0.2)1.1 (90.9)3.4E–3 (0.3)1.21
PolychaetesPCB3.5E–6 (0.0)6.7E–6 (0.0)0.73 (100.0)0.73
 Pyrene3.6E–2 (0.0)3.3E–3 (0.0)83.9 (100.0)83.9
  WaterYoung fishb(diet 1)Mayfly (diet 2)Chironomids (diet 3) 
Yellow perchPCB1.4E–3 (0.7)3.3E–2 (17.4)1.1E–1 (57.9)4.5E–2 (23.9)0.19
 Pyrene3.9E–3 (19.5)6.1E–5 (0.3)9.5E–3 (47.5)3.0E–3 (15.0)0.02
  AirWorm diet (diet 1)Beetle diet (diet 2)Mouse diet (diet 3) 
Little owlPCB3.6E–8 (0.0)3.38 (53.2)2.68 (42.2)0.29 (4.6)6.35
 Pyrene3.6E–8 (0.0)5.89 (66.9)2.68 (30.5)0.23(2.6)8.80

Simulated steady-state (SS) concentrations in mayfly larvae ranged from 0.66 to 2775 ng/g wet wt with a mean value of 53.3 ± 89.2 ng/g wet wt (median = 28.0) for PCB-153, and from 0.07 to 291.4 ng/g wet wt with a mean value of 9.95 ± 14.5 ng/g wet wt (median = 5.37) for pyrene (Table 3). The PCB-153 concentrations are within the range of mean values found in the Great Lakes (Corkum et al. 1997) and other large lakes in Canada receiving atmospheric and point-source inputs of PCBs (Stewart et al. 2003). Simulated log BAF values ranged from 5.24 to 8.97 with a mean value of 6.81 (±0.49) for PCB-153, and from 4.03 to 7.93 with a mean value of 6.08 (±0.53) for pyrene. Simulated log BSAF values differed between the 2 compounds and ranged from −0.92 to 1.54 with a mean of 0.48 (±0.26) for PCB-153, and from −1.79 to 0.91 with a mean of −0.22 (±0.37) for pyrene (Table 3). For the mayflies, the modeled mean PCB-153 BSAF was 3.0 (±1.8) and compared favorably with the empirical data of 4.7 ± 4.4 (Figure 1A). The coefficients of variation were 61.1 and 93.6% between the simulated and empirical data sets, respectively. The 1st to 99th percentiles of the simulation trials (0.64–9.60) overlapped with the 6th to 86th percentiles of the empirical data set, indicating that that as much as 80% of the variation in the empirical data could be explained by the simulation model (Figure 1A).

Table 3. Percent contribution of parameters to variation in model simulated steady-state concentrations of PCB-153 and pyrene in mayflies, polychaetes, little owl, and yellow percha
ParameterParameter typeMayflyPolychaeteLittle OwlYellow perch
PCB-153PyrenePCB-153PyrenePCB-153PyrenePCB-153Pyrene
  • a

    Parameters covering a minimum of 95% of the variation in Corg are included for each simulation, and the 3 most important parameters are highlighted in gray. Parameters are described in detail in the text and in Supplemental Data 2. The table presents 26 of a total of 56 parameters. Steady-state concentrations (Corg(ss): ng/g wet wt), log BAF (L/kg), log BSAF (kg lipid/kg organic carbon) and log BMF (kg wet wt/kg wet wt) values are presented as mean ± 1 SD. BAF = bioaccumulation factor; BMF = biomagnification factor; BSAF = biota–sediment accumulation factor; Chem./Phys. = chemical/physical; NLOM = nonlipid organic matter; PCB = polychlorinated biphenyl.

  • b

    The term Cw(pw) is a surrogate measure of bulk sediment concentration: Csed (see text for further explanation).

Mean Corg(ss) 53.3 ± 89.29.95 ± 14.54.97 ± 12.622.2 ± 39.9251.4 ± 156.41.63 ± 1.5245.4 ± 29.60.01 ± 0.010
Range Corg(ss) 0.66–27750.07–291.40.00–308.440.01–948.66.01–16470.1–22.22.1–426.30.001–0.27
Mean log BAF 6.79 ± 0.496.08 ± 0.538.60 ± 0.956.31 ± 0.6510.9 ± 0.448.63 ± 0.496.93 ± 0.333.25 ± 0.32
Mean log BSAF or BMF 0.48 ± 0.26–0.22 ± 0.370.18 ± 0.41–0.70 ± 0.451.10 ± 0.18–1.16 ± 0.270.11 ± 0.28–2.40 ± 0.32
Cw (pw), ng/mLbEnvironmental8.010.717.840.4    
fBCEnvironmental7.57.3      
C(diet1)Environmental    15.137.04.6 
C(diet2)Environmental    13.55.629.817.6
Cair, or C(o,w)Environmental       7.8
nfBCChem./Phys.35.222.267.420.2    
KfBCChem./Phys.17.715.3 7.1    
KowChem./Phys.15.529.910.613.6  12.7 
NLOM partitioning in organic matterChem./Phys.2.21.8  10.0   
NLOM partitioning diet1Chem./Phys.    1.0   
NLOM partitioning diet3Chem./Phys.    3.3   
p(diet1)Ecological    20.9   
p(diet2)Ecological    16.8 7.54.5
Fsed, feed (fraction of sediment ingested)Physiological2.1       
AElip(diet3)Physiological    4.0   
AEwat(diet1)Physiological    0.8   
AEwat(diet3)Physiological    3.3   
Esed(BC)Physiological2.01.8      
E(diet1)     1.12.2  
E(diet2)Physiological    1.1 9.16.6
E(diet3)     4.5   
QV (total)Physiological      4.9 
QfeedPhysiological 3.2 14.8 14.020.59.0
flip(org)Physiological4.93.7    3.6 
kmPhysiological     36.4 51.6
kgPhysiological      3.1 
Total variance % 95.195.995.896.195.495.295.897.1
thumbnail image

Figure 1. Comparison of PCB-153 bioaccumulation metrics between selected field data sets and model simulations for the model organisms (A) mayfly, (B) polychaete, (C) yellow perch, and (D) birds. Box charts present mean (▪), median (horizontal line), 25th and 75th percentiles (box edges), 5th and 95th percentiles (whiskers), and 1st and 99th percentiles (×). Mayfly and yellow perch raw field data from the Detroit River (n = 13 mayfly BSAFs; Drouillard 2010; n = 24 yellow perch; Kashian et al. 2010); Raw polychaete BSAF data from Nesto et al. 2010; Guillemot data generated from Lundsted-Enkel et al. (2005).

Download figure to PowerPoint

The differences in chemical bioaccumulation in this species were mostly related to the stronger association of pyrene with BC. A total of 94.4% of pyrene in sediments was predicted to be associated with BC whereas for PCB-153, BC contained 82.9% of the chemical, whereas 17.1% was associated with labile organic matter (LOM). Because only LOM in the model is allowed to undergo digestion in the gastrointestinal tract of the mayfly, this is the only fraction of chemical subject to gastrointestinal magnification. Thus, differential scavenging of the chemical by BC played a major role in bioaccumulation by mayflies as did differences in chemical hydrophobicity which influenced the fraction of chemical exposure to mayflies and their depuration to water. For both compounds, chemical and/or physical parameters had the greatest impact on variation in Corg (∼70%), followed by environmental parameters (∼17%), and physiological parameters (∼9%). Ecological parameters did not contribute (0%) to variation in accumulation in mayfly larvae (Figure 2 and Tables 3 and 4). The lack of influence by ecological factors may have been a consequence of only having 2 diet types (sediment and seston) and simulation conditions were such that sediment exposures dominated over seston exposures. A different conclusion may have been reached if seston had much higher contaminant concentrations than simulated. The presence of BC strongly influenced the variation in steady-state PCB-153 concentrations in mayflies. Model parameters linked to the dynamics of chemical sorption onto BC in sediment including the Freundlich coefficient (nfBC) and constant (KfBC) as well as the proportion of sediment that is BC (fBC), and the resulting sediment concentration (Cw(pw) in Table 3) of PCB-153 concentrations accounted in total for 68.4% of the variation in PCB-153 concentrations in mayflies. Parameters linked to the partitioning of PCB-153 into lipid (fLipid(org) and KOW) accounted for approximately 20.4% of the variation in steady-state mayfly PCB-153 concentrations (Table 3). As with PCB-153, the presence of BC strongly influenced the variation in steady-state pyrene concentrations in mayflies. In particular, the Freundlich coefficient (nfBC) and constant (KfBC), the proportion of sediment that is BC (fBC), and the resulting porewater dissolved pyrene concentrations (Cpw) accounted for 56% of the variation in pyrene concentrations in mayflies. Parameters linked to the partitioning of pyrene into lipid (fLipid(org) and KOW) accounted for approximately 33.6% of the variation in steady-state mayfly pyrene concentrations (Table 3).

thumbnail image

Figure 2. Contribution of parameter type to variation in steady-state concentration in 4 selected fauna groups (mayfly larvae, deposit-feeding polychaetes, fish, and birds), using 2 model compounds: pyrene and PCB-153. Black = environmental parameter; light gray = chemical and/or physical parameter; dark gray = ecological parameter; white = physiological parameter.

Download figure to PowerPoint

Table 4. Percent contribution of parameter type to variation in model simulated steady-state concentrations of PCB-153 and pyrene in mayflies, polychaetes, little owl, and yellow percha
ParameterMayflyPolychaeteLittle OwlYellow perch
PCB-153PyrenePCB-153PyrenePCB-153PyrenePCB-153Pyrene
  • a

    Mean values are ± 1 SD. PCB = polychlorinated biphenyl.

Environmental15.518.017.840.428.642.634.425.4
Chemical/physical70.669.278.040.914.3012.70
Ecological000037.707.54.5
Physiological9.08.7014.814.852.641.267.2
Total variance %95.195.995.896.195.495.295.897.1

In summary, approximately 99% of the influx of the model compounds could be explained by influx from sediment and seston (PCB-153) or sediment and overlying water (pyrene). The main drivers of variation in bioaccumulation for both model compounds were chemical and/or physical (∼70%: primarily nfBC, KfBC, and KOW), environmental (16%–18%: sediment concentration, fBC), and physiological (∼9%: divided on 4 parameters). The simulated BSAF could explain as much as 80% of the variation in the empirical data.

Polychaetes

Bulk sediment was the dominant input to polychaetes, giving approximately 2.4 × 104 and approximately 2.1 × 105 higher weight-specific uptake (ng/g wet wt/d) from ingested sediment compared to respiratory uptake via porewater and overlying water for pyrene and PCB-153, respectively (Table 2). This is consistent with other studies of deposit-feeding polychaetes showing that the sediment pool contributes up to 95% of the body burden in deposit-feeding polychaetes (e.g., Selck et al. 2003a).

The simulated steady-state concentrations of PCB-153 in polychaetes ranged from 0 to 308.4 ng/g wet wt with a mean value of 4.97 (±12.6) ng/g wet wt (median = 1.45) (Table 3). This range covers the concentrations measures in N. diversicolor collected from Weser estuary at Bremerhaven (∼12.5 ± 2.5 ng PCB-153/g wet wt; from figure 6 in Goerke and Weber 2001) and exposed to 2 Norwegian harbor sediments (6.6 ± 1.7 ng PCB-153/g wet wt; Ruus et al. 2005), as well as in another polychaete, P. rullieri collected from 2 stations over 4 seasons in a lagoon of Venice, Italy (7.4–35 ng PCB-153/g dry wt; Nesto et al. 2010). Simulated pyrene concentrations in polychaetes ranged from 0.01 to 948.6 ng/g wet wt with a mean value of 22.2 (±39.9) ng/g wet wt (median: 9.9 ng/g wet wt), which corresponds with measured concentrations in P. rullieri collected from 2 stations over 4 seasons in a lagoon of Venice, Italy (6.6–39 ng pyrene/g dry wt; ∼47.1–278.6 on a wet-weight basis, assuming a water content of 86%; Supplemental Data 2), and N. diversicolor exposed to 2 Norwegian harbor sediments (242 ± 7.9 ng pyrene/g wet wt and 40 ± 12.3 ng pyrene/g wet wt; Ruus et al. 2005). Simulated log BAF values ranged from 5.00 to 12.35 with a mean value of 8.56 (±0.95) for PCB-153, and from 3.29 to 8.82 with a mean value of 6.31 (±0.65) for pyrene. Simulated log BSAF values ranged from −1.54 to 1.57 with a mean of 0.22 (±0.41) for PCB-153, and from −2.58 to 0.75 with a mean of −0.69 (±0.45) for pyrene. These ranges covers values presented for P. rullieri and N. diversicolor (P. rullieri; log BSAF: −0.70–0.80; pyrene: log BSAF: −1.22–0.30; Nesto et al. 2010), (N. diversicolor; log BSAF: 0.01; pyrene: log BSAF: −0.66 and −1.03) (Ruus et al. 2005) (Figures 1B and 3). For polychaetes, modeled mean PCB-153 BSAF for P. rullieri was 1.7 ± 2.0 and compared favorably to model-simulated value of 2.2 ± 1.7, respectively (Figure 1B). The model-simulated BSAF for PCB-153 was also consistent with laboratory BSAF data reported for PCB-153 in N. diversicolor, which had a mean of 1.38 ± 0.46 (Ruus et al. 2005). Comparing the box and whisker plots between field and simulated PCB-153 BSAFs, the simulated output distribution had a larger range of values compared to the field data, but the coefficient of variation was found to be higher for the field data at 115.8% compared to 79.9% for the modeled data. The 1st to 99th percentiles of the simulation trials fully overlapped the range of data associated with the empirical data set (Figure 1B). Figure 3 summarizes the box and whisker chart comparing model uncertainty and empirical variation of pyrene in polychaetes exposed under laboratory bioassay conditions or in the field. The mean simulation BSAF was 0.31 ± 0.32 and compared favorably to the empirical field BSAF of 0.49 ± 0.66 for P. rullieri and fully covered the laboratory bioassay BSAF of 0.15 ± 0.08 for N. diversicolor (low variation). The coefficient of variation was similar at 105.1% and 135.0% for simulation and field data, respectively, but higher than that for the laboratory bioassay at 52.6%. The uncertainty associated with pyrene simulations was somewhat greater than field variability, particularly at the lower range and considerably greater than that associated with laboratory bioassay data. The 1st to 99th percentiles of model trials overlapped with the 1st to 95th percentiles of the field data and fully overlapped with the laboratory bioassay trials (Figure 3). Thus, much of the variation in field and laboratory bioassay data could be explained by uncertainty in model simulations. The higher variation in simulation data compared to empirical data is likely a result of how the model includes variation for deposit-feeding polychaetes as a group of organisms which evidently will result in a higher variation compared to the empirical data covering a single species (P. rullieri). In addition, a very large range of environmental concentrations (0.03–5 ng PCB-153/g wet sediment; <0.03–1373 ng pyrene/g wet sediment) was used as input for the polychaete trials and would contribute to the large variability. The constrained variation in the laboratory bioassays are likely related to the more controlled environment of bioassay conditions and use of limited number of sediment sites from the same region. This implies that laboratory studies may be less suited to assess complex accumulation processes that involve ecological interactions between organisms and their surrounding, for instance in case of food web accumulation. In such cases, field studies or the use of accumulation models may be recommended.

thumbnail image

Figure 3. Comparison of pyrene bioaccumulation metrics between selected field data sets and model simulations for polychaetes. Box charts present mean (▪), median (horizontal line), 25th and 75th percentiles (box edges), 5th and 95th percentiles (whiskers), and 1st and 99th percentiles (×). Raw polychaete BSAF data from Nesto et al. 2010 and in-house data set from A. Ruus.

Download figure to PowerPoint

The chemical and/or physical parameters was the predominant parameters category contributing to variation in bioaccumulation for PCB-153 (78%) followed by environmental parameters (∼18%) (Figure 2 and Tables 3 and 4). This distribution was different for pyrene where environmental parameters were equally significant (∼40%) as chemical and/or physical (∼41%) followed by physiological (∼15%) parameters. Ecological parameters did not affect variation for either of the compounds (Figure 2 and Tables 3 and 4), and physiological parameters were only significant for pyrene (feeding rate). A total of 96% of the variation in the steady-state PCB-153 concentrations in polychaetes could be explained by only 3 parameters, where the presence of BC was by far the most important. Model parameters linked to the dynamics of chemical sorption onto BC (i.e., the Freundlich coefficient: nfBC) was the most significant contributor to variation (67%) followed by the sediment concentration of PCB-153, contributing 18% to the variation, and KOW with approximately 11%. The physiology of the polychaetes did not contribute to variation (0%) in the model simulation. As with PCB-153, the presence of BC influenced the variation in steady-state pyrene concentrations in polychaetes. However, the contribution of the Freundlich coefficient (nfBC) and constant (KfBC) was lower at 27% for pyrene compared to 67% for PCB-153. The sediment concentration accounted for 40% and KOW for approximately 14% of the variation in the pyrene concentration in polychaetes. Polychaete physiology, presented as feeding rate and thus the amount of contaminated sediment passing through the gut of the polychaetes, also played a bigger role for pyrene accounting for ∼15% of the variation compared to PCB-153. The influence of feeding rate may have been linked to the influence of sediment concentration.

In summary, the uptake of both model compounds by the polychaetes could be explained by sediment-associated factors (100%). Variation in bioaccumulation was mainly driven by chemical and/or physical (67%: primarily nfBC) and environmental (18%: sediment concentration) parameters for PCB-153, whereas physiological and ecological parameters contributed very little. The main drivers of variance for pyrene were environmental (40%: sediment concentration), chemical and/or physical (41%: primarily nfBC and KOW) and physiological (15%: feeding rate) parameters. The simulated PCB-153 BSAF and pyrene BSAF exceeded the variation in empirical data, which was mainly attributed to the model that treated deposit-feeding polychaetes as a group (i.e., covered a large number of species) and a large variation in Csed, whereas the empirical data accounts for only one species. The variation in laboratory-based BSAF corresponded well with the simulated PCB-153 BSAF but was somewhat lower for the simulated pyrene BSAF, which may be related to the more controlled conditions in the laboratory.

Yellow perch

Uptake of pyrene (ng/g wet wt/d) was dominated by diet 2 (mayfly) followed closely by water and diet 3 (chironomids), whereas diet 1 (young fish) contributed about a factor of 100 less to uptake (Table 2). PCB-153 uptake flux was dominated by diet 2 followed by diet 3 and diet 1 (Table 2).

Simulated PCB-153 concentrations for yellow perch ranged from 2.1 to 426.3 ng/g, with a mean value of 45.4 (±29.6) ng/g wet wt (median: 38.3 ng/g wet wt). The empirical data exhibited a PCB-153 concentration of 92.1 ± 139.6 ng/g wet wt and median value of 64.2 ng/g wet wt (range: 1.0–685.4 ng PCB-153/g wet wt). For the yellow perch, the modeled mean steady-state concentration was lower by almost a factor of 2 compared to the empirical data (Figure 1C). The coefficients of variation were 65.2% and 151.6% for the modeled and empirical data sets, respectively. This corresponds well with the uncertainty and sensitivity analysis performed by Ciavatta et al. (2009). These authors simulated steady-state concentrations in the fish, Zosterisessor ophiocephalus, for PCB-15, PCB-101, and PCB-194 and reported that the coefficient of variation ranged from 65% to 119%. The 1st to 99th percentiles of the yellow perch model simulation output overlapped with the 14th to 87th percentiles, indicating that model uncertainty was capable of explaining approximately 73% of the variation in the empirical data. The empirical data also exhibited a greater range of concentrations than the predicted range. The lower variance in the predicted concentration could be due to a number of things. These include fish movements outside of the Detroit River (model inputs were set to Detroit River) or much larger variation in diet preferences in real populations compared to the simple 3-item diet incorporated into model simulations. In addition, the yellow perch data set included a wide range of fish sizes, some of which may have achieved the steady state and others which have not. Neither fish size nor non–steady state bioaccumulation dynamics were considered in model simulations. For yellow perch, the model predicted pyrene concentrations ranging from 0.001 to 0.27 ng/g wet wt with a mean of 0.010 (±0.010) ng/g wet wt (Table 3). These predicted concentrations generally approach the analytical detection limit or are below typical detection limits for PAHs in biological tissues. This prediction is consistent with the fact that PAHs are rarely detected in tissues of freshwater fish (Leadley et al. 1998). The predicted log BAF ranged from 5.53 to 8.16 with a mean of 6.93 (±0.33) for PCB-153, and from 2.46 to 4.76 (mean: 3.21 ± 0.32) for pyrene. The PCB-153 value closely approximated the observed BAF reported for yellow perch by Arnot and Gobas (2004) from Lake St. Clair, Canada, and Lake Erie in the range of 6.4 to 7.04. Hauck et al. (2011) simulated a model fish using 24 parameters and a range of KOW values. Based on their figure 2 and the KOW values corresponding to pyrene and PCB-153, we estimate that the 10th and 90th percentile for log BAF for the model fish was 3.15 and 4.48, and 4.60 and approximately 6.26 for pyrene and PCB-153, respectively. Our simulations had a much narrower range for pyrene (2.86 for pyrene and 3.67 for PCB-153) compared to Hauck and coworkers (2011), which probably is because these authors did not include biotransformation in their simulations. For PCB-153, our simulations had a wider range (6.50 and 7.35), which probably reflects that we included 3 diets, whereas Hauck et al. (2011) only included 1 diet. The mean biomagnification factor, expressed as the fugacity ratio of animal-to-average diet, was 1.28 (±1.90: range: 0.06–7.2) for PCB-153, and the mean predicted biomagnification factor was 0.004 (±2.09; range: 0.0002–0.108) for pyrene (Table 3). The very large range in PCB-153 BMFs results from high variability predicted for the organism as well as propagation of additional uncertainty associated with dietary proportions among the model simulations.

Environmental and physiological parameters contributed equally to variation in PCB-153 concentration (∼34%–41%) followed by the chemical–physical (∼13%) and ecological- (∼8%) parameters. Physiological parameters contributed most to the variation of steady-state pyrene concentration in yellow perch (∼68%), followed by environmental (∼25%), and ecological parameters (∼5%) (Figure 2 and Tables 3 and 4). Steady-state PCB-153 concentrations in fish were most strongly related to model parameters governing dietary exposures. The 4 most important parameters associated with model uncertainty, cumulatively explaining 62% of the model variation, consisted of chemical concentrations in diet 1 and diet 2, fish feeding rates, and the magnitude of the proportion of diet 2. Fish ventilation rate and growth rate each explained approximately 8% of the variation in model output and were negatively related to concentrations predicted in fish. Parameters linked to the partitioning of PCB into lipid (fLipid(org) and KOW) accounted for approximately 17% of the variation in steady-state concentrations (Table 3). For pyrene, the uncertainty associated with metabolic biotransformation was the most important factor contributing to variation in model output (52% of model variation associated with km). The other major contributing factors to model uncertainty included dietary exposure terms (pyrene concentration in diet 2, animal feeding rate and proportion of diet 2 ingested collectively contributing to 31% of model uncertainty along with pyrene concentration in overlying water and chemical assimilation efficiency from diet 2 (14% of model uncertainty).

In summary, uptake of pyrene was highest from diet 2 (48%: mayfly) followed by water (20%) and diet 3 (15%: chironomids), whereas diet 1 (young fish) contributed about a factor 100-fold less to uptake. PCB-153 uptake flux was dominated by diet 2 (58%: mayfly) followed by diet 3 (24%: chironomids) and diet 1 (17%: young fish). Variation in pyrene bioaccumulation was driven by physiological (67%: primarily biotransformation) and environmental (25%: concentration in mayfly) factors, and variation was almost equally driven by environmental (34%) and physiological (41%: primarily feeding rate) factors for PCB-153. The simulated PCB-153 steady state Corg could explain 73% of the variation in the empirical data. This likely has to do with the greater heterogeneity of actual exposure conditions, greater diversity of diets, and other uncertainties not accounted for in the model simulations.

Little owl

Uptake of both compounds (ng/g wet wt/d) was highest from worm ingestion (diet 1) followed by that of the beetle (diet diet 2) and mice (diet 3) (Table 2). Uptake from air was negligible.

Simulated PCB-153 concentrations in little owls ranged from 6 to 1647 ng/g wet wt with a mean of 251 ± 156 ng/g wet wt (Table 3). Liver concentrations of PCB-153 in little owl chicks, 2 to 5 weeks of age, collected in the field were approximately 46 ± 30 ng/g wet wt (van den Brink et al. 2001), which is at the lower end of the simulated concentrations. It should be noted however, that the simulated concentrations relate to adult birds, which are likely to have much higher concentrations than chicks. To our knowledge, no relevant data on PCB-153 are available in adult little owl tissues. The simulated concentrations of pyrene in the little owl are low and range from 0.1 to 22.2 ng/g wet wt (mean 1.63 ± 1.52 ng/g wet wt). Such low concentrations were to be expected due to biotransformation of pyrene both in the birds themselves and in their diet (e.g., mice; Ronis and Walker 1989). The simulated log BMFs ranged from −0.26 to 1.61 with a mean of 1.10 ± 0.18 for PCB-153, corresponding to a range in absolute BMFs of 1.5 to 47.0 and mean of 13.6 ± 5.0. The calculated BMFs from the present study are well within the same range reported by Van Wezel et al. (2000) on congener-specific BMFs for PCB-153 (lipid–lipid) ranging from 2 to 180 from field studies on the range of species. For the avian comparison, the magnitude of BMF differed by a much larger extent than observed for the other 3 PCB-153 simulations. The guillemot mean BMF was 30.2 ± 20.2 whereas the little owl simulations predicted a BMF of 13.6 ± 4.9, respectively (Figure 1D). The coefficient of variation for the little owl was 36.6% compared to 67.1% for the guillemot. The 1st to 99th percentiles of the model simulations overlapped with the 1st to 56th percentiles of the guillemot data but failed to overlap with the upper range of the empirical data. The slightly higher variation in the guillemot BMF may be related to their variable diet (Wilson et al. 2004). Because diet composition and prey concentrations are the main drivers for the variation in little owl BMF, it is essential to capture the true variation in its diet. The modeled diet of the little owl is a simplification of the real diet, which may have resulted in an underestimation of the variation of the modeled BMF and field BMF values. The magnitude of this underestimation is unknown, but given the fact that the guillemot field BMFs are only slightly higher, it is expected that most relevant field-related variance is included in the modeled little owl BMFs. In order to compare variation in outputs, the little owl simulation was adjusted for bias by multiplying individual trial values by the ratio of the guillemot median BMF to the little owl median BMF. For this centered data set, the 1 to 99% adjusted little owl data overlapped with the 1st to 89th percentiles of the guillemot data. The log BMF range from −2.1 to −0.20 for pyrene with a mean of −1.16 (±0.27) (median BMF: −1.16), and the data cover more than an order of magnitude, which is approximately 1 unit lower than for PCB-153, indicating a much lower accumulation rate. Similar to the concentrations, no relevant empirical data for pyrene BMF are known.

When clustered according to parameter category, and restricted to those factors that add up to 95% (Table 3), ecological factors explain 38%, environmental factors 29%, chemical and/or physical factors 14%, and physiological factors 15% of the variation in PCB-153 accumulation in little owl (Figure 2 and Tables 3 and 4). The ratios between ecological, environmental, and physiological factors determining the Corg is very different for pyrene compared to PCB-153: environmental factors explain 43% and physiological factors 53%, whereas ecological and chemical and/or physical factors do not contribute to variation (Figure 2 and Tables 3 and 4). The main factors explaining the variability in PCB-153 concentrations in the little owl are diet fractions of worms and beetles and their concentrations (approximately 66% contribution). The fraction of the diets (p(diet1) and p(diet2)) are independent, and can be regarded as ecological parameters, reflecting the foraging ecology of the birds. It should be noted that p(diet3) is calculated as 1 − p(diet1) − p(diet2). The concentrations in the diet items (C(diet1) and C(diet2)) reflect environmental conditions, e.g., the environmental contaminant levels. The following parameters driving the variance of Corg are chemical and/or physical: the partition coefficient of NLOM for diet 1 and diet 3 and physiological: assimilation efficiency (AElip(diet 2,3)). The main factor explaining the variability in pyrene concentrations in the little owl is the concentration in diet 1 (earthworms), which is a diet item with high fresh weight concentrations (20.4 ng pyrene/g wet wt) in combination with a low fat percentage (2%), and the metabolic capacity (km). The feeding rate (Qfeed) of the little owl have a great influence on Corg, and the concentrations in the beetles are also important; this diet item contains relatively high concentrations of pyrene (22 ng pyrene/g wet wt) but its lipid content is somewhat higher (4.8%). Overall, the composition of the diet and the concentrations in the diet are the main driving forces in explaining the variance in PCB-153 concentrations in little owl. These results indicate that in contrast to PCB-153, ecological factors are less important in explaining accumulation of pyrene by the little owl. In this case, environmental concentrations (in the diet) and the physiology of the bird (metabolic capacity) are more important.

In summary, uptake of both compounds was highest from worm ingestion followed by beetle and field mice ingestion. Environmental (43%: primarily concentration in worm diet) and physiological (53%: feeding and biotransformation rates) factors were the main drivers of variation in pyrene bioaccumulation for the little owl. Ecological (38%: diet fraction of worm and beetle) and environmental (29%: concentration in worm and beetle diets) factors contributed most to variation in PCB-153 bioaccumulation and chemical and/or physical and physiological factors contributed (15%) equally to variation. The simulated BSAF could explain as much as 88% of the variation in the empirical data for PCB-153 after centering simulation output on BMF to the empirical data set. The slightly higher variation in the guillemot BMF may be related to their variable diet and that the lipid concentrations in the guillemot diet are expected to be higher than in the owl diet, resulting in higher BMF.

Model uncertainty and knowledge gaps

From the modeling results of the 8 organism–chemical combinations, some clear patterns have emerged. First, it was evident that uptake of both model compounds was clearly dominated by uptake from bulk sediment and diet in all 8 simulations. Second, the variation in bioaccumulation was mainly covered by 10 of the 56 input parameters for all 4 model organisms (Table 3). Variability in bioaccumulation appears to be mainly driven by sediment exposure, sediment composition, and chemical partitioning to sediment components at the lower trophic levels, as illustrated by the mayfly and polychaete. At higher trophic levels, food web structure (i.e., diet composition) and diet concentrations become more important particularly for highly persistent compounds such as PCB-153. Thus, it was evident that at higher trophic levels, ecological, environmental, and physiological factors played a dominant role in explaining the variability in bioaccumulation, whereas the physical, chemical, and environmental properties of the compounds were more important at lower levels in the food web (Figure 2 and Tables 3 and 4). Third, the difference between the 2 chemicals was far less than differences among the organisms chosen, and this observation was most pronounced in the species at the lower trophic levels (Table 3) which may reflect the high degree of similarity between the model chemicals. However, a few differences were observed. Physiological factors appeared to play a larger role for nonpersistent compounds (i.e., pyrene) compared to persistent compounds (i.e., PCB-153). These observations are tied to both the chemical attributes that specify time to the steady state for a chemical in the organism as well as animal life history of attributes (i.e., body size, longevity, spatial movements, and feeding ecology).

For the 2 benthic species, sediment concentrations and bioavailability contributed considerably to model uncertainty. One of the most apparent challenges in estimating bioaccumulation including field studies is to accurately estimate the “dose” or the amount of substance taken into an organism per body mass per day (mg · kg · d). The Csed variable is more straightforward to estimate and can principally take any value or range dependent on site and/or pollution history and characteristics. For simulations of the polychaetes, we included a substantial range in sediment concentration for both compounds which was based on field samples from 2 sites covering concentrations ranging from uncontaminated to highly contaminated sediment. Evidently, this led to a substantial uncertainty in Csed, which eventually affected the uncertainty in the estimated bioaccumulation. BC and its ability to sequester contaminants and influence chemical bioavailability have only recently been considered in benthic invertebrate modeling of hydrophobic organic pollutant bioaccumulation (Moermond et al. 2005; Koelmans et al. 2006; Hauck et al. 2007; Moermond et al. 2007). Incorporating these terms into model simulations was difficult due to data gaps on BC distributions in natural environments and chemical partitioning to this and other sediment components as it applies to the specific cases of species and ecosystems being modeled. Although literature reviews on the fraction of BC in different sediment types were available (Cornelissen et al. 2005; Koelmans et al. 2006), site-specific estimates would enable model simulations to be set up to produce more accurate predictions and constrain model uncertainty (Koelmans et al. 2009). With regard to chemical partitioning to sediment components, there remain major data gaps. BC is not a homogeneous matrix but can be further categorized into different components, each with different sorptive characteristics and associated Freundlich parameters (Jonker and Koelmans 2002; Hawthorne et al. 2007; Poot et al. 2009). Although not specifically flagged in the model sensitivity priorities, several model parameters related to chemical bioavailability from ingested sediment had to be estimated without strong empirical data to support such estimates. For example, the ability of organisms to assimilate chemical from individual sediment components (LOM versus BC) is poorly characterized, and estimates were generally derived in order to match the bulk sediment chemical assimilation efficiencies more widely reported in the literature. Available data for Efood and species or chemical combinations are relatively scarce. This is further complicated for sediments because we need to establish Efood for LOM and BC, and this type of information is very limited. Empirical studies to demonstrate how, for example, chemical assimilation from sediment differs across gradients in LOM and BC would be useful to address these uncertainties. Other problems related to a more fundamental understanding of the digestive physiology of benthic invertebrates, e.g., how they assimilate nutritional components from different sediment components, also resulted in broad estimates of many parameters.

The little owl and yellow perch simulations exhibited many commonalities even though 1 organism was terrestrial and the other was aquatic. This is mostly related to the fact that diet represented the major exposure route for both chemicals and species. Both species were modeled to include multiple diet items, and the combination of environmental factors related to different chemical concentrations among different diet items and ecological factors relating to dietary proportions were of high relevance to model uncertainty. Unlike the benthic invertebrates, many model parameters related to animal bioenergetics, such as proximate composition and digestion physiology, are better understood for fish (e.g., Hewett and Johnson 1992; Hendersen et al. 2000) and birds (see Materials and Methods). However, empirical data related to ecological parameters associated with diet choice and how this varies in time and space remain challenges to both the fields of ecology and ecotoxicology. Animal movements over time and space are directly coupled to time-integrated chemical exposures, and this will vary by chemical depending on the time required for the organism to achieve steady state with its diet and environment. Food web bioaccumulation models often include a feeding matrix of multiple diet items, although assigning uncertainty to large numbers of diet items can prove challenging in model uncertainty and sensitivity evaluations. One feature not recognized in the simulations considered in this study were time-dependent, or seasonal, trends in prey availability commonly described for fish and wildlife populations (Korpimaki and Norrdahl 1991; Harnois et al. 1992; Dauwalter and Fisher 2008). These structural components (i.e., diet shifts, partitioning of diet, variation in prey availability with time, individual differences in feeding behavior) are rarely considered in bioaccumulation model applications. At this stage, it is difficult to understand the importance of these factors which would require development of life history–based, non–steady state bioaccumulation model approaches (Paterson et al. 2006; Burtnyk et al. 2009).

The model simulations provide a framework in which to evaluate the variability in accumulation between laboratory and field. Unfortunately, the range of ecological conditions that organisms typically experience in nature are not easily simulated in (simple) laboratory studies, making it difficult to assess the importance of ecological factors on bioaccumulation. Indeed, laboratory studies may be less suited to address complex accumulation processes that involve ecological interactions between organisms and their surrounding, for instance in case of food web accumulation. In such cases, field studies or the use of accumulation models is recommended. On the other hand, laboratory studies are excellent when studying fewer variables without confounding factors encountered in field situations, and such data are crucial for sound model simulations.

Finally, for chemicals subject to rapid biotransformation, individual, species, and environment biotransformation interactions are rarely understood and methods to estimate the magnitude of km for a given species are often lacking. In the present simulations, the focus was on predicting whole-body concentration of the chemical using a simplified single-compartment framework. Such a framework often works well (Clark et al. 1987) for persistent chemicals that exhibit relatively long half-lives in animals on the order of weeks or longer. However, a single-compartment approach introduces several errors when extended to chemicals that are rapidly cleared from the animal. Artifacts are introduced when tissue-specific clearance exceeds the time required for intertissue distribution (often on the order of days; Stehly and Hayton 1988; Barron et al. 1990). In such cases, multicompartment clearance models or physiologically based toxicokinetics models become necessary to tease out consequences of different exposure routes (e.g., respiratory exchange versus ingestion), and to establish differences in chemical concentrations among tissues that occur not only because of differences in chemical partitioning capacity among tissues but also related to between tissue transport lags. Clearly, the model simulations for pyrene in yellow perch and little owl were strongly subject to such problems because of the single compartment model used. Thus, the uncertainty analysis that flagged the high importance of km for these chemical and/or species combinations suggests a need to consider not only improvement in the biotransformation rate coefficient itself, but also adoption of a different model framework altogether.

GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

Despite the aforementioned limitations of the present models and simulated output, their overall correspondence with field measures of bioaccumulation provide a basis from which to offer recommendations and guidance for the interpretation and use of variable bioaccumulation data.

Applicability of toxicokinetic models in a “B” assessment

The uncertainty analysis provided in this article illustrates the potential for probabilistic modeling approaches as a means to identify the parameters and variables that contribute most to bioaccumulation uncertainty in specific cases, as well as to quantify propagation of error in the metrics used in a B-assessment. Another rationale for probabilistic modeling of B-metrics, rather than calculating them as discrete values without error, relates to their role in evaluation of risk criteria. If B-assessment depends on exceeding some bright line criterion (e.g., BAF > 5000), Stockholm convention, Annex D, 2001), “exceeding” should be statistically meaningful, that is, the B-metric should exceed the criterion with 95%, 50%, or 5% confidence, or whatever is deemed appropriate. Probabilistic approaches such as applied in this article, can be used to include measures of variance in calculation and interpretation of risk criteria (Figure 4). Figure 4 is included to illustrate how use of a simple model can provide this type of information, though it should be considered with care based on the bounds of the model parameters. This figure shows that the range between the 10th percentile and 90th percentile is generally similar between the compounds for a given species, but varies significantly among organism groups, with the largest range for polychaetes and the smallest for yellow perch. This may be related to the fact that simulations of the mayfly, yellow perch, and the little owl parameters are mostly species-specific, which was not the case for the polychaete simulations. Hence, the results of the polychaete simulations may be more variable and have the advantage of including a group of functionally similar species (i.e., deposit-feeding polychaetes) instead of 1 representative species. Nevertheless, the results indicate that the level of uncertainty cannot easily be extrapolated between organism groups.

thumbnail image

Figure 4. Whisker–box plot displaying log BAF of PCB-153 and pyrene for the 4 species as calculated in the simulations. The box charts present the 25th to 75th percentile, and the whiskers indicate the 5th to 95th percentile. Upper line denotes the mean, and lower line shows median. Line is BAF = 5000.

Download figure to PowerPoint

Evidently, there is a requirement for several criteria to be fulfilled prior to implementing a model of this type as a tool in “B” assessments including transparency, robustness, and validation of the model (i.e., comparative studies between field measures and simulation output). The latter is complicated by the scarcity of this kind of information in the literature (see above), which is reflected in the difficulty encountered in the present study comparing simulation output to empirical field data. The simulations on PCB-153 showed that in 3 of the 4 organisms, the amount of variation observed for the empirical data exceeded the variation predicted by model simulation trials. The model variation exceeded the empirical variance for pyrene in the polychaete model. However, in most cases, the model uncertainty reflected to a large degree the field data variability measured. Across the organism groups, the range of uncertainty associated with model simulations overlapped with between 73% and greater than 100% of the empirically measured variation in field data sets. Only the simulations for the polychaete exceeded the field variation associated with the matched empirical data set. This is related to the higher variance associated with implementing results from a group of organisms in the model compared to variance associated with a single species (i.e., in empirical data). The yellow perch simulations had the lowest amount of uncertainty relative to field variability, explaining only 73% of the variation in field data. This could be due to fish movements outside of the Detroit River for which model inputs were set or much larger variation in diet preferences in real populations compared to the simple 3-item diet incorporated into model simulations. Because the variation associated with the simulation output is a function of the model structure and range of error provided for the different parameters, the models can never provide a perfect representation of reality. However, to get closer to reality, we should engage in more robust model validation exercises by, e.g., focusing on the main contributors to variation and setting up experiments to test the effect of manipulating these parameters on bioaccumulation endpoints. For mayflies and polychaetes, a clear focus should be on BC, whereas for little owl and yellow perch, perturbations to feeding rate in conjunction with exposures to heterogeneous diets differing in chemical concentration and proximate composition should be examined. In addition, to validate and implement models in B-assessment, there is a research need for comparative studies between laboratory and field and field and model output.

Guidance on the use of field data in a “B” Assessment

In developing this article, we envisioned the use of field data as part of an overall assessment of the propensity for a specific substance (or group of substances) to accumulate in individual organisms and/or magnify or dilute in a food chain. No single study necessarily provides definitive information regarding this data. The question then arises, how are studies, which may be collected in different spatial and temporal contexts and for different species, best compared? Therefore, it becomes important to provide some basic guidance for the use of field data, based on the knowledge gained from the model simulations that addressed this question. Acknowledging the variability present between studies (laboratory to field and field to field) can facilitate a better understanding of a substance's ability to bioaccumulate. The substances under study (i.e., PCB-153 and pyrene) represent 2 material classes that cover a range of hydrophobic substances. Although it is apparent that these 2 materials do not represent the universe of chemicals in the environment, they are of importance in “B” assessments due to their hydrophobicity and difference in likelihood of biotransformation. Substances with lower hydrophobicity are expected, on the whole, to be less bioaccumulative than the model compounds used here.

The guidance presented is meant to be used in a PBT assessment. However, the information provided by the analysis in this article does raise some important considerations for sediment risk assessment. Often, sediment risk assessment is performed by extrapolating between endpoints using Equilibrium Partitioning Theory (EqP), assuming uptake from water only (EU TGD 2003). Recognizing that there are situations where porewater concentrations correlate well with organism concentration, the general assumption that porewater concentration reflects the available fraction of a contaminant (EqP theory) is not in agreement with empirical data as well as modeled data from this study. The model applied in the simulations includes a gastrointestinal magnification model that allows for chemical biomagnifications dependent on the degree of sediment digestion and change in chemical partition capacity of feces relative to ingested bulk sediments. Through the model parameterization, we allowed for sediment (i.e., sediment organic matter) digestion to occur, which resulted in a more important chemical uptake from sediment ingestion relative to porewater exchange. Indeed, gut fluid extraction and subsequent contaminant accumulation during sediment ingestion have been observed in many sediment-dwelling organisms (Mayer et al. 1996; Voparil et al. 2004; Voparil and Mayer 2004). By adopting an EqP approach, between animal and porewater and the BC concept (regarding solid–water partitioning), we would greatly underpredict contaminant accumulation and discard biomagnification, especially for chemicals with log KOW > 5. Thus, our results contradict present recommendations to base accumulation on freely dissolved concentrations for the 2 compounds employed. Instead, they show that uptake of contaminants through ingestion and the bioavailability of contaminants from sediment particles (especially BC) is an important consideration in determining exposure for sediment-associated organisms.

Although not exhaustive, the 4-organism, 2-chemical simulation runs identify some specific parameters, that, when properly measured and reported, provide the means to estimate the uncertainty in the reported “B” value, and therefore determine the usefulness of that measurement in the overall “B” assessment. Overall, in evaluating field “B” studies, the data presented must be considered in light of the life history traits of the organism, food sources within its habitat, and its ability to biotransform the chemical of interest. It would appear that for lower taxa within a food web (in this case, the mayfly and polychaete), sediment concentrations, and sorptive capacity to and presence of black carbon are important parameters to be reported. Precision and accuracy of the analytical method to determine these data should therefore be reported. For organisms in higher taxa (here, the little owl and perch), measurement of the chemical in food sources from within the habitat need to be accurately reported as well. Distribution of pdiet plays an important role particularly for chemicals of higher log KOW (e.g., PCB-153) that are poorly biotransformed. Feeding rates play a role (lesser in the mayfly; greatest in the owl and polychaete) for biotransformable chemicals with metabolic rate being very important for the higher taxa for biotransformable chemicals. Data other than the ones measured would be useful for assessing the value of the field study, but their absence would not necessarily preclude its use. For example, NLOM partitioning appears to be important for nonbiotransformable chemicals in high-tiered terrestrial taxa but is of little to no importance in other species.

Methodological approaches to minimize variation

In the previous sections (and Table 3), we have shown that 10 input parameters will largely cover the 3 main drivers of variation for all 4 model organisms (8 organism–chemical combinations). These 10 parameters can be further grouped into 6 overall processes or characteristics, namely partitioning to black carbon, sediment concentration, diet concentration, diet choice, feeding rate and biotransformation. In the following, we will address each of these processes or characteristics individually to see how variation can be minimized by different methodological approaches.

Concentration in diet, C(diet)

The simulations showed that concentration in diet was generally the most important contributor to variation in the 4 vertebrate–chemical combinations. Variation is minimized by measuring the actual chemical concentration in the diet, with the assumption that the main diet items for the particular organism are known (see section on p(diet) below). The measurement of diet concentrations in the field require a priori knowledge of diet choice in that particular setting and/or habitat. Information on concentrations in the different diet items can be obtained either by direct measurements, or by applying BSAFs or regression models on soil or water concentrations. However, direct measurements in relevant prey items are more favorable and recommended when possible. In order to assess the main diet composition of predators, several approaches may be applied, all with their limitations. Consequently, minimizing variation by assessing diet concentration will increase sampling effort and time requirements for field studies. Further complications include the fact that concentrations in diets often fluctuate in time and space, and species do not exploit their environment in an uniform way, depending on their foraging ecology. When considering migratory organisms or temporally changing diet choices, we are obviously faced with even larger challenges, though the same complexity problems exist, with regard to spatial and temporal heterogeneity, when bioaccumulation metrics are based solely on concentrations in water, sediment, air, and/or soil. These challenges are not easily addressed (except perhaps through modeling; Luoma and Presser 2009) but could nevertheless have a large influence on the outcome of a field measured bioaccumulation value.

Proportion of a given diet item to the total feeding rate, p(diet)

Not surprisingly, given the importance of diet concentration for variation in organism steady-state concentration, the simulations showed that diet choice was among the 3 largest contributors to variation in 2 out of 4 vertebrate–chemical combinations. This underlines the need for good understanding of the feeding ecology for the species used in bioaccumulation studies. Ecological parameters such as diet choice are not easily simulated in laboratory situations, and information is therefore largely dependent on what can be collected in (semi)field situations. Choice of diet can be estimated using different techniques such as observations (studying prey selection in the field or feces and pellets of predators), analysis of stable isotopes (for carbon source: δ13C, trophic level: δ15N; Peterson and Fry 1987; Post 2002; Hoekstra et al. 2003), and separation of pelagic and benthic food sources (δ34S; Connolly et al. 2004), or fatty acids (Budge et al. 2002; Iverson et al. 2004). This information can then be used to construct a food web model, helping to understand prey–predator relationships and defining appropriate food baskets for selected predators (Borgå et al. 2004 and references therein).

Concentration in sediment, Cw(pw)

For all 4 invertebrate–chemical combinations, Cw(pw), which is a surrogate measure for bulk sediment concentration (see Materials and Methods), is among the 3 most important contributors to variation. Because both of the modeled invertebrate groups are deposit-feeders, this again emphasizes the contribution of the diet chemical concentration to the observed variation. Measuring sediment concentration and calculating B-metrics (i.e., BSAF or sediment-based BAF) should be easy to implement for sediment-associated organisms, especially because these are often much more stationary than pelagic species. However, as discussed below, this obviously also has to be related to bioavailability constraints such as the presence of BC. There may be an added complexity when dealing with sediment-associated organisms that switch between feeding strategies, which is only partly explored in the present simulations by the mayfly example. In cases where the organism, for example, switches between deposit-feeding and predation (as is the case for some errant polychaetes), other food items (i.e., prey or recently settled particles) may also play a role in determining steady-state concentrations. Therefore, the availability of, and preference for, other food items than sediment could in some cases contribute to variation in steady-state concentrations, as seen for the vertebrate species, even though this was not thoroughly explored by the present simulations. In that case similar approaches with identifying preferred food items, as suggested above, could be applied also for sediment-associated invertebrates.

Black carbon parameters; the Freundlich coefficient, nfBC, and constant, KfBC

Partitioning to black carbon, parameterized as the Freundlich coefficient and constant, was the most (in one case, the second most) important process driving variation in chemical accumulation in the 4 invertebrate–chemical combinations. In both field and laboratory settings, the uncertainties related to this variability can be strongly reduced by measuring appropriate BC concentrations on a site-specific basis in conjunction with field or bioassay results. Sediment-ingesting organisms take up contaminants during digestion. This process will also be influenced by sediment composition (e.g., BC content) (Janssen et al. 2010). Increased desorption and/or uptake due to gut fluids can be estimated by extracting contaminated sediments with synthetic gut fluids shown to mimic polychaete gut fluids (Voparil and Mayer 2004). Increased uptake from sediment due to digestion of particles is strongly influenced by sorption of the contaminant to the particle type and digestibility of the particle itself (Gunnarsson et al. 1999; McLeod et al. 2004). Data gaps related to the digestability and nutritional contributions of individual sediment components were identified in this study and should be further explored to improve model parameterization in future applications.

Feeding rate, Qfeed

For some of the simulated organism–chemical combinations, the feeding rate was among the 3 most important contributors to variation. Given that diet and chemical bioavailability are generally very important for determining variability in B-measures, it is not surprising that the amount fed per time unit and organism weight is important for the achieved steady-state concentration, because feeding rate determines the residence time in the gut and therefore time for chemical assimilation. Feeding rate can be measured in the laboratory for some (especially deposit feeding) organisms by relatively simple experiments, whereas it is not straightforward for others. Laboratory-measured feeding rates contain an additional complexity preventing them from being easily extrapolated directly to field conditions, because feeding rates vary considerably with different factors, such as food quality and quantity as well as temperature and metabolic rate relationships (e.g., Chipps 1998; Granberg and Forbes 2006).

Chemical biotransformation, km

Not surprisingly, chemical biotransformation is an important contributor to observed variation for some organism and chemical combinations. More specifically, in the 2 vertebrate model organisms, biotransformation was a major player in the variation observed for pyrene. Chemical biotransformation can be tested in the laboratory using standard protocols and should involve a variety of species, at least 1 or 2 invertebrates and a fish. Biotransformation capacity of organisms at higher trophic levels can be tested using in vitro bioassays using liver microsomes and appropriately scaled up to an in vivo estimate (Boon et al. 1998). Even though biotransformation capacity of chemicals is dependent on both chemical type and species, results will roughly show whether a chemical is resistant to biotransformation or can be metabolized by invertebrate or vertebrate species. In cases where initial simulations indicate very rapid biotransformation rates (e.g., less than 1 d), alternative models such as physiologically based toxicokinetic models should be considered.

SUMMARY AND CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

In this article, we evaluated why laboratory and field bioaccumulation measures do not align, by examining the parameters contributing to the observed variance in bioaccumulation measures in the field, and identified the most important factors that lead to the observed divergence between laboratory and field bioaccumulation measures using a modeling approach. The advantage of the probabilistic modeling approach employed in the present study is that it allowed us to integrate all selected parameters (up to 56) in a simple and transparent manner and directly examine the influence of parameter error and mean on bioaccumulation (i.e., contribution of each parameter to variation in steady-state concentration in the organism) in all 8 organism–chemical bioaccumulation scenarios.

The results from the simulations clearly demonstrated that the input flux of both model compounds (pyrene and PCB-153) was dominated by uptake from sediment and diet in all 8 organism–chemical simulations, which add to the discussion on the appropriateness of BCF as an endpoint for hydrophobic contaminants. Because food is also the main contributor to variance for all 4 organisms, we suggest that the importance of food is a more general trend when considering bioaccumulation risk for moderate to highly hydrophobic chemicals. We found that uptake of contaminants through ingestion and the bioavailability of contaminants from sediment particles (especially BC) is important. The importance of ingestion was further emphasized by the assessment of the most critical parameters for variation in bioaccumulation of the 2 model compounds. The variation in bioaccumulation was mainly covered by 10 of the 56 input parameters for all 4 model organisms. These critical parameters explained between 66% and 96% of variation in bioaccumulation for a diverse group of organisms covering both aquatic and terrestrial organisms with different feeding modes (sediment and/or seston ingestion versus prey items). The relative contributions of the different parameters to the predicted variability in organism bioaccumulation were strongly related to feeding mode (sediment versus prey) among species, whereas the impact of chemical was moderate. At lower trophic levels, as illustrated by the mayfly and polychaete, variability in bioaccumulation was mainly driven by sediment exposure, sediment composition, and chemical partitioning to sediment components. At higher trophic levels, illustrated by yellow perch and the little owl, food web structure (i.e., diet composition) and diet concentrations become more important, particularly for the highly persistent compound PCB-153. Thus, it was evident that at higher trophic levels, environmental and physiological factors played an especially dominant role in explaining the variability in bioaccumulation, whereas the physical and/or chemical and environmental properties of the compounds were more important at lower levels in the food web. Although this may seem trivial and also may be considered a conditional result dependent on the chosen model structures, it is very insightful that a B-assessment for benthic species suffers most from uncertainty in partitioning, whereas an accurate B-assessment for higher predatory species like owl and perch is mainly limited by uncertainty with respect to concentrations in diet and diet composition (including food web dynamics). Because these 2 issues are highly variable, bioaccumulation always is highly conditional. In addition, uncertainty in accumulation in prey items at lower trophic levels may also affect the uncertainty of accumulation in predators at higher trophic levels. This was partially considered in the yellow perch simulations through the model linkage to mayfly output but not included explicitly in the simulations. For sediment-dwelling organisms, chemical bioavailability from ingested food (chiefly sediment components) played a major role contributing to variability in steady state concentrations in the simulations. Therefore, we suggest that the way to minimize uncertainty in B-assessments in the future is to identify food sources, base the B-metrics on the actual bioavailable chemical concentrations in the food items, and direct future research to the distribution and type of BC in natural sediments and the effect of different sediment components (BC, LOM, and so forth) for chemical absorption efficiencies in different organisms.

It should be noted that sensitivity analyses reported in this article are strongly determined by the inputs and parameter error ranges associated with a given simulation trial. Attempts were made to base simulations on realistic situations where possible; however, data gaps and lack of systematic investigation of the actual boundaries among all parameters included in the model limit the degree of comprehensiveness of model simulation output. In addition, many parameters included in the model exhibit interactions with one another, and these interactions are not considered in the uncertainty simulations. We can only guess which parameters are likely to covary and which are truly independent of one another. For example, lower feeding rates (Qfeed) are likely to be associated with lower growth rates, which in turn may affect other parameters such as diet digestibility and chemical assimilation efficiency from ingested food. Another obvious example is ventilation rate (Qv) and feeding rate (Qfood), which are both functions of metabolic rate. However, in none of the simulations were these parameters significant contributors to model output variability in combination with one another. Another condition of the model is that the simulations acted at different scales in time and space regarding interaction among chemical properties, environmental conditions, and animal behavioral differences between the organisms and compounds. These interactions would presumably constrain uncertainty, and therefore failure to account for such parameter correlations in the Monte Carlo simulations would be expected to result in overestimates of overall model uncertainty. However, across the organism groups, the range of uncertainty associated with model simulations overlapped with between 73% and >100% (polychaetes only) of the empirically measured variation in field data sets. Thus, much of the field and laboratory bioassay data could be explained by uncertainty in model simulations.

The uncertainty analysis provided in this article illustrates the potential of applying models in B assessments to provide concentration estimates and uncertainty around such estimates for a given set of predefined conditions and serve as a tool to identify those model parameters contributing most to variation in bioaccumulation. Use of probabilistic models allows us to determine bioaccumulation in a quantitative manner which undoubtedly has advantages regarding the role of B metrics in evaluation of risk criteria. Evidently, several criteria are required to be fulfilled prior to implementing the model as a tool in “B” assessments including factors such as transparency, robustness, and validation (e.g., testing model output against empirical data). As discussed in this article, the latter is complicated by the scarcity of this kind of information in the literature.

The implication from the present study is that variation in bioaccumulation measures can be significantly reduced by implementing a few methodological modifications and/or changes. Furthermore, probabilistic models may play an important role scientifically as well as in a regulatory context by improving our understanding of the underlying processes that control bioaccumulation, directing sampling methods to improve accuracy of bioaccumulation measures in field samples, providing information on sampling design (i.e., consideration of environmental heterogeneity), as well as providing a basis for estimating the numbers of samples required to evaluate the field bioaccumulation metric against the regulatory criteria. Explaining variation will further increase the confidence in using models as extrapolation tools (laboratory to field); however, risk assessors should not use these models to replace field sampling at least until the models have been robustly validated through use of focused experiments.

EDITOR'S NOTE

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

This paper is 1 of 5 articles resulting from the “Lab-Field Bioaccumulation Workshop” held in November 2009 in New Orleans, Louisiana, USA. Workshop participants focused on three objectives: 1) compare laboratory and field measurements of bioaccumulation endpoints, 2) evaluate the reasons why laboratory and field bioaccumulation data may not align, and 3) explore the measurement and application of TMFs.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

We thank the International Life Sciences Institute, Health and Environmental Sciences Institute (ILSI–HESI), the United States Environmental Protection Agency (USEPA), and the Society of Environmental Toxicology and Chemistry, who sponsored the workshop. United States Geological Survey Priority Ecosystem Science (San Francisco Bay) provided support to R.S. Grant support from the Dutch Ministry of Agriculture, Nature Conservation and Food Quality (BO-11-006.03-003) was given to N.v.d.B.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information
  • Accardi-Dey A, Gschwend PM. 2002. Assessing the combined roles of natural organic matter and black carbon as sorbents in sediments. Environ Sci Technol 36: 2129.
  • Arnot JA, Gobas FAPC. 2004. A food web bioaccumulation model for organic chemicals in aquatic ecosystems. Environ Toxicol Chem 23: 23432355.
  • Arnot JA, Gobas FAPC. 2006. A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in fish. Environ Rev 14: 257297.
  • Barron MG. 1990. Bioconcentration. Environ Sci Technol 24: 16121618.
  • Boon JP, Sleiderink HM, Helle MS, Dekker M, van Schanke A, Roex E, Hillebrand MTJ, Klamer HJC, Govers B, Pastor D., et al. 1998. The use of a microsomal in vitro assay to study phase I biotransformation of chlorobornanes (toxaphene(R)) in marine mammals and birds: Possible consequences of biotransformation for bioaccumulation and genotoxicity. Comp Biochem Phys C 121: 385403.
  • Borgå K, Fisk AT, Hoekstra PF, Muir DCG. 2004. Biological and chemical factors of importance in the bioaccumulation and trophic transfer of persistent organochlorine contaminants in arctic marine food webs. Environ Toxicol Chem 23: 23672385.
  • Budge SM, Iverson SJ, Bowen WD, Ackman RG. 2002. Among- and within-species variability in fatty acid signatures of marine fish and invertebrates on the Scotian Shelf, Georges Bank, and southern Gulf of St. Lawrence. Can J Fish Aquat Sci 59: 886898.
  • Burkhard LP, Arnot JA, Embry MR, Farley KJ, Hoke RA, Kitano M, Leslie HA, Lotufo GR, Parkeron TF, Sappington KG, Tomy GT, Woodburn KB. 2012. Comparing laboratory and field measured bioaccumulation endpoints. Integr Environ Assess Manag 8: 1731.
  • Burtnyk MD, Paterson G, Drouillard KG, Haffner GD. 2009. Steady and non-steady state kinetics describe polychlorinated biphenyl bioaccumulation in natural populations of bluegill (Lepomis macrochirus) and cisco (Coregonus artedi). Can J Fish Aquat Sci 66: 21892198.
  • Chipps SR. 1998. Temperature-dependent consumption and gut-residence time in the opossum shrimp Mysis relicta. J Plankton Res 20: 24012411.
  • Ciavatta S, Lovato T, Ratto M, Pastres R. 2009. Global uncertainty and sensitivity analysis of a food-web bioaccumulation model. Environ Toxicol Chem 28: 718732.
  • Clark KE, Gobas FAPC, Mackay D. 1990. Model of organic chemical uptake and clearance by fish from food and water. Environ Sci Technol 24: 12031213.
  • Clark TP, Norstrom RJ, Fox GA, Won HT. 1987. Dynamics of organochlorine compounds in herring gulls (Larus argentatus): II. A two-compartment model and data for ten compounds. Environ Toxicol Chem 6: 547559.
  • Connolly R, Guest M, Melville A, Oakes J. 2004. Sulfur stable isotopes separate producers in marine food-web analysis. Oecologia 138: 684691.
  • Corkum LD, Ciborowski JJH, Lazar R. 1997. The distribution and contaminant burdens of adults of the burrowing mayfly, Hexagenia, in Lake Erie. J Great Lakes Res 23: 383390.
  • Cornelissen G, Gustafsson Ö, Bucheli TD, Jonker MTO, Koelmans AA, van Noort PCM. 2005. Critical Review. Extensive sorption of organic compounds to black carbon, coal, and kerogen in sediments and soils: Mechanisms and consequences for distribution, bioaccumulation and biodegradation. Environ Sci Technol 39: 68816895.
  • Council of the European Union. 2006. Regulation (EC) no …/2006 of the European Parliament and of the Council concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC of the European Parliament and of the Council and repealing Council Regulation (EEC) 793/93 and Commission Regulation (EC) 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. Brussels (BE): Council of the European Union. 673 p.
  • Dauwalter DC, Fisher WL. 2008. Ontogenetic and seasonal diet shifts of smallmouth bass in an Ozark stream. J Freshw Ecol 23: 113121.
  • De Laender F, Van Oevelen D, Middelburg JJ, Soetaert K. 2010. Uncertainties in ecological, chemical and physiological parameters of a bioaccumulation model: Implications for internal concentrations and tissue based risk quotients. Ecotoxicol Environ Saf 73: 240246.
  • Debruyn AMH, Gobas FAPC. 2006. A bioenergetic biomagnification model for the animal kingdom. Environ Sci Technol 40: 15811587.
  • Drouillard KG, Ciborowski JJH, Lazar R, Haffner GD. 1996. Estimation of the uptake of organochlorines by the mayfly Hexagenia limbata (Ephemeroptera:Ephemeridae). J Great Lakes Res 22: 2635.
  • Drouillard KG, Fernie KJ, Smits JE, Bortolotti GR, Bird DM, Norstrom RJ. 2001. Bioaccumulation and toxicokinetics of 42 PCB congeners in American kestrels (Falco sparverius). Environ Toxicol Chem 20: 25142522.
  • Drouillard KG, Norstrom RJ. 2003. The influence of diet properties and feeding rates on the bioaccumulation and toxicokinetics of PCBs in ring does. Arch Environ Contam Toxicol 44: 97106.
  • Drouillard KG, Paterson G, Haffner GD. 2009. A combined food web toxicokinetic and species bioenergetic model for predicting seasonal PCB elimination in yellow perch (Perca flavescens). Environ Sci Technol 43: 28582864.
  • Drouillard KG. 2010. Cause and effect linkages of contamination and delisting criteria in the Detroit River Area of Concern. Report submitted to the Great Lakes Sustainability Fund (Mayfly Data). Burlington (ON): Environment Canada. 117 p.
  • El Nemr A, Said TO, Khaled A, El-Sikaily A, Abd-Allah AMA. 2007. The distribution and sources of polycyclic aromatic hydrocarbons in surface sediments along the Egyptian Mediterranean coast. Environ Monitor Assess 124: 343359.
  • Environment Canada. 2003. Guidance manual for the categorization of organic and inorganic substances on Canada's Domestic Substances List: Determining persistence, bioaccumulation potential, and inherent toxicity to non-human organisms. Existing Substances Branch. [cited 2011 July 14]. Available from: www.ec.gc.ca/substances/ese/eng/dsl/cat:index.cfm
  • [EU] European Union. 2003. Technical Guidance Document on risk assessment. Ispra (IT): Institute for Health and Consumer Protection. European Chemicals Bureau. 112 p.
  • Fisk AT, Norstrom RJ, Cymbalisty CD, Muir DCG. 1998. Dietary accumulation and depuration of hydrophobic organochlorines: bioaccumulation parameters and their relationship with the octanol/water partition coefficient. Environ Toxicol Chem 17: 951961.
  • Gevao B, Beg MU, Al-Omair A, Helaleh M, Zafar J. 2006. Spatial distribution of polychlorinated biphenyls in coastal marine sediments receiving industrial effluents in Kuwait. Arch Environ Contam Toxicol 50: 166174.
  • Gewurtz SB, Lazar R, Haffner DG. 2000. Comparison of polycyclic aromatic hydrocarbon and polychlorinated biphenyl dynamics in benthic invertebrates of Lake Erie, USA. Environ Toxicol Chem 19: 29432950.
  • Gobas FAPC. 1993. A model for predicting the bioaccumulation of hydrophobic organic chemicals in aquatic food-webs: Application to Lake Ontario. Ecol Model 69: 117.
  • Gobas FAPC, Mackay D. 1987. Dynamics of hydrophobic organic chemical bioconcentration in fish. Environ Toxicol Chem 6: 495504.
  • Gobas FAPC, Bedard DC, Ciborowski JJH, Haffner GD. 1989. Bioaccumulation of chlorinated hydrocarbons by the mayfly (Hexagenia limbata) in Lake St. Clair. J Great Lakes Res 15: 81588.
  • Gobas FAPC, Morrison HA. 2000. Bioconcentration and biomagnification in the aquatic environment. In: Boethling RS, Mackay D, editors. Handbook of property estimation methods for chemicals: Environmental and health sciences. Boca Raton (FL): CRC. p 189231.
  • Gobas FAPC, de Wolf W, Burkhard LP, Verbruggen E, Plotzke K. 2009. Revisiting bioaccumulation criteria for POPs and PBT assessment. Integr Environ Assess Manag 5: 624637.
  • Goldman LI. 2002. Crystal ball software tutorial: Crystal ball professional introductory tutorial. In: Yücesan E, Chen CH, Snowdon JL, Charnes JM, editors. Proceedings of the 34th conference on Winter Simulation: Exploring New Frontiers. American Statistical Association. 2002 Dec 8-11; San Diego, CA: p 15391545.
  • Goerke H, Weber K. 2001. Species-specific elimination of polychlorinated biphenyls in estuarine animals and its impact on residue patterns. Mar Environ Res 51: 131149.
  • Granberg ME, Forbes TL. 2006. Role of sediment organic matter quality and feeding history in dietary absorption and accumulation of pyrene in the mud snail (Hydrobia ulvae). Environ Toxicol Chem 25: 9951006.
  • Greulich K, Pflugmacher S. 2004. Uptake and effects on enzymes of cypermethrin in embryos and tadpoles of amphibians. Arch Environ Contam Toxicol 47: 489495.
  • Gunnarsson JS, Granberg ME, Nilsson HC, Rosenberg R, Hellman B. 1999. Influence of sediment-organic matter quality on growth and polychlorobiphenyl bioavailability in echinodermata (Amphiura filiformis). Environ Toxicol Chem 18: 15341543.
  • Harnois E, Couture R, Magnan P. 1992. Seasonal-variation in food-resource partitioning of 5 fish species in relation to prey availability. Can J Zool 70: 796803.
  • Hawthorne SB, Grabanski CB, Miller DJ. 2007. Measured partition coefficients for parent and alkyl polycyclic aromatic hydrocarbons in 114 historically contaminated sediments: Part 2. Testing the two carbon-type model. Environ Toxicol Chem 26: 25052516.
  • Hauck M, Huijbregts MAJ, Koelmans AA, Moermond CTA, Van den Heuvel-Greve MJ, Veltman K, Hendriks AJ, Vethaak AD. 2007. Including sorption to black carbon in modeling bioaccumulation of polycyclic aromatic hydrocarbons: Uncertainty analysis and comparison to field data. Environ Sci Technol 41: 27382744.
  • Hauck M, Hendriks HWM, Huijbregts MAJ, Ragas AdMJ, van de Meent D. 2011. Parameter uncertainty in modeling bioaccumulation factors of fish. Environ Toxicol Chem 30: 403412.
  • Hewett AJ, Johnson LB. 1992. A generalized bioenergetics model of fish growth for microcomputers. Madison (WI): Univ of Wisconsin Sea Grant Institute. 47 p.
  • Henderson BA, Trivedi T, Collins N. 2000. Annual cycle of energy allocation to growth and reproduction of yellow perch. J Fish Biol 57: 122133.
  • Hendriks JA. 1995. Modelling non-equilibrium concentrations of microcontaminants in organisms: comparative kinetics as a function of species size and octanol water partitioning. Chemosphere 30: 265292.
  • Hoekstra PF, O'Hara TM, Fisk AT, Borgå K, Solomon KR, Muir DCG. 2003. Trophic transfer of persistent organochlorine contaminants (OCs) within an Arctic marine food web from the southern Beaufort-Chukchi Seas. Environ Pollut 124: 509522.
  • Hounsome T, O'Mahony D, Delahay R. 2004. The diet of little owls Athene noctua in Gloucestershire, England. Bird Study 51: 282284.
  • Iverson SJ, Field C, Bowen WD, Blanchard W. 2004. Quantitative fatty acid signature analysis: a new method of estimating predator diets. Ecol Monogr 74: 211235.
  • Janssen EML, Croteau M-N, Luoma SN, Luthy RG. 2010. Measurement and modeling of Polychlorinated biphenyl bioaccumulation from sediment for the marine polychaete Neantes arenaceodentata and response to sorbent amendment. Environ Sci Technol 44: 28572863.
  • Jorgensen A, Giessing AMB, Rasmussen LJ, Andersen O. 2008. Biotransformation of polycyclic aromatic hydrocarbons in marine polychaetes. Mar Environ Res 65: 171186.
  • Jørgensen SE. 1990. Modelling concepts. In: Jørgensen SE, editor. Modelling in ecotoxicology: Developments in environmental modelling 16. Amsterdam (NL): Elsevier. p 1533.
  • Jonker MTO, Koelmans AA. 2002. Sorption of polycyclic aromatichydrocarbons and polychlorinated biphenyls to soot and soot-like materials in the aqueous environment: Mechanistic considerations. Environ Sci Technol 36: 37253734.
  • Kane-Driscoll SB, McElroy AE. 1996. Bioaccumulation and metabolism of benzo[a]pyrene in three species of polychaete worms. Environ Toxicol Chem 15: 14011410.
  • Kashian DR, Drouillard K, Haffner D, Krause A, Liu Z, Sano L. 2010. What are the causes, consequences and correctives of fish contaminants in the Detroit River AOC that cause health consumption advisories? Final report submitted to Michigan Sea Grant (Yellow Perch Data). 523 p.
  • Kofoed L, Forbes VE, Lopez G. 1989. Time-dependent absorption in deposit feeders. In: Lopez G, Taghon G, Levinton J, editors. Ecology of marine deposit feeders. New York (NY): Springer. p 129148.
  • Koelmans AA, Jonker MTO, Cornelissen G, Bucheli TD, Van Noort PCM, Gustafsson Ö. 2006. Black carbon: The reverse of its dark side. Chemosphere 63: 365377.
  • Koelmans AA, Kaag K, Sneekes A, Peeters ETHM. 2009. Triple domain in situ sorption modelling of organochlorine pesticides, polychlorobiphenyls, polyaromatic hydrocarbons, polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans in aquatic sediments. Environ Sci Technol 43: 88478853.
  • Korpimaki E, Norrdahl K. 1991. Numerical and functional responses of kestrels, short-eared owls and long-eared owls to vole densities. Ecol (Wash D C) 72: 814826.
  • Landrum PF, Poore R. 1988. Toxicokinetics of selected xenobiotics in Hexagenia limbata. J Great Lakes Res 14: 427437.
  • Leadley TA, Balch G, Metcalfe CD, Lazar R, Mazak E, Habowsky J, Haffner GD. 1998. Chemical accumulation and toxicological stress in three brown bullhead (Ameiurus nebulosus) populations of the Detroit River, Michigan, USA. Environ Toxicol Chem 17: 17561766.
  • Liu J, Haffner GD, Drouillard KG. 2010. The influence of diet on the assimilation efficiency of 47 polychlorinated biphenyl congeners in Japanese koi (Cyprinus carpio). Environ Toxicol Chem 29: 401409.
  • Lopez GR, Levinton JS. 1987. Ecology of deposit-feeding animals in marine sediments. Q Rev Biol 62: 235260.
  • Lundstedt-Enkel K, Tysklind M, Trygg J, Schüller P, Asplund L, Eriksson U, Häggbert L, Odsjo T, Hjelmberg M, Olsson M., et al. 2005. A statistical resampling method to calculate biomagnifications factors exemplified with organochlorine data from herring (Clupea harengus) muscle and guillemot (Uria aalge) egg from the Baltic Sea. Environ Sci Technol 39: 83958402.
  • Luoma SN, Johns C, Fisher NS, Steinberg NA, Oremland RS, Reinfelder JR. 1992. Determination of selenium bioavailability to a benthic bivalve from particulate and solute pathways. Environ Sci Technol 26: 485491.
  • Luoma SN, Presser TS. 2009. Emerging opportunities in management of selenium contamination. Environ Sci Technol 43: 84838487.
  • Luoma SN, Rainbow PS. 2005. Why is metal bioaccumulation so variable? Environ Sci Technol 39: 19211931.
  • Mayer LM, Chen Z, Findlay RH, Fang JS, Sampson S, Self RFL, Jumars PA, Quetel C, Donard OFX. 1996. Bioavailability of sedimentary contaminants subject to deposit-feeder digestion. Environ Sci Technol 30: 26412645.
  • McLeod PB, Van Den Heuvel-Greve MJ, Allen-King RM, Luoma SN, Luthy RG. 2004. Effects of particulate carbonaceous matter on the bioavailability of benzo[a]pyrene and 2,2′,5,5′-tetrachlorobiphenyl to the clam, Macoma balthica. Environ Sci Technol 38: 45494556.
  • Mehlum F. 2001. Crustaceans in the diet of adult Common and Brünnich's Guillemots Uria aalge and U. lomvia in the Barents Sea during the breeding period. Mar Ornithol 29: 1922.
  • Merritt RW, Cummins KW, editors. 1996. An introduction to the aquatic insects of North America 3rd ed. Dubuque (IA): Kendall/Hunt Publishing.
  • Moermond CTA, Traas TP, Roessink I, Veltman K, Hendriks AJ, Koelmans AA. 2007. Modeling decreased food chain accumulation of PAHs due to strong sorption to carhonaceous materials and metabolic transformation. Environ Sci Technol 41: 61856191.
  • Moermond CTA, Zwolsman JJG, Koelmans AA. 2005. Black carbon and ecological factors affect in situ biota sediment accumulation factors for hydrophobic organic compounds in flood plain lakes. Environ Sci Technol 39: 31013109.
  • Morrison HA, Gobas FAPC, Lazar R, Whittle MD, Haffner GD. 1997. Development and verification of a benthic/pelagic food web bioaccumulation model for PCB congeners in Western Lake Erie. Environ Sci Technol 31: 32673273.
  • Namdari R, Law FCP. 1996. Toxicokinetics of waterborne pyrene in rainbow trout (Oncorhynchus mykiss) following branchial or dermal exposure. Aquat Toxicol 35: 221235.
  • Nesto N, Cassin D, Da Ros L. 2010. Is the polychaete, Perinereis rullieri (Pilato 1974), a reliable indicator of PCB and PAH contaminants in coastal sediments? Ecotoxicol Environ Saf 73: 143151.
  • Nichols JW, Bonnell M, Dimitrov SD, Escher BI, Han X, Kramer NI. 2009. Bioaccumulation assessment using predictive approaches. Integr Environ Assess Manag 5: 577597.
  • Nichols JW, Fitzsimmons PN, Whiteman FW. A physiologically based toxicokinetics model for dietary uptake of hydrophobic organic compounds by fish. II. Simulation of chronic exposure scenarios. Toxicol Sci 77: 219229.
  • Paterson G, Drouillard KG, Haffner GD. 2006. Quantifying partitioning in centrarchids using stable isotope analysis. Limnol Oceanogr 51: 10381044.
  • Paterson G, Drouillard KG, Haffner GD. 2007. PCB elimination by yellow perch (Perca flavascens) during an annual temperature cycle. Environ Sci Technol 41: 824829.
  • Peterson BJ, Fry B. 1987. Stable isotopes in ecosystem studies. Annu Rev Ecol Syst 18: 293320.
  • Poot A, Quik JTK, Veld H, Koelmans AA. 2009. Quantification of black carbon in sediments and soils: Comparison of Rock-Eval analysis with traditional methods. J Chromatogr A 1216: 613622.
  • Post DM. 2002. Using stable istopes to estimate trophic position: models, methods and assumptions. Ecology 83: 703718.
  • Ronis MJJ, Walker CH. 1989. The microsomal monooxygenases of birds. Rev Biochem Toxicol 10: 301385.
  • Ruus A, Schaanning M, Øxnevad S, Hylland K. 2005. Experimental results on bioaccumulation of metals and organic contaminants from marine sediments. Aquat Toxicol 72: 273292.
  • Selck H, Palmqvist A, Forbes VE. 2003a. Uptake, depuration, and toxicity of dissolved and sediment-bound fluoranthene in the polychaete, Capitella sp. I. Environ Toxicol Chem 22: 23542363.
  • Selck H, Palmqvist A, Forbes VE. 2003b. Biotransformation of dissolved and sediment-bound fluoranthene in the polychaete, Capitella sp. I. Environ Toxicol Chem 22: 23642374.
  • Stehly GR, Hayton WL. 1988. Detection of pentachlorophenol and its glucuronide and sulfate conjugates in fish bile and exposure water. J Environ Sci Health B 23: 355366.
  • Stewart AR, Stern GA, Lockhart WL, Kidd KA, Salki A, Stainton M, Koczanski K, Rosenberg DM, Savoie DA, Billeck BN, Wilkinson P, Muir DCG. 2003. Assessing trends in organochlorine concentrations in Lake Winnipeg fish following the 1997 Red River flood. J Great Lakes Res 29: 332354.
  • Stockholm Convention on Persistent Organic Pollutants. 2001. Annex D, p 35-36. [cited 2010 April 20]. Available at: www.pops.int/
  • Thomann RV, Connolly JP. 1984. A model of PCB in the Lake Michigan lake trout food chain. Environ Sci Technol 18: 6571.
  • [USEPA] US Environmental Protection Agency. 1976. Toxic substances control act (1976). Washington (DC): USEPA.
  • Van den Brink NW, Groen NM, de Jonge J, Wegener JWM, Bosveld ATC. 2001. Ecotoxicologisch onderzoek naar effecten van verontreinigingen in uiterwaarden op steenuilen (Athene noctua): een integratie. Alterra report (in Dutch). 79 p.
  • Van den Brink NW, Groen NM, De Jonge J, Bosveld ATC. 2003. Ecotoxicological suitability of floodplain habitats in The Netherlands for the little owl (Athene noctua vidalli). Environ Pollut 122: 127134.
  • Van Wezel AP, Traas TP, van der Weiden MEJ, Crommentuijn TH, Sijm DTHM. 2000. Environmental risk limits for polychlorinated biphenyls in the Netherlands: Derivation with probabilistic food chain modeling. Environ Toxicol Chem 19: 21402153.
  • Van Zoest JGA, Fuchs P. 1988. Hunting behaviour and prey supply by a breeding pair of the Little Owl, Athene noctua. Limosa 61: 105112.
  • Voparil IM, Burgess RM, Mayer LM, Tien R, Cantwell MG, Ryba SA. 2004. Digestive bioavailability to a deposit feeder (Arenicola marina) of polycyclic aromatic hydrocarbons associated with anthropogenic particles. Environ Toxicol Chem 23: 26182626.
  • Voparil JM, Mayer LM. 2004. Commercially available chemicals that mimic a deposit feeder's (Arenicola marina) digestive solubilization of lipids. Environ Sci Technol 38: 43344339.
  • Wang WX, Fisher NS. 1999. Assimilation efficiencies of chemical contaminants in aquatic invertebrates: a synthesis. Environ Toxicol Chem 18: 20342045.
  • Wang WX, Rainbow PS. 2008. Comparative approaches to understand metal bioaccumulation in aquatic animals. Comp Biochem Phys C 148: 315323.
  • Weisbrod AV, Woodburn KB, Koelmans AA, Parkerton TF, McElroy AE, Borgå K. 2009. Evaluation of bioaccumulation using in vivo laboratory and field studies. Integr Environ Assess Manag 5: 598623.
  • Wilson LJ, Daunt F, Wanless S. 2004. Self-feeding and chick provisioning diet differ in the Common Guillemot Uria aalge. Ardea 92: 197208.

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS AND DISCUSSION
  6. GUIDANCE AND RECOMMENDATIONS IN A “B” ASSESSMENT
  7. SUMMARY AND CONCLUSIONS
  8. EDITOR'S NOTE
  9. SUPPLEMENTAL DATA
  10. Acknowledgements
  11. REFERENCES
  12. Supporting Information

All Supplemental Data may be found in the online version of this article.

FilenameFormatSizeDescription
ieam_217_sm_SuppTabS1.doc435KSupplemental Table S1
ieam_217_sm_SuppTabSI1.doc290KSupplemental Table SI 1

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.