SEARCH

SEARCH BY CITATION

Keywords:

  • Bioaccumulation;
  • Trophic magnification factor;
  • Persistent organic pollutant;
  • Food web

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information

Recent technical workgroups have concluded that trophic magnification factors (TMFs) are useful in characterizing the bioaccumulation potential of a chemical, because TMFs provide a holistic measure of biomagnification in food webs. The objectives of this article are to provide a critical analysis of the application of TMFs for regulatory screening for bioaccumulation potential, and to discuss alternative methods for supplementing TMFs and assessing biomagnification in cases where insufficient data are available to determine TMFs. The general scientific consensus is that chemicals are considered bioaccumulative if they exhibit a TMF > 1. However, comparison of study-derived TMF estimates to this threshold value should be based on statistical analyses such that variability is quantified and false positive and false negative errors in classification of bioaccumulation potential are minimized. An example regulatory decision-making framework is presented to illustrate the use of statistical power analyses to minimize assessment errors. Suggestions for considering TMF study designs and TMFs obtained from multiple studies are also provided. Alternative bioaccumulation metrics are reviewed for augmenting TMFs and for substituting in situations in which field data for deriving TMFs are unavailable. Field-derived, trophic level-normalized biomagnification factors (BMFTLs), biota–sediment accumulation factors (BSAFTLs), and bioaccumulation factors (BAFTLs) are recommended if data are available, because these measures are most closely related to the biomagnification processes characterized by TMFs. Field- and laboratory-derived BAFs and bioconcentration factors are generally less accurate in predicting biomagnification. However, bioconcentration factors and BAFs remain useful for characterizing bioaccumulation as a result of the transfer of chemicals from abiotic environmental compartments to lower trophic levels. Modeling that incorporates available laboratory and field data should also be considered for augmenting assessments of bioaccumulation potential. Modeling can provide a TMF-focused assessment for new or unreleased chemicals in the absence of field data by estimating TMF values and theoretical relationships between physical-chemical properties and TMF values (quantitative structure–activity relationships). An illustration of the use of physicochemical properties for estimating TMFs is provided. Overall, TMFs provide valuable information regarding bioaccumulation potential and should be incorporated into regulatory decision making following the suggestions outlined in this article. Integr Environ Assess Manag 2012;8:?–?. © 2011 SETAC


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information

Screening and exposure assessments for bioaccumulation use a suite of measures, including laboratory-derived bioconcentration factors (BCFs), field-determined bioaccumulation factors (BAFs), and octanol–water partition coefficients (KOW) (European Commission 2003; Arnot and Gobas 2006; Muir and Howard 2006; Gobas et al. 2009). However, in recent workshops, the use of trophic magnification factors (TMFs) and biomagnification factors (BMFs) has been encouraged. TMFs are calculated from the slope of logarithmically transformed concentrations of chemicals versus trophic level of organisms in the food web (often determined from stable N isotope ratios), whereas BMFs are calculated by dividing concentrations of chemicals in predators by concentrations in their prey (Fisk et al. 2001; Borgå et al. 2012). The TMF and BMF approaches assume that the diet is the major route of exposure to contaminants, such that bioaccumulation potential for a particular organism is directly related to its trophic position.

TMF and BMF values explicitly account for biomagnification resulting from trophic transfer, a special case of bioaccumulation in which the chemical concentration in the organism exceeds that in the organism's diet due to the magnification that occurs in the gastrointestinal tract when food is being digested and absorbed (Connolly and Pedersen 1988; Gobas et al. 1993; Gobas et al. 1999). Biomagnification can produce concentrations that are in excess of what is expected based on chemical equilibrium between the organism and its environment. Successive occurrences of biomagnification in food chains and food webs result in a scenario in which concentrations of chemicals in biota at the apex of food webs (e.g., piscivorous fish, birds, and mammals) greatly exceed those at the bottom (e.g., zooplankton and small planktivorous fish). Stakeholders (regulatory and regulated communities) are often most concerned with avoiding this scenario due to the possibility that concentrations in avian and mammalian species in the upper trophic levels, such as raptors, polar bears, and humans, for example, may reach levels associated with adverse effects. Because this pattern of accumulation is primarily influenced by biomagnification resulting from trophic transfer, rather than simple bioconcentration of the chemical from the surrounding environment, bioaccumulation potential is best measured by metrics that explicitly account for biomagnification, such as TMFs and BMFs.

BCFs, the typical metric used to quantify bioaccumulation potential, are less accurate than TMFs and BMFs in quantifying biomagnification (Gobas et al. 2009; Weisbrod et al. 2009; Borgå et al. 2012). First, chemical bioaccumulation that occurs as a result of dietary accumulation cannot be measured directly in bioconcentration tests, where exposure is via the respiratory and/or dermal routes rather than the dietary route. Second, bioconcentration is due to passive exchange of chemical between water and organism. Thus, BCFs reflect a chemical equilibrium between water and organism as long as chemicals are not quickly eliminated by biotransformation or other pathways. If chemicals are biotransformed or eliminated at significant rates (i.e., greater than the rate of elimination via the respiratory route) via other routes (e.g., fecal egestion, growth dilution), then BCFs will reflect concentrations in biota that are less than expected based on chemical equilibrium.

In addition, BCFs apply only to aquatic organisms in a laboratory context. Kelly and Gobas (2001, 2003), Kelly et al. (2007), Czub and McLachlan (2004), and Kitano (2007) have shown that BCFs and KOW-predicted BCFs are inadequate for assessing bioaccumulation potential in food webs that include air-respiring organisms. Additional reasons for the preference of TMFs and BMFs over BCFs include the difficulties associated with making reliable analytical measurements of very hydrophobic chemicals in water and accounting for the slow uptake of very hydrophobic chemicals in laboratory experiments.

Although BMFs explicitly describe enrichment of chemicals between predator and prey, BMFs represent only a single trophic transfer. Variation in the ability of organisms to biotransform and eliminate chemicals can produce variation in BMFs among predator-prey relationships in a food web, and this can obscure the chemical's overall food web biomagnification behavior (Hop et al. 2002). TMFs provide a characterization of the average degree of biomagnification that occurs in an entire food web by incorporating multiple food web interactions (Jardine et al. 2006; Borgå et al. 2012). It is therefore considered to provide the most conclusive evidence of the occurrence of food web biomagnification (Gobas et al. 2009).

Despite the advantages of TMFs for quantifying biomagnification, TMFs are more difficult to determine than other measures of bioaccumulation potential and can exhibit substantial variation that needs to be considered when the TMF is used for regulatory bioaccumulation screening purposes. At present, TMFs are empirically determined using field data. This limits the usability of the TMF to currently used chemicals that have been in commerce for a duration sufficient for concentrations in species in upper trophic levels to reach detectable levels. Data needed for TMF estimation can be absent entirely for chemicals that have not yet been released into the environment, or for which analytical methods do not exist. There is also a desire to replace destructive sampling, which sacrifices animals, with noninvasive methods in bioaccumulation assessments, such as is found in the European Union REACH (Registration, Evaluation, Authorization and restriction of CHemicals) regulation.

The objectives of this article are 1) to provide a critical analysis of the application of TMF in regulatory screening for bioaccumulation potential; and 2) to discuss alternative methods for assessing trophic biomagnification for cases where insufficient data are available to determine TMFs. This article is a product of a workshop sponsored by the Health and Environmental Sciences Institute of the International Life Sciences Institute (ILSI–HESI), the Society for Environmental Toxicology and Chemistry (SETAC), and the US Environmental Protection Agency (USEPA), held 18 to 19 November 2009 that focused on bioaccumulation science. The article is recommended to be read in conjunction with Borgå et al. (2012), provided in this issue of Integrated Environmental Assessment and Management. The article by Borgå et al. (2012) is intended for a technical audience and provides detailed technical definitions, guidance for derivation and analysis of TMF values, and suggestions for future research to improve TMF derivation techniques. Although many of these topics are relevant to the incorporation of TMFs into decision-making approaches to understand bioaccumulation potential, they are not reproduced in full within this article.

CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information

Evaluation a TMF value within a single study

When a solid study design incorporating sufficient field data is available (see recommendations in Borgå et al. 2012), TMFs are robust measures of the bioaccumulation behavior of chemicals in food webs. Scientific studies have traditionally associated bioaccumulation potential for chemicals with TMF values greater than 1 (i.e., a TMF value >1 indicates that biomagnification within a food web is occurring). However, TMF values reported in studies only represent estimates of the “true” TMF for the entire food web, a value that could only be obtained if all individuals in a food web were sampled. There is uncertainty associated with the confidence that the TMF estimate calculated by a study's limited sampling of the food web represents the TMF of the entire food web, because concentrations of chemicals and trophic position vary among individual organisms within the food web. Thus, variability associated with a study-derived TMF estimate must be evaluated statistically prior to concluding that the “true” TMF value is >1 for the entire food web. TMF values reported in biomagnification studies cannot be used to classify bioaccumulation potential without this evaluation, regardless of the degree to which they are above or below the threshold of 1.

Statistical modeling of a study's TMF estimate is usually conducted via null hypothesis testing. Statistical procedures are applied to the TMF data generated by the investigation to calculate the probability of being incorrect if one were to reject the (null) hypothesis that the TMF value for the entire food web is less than or equal to 1. A statistical significance level of 0.05 is most frequently used as a benchmark in this evaluation (Fisk et al. 2001; Hop et al. 2002; Hoekstra et al. 2003; Mackintosh et al. 2004; Houde et al. 2006; Wan et al. 2007; Houde et al. 2008; Kelly et al. 2008; Wan et al. 2008; Tomy et al. 2009; Gobas et al. 2009; Gantner et al. 2010; Borgå et al. 2012). This approach implies that investigators are confident in identifying a TMF significantly greater than 1 only when there is less than a 5% chance of committing a Type I error (i.e., identifying a chemical to be bioaccumulative whereas in reality it is not). For chemicals with TMFs that are statistically greater than 1, the absolute value of the TMF is not usually relevant in the bioaccumulative screening process, although higher TMF values indicate higher bioaccumulation potential. The absolute value of TMF may be relevant in risk assessment of species at specific trophic levels of a food web, where the regression model including baseline exposure of contaminants and trophic level (intercept) and regression slope may be able to be used to assess the risk of exceeding a threshold for negative effects in animals (Borgå et al. 2012).

Regulators also have an interest in Type II errors because of the precautionary principle applied to many regulatory decisions. Type II errors represent the likelihood of a false negative (e.g., the circumstance in which the TMF for a chemical is found to be not significantly greater than 1, whereas in reality the chemical is bioaccumulative). Hence, the statistical power of the study from which the TMF is derived is an important consideration for technical and regulatory communities. For example, a review of food web biomagnification studies (Borgå et al. 2012) found typical study conditions and experimental designs (approximately 30–40 samples) are such that a statistically significant difference from 1 is usually found only when TMF values exceed a range of 2 to 3. Thus, many studies may be unable to indicate that the TMF is significantly greater than 1, even if the “true” TMF of the entire food web is actually greater than 1 (e.g., in the range of 1 to 2). This case can lead to bioaccumulative chemicals being labeled as “not bioaccumulative.”

To account for the likelihood of false positives in the evaluation of TMFs, it is important to consider the statistical power of the study from which the TMF estimates are derived. If a TMF is reported as being not significantly different than 1, then the statistical power (or its analog, Type II error rate (β)) of the food web biomagnification data set from which it is developed should be evaluated before a chemical should be considered not bioaccumulative. For example, Fisk et al. (2001) reported a TMF 1 of 0.9 for gamma-hexachlorocyclohexane (γ-HCH) (concluding that the TMF was not significantly greater than 1). This is not surprising, because γ-HCH has not been found to biomagnify in other aquatic food web studies (Hop et al. 2002; Hoekstra et al. 2003; Kelly et al. 2007). Hence, this information indicates that the TMF for γ-HCH is less than 1 in water-respiring organisms of aquatic food webs. Despite the concurrence of multiple studies, confidence in the conclusion that γ-HCH is not bioaccumulative should ideally be evaluated in terms of the statistical power of the studies. For example, power analysis of the Fisk et al. (2001) γ-HCH data set suggests that the variability and sample size in the data set would enable a TMF as low as 1.4 to be detected as being significantly greater than 1, through use of a typical error rate for scientific studies of β = 0.20 (i.e., the occurrence of false negatives is possible in 20% of cases). Whereas many studies can only identify TMFs greater than 2 or 3 as being statistically greater than 1 suggests that the Fisk et al. (2001) data set is above average in statistical power. In contrast, the variability and study design for the γ-HCH data set from Hoekstra et al. (2003), which indicated a TMF of 0.65, are only able to detect statistical significance for TMF values of approximately 9 or higher. The lower statistical power is primarily due to the higher variability and lower sample size of the study by Hoekstra et al. (2003) compared to that of Fisk et al. (2001). Stakeholders can thus be more confident in classifying γ-HCH as not being bioaccumulative using the results of the Fisk et al. (2001) data set than when using the results of Hoekstra et al. (2003).

Power analyses are a necessary component of studies that do not detect a TMF value to be statistically significant from 1 and should be conducted when a chemical is considered for classification as not bioaccumulative, as shown in the example decision-making framework in Figure 1. If a study fails to detect bioaccumulation potential, it does not necessarily indicate that the chemical is not bioaccumulative. If power analysis indicates a study does not meet an acceptable level for detecting statistically significant TMFs greater than 1, then the use of the study for classifying a chemical as being “not bioaccumulative” is tenuous and the chemical should be evaluated further.

thumbnail image

Figure 1. Example decision-making procedure to evaluate the bioaccumulation potential for a chemical using a trophic magnification factor (TMF) associated with a particular data set.

Download figure to PowerPoint

Levels of statistical confidence used for classifying a TMF as greater than or less than 1 can be adjusted to result in more or less conservatism (Figure 1). For example, significance level of the statistical test (α) can be increased from 0.05 to 0.1 (or higher values). This will increase the likelihood of false positives (Type I errors) but lowers the likelihood of false negatives, where a conclusion is reached that a chemical is not bioaccumulative whereas in reality, the chemical is bioaccumulative (Type II errors). Balancing false positive and false negative error rates is an important issue of policy that should not be ignored by stakeholders when using TMFs to classify bioaccumulation potential.

Evaluation of TMF values among multiple studies

In addition to evaluating the statistical power of individual TMF estimates, stakeholders must often evaluate the conclusions of multiple studies. Results of various studies should be weighted according to statistical power of the results, and stakeholders should also consider the underlying practicalities of the study design. Considerations include proper balancing of samples among multiple trophic levels in a food web (e.g., benthic versus pelagic), consideration of variables such as sex, age, feeding strategy, and other life history and ecological attributes of the species sampled. Borgå et al. (2012) provides recommendations on experimental design and data analysis considerations that can be used to aid in the evaluation and comparison of results from multiple studies.

Practical considerations for comparing TMF values from multiple studies should also consider the relevance of the food webs studied. TMF values can vary among food webs to the degree that studies can arrive at different conclusions as to a chemical's bioaccumulation potential, due to the intrinsic biological characteristics of the species comprising the food webs, which may result in different accumulation patterns (Borgå et al. 2012). This disparity is highlighted by comparing biomagnification in terrestrial mammalian and aquatic poikilothermic food webs (Kelly et al. 2007). Bioaccumulation of chemicals in terrestrial ecosystems can be much higher than that observed in some aquatic food webs. Respiration is a major component of this disparity. For example, Kelly et al. (2007) found that chemicals with a high rate of respiratory elimination to water exhibit TMFs < 1 in food webs that are comprised of water-respiring organisms. In contrast, the same chemicals exhibit TMFs >1 in terrestrial food webs that include air-breathing organisms.

Even within aquatic food webs, there are often great differences in TMF values for the same food webs with and without the inclusion of samples from avian and mammalian predators (Hop et al. 2002). A TMF value developed for a food web of benthic invertebrates and small forage fish may be insufficient to portray the bioaccumulation behavior of that chemical in a more complex food web that includes higher trophic level organisms such as large piscivorous fish, birds, and mammals. For example, perfluorooctane sulfonate was shown to not biomagnify in a Lake Ontario food web that included both pelagic and benthic organisms (benthic invertebrates and fish) (Martin et al. 2004). However, when benthic organisms were excluded from the analysis, enabling an evaluation of only the pelagic food web, TMF values >1 were observed.

Including an apex predator, particularly a homeotherm, can also produce a TMF that is different from the TMF found in food webs that only include poikilotherms, because homeotherms have higher energy demands in combination with the ability to metabolize some chemicals (Fisk et al. 2001; Hop et al. 2002). TMF values generated in technically sound studies that include apex predators in relevant food webs should be given a high weight of evidence when evaluating results of multiple studies, especially if the effects of inclusion or exclusion of the apex predator upon the TMF is discussed (Hop et al. 2002; Borgå et al. 2012).

In addition to the characteristics of the species comprising the food web, a variety of other variables can also influence TMFs such that a range of TMF values may be likely among studies. Important considerations include an ecosystem's nutrient status, aquatic strata (benthic and pelagic), and geographic location (Borgå et al. 2012). Stakeholders should evaluate the relevancy of study conditions when comparing TMF results of multiple studies and/or extrapolating results to other food webs of interest.

Consideration of bioaccumulation within individual trophic levels

Because the TMF can be thought of as an “average” indication of biomagnification within a food web, limiting the evaluation of a food web data set to an evaluation of only the TMF value can obscure potentially important relationships regarding the biomagnification of chemicals between certain levels within the food web. For example, although Hop et al. (2002) did not identify significant food web biomagnification for γ-HCH, their data does suggest some potential biomagnification of γ-HCH (TMF not significantly different than 1), by seals from a proportionally weighted diet of fish and invertebrates, with BMF values of 3 to 7 (Figure 2). The apparent bioaccumulation may be due to the more efficient elimination of γ-HCH from the aquatic poikilotherms compared to that of the homeothermic seals. Although the BMFTL value does not represent the bioaccumulation potential across the entire food web, certain predator-prey relationships (such as ingestion of fish and invertebrates by seals) may be of interest to stakeholders and may be useful in screening for focused risk assessment or other evaluations.

thumbnail image

Figure 2. Average log10 lipid-normalized concentrations of gamma-hexacyclohexane (γ-HCH) in Arctic wildlife versus trophic level (data from Hop et al. 2002). Data within with dashed outlines were used by Hop et al. (2002) to calculate biomagnification factor (BMF) values for seals, assuming a mixed diet of fish and invertebrates.

Download figure to PowerPoint

TMFs based on chemical analysis of organisms do not account for transfer of chemicals from the water or air into organisms. This phenomenon is best represented by the BCF or BAF, and it is possible for BAF and BCF values to suggest bioaccumulation potential when TMF values do not. For example, although pyrene is as hydrophobic as many of the persistent organic pollutants (POPs), it is not considered a chemical that biomagnifies (Wan et al. 2007) because most fish and higher mammals can efficiently metabolize pyrene. However, many invertebrates lack the ability to metabolize pyrene, resulting in invertebrate BAFs and BCFs greater than 1 (Burkhard et al. this issue2012). Thus, bioaccumulation potential and ecological risk of pyrene accumulation may be considerable to invertebrates despite a TMF < 1.

As highlighted by the previous examples, the TMF is not always able to characterize bioaccumulation potential for a single trophic level or organism in a food web. In some cases, additional considerations and exercises (e.g., receptor-specific risk assessment, consideration of critical body residues, quantification of bioavailability, or bioaccumulation potential to invertebrates) may be necessary to more accurately classify bioaccumulation potential and characterize risk. However, metrics based on a TMF approach may be possible, such as evaluating the intercept of the regression model used to develop the TMF, or including concentrations in abiotic exposure media such as sediment in the regression (Borgå et al. 2012).

TMF ALTERNATIVES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information

If TMFs can be derived via a solid study design, they are considered the most preferred measure of the bioaccumulative nature of chemicals under ecologically relevant conditions (Gobas et al. 2009; Borgå et al. 2012). This is because the TMF appears to be the best metric for indentifying the scenario of most concern to human and environmental health (i.e., biomagnification leading to high concentrations of chemicals in species at high trophic levels). However, TMFs can only be measured in environments in which the chemical is present in detectable concentrations. This poses a significant limitation to the use of the TMF for regulatory purposes, as there are many chemicals that require bioaccumulation screening that have not been released to the environment. Also, it is possible for a newly released bioaccumulative chemical to have not reached detectable concentrations in organisms. This can occur simply because of a lag period between the release of a chemical and its presence at higher trophic levels at concentrations above detection limits. This does not imply that it is only a matter of time before a newly released chemical is observed to be bioaccumulating in food web, only that the time scales of bioaccumulation, chemical production, and other relevant fate processes be considered when evaluating TMFs for newly released chemicals. In addition, the historical and/or geographic pattern of a chemical's use may have precluded entry into routinely studied food webs (e.g., aquatic food webs in the Antarctic).

If field TMF data are not available because the TMF has not or cannot be measured reliably, there are several other methods and measures that could be explored to assess whether chemicals can be expected to biomagnify in food webs and result in a TMF > 1. These measures can be divided into: 1) field-based measures, including the field-based BMF the biota–sediment accumulation factor (BSAF) and the BAF; 2) laboratory assay results, including the laboratory-derived BMF and the BCF; 3) models, including food web bioaccumulation models that can calculate concentrations of chemicals as a function of trophic position under varying environmental conditions; and 4) quantitative structure–property relationships and related quantitative structure–activity relationships (QSARs), which relate bioaccumulation to chemical properties. These approaches are also effective for supporting the conclusions of TMF evaluations when TMF data are available, providing a multiple lines of evidence assessment of bioaccumulation potential. For these reasons, TMF alternatives and a review of proposed tiered approaches for their use in bioaccumulation potential assessments are reviewed below.

Field-derived BMF

The derivation of a field-derived BMF requires the existence of appropriate field data and can be useful in situations where field data are available but fail to meet the criteria (i.e., insufficient sample size, incomplete characterization of a food web, and so forth) for developing a TMF. BMF values can also be used to examine individual predator–prey relationships to determine if certain combinations result in greater or less accumulation than that indicated by a TMF. Although BMF values have been calculated using different mathematical approaches (Fisk et al. 2001; Borgå et al. 2004; Weisbrod et al. 2009), only the BMFTL derived by Equation 1 is directly comparable to TMF (Figure 3).

  • equation image(1)

where Cpredator and Cprey are appropriately normalized, if needed, (e.g., lipid- or protein-normalized), and chemical concentrations in the predator and prey and TLpredator and TLprey are the trophic levels of the predator and prey. When the BMFTL is used to determine the bioaccumulative nature of lipophilic chemicals, it is preferable to use the concentration of chemicals on a lipid weight basis (instead of a wet or dry weight basis), as organs and tissues can vary substantially in their lipid content.

thumbnail image

Figure 3. Biomagnification factors (BMFs) and trophic level–normalized BMFs (BMFTLs) calculated by various approaches (see text) for a hypothetical example data set where trophic magnification factor (TMF) is equal to 2.5 and log-transformed concentrations are perfectly correlated with trophic level. Only the BMFTL calculated by Equation 1 (BMFTL = 2.5) is mathematically equivalent to TMF regardless of predator–prey pairing.

Download figure to PowerPoint

The BMFTL represents a biomagnification factor normalized to a single trophic level increase in the food web. As demonstrated for a hypothetical “perfect” (i.e., r2 = 1.0) data set where TMF is equal to 2.5 (Figure 3), Equation 1 is the only approach the yields BMFs mathematically equivalent to the TMF value of 2.5, regardless of the trophic level differences in predator and prey with which is the BMFTL values are calculated. In cases in which the proportions of prey items in a predator's diet can be quantified (Hop et al. 2002), a weighted average concentration in the diet can be used for Cprey if all concentrations of the prey items are measured.

To explore the strengths and limitations of the field-derived BMFTL values to act as a surrogate of the TMF, BMFTL values were calculated (using Equation 1) from food web biomagnification studies summarized in Borgå et al. (2012). This compilation, shown in Supplemental Table S1, included data for bis(2-ethylhexyl)phthalate (DEHP), pyrene, perfluorooctane sulfonate, beta-hexachlorocyclohexane (β-HCH), γ-HCH, hexachlorobenzene (HCB), bromodiphenyl ether (BDE) congeners BDE-47 and BDE-153, polychlorinated biphenyl (PCB) congeners PCB-52, PCB-153, and PCB-209, and methylmercury compiled from several studies conducted in a variety of aquatic ecosystems (Fisk et al. 2001; Hop et al. 2002; Hoekstra et al. 2003; Mackintosh et al. 2004; Tomy et al. 2004; Houde et al. 2006; Wan et al. 2007; Houde et al. 2008; Kelly et al. 2008; Wan et al. 2008; Gantner et al. 2010). BMFTL values were calculated from predator–prey relationships for fish and invertebrate species (i.e., poikilotherms), which differed in their trophic position by 0.5 or more. Figure 4 illustrates the BMFTL values in relation to the TMF for chemicals for which the TMFs were not significantly greater than 1 (Figure 4a) and were significantly greater than 1 (Figure 4b). BMFTLs for any particular chemical show a large variability, but the majority of cases suggest that the range in BMFTL values includes the TMF. The latter can be expected, because the TMF can be viewed as the average BMF for the food web that is studied.

thumbnail image

Figure 4. Trophic magnification factor (TMF; ±95% confidence interval) and trophic level–normalized biomagnification factor (BMFTL) values (individual BMFTL values plotted as “–” below/above corresponding TMF value) calculated from datasets obtained from the selected food web biomagnification studies where TMFs were (a) not significantly greater than 1 and (b) significantly greater than 1.

Download figure to PowerPoint

Most of the data sets in Figure 4 exhibited at least 1 BMFTL value greater than 1. Ten of the 65 data sets shown in Figure 4b (bioaccumulation potential noted due to TMF significantly greater than 1) contained at least 1 BMFTL below 1. However, in these 10 data sets, the majority of BMFTL values were greater than 1. For data sets that exhibit a TMF that is not significantly different from 1 (Figure 4a), there appears to be less agreement between the BMFTL and the TMF. For example, 15 of the 18 data sets in Figure 4a contained a BMFTL > 1, whereas the TMF was not significantly different from 1. Unless a specific focus on an individual trophic level is desired, these BMFTL values could be considered “false positives” for predicting TMFs, suggesting chemicals that appear to biomagnify when, in reality, TMF values were less than 1. For example, the BMFTL for pyrene ranged from 0.006 to 200, with only 1 BMFTL value in excess of 1. Pyrene is a moderately hydrophobic polycyclic aromatic hydrocarbon (PAH) that is easily metabolized by many organisms and has not been shown to biomagnify in aquatic food webs.

Similar conclusions regarding the ability of BMFTL values to accurately predict TMFs have been reached in other studies. For example, in Arctic marine food web studies, 70% to 90% of the invertebrate and fish BMFTL values concurred with their corresponding TMF values in terms of being greater than or less than 1 (Fisk et al. 2001; Hop et al. 2002; Hoekstra et al. 2003). Discrepancies between TMF and BMFTL reflect the difficulty of a single trophic interaction being able to represent the overall degree of biomagnification that may occur in the food web. This limited review of the data confirms that BMFTL values may be an appropriate surrogate approach when TMF data are unavailable, although BMFTL values may tend to overestimate TMFs in cases where biomagnification would not be indicated by a TMF. Additional evaluation of BMFTL values (including guidance for statistical consideration) is needed. As with other measures of bioaccumulation potential, BMFTL values should not be used independently to assess the degree of biomagnification in entire food webs.

Field-derived BSAF

Biota–sediment accumulation factors (BSAFs) are field-based measurements of the chemical concentrations in the organism and the sediments calculated according to the following equation:

  • equation image(2)

where CB is the chemical concentration in the organism at steady-state, and CS is the sediment chemical concentration at steady-state. BSAFs are relatively easy to measure but are subject to propagation of error associated with the measurements in sediment and biota as discussed in Burkhard et al. (2005). Databases of BSAFs exist (Burkhard et al. 2005) and the BSAF is increasingly used in risk assessments and to develop targets for remediation and sediment quality guidelines (Gobas and Arnot 2010).

The BSAF is typically expressed in units of kilogram sediment per kilogram organism, if the concentration in the organism is expressed in units of gram chemical per kilogram of organisms and the concentration in the sediment is expressed in units of gram chemical per kilogram of sediment. The BSAF can also be expressed in units of kilogram organic C per kilogram lipid if the concentration in the organism is expressed in units of gram chemical per kilogram of lipid and the concentration in the sediment is expressed in units of gram chemical per kilogram of organic C. This lipid and organic C normalization is most appropriate for lipophilic organic chemicals. When expressed in units of kilogram organic C per kilogram lipid, the BSAF can be viewed as a fugacity ratio, measuring the increase in fugacity of the chemical in the organism over that in the sediment. The BSAF may also be normalized by trophic level to calculate a trophic level–normalized BSAF (BSAFTL) via the following equation:

  • equation image(3)

where Cbiota is appropriately normalized, if needed, (e.g., lipid- or protein-normalized), Csediment is the organic C normalized or dry weight concentration in the sediment, expressed in units of gram chemical per kg of organic C or kg sediment, respectively, TLbiota is the trophic level or position of the sampled organism or organisms, and TLsediment is the trophic position of the sediment which is typically set to 1 (Vander Zanden and Rasmussen 1996). It may be more appropriate to calculate BSAFTLs for nonlipophilic organic chemicals without lipid or organic C normalization or with normalization to another parameter (e.g., tissue protein content). Because more than 1 species in a food web can be used for the calculation of BSAFs, the BSAFTL may be useful for normalizing for the effect of trophic position and thereby serve as a metric more analogous to the TMF.

The BSAF can be a good surrogate for the TMF for persistent chemicals that are not metabolized as long as organic C and lipid have similar affinities (i.e., equal fugacity capacities) for the chemical. Seth et al. (1999) showed that organic C has, on average, approximately one-third of the sorptive capacity of octanol, a commonly used lipid surrogate. Other studies have suggested that “black carbon” (a type of C that can be found in sediment) can have a much greater sorptive capacity than octanol and lipid, suggesting that the sorptive capacity of organic C in sediment can vary greatly according to the types of C present (Cornelissen et al. 2005, 2006). In addition, nonequilibria between chemical concentrations in sediment and water can have a large effect on the value of the BSAF (Gobas and Maclean 2003). For example, high chemical fugacities in sediments relative to surface water can cause trophic transfer of chemical from sediment (via the benthic food chain) to be the main source of uptake to many organisms in the food chain. However, the low chemical fugacity in surface water can produce a high rate of chemical elimination from organism to water, which will lead to a degree of trophic magnification less than that expected under equilibrium conditions. Thus, as field-derived BSAFs are particularly sensitive to assumptions regarding steady state, it would be difficult to derive an appropriate BSAF criterion value for identifying bioaccumulative substances. Also, it may not be appropriate to use BSAFs for benthic macroinvertebrates, because the degree of biomagnification can be small compared to the error in concentration measurements, trophic level determinations, and normalization of organic C to lipid. In other words, BSAFs for macroinvertebrates can have low predictive power for biomagnification in higher trophic levels. However, BSAFs for species at higher trophic levels (e.g., seals, herring gulls) may be more informative as they can reflect a high degree of biomagnification in relation to the error in the required measurements (Gobas and Arnot 2010).

Despite some of the limitations of BSAFs for acting as surrogates for the TMF, the BSAFTL correlates well with the TMF (i.e., log BSAFTL = 0.89 (± 0.09) log TMF – 0.31 (± 0.04), n = 54, r2 = 0.67, p < 0.05) (Figure 5; data from Mackintosh et al. [2004, 2006]). The chemicals evaluated include phthalate esters, which have TMFs ≤ 1, and PCB congeners, which have TMF values much greater than 1. Figure 5 shows that the absolute values of the BSAFTL values were consistently smaller than the TMF values, likely due to differences in sorptive capacities between organic C and lipids and/or chemical disequilibria between sediment and water (Mackintosh et al. 2004, 2006). Thus, when using BSAFTL data for identifying bioaccumulative substances, it is recommended to include measurement of certain substances with well-known bioaccumulation behavior such as PCBs (e.g., PCB-153), because BSAFTL values for these substances should be >1.

thumbnail image

Figure 5. Relationship between the trophic magnification factor (TMF) and the trophic level–normalized biota–sediment accumulation factor (BSAFTL) for phthalate esters (TMF values < 1) and PCB congeners (TMF values > 1) in striped seaperch (Embiotoca lateralis) (equation image), white spotted greenling (Hexagrammos stelleri) (▴), and spiny dogfish (Squalus acanthias) liver (●) and muscle (▪).

Download figure to PowerPoint

Field-derived BAF

The bioaccumulation factor (BAF) is typically measured in the field as the ratio of the chemical concentrations in the organism and the water:

  • equation image(4)

where chemical concentration in the organism (Cbiota) is usually expressed in units of gram of chemical per kilogram of organism. The weight of the organism can be expressed on a wet weight basis or appropriately normalized, if needed, (e.g., lipid- or protein-normalized).

The BAF is a field measurement and hence different from the BCF, which is measured under controlled laboratory conditions often according to specific protocols such as OECD 305E bioconcentration flow-through fish test (OECD 1996) or the USEPA OPPTS 850.1730 fish bioconcentration test (USEPA 1996). Another difference between the BCF and the BAF is that the BCF is measured under conditions in which the test organisms are not exposed to chemicals via the diet whereas the BAF, being a field-based measure, includes chemical exposure via all routes including the diet. This difference makes it possible to infer the rate of trophic magnification (accumulation due to dietary transfer) via calculation of a trophic level-normalized BAFTL using a laboratory-derived BCF for a prey species and a field-derived BAF for a prey species:

  • equation image(5)

where the BAF is the field-derived bioaccumulation factor, in units of liters per kilogram of organism (e.g., wet weight, lipid weight, protein weight), BCF is the laboratory-derived bioconcentration factor, also in units of liters per kilogram of organism (e.g., wet weight, lipid weight, protein weight), TLbiota is the trophic level of the organism from the field, and TLlab is the trophic level of the laboratory organism. TLlab is set to 1. This value assumes that only bioconcentration, not biomagnification, is responsible for chemical accumulation in the lowest trophic level of the food web. The BAFTL approach requires measured concentrations of chemical in a field organism and field water, and, ideally, a measurement of TLbiota. If measured TLbiota is unavailable, it may be possible to substitute a value measured in other studies. BCFs can be obtained from laboratory studies or estimated from QSARs.

The BAFTL should not be confused with the BAF, which measures the increase in chemical concentration in the organism relative to that in the water. The increase in the BAF over the BCF that is quantified by BAFTL measures the accumulation that has taken place as a result of dietary biomagnification. The BAFTL accounts for the trophic magnification that has occurred in the food web and can be viewed as a surrogate for and supplement to the TMF. As with all non-TMF measures of bioaccumulation, it should not be used alone to indicate bioaccumulation potential.

Laboratory-derived BMF

In laboratory tests, the BMF can be measured in a fashion similar to that used in the OECD and USEPA bioconcentration test protocols (OECD 1996; USEPA 1996). In BMF laboratory tests, organisms are exposed to a chemical primarily via diet (Bruggeman et al. 1981; Fisk et al. 1998). The BMF test typically includes an uptake phase, where chemical concentrations are followed over time, ideally until the chemical concentration in the organism no longer changes with time (i.e., reaching the steady state). If a steady state cannot be reached in the experiment, the uptake phase is followed by a depuration phase where organisms are exposed to uncontaminated food. The rate of decline in chemical concentration over time measured in the depuration phase can then be used to derive the chemical uptake rate from which a hypothetical steady-state concentration can be estimated.

The laboratory-derived BMF is calculated using the ratio of the chemical concentrations in the test animals at steady-state and their diet:

  • equation image(6)

where chemical concentration in the organism (Cbiota) and its diet (Cdiet) are appropriately normalized, if needed, (e.g., lipid- or protein-normalized).

Because trophic transfer in a laboratory test is controlled to a single diet-to-test organism transfer, there is no need to normalize the BMF for differences in trophic position between diet and test organism. The laboratory-derived BMF is equivalent to the TMF as long as the degree of biomagnification can be assumed to be the same across all trophic levels in the food web. The main limitation of the BMF compared to the TMF is that the laboratory-derived BMF only includes 1 species such that results may not represent trophic magnification for all species in a food web. It is possible to include multiple species BMF or TMF studies via laboratory or controlled mesocosm experiments (Liber et al. 2009); however, such experiments are much more difficult to conduct than single-species laboratory BMF experiments. Although controlled BMF experiments do not capture the full variation in trophic magnification as represented by a TMF, the BMF represents one of the best surrogates for the TMF in cases where field data are unavailable.

Bioconcentration factor

Although a previous workshop (Gobas et al. 2009) and discussion in this article have emphasized the limitations of using BCFs for bioaccumulation screening, it is important to recognize that BCFs are not without merit and do contain useful information that may be able to aid in understanding part of the process related to food web biomagnification. Although the BCF test is unable to measure the increase in chemical fugacity due to food absorption and food digestion, it is able to measure the rate of elimination. The rate of chemical elimination determines to what degree the fugacity of the chemical in the organism approaches that in the gastro-intestinal tract. If this elimination rate is high, then the chemical concentration in the organism cannot reflect the fugacity in the intestines of the consumer organism, and biomagnification will not occur. However, if the elimination rate is low, then the chemical concentration in the organisms will approach the fugacity in the intestines of the consumer organism, which is greater than that in the diet. In this case, biomagnification is expected to occur. Elimination rates that can be determined via BCF tests can therefore provide useful information about whether or not biomagnification can occur. Elimination rate constants are relatively easy to measure in bioconcentration tests as they do not necessarily require a measurement of the chemical concentration in the water, the latter of which is often analytically challenging and subject to considerable error.

Elimination rates obtained from BCF with small fish tests can be used to estimate the dietary uptake rate constant and can thus provide information on determining whether the TMF has the potential to be >1. For example, assuming the dietary consumption rate of the fish is 2% of the organism's body weight per day (i.e., 0.02 kg food · kg fish−1 · d−1), and that chemical uptake efficiency from the diet is 50% (Gobas et al., 1993), one can calculate a dietary uptake rate constant (kD) of 0.02 × 0.5 = 0.01 kg food · kg fish−1 · d−1. A BMF can be calculated from dividing kD by kE, where elimination rate constant kE is derived from a laboratory BCF experiment. For the BMF to fall below 1 (corresponding to a TMF of 1), kE has to be <0.01/d. Hence, for this example, elimination rate constants measured in BCF tests showing values >0.01/d would not indicate the potential for biomagnification in a food web.

Bioaccumulation models

In the absence of empirical data, food web bioaccumulation models can be useful tools to estimate TMFs. Food web bioaccumulation models can calculate chemical concentrations for organisms among different trophic levels. If values for trophic levels are assumed, model predictions can be used to produce estimated TMFs. Several food web bioaccumulation models exist, including bioaccumulation models for aquatic (Gobas 1993; Arnot and Gobas 2004; Gobas and Arnot 2010) and terrestrial (Czub and McLachlan 2004; Armitage and Gobas 2007) food webs. The models include the chemical properties KOA and KOW and several classes of organisms, including primary producers, zooplankton, benthic macroinvertebrates, fish, mammals, birds and humans. The models also include algorithms to calculate the effect of several biological and environmental factors on the degree of bioaccumulation. The models recognize the role of membrane permeation, bioavailability, fecal egestion, growth of the animals and the biochemical composition of the organisms (e.g., lipid content) and can incorporate environmental parameters such as temperature, organic C content of bottom and suspended sediments, and suspended particle concentrations in water and air. The models can also take into account the effect of metabolic transformation on bioaccumulation potential as long as metabolic transformation has been determined experimentally (Cowan-Ellsberry et al. 2008). However, there are currently no models, to the authors' knowledge, that can predict metabolic transformation using only chemical properties.

Figure 6 illustrates the ability of the models to estimate TMFs. The TMFs depicted in Figure 6 were calculated by Fan (2008) for a number of PCB congeners, mirex, photomirex, gamma-chlordane, α-HCH, β-HCH, γ-HCH, and HCB in the Lake Ontario aquatic food web and in the lichen–caribou–wolf terrestrial food chain in Northern Canada based on reported concentrations and trophic levels calculated from dietary analyses. All of the chemicals used in this analysis can be considered persistent chemicals and were assumed not to degrade. The model-predicted TMFs provide reasonable estimates of the measured TMFs for the test chemicals in the 2 food webs (Figure 6). The ability of food web bioaccumulation models to estimate TMFs for chemicals that can be metabolized remains untested; however, preliminary findings of ongoing studies suggest that in vitro biotransformation assays can provide useful information on metabolic transformation rates for bioaccumulation models for selected species (Han et al. 2007; Cowan-Ellsberry et al. 2008).

thumbnail image

Figure 6. Relationship between model-calculated and observed trophic magnification factors (TMFs; Fan 2008) for persistent organochlorines in the Lake Ontario aquatic food web (●) and in the lichen–caribou–wolf terrestrial food chain (▴). The line represents model-calculated TMFs being equal to observed TMFs (unity).

Download figure to PowerPoint

Octanol–water partition coefficient

The octanol–water partition coefficient (KOW) of the chemical has played a key role in bioaccumulation screening. Because empirical bioaccumulation data are only available for a very small fraction of the chemicals that require bioaccumulation screening (Arnot and Gobas 2006), the majority of chemicals are evaluated based on their KOW. Measured KOW values are available for many commercial chemicals. If not available, KOW values can be calculated based on chemical structure using models.

The KOW relates to the TMF in water-respiring organisms of aquatic food webs because for many organic chemicals, KOW provides a good surrogate value for the rate of chemical elimination from the organism to the water (Bruggeman et al. 1981). If the rate of elimination is sufficiently high (predicted for chemicals with low KOW), then biomagnification would not be expected to occur such that the TMF would be ≤1.0. Data analyses have indicated that organic chemicals with a log KOW < 4 eliminate to the water at a sufficiently high rate to prevent the occurrence of biomagnification in water-respiring organisms (Arnot and Gobas 2006). This information is relevant from a regulatory point of view, because it implies that many commercial chemicals with a log KOW < 4 can be quickly screened for biomagnification in aquatic water-respiring biota based on KOW. Several studies indicate that chemicals with a log KOW ≥ 4 have the potential to biomagnify in the food web and can exhibit a TMF > 1 (Borgå et al. 2012).

However, it is well recognized that not all chemicals with a log KOW ≥ 4 realize their inherent biomagnification potential. Chemicals that have the ability to be biotransformed by organisms at a sufficiently high rate will be eliminated from organisms at a rate high enough to prevent biomagnification in the food web. For this reason, KOW should be applied cautiously to assess the TMF for chemicals with a log KOW greater than or equal to 4 because the application of KOW-based criteria has the potential to overestimate the degree of biomagnification of chemicals in water-respiring organisms of aquatic food webs. Knowledge of the biotransformation rates of these high-KOW chemicals is required to conduct accurate assessments of the biomagnification behavior and the TMF. Standardized methods to measure in vivo biotransformation rates of high KOW chemicals in biota do not currently exist. Recently, in vitro biotransformation methods have been developed and tested to measure biotransformation rates (Han et al. 2007; Dyer et al. 2009). These methods are promising and have the potential to include biotransformation in bioaccumulation screening when methods for extrapolation of in vitro to in vivo biotransformation rates become available in the future.

Octanol–air partition coefficient

The octanol–air partition coefficient (KOA) is the ratio of the chemical concentrations in octanol (Coctanol) and air (Cair), expressed as:

  • equation image(7)

from experimental methods developed by Harner and Mackay (1995) and others. The KOA can also be predicted from chemical structure using models. The KOA is useful in predicting bioaccumulation in terrestrial food webs in a similar manner as KOW is useful for predicting bioaccumulation in aquatic food webs. Although KOW can predict chemical elimination from aquatic organism to water and can be used to predict elimination via urinary routes in terrestrial organisms, KOW cannot account for elimination of chemicals from the organism to the air through exhalation. The KOA can be used to predict this elimination rate, such that a chemical with a high KOA is expected to eliminate slowly by exhalation in air-respiring organisms (Gobas et al. 2003). If a chemical is also slowly eliminated by other routes, such as urine excretion, biotransformation, transfer to offspring, and/or growth dilution, then it exhibits the ability to biomagnify in food webs

Empirical data and models have suggested KOA can be used to predict bioaccumulation potential in terrestrial food webs. Kelly and Gobas (2001, 2003) and Kelly et al. (2007) showed that in Arctic terrestrial and marine mammalian food webs, chemicals that have a log KOW < 5 and no biomagnification potential in food webs comprising water-respiring organisms can exhibit a TMF in excess of 1 if KOA is high. Terrestrial bioaccumulation models by Gobas et al. (2003), Czub and McLachlan (2004), and Armitage and Gobas (2007) are consistent with these observations and indicate that chemicals with a log KOA greater than 5 to 6 can biomagnify in air-respiring organisms as long as the chemical is not too hydrophilic (i.e., exhibits a log KOW greater than 2) such that it is not quickly eliminated by urinary excretion. Similar KOA benchmarks for biomagnification potential were proposed by Czub and McLachlan (2004) based on models of chemical distribution in human agricultural food webs. Thus, although there are few empirical measurements that can be used to confirm model results, it can be concluded that nonionizable chemicals with a log KOA less than approximately 5 are not expected to have a potential for biomagnification in terrestrial food webs. However, because approximately two-thirds of commercial organic chemicals have a high log KOA (greater than 5) and half of these chemicals have a log KOW between 2 and 5, it is important that bioaccumulation screening considers biomagnification in food webs other than aquatic food webs (Kelly et al. 2007).

Frameworks for the application of TMF and TMF alternatives

Weisbrod et al. (2009) and Gobas et al. (2009) provide frameworks for the consideration of TMF and TMF alternative metrics in a tiered approach with regards to evaluating bioaccumulation potential, and our review is in agreement with both tiered approaches. When sufficient field and laboratory data are available for a given chemical, the following preference for metrics of bioaccumulation potential is recommended (listed highest weight of evidence to lowest): 1) Field-derived TMFs; 2) field-derived BMFTL values; 3) field-derived BSAFTLs and/or BAFTL values; 4) laboratory-derived BMFs and/or laboratory- or mesocosm-derived TMFs; 5) laboratory-derived BCFs and/or BAFs, including elimination rate information; and 6) models/QSARs based on a chemical's physical and chemical properties (e.g., KOW, KOA). Although models are noted to have a relatively low weight of evidence, models can be modified to integrate available field or laboratory information from any of the tiers and can be an extremely useful in supporting empirical evidence of bioaccumulation potential. Although we indicate that TMFs are the most accurate metric for characterizing bioaccumulation potential, consideration of all available lines of evidence is prudent.

The preference for TMF values does not imply that a full field TMF or extensive laboratory BMF study is required for all chemicals for which there may be little or no data. Regulatory decision making should continue to rely upon frameworks that follow a step-by-step process, proceeding to more intensive information collection steps only in cases where the available information suggests bioaccumulation potential. However, we advocate that TMFs and other metrics most indicative of food web biomagnification (e.g., TMF alternatives such as modeled TMFs, BMFs, and BMFTL values) should be incorporated at appropriate steps in existing decision-making frameworks.

For chemicals for which little or no data are available, we suggest that the initial step in evaluating bioaccumulation potential include an evaluation of a chemical's uses and release mechanisms to the environment relative to the likelihood of reaching aquatic and/or terrestrial food webs. If food web interaction is suggested or possible, information on a chemical's behavior in the environment, such as hydrophobicity, acid-dissociation, transformation under environmental (abiotic and biotic) conditions, and so forth, should be compiled and evaluated using existing relevant QSARs and TMF-modeling approaches. As discussed above, evidence suggesting bioaccumulation potential would be indicated by model-predicted TMFs or BMFs greater than 1. QSARs can be used to evaluate nonionizable, nonpolar organics (bioaccumulation potential suggested in cases where KOW values are greater than 4 and/or KOA values are greater than 5 (Gobas et al. 2009), although this approach may not be suitable for ionizable and/or polar chemicals (e.g., pesticides, pharmaceuticals, surfactants).

Chemicals identified as possible bioaccumulative substances by initial QSAR and modeling evaluations would then likely require empirical data collection. Initial evaluation could include traditional bioaccumulation and bioconcentration tests. BCF and BAF data from these tests could be compared to traditional bioaccumulation benchmarks (e.g., 5000 L/kg, wet weight); however, these metrics are best suited for understanding bioaccumulation for particular trophic levels (e.g., lower trophic levels). As discussed above, elimination rates derived from BCF studies and BAFTLs (calculated from BCFs and BAFs) are more informative as they are more directly analogous to the TMF. A robust laboratory evaluation is also possible via laboratory biomagnification studies (or multiple-species biomagnification studies), where bioaccumulation potential is indicated by BMF values greater than 1. All laboratory information could be used to enhance TMF model estimates for relevant food webs.

For very new or unreleased chemicals identified as possibly bioaccumulative substances by initial empirical information, it is likely that bioaccumulation potential would be assumed given that field data would be unavailable. For chemicals present at detectable levels in field organisms, limited field investigations to quantify BMFTL values, and/or full field investigations to quantify TMF values would be warranted to confirm bioaccumulation potential.

CONCLUSIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information

TMFs measure the degree of biomagnification (i.e., dietary accumulation) that occurs in food webs. Because they represent a holistic measure of biomagnification and are derived from field observations, they can represent some of the most conclusive evidence of the biomagnification behavior of a chemical substance in food webs. The ability of chemicals to biomagnify is important from a regulatory point of view because it relates to the potential for chemical accumulation in higher trophic level species such as birds, marine mammals, and human beings. We advocate that TMFs and other metrics most indicative of food web biomagnification be incorporated into existing decision-making frameworks for evaluating bioaccumulation potential.

It is important to include quantitative statistical evaluations of TMF values reported by field studies, as these TMF estimates cannot be directly compared to TMF criteria indicative of bioaccumulation potential. Variability associated with TMF estimates must be evaluated. Studies of the food web bioaccumulation behavior of chemicals usually focus on a statistical assessment that identifies TMFs greater than 1 in an attempt to minimize the probability of Type I errors (i.e., an error in which the chemical is considered to be biomagnifying (TMF greater than 1) whereas in reality, the substance is not biomagnifying). Stakeholders should also consider the probability of making Type II errors (i.e., an error in which the chemical is considered to not to be biomagnifying (TMF less than 1) whereas in reality it is biomagnifying), which is measured by the statistical power of the study. We recommend that stakeholders fully evaluate statistical hypothesis testing results when the TMF value for a chemical in a particular study is greater than 1 and engage in statistical power analysis when the TMF value is found to be not significantly greater than 1. Statistical power analyses are also helpful in evaluating the results of multiple studies. Considerations of food web relevance, as well as variables such as sex, age, feeding strategy, and other characteristics of the species comprising the study should also be evaluated when evaluating individual TMF data sets and comparing TMFs from among multiple studies. An in-depth review of the statistical procedures for evaluating TMF values (hypothesis testing, treatment of concentrations in biota that are below the detection limit, etc.) and technical considerations of ecological variables are discussed in detail in Borgå et al. (2012).

The application of TMFs for addressing bioaccumulation potential is not without caveats. A major, potential disadvantage of the TMF approach is that it reduces a number of potentially relevant interactions to a single value. This may obscure certain bioaccumulation information for trophic levels of particular interest. Supplemental analysis of food web data sets (e.g., exploration of the data with BMFTL analyses) is helpful in addressing this potential concern. In addition, TMF values based on food web studies derived from biological samples do not describe the initial transfer of chemicals from the abiotic compartment (e.g., soil, sediment, water) to organisms in lower trophic levels. This initial transfer may be of interest to stakeholders. Useful approaches to address this concern include supplementing TMF values with BCF and/or BAF data, examining the y-intercept of the regression model used in the TMF derivation, incorporation of sediment in the regression model used in the TMF derivation, and/or use of a BSAFTL metric.

Because TMFs cannot be measured for chemicals that have not been in the environment for a sufficient time, chemicals for which methods of analysis are inadequate, or for chemicals for which TMF data do not exist, we propose several TMF alternatives that can be useful in bioaccumulation screening. They include the trophic position normalized biomagnification factor (BMFTL), the trophic position normalized biota–sediment accumulation factor (BSAFTL), and the trophic position normalized bioaccumulation factor (BAFTL), as well as the laboratory-derived biomagnification factor (BMF). Other lines of evidence, such as food web bioaccumulation models and physicochemical property data, are also useful in bioaccumulation screening. Data from laboratory BCF and BAF experiments can also be incorporated into models, although BCF and BAF values are not accurate predictors of TMFs. Although none of these approaches should be used individually to substitute for a TMF, they may be combined as multiple lines of evidence in an approach to evaluate bioaccumulation potential in a manner that is analogous to an assessment of a TMF. In addition, they can also serve as useful supporting lines of evidence when field-derived TMF values are available.

EDITOR'S NOTE

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. 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. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information

The authors thank ILSI–HESI, USEPA, and SETAC for sponsoring the “Laboratory-Field Bioaccumulation Workshop” (18–19 November 2009, New Orleans, LA, USA, prior to the SETAC North America 30th annual meeting).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information
  • Armitage J, Gobas FAPC. 2007. A terrestrial food-chain bioaccumulation model for POPs. Environ Sci Technol 41: 40194025.
  • 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.
  • Borgå K, Kidd K, Muir DCG, Berglund O, Conder JM, Gobas FAPC, Kucklick J, Malm O, Powell DE. 2012. Trophic magnification factors: Considerations of ecology, ecosystems and study design. Integr Environ Assess Manag 8: 6484.
  • 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 10: 23672385.
  • 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.
  • Burkhard LP, Cook PM, Lukasewycz MT. 2005. Comparison of biota-sediment accumulation factors across ecosystems. Environ Sci Technol 39: 57165721.
  • Bruggeman WA, Martron LBJM, Kooiman D, Hutzinger O. 1981. Accumulation and elimination kinetics of di-, tri-, and tetra chlorobiphenyls by goldfish after dietary and aqueous exposure. Chemosphere 10: 811832.
  • Connolly JP, Pedersen CJ. 1988. A thermodynamic-based evaluation of organic chemical accumulation in aquatic organisms. Environ Sci Technol 22: 99103.
  • Cornelissen G, Breedveld GD, Kalaitzidis S, Christanis K, Kibsgaard A, Oen AMP. 2006. Strong sorption of native PAHs to pyrogenic and unburned carbonaceous geosorbents in sediments. Environ Sci Technol 40: 11971203.
  • Cornelissen G, Gustafsson O, Buchelli TD, Jonker MTO, Koelmans AA, Van Noort PCM. 2005. 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.
  • Cowan-Ellsberry CE, Dyer SD, Erhardt S, Bernhard MJ, Roe AL, Dowty ME, Weisbrod AV. 2008. Approach for extrapolating in vitro metabolism data to refine bioconcentration factor estimates. Chemosphere 70: 18041817.
  • Czub G, McLachlan MS. 2004. Bioaccumulation potential of persistent organic chemicals in humans. Environ Sci Technol 38: 24062412.
  • Dyer SD, Bernhard MJ, Cowan-Ellsberry C, Perdu-Durand E, Demmerle S, Cravedi JP. 2009. In vitro biotransformation of surfactants in fish. Part II – Alcohol ethoxylate (C16EO8) and alcohol ethoxylate sulfate (C14EO2S) to estimate bioconcentration potential. Chemosphere 76: 989998.
  • European Commission. 2003. Technical guidance document on risk assessment. Ispra (IT): European Commission Joint Research Centre. 1044 p.
  • Fan Y. 2008. Development and testing of a new bioenergetic/bioaccumulation model for persistent organic pollutants in aquatic and terrestrial food webs [Master thesis]. Burnaby (BC): Department of Biological Sciences, Simon Fraser Univ.
  • Fisk AT, Hobson KA, Norstrom RJ. 2001. Influence of chemical and biological factors on trophic transfer of persistent organic pollutants in the northwater polynya marine food web. Environ Sci Technol 35: 732738.
  • 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.
  • Gantner N, Power M, Lawson G, Iqaluk D, Meili M, Köck G, Borg H, Sundbom M, Solomon KR, Muir DCG. 2010. Mercury concentrations in landlocked Arctic char (Salvelinus alpinus) in the Canadian high Arctic. Part I: insights from trophic relationships in 18 lakes. Environ Toxicol Chem 29: 621632.
  • Gobas FAPC, Arnot JA. 2010. Food web bioaccumulation model for polychlorinated biphenyls in San Francisco Bay, California, USA. Environ Toxicol Chem 29: 13851395.
  • Gobas FAPC, de Wolf W, Burkhard L, Verbruggen E, Plotzke K. 2009. Revisiting bioaccumulation criteria for POPs and PBT assessments. Integr Environ Assess Manag 5: 624637.
  • Gobas FAPC, Kelly BC, Arnot JA. 2003. Quantitative structure-activity relationships for predicting the bioaccumulation of POPs in terrestrial food webs. QSAR Comb Sci 22: 329336.
  • 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, Maclean LG. 2003. Sediment-water distribution of organic contaminants in aquatic ecosystems: the role of organic carbon mineralization. Environ Sci Technol 37: 735741.
  • Gobas FAPC, Wilcockson JWB, Russell RW, Haffner GD. 1999. Mechanism of biomagnification in fish under laboratory and field conditions. Environ Sci Technol 33: 133141.
  • Gobas FAPC, Zhang X, Wells RJ. 1993. Gastro-intestinal magnification: The mechanism of biomagnification and food-chain accumulation of organic chemicals. Environ Sci Technol 27: 28552863.
  • Han X, Nabb DL, Mingoia RT, Yang CH. 2007. Determination of xenobiotic intrinsic clearance in freshly isolated hepatocytes from rainbow trout (Oncorhynchus mykiss) and rat and its application in bioaccumulation assessment. Environ Sci Technol 41: 32693276.
  • Harner T, Mackay D. 1995. Measurement of octanol-air partition coefficients for chlorobenzenes, PCBs, and DDT. Environ Sci Technol 29: 15991606.
  • Hoekstra PF, O'Hara TM, Fisk AT, Borgå K, Solomon KR, Muir DC. 2003. Trophic transfer of persistent organochlorine contaminants (OCs) within an Arctic marine food web from the southern Beaufort-Chukchi Seas. Environ Pollut 124: 509522.
  • Hop H, Borgå K, Gabrielsen GW, Kleivane L, Skaare JU. 2002. Food web magnification of persistent organic pollutants in poikilotherms and homeotherms from the Barents Sea. Environ Sci Technol 36: 25892597.
  • Houde M, Bujas TAD, Small J, Wells RS, Fair PA, Bossart GD, Solomon KR, Muir DCG. 2006. Biomagnification of perfluoroalkyl compounds in the bottlenose dolphin (Tursiops truncatus) food web. Environ Sci Technol 40: 41384144.
  • Houde M, Muir DCG, Kidd K, Guildford S, Drouillard K, Evans M, Wang X, Whittle M, Haffner D, Kling H. 2008. Influence of lake characteristics on the biomagnification of persistent organic pollutants in lake trout and their food webs. Environ Toxicol Chem 27: 21692178.
  • Jardine TD, Kidd KA, Fisk AT. 2006. Applications, considerations, and sources of uncertainty when using stable isotope analysis in ecotoxicology. Environ Sci Technol 40: 75017511.
  • Kelly BC, Gobas FAPC. 2001. Bioaccumulation of persistent organic pollutants in lichen-caribou-wolf food chains of Canada's central and western arctic. Environ Sci Technol 35: 325334.
  • Kelly BC, Gobas FAPC. 2003. An Arctic terrestrial food chain bioaccumulation model for persistent organic pollutants. Environ Sci Technol 37: 29662974.
  • Kelly BC, Ikonomou MG, Blair JD, Gobas FAPC. 2008. Bioaccumulation behavior of polybrominated diphenyl ethers (PBDEs) in a Canadian Arctic marine food web. Sci Total Environ 40: 6072.
  • Kelly BC, Ikonomou MG, Blair JD, Morin AE, Gobas FAPC. 2007. Food web–specific biomagnification of persistent organic pollutants. Science 317: 236238.
  • Kitano M. 2007. Discussion paper on bioaccumulation evaluation. UNEP/POPS/POPRC.3/INF/8. Geneva (CH): UN Environment Programme. 17 p.
  • Liber K, Gangl JA, Corry TD, Heinis LJ, Stay FS. 2009. Lethality and bioaccumulation of 4-nonylphenol in bluegill sunfish in littoral enclosure. Environ Toxicol Chem 18: 394400.
  • Mackintosh CE, Maldonado J, Hongwu J, Hoover N, Chong A, Ikonomou MG, Gobas FAPC. 2004. Distribution of phthalate esters in a marine aquatic food web: Comparison to polychlorinated biphenyls. Environ Sci Technol 38: 20112020.
  • Mackintosh CE, Maldonado JA, Ikonomou MG, Gobas FAPC. 2006. Sorption of phthalate esters and PCBs in a marine ecosystem. Environ Sci Technol 40: 34813488.
  • Martin JW, Whittle DM, Muir DC, Mabury SA. 2004. Perfluoroalkyl contaminants in a food web from Lake Ontario. Environ Sci Technol 38: 53795385.
  • Muir DCG, Howard PH. 2006. Are there other persistent organic pollutants? A challenge for environmental chemists. Environ Sci Technol 40: 71577166.
  • [OECD] Organisation for Economic Co-operation and Development. 1996. OECD guidelines for testing of chemicals. Proposal for updating guideline 305. Bioconcentration: Flow-through fish test. Paris (FR): OECD. 23 p.
  • Seth R, Mackay D, Muncke J. 1999. Estimating the organic carbon partition coefficient and its variability for hydrophobic chemicals. Environ Sci Technol 33: 23902394.
  • Tomy GT, Budakowski W, Halldorson T, Helm PA, Stern GA, Friesen K, Pepper K, Tittlemier SA, Fisk AT. 2004. Fluorinated organic compounds in an eastern Arctic marine food web. Environ Sci Technol 38: 53795385.
  • Tomy GT, Pleskatch K, Ferguson SH, Hare J, Stern G, Macinnis G, Marvin CH, Loseto L. 2009. Trophodynamics of some PFCs and BFRs in a Western Canadian Arctic marine food web. Environ Sci Technol 43: 40764081.
  • [USEPA] US Environmental Protection Agency. 1996. Ecological effects test guidelines. OPPTS 850.1730, Fish BCF. USEPA guideline 712-C-96-129. April 1996. 25 p.
  • Vander Zanden MJ, Rasmussen JB. 1996. A trophic position model of pelagic food webs: impact on contaminant biomagnification in lake trout. Ecol Monogr 66: 451477.
  • Wan Y, Hu J, Zhang K, An L. 2008. Trophodynamics of polybrominated diphenyl ethers in the marine food web of Bohai Bay, North China. Environ Sci Technol 42: 10781083.
  • Wan Y, Jin X, Hu J, Jin F. 2007. Trophic dilution of polycyclic aromatic hydrocarbons (PAHs) in a marine food web from Bohai Bay, North China. Environ Sci Technol 41: 31093114.
  • 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.

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. CONSIDERATIONS FOR REGULATORY APPLICATION OF EMPIRICAL TMF VALUES
  5. TMF ALTERNATIVES
  6. CONCLUSIONS
  7. EDITOR'S NOTE
  8. SUPPLEMENTAL DATA
  9. Acknowledgements
  10. REFERENCES
  11. Supporting Information

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

FilenameFormatSizeDescription
ieam_216_sm_SuppInfo.doc435KSupplemental Information

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.