ECOTOXICOLOGY: A Comprehensive Treatment - Chapter 34 pdf

Mathematical models con-Benthic invertebrates Particulates and suspended sediment Water column Mountain whitefish Longnose sucker Pulp effluents Sediments FIGURE 34.2 Food chain model sh

34 Fate and Transport of

Contaminants in

Ecosystems

34.1 INTRODUCTION

Food web investigations have a relatively long history in ecotoxicological research Rachel Carson’s

Silent Spring (1962) placed bald eagles and other birds of prey at the top of Elton’s trophic pyramid

and introduced the lay public to the important, but often misunderstood, concept of tion Since the publication of Carson’s influential book, literally hundreds of studies have reportedconcentrations of contaminants across trophic levels and attempted to relate trophic position to bio-magnification The goal of this chapter is not to provide a comprehensive review of these studies,which have been adequately described in several recent publications (Barber 2003, Borgå et al

biomagnifica-2004, Fisher and Wang 1998, Iannuzzi et al 1996, Zaranko et al 1997) Instead, the primary goal ofthis section is to characterize the ecological factors that influence transport of contaminants throughecosystems Because of the difficulty developing reliable food web models, researchers are keenlyaware that predicting food chain transport requires more than an understanding of the physicochem-ical properties of contaminants Quantification of feeding habits of organisms, especially those withmixed diets or that show ontogenetic changes, is often challenging The structure of food webs andthe dynamics of energy and contaminant flow also vary greatly among locations Consequently,predictive models have become increasingly sophisticated as investigators attempt to quantify theinfluence of ecological factors, such as feeding habits, food chain length, and habitat characterist-ics, on contaminant transport and biomagnification The inclusion of these ecological factors intotransport models represents a major improvement in our understanding of how contaminants aredistributed in ecosystems However, knowing the concentration of contaminants in a particular spe-cies or trophic level tells very little about the consequences of exposure The next logical step in therefinement of food web models is to relate predicted tissue concentrations to ecologically significanteffects (Cain et al 2004, Toll et al 2005)

con-in water Bioaccumulation is defcon-ined as the uptake of chemicals from either biotic (food) or otic (sediment) compartments, and bioaccumulation factors (BAFs) are calculated as the ratio ofthe concentration in organisms to the concentration in these compartments Biomagnification refersspecifically to the increase in contaminant concentration with trophic level (often after adjustingfor lipid content of the organism) If biomagnification occurs, we would expect that lipid-based

abi-737

concentrations of lipophilic contaminants should increase with trophic level Although the highestlevels of contaminants such as polychlorinated biphenyls (PCBs) and other lipophilic chemicalsare frequently measured in top predators, biomagnification is a complex phenomenon influenced

by many physicochemical, physiological, and ecological factors (Moriarty et al 1984, arty and Walker 1987) In addition to feeding habits, factors such as metabolism, growth rates,and habitat preferences of predators and prey may regulate contaminant transfer to higher trophiclevels

Mori-Bioaccumulation and bioconcentration of chemical substances are widely recognized as usefulindicators of biological effects BCFs and BAFs have been employed to predict hazard of hydrophobicorganic chemicals to aquatic organisms Persistent organic compounds with relatively large BCF orBAF values are generally considered to be of greater environmental concern than less recalcitrantmaterials The application of these concepts to predict effects of other compounds, especially metalsand other inorganic substances, is problematic Physicochemical differences between hydrophobicorganic chemicals and heavy metals limit the applicability of the BCF/BAF approach for heavymetals Furthermore, many aquatic organisms are capable of regulating internal metal concentrations,especially essential metals such as Cu and Zn, through a variety of physiological processes McGeer

et al (2003) observed extreme variability in BCF/BAF values for several metals and an inverserelationship between BCF/BAF and exposure concentrations Assuming that high values should

be indicative of greater hazard, the observed inverse relationship between BCF/BAF values andexposure concentration is inconsistent with known toxicological data These results indicate thatapplication of BCF and BAF values to assess hazard is inappropriate for metals (McGeer et al 2003)and possibly other classes of contaminants

Criticism of the use of BCFs and BAFs in hazard assessment highlights a more fundamental issueconcerning the significance of contaminant bioaccumulation Although observing elevated levels of

a contaminant in organisms is a reasonable indicator of exposure, few studies have attempted toquantify the ecological effects of bioaccumulation This is a particularly important issue for heavymetals and other classes of contaminants that are regulated What is often lacking is a fundamentalunderstanding of the mechanisms associated with bioaccumulation and a direct link to biologicaleffects Studies conducted by Cain et al (2004) and Buchwalter and Luoma (2005) have providedimportant insight into the mechanisms of metal bioaccumulation in invertebrates and attempted toexplain differential sensitivity among species based on these mechanisms These researchers relatedinterspecific variation in morphological characteristics of aquatic insects to heavy metal uptake andsensitivity Cain et al (2004) quantified interspecific variation in subcellular distributions of heavymetals between metal-sensitive and detoxified compartments in aquatic insects These differenceswere then related to observed distributions of sensitive and tolerant invertebrate species in thefield Longitudinal distributions of most species were explained by partitioning of metals betweenmetal-sensitive and detoxified fractions These two studies represent important steps in improvingour understanding of the relationship between metal bioaccumulation and ecological effects Theyalso demonstrate that important insights can be achieved by linking mechanistic-based studies ofphysiology and toxicology to ecological investigations conducted at higher levels of biologicalorganization

34.2.1 LIPIDSINFLUENCE THEPATTERNS OFCONTAMINANT

DISTRIBUTION AMONGTROPHICLEVELS

The positive relationship between the concentration of lipophilic chemicals and trophic level is

a consistent pattern reported in the literature However, the precise mechanistic explanation for thisphenomenon is not well understood The high concentration of contaminants often observed in uppertrophic levels may simply be explained by the greater levels of lipids in these organisms Kiriluk

et al (1995) reported a significant positive relationship between lipid content and trophic position

in a pelagic food web Similar results were reported by Rasmussen et al (1990) for lake trout

The observation that organisms representing higher trophic levels often have greater levels of lipidscomplicates assessments of biomagnification and requires that lipid content be considered If lipidsincrease with trophic level, the greater concentration of hydrophobic contaminants observed in toppredators reported by many studies may simply be a result of equilibrium partitioning One altern-ative is to measure lipid content in different compartments and then simply express all contaminantconcentrations on a lipid basis Using this approach, our definition of biomagnification is restrictedonly to those instances where lipid-based concentrations increase with trophic level However, ifthe concentration of a chemical does not vary in direct proportion with lipids, this approach canprovide biased results (Hebert and Keenleyside 1995) Various statistical approaches, such as ana-lysis of covariance (ANCOVA), have been employed to estimate the influence of lipid content andfood chain length on organochlorine concentrations in fish (Bentzen et al 1996) Kidd et al (1998)observed a strong positive relationship between food chain length and organochlorine concentra-tion after accounting for lipid content in fish from subarctic lakes The strength of the relationshipbetween contaminant concentration and trophic position will also be influenced by lipophilicity of thechemicals (Figure 34.1) In general, more lipophilic chemicals show stronger relationships betweenconcentration and trophic level (Kiriluk et al 1995).

Physicochemical characteristics, such as that reflected by the octanol–water partition coefficient(Kow), greatly influence uptake and transport of contaminants through food webs There is consid-erable evidence that the molecular configuration of PCBs, particularly the number and arrangement

of chlorine molecules, significantly influences uptake (Oliver and Niemi 1988) Trowbridge andSwackhamer (2002) observed preferential biomagnification of dioxin-like PCB congeners in a LakeMichigan food web Because of preferential uptake, the ratio of these highly toxic PCBs to totalPCBs increased with trophic level Because of this relationship, ecological risk assessments based

on food web models using total PCBs may underestimate potential effects on higher trophic levels.Russell et al (1999) examined the roles of chemical partitioning and ecological factors in determ-ining transfer of organic contaminants in the Detroit River Biomagnification of high-Koworganicchemicals (log10Kow > 6.3) was observed in this food web, but simple equilibrium partitioning

between lipids and water explained patterns for low-Kow chemicals (log10Kow < 5.5) Principal

component analysis (PCA) based on chemical concentrations in organisms showed greater similarity

to the observed diets of these organisms than assigned trophic positions Similar results were ted by Kucklick et al (1996) for a pelagic food web in Lake Baikal BAFs, defined as the ratio oflipid-corrected PCB concentrations in predators to those in prey, increased with log10Kowfor bothpredatory zooplankton and fish

repor-High lipophilic compound

Less lipophilic compound

15 N ( )

0.1 0.2 0.5 1 2 5 10 20 50

34.2.2 RELATIVEIMPORTANCE OFDIET ANDWATER IN

AQUATICECOSYSTEMS

Much of the debate regarding the significance of food chain transfer of contaminants in aquaticsystems focuses on the relative importance of food and water pathways For many lipophilic organiccontaminants, especially PCBs and other organochlorines, accumulation from food is generally con-sidered the primary route of exposure Although sophisticated models have been developed to predictbioconcentration from water, models that ignore aqueous exposure can provide reasonably accur-ate estimates of contaminant levels in fish (Jackson and Schindler 1996) In contrast, attempts topredict chemical concentrations in predators based only on physiological features of organisms andphysicochemical characteristics of contaminants are fraught with uncertainty (Owens et al 1994,Russell et al 1999) Failure to account for food chain transport will significantly underestimateconcentrations of organochlorines and other lipophilic chemicals (Zaranko et al 1997) Indeed, con-temporary models describing fate and transport of highly lipophilic contaminants generally include

a food chain component and account for input from sediment (Figure 34.2) Comparative studies

of different food webs have been conducted to quantify the relative importance of trophic fer and passive uptake Wallberg et al (2001) compared uptake and food chain transfer of a PCB(2,2
,4,4
,6,6
-hexachlorobiphenyl) in an autotrophic food web consisting of algae and bacteria and

trans-a heterotrophic food web consisting of btrans-acteritrans-a, fltrans-agelltrans-ates, trans-and cilitrans-ates Results showed thtrans-at trophictransfer was the dominant pathway in the heterotrophic food web, resulting in significantly elev-ated concentrations in higher trophic levels Russell et al (1999) employed multivariate analyses

to investigate the relationship between trophic level and organochlorine concentrations in a DetroitRiver food chain Lipid-based concentrations of organochlorines increased with trophic level, sup-porting the hypothesis that these chemicals biomagnified through the food chain In addition to anincrease in concentration with trophic level, PCA showed that the specific constituents of organo-chlorines varied among trophic groups Morrison et al (1997) developed and field validated a model

to predict transfer of PCBs in a pelagic food chain Results showed that 95% of the observed centrations in invertebrates and fish were within a factor of two times the predicted concentrations.The close agreement between measured and predicted concentrations suggests that the model ulti-mately may be useful for assessing effects of PCBs on aquatic organisms Mathematical models

con-Benthic invertebrates

Particulates and suspended sediment

Water column

Mountain whitefish

Longnose sucker

Pulp effluents

Sediments

FIGURE 34.2 Food chain model showing transport of contaminants in an aquatic ecosystem The size of the

arrows indicates the relative importance of each pathway (Modified from Figure 6 in Owens et al (1994).)

developed by Thomann (1981) that quantify the relative importance of exposure from diet and waterare discussed in Section 34.3.2.

Unlike the situation for PCBs and many other lipophilic organic contaminants, the relative ance of aqueous and dietary exposure to heavy metals is uncertain Most of the evidence derivedfrom laboratory studies indicates that uptake from water is a more important route of exposure thanfood, particularly for fish.1However, some investigators have suggested that dietary uptake may alsocontribute significantly to total body burdens of heavy metals (see review byDallingeret al 1987).For example, Hatakeyama and Yasuno (1987) reported that 90% of cadmium (Cd) accumulation

import-in the guppy, Poecilia reticulata, was derived from feedimport-ing on contamimport-inated chironomids

Simil-arly, Dallinger and Kautzky (1985) demonstrated that rainbow trout accumulated metals primarilythrough the diet, particularly when levels in the water were low Munger and Hare (1997) meas-ured the relative importance of diet and water as sources of Cd uptake for the predatory insect

Chaoborus in a laboratory food chain They reported no significant difference in organisms exposed

to Cd in food alone versus Cd in food and water, indicating that uptake from water was relativelyunimportant

Although food chain transfer of most metals is probably a less serious issue than for lipophilicorganic contaminants, dietary exposure should not be ignored when assessing ecological risk of heavymetals (Hansen et al 2004) Dietary exposure to heavy metals is especially contentious becausewater quality criteria are based exclusively on aqueous exposure and assume no effects from dietaryuptake Because concentrations of metals in certain biotic and abiotic compartments may be veryhigh, relatively inefficient transfer of metals through food chains can result in harmful levels Forexample, periphyton and attached algae in streams concentrate metals and other contaminants severalorders of magnitude above aqueous levels Organisms grazing these materials, such as mayflies andother benthic macroinvertebrates, are exposed to significantly elevated concentrations Irving et al.(2003) compared effects of aqueous and dietary cadmium on grazing mayflies Organisms were verytolerant of aqueous exposure (96-h median lethal concentration= 1611 µg/L), whereas exposure to

Cd through the diet significantly inhibited feeding and reduced mayfly growth Several researchershave reported that despite low transfer efficiencies for some metals, dietary exposure may havenegative effects on upper trophic levels (Farag et al 1998, Woodward et al 1994, Woodward et al.1995) This point was demonstrated convincingly in a series of laboratory experiments in whichrainbow trout were fed benthic invertebrates collected from a metal-contaminated stream (Woodward

et al 1994) Fish consuming metal-contaminated prey showed reduced growth and greater mortality

as compared to fish feeding on organisms collected from an unpolluted stream

At least part of the controversy surrounding the relative importance of aqueous versus dietaryexposure to metals involves differences in experimental designs used to expose organisms Somestudies using artificial diets have reported relatively minor effects (Mount et al 1994), whereas thoseusing field-collected organisms have observed increased mortality and reduced growth (Woodward

et al 1994) Although natural diets collected from reference and contaminated sites are more logically realistic, differences in prey composition between locations confound interpretation ofgrowth effects because of potential differences in nutritional quality An alternative experimentaldesign that addresses this problem is to expose prey species to contaminated media (e.g., periphyton

eco-or sediments) collected from field sites and then feed these prey to fish predateco-ors Hansen et al (2004)used this experimental design to assess the effects of dietary exposure to metals on the growth ofrainbow trout Fish were fed freshwater oligochaetes that had been exposed to reference and metal-contaminated sediments collected from the Clark Fork River (Montana, USA), a stream receivingmetals from historic mining and mineral processing facilities Significant reductions in growth of fishfeeding on metal-contaminated prey were attributed to elevated levels of arsenic in tissues This isone of the first studies to demonstrate a relationship between contaminated sediments and effects on

1 The notable exception is mercury that, as methylmercury, has a dominant food-linked transfer among species.

fish through dietary exposure of metals It is important to note that, from a management perspective,concerns over differences in prey nutritional quality between reference and metal-contaminated sitesmay be relatively unimportant While differences in community composition of prey may confoundour understanding of mechanisms of toxicity of dietary exposure, effects on fish are ultimately a res-ult of heavy metals, either through direct dietary exposure or because of metal-induced alterations

in prey nutritional quality

34.2.3 ENERGYFLOW ANDCONTAMINANTTRANSPORT

Quantitative approaches developed to measure energy flow in ecosystems can also be employed

to estimate the movement of contaminants across trophic levels and between biotic and abioticcompartments Odum’s (1968) box and arrow diagrams showing energy and material flow amongtrophic levels are the predecessors of contemporary contaminant transport models Although ecotox-icologists have done a reasonable job quantifying contaminant concentrations in biotic and abioticcompartments, validation of transport models requires accurate estimates of transfer rates betweentrophic levels Because these estimates are typically obtained from laboratory studies, there is someuncertainty concerning their relevance to conditions in the field Jackson and Schindler (1996)used a long-term monitoring program to estimate transfer efficiencies of PCBs from prey fishes

to salmonids in Lake Michigan Despite significant temporal changes in concentrations of PCBs

in prey, transfer efficiencies remained relatively constant over the 15-year study These findingsdemonstrate that temporal changes in PCB levels in top predators are determined primarily by con-centrations in prey species Thus, the steady decline in PCB levels in Lake Michigan salmonids overthe past 20 years (Stow et al 1995) is likely a direct result of both reduced inputs and lower PCBconcentrations in prey species

Alterations in food web structure resulting from anthropogenic perturbations have importantimplications for energy flow and trophic dynamics in aquatic ecosystems Some of the most compre-hensive examples demonstrating the cascading influences of contaminants on predator populationsand energy flow are from estuaries subjected to hypoxia (Buzzelli et al 2002, Peterson et al 2000).Loss of oysters and other benthic suspension feeders reduces the capacity of estuarine ecosystems

to regulate phytoplankton, making these systems more susceptible to nutrient enrichment Baird

et al (2004) used network analysis to quantify the movement of energy through the Neuse RiverEstuary (North Carolina, USA), a eutrophic system receiving high levels of N from agricultural,industrial, and urban sources By taking advantage of annual variation in the level of hypoxia overtwo consecutive summers (1997 and 1998), researchers demonstrated that impairments in waterquality cascaded through several trophic levels and diverted energy from consumers to microbialpathways These researchers also speculated that reduced transfer of energy to higher trophic levelsincreased the susceptibility of the Neuse River estuary to other stressors

34.3 MODELING CONTAMINANT MOVEMENT IN

FOOD WEBS

In the past several decades, there has been significant progress in the development of food web models

to predict contaminant concentrations in aquatic organisms and transport among compartments Thegoal of these models is often to estimate concentrations in organisms at different trophic levels based

on measured concentrations in abiotic compartments such as water or sediments Alternatively,researchers often use food web models to predict events outside the range of existing empirical data.The relatively simple equilibrium partitioning models based on physicochemical characteristics oforganic contaminants (e.g., Kow) have been replaced by more sophisticated compartmental, kinetic,bioenergetic, and physiological models (Landrum et al 1992) Much of this research has focused onimproving our understanding of factors that contribute to variation among species In their simplest

forms, these steady-state models predict that the concentration of contaminants in organisms is

a function of uptake from water and food minus loss due to depuration, growth dilution, metabolism,and excretion Recognition of the importance of dietary contributions to total body burdens and theincorporation of biological factors such as lipid content, reproduction, body size, age, sex, life cycle,habitat use, feeding ecology, and trophic position into these models represent major improvements

in their predictive capability However, as with the development of any mathematical model, theseimprovements have a cost Incorporating these additional parameters increases the complexity of foodweb models, thereby reducing their generality and increasing uncertainty of predictions (Borgå et al.2004) Researchers also recognize that because of the large number of species and potential feedinginteractions in most ecosystems, predicting contaminant concentrations in all species is not practical.Consequently, it is often necessary to select representative taxa from different functional groups whenconstructing contaminant transport models (Arnot and Gobas 2004) Finally, comparison of modelresults with empirical data is a critical step in this process and is required to give food web modelsthe necessary environmental realism

34.3.1 KINETICFOODWEBMODELS

Food web models developed by Thomann et al (1992) and Gobas et al (1993) have been widelyemployed to predict the bioaccumulation and transport of hydrophobic organic compounds (HOCs)

in aquatic ecosystems The models are similar in the use of lipid-normalized contaminant levels

in organisms and expressing sediment contaminant concentrations based on organic carbon levels.There are important differences between the models in the treatment of contaminant dynamics inthe benthic and planktonic compartments that may result in different estimates of bioaccumulation.Using empirical data collected from Lake Ontario, Burkhard (1998) compared the ability of bothmodels to predict BAFs of HOCs in phytoplankton, zooplankton, macroinvertebrates, and fish BAFswere generally similar for most groups; however, BAFs for compounds with log10Kowvalues>8.0

diverged significantly Although the Thomann model had greater predictive ability for phytoplankton,zooplankton, and benthic invertebrates, predicted BAFs had lower uncertainty in the Gobas model(Burkhard 1998)

Although kinetic food web models have been validated using data from several freshwater systems, especially Lake Ontario, these approaches have received considerably less attention in otherecosystems Borgå et al (2004) conducted an extensive review of biological factors that determineduptake and food chain transfer of HOCs in Arctic marine food webs They note that Arctic eco-systems offer unique advantages for the study of trophic transfer of contaminants because of theirremote location and distance from point sources, relatively simple but long food chains, and highdependence on lipid levels in most organisms The relative importance of various biological factorsvaried among HOCs and among different species, but diet and trophic levels were the most importantbiological factors for seabirds and marine mammals

eco-Parameters included in most food web models are based on point estimates of organism bodyweight, lipid content, ingestion rate, metabolism, growth, and other physiological characteristics thatdetermine bioaccumulation However, it is generally recognized that there is considerable variability

in estimates of these exposure factors, even at specific locations Iannuzzi et al (1996) conducted

a comprehensive literature review to develop probabilistic distributions for factors that determinecontaminant exposure and uptake Mechanistic food web models developed by Thomann et al (1992)and Gobas et al (1993) were applied to a relatively simple estuarine food web that included poly-chaetes, benthic forage fish, blue crabs, and stripped bass Exposure factors were represented by one

of four distributional forms (uniform, triangular, beta, or truncated normal) to derive a probabilisticfood web model Estimated concentrations of five PCB congers were within an order of magnitude

of measured concentrations, suggesting this probabilistic approach is appropriate for screening levelrisk assessment (Iannuzzi et al 1996)

Compounds that may be rapidly metabolized by aquatic organisms, such as polycyclic matic hydrocarbons (PAHs), pose significant challenges to the development of food web models.Iannuzzi et al (1996) argue that because metabolites are generally more toxic than parent compoundsand because metabolites are often detected in specific target organs, food web models developedfor these and other compounds are not very effective Nonetheless, PAHs are widely distributed inaquatic ecosystems and pose significant risks to many aquatic organisms, especially higher trophiclevels Thus, some understanding of the potential transfer of these contaminants among trophiclevels is critical for developing ecological risk assessments Using a similar framework employedfor PCBs, Thomann and Komlos (1999) developed a steady-state food web model for PAHs andapplied this model using data from a small creek in Alabama (USA) Biota-sediment accumulationfactors (BSAF), defined as the ratio of the lipid-normalized concentration of PAHs in the organism

aro-to the organic-carbon normalized concentration in the sediment, were calculated for PAHs over

a range of Kowvalues Measured concentrations of PAHs in crayfish and fish were considerably lessthan in sediments, indicating significant loss due to metabolism of the parent compounds Modelcomponents to account for this loss of PAHs included organism weight, lipid content, growth rate,respiration rate, food assimilation efficiency, and food ingestion rate

Sensitivity analysis of the model showed that metabolism in fish had a large effect on cumulation of PAHs with log10Kow > 4.5 In contrast, relatively low metabolism of the crayfish

bioac-resulted in much higher BSAF values The analysis also showed that relative contributions of foodand water varied with Kowvalues for the unsubstituted PAHs Water was the predominant route ofexposure for PAHs with log10Kowvalues between 4 and 6, and food was the predominant route atlower and higher values

Arnot and Gobas (2004) described an innovative bioaccumulation model that represented nificant improvement in the original kinetic model developed by Gobas et al (1993) These newelements included: (1) a new model to predict contaminant partitioning; (2) a new model to predictcontaminant levels in algae and phytoplankton; (3) improved estimates of gill ventilation rates based

sig-on allometric relatisig-onships; and (4) a mechanistic model to predict gastrointestinal magnificatisig-on.Improvements in the model were evaluated using empirical data collected for 64 chemicals in 35 spe-cies from three different ecosystems The modifications in the original model significantly reducedmodel bias and improved predictions for each ecosystem Arnot and Gobas (2004) note that furtherimprovements in the model will be challenging because of the large amount of variation amongindividuals within a population

34.3.2 MODELS FORDISCRETETROPHICLEVELS

Trophic exchange of contaminants can be defined with a simple model that includes contaminantconcentration, biomass in the trophic level of interest, biomass consumed from the lower trophiclevel, contaminant bioavailability, and the fraction of contaminant excreted daily (Ramade 1987) byorganisms in the trophic level of interest To begin developing such a model, the BAF is defined as

the ratio of the contaminant concentration (C) at trophic level n+ 1 and the concentration in the next

lowest trophic level, n:

The BAF for transfer n → n + 1 can be described in more detail by inclusion of the weight

of organisms in Level n + 1 (b n ), the weight of level n organisms consumed (a n), the fraction of

contaminant absorbed from ingested food ( f n), and the fraction of accumulated contaminant that is

excreted daily (k n):

BAFn,n+1=a n f n

b n k n

Substituting this more detailed version of BAFn,n+1into the relationship between C n and C n+1

given above, the following model is generated:

Generalizing this approach, one could theoretically predict the concentration in any trophic

level (r) knowing the contaminant concentration at the lowest level (C0) and the variables a i , f i , b i,

and k ifor each trophic level:

parsi-Thomann (1981) expanded this steady-state approach by including organism growth rate anduptake of contaminants from water Organism growth was incorporated because any increase inbody mass has an apparent dilution effect on contaminant concentration Inclusion of uptake fromwater allowed comparison of the relative importance of food and water sources A food chain transfer

number ( f ) serves this purpose.

In this equation,α = the chemical absorption efficiency ( f i in the simple BAF model above),

predator× day), k = excretion rate (k i in the BAF model above), and G= net organism growth rate

Thomann (1981) generalized that significant food chain transfer was indicated if f > 1, but uptake of contaminants from water was more important than food sources if f < 1 Applying this rule to PCBs,

239Pu and137Cs data, he concluded that PCB and radiocesium concentrations in top predators werepredominantly determined by food sources but accumulated plutonium came primarily from watersources Thomann (1981) also added explicit details to this steady-state model for predicting water

→ phytoplankton, phytoplankton → zooplankton, zooplankton → small fish, and small fish →

large fish transfers of contaminants Later (Thomann 1989), this approach was focused on predictingtransfer of organic chemicals in food chains by relating relevant model parameters to Kow Trophic

transfer in simple aquatic systems was predicted to be insignificant if log10Kow < 5 Food chain

transfer was important for top predators in aquatic systems if log10Kowwas between 5 and 7

34.3.3 MODELSINCORPORATINGOMNIVORY

A major shortcoming of the approaches described above is the assumption that no species feeds onmore than one trophic level Although unrealistic in many cases, this assumption allows a level ofaccuracy in predicting trophic transfer of some contaminants After noting that such an approach wasinsufficient to define trophic transfer in a pelagic food web, Cabana and Rasmussen (1994) expandedtrophic models to include “omnivory.” Here, omnivory means that a species is feeding on food itemscoming from several trophic levels Although the approach is similar to that described above, matrixformulation accommodates the increased number of trophic exchanges In this approach, fractions

of the total amount of the ith level’s diet coming from specific trophic levels ( j) are designated ρ ij.Obviously, allρ ij fractions sum to 1 in order to include the entire diet of level i The total ration to the ith level (C i) is defined as follows:

The fractions of the ith level’s diet coming from the different sources ( j levels) can be placed

into a matrix with the subdiagonal reflecting the fractions for the simple Level 1→ Level 2,

Level 2→ Level 3, Level 3 → Level 4, and so forth transfers The fractions entered below the

sub-diagonal are those for the transfers not accommodated in Thomann’s model (e.g., Level 1→ Level 3

and Level 2→ Level 4 transfers) The following relationship describes a vector of the total rations

for all trophic levels i in a trophic scheme with four levels:

Using matrix notation and omitting accumulation for all sources except food, Cabana andRasmussen (1994) redefined Thomann’s steady-state model as the following:

where, for the different trophic levels, B = a vector of BAFs, α = a vector of assimilation (chemical

absorption) efficiencies, C = a vector of rations, K = a vector of excretion rates, G = a vector

of growth rates, and I= the identity matrix They expanded this formulation to include exchanges

other than those depicted in the matrix subdiagonal (e.g.,ρ42andρ43) in the example above The

following matrix-formulated model predicts a vector of contaminant concentrations (ν) expected for the i trophic levels in a food web incorporating omnivory:

In this model,ρ is the omnivory-adjusted mean dietary concentration for each trophic level.

A major challenge to applying this approach is to obtain estimates of elements in the omnivorymatrix Some estimates of the trophic position must be obtained that includes the possibility thatspecies are feeding at various lower levels In Section 34.4.4, a technique will be described thatcan be applied to these estimates.

34.3.4 THEINFLUENCE OFLIFEHISTORY, HABITATASSOCIATIONS,

ANDPREYTOLERANCE ONCONTAMINANTTRANSPORT

Species-specific feeding habits, habitat associations, and tolerance of prey will greatly influence foodchain transfer and levels of contaminants in top predators Gewurtz et al (2000) reported significantvariation in PAH and PCB concentrations among benthic macroinvertebrate taxa collected from LakeErie, USA (Figure 34.3) The highest concentrations of both classes of compounds were observed

in the mayfly Hexagenia, organisms that inhabit and consume highly contaminated sediments and detritus Because of the importance of Hexagenia in the diet of both aquatic and terrestrial predators,

and because abundance of these organisms is increasing as a result of improvements in water quality(primarily reduced anoxia), it is likely that greater PAH and PCB exposure to the Lake Erie foodweb will occur (Gewurtz et al 2000) Differences in organochlorine concentrations among water-

fowl species from the Great Lakes were directly related to consumption of zebra mussels (Dreissena

polymorpha), an introduced species that has dramatically altered food chains in this region (Mazak

et al 1997) Variation in contaminant concentrations within populations were also explained by theproportion of zebra mussels in the diet Similarly, differences in feeding habits between populations

of small mammals also accounted for large variation in Hg bioaccumulation (Figure 34.4) Higherlevels of contamination in prey and greater transfer efficiency resulted in a 20 times higher con-centrations of Hg in insectivorous mammals (shorttail shrew) compared to omnivorous mammals(white-footed mouse) (Talmage and Walton 1993) Finally, several investigations have reported thatconcentrations of contaminants in aquatic systems are often higher in small prey organisms compared

to larger individuals (van Hattum et al 1991, Kiffney and Clements 1993) This phenomenon may bepartially explained by the greater surface area to volume ratio of small individuals Regardless ofthe explanation, predators that select smaller prey species, such as juveniles and early life stages,may be at greater risk from contaminant exposure (Farag et al 1998)

Habitat associations of prey species will contribute to variation in contaminant levels amongpredators Contaminated habitats are typically characterized by reduced species diversity and a shift

in community composition from sensitive to tolerant species Prey species directly associated with

Mussels Amphipods Crayfish

FIGURE 34.3 Concentrations (µg/g, lipid basis) of total PAHs and total PCBs measured in benthicmacroinvertebrates collected from Lake Erie, USA (Data from Table 1 in Gewurtz et al (2000).)

Litter Soil

Vegetation Herbiv orous in ver ts

Omniv orous mammals

Detr itivores

Car nivorous in

ver ts

Insectiv orous mammal

0.3 1 3 10 30 100 300

mam-the most contaminated compartments in mam-these systems (e.g., sediments, periphyton) are likely to havesignificantly elevated levels of chemicals Several investigators have shown that feeding habits ofpredators at impacted sites may be modified to include these tolerant and highly contaminated preyspecies (Clements and Livingston 1983, Jeffree and Williams 1980, Livingston 1984) For example,Jeffree and Williams (1980) reported that fish switched from pollution-sensitive to pollution-tolerantprey in streams polluted by mining effluents As described above, these shifts in feeding habits arelikely to influence contaminant levels in top predators

Pollution-tolerant species employ a variety of mechanisms to detoxify contaminants, ing increased excretion, storage, and compartmentalization The specific method of detoxificationemployed by prey species in polluted environments may influence bioavailability and food chaintransfer In particular, organisms that store or compartmentalize contaminants may pose a signi-ficant risk to predators This phenomenon, called the “food chain effect,” has been reported forspecies inhabiting metal-polluted environments (Dallinger et al 1987) In a laboratory study, fishfed Cd-contaminated mussels accumulated approximately two times higher metal levels than fish fedCd-contaminated chironomids, despite greater metal concentrations in the chironomids (Langevoord

includ-et al 1995) These differences were related to differences in dinclud-etoxification mechanisms binclud-etween thetwo species Wallace et al (1998) showed that metal-tolerant oligochaetes accumulated four timesmore Cd than nonresistant organisms when exposed in the laboratory However, because of dif-ferences in regulatory mechanisms employed by resistant and nonresistant prey (storage in metalrich granules vs metallothionein), metals in nonresistant oligochaetes were more bioavailable topredators Cain et al (2006) noted that rates of Cd uptake by caddisflies were similar in organismscollected from reference and metal-polluted streams However, a larger fraction of Cd was associatedwith metallothionein-like proteins in caddisflies from the metal-polluted stream

Finally, variation in life history characteristics of dominant prey species may control contaminantuptake and transfer to higher trophic levels Elevated concentrations of persistent organic pollutants

in alpine and subalpine lakes compared to montane lakes have been attributed to greater deposition

by snowfall (Blais et al 1998) Life history characteristics of dominant prey species in these systemsplay an important role in determining uptake and transport of organochlorines Blais et al (2003)measured levels of persistent organic contaminants in amphipods from lakes along a 1300 m eleva-

tion gradient in Alberta, Canada Concentrations of semivolatile organic compounds in Gammarus

lacustris increased with elevation Most of the variation in contaminant accumulation was explained

by the slower growth rates and higher lipid content of amphipods from alpine lakes Because pods are an important component of the food web in these lakes, it is likely that top predators willalso be exposed to higher levels of these persistent contaminants

amphi-34.3.5 TRANSPORT FROMAQUATIC TOTERRESTRIALCOMMUNITIES

While the majority of studies investigating food chain transport of contaminants have focused oninvertebrates and fish, a few researchers have attempted to quantify movement from aquatic systems

to avian and mammalian predators Export of contaminants from aquatic to terrestrial ecosystemscan be significant in some situations, posing risks to terrestrial predators By integrating estimates

of secondary production with measures of Cd concentration in emerging insects, Currie et al (1997)calculated that 1.3–3.9 g Cd was exported annually by aquatic insects (dipterans, dragonflies, andmayflies) from Cd-treated Lake 382 in the Experimental Lakes Area, Ontario Fairchild et al (1992)estimated that as much as 2% of 2,3,7,8-tetrachlorodibenzofuran (TCDF) in sediments are exportedannually by emerging insects, posing a significant risk to terrestrial predators (primarily birds andbats) Froese et al (1998) measured transport of PCBs from emerging aquatic insects to tree swallows

in Saginaw Bay, Michigan Relative concentrations of PCB congeners were markedly differentbetween sediments, benthic invertebrates, and swallows, possibly reflecting metabolic differencesamong trophic levels This relationship between contaminant concentrations in sediments and levels

in terrestrial predators is often complex and will be influenced by trophic relationships and lifehistory characteristics of emerging aquatic insects Maul et al (2006) reported that biomagnification

of PCBs in nestling tree swallows was dependent on feeding habits of adults birds, which were quitevariable These researchers cautioned that risk assessments based exclusively on a single component(e.g., contaminant concentrations in emerging insects) that do not consider life history characteristics

of prey species and feeding habits of predators may be biased Muir et al (1988) measured PCBs

and other organochlorines in a marine food chain consisting of arctic cod (Boreogadus saida), ringed seals (Phoca hispida), and polar bears (Ursus maritimus) In addition to increased concentrations

with trophic level, major differences in the constituents of PCBs and chlordane-related compoundswere observed among species Elevated levels of organochlorines in bald eagles collected from LakeSuperior were attributed to consumption of highly contaminated gulls (Kozie and Anderson 1991),which feed predominately on fish Finally, food chain transport and biomagnification of PCBs have

likely contributed to the decline of otter (Lutra lutra) populations in western Europe (Leonards et al.

1997) In addition to significant biomagnification of PCBs, results of multivariate analyses showedchanges in the distribution of PCB congeners among trophic levels and enrichment of the most toxicconstituents in otters (Figure 34.5)

34.3.6 FOODCHAINTRANSFER OFCONTAMINANTS FROM

SEDIMENTS

Because sediments are an important sink for contaminants in aquatic ecosystems, models of inant transport should include a sediment compartment Concentrations of contaminants in sedimentsare often several orders of magnitude greater than in overlying water, and benthic organisms asso-ciated with sediments may influence the transport of these contaminants In addition to their role infood chain transport of contaminants to higher trophic levels, the activities and movements of benthicorganisms may indirectly affect bioconcentration and bioaccumulation For example, Reynoldson(1987) reported that 0.2–7.4 g/m2/year PCBs are ingested by oligochaete worms in contaminatedsections of the Detroit River Similarly, Evans et al (1991) estimated that 30% of the PCBs depos-ited annually in Lake Michigan sediments are recycled by amphipods Vertical migration of the

contam-invertebrate planktivore, Mysis relicta, transports sediment contaminants back to the water column

where they are available to higher trophic levels (Bentzen et al 1996) Bioturbation, defined as the

Congener type

Congener type

Congener type

Congener type 0.4

FIGURE 34.5 Results of PCA showing the relationship between trophic level and patterns of PCB constituents

in an aquatic food web (Modified from Figure 4 in Leonards et al (1997).)

Day 18

Day 2

0 10 20 30 40 50 60

FIGURE 34.6 The influence of bioturbation by benthic invertebrates on concentration of fluoranthene in

filter-feeding mussels The figure shows results after 2- and 18-day exposure (Modified from Figure 4 in Ciarelli

copepods was transferred to fish Interestingly, PCB levels in predators foraging on clean prey incontaminated sediments were five times greater than those in fish feeding on contaminated prey inclean sediments These results suggest that incidental ingestion of sediments is a significant route ofexposure in benthic-feeding fish

Comparative studies of food webs in different ecosystems provide an opportunity to evaluatethe relative importance of sediment and aqueous exposure to contaminants Morrison et al (2002)

compared transport and fate of PCBs in the eastern and western basins of Lake Erie, areas that differ inimportant limnological and geomorphological characteristics related to sediment–water interactions.Compared to the deeper eastern basin, the western basin of Lake Erie is relatively shallow, highlyproductive, and subjected to high winds that result in sediment resuspension Concentrations of PCBs

in organisms were much higher in the western basin, and PCBs in water contributed significantly

to these body burdens compared to organisms from the eastern basin These differences also haveimportant implications for the responses of organisms to hypothetical decreases in PCBs in waterand sediment In the eastern basin, fugacity of PCBs in sediment was much greater than fugacity inwater, indicating that organisms accumulate most of their PCBs from sediment Thus, remediationefforts to reduce PCB levels in sediments would likely be successful In contrast, because organismsfrom western Lake Erie receive significant amounts of PCBs from water, remediation efforts shouldfocus on reducing levels of dissolved PCBs

34.3.7 BIOLOGICALPUMPS ANDCONTAMINANTTRANSFER IN

ECOSYSTEMS

Persistent organic pollutants such as PCBs, HCB, and dichlorodiphenyltrichloroethane (DDT), aswell as Hg are widely distributed by the atmosphere and oceans The global distribution of thesecontaminants is indicated by their elevated levels in food webs of remote arctic and subarctic eco-systems Transport of persistent pollutants in remote marine ecosystems is facilitated by migratorysalmon that accumulate contaminants from the ocean and deliver them to their native lakes whenthey return to spawn These migrating organisms act as biological pumps, delivering contaminantsupstream where they may accumulate in aquatic food webs Krummel et al (2003) observed a highly

significant relationship (r2≥ 9) between the density of spawning sockeye salmon (Oncorhynchus nerka) and PCB concentrations in lake sediments Concentrations of PCBs in lakes with spawning

salmon were approximately six times greater than in lakes without fish, and the pattern of PCBcongeners in lake sediments was very similar to those in fish Persistent pollutants that are pumpedupstream may be accumulated in arctic food webs of receiving systems Ewald et al (1998) reported

elevated levels of PCBs and DDT in arctic grayling (Thymallus arcticus) collected from lakes with

returning migratory salmon Similar transport of marine-derived contaminants has been reported

in arctic seabirds Blais et al (2005) collected sediments from ponds at the base of cliffs along

a gradient of petrel (Fulmarus glacialis) use in the Canadian Arctic Concentrations of Hg, DDT,

and HCB were 10–60 times greater in sediments collected from ponds with high petrel use as a result

of inputs from guano This research indicates that in some instances biological transport can have

a much greater influence on levels of organic contaminants in arctic and subarctic ecosystems thanatmospheric deposition

34.4 ECOLOGICAL INFLUENCES ON FOOD CHAIN

TRANSPORT OF CONTAMINANTS

Most studies that describe uptake and food chain transport of contaminants usually do not focus onthe ecology of these systems, but simply report tissue concentrations in biotic and abiotic compart-ments More recently, researchers have recognized that ecological characteristics of communitiesinfluence contaminant transfer and the concentrations in upper trophic levels Because food webinteractions strongly influence energy flow and biogeochemical cycling, understanding the relativeimportance of consumer versus resource control is important for predicting chemical transport Forexample, the concentration of lipophilic contaminants in top predators will be influenced by foodweb interactions and the relative strength of top-down versus bottom-up controls The development

of new techniques to quantify feeding preferences, such as stable isotope analyses, allows igators to better characterize relationships between trophic level and contaminant concentrations

invest-In addition, the larger spatial scale of many contemporary food web studies provides an opportunity

to investigate how landscape features influence food chain transport of chemicals Quantifying therelative importance of ecological factors on contaminant transport is greatly improved by makingcomparisons across communities For example, studying contaminant levels in systems that lackpoint source discharges allows investigators to isolate the relative importance of ecological andhabitat features The best examples of this research have been conducted in remote systems whereatmospheric deposition is the primary source of contamination (Berglund et al 1997, Kidd et al

1995, Kidd et al 1998, Larsson et al 1992, Rasmussen et al 1990) Better integration of ecologicaland landscape concepts into kinetic and bioenergetic models will allow for a more comprehensiveunderstanding of contaminant transport in communities

34.4.1 FOODCHAINLENGTH ANDCOMPLEXITY

Understanding the relative importance of ecological factors such as food chain length, primary andsecondary productivity, and linkage strength will help explain the large amount of variability incontaminant concentrations often observed in predators collected from different ecosystems Theearly work by Rasmussen et al (1990) stimulated a significant amount of interest in the relationshipbetween food web structure and contaminant transport These investigators classified lakes into

three types based on the presence of invertebrate planktivores (Mysis) and pelagic forage fish Trout

collected from lakes with long food chains (i.e., more trophic levels) generally had higher PCBlevels than fish from lakes with simple food chains (Figure 34.7) Similar results were reported byKidd et al (1995) in which elevated levels of toxaphene in fish collected from a subarctic lake wereattributed to an “exceptionally long” food chain

The influence of food chain length on contaminant levels in top predators may have importantimplications for systems where food webs are altered by exotic species Introduced species thatlengthen food chains may increase levels of persistent chemicals in top predators (Cabana et al 1994,Cabana and Rasmussen 1994, Kidd et al 1995), especially if these species link contaminated benthichabitats to pelagic consumers However, results of studies attempting to demonstrate enhanced food

chain transport in ecosystems where exotic species have invaded are mixed Rainbow smelt (Osmerus

mordax) have recently invaded many freshwater ecosystems of North America Because rainbow

0 200 400 600 800

1 000

1 200

0 5 10 15 20

FIGURE 34.7 Influence of trophic structure on concentrations of PCBs in lake trout from central Ontario

lakes Data are shown as total PCBs (solid bars) and after correcting for lipid content (open bars) Class 1 lakes

with short food chains lack Mysis and pelagic forage fish Class 2 lakes with intermediate length food chains lack Mysis but have pelagic forage fish Class 3 lakes with long food chains have both Mysis and pelagic forage

fish (Data from Table 1 in Rasmussen et al (1990).)