Biomarkers and integrated environmental risk assessment: Are there more questions than answers?



The introduction of the European Commission's Water Framework Directive (WFD; 2000/60/EC) established a new era in environmental risk assessment. In addition to incorporating the compliance of chemical quality standards, the key objective of the WFD is the general protection of the aquatic environment in its entirety. This new approach emphasizes the need for an integrated environmental risk assessment and offers the potential for the incorporation of biological effects measures, including the use of biomarkers in this process. Biomarkers have been suggested as practical tools for environmental management for a number of decades, but their inclusion has not been universally accepted because of a number of unanswered questions regarding sensitivity, practicality, and reproducibility. With this in mind, this paper addresses these potential questions and shows how, by taking a weight-of-evidence approach, biomarkers may be successfully incorporated within environmental risk assessment frameworks such as the WFD.


The use of biomarkers (functional measures of exposure to stressors expressed at the suborganismal, physiological, or behavioral level; McCarty and Munkittrick 1996) as surrogate measures of biological impact within laboratory and field studies has been prevalent for many years. However, the incorporation of biomarkers into regulatory legislation for environmental risk assessment (ERA) has generally been lacking. The focus has continued to be on chemical measurements as standard practice, made in the context of environmental quality standards (EQSs) for risk assessment of, for example, hazardous substances. The introduction of the Water Framework Directive (WFD) by the Commission (EC) of the European Union (EU; 2000/60/EC) has shifted the emphasis away from primarily monitoring chemicals to an approach that incorporates both chemical and ecological objectives, designed to protect and, where necessary, restore the structure and function of aquatic ecosystems themselves (Environment Agency 2002). Key features of the WFD are to ensure that bodies of water are maintained or restored to “good ecological and chemical status” as well as ensuring the protection of highly valuable aquatic ecosystems. To achieve and assess good status, the WFD advocates the integration of various disciplines, analyses, and expertise as 1 of their central concepts (Littlejohn et al. 2003). In terms of water management and ERA, it is clear that an integrated approach offers the potential for the incorporation of biomarkers within this framework, notably in the area of risk assessment validation for pressures, such as point-source and diffuse hazardous chemicals. Indeed, Annex V of the WFD highlights the need for biological elements as well as physicochemical and hydromorphological components for the determination of good ecological status for different water bodies. Thus, the potential inclusion of biomarkers into risk assessment may now be possible with this groundbreaking new approach to environmental management. Although previously recommended for use in environmental monitoring (Peakall 1992) and despite the interest of regulatory agencies (Burgeot et al. 1996; OSPAR 2000; Littlejohn et al. 2003; Long et al. 2004; Lehtonen 2005a), relatively few areas of environmental legislation have successfully implemented the use of biomarkers, and, as a result, there is little published work on their systematic use in environmental management. Because of the inclusion of biological elements within the EU WFD, bioassays may also have their place within the ERAs. Bioassays are tools used to assess effects of exposure to sediment or water samples and generally measure changes in mortality, growth, or reproduction. The use of bioassays for environmental assessment is well established in European programs such as the OSPAR Coordinated Environmental Monitoring Program and in the UK National Marine Monitoring Program. As with biomarkers, the inclusion of bioassays presents an opportunity for a more holistic (and therefore more meaningful) way of assessing effects of environmental samples and wastes on ecosystems than is possible using chemical-based monitoring alone. Both bioassays and biomarkers have their place in an integrated approach in ERA; however, this review will focus on the use of biomarkers in aquatic environments and previous attempts to incorporate them into risk assessment, focusing primarily on EU programs.


The Oslo and Paris (OSPAR) commissions are responsible for the regulation, monitoring, and assessment of pollution in the maritime area (Stagg 1998). Established in 1992, the OSPAR convention is obliged to prevent and eliminate pollution of the marine environment (from dumping, land-based, and offshore sources) and to conduct assessments of the quality of the marine environment (Stagg 1998). The Joint Assessment and Monitoring Program (JAMP), adopted by the OSPAR commission in 1995, aims to assess the concentrations, trends, and effects of substances in and inputs to the marine environment. The measurement of biological effects of contaminants is a key component within JAMP (OSPAR/ICES 1995) and has also been identified as a topic of common interest by other European organizations, including the International Council for the Exploration of the Sea (ICES), the European Environment Agency, the Helsinki Commission, the Artic Monitoring and Assessment Program, and the United Nations Environment Program Mediterranean Plan (UNEP-MED). The notable use of biomarkers within JAMP illustrates an important aspect that has hampered the implementation of biomarkers within international programs; that is the need for a comprehensive quality assurance program to ensure comparability of data (Stagg 1998). Several programs have been initiated to develop standard operating procedures and to undertake interlaboratory comparisons in response to this shortfall. For example, the Biological Effects Quality Assurance in Monitoring Programs (BEQUALM), an EU-funded project, was set up to develop quality assurance and control procedures for marine biological effects measures in order that laboratories contributing to international marine monitoring programs attained defined quality standards. The biomarkers recommended by OSPAR and ICES are listed in Table 1 along with their respective quality assurance measures. It is clear that the use of biomarkers is expanding within legislation and that through programs such as BEQUALM, a quality assurance regime is under way. However, this progress is slow, and many concerns are still limiting or preventing biomarkers from being incorporated into frameworks of ERA. Thus, before we can fully address why biomarkers have failed to be used extensively in risk assessment, we must 1st analyze what drives the selection of biomarkers that will be most practical and, more importantly, provide us with the most relevant information.


Biomarkers developed in response to the need for more sensitive indicators of sublethal ecological effects (Bickham et al. 2000). One definition of the term “biomarker” is “biochemical, cellular, physiological or behavioural variations that can be measured in tissue or body fluid samples, or at the level of whole organisms, to provide evidence of exposure and/or effects from one or more contaminants” (Depledge 1994).

Many molecular, cellular, biochemical, and physiological processes occur within an organism, and, potentially, any alterations in any of these processes may be used as biomarkers. However, biomarkers, in general, may be classified into 3 types: Those of exposure, those of effect, and those of susceptibility (Chambers et al. 2002). Biomarkers of exposure indicate that an organism has experienced exposure to a toxicant or other stressor; however, the change in the biomarker is not necessarily related directly to the toxicant's specific mechanism of action and may not be predictive of the degree of adverse effects either on the organism or at a population level (Chambers et al. 2002). Biomarkers of exposure can provide quantitative and qualitative estimates of exposure to various compounds and may potentially be used to replace expensive chemical analyses or to measure effects of short-lived chemicals. Biomarkers of effect are associated specifically with the toxicant's mechanism of action and are sufficiently well characterized to relate the degree of biomarker modification to the degree of adverse effects (Chambers et al. 2002). Thus, in terms of ERA, biomarkers of effect can shed light on qualitative aspects of hazard identification by both demonstrating that a hazard is occurring and elucidating probable mechanisms of action. These biomarkers can provide insights into both the causal factors of the hazard and its ecological consequences, depending on the degree of specificity of the biomarker with the pressure and the degree to which its expression to higher-order effects is understood (e.g., imposex in gastropods as a biomarker of organotin exposure, reproductive impairment, and population-level consequences). In contrast to biomarkers of effect or exposure, biomarkers of susceptibility do not represent stages along the dose-effect continuum but reflect an increase in the rate of transition between steps along that continuum (Schlenk 1999). Therefore, biomarkers of susceptibility can be used in ERA to provide a characterization of variability that can be used in defining uncertainty factors (Schlenk 1999).

Although all biomarker types can provide useful information in terms of knowledge of exposure or effects of toxicants, not all biomarkers, whether they are indicators of effect, exposure, or susceptibility, are suitable for use in ERA. In the past, certain criteria have been developed to address the selection of the most useful/relevant biomarkers to use (Mayer et al. 1992). These include whether a biomarker is easy to measure; whether it responds in a dose- or time-dependent manner to the toxicant, including its transience and biochemical memory (how long after exposure the response lasts); and whether it is sensitive. Additional criteria include that the variability in biomarker response due to natural variation should be understood (i.e., season, temperature, sex, weight, and handling). Finally, for ERA, a biomarker should have biological significance; only biomarkers that can be linked to important biological processes and for which changes can be interpreted should be used (Mayer et al. 1992). The relevant question therefore is, Do these criteria still define the use of biomarkers for use in the field of ERA, or are they unnecessary limitations? In the following section, we address the questions applied to biomarker inclusion in ERA.

Should we use only biomarkers that are easy to measure?

One suggestion is that only biomarkers that are easy to measure should be included in ERAs to allow quantification of larger numbers of multiple individuals. This approach has been successfully utilized in the Rapid Assessment of Marine Pollution (RAMP) program. RAMP is an ongoing pilot project of the UN Coastal Oceans Observing System initiated in the late 1990s to develop suites of rapid, robust biomarkers for the detection of exposure to specific classes of chemicals, sublethal biological damage, and organismal health in marine biota in areas of the world where only basic laboratory equipment was present and where the assays could be performed by personnel with nonspecialist laboratory skills. This rapid assessment approach has undergone preliminary testing in Brazil, Costa Rica, and Vietnam as part of the UNEP-funded Global Investigation of Marine Pollution in the Marine Environment program, where it has shown considerable promise (Depledge 2000). In addition to practical use in developing countries, the RAMP techniques have also been used in detailed studies performed in the United Kingdom, the United States, and Norway to determine the extent of environmental degradation in contaminated estuaries (Galloway et al. 2002; Dissanayake and Galloway 2004; Galloway et al. 2004a, 2004b; Watson, Andersen, Depledge, et al. 2004; Watson, Andersen, Galloway, et al. 2004). The results from these studies have illustrated how sensitive, rapid assessments can be of value as a means of identifying the toxic impacts of pollutants on biota in situ. However, it is important to emphasize that the aim of these rapid assessment designs was not to be definitive but rather to contribute to a hierarchal process of risk assessment. Results from these easy-to-use biomarkers determine if more detailed studies are required or may be used to assess the progress of any remedial action. Clearly, easy-to-use biomarkers have a place in ERA but may not totally replace biomarkers that require more sophisticated expertise, that are more expensive, and/or that are more time consuming (Huggett et al. 1992). For example, many techniques, such as P32 postlabeling (used to measure DNA adducts), intersex and liver nodules, are expensive, time consuming, and not easy to measure and require a lot of expertise; however, these techniques play an important role in answering specific questions on exposure to particular chemicals and provide a greater mechanistic understanding of cellular and molecular responses. Therefore, it is a general recommendation that biomarkers should be chosen not purely on their ease of use but rather to address specific questions. Despite this, in some parts of the world, ease of use will be 1 of the most important aspects of biomarker selection.

Table Table 1a.. Techniques recommended in fish for biological effects monitoring programs at the national or international level. Part of the continuing International Council for the Exploration of the Sea (ICES) work to review and evaluate marine biological effects monitoring techniques in relation to programs in member countries and the OSPAR Coordinated Environmental Monitoring Program. Adapted from ICES (2004)
MethodOrganismQAaPressure(s) addressedBiological significanceReferences
  1. a QA = quality assurance; B = BEQUALM; Q = QUASIMEME; O = OTHER (EU projects BEEP, COMPREHEND, UNEP, MEDPOL).

Bulky DNA adduct formationFishBPolycyclic aromatic hydrocarbons (PAHs), other synthetic organics (e.g., nitro-organics, amino triazine pesticides [triazines])Measures genotoxic effects; possible predictor of pathology through mechanistic links; sensitive indicator of past and present exposurePfau (1997), Bodin et al. (2004)
Acetylcholinesterase (AChE) inhibitionFishOOrganophosphates and carbamates (e.g., pesticides) or similar molecules, possibly algar toxinsMeasures exposureStein et al. (1992), Adams et al. (1992), Cajaraville et al. (2000), Kirby et al. (2000), Aarab, Champeau, et al. (2004)
Metallothionein inductionFishBMeasures induction of metallothionein protein by certain metals (e.g., Zn, Cu, Cd, Hg)Measures exposure to and disturbance of copper and zinc metabolismKillie et al. (1992), Hylland (1999)
EROD or P4501A inductionFishBMeasures induction of enzymes that metabolize planar organic contaminants (e.g., PAHs, planar PCBs, dioxins)Possible predictor of pathology through mechanistic links; sensitive indicator of past and present exposureMoore (1992), Stagg and Mcintosh (1998), Viarengo et al. (2000), George et al. (2004)
ALA-D inhibitionFishBLeadIndex of exposureHylland (2004), Schmitt et al. (2005)
PAH bile metabolitesFishQPAHsMeasures exposure to and metabolism of PAHsGeorge et al. (2004), Hoeger et al. (2004), Page et al. (2004)
Lysosomal stabilityFish Not contaminant specific but responds to a wide variety of xenobiotic contaminants and metalsMeasures cellular damage and is a good predictor of pathology; provides a link between exposure and pathological end points; possibly a tool for immunosuppression studies in white blood cellsKohler (1991), Diamant et al. (1999), Kohler et al. (2002)
Early toxicopathic lesions, preneoplastic and neoplastic liver lesions by histopathologyFishBPAHs, other synthetic organics (e.g., nitro-organics, amino triazine pesticides [triazines])Diagnosis of pathological changes associated with exposure to genotoxic and nongenotoxic carcinogensKohler (1990), Kohler and Van Noorden (1998)
Enzyme-altered liver foci by NADPH-producing enzymesFishBPAHs, other synthetic organics (e.g., nitro-organics, amino triazine pesticides [triazines])Diagnosis of enzymatic markers of carcinogenesis associated with exposure to genotoxic and nongenotoxic carcinogensKohler and Van Noorden (1998)
Liver nodulesFishBPAHs and other carcinogenic compoundsDiagnostic of pathological changes associated with exposure to genotoxic and nongenotoxic carcinogensVethaak (1992), Bogovski et al. (1999)
Externally visible lesions and parasitesFishBResponds to a wide variety of environmental contaminants and nonspecific pressuresIntegrative response; measures general fish health; elevated prevalence may indicate exposure to contaminantsVethaak et al. (1992), Bogovsk et al. (1999)
Vitellogenin inductionMale and juvenile fish Estrogenic substancesIndicates exposure and endocrine disruptionJobling et al. (1998), Tyler and Routledge (1998), Kirby et al (2004), Vethaak et al. (2005)
IntersexMale flounder Endocrine-disrupting substancesMeasures feminization of male fish and reproductive impairmentJobling et al. (1998), Tyler and Routledge (1998), Kirby et al (2004), Lyons et al. (2004)
Reproductive success in Zoarces viviparusZoarces viviparusBNot contaminant specific but will respond to a wide range of environmental contaminantsMeasures reproductive output and survival of eggs and fry in relation to contaminants; restricted to period when young are carried by female viviparous fishJacobsson and Neuman (1991), Mattsson et al. (2001), Strand et al. (2004)

Table 1b

Table Table 1b.. Recommended techniques, in invertebrate species, for biological effects monitoring programs at the national or international level. Part of the continuing International Council for the Exploration of the Sea (ICES) work to review and evaluate marine biological effects monitoring techniques in relation to programs in member countries and the OSPAR Coordinated Environmental Monitoring Program. Adapted from ICES (2004)
MethodOrganismQAaIssues addressedBiological significanceReferences
  1. a QA = quality assurance; B = BEQUALMS; Q = QUASI ME ME; O = OTHER (EU projects BEEP; COMPREHEND; UNEP; MEDPOL).

Acetylcholinesterase inhibitionMollusks and crustaceansOOrganophosphates and carbamates or similar molecules, possibly algal toxinsMeasures exposureRadenac et al. (1998), Narbonne et al. (1999), Cajaraville et al. (2000), Bonacci et al. (2004), O'Neill et al. (2004), Rickwood and Galloway (2004)
Metallothionein inductionMytilus spp.OMeasures induction of metallothionein protein by certain metals (e.g., Zn, Cu, Cd, Hg)Measures exposure and disturbance of copper and zinc metabolismLeung and Furness (1999) Cajaraville et al. (2000), Porte, Sole, et al. (2001), Geffard et al. (2002), Chevre et al. (2003), Galloway et al. (2004b)
Lysosomal stability (including neutral red retention)Mytilus spp. OysterO/BNot contaminant specific but responds to a wide variety of xenobiotic contaminants and metalsMeasures cellular damage and is a good predictor of pathology; provides a link between exposure and pathological end points; possibly a tool for immunosuppression studies in white blood cellsColes et al. (1995), Pipe et al. (1999), Cajaraville et al. (2000), Wedderburn et al. (2000), Brown et al. (2004), Ringwood et al. (2004)
Scope for growthBivalve mollusks (e.g., Mytilus spp. and oysters)QResponds to a wide variety of contaminantsIntegrative response; a sensitive sublethal measure of energy available for growthWiddows et al. (2002), Toro et al. (2003), Olsson et al. (2004), Halldorsson et al. (2005)
ImposexNeogastropod mollusksQSpecific to organotinsReproductive interference; estuarine and coastal littoral waters and offshore watersGibbs et al. (1991), Matthiessen and Gibbs (1998), Oberdorster et al. (1998), Alzieu (2000), Cajaraville et al. (2000)
IntersexLittorina LittoreaBSpecific to reproductive effects of organotinsReproductive interference in coastal (littoral) watersBauer et al. (1997), Oehlmann et al. (1998), Davies et al. (1999), De Wolf et al. (2001), Galloway et al. (2004b)
Induction/inhibition of multidrug/multixenobiotic resistanceMytilus edulis Multiple contaminants (organics and metals)Adaptation/inhibition in response to xenobiotic stressKurelec (1995), Eufemia and Epel (2000), Minier et al. (2002), Pain and Parant (2003), Bodin et al. (2004)
HistopathologyBlue mussels Not contaminant specificGeneral responsesCalabrese et al. (1984), Sunila (1987), Peters et al. (1994), Wedderburn et al. (2000), Lajtner et al. (2003), Aarab, Minier, et al. (2004)
Embryo aberrations in field-collected amphipod crustaceansAmphipods Contaminant specificMeasures frequency of different types of lethal embryo aberrations; allows for separating effects of contaminants and environmental climate variablesSundelin and Eriksson (1998)

Should we use only biomarkers that detect dose- and time-dependent responses?

Ideally, the degree to which the magnitude of the biological response relates to the dose of the exposure will be known, enabling severity of the exposure to be assessed directly (Long et al. 2004); however, the relationship is often complex (Walker et al. 1996). In general, the extent to which a molecular interaction occurs is generally related to the dose of chemical received. Nevertheless, exposure to low doses may produce no effects because of the presence of a threshold level of effect. In other cases where the threshold level is exceeded, protective mechanisms may mask the effects, such as the induction of metallothionein or stress proteins. In addition to dose responses, it is also suggested that the temporal relationships of biomarker responses need to be known. In general, biomarker responses indicative of sublethal effects should be detectable sufficiently in advance of pathological events that lead to death but not so far in advance as to decouple the early response from the late effect. Unless these exposure/adverse effect time relationships are understood, the biomarker approach will have limited practical utility for recognizing or predicting ecological significant events (Depledge 1994).

Likewise, the temporal nature (or transience) of the biomarker response after exposure of the organism to the pressure should also be known. This issue was addressed recently, and biomarker induction, adaptation, and recovery times were collated for a range of suborganismal biomarkers (Wu et al. 2005). Interestingly, 2 main response types were apparent: “Fast induction-fast recovery,” characteristic of responses to metals, polycyclic aromatic hydrocarbons (PAHs), and organochlorines, and “fast induction-slow recovery,” typified by responses to endocrine disruptors, oils, and organophosphates. The review also highlighted the wide variation observed in induction and recovery times; for example, crustaceans typically responded more quickly than fish, while responses at the molecular and cellular levels were generally reversed more quickly than those at the whole-organism level (Wu et al. 2005). It follows from this kind of analysis that we should not always expect to find a simple relationship between the level of a biomarker response and the tissue residue concentration of a pollutant (Depledge and Galloway 2005). It is important, however, to note that although it is useful to understand the toxic responses of contaminants and how they alter with dose and over time, in terms of ERA it is harder to quantify these responses. Even biomarkers that respond to specific contaminants, such as metallothionein, have been shown to be influenced by other contaminants as well as natural fluctuations due to temperature and salinity changes. Furthermore, the aquatic environment typically contains a complex mixture of contaminants that often interact, rendering the knowledge of a typical single toxic response useless. This does not mean, however, that we cannot use biomarkers in ERA. What is needed when using biomarkers in field environments is a greater understanding of the variations in biomarker response due to natural abiotic and biotic factors (e.g., reproductive status and seasonality).

Table Table 1c.. Promising biological effects monitoring methods that require further research before they can be recommended for monitoring (both fish and invertebrates). Part of the continuing International Council for the Exploration of the Sea (ICES) work to review and evaluate marine biological effects monitoring techniques in relation to programs in member countries and the OSPAR Coordinated Environmental Monitoring Program. Adapted from ICES (2004)
MethodOrganismIssues addressedBiological significanceReferences
DNA strand breaks including Comet assayFish, mussels, cellsNot contaminant specific but will respond to a wide range of environmental contaminantsMeasures genotoxic effects but is also extremely sensitive to other environmental parametersNacci et al. (1996), Steinert (1996), Bolognesi et al. (2004), Hagger et al. (2005)
ApoptosisFish cellsResponds to a wide range of contaminantsGeneral responseBurkhardt-Holm et al. (1997), Lundebye et al. (1999), Frenzilli et al. (2004), Lyons et al. (2004)
Acetylcholinesterase inhibitionOther invertebratesOrganophosphates and carbamates or similar molecules, possibly algal toxinsMeasures exposureBocquene et al. (1993), Forget and Bocquene (1999)
Polycyclic aromatic hydrocarbon (PAH) urine metabolitesCrustaceansPAHsMeasures exposure to and metabolism of PAHsDissanayake and Galloway (2004), Watson, Andersen, De pledge, et al. (2004), Watson, Andersen, Galloway, et al. (2004)
OncogenesFishPAHs, other synthetic organics (e.g., nitro-organics, amino triazine pesticides [triazines])Activation of oncogenes (ras) or damage to tumor-suppressor genes (p53); measures genotoxic effects leading to carcinogenesisGoodwin and Grizzle (1994), Rotchell et al. (2001), Cronin et al. (2002)
P4501 A-like enzymesInvertebratesInduced enzymes response to PAHs, planar polychlorinated biphenyls, dioxins, and/or fu-ransMeasures exposure to organic contaminantsPeters et al. (1999), Cajaraville et al. (2000), Porte, Biosca, et al. (2001), Shaw et al. (2002)
Induction/inhibition of multidrug/multixenobiotic resistanceFish and invertebrates other than MytilusMultiple contaminants (organics and metals)Adaptation/inhibition in response to xenobiotic stressKohler et al. (1998), Schroder et al. (1998), van der Oost et al. (2003), Shuilleabhain et al. (2005)
Glutathion-S- transferaseFish, musselsPredominantly organic xeno-bioticsMeasures exposure and the capacity of the major group of phase II enzymes; considered most promising for isoenzyme-specific measurementsNarbonne et al. (1999), Devier et al. (2005), Moreira and Guilhermino (2005)
Oxidative stressFish Mytilus spp.Not contaminant specific but will respond to a wide range of environmental contaminantsMeasures the presence of free radicalsLivingstone et al. (1992), Moore (1992), Pipe et al. (1993), Camus et al. (2004), Regoli et al. (2004)
ImmunocompetenceFish, invertebratesNot contaminant specific but will respond to a wide range of environmental contaminantsMeasures factors that influenceColes et al. (1994), Dyrynda et al. (1998), Auffret et al. (2004), Parry and Pipe (2004)
Online monitoringMussels and crabsNot contaminant specific but will respond to a wide range of environmental contaminantsMeasures the effects of chemicals on heart rate using a simple and inexpensive remote biosensor; gives an integrated responseStyrishave and Depledge (1996), Bloxham et al. (1999), Galloway et al. (2002), Brown et al. (2004), Abessa et al. (2005)
Abnormalities in wild fish embryos and larvaeFish, including demersal and pelagic speciesNot linked unequivocally to contaminantsMeasures frequency of probably lethal abnormalities in fish larvae; mutagenic, teratogenicCameron et al. (1996), Klumpp et al. (2002), Strand et al. (2004), Hallare et al. (2005)
Bulky DNA adduct formationMussels, invertebratesPAHs, other synthetic organicsMeasures genotoxic effectsDunn et al. (1987), Liu et al. (1991), Pfau (1997), Akcha et al. (2000), Malmstrom et al. (2000), Bodin et al. (2004), Lyons et al. (2004)
Gene arraysFishVariousCombined responses from various biomarkersLam et al. (1998), Larkin et al. (2003), Cheung et al. (2004), Roling et al. (2004)
HistopathologyInvertebrates (other than Mytilus)Not contaminantGeneral responsesPeters et al. (1994), Lajtner et al. (2003), Binelli et al. (2004)
SpigginThree-spined sticklebackAndrogensMeasures environmental androgensKatsiadaki et al. (2002), Hahlbeck et al. (2004)
MicronucleiFish, bivalve mollusksNot contaminant specificGeneral responsesHeddle et al. (1983), Scarpato et al. (1990), Bahari et al. (1994), Burgeot et al. (1995), Sugg et al. (1996), Bolognesi et al. (1999), Pavlica et al. (2000), Perez-Cadahia et al. (2004), Hagger et al. (2005)
Peroxisomal proliferation (enzyme assays)Fish and invertebratesContaminant specificPotential alterations in lipid metabolism, nongenotoxic carcinogenesisCajaraville et al. (1997), Porte, Solé, et al. (2001), Orbea et al. (2002), Cajaraville et al. (2003)
Delayed reproduction/ gonadal maturationFishNot contaminantReproductive disruptionMoiseenko (2002), Cardinali et al. (2004), Weber and Banner-man (2004)

How sensitive should biomarker responses be?

Two aspects are embedded in this question: 1) How sensitive is the biomarker compared to classical endpoints, such as lethality, reproduction, or growth impairment? And 2) how sensitive is the biomarker compared to other candidate biomarkers? (Huggett et al. 1992). A highly sensitive biomarker could be useful in identifying exposure situations if changes in the biomarker can be measured at earlier time points or at lower exposure levels than traditional toxic endpoints (Chambers et al. 2002). The effects of contaminants at lower levels of biological organization (e.g., molecular effects) in general occur more rapidly than those at higher levels (e.g., ecological effects) and therefore may provide a more sensitive early warning of toxicological effects within populations (Clements 2000). In ERA frameworks, it is recommended that a suite of sensitive biomarkers be used that detect short-term responses as well as longer-term ecologically relevant end points to provide a weight-of-evidence approach for establishing relationships between environmental stressors and ecological effects (Adams et al. 2001; Galloway et al. 2004a).

How do biomarker responses alter with natural variation?

As previously highlighted, it is important when selecting biomarkers for ERA that the naturally occurring variation in biomarker response is clearly understood. This is 1 of the most difficult aspects when making ecological predictions based on biomarker analyses (Depledge 1994). Relatively few studies have taken into account abiotic- and biotic-driven differences in biomarker response within the population of indicator organisms. Table 2 shows that environmental factors such as temperature, salinity, turbidity, diet, and season may contribute to biomarker variation either by altering the pollutant bioavailability or resulting from natural biological changes such as growth and reproductive stage. In a study analyzing biomarker expression in mussels every month for 2 y, Bodin et al. (2004) showed seasonal variation in biomarkers (metallothionein [MT] and heat shock protein 70) with coefficients of variations in the range 16% to 89%. In addition to seasonal effects for these 2 biomarkers, chemical monitoring also detected a relationship between heavy metal concentration and MT and heat shock protein (Bodin et al. 2004). The utility of some biomarkers in detecting interregional differences may be very limited because of alterations of abiotic factors contributing to their response, such as salinity.

In addition to abiotic influences, it is important to address what, if any, variation in a biomarker response is due to the inherent differences in morphology and biochemical/physiological status of exposed organisms. Such biologically driven variation may include differences in biomarker responses between sexes, ages, sizes, or genotypes or at different stages of reproductive output or growth (Bodin et al. 2004). As well as biologically driven variation, variability between individuals may be due to the intrinsic susceptibility of that individual, which in turn may be influenced by inherited mutations in genes involved in the predisposition to specific diseases (Schlenk 1999). Factors such as previous exposure to an environmental agent (i.e., imprinting before development), the physiological state of the individual (i.e., starvation, disease state), and developmental or age-related processes that control pathways that impact genetic expression are some examples of environmental factors that may influence susceptibility (Schlenk 1999). Barrett et al. (1997) also highlighted that interindividual variation may occur in the uptake and effects of chemicals, associated with variation both within and between organisms in bioactivation and detoxification enzyme activity, including the activity of the cytochrome P450 enzyme system, cell membrane transporters such as P-glycoproteins (which may confer multixenobiotic resistance), and DNA repair enzymes.

To be confident that assessments of environmental health based on biomarker measurements represent a true measurement, it is vital that the range of inherent variability in the biomarker measurements in healthy organisms is known (Depledge 1994). Lack of knowledge of the normal physiological ranges of biomarkers has impeded their applied use, and, in many cases, this gap in understanding reflects a lack of understanding of the fundamental biochemistry of many aquatic organisms. In a medical context, the normal ranges of values quoted are ranges from a skewed distribution calculated to include 95% of the values obtained from a healthy population (Depledge 1994). Schlenk (1999) recommends the use of a “safety factor” that takes into account variability in the susceptibility of individuals, whereas Depledge (1994) suggests that the “normal range” of biomarker values is best described as the mean ±2 times the standard deviation. Both recommendations may not be necessary if a suite of biomarkers is used to examine the relationship between biomarkers, as any signal of abnormality is taken to reflect impact rather than focusing on absolute values.

Should we use only biomarkers that are linked to higher levels of biological relevance?

A key issue for the use of the biomarker approach in ERA is that of biological relevance, that is, that biomarkers manifested at lower levels of biological organization are linked to population, community, or ecosystem consequences and, ideally, that they provide an early warning before disturbances are realized at these higher levels (Adams 1990). One of the major criticisms of biomarkers has been that the link between molecular and cellular effects and impacts at or above the population level has not been demonstrated. However, an area where this has been successfully addressed is the effects of endocrine-disrupting chemicals. The induction of a pseudohermaphroditism in the gonochoristic marine gastropod Nucella lapillus remains the best characterized example of endocrine disruption in wildlife and the only example where an unequivocal causal association with a single chemical has been proven. Smith (1971) 1st named the superimposition of male genitalia in female gastropods as imposex, and subsequent research has undisputedly linked the condition in N. lapillus to exposure to tributyltin (Gibbs and Bryan 1986; Bryan and Gibbs 1991; Schulte-Oehlmann et al. 2000). In its most severe form, imposex (measured as Vas Deferens Sequence Index) prevents the reproductive (egg laying) capacity of affected females, leading to sterility and thus the ecologically significant decrease in population (Bryan et al. 1986; Gibbs and Bryan 1986). Because of the unprecedented link between the presence of imposex and the decline in dogwhelk populations, the presence of imposex has been included as a biological effects technique in many national and international regulatory monitoring programs and is the only biomarker currently included in the WFD.

Table Table 2.. Studies where abiotic and biotic factors have been identified as influencing biomarkers
 Environmental variationBiological variationEither
BiomarkerSalinityTemperatureReproductive stageSexGrowth/sizeSeasonal variations
AcetylcholinesteraseMussels (Pfeifer et al. 2005) Oyster (larvae) (Damiens et al. 2004)Mussels (Bocquene et al. 1990; Pfeifer et al. 2005)  Mussels (Radenac et al. 1998)Musseis (Romeo et al. 2003; Leinio and Lehtonen 2005; Moreira and Guilhermino 2005; Pfeifer et al. 2005) Sea urchins (Paracentrotus lividus) (Cunha et al. 2005) Macoma balthica (Leinio and Lehtonen 2005)
MetallothioneinCrabs (Legras et al. 2000; Mouneyrac et al. 2001) Clams (Martin-Diaz et al. 2005) Dogwhelks (Leung et al. 2002)Mussels (Serafim et al. 2002) Oligochaetes (Gillis et al. 2004) Dogwhelks (Leung et al. 2000)Clams (Hamza-Chaffai et al. 1999)Flounder (Hylland et al. 1998) Crabs (Mouneyrac et al. 2001) Oysters (Geffard et al. 2002)Clam (Bordin et al. 1997; Hamza-Chaffai et al. 1999) Crabs (Legras et al. 2000) Mussels (Serafim et al. 2002)Musseis (Romeo et al. 2003; Leinio and Lehtonen 2005; Moreira and Guilhermino 2005) Macoma balthica (Leinio and Lehtonen 2005)
Ethoxy reosrufin-O-deethylaseEuropean eel (Anguilla anguilla) (Livingstone et al. 2000)Dab (Lange et al. 1998)Flounder (Hylland et al. 1998) Carp (Sole et al. 2002)Fish (Kruner and von Westernhagen 1999) Flounder (Hylland et al. 1998) Carp (Solé et al. 2002) European eel (Anguilla anguilla) (Rotchell et al. 1999) Flounder (Pleuronectes flesus) (Rotchell et al. 1999)

Of the molecular and cellular biomarkers in current use, lysosomal stability is a good indicator of the degree of stress or disease and health status of the animal (Moore 2002) and has been shown to be linearly correlated with scope for growth, indicating that it is a good indicator of the overall health status of the organism (Widdows et al. 1981, 1982; Moore 1992, 2002). Although there is a continual search for this “holy grail” linkage of responses, it is important to realize that the lack of clear-cut linkages does not completely invalidate the application of biomarkers (Walker et al. 1996). For example, exposure biomarkers may have the potential to offer an alternative to some chemical analyses or to measure effects of short-lived chemicals as well as giving a more biologically relevant indication of exposure (Walker et al. 1996). The detection of PAH metabolites in fish and crustaceans using fluorimetric analysis of urine and bile has been successfully used for the rapid and easy detection of exposure to PAHs. Using this approach, Watson, Andersen, Depledge, et al. (2004) detected a gradient of PAH exposure in the urine of crabs along a Norwegian coastline, and Stein et al. (1992) used PAH bile metabolites to detect contaminant exposure in benthic fish from Puget Sound, USA. In addition, Kirby et al. (2000) demonstrated a strong gradient of cholinesterase enzyme activity in flounder collected along the Tyne estuary in the United Kingdom, suggesting a source of neurotoxic contamination upstream. Thus, there is a potential use for exposure biomarkers as a cost-effective primary screening tool in order to determine if more detailed chemical analyses are necessary.

From the previous discussion, it is clear that our knowledge of biomarkers has improved vastly over the past several decades and that their usefulness in field studies has been demonstrated. However, what is also clear is that, while biomarkers have been helpful in addressing some environmental risk assessment and risk management issues, the applied use of a given biomarker in a given candidate species often requires the answering of many questions. This leads us to question how biomarkers may be incorporated within environment frameworks for risk assessment, minimizing problems associated with their interpretation while emphasizing their usefulness.


One biomarker or a suite?

No single biomarker can unequivocally measure environmental degradation. The ability to differentiate between clean and polluted sites would be at best incomplete using a single-biomarker approach (Galloway et al. 2004a). A suite of end points, at different levels of biological organization, allows for a better evaluation of the hazard (Schlenk 1996). Sole (2000) found when determining the effects of environments chronically exposed by organic pollution that no single biomarker could be adopted as a surrogate indicator but that a battery of biomarkers coupled with chemical analysis provided the most comprehensive indicator of ecosystem health. Furthermore, by using a suite of biomarkers, a weight-of-evidence approach can be adopted that will minimize the influence of natural variation and allow the discrimination of clean healthy and polluted/unhealthy sites. Indices of environmental quality are often used by environmental managers as a practical approach to classifying sampled sites on a scale such as “clean to highly polluted” or “good to poor.” Usually, 4 or 5 levels of chemical or ecological classifications are used to minimize errors. Various attempts have been made to try and develop a biomarker-based index for minimizing random errors and variations so that biomarkers can be used in ERA. Bodin et al. (2004) used an integrated biomarker response index to compare the physiological state of native and transplanted mussels from 2 polluted coastal areas in southern France. The integrated response was carried out on data taken every month over 2 y for 6 different biomarkers (acetylcholinesterase, DNA adducts, benzo[a]pyrene hydroxylase, heat-shock proteins, MT, and P-glycoprotein-mediated multixenobiotic resistance). Aarab, Champeau, et al. (2004) used a scoring-based approach with biomarkers from fish collected from 11 sites in rivers in southwest France to complement freshwater monitoring programs carried out by the Water Agency Adour-Garonne. The biomarkers (acetylcholinesterase in brain and gills and glutathione S-transferase activities, catalase, and ethoxyresorufin-O-deethylase activity in liver samples) were selected to reflect early molecular mechanisms of action of contaminants, that is, phases I and II of drug metabolism, oxidative stress, and neurotoxicity (Aarab, Champeau, et al. 2004). Different sites were separated by discriminant analyses, and all the biomarkers were gathered in a global index called the multibiomarker pollution index (previously applied in the European BIOMAR program) (Narbonne et al. 1999) to give a summary of the pollution level undergone by the animals (Aarab, Champeau, et al. 2004). From the examples mentioned previously, it is clear that by using a biomarker-based index, random errors and variations can be minimized.

One species or multiple species?

From the previous discussion it is apparent that maximizing discriminatory potential requires the use of a suite of biomarkers. Given different routes of exposure and interspecies variation in effects/susceptibility, measurement from a biomarker suite should be measured ideally in a wide range of species at a location.

Figure Figure 1..

Description of hierarchal approach to risk assessment. Adapted from Stagg and Mcintosh (1998), Environment Agency (2002), Galloway et al. (2004a), and Lehtonen (2005b).

Historically, species for biomarker studies have been chosen primarily for their ease of culture and sampling rather than for their ecological relevance (Galloway et al. 2004a). However, in a recent study conducted in Southampton Water, United Kingdom (Galloway et al. 2004b), ecological relevance drove the selection of key species to determine the risk to the ecosystem. They hypothesised that by selecting diverse phyla exhibiting different feeding strategies (filter feeding, grazing, omnivory, predation) and measuring suites of biomarkers at the suborganismal and physiological levels, the ecological relevance of pollutant exposures may be more readily determined and integrated into environmental management strategies (Galloway et al. 2004a). This weight-of-evidence approach provides a more holistic attitude to environment management. Figure 1 summarizes this hierarchal approach to risk assessment.

Identifying specific areas of legislation where biomarkers can add value within ERA

One question that needs to be addressed is when to monitor for biological effects, notably in the context of key legislative drivers, identifying specific areas where the use of biomarkers can add significant value. For example, under Annex V of the WFD, 3 types of monitoring are recommended (Figure 2). These include surveillance monitoring to validate the risk assessments, to detect long-term trends, to assess impacts, and to influence the design of the operational monitoring strategy; operational monitoring to classify (in terms of status) those water bodies that have been identified as being at risk of failing good status and to ascertain any changes in status resulting from programs of measures; and investigative monitoring to ascertain the cause and effects of a failure of status when this is not clear and to ascertain the magnitude and impacts of accidental pollution. These 3 monitoring regimes highlight the usefulness of routine monitoring to check for long-term trends as well as any sudden alterations, an emphasis that has been lacking in most biomarker studies to date.

Figure Figure 2..

Description of monitoring programs for the Water Framework Directive. Adapted from Environment Agency (2002).

Table Table 3.. An example list of biomarkers that may be potentially useful in a tiered approach for the integration monitoring of hazardous substances. As suggested in Table 1C, some of these biomarkers need further research before they can be recommended by governing bodies
Lysosomal stabilityEarly warning indication of exposure to contaminants
Immunotoxicity: macrophage activityGeneral stress biomarkers (related to exposure to diverse contaminants)
Oxidative stress: Catalase Liver detoxification enzymes: glutathion S-transferase Cardiac output
External visible diseases General histopathologyBiological effects of contaminants on a longer time scale
Polycyclic aromatic hydrocarbons: ethoxyresorufin-O-deethylase, DNA adducts, bile metabolites Endocrine disruption: vitellogenin Lead: Ala-DMetals: metallothionein Genotoxicity: micronucleus Pesticides: acetylcholinesterase inhibitionSpecific biomarkers to determine causality and direct more detailed chemical analyses

In the context of hazardous substances and the WFD, surveillance monitoring offers an opportunity for the use of biomarkers as risk assessment tools, notably within the pressures and impacts phase of river basin characterization. Indeed, imposex Vas Deferens Sequence Index is used currently within the WFD in the United Kingdom as a measure of impact, associated with diffuse pesticide pressures in transitional and coastal waters (in the case of tributyltin). Exceedance of the EQS for tributyltin and Vas Deferens Sequence Index are used together in an integrated approach to characterize risk for this pressure, and, in this sense, the use of the biomarker provides a measure of validation, reducing uncertainty in the risk assessment process. However, for other important pressures, such as metals, while biomarkers exist that could be used in a similar way, they have not as yet been incorporated into the risk characterization process. In this regard, despite the emphasis of the WFD on the use of biological effects, the opportunity to reduce uncertainty in the risk assessment process for this and other pressures has not been fully realized. Ideally, a suite of both pressure-specific and general health biomarkers could be measured, including markers that provide an early warning of deleterious effects and that are capable of detecting current and long-term impact associated with key pressures. This requires the consideration of biomarkers that are sensitive to a wide range of contaminants and that are cost effective enough to be used in combination on a regular basis (Stagg 1998). Table 3 lists examples of biomarkers that may be potentially useful in a tiered approach for the integrated monitoring of hazardous substances.

Operational monitoring will be carried out on water bodies that are identified as being at risk of failing good status to confirm that status is indeed compromised, informed by surveillance monitoring. In general, monitoring techniques required for operational monitoring are at the community level of biological organization. This emphasizes the importance of an integrated approach since, at this stage, multiple lines of evidence (including a knowledge of pressures and impacts at various levels of biological organization) can provide evidence regarding causal risk relationships between the pressures and ecological status to underpin programs of measures (i.e., risk mitigatory action). The incorporation of an effective suite of biomarkers of exposure and effect could therefore provide insights into the causality of observed higher-level adverse effects (i.e., measures of ecological status), which community and ecosystem measures alone may not be able to provide since they integrate the effects of multiple pressures (Schlenk 1999). In terms of measuring environmental improvement, the continued assessment of ecosystem health using a suite of biomarkers can also help demonstrate impact reduction and progress toward better ecological status (Depledge and Fossi 1994). Thus, operational monitoring might also include the application of contaminant-specific biomarkers (Stagg 1998). Investigative monitoring could encompass a whole suite of general and specific biomarkers, depending on whether the nature of the investigation is specific or unidentified, notably in establishing causality of poor status.

Committing to an integrated ecosystem approach to ERA

The recent adoption by UK government departments and agencies of the principle of an ecosystem or ecosystem-based approach to managing the seas (DEFRA 2002; Laffoley et al. 2004) is an important step for integrated ERA. It is increasingly important with this approach that biomarkers are considered and implemented as part of a single framework of adaptive management that informs on the state of the ecosystem (Rogers and Greenaway 2005). The 2000 Quality Status Report (OSPAR 2000) recommends monitoring of both chemical concentrations and biological effects to strengthen the information regarding long-term ecological change. This integrated approach is important. As an example, although the numerous biological effects detected during the Baltic Sea component of the BEEP project are currently being used as a central source of information for the planning of future monitoring strategies within the Baltic Sea (Lehtonen 2005b), their use in terms of risk assessment is limited because of the lack of direct linkage to environmental contamination (Lehtonen 2005a). Thus, it is a commitment to the use of integrated assessments that can provide the best information in the context of decision making (Chapman 1993).


Under field conditions, organisms are exposed to a multiplicity of chemical and physical pressures against a background of naturally occurring seasonal fluctuations. Biomarkers have the potential to act as rapid integrative measures that indicate adverse conditions at a biologically relevant level (Huggett et al. 1992) and to provide a more proactive approach to risk assessment. The use of biomarkers has proven useful in establishing evidence of exposure to pollutant chemicals and damage to the health of sentinel organisms. In addition, biomarkers have helped establish causal relationships. However, a better understanding of the role of biomarkers within current legislative frameworks of integrated ERA is now needed. In ERA frameworks, it is recommended that a suite of sensitive biomarkers be used that detect short-term responses as well as longer-term ecologically relevant end points to provide a weight-of-evidence approach for establishing relationships between environmental stressors and ecological effects. However, we feel that biomarkers should be viewed as 1 part of the important “tool in the toolbox” for environmental management. It is vital that an integrated approach to ecosystem health be implemented, with biomarkers and bioassays adding value and providing complementary information to that provided by the chemical and ecological community measures currently undertaken.


Funding for the Ecosystems Management Bioindicators (ECOMAN: EVIDENCE) project is provided by DEFRA (ME3110) and the UK Environment Agency (SC050062). The authors would like to thank the International Council for the Exploration of the Sea (ICES) for their permission to reproduce Table 1 of this manuscript.

Disclaimer—The views expressed in this article are not necessarily those of the Environment Agency or DEFRA.