A new risk assessment approach for the prioritization of 500 classical and emerging organic microcontaminants as potential river basin specific pollutants under the European Water Framework Directive

https://doi.org/10.1016/j.scitotenv.2011.01.054Get rights and content

Abstract

Given the huge number of chemicals released into the environment and existing time and budget constraints, there is a need to prioritize chemicals for risk assessment and monitoring in the context of the European Union Water Framework Directive (EU WFD). This study is the first to assess the risk of 500 organic substances based on observations in the four European river basins of the Elbe, Scheldt, Danube and Llobregat. A decision tree is introduced that first classifies chemicals into six categories depending on the information available, which allows water managers to focus on the next steps (e.g. derivation of Environmental Quality Standards (EQS), improvement of analytical methods, etc.). The priority within each category is then evaluated based on two indicators, the Frequency of Exceedance and the Extent of Exceedance of Predicted No-Effect Concentrations (PNECs). These two indictors are based on maximum environmental concentrations (MEC), rather than the commonly used statistically based averages (Predicted Effect Concentration, PEC), and compared to the lowest acute-based (PNECacute) or chronic-based thresholds (PNECchronic). For 56% of the compounds, PNECs were available from existing risk assessments, and the majority of these PNECs were derived from chronic toxicity data or simulated ecosystem studies (mesocosm) with rather low assessment factors. The limitations of this concept for risk assessment purposes are discussed. For the remainder, provisional PNECs (P-PNECs) were established from read-across models for acute toxicity to the standard test organisms Daphnia magna, Pimephales promelas and Selenastrum capricornutum. On the one hand, the prioritization revealed that about three-quarter of the 44 substances with MEC/PNEC ratios above ten were pesticides. On the other hand, based on the monitoring data used in this study, no risk with regard to the water phase could be found for eight of the 41 priority substances, indicating a first success of the implementation of the WFD in the investigated river basins.

Research highlights

► This study assessed the risk of 500 organic substances based on environmental observations in four European river basins. ► A decision tree first classifies chemicals into six categories, which allows water managers to focus on the next steps (e.g. derivation of thresholds, improvement of analytical methods, etc.). ► Prioritization revealed that about three-quarter of substances with “PEC/PNEC” ratios above ten were pesticides.

Introduction

The WFD aims to achieve good status of the European surface waters and groundwater by 2015 and to prevent their further deterioration, which might be caused by a variety of stressors, including toxic chemicals. The Chemical Status of surface water is therefore assessed according to Article 16 of the WFD on the basis of a limited set of 33 priority or priority hazardous substances (PS, daughter Directive 2008/105/EC (Commission, 2008)), including eight priority hazardous substances (PHS) coming from previous legislation, which are regulated and monitored at the European scale. If the remaining chemicals are discharged in significant quantities, they are considered under the Ecological Status assessment (Commission, 2000). To this purpose, Annex VIII of the WFD provides an “Indicative list of the main pollutants” that Member States should use as a basis for identification of the chemicals of potential concern for the Ecological Status assessment, referred to as “specific pollutants” (Commission E, 2000, Wilkinson et al., 2007). It should be noted that these specific pollutants may also consist of pesticides, despite a rather strict pre-market pesticide approval process, which are regulated under Directive 91/414, to prevent potential environmental effects a priori. However, these compounds may end up in the environment in undesirable concentrations and they are therefore also considered under the WFD.

Whether a compound is discharged in significant quantities is commonly decided based on the substance's exposure level, referred to as Predicted Environmental Concentration (PEC), compared to an ecological safety threshold, referred to as Predicted No-effect Concentration (PNEC) (Commission, 2003). PEC/PNEC risk ratios above one would trigger the substance's inclusion into the routine monitoring and the derivation of legally binding thresholds, referred to as environmental quality standards (EQS).

However, considering the vast number of existing chemicals (presently more than 14 million), from which more than 100,000 are produced at industrial scale, as well as their environmental transformation products makes it necessary to reduce the number of candidate substances for both monitoring and RA. For this reason, a number of screening and prioritization exercises have been conducted in the last years (Daginnus et al., 2010, Götz et al., 2010, James et al., 2009, Klein et al., 1999, Muir and Howard, 2006, Wilkinson et al., 2007), giving different emphasis on the required input data. In Switzerland, for example, an exposure-based methodology was developed to rank microcontaminants for monitoring according to their potential occurrence in surface waters (Götz et al., 2010). The methodology was based on the chemicals' distribution behavior between different environmental media, degradation data and input dynamics, while the hazard aspect was ignored. Although this allows for prioritizing chemicals for monitoring based on their potential presence in the environment, chemicals with high toxicity but rather low exposure levels might be overlooked. Similarly, a study by Muir and Howard (2006) focused their prioritization on the persistence, bioaccumulation and the long range transport potential of substances, disregarding their potential toxicity. Again, these properties might justify the inclusion of these chemicals in monitoring programs, as they cover other properties of chemicals that are likewise undesirable, but the methodology disregards the potential toxicity of the compounds and thus hampers a risk assessment and subsequent prioritization with regard to effects on the Ecological Status of river basins.

Most of the proposed prioritization methodologies left approximately half of the candidates unevaluated, because of insufficient data (Wilkinson et al., 2007). One example was the Combined Monitoring-based and Modeling-based Priority setting (COMMPS) procedure, the first European wide prioritization exercise that resulted in the current list of PS (Klein et al., 1999). In that study, emphasis was given to the availability of complete exposure and hazard information, reducing the list of evaluated substances to only 279, disregarding potentially problematic substances with limited data sets. The analytical techniques and the limits of quantification available at that time further limited the number of possible detections (Klein et al., 1999).

A similar approach was applied in the prioritization study carried out for the revision of the first list of PS. According to Article 16 of the WFD, the list of PS needs to be reviewed every four years. The revision of the first list of PS (which is still under way at the time of writing) involved two distinct prioritization procedures: a monitoring-based prioritization study conducted by L'Institut National de l'EnviRonnement Industriel et des RiSques (INERIS), hereafter referred to as “INERIS study”, and a modeling-based prioritization study conducted by JRC (Daginnus et al., 2010). The monitoring-based study, which is the one most frequently referred to in this article, assessed a larger number of substances (339) compared to the COMMPS study and could rely on a more extensive database of environmental observations and a refined hazard assessment, which is discussed more closely in the “Materials and methods” section.

The modeling-based prioritization exercise again evaluated a total of 2034 compounds according to pre-defined hazard and exposure criteria that yielded 78 substances of potential high concern, for which a more intensive assessment was performed (Daginnus et al., 2010). The modeling-based approach used a risk scoring that ranged from 1 to 5, which therefore did not allow for a quantitative assessment based on PEC/PNEC ratios.

It was adapted from another study, performed in the UK (Wilkinson et al., 2007), which also depended on the integration of hazard and exposure predictions and aimed to develop a robust and transparent methodology for identifying and prioritizing Annex VIII chemicals in the UK, referring to Specific Pollutants.

As mentioned earlier, every risk assessment process considers the potential effect of a given substance and its exposure level (Daginnus et al., 2010). With regard to the latter, certain aspects have to be taken into consideration when deriving meaningful indicators for the environmental exposure level of organic pollutants with regard to a river basin, a country or even the whole of Europe. These include: (a) how to treat data below the limit of quantification (LOQ), (b) whether to use maximum or average concentrations, (c) how to consider the bioavailability of a substance (i.e. freely dissolved concentrations) and (d) how to aggregate the data based on a certain percentile.

In chemical databases, the LOQ and LOD (limit of detection) are often not reported, leading to uncertainties whether a concentration is just below LOQ or even below LOD. In Environmental Risk Assessments, half of the reported LOQ value is often used to consider the worst case, i.e. concentrations just below the LOQ. However, considering that some chemicals have expected no-effect levels (e.g. EQS) even below the LOQ, the risk of these substances may be overestimated (James et al., 2009) by assuming they are present at half of the LOQ, when they are actually not.

Moreover, organic chemicals often show high fluctuations in their concentrations over time. Thus, there is a high chance of overlooking short term peak concentrations, especially for compounds with an expected intermittent release, such as pesticides (Pietsch et al., 1995), when applying the widely used method of monthly or even quarterly water grab samples providing WFD-compliant sampling frequencies (Commission, 2000). In the INERIS study, two simple measures were used for the ranking of priority substances: (a) all data (both quantified and non quantified data) at each site, considering half the reported LOQ value for the non quantified data and (b) the arithmetic mean of the data above the LOQ only, at each site (James et al., 2009). Further to that, the 90th percentile was calculated from the monitoring sites' arithmetic means, in order to avoid that single “false” concentrations at single sites would bias the whole prioritization. However, such a low percentile, allowing 10% of sites to exceed the PNEC, seems inappropriate to address relevant local effects on Ecological Status with regard to specific pollutants.

Moreover, hazard data for many compounds are still absent (von der Ohe et al., 2009). Apart from the problem of missing data, certain limitations within the current hazard assessment methodologies, especially with regard to emerging substances, should be addressed. For example, the Technical Guidance Document (TGD) on Risk Assessment (Commission, 2003), as well as the manual for the derivation of EQS (Lepper, 2005), use decreasing assessment factors with increasing data availability. The approach to reduce assessment factors by adding more data is justified with regard to the true uncertainty. However, problem definition uncertainty (e.g. biota are subject to multiple stressors) and variability (e.g. different strain sensitivity for the same test species) are not incorporated in current risk assessment approaches. These uncertainties can be described by means of scientific research, but they cannot be reduced in the context of RA (Ragas et al., 2009).

Even more exhaustive risk assessments, such as those required for pesticides registration, are frequently conducted with a small number of test species of different trophic levels, despite significant differences in the complex species composition of natural communities, e.g. species with generation times of up to 3 years (Kefford et al., 2005). These communities consist of diverse populations of species interacting in complex ways both within and between populations. The strength of these interactions depends on how individuals and populations are affected indirectly by direct toxicant effects on other species which they interact with (Preston, 2002). Therefore, it has been suggested that indirect effects of toxicants have similar or even greater influence on species abundance than direct effects (Lampert et al., 1989).

The aspect of indirect effects is also not fully covered by using simulated ecosystem studies, often referred to as micro- or mesocosm studies, which are typical for pesticide risk assessments (Van Wijngaarden et al., 2005). In these mesocosms, the impact of a chemical on populations or communities of aquatic ecosystems could be assessed under environmentally more realistic conditions than achievable with standard single-species laboratory studies (Lepper, 2005). However, the indirect effects of toxicants are still limited to a relatively small number of test species with typically low generation times (Beketov et al., 2008).

In the light of these considerations, the aim of the present study was to assist water managers in prioritizing organic pollutants of high concern, which may be considered as river basin specific pollutants under Annex VIII of the WFD, while taking into account the current knowledge gaps in the prioritization process. This means that with the prioritization methodology discussed here chemicals are prioritized by action needed (prioritization of substances by action category). To our knowledge, this is the first study that assesses the risk of 500 organic chemicals based on analytical observations which takes systematically into account also the substances for which hazard assessment information is still missing. The dataset comprised the four river basins of Danube, Elbe, Scheldt and Llobregat and was compiled in the project database of the European Research Project MODELKEY (Brack et al., 2005). Firstly, all chemicals were classified into six categories, based on the quality and quantity of the available information, using a simple decision tree developed by the NORMAN network working group on prioritization of emerging substances. Depending on the outcome of the classification, major actions to be taken by water managers were suggested (e.g. derivation of EQS, improvement of analytical methods, etc.). Subsequently, the priority of each substance is assessed overall and within each category to identify the chemicals to be considered first. The methodology is based on adjusted exposure and hazard assessments compared with the approaches mentioned previously. The 95th percentile of the maximum concentrations at each site, referred to as MEC95, was considered as PEC for the exposure assessment, compared with the 90th percentile of the average concentrations as suggested for PS (James et al., 2009). The lowest value of the available acute-based environmental thresholds (PNECacute) and the commonly used chronic-based standards (PNECchronic) were used for the hazard assessment, instead of preferring chronic over acute data per se (Lepper, 2005). If sufficient effect data were not available, provisional Predicted No-effect Concentrations (P-PNEC) were established from read-across models for the acute toxicity to the standard test organisms Daphnia magna, Pimephales promelas as well as Selenastrum capricornutum (von der Ohe et al., 2009), based on a k-Nearest Neighbors (KNN) approach introduced by Kuhne et al. (2007). A safety factor of 1000 was then applied, as suggested by the EQS methodology (Lepper, 2005), which accounts both for problem formulation uncertainty and variability, but also for uncertainty regarding the predictions. Two indicators were used for the prioritization of substances within the six categories: (i) the frequency of sites whose maximum concentration (MECSite) exceeds the lowest PNEC, to address the spatial aspect of exposure, and (ii) the risk ratio (MEC95/lowest PNEC) to assess the intensity of local impacts.

Section snippets

Database

The data sets used were provided by the International Commission for the Protection of the Danube River (ICPDR), which carried out a sampling cruise along the Danube River in 2007 (Joint Danube Survey 2; JDS 2 (Liska et al., 2008)), and by four regional water authorities responsible for the implementation of the WFD: The Landesbetrieb für Hochwasserschutz und Wasserwirtschaft Sachsen-Anhalt (LHW, Magdeburg) and the Sächsisches Landesamt für Umwelt und Geologie (LfUG, Dresden) provided data from

Database

From a total of 500 substances for which chemical monitoring data was available between 2000 and 2008, 114 compounds ceased monitoring in 2004, while another 101 started monitoring only in 2005. The largest group of compounds originated from industrial production processes (29% base products and 11% intermediates), containing 201 substances. Another 189 substances belonged to the group of pesticides and their metabolites (33%) as well as biocides (5%), while pharmaceuticals accounted only for

Improved exposure assessment

It was recognized in the INERIS study that PEC values can be greatly influenced by the number and level of measurements below the analytical determination limit (James et al., 2009). Therefore, two exposure indicators were employed in that study to rank substances according to the arithmetic mean of (1) all analytical measurements for a given substance, a given station, and a given analytical fraction as well as (2) only for analytical measurements above the LOQ for a given substance, a given

Summary

In the present paper, we introduced a classification approach that addresses especially the data scarcity for emerging substances and hence, would allow prioritizing substances as specific pollutants under Annex VIII of the WFD. Because emerging substances are not usually considered in conventional prioritization methodologies, they are monitored less often and as a result, little data are available to show evidence of risk. In other words, they are caught in a “vicious cycle”. Hence, six

Acknowledgements

The presented prioritization approach was developed within the NORMAN Association (No. W604002510) working group on prioritization of emerging substances (WG 1) and was approved by the WG members present at the WG meeting in Paris (22–23 November 2010). The work was supported by the European Commission through the Integrated Projects MODELKEY (Contract-No. 511237GOCE) and OSIRIS (contract No. 037017). Peter C. von der Ohe was financially supported through a Deutsche Forschungsgemeinschaft (DFG)

References (44)

  • Commission E
  • K. Daginnus et al.

    A modelling approach for the prioritization of chemicals under the water framework directive

  • D. De Zwart

    Observed regularities in species sensitivity distributions for aquatic species

  • D.M. Di Toro et al.

    Technical basis for establishing sediment quality criteria for nonionic organic chemicals using equilibrium paritioning

    Environ Toxicol Chem

    (1991)
  • D.L. Geiger et al.

    Acute toxicities of organic chemicals to fathead minnows (Pimephales promelas)

    (1990)
  • C.W. Götz et al.

    Targeting aquatic microcontaminants for monitoring: exposure categorization and application to the Swiss situation

    Environ Sci Pollut Res

    (2010)
  • IPCS

    Screening information data set (SIDS) for high production volume chemicals

  • A. James et al.

    Implementation of requirements on priority substances within the context of the Water Framework Directive

  • B.J. Kefford et al.

    What is meant by “95% of species”? An argument for the inclusion of rapid tolerance testing

    Hum Ecol Risk Assess

    (2005)
  • W. Klein et al.
  • R. Kuhne et al.

    Calculation of compound properties using experimental data from sufficiently similar chemicals

  • R. Kuhne et al.

    Estimation of compartmental half-lives of organic compounds — structural similarity versus EPI-Suite

    QSAR Comb Sci

    (2007)
  • Cited by (259)

    View all citing articles on Scopus
    View full text