The International Zinc Association-Europe (Van Assche 2005) is “surprised and disturbed at the premature publication by your journal of the preliminary results… (Bodar et al. 2005)… and conclusions of a draft risk assessment on zinc that remains under consideration within Europe.” Van Assche (2005) provides several examples to illustrate his concerns that the paper by Bodar et al. (2005) is incorrect, outdated, and misleading to the reader.

In the following response, we, as the authors, will explain why our paper is correct, dated, and not misleading. Notwithstanding those qualifications, we are happy to clarify further some issues that could have been subject to misinterpretation.

Van Assche (2005) criticizes the inclusion of preliminary European results. We clearly and openly stated in the article, however, that the environmental part of the risk assessment has to go through a final, written round of comments, mainly for editorial purposes, and that the Scientific Committee for Human and Environmental Risks (SCHER) will subsequently review the risk assessment. Because the general approach and the main conclusions of the zinc risk assessment were set at European Union (EU) technical meeting level—which is still the case—we felt that it was not opportunistic to bring this very important piece of work to the attention of the scientific community. Integrated Environmental Assessment and Management's peer reviewers apparently shared this viewpoint. Once the SCHER has released its opinion concerning the risk assessment report on zinc and zinc compounds, we will be happy to update the scientific audience about relevant aspects. If the journal will allow us, this could be done via a Letter to the Editor.

According to Van Assche (2005), the Abstract incorrectly states the assessment concluded that a regional risk was drawn for agricultural soils. This is in contrast to the contents of the paper itself. Indeed, a conclusion 2 (i.e., no risk) is drawn for the regional soil, contrary to the local situation for soil where potential risks may occur. The sentence in the Abstract may, in retrospect, have been subject to multiple interpretations from a linguistic point of view. However, we wrote it without the implicit link between, on the one hand, local and regional risks and, on the other hand, water, sediment, and soil.

With respect to zinc emissions to surface water from agricultural soils, Van Assche (2005) criticizes that even though these emissions are stated in the text, they are not integrated in the figures. At present, ongoing research is making clear that such emissions may, indeed, contribute significantly to the water levels of zinc. The magnitude of this emission route, however, is still not clear and, therefore, is not yet incorporated into the “official” Netherlands emission data from CCDM (Netherlands Emission Registration). In our paper, we referred to these official “firm” data (CCDM), as was done in the zinc risk assessment itself.

With respect to the crucial point of predicted no effect concentration (PNEC) setting, Van Assche (2005) draws attention to the paper by Van Sprang et al. (2004), in which a different PNEC for water was derived, using basically the same ecotoxicity data. Key aspects of the PNEC setting, discussed by Van Sprang et al. (2004), e.g. the use of the best-fit distribution versus use of a default distribution are ignored by Bodar et al. (2005). We would like to respond to two elements in this area. First, with regard to basically the same ecotoxicity data that were used by Van Sprang et al. (2004) in deriving a different PNEC for water, a reference is made to the ongoing zinc risk assessment described by Bodar et al. (2005). Because not exactly the same ecotoxicity data were used, another PNEC for water will evidently be derived as compared to the PNEC for water in Bodar et al. (2005). Why Van Sprang et al. (2004) did not use exactly the same set of ecotoxicity data is not clear, but Bodar et al. (2005) referred to the ecotoxicity data set that was adopted and agreed on by the technical meeting from member states, industry, and nongovernmental organizations. Because Bodar et al. (2005) described the outcome of the discussions from that technical meeting, it was of no use referring to other exercises, such as that by Van Sprang et al. (2004).

Second, with regard to use of the default distribution over the best-fit distribution, Bodar et al. (2005) again described the outcome of the discussions in the technical meeting. With respect to the choice or preference for the default distribution, Bodar et al. (2005) did not include the lengthy discussions that took place at the technical meeting, where we discussed the various possibilities and methods that could be employed in deriving a PNEC for water using probabilistic methods for this data-rich substance. In fact, the risk assessment report included the results of a log-logistic distribution, nonparametric method, triangular distribution, and extreme value distribution, such as that used by Van Sprang et al. (2004), in addition to the log-normal distribution. The technical meeting decided to base the PNEC for water on the outcome of the log-normal distribution, whereas the outcome of the other methods were used in the further decision regarding the magnitude of the assessment factor (see Table 3 in Bodar et al. 2005). The arguments for not choosing the methodology as proposed by Van Sprang et al. (2004) were of a more principal basis. Statistically, the method by Van Sprang et al. (2004) leads to a better fit. However, the rationale for the log-normal distribution was preferred by the technical meeting, because it assumes a type of distribution in which some species are more sensitive and others less sensitive to zinc, which can be described by a log-normal function.

In summary, we did not refer to Van Sprang et al. (2004), or to any other scientific publication that derived a PNEC for water or other protection levels, because they did not use the same ecotoxicity data set and to avoid repeating lengthy discussions in Bodar et al. (2005).

Regarding the PNEC sediment, Van Assche (2005) mentions that new toxicity data have been generated and, therefore, that the predicted environmental concentration (PEC) to PNEC ratios mentioned in Table 5 of Bodar et al. (2005) are outdated. At the time of the writing and acceptance of Bodar et al. (2005), we were not aware that new toxicity data had been generated. During the process of printing and proofreading, however, we fortunately became aware of that new information. Therefore, we included on page 314 of Bodar et al. (2005) a sentence drawing attention to a possible new PNEC for sediment. We did not explicitly include that a new PNEC for sediment also would affect the PEC to PNEC ratios for sediment as listed in Table 5.

Currently, we have proposed a new PNEC for sediment that recently was commented on by the technical meeting. Very soon, we will update the risk assessment report concerning this. Therefore, indeed, these PEC to PNEC ratios will be outdated soon. However, because the new PNEC for sediment does not significantly affect the conclusions—that is, whether the PEC exceeds the PNEC—there will be no dramatic implications. Although this may be self-evident, we should have mentioned more explicitly that not only the PNEC for sediment but also the PEC to PNEC ratios for sediment could change following the new toxicity data for sediment.

The point of deterministic versus probabilistic risk assessment also is referred to by Van Assche (2005). Of course, a risk assessment based on the same set of input data can be conducted in other ways (see above). Van Sprang et al (2004) is an example of an alternative approach, in this particular case containing much more probabilistic elements than the current EU risk assessment on zinc. The issue of state-of-the-art may then become semantic. Clearly, however, the current risk assessment reflects the outcome of both pragmatism and scientific insights. This is all within the settings and limitations of a multistakeholder process, in which the consensus approach is mostly prevalent. State-of-the-art then means as far as such processes may go. Gaps and delays will always exist between international decision processes like the EU Existing Chemicals Program and what science can offer. In the specific case of probabilistic risk assessments, we clearly refer in our paper to the SCHER conclusion of 2004 (CSTEE 2004) that no scientific consensus yet exists regarding how to carry out an uncertainty analysis within the EU so that it can adequately assist policy makers in decision making.

We do not see why the PNECs derived in the zinc risk assessment should not be useful for setting water-quality standards and sanitation levels, as contested by Van Assche (2005). This should include the newly derived PNEC for sediment. We furthermore emphasize that the bioavailability correction, as incorporated in the zinc risk assessment, should be an essential part of that process.


  1. Top of page
  2. References
  • Bodar CWM, Pronk MEJ, Sijm DTHM. 2005. The European Union risk assessment on zinc and zinc compounds: The process and the facts. Integr Environ Assess Manag 1: 301319.
  • CSTEE. 2004. Subgroup meeting of Technical Committee on New and Existing Substances and Scientific Committee for Toxicity, Ecotoxicity and the Environment (CSTEE) on the “Environmental Risk Assessments of Cadmium and Zinc”. ECBTCNES0804(ERAR_CD&ZN_meeting25.08.2004). JRC Ispra, Italy, August 25, 2004.
  • Van Assche F. 2006. Letter to the Editor. Integr Environ Assess Manag 2: 103.
  • Van Sprang PA, Verdonck FAM, Vanrolleghem PA, Vangheluwe ML, Janssen CR. 2004. Probabilistic environmental risk assessment of zinc in Dutch surface waters. Environ Toxicol Chem 23: 29933002.