ABSTRACT While water quality benchmarks for the protection of aquatic life have been in use in some jurisdictions for several decades (USA, Canada, several European countries), more and more countries are now setting up their own national water quality benchmark development programs. In doing so, they either adopt an existing method from another jurisdiction, update on an existing approach, or develop their own new derivation method. Each approach has its own advantages and disadvantages, and many issues have to be addressed when setting up a water quality benchmark development program or when deriving a water quality benchmark. Each of these tasks requires a special expertise. They may seem simple, but are complex in their details. The intention of this paper was to provide some guidance for this process of water quality benchmark development on the program level, for the derivation methodology development, and in the actual benchmark derivation step, as well as to point out some issues (notably the inclusion of adapted populations and cryptic species and points to consider in the use of the species sensitivity distribution approach) and future opportunities (an international data repository and international collaboration in water quality benchmark development).

ABSTRACT Environmental hazard assessments for chemicals are performed to define an environmentally 'safe' level at which, theoretically, the chemical will not negatively affect any exposed biota. Despite this common goal, the methodologies in use are very diverse across different countries and jurisdictions. This becomes particularly obvious when international scientists work together on documents with global scope, e.g. in the WHO International Program on Chemical Safety. In this paper, we present a study that describes the extent of such variability and analyse the reasons that lead to different outcomes in deriving a 'safe level' (termed the Predicted No Effect Concentration (PNEC) throughout this paper). For this purpose, we chose five chemicals to represent well-known substances for which sufficient high-quality aquatic effects data were available: ethylene glycol, trichloroethylene, nonylphenol, hexachlorobenzene and copper. From these data, two datasets for each chemical were compiled: the full dataset, that contained all information from selected peer-review sources, and the base dataset, a sub-sample of the full set simulating limited data. Scientists from the EU, United States, Canada, Japan and Australia independently performed hazard assessments for each of these chemicals using the same datasets. Their reasoning for, e.g., key study selection, use of assessment factors or use of probabilistic methods was comprehensively documented. The observed variation in the PNECs for all chemicals was up to three orders of magnitude and this was not simply due to obvious factors such as the size of the dataset or the methodology used. Rather this was due to individual decisions of the assessors within the scope of the methodology used, especially key study selection, acute versus chronic definitions and size of assessment factors. Awareness of these factors, together with transparency of the decision-making process, would be necessary to minimise confusion and uncertainty related to different hazard assessment outcomes, particularly in international documents. The development of a 'guideline on transparency in decision-making' ensuring the decision-making process is science-based, understandable and transparent, may therefore be a promising way forward. Integr Environ Assess Manag © 2013 SETAC.

Water quality guidelines for chemicals: learning lessons to deliver meaningful environmental metrics.