Using population level consequences as a basis for determining the “x” in ECx for toxicity testing


Recently, a stimulating and valuable discussion about the uses of NOEC/NOEL and ECx has taken place in Integrated Environmental Assessment and Management (e.g., Landis and Chapman 2011; Green et al. 2013). Landis and Chapman (2011) argued for dropping NOEC/LOEC as the metric for describing results from chronic toxicity tests, and instead focusing on using ECx (where “x” is a percentage effect). However, Green et al. (2013) pointed out that it may be difficult to determine which percent effect should be used (i.e., the x in ECx). If the x value is not chosen wisely, the use of ECx may, according to the authors, result in ecological risk assessments (ERAs) that are no better than when NOEC/NOEL is used. In this Learned Discourse (LD) we join the discussion and provide a recommendation on how to handle this issue by referring to current knowledge obtained from ERA studies.

Green et al. (2013) used several examples to show how difficult it is to find a universal x-value that can be used for all endpoints in ERA. As examples, the authors mention the biomarker vitellogenin (that is used to identify endocrine disruption) where increases of 1000% or more are common, and mean length of Daphnia where an effect of 2% to 3% is statistically detectable and may be considered biologically important. Using the same x value (e.g., 10%, a frequently-used value) for both of these endpoints would not be appropriate. In the present article, we argue that we need to use different x values for different endpoints, and that it is the population-level consequences that should determine the x value.

Management goals are typically developed at the population level (or higher). Therefore, it is desirable to relate ECx to population-level consequences. Fortunately, using population models, we can investigate how percent reductions in individual-level traits, such as survival and fecundity, are linked to population growth rate and extinction probability (Hanson and Stark 2011, 2012). By doing so, we can obtain information concerning how serious a given effect is from the perspective of environmental management goals. Given the diverse and numerous endpoints at individual (organismal) and suborganismal (biomarkers) levels that are available for ERAs, such information is valuable. However, these kinds of studies are surprisingly rare, and more research is required in this field.

Based on the acceptable population-level impacts (that should be determined by society or stakeholders, not by science), we can use population models to determine acceptable x values for different endpoints. For example, if the management goal is to prevent extinction of species in the environment, the chemical concentration that leads to zero net population growth (population threshold concentration [PTC]) can be used to determine ECx. Any chemical concentrations higher than the PTC will lead the population to decline and eventually to extinction. The estimation of PTCs for various species has been carried out previously in several studies (e.g., Kamo and Naito 2008; Iwasaki et al. 2010). By collecting such PTCs for a variety of species, it will be possible to set up the population-level species sensitivity distribution for a given chemical and choose an x level that protects a desirable proportion of the species (see Kamo and Naito 2008). This is a far more appealing strategy than the current strategy, where NOEC/NOEL is divided by different numbers (often called uncertainty or safety factors) to achieve more or less environmental protection, with no direct link to the protection goals. We recognize that it will take considerable effort and time to make the approaches that are mentioned in this LD available for practical ERAs, but we are convinced that they will play crucial roles in carrying out value-relevant ERAs in the not too distant future.