Risk assessment procedures use toxicity tests in which organisms are subjected to chemicals under otherwise constant and favorable experimental conditions. Because variable and suboptimal environmental conditions are common aspects of natural ecosystems, the hazard of underestimation of risk arises. Therefore, an uncertainty factor is used in the extrapolation of results of standard toxicity tests to field situations. The choice for these uncertainty factors is based on little ecological evidence. This review discusses studies on the toxicity of various chemicals to aquatic organisms, modified by temperature, nutritional state and salinity, excluding papers on changes in bioavailability of compounds. Collected data were analyzed quantitatively to evaluate the validity of toxicity data obtained from standard toxicity tests in the laboratory under field conditions. Generally, organisms living under conditions close to their environmental tolerance limits appeared to be more vulnerable to additional chemical stress. Usually, increasing temperature and decreasing food or nutrient level raised toxicity. The influence of salinity was less clear; metal toxicity increased with decreasing salinity, toxicity of organophosphate insecticides increased with higher salinity, while for other chemicals no clear relationship between toxicity and salinity was observed. The interactions can be explained by several physical and physiological processes, acting on factors such as bioavailability, toxicokinetics, and sensitivity of organisms. Quantitative analysis of data indicated that an uncertainty factor for the laboratory to field extrapolation should be smaller than one for an ecosystem in a temperate region, while a factor greater than one would be appropriate for systems nearby discharge points of cooling water. The factor should be greater than one when varying nutritional state is concerned, but smaller than one with respect to salinity. Dependent on the effect parameter used, the differences in toxicity between laboratory and relevant field situations ranged from a factor of 2.6 to 130 and 1.7 to 15 for the two temperature conditions and 1.2 to 10 for nutritional state. A salinity increase from freshwater to marine water decreased toxicity by a factor of 2.1. However, as less extreme salinity changes are more relevant under field conditions, the change in toxicity is probably much smaller. To obtain uncertainty factors that sufficiently protect natural systems without being overprotective, additional research is required.

The question whether environmental pollution affects genetic diversity in natural populations remains unanswered to date despite the fact that genetic variation is one of the three pillars of biodiversity recognized in the Rio convention of 1993. The loss of genetic diversity in populations subjected to anthropogenic stress can be designated as “genetic erosion” and may be considered as a factor of concern in risk assessment of toxic chemicals. Theoretically there are four different ways in which toxicants can affect genetic variation: (i) by increasing mutation rates, (ii) by directional selection on tolerant genotypes, (iii) by causing bottleneck events, and (iv) by altering migration. This paper reviews studies that have documented genetic change in animal populations exposed to environmental pollution. In these studies, genetic variation is measured in a variety of ways: heritability of quantitative characters, heterozygosity of allozyme loci, haplotype diversity in mitochondrial DNA, and variability in RAPD fingerprints. Studies on cadmium tolerance of Collembola living in metal-contaminated soil suggest that strong directional selection pressure may decrease genetic variability of traits immediately linked to tolerance. Allozyme studies in fish have documented a similar decrease of genetic variation in populations living in strongly acidified waters. A correlation between RAPD-PCR-based genetic similarity and site contamination has been documented in crayfish. Overall, there is significant support for the genetic erosion hypothesis, but the issue cannot be considered settled. In most studies insufficient attention is given to factors such as population size, bottlenecks and mutation, which may influence genetic variability in addition to the toxicant selection regime. At the moment, there does not seem to be a sound scientific basis for incorporating genetic diversity measurements into risk assessment, despite the variety of easily applicable molecular techniques available. It is often not known what kind of variation is measured by these techniques (neutral or selectable) and how the markers are inherited. Given the importance of the issue, as stressed by the Rio Convention, a concentrated research effort is necessary to better define the question and find a general approach to evaluate its importance in ecological risk assessment.