Relation of Survival to Other Endpoints in Chronic Toxicity Tests With Fish

Abstract

Hazard assessments of chemicals in aquatic organisms often include chronic toxicity testing. The evaluation of exposure duration and of the life stages tested according to standard test methods has led to the development of shorter chronic toxicity tests. A similar evaluation of biological endpoints (i.e., survival, growth and reproduction) could result in tests that are more economical. We analyzed endpoints for 28 chemicals and seven fish species in 34 chronic toxicity studies. When all endpoints were compared, survival was equal to or more sensitive than all other endpoints 56 to 69% of the time. Individual endpoints were more sensitive than survival 19 to 61% of the time, except for reproduction, which was always more sensitive (although there were few observations). The no observed effect concentration (NOEC) for growth could be predicted from the NOEC for survival by using interendpoint correlations (r = 0.949 to 0.974). Ratios of NOECs for survival to those for all other endpoints examined were 5 or less in 93 to 96% of the comparisons (specific endpoint comparisons ranged from 80 to 100%). The determination of the survival endpoint requires less time and money than does the determination of most other endpoints, and it appears adequate for hazard assessments in the initial stage of estimating chronic toxicity. However, a factor of at least 0.2 should be applied to the estimated no‐effect concentrations for survival to include other potential biologically significant effects at least 95% of the time. The factor of 0.2 is based on frequency analyses that resulted in the NOECs for survival being 5 times or less than the NOECs for most other endpoints about 95% of the time. Univariate analyses, however, indicated a range of 0.13 to 0.22 for the factor. A thorough evaluation of other published studies that contain endpoints other then survival should be conducted to define the appropriate factor more accurately.

Accuracy assessment of time–concentration–effect models in predicting chronic lethality from acute toxicity data

“…Acute-to-chronic models do not estimate chronic no-or low-effect values for growth and reproduction, but estimates for fish can be made from ACE estimates for chronic lethality. Growth (standard length) correlates well with lethality [25], but true chronic reproductive tests are few. If the user of ACE wants to estimate growth and reproductive effects, we provide the models below based on present data (CV ¼ geometric mean of NOEC-LOEC): log 10 growth CV ¼ À 0.093 þ (0.96 log 10 lethality estimate); r ¼ 0.98, n ¼ 69; and log 10 repro CV ¼ À 0.37 þ (0.98 log 10 lethality estimate); r ¼ 0.96, n ¼ 6.…”

Report of the OECD Workshop on Statistical Analysis of Aquatic Toxicity Data

“…ECx values and NOECs, as commonly estimated, can never be compared and nothing of use can come from trying to establish relationships between them. Mayer et al (1986) used elaborate statistical methods to compare NOECs between different studies involving different chemicals and different fish species. Again, the design dependence of the NOEC means that such an investigation cannot be relied upon to provide sensible conclusions.…”

A framework for ecological risk assessment at contaminated sites in Canada

“…Assuming that the untested species has a similar sensitivity to the test species, the untested species are expected to fall within the same range (Mayer et al, 1986; also, this assumption was the initial basis for development of water quality criteria). Species within a similar taxa can have a wide range of response concentrations, but the more data one compiles, the more confidence one can place in the range or interval.…”