Application of survival analysis methods to pulsed exposures: Exposure duration, latent mortality, recovery time, and the underlying theory of survival distribution models.

摘要

Ecotoxicologists adopted median lethal concentration (LC50) methods from mammalian toxicology. This conventional LC50 approach has shortcomings. Fixing the exposure duration and selecting the 50% mortality level result in loss of ecologically relevant information generated at all other times. It also ignores latent mortality that can manifest after exposure ends. As a result, it cannot adequately predict pulsed exposure effects in which concentration, duration, and frequency of pulses change through time. The underlying theory of the dose-response models used to calculate LC50 values, stochastic versus individual effective dose (IED) theory, has not been tested rigorously either.; In this study, the effects of exposure duration and concentration on mortality during and after exposures, and the effects of recovery time between two pulses on mortality during a second pulse were quantified. The influences of toxicant modes of action were discussed. The underlying theory for survival distribution models was further explored. Survival analysis was used to incorporate these factors into predictive models and to circumvent some of the aforementioned shortcomings. The experiments were conducted with two contrasting toxicants, copper sulfate (CuSO4) and sodium pentachlorophenol (NaPCP). The amphipod, Hyalella azteca, was used as the model organism.; Latent mortality is significantly affected by exposure concentration and an integral part of the lethal effects of toxicants that cause cumulative damage. For toxicants that cause minimal cumulative damage during the exposure, the latent mortality is not significant and can be ignored. Exposure duration did not show any significant effect on latent mortality for either toxicant. It is recommended that for other experimental conditions the effect still needs to be considered. Recovery time had significant effect on mortality during the second pulse for both toxicants. To recover to a similar background level mortality, the time an exposed organism needed to return to a stage similar to its original resistance was much longer for CuSO4 than for NaPCP. The hypothesis that IED is the dominant explanation for the dose-response models was rejected for both toxicants. By effectively incorporating exposure duration and other factors, survival analysis better predicted pulsed exposure consequences than did the conventional LC50 methods.