Many studies have investigated the consequences of exposure to fine-grained suspended sediments (SS) on aquatic organisms. Exposure has two components-concentration and duration-and can be expressed as dose, where we define suspended sediment dose (SSD: mg·h·L-1) as the product of suspended sediment concentration (SSC: mg·L-1) and duration of exposure (DoE: h). We evaluated these three measurement endpoints for managing SS effects on salmonids by assembling and analyzing all published SS dose-response observations. Despite a prevalence in SS management guidelines for using SSC as a primary endpoint to manage SS effects on salmonids, SSC was found to be less effective than SSD or DoE as a predictor variable for the available dose-response observations. We used data visualization to identify trends and distinct response categories that were then evaluated using a logistic regression model that accounts for nested observations by study. The model estimates the probability of moving from behavioural to major physiological and lethal responses in the available literature, as explained by one or more predictor variables, including ln(SSC), ln(DoE), ln(SSD), and life stage (adult versus juvenile). Akaike Information Criterion (AIC) and receiver operating characteristic (ROC) were used to compare model fit and classification performance, respectively, among alternative models. The best performing model as judged by AIC and ROC incorporated ln(SSD) as the predictor variable.
Keywords: Dose-response; Ecotoxicology; Fish; Multilevel logistic regression.
Copyright © 2021 Elsevier B.V. All rights reserved.