Critical review and analysis of aquatic toxicity data on oil spill dispersants

Abstract

Oil spill response requires consideration of several countermeasures including chemical dispersants, but their potential toxicity to aquatic species poses a concern. Considerable in vivo aquatic toxicity data from laboratory exposures have been generated since 2010 for current-use dispersants. The objective of the present review is to provide a synthesis of these data to improve dispersant hazard assessments. Data from multiple studies were evaluated based on reliability criteria. Although procedures, standards, endpoints, and statistical approaches were usually described, nearly a quarter of sources did not provide sufficient information to judge study quality but were considered on a case-by-case basis. Data were used to develop dispersant-specific species sensitivity distributions and hazard concentrations protective of 95% of the species (HC5). Given data limitations, post-2010 toxicity data were augmented with pre-2010 data and model predictions. The HC5s calculated for 54 dispersants fell mostly within the moderate to slightly toxic range and were compared to field dispersant-only concentrations estimated from operational application rates under conservative assumptions. Based on available evidence, dispersants may not pose a significant risk under field conditions to most aquatic species, if proper application and dilution are taken into account. Recommendations on improved toxicity testing and reporting as well as research needs are also provided. Environ Toxicol Chem 2018;37:2989-3001. © 2018 SETAC.

Further Development and Refinement of Interspecies Correlation Estimation Models for Current‐Use Dispersants

“…For example, only 13 of 189 observed observations fell below the HC5's lower 95% CI. Furthermore, in all but one case (the 24‐h value for Corexit 9500) time‐varying HC5s were lower or had CIs that overlapped with HC5s derived from observed data only (Table in Bejarano ).…”

“…Toxicity Even though these ICE-enhanced SSDs were comprised of data for only 5 species (2 observed; 3 predicted), HC5s were in most cases within 3-fold of those previously reported (Bejarano 2018), but from SSDs with much larger species diversity (i.e., observed data only or nondispersant ICE models). Although it is preferable to generate ICE-enhanced SSDs with predicted data from multiple surrogates, this practical application shows that predictions from a single surrogate produce results comparable to those obtained with more robust datasets.…”

“… a HC5s from SSDs developed with observed data only (b) or using nondispersant ICE models (Tables and 3 in Bejarano ) are reported as a reference. The NCP Product Schedule can be found at US Environmental Protection Agency .…”

“…For example, only 13 of 189 observed observations fell below the HC5's lower 95% CI. Furthermore, in all but one case (the 24-h value for Corexit 9500) time-varying HC5s were lower or had CIs that overlapped with HC5s derived from observed data only ( Table 2 in Bejarano 2018). Two of the practical applications just described could be potentially useful in facilitating laboratory-to-field comparisons.…”

“…The ICE models were developed for each unique pair of surrogate-predicted species using the methods and procedures outlined elsewhere (Bejarano andBarron 2014, 2016). Briefly, ICE models were developed for species pairs with at least 4 acute toxicity data points using linear regression models described by…”