Optimizing use of toxicity test data: Censored data in species sensitivity distributions and temporal data in mixture studies



Journal Title

Journal ISSN

Volume Title



Since toxicity testing is time and resource intensive, there is incentive to make more efficient use of the resulting data. This research investigates two approaches for optimizing use of toxicity test data: (1) the use of censored or implied censored data in species sensitivity distributions (SSDs), and (2) the application of time-to-event (TTE) analysis in detection of interactions among mixture components. In Chapter 1, SSDs were constructed for marine and tidally influenced species within the state of Texas in order to calculate dissolved oxygen (DO) concentrations protective of 95% of species (HC5s). Allowance of censored data, expressed as less-than, greater-than, or interval values was utilized as a way of including more of the available information in models. In addition to censored data reported as such, this study incorporated an additional source of censored data — that which is implied by non-standardized data. The study found median HC5 estimates ranging from 1.5 to 2.2 mg/L. Use of implicit censored data was shown to be an effective strategy for increasing data quantity, increasing the fish dataset from 10 species to 14 species, and the invertebrate dataset from four species to five. The effect of censored data on HC5s varied in degree and direction of shift, however no significant differences were observed. In Chapter 2, Ceriodaphnia dubia temporal toxicity data were analyzed from 96-h co-exposures to pesticides and predator cues. In contrast to typical methods of analysis, which focus on observations at a fixed time point (e.g., 96-h LC50), this study used TTE analyses, which facilitate inclusion of data collected over the time-course of a toxicity test. A novel application of TTE methods to the detection of mixture interaction effects was performed. The greater power afforded by use of temporal data allowed a more comprehensive characterization of mixture toxicity across time and concentration gradients. Interaction effects were demonstrated to be highly concentration specific, typically occurring below three toxic units and often in only one or a few concentrations. Detection of interactions was dependent on test length and number of observations. The study also provides some clues as to the possible mechanisms governing antagonistic and synergistic interaction between predator cues and pesticides.



Species sensitivity distributions, Censored data, Hypoxia, Dissolved oxygen, Gulf of Mexico, Mixtures, Mixture toxicology, Predator cues, Ceriodaphnia dubia, Time, Time-to-event