Shinyssd v1.0: Species sensitivity distributions for ecotoxicological risk assessment

Abstract

Living organisms have different sensitivities to toxicants. This variability can be represented by constructing a species sensitivity distribution(SSD) curve, where by the toxicity of a substance to a group of species is described by a statistical distribution. Building the SSD curve allows calculating the Hazard Concentration 5% (HC5), that is, the concentration at which 5% of the considered species are affected. The HC5 is widely used as an environmental quality criterion and a tool for ecological risk assessment (Posthuma, Suter II, & Traas, 2001). The shinyssd web application is a versatile and easy to use tool that serves to simultaneously model the SSD curve of a user-defined toxicity dataset based on four different statistical distribution models (log-normal, log-logistic, Weibull, Pareto). shinyssd directly calculate sthree estimators HC1, HC5 and HC10 associated to the four distribution models together with its confidence intervals, allowing the user to select the statistical distribution and associated HC values that best adjust the dataset. Thelevel of confidence of the result sobtained from a SSD curve will depend on the number of species used to produce the SSD. In this sense, the first tab of the user interface is used for visualizing the number of species for which toxicological data are available for each toxicant, species group, and endpoint combination in the uploaded dataset. A minimum of species is necessary to build a SSD curve varies according to the literature (Belanger et al., 2016; Newman et al., 2000; Plant Protection Products & Residues, 2013; Wheeler, Grist, Leung, Morritt, & Crane, 2002). After selecting the toxicant and species groups, the user can filter and select subsets of data from the whole database by applying different quality criteria (e.g., if the studies reported a chemical confirmation of the concentration sof the toxicanttested). The values enteredineach column of the data base serveas categories to filter the data basein relation to characteristics of the bioassays. The final SSD curve is fitted to different distributions using the package fitdistrplus and actuar. The HC is estimated for all the distributions. By facilitating and streamlining toxicity data analysis and the creation of SSD curves, the user interface proposed here should be useful for environmental managers and regulators conducting ecological risk assessments and scientific research.