Application of Bioavailability Models

The empirically based regression models predicting nickel toxicity in spiked sediments based on sediment properties can be used to normalize the available toxicity data set to specific bioavailability conditions. For example, a Realistic Worst Case (RWC) PNEC can be derived by recalculating the nickel sediment toxicity database toward a reference situation, such as 10th percentile of the regional AVS/ distribution, in case no actual or historical AVS data are available.  Alternatively, a PNEC can be derived for the actual AVS concentrations occurring at the site as follows:

  1. Link the NOEC/ECX values of the chronic ecotoxicity database (as total metal concentrations) with the relevant sediment parameters of the sediment (e.g., AVS) in which the test was performed.

  2. For the regression models [taking the form log(ECx) = intercept + slope * log(abiotic factor)], the corresponding organisms specific slopes can be used to normalize the NOEC/ECX values to “reasonable worst case” sediment properties (e.g., 10th percentile AVS) or to specific local/regional conditions (actual or historical AVS concentrations prevailing on the site under investigation). The normalization equations for RWC and site-specific conditions are given below in Equations 1 and 2.



The different bioavailability models have been used to normalize the final toxicity dataset to nine different bioavailability scenarios (eight specific sediments ranging from low to high AVS content and one hypothetical RWC condition). The different sediments are described in the paper of Vangheluwe et al. (2013). Bioavailability models are available for seven species. For all the derived EC10 values the oligochaete L. variegatus is the only test organism without a specific bioavailability model, and hence, this is the only data point in the Species Sensitivity Distribution (SSD) for which the bioavailability model for another species needs to be used. The choice of the model to use for L. variegatus was based on biological and practical considerations, including similarities in life history and behavior with other species for which bioavailability models were established, and the degree to which different bioavailability models reduced observed intra-species variability for L. variegatus. Oligochaetes such as L. variegatus alter their immediate environment through the formation of I-shaped burrows. Tubificids live in a similar way head-down in relatively permanent vertical burrows feeding on deposits on some depth. In contrast with Hexagenia, the micro-habitats of oligochaetes are not irrigated with oxygenated water; nevertheless, this behavior will minimize the mitigating capacity of a bioavailability factor such as AVS as reflected in the smaller slope of the AVS model developed for Hexagenia and T. tubifex. Because of the close similarity between tubificid/oligochaete worm behavior the T. tubifex model is used to normalize the L. variegatus data. This choice is considered precautionary as it has the smallest slope of all species tested (i.e. 0.125). Species to species "read across,” however, may introduce additional uncertainty. Figure 2 represents the lognormal functions normalized for AVS that were fitted through the eight data points for eight sediments and the RWC. All functions were accepted at P < 0.05.


A summary of the estimated HC5-50 values (with the 5-95% confidence limits) for the different log-normal distributions is provided in Table 2. The HC5-50 values obtained for the different bioavailability scenarios range with the AVS model from 109-305 mg/kg dry wt.  The PNEC was calculated as the HC5-50 multiplied by an AF (assessment factor) of 1.