Determining high-quality critical body residues for multiple species and chemicals by applying improved experimental...

Ecotoxicological effect data are generally expressed as effective concentrations in the external exposure medium and do thus not account for differences in chemical uptake, bioavailability, and metabolism, which can introduce substantial data variation. The Critical Body Residue (CBR) concept provides clear advantages, because it links effects directly to the internal exposure. Using CBRs instead of external concentrations should therefore reduce variability. For compounds that act via narcosis even a constant CBR has been proposed. Despite the expected uniformity, CBR values for these compounds still show large variability, possibly due to biased and inconsistent experimental testing. In the present study we tested whether variation in CBR data can be substantially reduced when using an improved experimental design and avoiding confounding factors. The aim was to develop and apply a well-defined test protocol for accurately and precisely measuring CBR data, involving improved (passive) dosing, sampling, and processing of organisms. The chemicals 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, 2,3,4-trichloroaniline, 2,3,5,6-tetrachloroaniline, 4-chloro-3-methylphenol, pentylbenzene, pyrene, and bromophos-methyl were tested on Lumbriculus variegatus (California blackworm), Hyalella azteca (scud), and Poecilia reticulata (guppy), which yielded a high-quality database of 348 individual CBR values. Medians of CBR values ranged from 2.1 to 16.1 mmol/kg wet weight (ww) within all combinations of chemicals and species, except for the insecticide bromophos-methyl, for which the median was 1.3 mmol/kg ww. The new database thus covers about one log unit, which is considerably less than in existing databases. Medians differed maximally by a factor of 8.4 between the 7 chemicals but within one species, and by a factor of 2.6 between the three species but for individual chemicals. Accounting for the chemicals' internal distribution to different partitioning domains and relating effects to estimated concentrations in the target compartment (i.e., membrane lipids) was expected to but did not decrease the overall variability, likely because the surrogate partition coefficients for membrane lipid, storage lipid, protein, and carbohydrate that were used as input parameters did not sufficiently represent the actual partitioning processes. The results of this study demonstrate that a well-designed test setup can produce CBR data that are highly uniform beyond chemical and biological diversity.