5 Summary and conclusions

Toxicity testing with the collembolans Folsomia fimetaria and Folsomia candida and the results of a ringtest

The collembolans Folsomia fimetaria L. and F. candida Willem are proposed for inclusion in the OECD guidelines for testing of chemicals programme. The present ISO guideline 11267 for the collembolan F. candida has been applied successfully for toxicity testing in soil since its release, but due to the restricted parthenogenetic reproductive biology of F. candida, a collembolan species with a sexual mode of reproduction is needed. The relative F. fimetaria reproduces sexually and fulfils basic requirements for performance and feasibility of laboratory test animals. Each female reproductive instar of F. fimetaria requires the presence of males, of which three instars are usually completed during the standard reproductive tests with F. candida and F. fimetaria.

Extensive experience of the ringtest coordinating laboratory with the F. fimetaria test demonstrates intra-laboratory repeatability during the years 1994 to 1999. Fifty-seven control reproduction data sets tests with F. fimetaria all fulfilled the validity criterion for reproduction while the survival and the CV criteria were not met in 14% and 7% of the tests, respectively.

To provide sufficient information for the adoption of these collembolans for the OECD test guideline programme an international ringtest was initiated in 2005. For both species, test validity is achieved with a mean maximum adult mortality of 20%, a mean minimum reproductive output of 100 juveniles in the controls with a maximum coefficient of variation (CV) of 30%. Due to less experience of the participants in performing the F. fimetaria test, the control reproduction validity criteria were not met in half of the test, while the F. candida test generally was successful.

As the proposed test for F. fimetaria is mechanistically and functionally similar to the previously validated ISO F. candida test method with established performance criteria, the reliabilities of the test methods were compared and the overall test performance was evaluated against a range of criteria: control survival and reproduction and their variability, variability of toxicity endpoints for the model chemicals, the precision of the LC₅₀ and EC₅₀ and intra- and interlaboratory variability. Most data was produced for boric acid, so it was selected for this analysis. Survival was successful for 44% of the F. fimetaria tests and 79% of the F. candida tests. F. fimetaria tests with a valid reproduction had a CV similar to F. candida. A mean control reproduction of 130 and 400 for F. fimetaria and F. candida, respectively, resulted in 43% F. fimetaria tests not meeting the validity criteria, while the F. candida tests were practically all valid. The precision of the EC₅₀ and LC₅₀ estimates were identical for the two species. Stable EC₅₀ and LC₅₀ estimates with low inter-laboratory differences were produced by both tests with no outliers, except for one F. candida LC₅₀ figure, and identical mean and variance for EC₅₀ values with boric acid. Thus, based on the overall inter- and intralaboratory validation results the following validity criteria are proposed for both species in the draft test guideline and should be met in the untreated controls for a test to be considered valid:

Adult mortality should not exceed a mean of 20 % at the end of the tests
An average minimum of 100 juveniles per vessel should be produced during the test
The coefficient of variation of the number of the juveniles per vessel should be less than 30%
The reference compound, boric acid, should cause a 50% decrease in reproduction at 100 mg kg^-1 in an OECD artificial soil substrate with 5% organic matter.

The proposed draft guideline includes comprehensive information and details to successfully perform toxicity testing with either of the two collembolan species. It is especially recommended to employ F. fimetaria for chemicals that are suspected to interfere with any parts of the reproductive biology of sexually reproducing species, while F. candida may be used more generally to assess less specific toxicity.