Experimental design for estimation of high lethal doses
by
Rosemary A. Kopittke
Queensland Department of Primary Industries
Coauthors: Anthony J. Swain (Queensland Department of Primary Industries)

Dose mortality regressions are commonly used in analysing results of bioassay experiments.

The toxicity of a treatment against several stages of an insect is typically compared by testing the equality of the regressions – to select the most tolerant stage. It is also standard practice to estimate lethal doses (doses required to produce a certain percentage mortality) and to compare the stages by examination of the confidence intervals for the lethal doses.

The traditional selection of doses has been based on Finney (1971). These guidelines, however, are designed to provide the best estimate of LD50. They are not generally applicable to disinfestation work where much higher LDs are estimated to meet export specifications and to ensure that the correct stage of a pest is targeted as the most tolerant.

Robertson et al (1984) examined by simulation a range of sample sizes and experimental designs in order to determine which is most appropriate for estimation of LD50 and LD90 using the logit model.

This work extends that of Robertson et al to higher lethal doses and to the complementary log-log model which has been used in preference to the logit model by many researchers.

Finney, D.J. (1971) Probit analysis. 3rd ed. Cambridge University Press, Cambridge.

Robertson, J.L., Smith, K.C., Savin, N.E. and Lavigne, R.J. (1984) Effects of dose selection and sample size on the precision of lethal dose estimates in Dose-mortality regression. Journal of Economic Entomology 77:833-837.

Date received: August 27, 2002

Copyright © 2002 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cajn-06.