Which value should be used as the lethal concentration 50 (LC(50)) with bacteria?

Abstract

Computations of lethal concentration 50 (LC(50)) of a data-set of a toxicity study on an herbicide against a cyanobacterium were performed by general linear regression, Spearman-Karber and probit transformation methods, for evaluation of the methods used. It is shown that the linear regression method yields some faulty LC(50) value, while both of Spearman-Karber and probit methods yield similar and statistically respectable LC(50) values. In the Spearman-Karber method, a prerequisite of some uniform dose-interval of test-chemical and tiring calculations were involved. But, the probit method is less tiring and additionally computed LC(25) and LC(75) values help assess the solicited accuracy of the LC(50) value and other test-statistics, including minimum inhibitory concentration (MIC), highest-permissive concentration (HPC), and a few more with respect to lethal concentration 100 (LC(100)) without prerequisite of any uniform dose-interval of test-chemical. Further, the redundancy of computations of standard error (SE) and 95% confidence limits (CL) of the LC(50) value is suggested, as CL values are so wide to spoil LC(50) accuracy that is solicited in toxicology.