EFFECTS OF ORGANISM ALLOCATION ON TOXICITY TEST RESULTS

Abstract

The purpose of this study was to investigate the effect of nonrandom allocation of organisms to treatments in aquatic toxicity tests on lethal concentration estimates. Random allocation of organisms is prescribed to avoid the introduction of systematic bias in these tests; however, completely random assignment is usually impractical. A source of systematic bias in a nonrandom allocation scheme is the potential inverse relationship between organism hardiness and the order in which organisms are captured for assigning to treatment (i.e., less hardy and hence more sensitive individuals are captured first). A Monte Carlo simulation experiment was performed to examine such an inverse relationship for a variety of random allocation schemes, focusing on the mortality endpoint in the fathead minnow larval survival and growth toxicity test. Results of the computer simulation experiment indicated that nonrandom allocation of organisms can produce significant bias in estimates of lethal concentration values. Any attempt at randomization resulted in approximately unbiased estimates of lethal concentration values; however, variance of the estimates increased as allocation schemes deviated from the completely random allocation. This study illustrates the importance of proper training of laboratory personnel to implement some form of random organism allocation. Furthermore, many partial random allocation schemes (i.e., those allocating more than one organism at a time) may still yield unbiased estimates of potency endpoints.