Variation, replication, and power analysis of Myriophyllum spp. microcosm toxicity data

Myriophyllum spp. have been proposed as a new standard laboratory aquatic macrophyte test species for the registration of pesticides. The main objectives of this investigation were to determine the power of Myriophyllum sibiricum and Myriophyllum spicatum toxicity data derived from an outdoor microcosm bioassay, to evaluate the variation of 10 different aquatic plant endpoints and to calculate the minimum detectable difference for these endpoints, to determine the replication required to detect ecologically significant changes from control for these endpoints, and to make recommendations for future field studies with Myriophyllum spp. Control data from four different studies that characterized haloacetic acid toxicity with Myriophyllum spp. for durations of three to six weeks during the summer of 1999 with five treatment levels (n = 3), including control, were examined. Endpoint coefficient of variation ranged, on average, from 6 to 28%. Node number and plant length endpoints were consistently the most statistically powerful for both plant species. It was possible to detect approximately 30% change from control in both endpoints with high statistical power (beta = 0.2, alpha = 0.05, n = 3). The range of minimum detectable differences was 40 to 60% for the other endpoints. Replication to detect a > or = 25% change from control would require an n of 2 to 21, depending on the endpoint. Myriophyllum sibiricum had slightly lower coefficients of variation and thus required fewer replicates than M. spicatum to be statistically significantly different from control values. Variation within microcosm studies was not significantly different from that of controlled laboratory studies, implying that most of the variation observed in the field is inherent in the plants. Based on statistical sensitivity, ecological relevance, and toxicological sensitivity, we recommend using plant length and root endpoints as indicators of toxicity under field conditions.