Interlaboratory variability of amphipod sediment toxicity tests in a cooperative regional monitoring program

Marine sediment toxicity tests are widely applied in monitoring programs, yet relatively little is known about the comparability of data from different laboratories. The need for comparability information is increased in cooperative monitoring programs, where multiple laboratories (often with variable skill levels) perform toxicity tests. An interlaboratory comparison exercise was conducted among seven laboratories in order to document the comparability of sediment toxicity measurements during the Bight'98 regional sediment survey in southern California. Sediments from four stations in Los Angeles and Long Beach Harbors were tested using a 10-day survival test of the amphipod Eohaustorius estuarius. All laboratories successfully performed the sediment test and associated reference toxicant test. Statistically significant differences were found in mean amphipod survival rates among some laboratories for the field-collected sediments, but there was little evidence of a consistent bias among laboratories. Although the reference toxicant test indicated a five-fold variation in test sensitivity among laboratories, these results were not accurate predictors of interlaboratory performance for the sediment tests. The laboratories demonstrated excellent concordance (Kendall's W = 0.91) in ranking the field-collected sediments by toxicity. Agreement on classifying the sediments into categories (nontoxic, moderately toxic, and highly toxic) based upon the percent of survival was best for highly toxic sediments. An analysis of test precision based upon the variance among replicates within a test indicated that the measured survival rate for a sample may vary by up to 12 percentage points from the actual response.