In the development of the R. abrotiius test protocols, Swartz et al. (1985) observed that 

 with 5 replicates per treatment and 20 amphipods per replicate the test is 75 percent certain 

 of detecting statistical significance (p<0.05) if the difference in mean survival between 

 control and test sediments is 2.8. With the mean control survival of 19.0 that they suggest, 

 this difference corresponds with mean survival of 16.2 or less in test sediment, roughly a 15 

 percent reduction in survival. The relatively high sensitivity of R. abronius survival has 

 been reported in previous inter-method comparisons (Swartz et al, 1979, 1985; Williams et al, 

 1986; Chapman et al, 1987). In an inter-laboratory comparison of the R. abronius toxicity 

 test, Mearns et al. (1986) found that survival greater than 87 percent clearly indicated 

 sediments were not toxic and survival less than 76 percent clearly classified sediments as 

 toxic. An overall mean of 19.2 ±1.1 amphipods survived in control sediments. The minimum 

 mean survival was 15.8 ± 2.4 in a sample from Sinclair Inlet near Bremerton, Washington 

 and 3.8 ± 3.1 in a sample spiked with 12 mg/kg cadmium. Mean emergence in control 

 sediments was 0.4 amphipods out of 20. In that comparison, all five laboratories agreed in 

 the ranking of test samples according to survival, emergence, and reburial end-points. At 

 least four of the five laboratories agreed on mean values for the samples for the three end- 

 points. Based upon the results of these two evaluations of the methods, mean survival in 

 test sediments of 15.0 or fewer amphipods may be used as a conservative criterion for 

 classifying a sample as "toxic." 



Out of 129 sediment samples from the industrial waterways of Commencement Bay, 

 Washington, 92 (71.3 percent) were "toxic," i.e. 15 or fewer survivors (Swartz et al, 1982). 

 Using the same criterion, 6 of 8 samples (75%) were toxic in inner Everett Harbor, 

 Washington; 10 of 17 samples (59%) were toxic in Sinclair Inlet, Washington; and 5 of 10 

 samples (50%) were toxic off a major combined sewer overflow near Seattle (Battelle, 1986). 

 None of the 15 samples collected at sites on the Palos Verdes shelf and in Santa Monica 

 Bay, California resulted in 15 or fewer survivors (Swartz et al, 1986). Percent survival in 

 sediments collected in 1984 from the Commencement Bay waterways was 25 percent or less in 

 23 of 60 samples (38.3%) (Williams et al, 1986). There were no survivors in a sample taken 

 near a defunct metal smelter. Low survival of amphipods has been observed in samples from 

 parts of San Francisco Bay (Long et al, 1988). The lowest mean survival rate (34.5%) in two 

 deep cores taken off Hunters Point exceeded the mean (47%) observed in 10 samples from 

 Islais Waterway. A grand mean survival of 11.0 percent was observed in 26 samples from 

 South Bay tested with R. abronius (Baumgartner, unpublished manuscript). By comparison 

 with the results from other studies, mean survival in the present study exceeded 75 percent 

 at only 3 of the 15 stations sampled; therefore, 12 of 15 were "toxic," using the criterion 

 stated above. 



The average CV reported here (21.4%) was exactly the same as that for data from nine 

 samples reported by Chapman et al. (1987) and very similar to that (22.4%) for data from 

 seven samples reported by Mearns et al. (1986). Also, the high analytical variance (average 

 CV of 83.5%) for the avoidance end-point observed in this evaluation has been observed 

 elsewhere: 65.7 percent in nine samples (Chapman et al, 1987); and 111.2 percent in seven 

 samples (Mearns et al, 1986). 



Despite the correlations observed by Swartz et al. (1985) between sediment avoidance by 

 R. abronius and the cadmium concentrations and the relatively good agreement among five 

 laboratories in this end-point (Mearns et al, 1986), the avoidance end-point in this 

 evaluation performed relatively poorly. Within-sample variability was very high, 

 between-sample discriminatory power was low, the data did not correlate very highly with 

 chemical concentrations, nor did they correlate with the data from the other tests. Also, 

 the reburial end-point was not responsive at all, corroborating the observation of Swartz et 

 al (1985) that failure to rebury was very rare among survivors of the 10-d tests. 



The negative correlations between R. abronius survival and both percent clay and TOC 

 content corroborates this relationship demonstrated quantitatively in empirical experiments 

 conducted by DeWitt et al. (1988). R. abronius normally occurs in well-sorted, fine sands and 

 usually not in muddier sediments. Therefore, some degree of mortality observed in toxicity 

 tests is probably attributable to the presence of fine-grained sediments when such samples 

 are tested DeWitt et al. (1988) examined the possible role of particle size on R. abronius 



81 



