fish. All the tests under evaluation had been developed and tested to varying degrees in 

 laboratory and field research. 



Field sampling logistics, sample handling, and data analyses were performed 

 separately for the work with fish and sediments, though some of the sites overlapped. 

 Sampling sites were selected to represent a range in contamination from relatively highly 

 contaminated conditions to pristine. Data from previous studies, including those by the 

 NS&T Program (e.g., NOAA, 1987), were used to select the sampling sites. 



In an attempt to reduce costs, as many sediment sampling sites as possible that are 

 routinely sampled in the NS&T Program annually were used in the evaluation. As a result, 

 the sediments were sampled using the NS&T Program protocol: three stations sampled at 

 each site. Four sites from the NS&T Program were sampled in the San Francisco Bay area: 

 Yerba Buena Island (YB), San Pablo Bay (SP), Vallejo (VA), and Tomales Bay (TB). Three 

 samples (numbers 1, 2, and 3) from the Oakland Inner Harbor (near 37°47'11"N, 

 122°14'50"W) were expected to be relatively highly contaminated. Three samples each 

 collected near Yerba Buena Island (numbers 4, 5, and 6; near 37°50T6"N, 122°20"W), near 

 Vallejo (numbers 7, 8, and 9; near 38°04'10"N, 122°14'17"W), and in southwestern San Pablo 

 Bay (numbers 10, 11, and 12; near 38°01'35"N, 122°25'36"W) were expected to be moderately 

 contaminated. Three samples (numbers 13, 14, and 15) from Tomales Bay (near 38°09'02"N, 

 122°53'55"W), a remote embayment located north of San Francisco Bay, were expected to be 

 minimally contaminated. The locations of the sites are illustrated in Figure 1. In the data 

 evaluation, the 15 samples were treated either as (1) 15 independent sampling stations or (2) 

 three replicates of each of the five sites in accordance with NS&T Program sampling 

 protocols. Tests of samples from the respective animal collection sites were also performed 

 concurrently with each toxicity test and treated as laboratory controls. Since the 

 participating laboratories were scattered, the same material was not used as controls for all 

 the toxicity tests. This disparity is recognized as a weakness in the study design, especially 

 since none of the control samples was analyzed for chemical concentrations. Nevertheless, 

 the controls served as independent test media for evaluating the viability and internal 

 consistency of the test organisms and for determining which test samples were "toxic", i.e., 

 significantly different from respective controls. Five separate grab samples were collected 

 at each station for the benthic community analyses. Finally, an independent survey of 69 

 sites in the San Francisco Bay estuary was performed, using a sediment profiling camera. 



The fish sampling sites were: in the Oakland Outer Harbor (OK), off Berkeley (BK), 

 in San Pablo Bay (SP), off Vallejo (VJ), off the mouth of the Russian River (RR) (reference 

 site), and off Santa Cruz (SC) in Monterey Bay. The former two (OK, BK) were expected to 

 be relatively highly contaminated, VJ was expected to be moderately contaminated, and RR, 

 SP, and SC were expected to be minimally contaminated reference sites. The species selected 

 for the fish analyses was the Platichthys stellatus. It had been the subject of extensive 

 research on availability, contamination, and measures of biological effects in the Bay (Spies 

 and Rice, 1988). 



All analyses were performed "blind," i.e., without knowledge of the station or site from 

 which the samples were collected. Following the calculation of mean values for each 

 station or site for each measure, the data were subjected to a variety of statistical 

 procedures to comparatively evaluate the performance of the biological measures. 



All measures of biological effects performed with feral organisms or complex 

 environmental samples (e.g., sediments) are subject to several sources of variability. Each 

 species has different degrees of tolerance to contaminants and natural factors. Also, the 

 tolerance of individual organisms varies depending upon age, condition, natural sources of 

 stress, genotype, etc. Chemical analyses of environmental samples do not provide 

 information for all chemicals that are potentially toxic. They do not indicate which 

 chemicals or portions of chemical concentrations are bioavailable. Good reliable measures of 

 effects are needed to stand alone as independent indicators of biologically significant stress 

 without the benefit of or confusion with synoptic chemical data. Nevertheless, the most 



