Based upon all of the data from both phases and all four test end-points, there are several patterns in 

 toxicity in the study area. First, toxicity was very high in the upper and lower East River samples, and 

 generally (but not consistently) diminished eastward into western Long Island Sound and southward 

 into the New York Harbor. Second, toxicity was relatively high in the lower Passaic River, Newark 

 Bay, and Arthur Kill, gradually diminished somewhat into western Raritan Bay, and diminished addi- 

 tionally into north-central Raritan Bay and the mouth of the estuary. Third, toxicity was high in the 

 inner portions of Sandy Hook Bay, diminished into lower New York Harbor, the outer bay, and the 

 mouth of the estuary. 



Areas in which the sediments were toxic in the present survey also were toxic in one or more historical 

 surveys in which sediments were tested with either amphipods or nematodes (Tietjen and Lee, 1984; 

 Schimmel et al., 1994; Scott et al., 1990; Brosnan and O'Shea, 1993; Aqua Survey, 1990a, 1990b; 

 Tatemetal., 1991). These areas included the lower East River, Newark Bay, Arthur Kill, Kill van Kull, 

 lower Passaic River, western Raritan Bay, and Sandy Hook Bay. 



Correlations Among Toxicity Tests. It was apparent from these data that the four test end-points did 

 not always agree on the relative toxicity of all the samples. While some stations and sites were toxic in 

 more than one test, there were many cases where the tests did not agree as to which samples were toxic. 

 Table 39 summarizes the Spearman-rank correlations (Rho) among the four test end-points. All except 

 the correlations between amphipod survival and bivalve normal development were significant. The 

 strongest relationship, as expected, was between normal development and survival of the bivalve lar- 

 vae (Rho = 0.741, p<0.0001). The results of the Microtox test were correlated with the results of the 

 three other tests. These data indicate that, while the four tests suggested different patterns in toxicity, 

 they did overlap to a significant degree. 



These correlation coefficients illustrate the advantage of determining toxicity with a battery of tests 

 and end-points. The spatial pattern of toxicity indicated with one assay may not necessarily represent 

 patterns in toxicity to other organisms and/or end-points. The study area has many different sources of 

 contamination, the mixtures and concentrations of contaminants differ remarkably from place to place, 

 and the relative bioavailability of the different chemicals probably varies spatially. The different tox- 

 icity tests, therefore, would be expected to differ spatially in their responses to the various sources of 

 contamination. 



Table 39. Spearman rank correlation coefficients (Rho) for the four toxicity test end-points tested 

 in Phase 1 as percent of controls (n=117). 



*p<0.05, **p<0.01, ***p<0.001 



121 



