A field validation of two sediment-amphipod toxicity tests.

A field validation study of two sediment-amphipod toxicity tests was conducted using sediment samples collected subtidally in the vicinity of a polycyclic aromatic hydrocarbon (PAH)-contaminated Superfund site in Elliott Bay (WA, USA). Sediment samples were collected at 30 stations with a 0.1 m2 grab from which subsamples were taken for sediment toxicity testing and geochemical and macrofaunal analyses. Standard 10-d sediment-amphipod toxicity tests were conducted with Rhepoxynius abronius and Leptocheiros plumulosus. Sediments were analyzed for 33 PAHs, pentachlorophenol, polychlorinated biphenyls, acid-volatile sulfide, simultaneously extracted metals (Cd, Cu, Zn, Pb, Ni), total organic carbon, and grain size. Sediment temperature, oxygen-reduction potential, water depth, and interstitial water salinity were also measured. Polycyclic aromatic hydrocarbons, quantified as total PAH toxic units (TU(PAH)), were confirmed to be an important common causal agent of the changes in the two toxicity test (% survival R. abronius, % survival L. plumulosus) and five macrofaunal community (number of species, S; numerical abundance, A: total biomass, B: Swartz's dominance index, SDI; Brillouin's index, H) endpoints. Two other macrofaunal community metrics (the complement of Simpson's index, 1 - SI, and McIntosh's index, MI) were less sensitive to TU(PAH) than the two toxicity test endpoints. The sensitivities of R. abronius and L. plumulosus to TU(PAH) were statistically indistinguishable. Field validations were conducted by testing the association between or among each toxicity test endpoint, each of seven macrofaunal community metrics (S, A, B, SDI, H, 1 - SI, MI), and TU(PAH) by (1) Spearman's coefficient of rank correlation, (2) Kendall's coefficient of concordance, (3) G tests of independence, and (4) regression analysis. Some field validations based on multivariable tests of association (e.g., points 2 and 3) among toxicity test, field, and stressor endpoints produced false positive results. Both toxicity test endpoints were validated as indicators of changes in S, A, SDI, and H by all the methods tested. The resolution power of the relationships between the laboratory toxicity test and macrofaunal field endpoints was low (< or = three classes) but sufficient to discriminate ecologically important effects. We conclude that standard sediment-amphipod toxicity tests are ecologically relevant and that, under the proper conditions, their results can be used for lab-to-field extrapolation.