Evaluating Consistency of Best Professional Judgment in the Application of a Multiple Lines of Evidence Sediment Quality Triad Steven Bay,*À Walter Berry,` Peter M Chapman,§ Russell Fairey,// Tom Gries,# Edward Long,ÀÀ Don MacDonald,`` and Stephen B WeisbergÀ ÀSouthern California Coastal Water Research Project, 3535 Harbor Boulevard, Suite 110, Costa Mesa, California 92626, USA `US Environmental Protection Agency, Narragansett, Rhode Island 02882 §Golder Associates, 195 Pemberton Avenue, North Vancouver, British Columbia V7P 2R4, Canada //Moss Landing Marine Laboratories, Moss Landing, California 95039, USA #Washington Department of Ecology, 300 Desmond Drive, Lacey, Washington 98503, USA ÀÀERL Environmental, 3691 Cole Road, South Salem, Oregon 97306, USA ``MESL, 24-4800 Island Highway North, Nanaimo, British Columbia V9T 1W6, Canada (Received 11 January 2007; Accepted 20 May 2007) ABSTRACT The bioavailability of sediment-associated contaminants is poorly understood. Often, a triad of chemical concentration measurements, laboratory sediment toxicity tests, and benthic infaunal community condition is used to assess whether contaminants are present at levels of ecological concern. Integration of these 3 lines of evidence is typically based on best professional judgment by experts; however, the level of consistency among expert approach and interpretation has not been determined. In this study, we compared the assessments of 6 experts who were independently provided data from 25 California embayment sites and asked to rank the relative condition of each site from best to worst. The experts were also asked to place each site into 1 of 6 predetermined categories of absolute condition. We provided no guidance regarding assessment approach or interpretation of supplied data. The relative ranking of the sites was highly correlated among the experts, with an average correlation coefficient of 0.92. Although the experts’ relative rankings were highly correlated, the categorical assessments were much less consistent, with only 1 site out of 25 assigned to the same absolute condition category by all 6 experts. Most of the observed categorical differences were small in magnitude and involved the weighting of different lines of evidence in individual assessment approaches, rather than interpretation of signals within a line of evidence. We attribute categorical differences to the experts’ use of individual best professional judgment and consider these differences to be indicative of potential uncertainty in the evaluation of sediment quality. The results of our study suggest that specifying key aspects of the assessment approach a priori and aligning the approach to the study objectives can reduce this uncertainty. Keywords: Sediment quality triad Contaminated sediments Best professional judgment Uncertainty INTRODUCTION Sediment is composed of a complex matrix of constituents that makes interpretation of chemical contamination data challenging. Bulk measures of chemical concentration fail to differentiate between the fraction that is tightly bound to sediment and the fraction available for transport across biological membranes via interstitial water. Furthermore, some benthic organisms ingest sediment and can uptake chemicals sorbed onto particles. Thus, even measurement approaches that differentiate interstitial water chemical concentrations, such as equilibrium partitioning models or direct measurement of porewater chemistry, do not fully describe chemical bioavailability in the sediment (Wenning et al. 2005). Consequently, assessments of sediment quality conditions are often conducted by augmenting chemical measurements with toxicity tests, measures of benthic infaunal condition, or both. Chemical measurements can be enhanced by toxicity tests that integrate the effects of multiple contaminants. However, toxicity tests are typically conducted under laboratory conditions and use species that might not occur naturally at the test site, making it difficult to interpret ecological significance of the results when used alone. Benthic community condition is a good ecological indicator because benthic animals readily exhibit the effects of sediment contamination. Conversely, the use of benthic community condition alone is problematic because the benthos is potentially affected by a diverse battery of noncontaminant variables, including depth, texture, organic carbon content, salinity, dissolved oxygen concentration, currents, tides, and physical habitat disturbances. Benthic conditions are also affected by biotic interactions, such as predation and competition. For these reasons, benthic communities are naturally highly variable. Habitat measures are often combined into a multiple lines of evidence (MLOEs) triad that integrates exposure and effect to assess chemical concentration levels in terms of biological concern (Long and Chapman 1985). Presently, no single, universally accepted method for interpreting triad data and classification of sediments on the basis of an MLOE approach exists (Chapman et al. 2002; Long and Sloane 2005; Wenning et al. 2005). Each regulatory or monitoring program uses an * To whom correspondence may be addressed: steveb@sccwrp.org Published on the Web 6/12/2007. Integrated Environmental Assessment and Management — Volume 3, Number 4—pp. 491–497 Ó 2007 SETAC 491 Original Research Review