|HOME SCHEDULE AUTHOR INDEX SUBJECT INDEX|
PM08 Sediment Quality Assessment
(PM108) Quantile regression − another tool for examining the predictive ability of sediment quality guidelines.
Preziosi, D.1, Williams, L.2, 1 Integral Consulting, Inc., Annapolis, MD, USA2 Integral Consulting, Inc., Seattle, WA, USA
ABSTRACT- One approach used in the development of numerical sediment quality guidelines (SQGs) is to derive single-point predictive estimates of adverse effects to sediment-dwelling organisms based on a match between synoptically collected sediment bulk chemistry and sediment toxicity test data. Recently, a number of statistical approaches (e.g., logistic regression, receiver operating characteristic curves) have been explored as alternate approaches to the single-point estimate approach. These alternate statistical approaches permit a more complete examination of the full distribution of possible effect levels and their attendant probabilities rather than simply relying on a single-point estimate on the response curve. As a follow-on to these approaches, we explore the use of quantile regression to evaluate potential relationships between sediment chemistry and toxicity data. Quantile regression uses various portions of the response curve to evaluate the presence/strengths of relationships between effect concentrations and incidence of toxicity. This approach is gaining increasing attention in the field of ecology because it acknowledges that all the factors that affect ecological processes are not measured and included in the statistical models used to investigate relationships. We present an example of how quantile regression may be used to examine matched sediment chemistry and toxicity data, and discuss possible advantages offered by this approach for addressing heterogeneous variances and confounding factors in sediment data and biological responses. Ultimately, quantile regression models may provide information to risk assessors and managers beyond that afforded by single-point predictive estimates by offering insight on the probability, strength, and reliability of SQGs.
Key words: toxicity, sediment, quantile, regression
Internet Services provided by|
Allen Press, Inc. | 810 E. 10th St. | Lawrence, Kansas 66044 USA
e-mail email@example.com | Web www.allenpress.com
All content is Copyright © 2004 SETAC