You are here:
Using a RIVPACS model to predict expected macrofaunal species richness in Puget Sound
Citation:
LEE, II, H., M. R. FRAZIER, D. REUSSER, W. G. NELSON, M. Dutch, AND K. Welch. Using a RIVPACS model to predict expected macrofaunal species richness in Puget Sound. Presented at Marine Bioinvasions Conference, Portland, OR, August 24 - 27, 2009.
Impact/Purpose:
As part of a project to develop regional indicators for Pacific coastal environments using soft-bottom benthic species, we are evaluating a RIVPACS predictive model (River InVertebrate Prediction and Classification System).
Description:
As part of a project to develop regional indicators for Pacific coastal environments using soft-bottom benthic species, we are evaluating a RIVPACS predictive model (River InVertebrate Prediction and Classification System). This approach was originally developed for rivers and streams to predict the expected number of benthic macrofauna taxa (E) at a site assuming minimal disturbance (i.e., reference condition). A significant difference between the observed (O) and expected (E) number of taxa indicates deviation from reference conditions. The RIVPACS approach was recently found to be effective in assessing estuarine sediment conditions in Southern California. Our objective was to evaluate this approach in other marine environments. The initial RIVPACS model included > 4,300 benthic samples from the entire US west coast from northern Mexico to northern Washington State, and the data was compiled from several studies and agencies including EPA’s Environmental Monitoring and Assessment Program, WA Dept. of Ecology, and SCCWRP. Because this model performed poorly in some areas, we are now testing RIVPACS within ecoregions and biomes (i.e. estuarine vs oceanic), starting with Puget Sound (N = 1888). Model performance can be evaluated by the standard deviation of O/E (SDOE) for reference sites (lower SDOE’s reflect better model prediction of reference conditions) and the degree of agreement with best professional judgment (BPJ). The SDOE for Puget Sound reference samples was ~0.24, which is slightly higher than the 0.2 guidelines for freshwater systems, but appears acceptable for marine environments. The sites classified as disturbed by RIVPACS (operationally defined as O/E <0.75) were mostly classified as disturbed by BPJ (>80% agreement); however, BPJ classified many more sites as disturbed (63% BPJ samples vs. 38% RIVPACS). We are now exploring ensemble models which combine RIVPACS and other indicators, such as percent sensitive/tolerant species.