Probabilistic Assessment of the Biotic Condition of Perennial Streams and Rivers in California

States are required under the Clean Water Act §305(b) to report annually to Congress on the condition of their waterbodies, but satisfaction of this requirement has been confounded by the lack of resources (both technical and financial) to adequately assess these waterbodies. Partly because of this limitation, most regions of the nation have been unable to answer the most basic questions about their water quality, even 35 years after the passage of the act.

To address these constraints, the U.S. EPA established its Environmental Monitoring and Assessment Program (EMAP), a long term research program designed to develop the tools and techniques needed for cost-effectively answering the fundamental status and trends questions in the Clean Water Act. The EMAP studies are based on a probabilistic survey design in which each sampling location represents a known length of stream with known statistical confidence. This design permits the inference of stream conditions for large geographic regions with a relatively small investment in sampling. After completing an assessment of the condition of wadeable streams in the Middle-Atlantic states, the EPA initiated a similar assessment of streams in the western states (WEMAP), which included a high density of sites in California. Now that the first round of WEMAP studies has been completed, California has the ability to include the first statistically defensible condition assessments in its 305b reports.

Because they provide a direct measure of the biotic integrity, bioassessment data are a key component of EMAP water quality monitoring. Bioassessment, the science of describing the ecological condition of waterbodies from the assemblages of organisms they contain, is well established as a valuable tool for water resource management (Karr 1981, Yoder and Rankin 1995, Barbour et al. 1996, Wright et al. 2000, Bailey et al. 2004). Because assemblages of aquatic organisms (e.g., fish, benthic macroinvertebrates (BMIs) and algae) are comprised of taxa that are differentially responsive to different environmental stressors, bioassessments provide a direct means of measuring compliance with the goal of biotic integrity stipulated under the Clean Water Act. Although comprehensive condition assessments will ultimately include assessments of the physical and chemical conditions of waterbodies, condition assessments based on bioassessment data can stand alone as effective measures of the ecological condition of the state’s waterbodies.

There are many different approaches to translating a list of organisms present at a site into an assessment of its ecological condition (Wright et al. 1984, Kerans and Karr 1994, Hawkins et al. 2000, Van Sickle et al. 2005, Ode et al. 2005). We have demonstrated elsewhere (Rehn and Ode 2004, Rehn and Ode 2005) the use of BMI data to produce regional 305(b)-type stream condition assessments using multimetric techniques to calculate site condition scores. These regional assessments were possible because we had previously developed benthic indices of biotic integrity (B-IBIs) for northern and southern coastal California; statewide assessments require a tool for scoring sites that can be applied statewide.

We present here a statewide condition assessment using predictive models based on the River Invertebrate Prediction and Classification System (RIVPACS, Wright 1984). Like multimetric approaches (Kerans and Karr 1994, Ode et al. 2005, Rehn and Ode 2005), predictive modeling techniques establish thresholds of ecological impairment based on a characterization of the biotic assemblages expected to occur under minimal human disturbance (Wright et al. 1984, 1989, 2000). However, predictive models compare assemblages at test sites to an expected taxonomic composition rather than expected metric values. Taxon-based models have seen widespread use since the first BMI models were created in Great Britain in the late 1970s (Norris and Georges 1993, Hawkins et al. 2000, Van Sickle et al. 2005) and have been promoted in the US (Hawkins et al. 2000, Hawkins and Carlisle 2001) as an alternative to the multimetric approach initially endorsed by the EPA (Barbour et al. 1999). For this analysis, we employed newly developed California RIVPACS models (C. Hawkins unpublished) that can be used to score sites throughout the state.