The influence of sampling and taxonomic effort on the accuracy of biological water quality assessments
The enactment of the Clean Water Act in 1972 required that states assess and report water quality conditions biennially to the U.S. Environmental Protection Agency. Rapid bioassessment (RBA) methods were developed to identify impaired streams in a timely and cost efficient manner. With disagreements regarding the accuracy of rapid protocols and the perceived need to standardize across programs, states began to implement more expensive and complex protocols. In 2000, California established the Surface Water Ambient Monitoring Program (SWAMP), a standardized, statewide monitoring program that requires the collection, assessment and interpretation of samples at an expert/professional level, adding significant amounts of time and costs to the protocol. The objectives of this study were to determine (1) if a simpler, modified RBA method produced a similar classification of impairment as the more resource-intensive SWAMP method, and (2) if identification of macroinvertebrates in the laboratory improved the accuracy of the field-based modified RBA method. I hypothesize that the modified RBA method will produce similar classifications of impairment as the SWAMP method while lowering time and costs of the entire assessment process. I performed a modified RBA method at 12 sites on urban streams in the eastern San Francisco Bay Area that had previously been sampled with the SWAMP method, and identified macroinvertebrates in the field (mField) and in the lab (mLab). Candidate biological metrics were screened against a human disturbance index (HDI) I developed that incorporated water quality conditions, an assessment of physical habitat, and a GIS analysis of land use. Four metrics were included in each of the three Indices of Biotic Integrity (hereafter, IBI). There was no significant difference in IBI scores among the three methods. IBI scores were strongly correlated between mField and mLab and moderately correlated between the modified RBA protocols and the SWAMP protocol. Condition classifications were very similar between the SWAMP and mLab IBIs. The mField IBI produced the most similar condition classifications (i.e., good, fair, bad) to the HDI. The results of this study support my hypothesis that the modified RBA methods produce similar classifications of impairment as the SWAMP method while lowering time and costs of the entire assessment process.