Precision of benthic macroinvertebrate indicators of stream condition in Montana

The Montana Department of Environmental Quality (MDEQ) uses 2 forms of benthic macroinvertebrate indicators for detection of stream impairment, a multimetric index (MMI) and a predictive model of observed to expected taxa (O/E), each of which is calibrated to streams across the state. As part of the routine monitoring program, some sample locations were subjected to repeated sampling, i.e., multiple samples were collected from stream reaches in spatial and temporal proximity. Results from repeated sampling allow calculation of precision estimates, which are important for describing a portion of the uncertainty (systematic error) associated with field sampling and site assessments. In this project, we evaluated 131 and 77 repeated-sample pairs for the MMI and O/E, respectively, using 4 different measures of precision: coefficient of variability (CV), 90% confidence intervals, relative % difference (RPD), and % difference for the final assessments. MMI and O/E had similar consistency and repeatability. Segregating the data set and calculations by region or field method yielded generally similar precision estimates for the indicators, although precision was slightly better in the mountains using the Hess field-sampling method than in other regions or with other field methods. Evaluation of RPD showed that assessments (impaired/nonimpaired) on the basis of the MMI differed between samples in 18.3% of repeated-sample pairs and assessments on the basis of O/E differed between samples for 19.5% of repeated-sample pairs. Recommended measurement quality objectives were 10 to 15% for CV and 15 to 20% for RPD for both indicators. Field-sampling precision was the focus of our paper, but we emphasize that detecting the presence of stressors or degraded conditions is the primary objective of the MDEQ stream condition indicators.