Field and laboratory performance characteristics of a new protocol for sampling riverine macroinvertebrate assemblages

Measurement and estimation of performance characteristics (i.e. precision, bias, performance range, interferences and sensitivity) are often neglected in the development and use of biological sampling methods. However, knowledge of this information is critical in enabling potential users to assess data quality and make comparisons among different sampling methods. In this study, the performance characteristics were evaluated for both the field and laboratory components of a new large river macroinvertebrate bioassessment protocol (mLR‐BP) for non‐wadeable streams. We sampled 19 sites across two depth classes, collecting three replicate samples at each site and sorting three 300‐organism subsamples from each sample. The replicate samples provided data for estimates of precision in the laboratory and field, and abiotic variables allowed for measurements of overall sensitivity. Precision and performance range differed between shallow and deep sites, particularly for the field component. As compared with precision measured in other studies of bioassessment methods, the field component of the mLR‐BP performed similarly, particularly in shallow sites. Based on the measures of combined field and laboratory sensitivity, this protocol should be able to detect differences of approximately 20–25% in the metrics evaluated in this study, if used for bioassessment in similar types of rivers. With all sites and the field and laboratory components combined, metrics were most responsive to a gradient of urban land cover but also showed some relationship with agricultural land cover. However, metric responsiveness does not necessarily correlate with precision, and metric selection can influence the performance characteristics of the method. Overall, the sampling protocol shows great utility for bioassessment and monitoring of non‐wadeable rivers, as well as for measuring the success of restoration efforts. In addition, the design of this study provides a template for estimating performance characteristics in other non‐wadeable systems. Copyright © 2008 John Wiley & Sons, Ltd.