The Effects of Sampling Design on Stream Habitat Characteristics and Habitat-Based Salmon and Steelhead Production Estimates

Accurately estimating stream channel characteristics is essential for managing and restoring  populations and aquatic ecosystems.  Likewise, there is a need to understand the tradeoffs between bias and precision when choosing a sampling design, particularly given limited resources available for assessments and monitoring. Reach-based sampling designs have been used extensively to collect fisheries related data; however, few studies have examined how reach based inferences may be biased, a particular concern given the non-random distribution of factors such as woody debris and the magnitude of site-to-site variability.  Here, we used continuous habitat surveys to census stream attributes in tributaries in the upper Lewis River, WA. We delineated the surveys into reaches and integrated our data with the EDT model for reach-based estimates of salmon and steelhead production.  We then bootstrapped our data to evaluate potential outcomes of different sampling designs including simple random, stratified random, and a generalized random tessellation stratified design and to identify potential bias and precision using sampling designs.  Our results indicate the extent of bias and levels of precision varied across habitat metrics.  Our analyses suggest the use of reach-based approaches, particularly with low sampling efforts can result in different estimates of habitat characteristics and juvenile fish production.