Development of the Asotin Creek Intensively Monitored Watershed Project with Specific Emphasis on Experimental Design and Implementation Considerations

The effectiveness of past anadromous stream habitat restoration actions has been hard to determine because few restoration projects were implemented in an experimental fashion, or were not large enough to produce a detectable effect size. Recently, a series of Intensively Monitored Watersheds (IMWs) have been established with the explicit intent to determine the population response of anadromous salmonids to watershed scale restoration actions. Paramount to the successful development and implementation of an IMW is a robust and logistically feasible experimental design. Optimal experimental designs seek to account for the significant sources of variability while achieving large enough effect size. Because replication at a watershed scale is often not logistically feasible, before-after-control-impact (BACI) designs are often used to test the effect of restoration treatments.  However, implementation of a large treatment within a single year can also be logistically infeasible, and the analysis results of a BACI design can be ambiguous due to failure to account for the random effects of year-treatment interactions. An alternative to BACI designs is a staircase design whereby treatment applications are staggered over time. We use historic steelhead monitoring data (redd counts and juvenile abundance estimates) and data from an ongoing IMW in the Asotin Creek in southeast Washington to assess the power of different experimental designs to detect treatment responses. The Asotin IMW monitors summer steelhead populations and stream habitat in three tributaries in Asotin Creek. Each tributary has three 4 km long sections that are subsampled with 500 m reaches. A series of simulations were used to compare the power of the staircase design to a BACI design, and variations of a simple staircase design. The simulation results demonstrate that the power of different experimental designs is strongly influenced by sources of variance within sections of streams, between sections within the same stream, and between streams. All designs we tested were able to detect a 25% change in juvenile steelhead abundance when low-to-moderate variances were assumed.  However, an alternative staircase design that treats one section of each tributary in a staggered fashion had greater power (60-70%) to detect a 25% change than all the other designs (25-30% power) when variability was high. We show how these simulations can inform the selection of appropriate experimental designs to detect population level responses to restoration actions.