Comparison of staircase and asymmetrical before–after, control–impact (aBACI) experimental designs to test the effectiveness of stream restoration at increasing juvenile steelhead density

Thomas M Loughin,Nicolaas Bouwes,Stephen N Bennett

doi:10.1139/cjfas-2020-0096

Abstract

Before–after, control–impact (BACI) experimental designs are commonly used in large-scale environmental experiments, but these designs can be confounded by location and time interactions. Staircase designs, where replicate treatments are staggered temporally, have been suggested as an alternative to BACI designs. We performed a simulation study based on data from an ongoing watershed-scale restoration experiment within three streams to test the effectiveness of adding large wood to increase habitat complexity and abundance and productivity of juvenile steelhead (Oncorhynchus mykiss). We compared the power of two asymmetric BACI (aBACI) designs with two staircase designs for detecting changes in the density of steelhead (fish·m–2). A staircase design where treatments were temporally staggered in one treatment section in each stream had the highest power and best precision, especially when the innate spatial and temporal variances of steelhead density were large. A traditional BACI performed the worst, and a variation on another BACI and staircase design had intermediate performance. Multistream staircase designs are also more logistically and economically feasible and can maximize learning by replicating experiments across different stream types.

Highlights

Ecosystem experiments using impacts have led to a greater understanding of the influence of management actions on ecosystem processes and biological populations (Carpenter 1990; Hartman et al 1996; Likens et al 1970)
Under the best-case variability, all designs had 100% power to detect a 25% increase in juvenile fish density, so there is no basis for comparisons of the designs in this case, and we focus on expected and worst-case variability scenarios hereinafter
The traditional BACI design (BACI-3) had the lowest power to detect a 25% increase in fish density

Summary

Introduction

Ecosystem experiments using impacts (e.g., perturbations such as logging, addition of nutrients) have led to a greater understanding of the influence of management actions on ecosystem processes and biological populations (Carpenter 1990; Hartman et al 1996; Likens et al 1970). Ecosystem-scale experiments are expensive, difficult to replicate, and require large changes (impacts) to reliably detect a response (Roni et al 2010) Despite these challenges, ecosystem-scale experiments continue to be advocated for determining the effectiveness of management, such as restoration actions, because of the amount of funding being invested (e.g., Katz et al 2007) and the difficulty in determining how habitat and populations change due to natural variability in space and time (Bennett et al 2016; Roni et al 2010). BACI designs are useful when there is a single treated or impacted location and a single control location They have been used to detect large and permanent changes in the mean of a population resulting from an impact (Smith et al 1993). BACI designs become more powerful when more replicates are used, sometimes referred to as Multiple BACI or mBACI (Downes et al 2002). Underwood (1994) argues that asymmetric BACI (aBACI) designs, where a treatment system

Objectives

Methods

Results

Discussion

Conclusion