How Do Spatial and Temporal Scales of Post‐Restoration Studies Affect Conclusions About the Effectiveness of Habitat Restoration?

ABSTRACT Evaluating biological responses in fish following restoration is a complex undertaking with myriad possible approaches and methodologies. Nevertheless, results vary greatly depending on the species studied and the temporal (single‐year vs. multi‐year data and time since restoration) and spatial (habitat vs. whole‐valley segment) scale of post‐restoration studies. The Entiat River, Washington, USA is a system in which fish habitat restoration is coupled with intensive post‐restoration monitoring to evaluate restoration efficacy. Extensive previous research on the response of juvenile Chinook salmon and steelhead to restoration allows for (1) comparisons of the results of continuing studies over multiple years, and (2) the analysis of how all results are influenced by the spatial and temporal scales of the study. From a set of study reaches restored by addition of Engineered Log Jams (ELJs) and unrestored control reaches, we surveyed fish abundance, replicating the earlier methods. We then examined the results for both species by comparing results from methods applied to single years of data with those applied to multiple years of data. To compare data from habitat units (~10 m stream length), we used Bayesian Generalized Linear Mixed Models (GLMMs) to compare salmon density in ELJ‐treated habitat units with that in untreated habitat units. We added restoration age as a model parameter to understand how density differences between treated and untreated habitat units change over time. We used model selection to test the models with the age parameter against models that considered habitat restoration only. GLMM results showed that a positive restoration response was detectable for Chinook independent of the temporal scale. For steelhead, 1‐year and multi‐year results were consistent in the lower valley segment of the river but only with data collected within a few years after restoration. Data collected in 2022 did not confirm previous results. Time since restoration had a variable effect on the observed density difference between treated and untreated habitat units but this depended on where in the river basin the reaches occurred. For steelhead, older ELJs sometimes showed a smaller effect and sometimes a larger effect. For Chinook, the difference was larger at older ELJs but models excluding a time since restoration term were a better fit to the data. Total steelhead density decreased with time, which may explain cases with a lack of restoration responses. A single year of data analysis can show results that are consistent with those shown by multiple years of data for Chinook salmon on the habitat scale. Steelhead results are more variable, which is also consistent with previous findings. Species that respond differently to older restoration projects might show unexpected longer‐term results.