Landscape attributes explain salmonid ecological neighbourhoods across a complex river network. (26th July 2021)

• Record Type: Journal Article
• Authors: Walther, Eric J.; Zimmerman, Mara S.; Westley, Peter A. H.
• Is Part Of: Freshwater biology. Volume 66:Number 10(2021)
• Page Start: 1930
• Page End: 1948
• Publication Date: 2021-07-26
• Subjects: aquatic habitat -- connectivity -- distribution models -- Pacific salmon -- restoration planning; Freshwater biology -- Periodicals; Biologie d'eau douce -- Périodiques; 577.605
• DOI: 10.1111/fwb.13801
• Languages: English
• ISSNs: 0046-5070

Abstract

Abstract: Defining population neighbourhood boundaries provides a critical spatial framework for understanding ecological processes and informing species‐specific conservation and restoration planning. The population neighbourhood boundary demarcates the total quantity of available habitat across life stages where specific ecological processes (e.g., breeding, foraging, migration) occur for all individuals within a population. For mobile species within dendritic freshwater networks, the population neighbourhood boundary can be defined by the upper limit of occurrence (ULO) of an individual within a population. We sought to define contemporary neighbourhood boundaries across a complex river network in southwestern Washington State, U.S.A. by: (1) identifying ULOs for three anadromous fish ( Oncorhynchus spp.); and (2) quantifying the relationship of the ULO with landscape attributes and local features for these species of profound social and ecological importance. Extensive field surveys covering 669 river km across 2 years documented the ULOs for coho salmon ( Oncorhynchus kisutch ), steelhead trout ( Oncorhynchus mykiss ), and chum salmon ( Oncorhynchus keta ) as well as any local features, such as natural barriers, that influenced the ULOs. Generalised linear mixed models quantified the relationships between the ULOs and landscape attributes derived from remote sensing data. We identified the ULO for 115 coho salmon terminal streams, 97 steelhead streams, and 57 chum salmon streams. Natural barriers determined the ULO in just 5%–16% of the terminal streams, depending on species. Landscape attributes associated with the ULOs varied among species; however, drainage area, elevation, and geology were important landscape predictors for all species. Models for each species had good to excellent predictive performance (percent correct classification: 78%–89%; AUC: 0.87–0.96). We demonstrated that large‐scale landscape attributes can accurately and consistently detect species‐specific differences in population neighbourhoods across broad and diverse habitats. In comparison, local scale features (e.g., natural barriers) played a minor role in determining the neighbourhood boundaries in our study catchment. The modelled relationships between landscape attributes and the population neighbourhood provide a framework to guide conservation planning and predict benefits of habitat reconnections.