Landscape models to understand steelhead (Oncorhynchus mykiss) distribution and help prioritize barrier removals in the Willamette basin, Oregon, USA

• Published in: Canadian journal of fisheries and aquatic sciences Vol. 61; no. 6; pp. 999 - 1011
• Main Authors: Steel, E Ashley, Feist, Blake E, Jensen, David W, Pess, George R, Sheer, Mindi B, Brauner, Jody B, Bilby, Robert E
• Format: Journal Article
• Language: English
• Published: Ottawa, Canada NRC Research Press 01.06.2004; National Research Council of Canada; Canadian Science Publishing NRC Research Press
• Subjects: Agnatha. Pisces; Agricultural land; Alluvium; Animal and plant ecology; Animal migration; Animal populations; Animal, plant and microbial ecology; Animals; Biological and medical sciences; Climate change; Comparative studies; Decision making; Demecology; Density; Fish; Freshwater; Fundamental and applied biological sciences. Psychology; Geology; Habitats; Land management; Land use; Landscape; Landslides; Life history; Oncorhynchus mykiss; River basins; Rivers; Salmon; Scale models; Spawning; Upstream; Vegetation; Vertebrata; Oregon; United States; North America; America; Willamette River; USA, Oregon; USA, Oregon, Willamette R; Salmonidae; Aquatic environment; Vertebrata; Landscape; Spatial distribution; Migratory; Pisces; Models; Oncorhynchus mykiss; Population density; Linear model; Anadromy
• DOI: 10.1139/f04-042
• ISSN: 0706-652X

We use linear mixed models to predict winter steelhead (Oncorhynchus mykiss) redd density from geology, land use, and climate variables in the Willamette River basin, Oregon. Landscape variables included in the set of best models were alluvium, hillslope < 6%, landslide-derived geology, young (<40 years) forest, shrub vegetation, agricultural land use, and mafic volcanic geology. Our approach enables us to model the temporal correlation between annual redd counts at the same site while extracting patterns of relative redd density across sites that are consistent even among years with varying strengths of steelhead returns. We use our model to predict redd density (redds per kilometre) upstream of 111 probable migration barriers as well as the 95% confidence interval around the redd density prediction and the total number of potential redds behind each barrier. Using a metric that incorporates uncertainty, we identified high-priority barriers that might have been overlooked using only stream length or mean predicted fish benefit and we clearly differentiated between otherwise similar barriers. We show that landscape features can be used to describe and predict the distribution of winter steelhead redds and that these models can be used immediately to improve decision-making for anadromous salmonids.