Using a Stream Network Census of Fish and Habitat to Assess Models of Juvenile Salmonid Distribution

Abstract

We censused juvenile salmonids and stream habitat over two consecutive summers to test the ability of habitat models to explain the distribution of juvenile Coho Salmon Oncorhynchus kisutch, young-of-the-year (age-0) steelhead O. mykiss, and steelhead parr (age ≥1) within a network consisting of several different-sized streams. Our network-scale habitat models explained 27, 11, and 19% of the variation in density of juvenile Coho Salmon, age-0 steelhead, and steelhead parr, respectively, but strong levels of spatial autocorrelation were typically present in the residuals. Explanatory power of base habitat models increased and spatial autocorrelation decreased with the sequential inclusion of variables accounting for the effects of stream size, year, stream, reach location, and a tertiary interaction term. Stream-scale models were highly variable. Fish–habitat associations were rarely linear and ranged from negative to positive; the variable accounting for location of the habitat within a stream was often more important than the habitat variables. The limited success of our network-scale models was apparently related to variation in the strength and shape of fish–habitat associations across and within streams and years. These results indicate that there are several potential limitations to extrapolating models to broader areas based only on spatially limited surveys.

Received October 11, 2012; accepted March 21, 2013

ACKNOWLEDGMENTS

Project funding was provided by the Wild Salmon Center, the National Oceanic and Atmospheric Administration (NOAA) Northwest Fisheries Science Center, and the NOAA Open Rivers Initiative. Sincere thanks goes to Guido Rahr of the Wild Salmon Center for his fundraising and support. The field work portion of the study was conducted by J. R. McMillan and J. Starr while employed by the Wild Salmon Center. We also thank Phil DeCillis and Jim Jacoby for helping us to access some of the streams, and we are grateful to Steve Katz, Gordie Reeves, Becky Flitcroft, Dave Hockman-Wert, Phil Roni, Tom Quinn, Jason Dunham, and Bill McMillan for providing helpful suggestions and reviews on early versions of the manuscript. Use of trade or firm names is for reader information only and does not constitute endorsement of any product or service by the U.S. Government.