1. Introduction

The Columbia Basin is an extensive region extending into the states of Washington, Oregon, Idaho, and Montana, and the province of British Columbia (Figure 1.1). In addition to the Columbia River and its major tributary, the Snake River, many tributaries, including the Yakima, Wenatchee, Methow, Clearwater and Salmon Rivers, comprise the Columbia River system. Several species of anadromous¹ salmonids² inhabit the Columbia River system - sockeye (Oncorhynchus nerka), coho (O. kisutch), chinook (O. tshawytscha), and steelhead (O. mykiss).

Anadromous salmonids spend the first part of their lives in rivers or lakes and then migrate downstream to saltwater as juveniles. After an extended period of growth in saltwater, they return to a freshwater environment to spawn. Because of their migratory nature, they encounter a variety of habitats and thus are exposed to several different sources of mortality. The accumulation of these effects has greatly reduced their numbers in the Columbia River system. Spawning and rearing habitat has been degraded due to development, irrigation, and logging practices. Also, some dams, such as the Grand Coulee, are impassable to fish, and rearing habitat has been entirely lost upstream from the dam. As a result, over fifty percent of spawning habitat has been eliminated above McNary Dam (Raymond, 1988). In the ocean, harvest of adults by sport and commercial fisheries is an additional source of mortality.

The downstream migration of juvenile salmon is a particularly critical stage of the salmon life history (NPPC, 1992), but migratory behavior is not well understood. Some species of salmon migrate for hundreds of miles as juveniles and in doing so incur heavy mortality due to factors such as predation and disease. In addition, during outmigration juvenile salmon undergo smoltification, a series of physiological, behavioral and biochemical changes preparing them for a saltwater habitat (Hoar, 1976). Since arrival to the estuary is coordinated with smoltification (Folmar and Dickhoff, 1980), the timing of outmigration is important to ensure that the smolts reach saltwater when they are physiologically ready.

In the Columbia River system, the downstream migrants are exposed to further hazards due to the presence of dams - some runs must pass nine dams during their migration. In addition to being a direct source of mortality, dams complicate the migration process by creating large reservoirs in which river velocity is significantly reduced (Raymond, 1968), potentially disrupting the timing of migration. Also, the reservoirs have higher temperatures and less turbidity compared to free flowing rivers, potentially resulting in greater susceptibility to predation and disease (Park, 1969). In light of this, it is not surprising that mitigation efforts have targeted the downstream migration phase as crucial in terms of revitalizing salmon populations in the Columbia River system.

In this thesis, I develop spatial and temporal models of distributions of migrating juvenile salmonids. Model building and testing can be an important component in natural resource management. Models allow for the examination of various long term management scenarios without conducting costly experiments. The predictive ability of models is a useful tool in day to day operations. Also, the process of developing models and applying them to data enhances the understanding of the animal's behavior.

In all cases, the models I develop have practical applications, so comparing the models to data is important. In analyzing data, I have several objectives. First of all, I estimate parameters and construct confidence intervals. Secondly, I assess whether the models are consistent with the data; this involves goodness-of-fit tests. In some cases I evaluate which factors - biotic and abiotic - are important to the models. Finally, I attempt to use the models as predictive tools with independent data.

For the remainder of this chapter, I provide a brief overview of salmon biology and behavior relevant to modeling migrating populations. Chapter 2 reviews the use of models to describe dispersing animal populations. Chapter 3 covers the statistical procedures I follow and discusses the data used in the applications. Chapter 4 presents a model of the travel time of juveniles through a reservoir and includes applications to group releases of migrating chinook salmon and steelhead. Several extensions to the travel time model are presented in chapter 5 - time dependent mortality, delay in migration, a migrational effort component, and time variable parameters. In chapter 6, I develop methods to apply the travel time model to individuals, and in the process, assess which individual covariates are important to the model. In chapter 7, I develop models of the movement patterns of individuals and apply the models to radio-tracking data. Chapter 8 contains a model of the vertical distribution of fish in the water column in relation to environmental gradients.

All computational algorithms are written in the C programming language (Kernighan and Ritchie, 1978) and run on a Sun Sparcstation 2. Plots were constructed and some of the statistical analyses performed using the S-plus statistical/graphical software package (Becker, et al., 1988).

²Salmonids are salmon and their close relatives, including trout.

Spatial and Temporal Models of Migrating Juvenile Salmon with Applications.