Early Self Sorting Behavior In Chinook Salmon Is Correlated With Variation In Growth Behavior And Morphology Later In Life PDF Download

High mortality rates of Pacific salmon (Oncorhynchus spp.) in the nearshore ocean environment of the Columbia River (Northwest USA) is one of several key factors limiting recovery of these threatened and endangered fish. Several studies describe correlative relationships between environmental or biological factors and fish abundance. However, few mechanistic descriptions exist that describe the causes of growth and mortality during the early ocean life stage (i.e., the first two to four months in the ocean). Similarly, salmon navigation and behavior during early ocean migration is poorly understood. The purpose of this study was to build a spatially-explicit individual-based model (IBM) of yearling Chinook salmon migration in the nearshore ocean environment that mechanistically describes the biologically-relevant processes impacting salmon movement and growth during the early ocean life-history stage. The model domain covers about 1000 km of shoreline from northern California to Vancouver Island, BC and extends about 300 km offshore. Specific objectives were to: 1. Model yearling Chinook salmon spatial distribution through time as a function of environmental and geospatial covariates. Covariates were chosen and grouped according to the types of sensory capabilities salmon use to detect them. Results can therefore inform the behaviors and external cues used during migration. 2. Construct a spatially-explicit IBM that includes many of the basic ecological processes of early ocean migration and growth, relying on an existing external hydrodynamic model for environmental variables. The model tracked individual fish through space and time, recording location, size, and state (alive or dead, energetic content, etc.) for the first several months of ocean life. Several different migration strategies were simulated and compared to existing empirical estimates of spatially-explicit abundance data from a ten-year ocean cruise dataset. Only one strategy, defined here, was able characterize the observed spatial temporal distribution of fish. 3. Validate and fine-tune the model using existing empirical estimates of growth and migration rates derived from otolith microchemistry from about 200 individuals. For this validation, I used the behavior that was shown to be most reasonable in Objective 2 to test against empirical data. Chapter 1 provides a general background for the analyses described above and some of the reasoning that went into the project design. Chapter 2 describes the use of a zero-inflated Generalized Linear Model assuming a negative binomial error structure to describe catches of yearling Chinook salmon as a function of both environmental and geospatial covariates. I found that both types of information were associated with salmon abundance, but that the geospatial information was slightly more informative in the model. I conclude that environmental conditions experienced during out-migration can alter the genetically-driven, stock-specific migration patterns observed in the marine environment. By applying the model to multiple stocks over three months, I was able to show that spatial distributions vary among stocks and change through time. Chapter 3 compares catch data collected during May and June in three different years to simulations of fish distributions generated with five distinct migration strategies. Only two strategies produced fish distributions similar to those observed in May and only one of these mimicked the observed distributions through late June. In the strategies that result in matches with empirical data, salmon distinguish North from South (i.e., they must have a compass sense), and control their position relative to particular landmarks such as the river mouth (i.e., they must have a map sense). Salmon with these two abilities could follow spatially-explicit behavior rules and avoid entrapment in strong southward currents or advection offshore. To fit the relatively consistent interannual spatial distributions observed over the migration season, simulated swimming speed needed to vary among years, suggesting that salmon also have a clock sense to guide the timing of their migration. In Chapter 4, I applied the spatially-explicit individual based model of early marine migration designed in Chapter 3 on two stocks of yearling Chinook salmon to quantify the influence of external forces on estimates of swimming speed and consumption. Swimming speeds required in the model were higher than those estimated without taking into account ocean currents (and assuming a straight-line migration from the river mouth to the capture location). Moreover, the estimated variance in swimming speeds was significantly lower than the variance in movement rates, suggesting that ocean currents mask salmon behaviors and the role of genetically-determined movement may be more important in marine migration than previously thought. There was also a stock-specific response, as fish from the Snake River Basin swam faster than salmon from the Mid and Upper Columbia River. By taking into account experiences of individual fish, this approach incorporates both individual behavior and the influence of external physical factors such as ocean currents, allowing a more accurate estimation of biological parameters.