State-Dependent Migration Timing and Use of Multiple Habitat Types in Anadromous Salmonids

Anadromous salmonids vary considerably in their age at ocean entry, their timing of ocean entry within a year, and the extent to which they use multiple habitat types within freshwater. To better understand habitat use and movement timing, we developed a broadly applicable model of state-dependent movements among multiple habitats, which was parameterized based on a case study of steelhead Oncorhynchus mykiss in a California coastal watershed with a seasonally closed lagoonal estuary. The model correctly predicted population-level patterns, including predominance of anadromy and a dominant smolt age of 2 years. In addition, the new model predicted the occurrence of small, lagoon-rearing fish (displaying smoltlike migratory behavior) that returned upstream and did not enter the ocean until the next year, whereas large fish emigrated from the lagoon into the ocean. The new model predicted all-or-nothing habitat use for fish of a given size, but we observed a mix of strategies for fish of the same size. Our modeling suggests that a mortality–growth rate tradeoff can explain much of the life history variation, but this tradeoff alone cannot drive a mixture of habitat use strategies by fish of a similar state (i.e., length). We predicted that a mixed strategy may develop as a consequence of density-dependent reduction in growth rates, arising as more individuals recruit to the originally preferable habitat. Higher risk in the higher-growth habitat may halt recruitment to the high-growth habitat even before growth rates are equalized. Uncertainty in rewards associated with the higher-growth habitat may also favor a mixed strategy in which only some fish accept the higher risk associated with increased growth *Corresponding author: [email protected] Received July 21, 2011; accepted December 28, 2011 781 D ow nl oa de d by [ U ni ve rs ity o f C al if or ni a Sa nt a C ru z] a t 0 8: 38 2 5 M ay 2 01 2 782 SATTERTHWAITE ET AL. opportunity. This model framework can be used to predict movement timing and use of multiple habitats for other salmonids and in other systems. A relatively well-developed conceptual and computational theory describes variation in age-specific movements by salmonids between freshwater and ocean environments. First developed for Atlantic salmon Salmo salar (Mangel 1994; Thorpe et al. 1998) and then extended to Arctic char Salvelinus alpinus (Rikardsen et al. 2004) and steelhead Oncorhynchus mykiss (anadromous rainbow trout; Mangel and Satterthwaite 2008; Satterthwaite et al. 2009, 2010), these models allow predictions of life history patterns on the basis of tradeoffs between opportunities for growth and survival. These state-dependent life history models typically operate at the level of two habitats (i.e., freshwater versus ocean) and predict life history timing at the level of years (e.g., smolt at age 1 or smolt at age 2). Thus, they do not account for fine-scale variation important in some salmonid life histories. For example, many salmonids use estuarine habitats for varying time periods, and this phase of their life cycle is distinct from their use of either upstream or ocean habitats (Thorpe 1994; Miller and Sadro 2003; Koski 2009). In addition to variation in age at emigration, there can be considerable individual variability in the within-year timing of ocean entry (Fisher 1994; Bottom et al. 2005). In population-level tests, state-dependent models developed by Satterthwaite et al. (2009, 2010) successfully predicted the frequency of resident and anadromous steelhead in two California Central Valley populations and correctly predicted the predominance of anadromy and the most frequent age of smoltification on the central California coast (Sogard et al. 2012). These models predicted life histories of individuals on the basis of their current state (size) and expected survival probabilities and state dynamics (growth) associated with different habitats, with resultant effects on expected lifetime fitness. However, to date there are no published tests of state-dependent life history predictions from these models at the individual fish level. Such tests will increase the usefulness of models as management tools; by learning where current models fail, we can identify and then incorporate important mechanisms that have been omitted and we can identify life stages for which more data are needed. We revisited a state-dependent steelhead life history model that was developed by Satterthwaite et al. (2009), and we applied it to growth rates documented in the upper Scott Creek (central California coast) watershed by Hayes et al. (2008). Although it correctly predicted the approximate age structure of smolts, the model missed some details of life history in Scott Creek and nearby Waddell Creek (Shapovalov and Taft 1954; Hayes et al. 2011), namely the presence of small (<100 mm fork length [FL]) fish among the downstream migrants and some fish that rear in a downstream lagoon but then return upstream rather than entering the ocean. The model predicted a minimum size of 100–110 mm in December for fish to initiate the smoltification process, but smaller fish have been observed migrating downstream with elevated levels of gill Na+ , K+ ATPase (Hayes et al. 2011), an enzyme that provides an index of saltwater readiness (Zaugg and McLain 1972). In addition, some fish have been observed to migrate to and oversummer in the lagoon, migrate back upstream in fall, and overwinter upstream before smolting and emigrating in the following spring (Shapovalov and Taft 1954; Hanson 2008; Hayes et al. 2011). These discrepancies between model predictions and empirical observations may have arisen because Satterthwaite et al. (2009) neglected the potential for the high-risk–high-reward lagoon habitat to influence steelhead life history trajectories. Here, we apply a newly developed model to this lagoonal system as a case study for the treatment of state-dependent use of multiple habitat types by salmonids, but our methods can be generalized in a straightforward manner.