Incorporating Movement Patterns to Improve Survival Estimates for Juvenile Bull Trout

Populations of many fish species are sensitive to changes in vital rates during early life stages, but our understanding of the factors affecting growth, survival, and movement patterns is often extremely limited for juvenile fish. These critical information gaps are particularly evident for bull trout Salvelinus confluentus, a threatened Pacific Northwest char. We combined several active and passive mark–recapture and resight techniques to assess migration rates and estimate survival for juvenile bull trout (70–170 mm total length). We evaluated the relative performance of multiple survival estimation techniques by comparing results from a common Cormack–Jolly–Seber (CJS) model, the less widely used Barker model, and a simple return rate (an index of survival). Juvenile bull trout of all sizes emigrated from their natal habitat throughout the year, and thereafter migrated up to 50 km downstream. With the CJS model, high emigration rates led to an extreme underestimate of apparent survival, a combined estimate of site fidelity and survival. In contrast, the Barker model, which allows survival and emigration to be modeled as separate parameters, produced estimates of survival that were much less biased than the return rate. Estimates of age-class-specific annual survival from the Barker model based on all available data were 0.218 ± 0.028 (estimate ± SE) for age-1 bull trout and 0.231 ± 0.065 for age-2 bull trout. This research demonstrates the importance of incorporating movement patterns into survival analyses, and we provide one of the first field-based estimates of juvenile bull trout annual survival in relatively pristine rearing conditions. These estimates can provide a baseline for comparison with future studies in more impacted systems and will help managers develop reliable stage-structured population models to evaluate future recovery strategies. Knowledge of a species’ life history and associated vital rates is crucial for development of effective conservation and recovery strategies (Williams et al. 2002). For many fish species, population dynamics are extremely sensitive to changes in survival at early life stages (Houde 1994; Hilborn et al. 2003). However, demographic rates are often difficult to assess between egg deposition and subadult stages, in part because survival rates during early stages are typically relatively low and can be highly variable (Bradford 1995). Although they are sometimes costly to obtain, life-stage-specific estimates of survival can be used *Corresponding author: [email protected] Received February 1, 2012; accepted July 31, 2012 to evaluate the relative contribution of various subadult stages to overall population change and identify targets for management (Caswell 2001; Morris and Doak 2003; Gross et al. 2006). Further, precise estimates of survival can help managers comprehend the magnitude of variability that may occur naturally as a result of environmental factors, such as density-dependent interactions, relative to anthropogenic influences (e.g., Johnston et al. 2007). Mark–recapture studies provide a way to estimate survival and other key demographic information specific to individual 1123 D ow nl oa de d by [ U ta h St at e U ni ve rs ity L ib ra ri es ] at 1 5: 05 1 2 N ov em be r 20 12 1124 BOWERMAN AND BUDY cohorts or life stages (e.g., Lebreton et al. 1992; White and Burnham 1999). However, estimation of demographic rates may be complicated for highly migratory species, both because of the effort needed to recapture mobile individuals and because animal movement patterns can affect interpretation of survival estimates (Cilimburg et al. 2002; Horton and Letcher 2008). For example, estimates of apparent survival (φ) generated using the common Cormack–Jolly–Seber (CJS) model are a combined estimate of true survival and site fidelity, the probability that an animal remains available for recapture within the study area (White and Burnham 1999; Sandercock 2006). With CJS estimates, it is not possible to distinguish permanent emigration from mortality or temporary emigration from capture probability (Barker et al. 2004; Horton and Letcher 2008). As a result, frequent emigration of marked organisms from the study area can confound estimates of apparent survival, and this issue has previously limited studies that sought to estimate the survival of migratory stream-dwelling fishes (e.g., Paul et al. 2000; Letcher et al. 2002). However, recent advances in technology have allowed researchers to improve recapture and resighting probabilities, while new analytical techniques have improved the ability to incorporate movement patterns into mark–recapture survival analyses. The use of passive integrated transponder (PIT) tags has become increasingly common in fisheries research. Novel technology, including mobile PIT tag readers and passive (stationary) in-stream antennas, now often accompany the use of PIT tags. These technical advances offer a promising means of increasing the spatial and temporal extent of resight information (Zydlewski et al. 2006). Fish marked with PIT tags can be located by a researcher actively moving a mobile PIT tag reader through a study site (e.g., Roussel et al. 2000). In comparison, a passive in-stream antenna (PIA) can be operated continually to detect PIT-tagged fish as they swim past a stationary location in the stream. Both of these methods allow detection (i.e., resight) of marked individuals without handling or harassment. Although PIT tag data acquired at PIAs can help describe fish movement patterns within a stream system, resight data collected on a continual basis cannot be incorporated into many standard mark–recapture survival models. In the common CJS model, for example, captures and recaptures must take place over a short time period relative to the time between sampling events to ensure that survival probability is constant among individuals (Lebreton et al. 1992). A more recent model developed by Barker (1997) similarly requires captures during discrete events, but can also incorporate resights of marked animals during the intervals between discrete sampling periods. Whereas captures usually occur within a specific study area, resights of marked animals are assumed to take place throughout the range of the population of interest. Inclusion of this information allows for direct estimation of true survival and site fidelity as distinct parameters (Barker and White 2001; Barker et al. 2004). This model is uncommon in the fisheries literature (but see Buzby and Deegan 2004; Al-Chokhachy and Budy 2008), although it appears promising for studies that include numerous data types (Barker et al. 2004) or for fishes that exhibit coexisting life history strategies and diverse migration patterns (Buzby and Deegan 2004; Horton and Letcher 2008). One such fish species that demonstrates a range of movement patterns is the bull trout Salvelinus confluentus. The bull trout is a threatened species of stream-dwelling char that exhibits variability in life history types, migration patterns, and maturation schedules (Bahr and Shrimpton 2004; Johnston and Post 2009). Bull trout populations often include both migratory and nonmigratory (resident) life history types (McPhail and Baxter 1996; Homel et al. 2008). Adults typically spawn in cold headwater streams which also serve as rearing habitat for juveniles. Bull trout usually disperse between ages 1 and 4, migrating downstream into larger river systems and lakes where they may reside for several years before returning to natal waters to spawn, although resident adult bull trout may inhabit the upper portions of a watershed throughout their lives (Fraley and Shepard 1989; Ratliff 1992; Rieman and McIntyre 1993). For bull trout, high within-population variability and behavioral plasticity encumber the quantification of movement patterns and survival estimates. Bull trout migration distances can range from just a few kilometers to more than 200 km (McPhail and Baxter 1996; Hogen and Scarnecchia 2006), further complicating the estimation of demographic parameters. Considerable research has been conducted to describe migratory behavior and habitat use for individual bull trout populations (Swanberg 1997; Bahr and Shrimpton 2004; Watry and Scarnecchia 2008), but the majority of these studies have focused on adults. Information about bull trout life history requirements and vital rates is still relatively sparse, particularly for early life stages. Very few studies have assessed juvenile bull trout migration patterns, rates of survival, or the environmental factors affecting survival. Life-stage-based population projection models developed for bull trout suggest that population growth may be most sensitive to changes in the survival of large adults and early life stages (Rieman and McIntyre 1993; Al-Chokhachy 2006). However, the predictive ability of such models is currently limited by a lack of empirical survival estimates specific to subadult stages. To our knowledge, reliable estimates of survival for juvenile age-classes (<120 mm total length [TL]) are unavailable for bull trout. Previous studies assessed relative survival for early ageclasses of bull trout by comparing abundances between years but did not produce precise juvenile survival estimates (Paul et al. 2000; Johnston et al. 2007). Al-Chokhachy and Budy (2008) used mark–recapture methods to develop stage-specific survival estimates for bull trout larger than 120 mm TL, but their study did not include smaller individuals. Obtaining survival estimates specific to juvenile stage classes will help fill an important gap in our understanding of factors that determine bull trout survival at different life stages. Estimates of stagespecific survival rates will also aid in identifying the life stages to target for recovery and improve the ability of population D ow nl oa de d by [ U ta h St at e U ni ve rs ity L ib ra ri es ] at 1 5: 05 1 2 N ov em be r 20 12 IMPROVING SURVIVAL ESTIMATES FOR JUVENILE BULL TROUT 1125 models to predict population-level responses to environmental changes. To evaluate migration patterns and estimate survival rates for juvenile bull trout, we conducted an intensive mark–recapture study within one of several important spawning areas used by a relatively large population of bull trout in the South Fork Walla Walla River (SFWW), Oregon. The population of bull trout in the SFWW exhibits both migratory and resident life history forms (Homel et al. 2008), and migration distance and timing can be highly variable (Homel and Budy 2008). Prior to this study, little was known about juvenile bull trout dispersal and survival rates in this system. The overall goal of this research was to provide insight into a stage of bull trout life history which has previously not been well quantified and which has important implications for understanding how juvenile life stages affect population growth and persistence. To meet this goal, the specific objectives of this study were to (1) quantify and better understand the movement patterns exhibited by juvenile bull trout (70–170 mm TL) and (2) incorporate knowledge of juvenile migration rates into mark–recapture analyses to obtain the most precise estimates of survival for bull trout during these influential early life stages. METHODS