Flow and temperature effects on life history diversity of Oncorhynchus mykiss in the Yakima River basin

Дата канвертавання24.04.2016
Памер282.99 Kb.
1   2   3   4   5

Figure 4. Modeled habitat area as a function of flow for fry, juvenile and resident adult life stages in the nine LHRM reaches. Tributary reaches derived from PHABSIM studies conducted by the Yakima Nation and Washington Department of Fish and Wildlife (Frederiksen pers. comm.; Pacheco pers. comm.). Mainstem Yakima and Naches River reaches derived from 2D modeling conducted by the U.S. Geological Survey (Bovee et al. 2008).

Figure 5. Fry, juvenile and adult O. mykiss carrying capacity as a function of flow in Satus Creek. Capacities derived from habitat area estimates using average territory sizes for fry (0.08 m2), juveniles (1.48 m2) and adults (10.34 m2). Y-axis scaled down by 100x for fry and 10x for juveniles.

Figure 6. Relationship between stream-dwelling salmonid size and territory size (Grant and Kramer 1990) used to convert habitat area estimates to estimates of carrying capacity.


The effects of environmental conditions on fish growth in freshwater is captured through a bioenergetic model, whereby growth is influenced by food availability and stream temperature. Fish size information is utilized by numerous functions in the life cycle model including carrying capacity, overwinter survival, smoltification, marine survival, and fecundity. We developed a generalized growth model following the methods described by Mangel and Satterthwaite (2008) and Thorpe et al. (1998). According to this model, growth is defined as the difference between anabolic gains (i.e. consumption) and catabolic losses (i.e. energy lost through respiration). Daily growth for an average-sized fish on day t is given by:


Where Wi denotes fish mass in grams and T denotes daily average stream temperature (°C). The left-hand side of the equation represents anabolic gains, where qi denotes individual variation in food finding and processing ability and qe denotes environmental variation in food availability. The function describes the influence of stream temperature on food finding and processing ability and was based on the Thornton and Lessem algorithm described in Hanson (1997) and parameterized using data for juvenile steelhead described in Sullivan et al. (2000). According to this function, food finding and processing ability at full ration peaks between about 12 and18°C and declines rapidly at temperatures exceeding 20°C. An optimal temperature for consumption between 12 and 18°C is consistent with values reported in other salmonid studies (Elliott 1994; Tyler and Bolduc 2008). We assumed growth is described by the von Bertalanffy formula by setting the weight exponents to a = 2/3 and b = 1. The assumption that metabolic costs grow exponentially with a coefficient of x = 0.06818 was based on the work of Stewart (1980).

We calibrated the growth model to conditions in the Yakima River basin using size at age data provided by the Washington Department of Fish and Wildlife and the Yakama Nation Fisheries Program. We grouped data from the main stem Yakima River into sites between Roza Dam and the Cle Elum River confluence and sites between the Cle Elum confluence and Easton Dam to correspond with broad-scale differences in growth rates. Additionally, we modeled growth in three tributary sites including Taneum Creek, Satus Creek, and Toppenish Creek. The growth model was calibrated separately for each site using daily average temperature data obtained from the U.S. Bureau of Reclamation’s Hydromet database and from temperature data provided by the Yakama Nation Fisheries Program. For mainstem sites, we averaged the temperature data from 1990 to 1993 to correspond with the years that size at age data was collected. With the exception of Toppenish Creek, temperature data from the tributaries was not available for the years during which size-at-age data was collected. Therefore, we used average temperatures for all available years for model calibrations, assuming that the observed growth data was reflective of average stream temperatures.

The growth model was calibrated by solving for the values of qi and αi which minimized the variation between the model prediction and the observed size at age data. The qe parameter was held fixed at 1.0 for all sites to provide a consistent benchmark for food availability across sites and to aid in evaluation of model sensitivity to changes in food availability. We assumed a constant coefficient of variation of 16% based on the average variation in size-at-age data observed in mainstem Yakima River sites.

The growth model appeared to fit the data reasonably well, explaining approximately 76% of the variation in individual fish length in mainstem sites between Roza Dam and the Cle Elum confluence (Figure 7). According to the growth model, fish length increases rapidly from spring through summer, and is followed by periods of low or no growth during the winter months. Growth rate in mainstem sites peaked at approximately 19°C and then declined rapidly to values below zero at temperatures exceeding 23°C (Figure 8). Fish growth was considerably slower in the tributaries, with the growth model explaining approximately 50% of the variation in individual fish length on average.

Figure 7. Observed size-at-age data (circles) from mainstem Yakima River sites between Roza Dam and the Cle Elum River confluence (1990-1993) and the model-predicted growth curve with 95% confidence intervals (black lines).

Figure 8. Predicted growth rate (g ∙ day -1) in the main stem Yakima River between Roza Dam and the Cle Elum River as a function of stream temperature.

The age at which juvenile anadromous fish migrate to the ocean is estimated in the model as a function of fish size at emigration. For simplicity, we assume that all fish of sufficient size migrate to the ocean on May 1st, which was estimated as the median date of emigration for all smolts captured at the Chandler Juvenile Fish Facilities from 1999-2008. We assume that juvenile fish must achieve a length of at least 150 mm in order to smolt. This threshold length for smoltification was calculated as the fifth percentile of O. mykiss smolt lengths collected at the Chandler Juvenile Fish Facility (David Lind, pers. comm.) and is consistent with estimates of minimum O. mykiss smolt size at other sites sampled in the Columbia and Snake River Basins (Yanke et al. 2007; White et al. 2007). For each age class of anadromous juveniles, the model calculates the proportion of fish that exceed 150 mm on May 1st, assuming a growth trajectory as described above and a coefficient of variation of 16%. All anadromous juveniles are assumed to emigrate by the spring of their third year in freshwater because age-4 smolts are rarely observed in the Yakima Basin.


Survival of rearing O. mykiss in freshwater was estimated using two different methods depending on the time of year. During winter (November-February), when growth is very limited due to cold river temperatures and activity is expected to slow considerably, survival was assumed to be density-independent and size-dependent. Numerous studies have indicated that survival of juvenile salmonids during winter is positively related to fish size (Quinn and Peterson 1996; Ebersole et al. 2006; Smith and Griffith 1994). Lacking specific data from the Yakima River basin, we used the results from a study of rainbow trout survival in the Henrys Fork of the Snake River, Idaho (Smith and Griffith 1994) to develop a size-dependent logistic function for over-winter survival (Figure 9). We assumed an additional 30% mortality due to predation and other mortality factors that were not accounted for in the Smith and Griffith study. This additional mortality was determined by selecting the value which, when combined with other sources of mortality in the model such as post-spawning mortality and mortality from spring to fall, roughly corresponded with observed annual survival rates of resident O. mykiss in the upper Yakima River.

During spring through fall (March – October), daily survival rates were modeled as a function of carrying capacity and abundance. As fish grow, territory size increases (Figure 6) and habitat carrying capacity declines. In addition, changes in river discharge influence rearing capacity via changes in the quantity of available rearing habitat (Figure 4). Such changes in capacity directly influence freshwater survival via a hockey-stick function (Appendix A). When the estimated abundance for a given age class was less than the predicted rearing capacity for that age class, we assumed that fish survive at a maximum annual rate of 40% for fry, and 80% for all other age classes. Lacking specific data from the Yakima Basin on maximum survival rates of resident O. mykiss at low abundance, we selected these values based on professional judgment and evaluated model sensitivity to these values to determine their relative influence on key model results.

When abundance for a given age class exceeded the predicted rearing capacity, we assumed survival would decline in a density-dependent manner such that the abundance would quickly approach capacity. Specifically, survival S of fish of age m on day t is calculated as the ratio of capacity to abundance expressed over a period of d days and is given by:

Sm,t = (Km,t/Nm,t)(1/d),
where K is the estimated capacity, N is abundance, and d is an assumed lag period for survival. We apply a lag period of 14 days to density-dependent survival, such that, when abundance exceeds capacity, the mortality associated with this capacity limitation is expressed over an extended period of time (e.g. 14 days) instead of occurring instantaneously.

Significant mortality of resident trout occurs shortly after spawning as a result of increased energy expenditure associated with competition for mates and spawning territories and disproportionate allocation of resources for gonadal development (Schroeder and Smith 1989). In the model, we assume a post-spawning survival rate of 55% based on a estimates from Umtanum Creek, a tributary to the Yakima River (Wydoski and Whitney 2003). Egg-to-fry survival was assumed to be 20% for both resident and anadromous offspring based on average egg-to-fry survival estimates for steelhead in Snow Creek, Washington (Bley and Morning (1989)).

Survival of migrating smolts was based on smolt-to-adult return rates (SAR) estimated from smolt and adult steelhead counts at Prosser and Roza dams on the Yakima River and an assumed relationship between smolt size at emigration and marine survival. We developed a logistic relationship between smolt size at emigration and ocean survival based on data presented in Ward et al. 1989. We then scaled the logistic relationship to better reflect the lower average SAR rates observed for Yakima River steelhead compared with the coastal Keogh River population from which the logistic relationship was developed. Scaled relationships were developed separately for the lower Yakima Basin (below Roza Dam) and upper Yakima Basin (above Roza Dam) such that the SAR for an average-sized smolt (i.e., 175 mm) was equal to the geometric mean of the estimated SAR for steelhead smolts migrating from the Yakima River from 1985-2002 (Figure 10). Assuming a migration survival of 43.3% from the Upper Yakima River to Roza Dam, the geometric mean SAR for smolts originating from the upper basin was estimated to be 1.25% compared with a geometric mean of 2.88% for smolts originating from the lower basin (Chris Frederiksen Pers. Comm.).

Figure 9. Length-dependent over-winter survival function showing the original

Figure 10. Relationships between length at emigration (mm) and smolt-to-adult return rates for steelhead smolts originating from the upper Yakima Basin (above Roza Dam) and the lower Yakima Basin (below Roza Dam).

Maturity, Sex Composition, and Fecundity

Age-at-maturity for resident female spawners was estimated from age and maturation data collected in the main stem Yakima River above Roza Dam between 1990 and 1993 (WDFW, Gabriel Temple Pers. Comm.) (Appendix A). According to these data, the probability of a resident female attaining sexual maturity is roughly 8% at age 1, 22% at age 2, 47% at age 3, 73% at age 4, and 90% at age 5. We assumed that the resident O. mykiss population is composed of 60% males and 40% females based on average sex ratios observed in the upper Yakima Basin from 1990 to 1993 (Pearsons et al. 1993).

The age distribution and sex composition for returning anadromous spawners was based on fish sampled at Prosser and Roza Dams between 2002 and 2005 (Conley et al. 2008; Appendix A). The majority of the run (i.e., roughly 60-80%) consisted of fish that migrated to the ocean as two year old smolts and returned to spawn after 1 or 2 years in the ocean. The sex composition of anadromous spawners is heavily skewed towards females, with the percentage of females averaging 68 and 78% at Prosser and Roza dams respectively. To capture the observed differences in age and sex composition of spawners sampled at upper basin (Roza) and lower basin (Prosser) locations, we applied values based on Prosser data to lower basin reaches (i.e., Satus Creek, Toppenish Creek, Wapato, Union Gap, and Naches) and values based on Roza data to upper basin reaches (i.e., Kittitas, Easton, and Taneum Creek).

We used length-fecundity relationships developed from steelhead spawners captured at Prosser Dam (Conley et al. 2008) and resident O. mykiss sampled in tributaries and mainstem habitats in the upper Yakima Basin (Pearsons et al. 1993) to estimate the number of eggs produced by each spawning female (Appendix A; Figure 11). With an average fecundity of about 5,100 eggs per female (Fast and Berg 2001), steelhead in the Yakima Basin are substantially more fecund than their resident counterparts, which average about 800 eggs per female. Though steelhead produce considerably more eggs per female, resident trout have a higher rate of iteroparity, roughly 25-90% (Schroeder and Smith 1989 and Buchanan et al. 1990) versus roughly 5% for steelhead (Schroeder and Smith 1989 and Chilcote 2001), which slightly offsets the fecundity advantage of anadromous O. mykiss. In the model, we assumed a repeat spawning rate for resident fish of 50% based on the average value observed for rainbow trout in the Deschutes River (Schroeder and Smith 1989). We used a repeat spawning rate of 5.4% for anadromous fish based on the average value observed for steelhead returning to the Yakima River (Conley et al. 2008).

Figure 11. Fork length (mm) versus eggs per female for resident and anadromous O. mykiss spawners in the Yakima Basin

Mate Selectivity

Spawning between resident and anadromous O. mykiss has been well documented (Pearsons et al. 2007 and McMillan et al. 2007), and we expect mate selection to be affected by fish size and abundance. We estimated a spawner fidelity rate based on anadromous and resident male abundance to account for the higher probability of anadromous females spawning with resident males when anadromous male abundance is low. We calculated a baseline anadromous spawner fidelity rate of 0.76 when the ratio of anadromous to resident male abundance was 1:1. Anadromous spawner fidelity is defined as the probability that an anadromous female with mate with an anadromous male. Our baseline estimate was derived from studies of O. mykiss mating systems on the Olympic Peninsula, Washington (McMillan et al. 2007; John McMillan Pers. comm.). The baseline fidelity rate was then scaled between zero and one by assuming that the fidelity rate was proportional to the ratio of anadromous male to resident male spawners (Figure 12).

Figure 12: Estimated anadromous female spawner fidelity as a function of the proportion of the population comprised of anadromous males.

Cross-Ecotype Production

Evidence of rainbow trout producing anadromous offspring and steelhead producing resident offspring (Pascual et al. 2001; Zimmerman et al. In Press; Thrower and Joyce 2004) made it necessary to accommodate cross-ecotype production within the LHRM. Juvenile fish size or growth has been hypothesized as an indicator of which life history, residency or anadromy, a fish will adopt. However, state-dependent modeling approaches explaining resident and anadromous salmonid production remain largely theoretical and there is little data available to parameterize such a model. We took a simplified, empirically-based approach by assuming a fixed proportion of juveniles from each parental cross adopt anadromous and resident life history strategies. Barring data specific to Yakima stocks, we assumed O. mykiss adopt life history strategies proportional to observed values from the Grande Ronde River, a tributary to the Snake River (Carmichael pers. comm.) (Figure 13). Model sensitivity to changes in estimates of cross-ecotype production was explored.

Figure 13. Proportion of offspring smolting and residualizing from each of the four possible parent crosses in the LHRM. St=steelhead; Rb=resident rainbow; F=female; and M=male


Juvenile dispersal in the fall from tributary to mainstem habitats was built into the model. Dispersal was a significant uncertainty in the LHRM because data limitations prevent adequate modeling of this behavior; however, anecdotal and fall trapping data (Conley et al. 2008) suggest that a portion of juvenile O. mykiss leave tributary habitats prior to smoltification and continue rearing in the Yakima River main stem. To test the importance of this behavior for determining the balance between anadromy and residency, we built mainstem holding reaches into the model and allowed a fixed proportion of age-1 and age-2 anadromous juveniles to leave tributary habitats and continue rearing in the main stem. Fixed percentages of fall migrants (Appendix A) were approximated based on estimates from the Grande Ronde Basin (Carmichael pers. comm.), and model sensitivity to dispersal rates was explored. Flow and temperature conditions in mainstem holding reaches was assumed to be the same as conditions in mainstem sites nearest the tributary of interest. Survival in holding reaches was assumed to be density-independent and was modeled as a function of stream temperature using the same temperature suitability index that was used to estimate temperature effects on rearing capacity.

Model Sensitivity

We evaluated model sensitivity in terms of relative reproductive success in year 10 to changes in a variety of key model parameters including smolt-to-adult survival rates, fall-winter dispersal rates, anadromous spawner fidelity, cross-ecotype smolt production, growth (i.e., food availability), age-at-maturity, length threshold for smoltification, and maximum freshwater survival (Mar-Oct). We selected a range of reasonable values for each input parameter based on available empirical data, literature review, and/or professional judgment (Appendix C). For simplicity, we limited the sensitivity analysis to three model reaches (Kittitas, Taneum, and Toppenish Mid), representing different channel types, hydrologic regimes, and locations within the basin.

To evaluate model sensitivity to assumptions about cross-ecotype smolt production, we adjusted the proportion of offspring from the anadromous female and resident male cross that smolt, and examined resulting changes in relative reproductive success in simulation year 10. We focused specifically on mating pairs involving anadromous females and resident males because the frequency of interbreeding between different ecotypes, and the fate of their offspring represents one of the critical uncertainties in the life history dynamics of O. mykiss in the Yakima Basin.

After ten years of simulation using average flow and temperature conditions, comparisons of total egg production for resident and anadromous populations suggested that an anadromous life history strategy is favored in lower basin tributary reaches (e.g. Satus and Toppenish Creeks), while a resident strategy was most successful in all other reaches (Figure 2). Total anadromous egg production in lower tributary sites exceeded resident egg production by approximately 2-5 times, with relative reproductive success for anadromous fish being highest in the upper Toppenish and Satus Creek sites (Figure 14).

Model predictions of total resident spawner abundance at equilibrium exceeded anadromous spawner abundance in all model reaches (Table 2). Resident spawner abundance (fish per km) ranged from 47 in Taneum Creek to 1,246 in the Wapato reach (mean = 491), while anadromous spawner abundance ranged from only 0.4 to 61 (mean = 18). Resident spawners mature at an earlier age and much smaller size than their anadromous counterparts, such that a typical anadromous female deposits several times more eggs than a resident female. Sex ratios are also quite different with females composing approximately 75% of anadromous spawners, but only 40% of resident spawners. As a result of the difference in fecundity and sex ratios, the different metrics used to express the relative production of anadromous to resident offspring (egg ratios and spawner ratios) are not directly comparable. However, the metrics show the same trend in response of a given population to changes in environmental factors such as flow or smolt-to-adult survival.

Figure 14. Relative reproductive success in simulation year 10 for each model reach using baseline environmental conditions and default model parameters. The dashed line indicates the value at which resident and anadromous reproductive success is equal.

Table 2. Summary of key model results for each reach in simulation year 10 including anadromous and resident spawner abundance, egg production, and relative reproductive success. Model results are standardized by reach length to aid in comparisons.

Spawners ∙ km-1

Eggs ∙ km-1

Rel. Repr. Success






Anad Eggs / Res Eggs

Yakima (Easton)






Yakima (Kittitas)






Taneum Cr.












Yakima (Union Gap)






Yakima (Wapato)






Toppenish Cr. (upper)






Toppenish Cr. (mid)






Satus Cr.






Simulated changes in flow and temperature substantially influenced relative reproductive success in Taneum and Toppenish Mid sites, but had little effect on populations in the Kittitas site (Figure 15). Reducing summer-fall flow in the Taneum tributary site by approximately 3.5 cfs resulted in a substantial shift in the balance of total egg production towards anadromy. Additional reductions in flow beyond approximately 4 cfs had little effect on relative reproductive success at this site. Similarly, steady increases in relative reproductive success resulted from moderate decreases in summer base flow in the Toppenish Mid reach. In contrast, reproductive success in the main stem Kittitas site was insensitive to simulated changes in flow of up to 600 cfs above and below summer base flow conditions.

1   2   3   4   5

База данных защищена авторским правом ©shkola.of.by 2016
звярнуцца да адміністрацыі

    Галоўная старонка