Flow and temperature effects on life history diversity of Oncorhynchus mykiss in the Yakima River basin
Authors: Ian Courter, Casey Justice, and Steve Cramer
Oncorhynchus mykiss populations with ocean access display considerable life history plasticity. Resident (rainbow trout) and anadromous (steelhead) adults commonly produce offspring of the alternate ecotype, but environmental drivers of this life history variability are not well understood. To explore potential environmental factors influencing the distribution of resident and anadromous O. mykiss in the Yakima River basin, we used a life-cycle modeling approach to simulate flow and temperature effects on relative reproductive success of each ecotype. As a supplement to more traditional hypotheses about declining steelhead abundance, we propose a theory that more closely resembles evidence from pristine rivers. We hypothesize that flow regimes providing cool temperatures and maintaining depth and velocities necessary to sustain adult O. mykiss throughout the summer and fall seasons result in increased resident rainbow trout abundance and decreased steelhead abundance. Model results indicated a correlation between flow and temperature conditions and relative reproductive success of the two ecotypes and appear to explain in part why the upper Yakima Basin supports renowned resident rainbow trout populations, while tributaries in the lower basin continue to produce predominantly steelhead. Location within the basin also played a central role in determining the life history composition of O. mykiss. Upper basin sites, having a higher mortality cost for migration, favored a resident life history even when local environmental conditions promoted a migratory life history strategy. Channel type was also an important determinant of ecotypic dominance, with higher relative reproductive success of anadromous spawners occurring in tributary habitats. Alteration of flow conditions in mainstem habitats had little effect on the relative reproductive success of anadromous O. mykiss. Our modeling demonstrated that tributary habitats were most likely to support an anadromous ecotype, and management actions that improve tributary habitats have the greatest potential to increase abundance of steelhead in the Yakima Basin.
Table of Contents
Naturally spawning populations of both resident and anadromous Oncorhynchus mykiss populations coexist in sympatry throughout the Columbia and Snake River basins. The anadromous life history form (steelhead) is listed as threatened under the Endangered Species Act (ESA) while the resident form (rainbow trout) is not. This divergence in listing status sustains the lingering confusion surrounding the biology of the species and has proven problematic when attempting to evaluate anadromous O. mykiss population viability. Better explanation of the biological mechanisms that lead to predominance of anadromy, residency, or a balanced mixture is needed.
As in other parts of the Columbia and Snake River Basin, Yakima River steelhead are listed as threatened. A recovery plan, as required for species listed under the ESA, has been developed for Middle Columbia River (MCR) Steelhead, which includes populations in the Yakima, Walla Walla, Deschutes, John Day and Umatilla basins. According to the Plan, three populations in the MCR Evolutionarily Significant Unit (ESU) were determined to be at high risk of extinction, and two of those are in the Yakima Basin. The population of steelhead that spawns in the upper Yakima River is reportedly the most jeopardized independent steelhead population in the Columbia Basin, with a 10-year geometric mean abundance of only 85 adult spawners (Conley et al. 2008). The upper Yakima Basin is routinely cited as an example of imperiled steelhead habitat (Good et al. 2007).
Although few steelhead return to the upper Yakima River, resident rainbow trout are abundant there and support one of the most popular wild trout fisheries in Washington. Hatchery trout have not been stocked since 1991, and genetic sampling indicates the resident trout are similar to native steelhead, and quite distinct from the hatchery stocks formerly planted (Busack et al. 2005). Both mainstem and tributary rainbow trout populations in the upper basin have been stable in abundance and size since annual monitoring began in 1990 (Ham and Pearsons 2000; Temple et al. 2004). In contrast, O. mykiss populations in the lower basin, particularly Satus and Toppenish Creeks, are predominantly composed of anadromous fish (Conley et al. 2008).
Resident rainbow trout and steelhead are known to interbreed (Pearsons et al. 2007 and McMillan et al. 2007), and recent genetic analysis indicates a significant amount of genetic overlap between rainbow trout and steelhead in the upper Yakima River (Pearsons et al. 2007 and Blankenship pers. comm.). This evidence supports the conclusion that different life history forms of O. mykiss within a basin are likely interbreeding when spatial and temporal overlap in spawn timing occurs, congruent with genetic analyses conducted throughout the Pacific Rim (Busby et al. 1996, Docker and Heath 2003, and McPhee et al. 2007).
Several studies have confirmed that in addition to interbreeding, both ecotypes are capable of producing offspring of the alternate ecotype. For example, recent otolith microchemistry studies revealed that both resident and anadromous O. mykiss in the Sacramento-San Joaquin Basin can have maternal origins of the alternate ecotype (Zimmerman et al. In Press); a population of stocked rainbow trout from California has given rise to an anadromous ecotype in the Santa Cruz River, Argentina that are genetically indistinguishable from the resident population (Pascual et al. 2001); and breeding studies from the Grande Ronde River, Oregon and Sashin Creek, Alaska found that pure resident, pure anadromous and mixed spawner crosses all produced both resident and anadromous offspring (Carmichael pers. comm. and Thrower and Joyce 2004). Moreover, evidence suggests that while these life history traits are heritable (Thrower et al. 2004), the capacity of O. mykiss populations to produce steelhead is not lost after many generations of residency. For example, rainbow trout from Sashin Lake, Alaska maintain the capacity to produce smolts after over 70 years of isolation from anadromous recruitment (Thrower and Joyce 2004).
The need to understand the factors driving the relative abundance of anadromous and resident O. mykiss is particularly relevant in the Yakima Basin, where flows in mainstem reaches are largely controlled by releases from upstream storage reservoirs. Both the MCR Recovery Plan and ESA consultations over water operations have identified an urgent need to understand how the production of anadromous and resident O. mykiss is affected by flow management. Thus, a strong practical need exists to understand if changes in flow can be used to achieve desired abundances of steelhead.
Three primary mechanisms have been posed as explanations for the observed distributions of residency and anadromy in O. mykiss populations: genetic control, growth effects on conditional switching of life-histories, and differences in life history survival driven by environmental factors. Firstly, breeding experiments have established that the tendency to follow either a resident or anadromous life history is heritable in O. mykiss (Thrower et al. 2004); however, numerous studies point out that genetics is only one of several factors influencing life history response (Jonsson and Jonsson 1993; Hendry et al. 2004; Nussey et al. 2007; Nichols et al. 2008). Secondly, species of the family Salmonidae appear to have the ability to respond to environmental conditions by adopting either a resident or an anadromous life history. This ability is know as a state-dependent or conditional life history strategy (Houston and McNamara 1992 and Jonsson and Jonsson 1993), and allows individuals to maximize their fitness dependent upon their phenotypic condition (Gross 1996 and Jonsson and Jonsson 1993). Independent of the conditional ability to switch life histories, a third possibility is that environmental circumstances simply result in greater survival and reproduction of one ecotype over the other. Pavlov et al. (2008) concluded this was the explanation for differences in frequency of the two ecotypes between adjacent, pristine rivers in the Kamchatka Russia.
We hypothesize that the observed predominance of rainbow trout in the upper Yakima River main stem and steelhead in lower basin tributaries is controlled by environmental conditions. This same trend can be observed in numerous other regulated rivers throughout the Pacific Northwest and California. The typical explanation for this trend is centered on the higher cost of migration due to water storage and hydroelectric projects, predation, and commercial and sport fisheries. As a supplement to the more traditional hypotheses about declining steelhead abundance, our theory more closely resembles evidence from rivers with minimal anthropogenic effects, such as the Kamchatka Peninsula where unregulated rivers support thriving populations of six distinct O. mykiss life history types, including the traditional resident and anadromous forms as well as a range of estuarine and amphidromous ecotypes.
More explicitly, we theorize that flow regimes providing cool temperatures and maintaining depth and velocities necessary to sustain adult O. mykiss throughout the summer and fall seasons will result in increased resident rainbow trout populations and decreased steelhead abundance. This is consistent with a commonly referenced ecological principle that “when the animal’s needs are being met, it stays where it is; when they are not, it moves until it finds appropriate conditions for its current demands (Thorpe 1994).” Furthermore, this hypothesis may explain why basins like the upper Yakima support renowned resident rainbow trout populations and dwindling steelhead populations.
The shift of an O. mykiss population from residents to predominantly anadromous occurs when the benefits of freshwater survival no longer outweigh the increased fecundity associated with ocean migration. We expect non-anadromous individuals to experience the greatest survival pinch when flows drop to levels that would not sustain the depth and velocity requirements of adult rainbow trout and/or summer temperatures exceeded approximately 18oC (Todd et al. 2008). In these environments, migratory life-histories were expected to predominant. Both survival and fecundity are affected by body size, and growth is often hypothesized as a potential predictor of life history response in facultatively anadromous salmonids (Jonsson and Jonsson 1993; Hendry et al. 2004; Thorpe et al. 1998; Mangel and Satterthwaite 2008). In this paper, we examine the influence of environmental conditions (i.e. temperature and flow) on growth and survival, and simulate resident and anadromous relative reproductive success in O. mykiss populations throughout the Yakima Basin.
In this paper we describe an approach that integrated existing information into a life-cycle simulation model to predict resident and anadromous relative reproductive success in different portions of the Basin. The modeling framework utilized mathematical functions to account for temperature effects on growth, flow effects on juvenile and adult survival, and effects of both body size and life history type on fecundity. We then compared the relative reproductive success of steelhead and resident rainbow trout predicted by the Life History Response Model (LHRM) to the observed distribution of steelhead and rainbow trout in the Yakima River basin. Specifically, we address the following questions: (i) What factors best account for the observed distribution of the two ecotypes in the Yakima Basin? (ii) How are variation in flow and temperature likely to influence those relative distributions?
There are likely numerous valid modeling approaches for quantifying the effects of environmental conditions on life history traits. Roff (2002) presents an extensive review of quantitative approaches to account for life history variation, and in regard to determining relative fitness between life-histories, he concludes, “In circumstances in which there is both density-dependence and stochastic variation, a simulation approach is perhaps the only presently viable approach.” Life-cycle modeling is our preferred method because drivers can be appropriately quantified spatially and temporally. This approach allows one to examine the true balance of life history drivers operating in succession between life stages and across multiple generations with environmental conditions specific to a river segment of interest. Using data collected in the Yakima River basin, we developed a deterministic life-cycle model (LHRM) that combines survival and fecundity tradeoffs into a dynamic framework to predict the balance between resident and anadromous O. mykiss relative reproductive success under different environmental conditions.
A simple depiction of the linkages between primary functions of the LHRM is provided in Figure 1. The model operates on a daily time step. Flow and temperature conditions, specified for a specific river reach of interest, influence growth and available habitat area for fry, juvenile and resident adult O. mykiss. The model evaluates fish size and calculates the average territory size requirements of fish within each age-class. Based on availability of habitat, the model calculates carrying capacity for fry, juvenile and resident adults. Density dependent mortality acts on the fish during the summer and fall seasons when habitat area is limited and fish growth is at its highest. Winter mortality is fixed because fish metabolism is slowed and competition for space is no longer expected to limit survival. We assume no competition for space between the three freshwater life stages because they tend to occupy habitat with different depths and velocities. By the time juveniles have reached a sufficient size for smoltification, they have already made a life history decision to be either anadromous or resident. These decisions were predetermined by fixed rates of resident and anadromous offspring production estimated for each of the possible spawner crosses. Smolt-to-adult survival is dependent on smolt size at emigration and population location within the Yakima Basin. Fecundity of adult resident and anadromous female spawners is size dependent and ecotype dependent, resulting in residents that produce fewer eggs per female at size than anadromous females. A detailed description of various model functions and parameter values is provided in Appendix A.
Model results are expressed in terms of relative reproductive success. We define relative reproductive success as the ratio of total anadromous egg production to total resident egg production. Values of relative reproductive success greater than one indicate conditions that favor an anadromous life history. Each time a simulation is carried out under a specified set of flow and temperature conditions, the model quickly approaches an equilibrium ratio of mature resident and anadromous individuals with a measure of total egg production unique to the size distribution of fish within each life history category. As defined by Hendry et al. 2004, “the change in relative fitness conferred by a given behavior (e.g. migration) is determined by the influence of that behavior on survival to maturity, age at maturity, and reproductive output at maturity.” In our model, the ideal metric for rolling up all three of these parameters into a single measure, which allows us to make relative comparisons between fitness’s of the two life-histories, is egg production at equilibrium. For simplicity and consistency, we evaluated reproductive success at simulation year-ten because the model reached equilibrium within this timeframe under all specified flow and temperature conditions.
We compared predictions of relative reproductive success across model reaches to determine if expected patterns in the composition of anadromous and resident life history types were similar to the observed distribution of O. mykiss ecotypes within the Yakima Basin. Additionally, we examined the effects of environmental conditions on relative reproductive success of anadromous O. mykiss by simulating incremental changes in flow and temperature during the summer base-flow period (June – October) in three different reaches (Kittitas, Taneum, and Toppenish Mid). Our analysis of flow and temperature effects was limited to summer-fall because flow and temperature constraints on rearing capacity are likely to be greatest during this period.
Figure 1. Model flow chart showing key model inputs and relationships between model components.
Study Area Description
The Yakima Basin collects runoff from a network of streams draining the eastern slopes of the Cascade Mountain Range in the southern half of Washington. Most of these streams enter the two main branches off the basin, the Naches and the upper Yakima rivers, which join at rkm 183. The Yakima River flows southward through the arid Yakima Valley to eventually join the Columbia River 541 km upstream of the Pacific Ocean (340 ft msl). Precipitation varies from approximately 128 inches along the crest of the Cascades to less than 8 inches on the eastern valley floor. Much of the precipitation falls as snow in the upper basin, and flows generally peak as snowmelt in the spring. Flows in the upper Yakima Basin (above the Naches confluence) are augmented during summer by releases from three headwater storage reservoirs (~2,500 ft msl with ~840,000 acre-ft active storage), much of which is eventually diverted to irrigation networks downstream. Two additional storage reservoirs (~230,000 acre-ft active storage) modify the flow regime in the Naches branch.
Spawning distribution of steelhead in the Yakima Basin for brood years 1990-1992, as determined through radio-telemetry studies, was: 46% in the Satus Basin; 31% in the Naches Basin; 11% in the Toppenish Basin; 2% in the Marion Drain; 4% in the Yakima River main stem below Roza Dam; and 6% in the Yakima River or tributaries above Roza Dam (rkm 201). Anadromous adults can be counted passing through fish ladders at two irrigation diversions; Roza Dam and Prosser Dam (rkm 74). A fish facility near Prosser Dam also samples juveniles migrating downstream. Steelhead spawning is less widely distributed than rainbow trout, but is within the geographic range of rainbow trout spawning. The spawning time of rainbow trout and steelhead is similar, and peaks progressively later as elevation increases. Steelhead represent less than 1% of the O. mykiss spawners in the upper Yakima River above Roza Dam (Pearsons et al. 1998), but nearly 100% of the O. mykiss found in Satus and Toppenish creeks in the lower basin (Hubble 1992).
To study environmental drivers of anadromy and residency at an appropriate spatial scale, we selected spatial units (reaches) within each of the four Yakima Basin Steelhead Independent Populations identified in the steelhead Recovery Plan: Upper Yakima, Naches, Toppenish Creek and Satus Creek (Figure 2; Table 1). Reaches were chosen for which hydraulic modeling of fish habitat had been completed; either Physical Habitat Simulation (PHABSIM) (Bovee et al. 1998) or two-dimensional (2D) modeling. This information was necessary to quantify the effects of flow on habitat area. Additionally, daily temperature and flow data for each spatial unit were necessary as environmental data inputs to evaluate how the balance between residency and anadromy was influenced by environmental conditions. We simulated a total of nine model reaches—four tributary (Satus Creek, Toppenish Creek Mid and Upper, and Taneum Creek) and five mainstem reaches (Easton, Kittitas, Union Gap, Wapato and Naches).
Flow conditions differed dramatically between mainstem and tributary sites. Figure 3 shows the contrast in typical hydrographs between Toppenish Creek (lower basin tributary) and the main stem of the upper Yakima River. Flows in Toppenish Creek drop to critically low levels in the summer and fall, while flows in the main-stem, augmented by reservoir releases, actually increase slightly through the summer until delivery of irrigation water drops sharply in late August.
Figure 2. Map of the Yakima Basin showing the four independent steelhead population boundaries and model reaches within each population (highlighted in pink). Pie charts indicate the proportion of total egg production of anadromous and resident spawners at equilibrium predicted by the LHRM for each model reach.
Table 1. General characteristics of study reaches represented in the model.
Figure 3. Spring through fall 2007 discharge for the Upper Yakima and Toppenish Creek subbasins.
Numerous factors influence the carrying capacity of rearing O. mykiss in river environments including food availability, temperature, inter- and intraspecific competition, water depth and velocity, mesohabitat composition (i.e. pool, riffle, glide) and habitat complexity (e.g., instream cover, channel sinuosity). The model provides daily estimates of carrying capacity for each age class of rearing O. mykiss as a function of flow, temperature, and fish territory size (i.e., measure of intraspecific competition). Carrying capacity is then used in the model to estimate density-dependent survival rates for each age class via a hockey-stick relationship.
Using site-specific analyses that quantified habitat area as a function of stream discharge (Figure 4), we developed relationships between carrying capacity and stream flow for fry (age 0), juveniles (age 1), and resident adult (ages 2-5) O. mykiss (Figure 5) for each modeled reach. Habitat area curves were derived from PHABSIM and 2D modeling conducted throughout the Yakima Basin (Frederiksen pers. comm.; Pacheco pers. comm.; Bovee et al. 2008). Depth and velocity suitability criteria for fry, juvenile, and adult O. mykiss were based on data from the upper Klamath River (T.R. Payne and Associates 2004) and were modified into a binary format for input into the 2D model following a collaborative Delphi process with biologists and stakeholders working in the Yakima Basin (Appendix B). Habitat area was converted to an index of carrying capacity for each life stage using a relationship between fish size and territory size developed by Grant and Kramer (1990) (Figure 6; Appendix A).
Habitat area predicted from PHABSIM and 2D modeling was adjusted using a temperature suitability index to account for the effects of excessive stream temperatures on rearing capacity. The temperature suitability index was based on temperature criteria for juvenile steelhead developed by Sullivan et al. (2000) and is described by a hockey-stick relationship with values ranging from 1.0 at temperatures less than 17°C to 0 at temperatures exceeding the incipient lethal temperature of 26°C. After adjusting for temperature effects, habitat area was divided by territory size to determine the daily carrying capacity for a given age class. We then calculated a moving average of the daily capacity estimates over a period of 30 days based on the assumption that changes in capacity resulting from differences in flow, temperature, or territory size would not occur instantaneously, but instead would occur gradually over an extended period.
Estimates of suitable habitat area, such as Weighted Useable Area (WUA), are species and life-stage-specific measures of suitable depth, velocity and substrate area commonly used to quantify flow effects on fish habitat availability. It is not customary to convert habitat area estimates from PHABSIM and 2D modeling into estimates of carrying capacity because these indices of stream hydrodynamics are designed to examine relative effects of flow changes on habitat area—they are not predictions of true habitat carrying capacity. There are numerous other factors in addition to depth, velocity and substrate that influence production of fish (i.e. food availability, temperature, cover, etc.). Because of this, we caution readers not to draw conclusions about actual fish abundance or production potential from our carrying capacity estimates, and note that LHRM results are expressed in relative terms to avoid misinterpretation. Actual fish carrying capacity was not estimated in any of the spatial units analyzed, but it was necessary to convert habitat area estimates into a metric suitable for population modeling.