This section of the Methods addresses fish; it is organized as follows:

Introduction

Development of Current Potential and Historic Potential Landscapes

Development of Habitat Attributes

Development of Hatchery Attributes

Development of the Harvest Attributes

Population Structure

General Chinook Life Histories

Application to the Columbia River

Chinook Population Descriptions

Analysis Method

Uncertainty and World View Assumptions

                        World View Assumptions

                        Data Quality

Introduction

The Ecosystem Diagnosis and Treatment (EDT) process is a habitat–life history approach for evaluating potential future performance of fish and wildlife populations (Mobrand et al. 1997). The analytical tool used in the EDT process is an expert system—a compilation of data, information, and knowledge into a hypothesis describing past, present, and future performance of fish and wildlife populations. This expert system translates environmental attribute data into population survival parameters, thus creating a survival landscape. Population performance across this landscape is then estimated, based on life history characteristics of the species of interest. In this report, we detail the methods used to model the survival landscapes and perform the population analysis for the Multi-Species Framework coarse screening analysis. We define variables used, identify data sources, and describe how this data is used in the EDT model.

Development of Current Potential and Historic Potential Landscapes

This analysis defines and describes hypothetical future conditions, or alternatives, with reference to two baseline conditions, Current and Historic. The current baseline describes the average environment available to fish and wildlife over the most recent decade. The historic baseline is a hypothetical reconstruction of conditions prior to non-native human influences. The terms Current Potential and Historic Potential refer to the ability of the baseline conditions to support fish and wildlife. This section describes the methods used to estimate baseline environmental conditions for the coarse screening analysis.

Development of Habitat Attributes

Freshwater and marine environmental conditions are described in terms of habitat attributes. In the following subsection we identify these attributes and how they are estimated.

Freshwater

Tributaries and mainstem are discussed separately in the sections that follow.

Tributaries - Columbia and Snake River Tributaries

This section defines the environmental attributes used in Columbia River tributaries and describes the steps in translating them into the biometrics in the population analysis. A general discussion of the analytical framework and the data structure and terminology is found in Appendices A and B. In brief, three levels of data are used in EDT: Level 1 or environmental input, Level 2 or environmental attributes, and Level 3 or biological performance attributes.

Summarization and completion of Level 1 environmental input data

The initial step in the data organization procedure was to summarize all of the available Level 1 environmental input data at the 6-HUC scale. We assembled environmental data from various data sources, including: Interior Columbia Basin Ecosystem Management Project (ICBEMP) data, National Wetlands Inventory (NWI) data, EPA's STORET environmental data, StreamNet, and USGS stream flow data. Data elements that were identified as original source data (Table III.A.1) we summarized at the 6-HUC scale, preparing them for translation into conclusions about Level 2 environmental attributes. A more complete description of these data sources and how the data were summarized is found in a separate report by Battelle Pacific NW Laboratories (in preparation).

Some Level 1 environmental input data were classified as derived Level 1 variables (Table III.A.1) because they required separate calculations. For example, several channel variables (such as channel slope and channel length) required independent calculations with digital elevation maps in a Geographic Information System (GIS). Tributary runoff required use of a hydrologic model to estimate the flow coming from each of the 6-HUC units. Other variables associated with water quantity and quality (e.g., flow, temperature, sediment, and nutrients) required explicit flow routing calculations to incorporate the impact of upstream processes.

Stream channel morphometry (derived data)

Basic channel morphometry was characterized using Digital Elevation Models (DEM), Digital Line Graphs, and GIS techniques. Average channel length and slope of the main channel within each 6-HUC unit was derived using a 90-meter DEM, which was the scale of data available for the entire study domain at the time the analysis was done.[1] Channel slopes from the elevation of the main channel as it leaves and enters the 6-HUC unit were estimated, and channel lengths were estimated by summing the lengths of the stream segments composing the main channel and corrected for sinuosity (described below). In headwater 6-HUC units, the slope is based on the elevation of the highest main channel reach segment and the outlet elevation, placing constraints on both the maximum and minimum slopes. In certain 6-HUC units, inadequate data were available to estimate slope; in which case, we developed similarity relationships between other 6-HUCs based on area and elevation.

Estimates of sinuosity were needed to correct the estimates of channel length and slope described above. Channel lengths derived as straight lines between reach segment endpoints, as produced from the DEM, can significantly underestimate both total channel length and slope. We derived an approximation of sinuosity as a function of the combined density of urban and agricultural lands and the channel’s uncorrected estimate of slope. These land uses generally reduce sinuosity due to channel straightening and bank hardening practices. The function is expressed as a simple 2-dimensional lookup table, relating sinuosity to both land use and the uncorrected channel slope. Channels with high slopes are less likely to meander, resulting in lower sinuosity values.

Estimates of wetted channel width were derived using Manning’s equation, a standard hydrologic method. The equation relates wetted width to flow, cross sectional geometry, and Manning’s roughness coefficients. All channel cross-sections were assumed to have an inverted equilateral triangular shape, an assumption that tends to bias estimates on the low side. We believe that this technique, however, compensates for the possible overestimation of production in streams with wide shallow margins, resulting from the inclusion of all wetted area in the estimation of fish production. This method of estimating channel width incorporated no information about valley type—for example, whether the channel cuts across an alluvial plain or is tightly confined within a canyon. No reliable data were available to describe valley type for each 6-HUC unit. As these types of data become available, estimation of channel width using this method can be improved. We describe derivation of flow estimates, used in estimating width, below.

Data with temporal patterns and routed downstream (derived data)

Derivation of those environmental input data elements that are strongly related to stream flow, took into account seasonal runoff, temperature patterns, and flow route through the stream network. These stream flow-related attributes are runoff, natural flow, actual flow (minus diversions), water velocity, water temperature, fine sediment load, and nutrient load; they are discussed below.

Runoff defines the amount of flow generated from each 6-HUC unit into the stream network; runoff was estimated from the Distributed Hydrologic Soils Vegetation Model (DHSVM). In this model, flow is estimated before considering any anthropogenic activities (e.g. irrigation). Combining runoff from each 6-HUC unit with runoff from upstream 6-HUC units generates the estimates of natural streamflow, without consideration for water withdrawals. Each 6-HUC is divided into five elevation bands to estimate snowmelt where significant elevation changes occur, and estimates of daily minimum and maximum air temperatures and precipitation for Eastside drainages are incorporated. For the Westside drainages, data were obtained for several National Climate Data Center stations. Both of these records were combined to provide a 40-year daily climate record. The DHSVM was then used to simulate the entire 40-year record for all 6-HUC units in the basin with a 3-hour time interval. Runoff values were aggregated to a monthly time interval. A validation test performed with data from the Middle Fork Flathead River produced excellent agreement with empirical data (r²~0.90). Baseflow was assumed to be a fixed fraction of the total annual precipitation that is specified as a parameter for the entire study area (a report under preparation by Battelle Pacific NW Laboratories gives a complete description of the method).

Natural stream flow departing each 6-HUC unit (i.e., flow without consideration of any upstream diversion or regulation) was estimated by accumulating upstream flows and adding the 6-HUC-specific runoff on a monthly basis, providing simple mass conserving estimates of stream flow. This procedure could not be used to characterize extreme flow events; however, extreme flow event should be incorporated in future analyses. Also, recent improvements in DHSVMs should allow refinements in estimation of base flows. Currently, the DHSVM results are only validated for the Middle Fork Flathead River, Montana. Other sites with clean flow data records should be used to further improve confidence in the model's calibration.

Actual flows (i.e., flows as modified by upstream regulation and diversion) were more difficult to estimate because data for diversions are limited to annual values at the 4-HUC scale. The method employed required disaggregating procedures to distribute diverted water and water withdrawals to each of the 6-HUC units within their 4-HUC basin. In general, flow regulation resulting from upstream reservoir operations was simulated by extracting a fraction of each reservoir's storage capacity from the available streamflow during certain months (storage period) and returning it to the downstream 6-HUC units during other months (release period).

We accounted for diversion losses using the following rules:

·        Irrigation return flow was based on the relative 6-HUC areas defined as having a land use class of agriculture.

·        Conveyance loss return flows were distributed uniformly over all of the 6-HUC units within their 4-HUC.

·        Groundwater withdrawals were prorated based on total available base flow, including base flow from irrigation and conveyance losses.

·        Surface water withdrawals were prorated from each 6-HUC unit based on the available flow, including upstream flow.

Each of the forty years of runoff data were routed using the same diversion and regulation schedule to provide an estimate of the long-term interannual flow variability (Battelle Pacific NW Laboratories, in preparation).

Average water velocity within the main channel of each 6-HUC was estimated using the Manning’s equation referred to above, based on estimates of actual flow. Estimates of average velocity were developed for a specified slope, streamflow, channel cross-section geometry, and roughness coefficient. These estimates would be improved considerably by inclusion of valley type information, which was not available for this analysis.

Average water temperatures were estimated using a simple temperature model basedon the conservation of energy and the temperature equilibrium concept (Vail, 2000, pers. comm). This approach is most applicable in areas where irrigation withdrawals occur. Water temperature in each 6-HUC unit is estimated by adiabatic mixing of upstream flows with flows from the specific 6-HUC; water coming from various sources within the 6-HUC is assumed to have specific temperatures. Once the water from both local and upstream sources is mixed, the water is further altered by allowing it to transfer energy with the atmosphere based on the temperature equilibrium concept. Base flow (including irrigation return flow and conveyance losses) contributions are assumed to enter the river at the annual average air temperature. Surface water runoff (including irrigation) is assumed to take on the average monthly air temperature. Snowmelt is assumed to take on a temperature specified as a parameter. The temperature equilibrium concept allows for surface energy exchange based on the stream’s surface area (channel width times channel length), the stream’s residence time (channel length divided by velocity) within the 6-HUC, and the difference between the atmospheric and stream temperatures. The model did not incorporate effects of shading associated with riparian vegetation or unusually large inputs of natural groundwater; future uses of the model would require inclusion of these effects.

In estimating sediment load, supply of sediment was treated as a conservative, non-reactive constituent, neglecting important, but exceptionally difficult to accurately predict, processes of sediment deposition and re-suspension. Hence the estimation of Level 1 sediment load should be considered a relative measure without a specific metric.[2] This method assumes that the index employed behaves as a conservative, fully-mixed tracer of actual sediment loading. The local loading of sediment (from a 6-HUC unit) was assumed to be a function of the surface runoff and the relative fractions of various land use classes including urban, agriculture, range, and forest. The load from all land use classes was assumed proportionate to the fraction of the area with a high soil erosion hazard potential. Additionally, the sediment from forested lands was assumed to be proportionate to road density and the sediment from the range land proportionate to grazing intensity. Each land use class was given a separate sediment generation term, specified as a parameter. The resultant sediment load index was treated as a concentration, mixed with sources from upstream and routed downstream in a mass conserving procedure.

We estimated nutrient load associated with urbanization and agriculture using a similar procedure as that applied for sediment, treating nutrients as a single, conservative, non-reactive constituent. This estimation neglects the assimilative capacity of the stream to remove nutrients. Hence the estimates of Level 1 nutrient load should be regarded as relative and not tied to a specific metric. The local load of nutrients was assumed to be a function of surface runoff, baseflow, and the relative fractions of the land use classes urban, agriculture, and range. In addition, the input of nutrients from range land was assumed to be proportionate to grazing intensity. Each land use class was given a separate nutrient generation term, specified as a parameter. The resultant sediment load index was treated as a concentration, mixed with sources from upstream and routed downstream in a mass conserving procedure.

A more complete description of the methods applied to complete the Level 1 environmental input data sets can be found in the report from Battelle Pacific NW Laboratories (in preparation).

Translation to Level 2 environmental attributes

We translated Level 1 environmental input data into conclusions about Level 2 environmental attributes using a set of explicit rules, or in some cases, by summarizing directly into the categories defined by some environmental attributes (Table III.A.2). In the latter situation, the Level 1 data elements were the same as those defined by environmental attributes; hence no rule was required. The categorical conclusions defined for each environmental attribute are listed in Appendix B. The rules used for this translation procedure are described in a report under preparation by Battelle Pacific NW Laboratories.

Translation to Level 3 biological performance attributes

We formulated a set of rules for translating Level 2 environmental attributes into the survival-related values of Level 3 biological performance attributes (Table III.A.3) for chinook salmon (Appendix B).

We developed the rules by first identifying the specific Level 2 environmental attributes that were likely to be strongly, moderately, or weakly associated with each of the Level 3 performance attributes, for each life stage (Appendix B). These Level 2 attributes are referred to as the primary, secondary, and tertiary environmental attributes affecting biological performance, respectively.

Figure III.A.1 and Figure III.A.2 show an example using the Level 1 attribute, sediment yield, three sediment-related Level 2 environmental attributes, and the resulting Level 3 biological performance attribute, fine sediment (Table III.A.3, Figure III.A.1).

The Level 3 biometric, fine sediment, is an estimate of the contribution of all sources of fine sediment on survival during the egg incubation life stage (egg deposition to fry emergence), as shown in Figure III.A.1 above. Intragravel fine sediment is assumed to be the primary determinant of the effect of sediment on egg survival. Sediment effects are assumed to be increased in cases of high embeddedness or turbidity. Embeddedness would limit percolation into the area of the egg pocket, while high turbidity would overwhelm any gravel cleaning accomplished by the spawner in redd construction. Thus, the Level 2 attributes, embeddedness and turbidity, are considered as secondary, or modifying attributes.

Figure III.A.2 also shows the contribution of all sources of fine sediment on survival, but for the inactive life stage (overwintering). In this life stage, embeddedness is assumed to be the primary determinant of the effect of sediment on the survival of overwintering fingerlings (loss of interstitial space). Sediment effects are assumed to be increased in cases of high turbidity (secondary) due to impairment of respiration or feeding. Intragravel fine sediment (tertiary) is assumed to further reduce survival in this life stage due to a reduction in deeper interstitial spaces.

The rules for translating Level 2 environmental attributes into the Level 3 biological performance attributes are based on an extensive set of translation examples put together by Chris Frissell, with further input obtained from the BioRules Work Group.[3] We reformatted the information from these data sets into the rule set applied here for chinook salmon, taking into account refinements in the definitions and index values of the Level 2 environmental attributes. The rules should be considered still in a state of development and refinement. Further review of the rules by regional scientists will help ensure their adequacy and consistency with up-to-date thinking and research on the effects of the attributes on salmonids. Moreover, we propose that a forum be developed to routinely review and update the rule sets as a way of promoting learning about the effects of the ecological attributes on fish and wildlife performance. Such a forum would help gain wider acceptance of the application of these kinds of rule sets for assessing the effects of environmental change on biological performance.

The full set of Bio-rules, as they are currently configured, is provided in Appendix B. Similar rule sets are under development for bull trout, steelhead trout, chum and coho salmon. Extending the chinook rule set to other species is a straightforward task involving consideration of relative sensitivities of different species to the common set of environmental attributes based on species differences in behavior, physiology, and size.

Mainstem – Columbia River and Snake River Habitat

Methods used for developing habitat and survival attributes for fish utilizing the mainstem Columbia and Snake rivers are presented below for both the Historic Potential and Current Potential.

Habitat Quality and Quantity

Biological rules do not exist for deriving mainstem habitat ratings; therefore, we constructed the quality ratings for mainstem habitat, for the Historic Potential and Current Potential, based on existing literature and the professional expertise of fisheries biologists familiar with both Columbia and Snake river systems. The biologists used the existing data and their knowledge to rate the following biological performance attributes for each river reach of interest:

·        Habitat Quality

·        Temperature

·        Predation

·        Competition with Hatchery Fish

·        Competition with Other Species

·        Habitat Diversity

We adjusted these ratings up or down to meet the juvenile system survival values presented below. A summary of the ratings for all river reaches rated is presented in Appendix C.

The quantity of both riverine and reservoir habitat presented under both conditions were estimated from USGS Topo maps, average monthly river flow, and reservoir size and length data presented in the CRiSP 1.5 manual (Anderson et al., 1996).

River Flow

We obtained estimates of average monthly river flow for both the Columbia and Snake rivers under the Current Potential and Historic Potential from streamflow model runs developed by Council staff (Appendix D). The flow data used in modeling both conditions are summarized graphically in Table III.A.4 and Table III.A.5.

Juvenile Travel Time

We assumed that the time required for subyearling and yearling chinook to migrate through the mainstem corridor is affected by river flow (water velocity) and habitat types present (i.e., riverine or reservoir). Thus, juvenile migration speed is assumed to differ under the Current Potential  (primarily reservoir) and Historic Potential (riverine).

We developed subyearling and yearling chinook travel speeds for both conditions using CRiSP Model 4. A description of the model, inherent assumptions, formulas and inputs can be found in Zabel et al. (1997). In addition, for the Historic Potential, we estimated water velocity by dividing average monthly river flow by the average cross section of each stream reach. We used travel speed and timing data in this analysis to determine the survival conditions encountered by each juvenile as it migrates through the mainstem Snake and Columbia rivers.

Juvenile Migration Timing

We approximated subyearling and yearling juvenile migration timings from data developed by the Fish Passage Center (FPC 1998). These data are summarized in Table III.A.6 and were used in modeling both the Current Potential and Historic Potential.

Dam Survival (Juveniles)

Dam survival rates for juvenile salmonids are discussed below for the Current Potential only; dams do not exist for the Historic Potential, thus survival estimates are not needed for that condition.

The survival rate of juvenile salmonids migrating past Columbia and Snake river hydroelectric projects is dependent on riverine conditions, juvenile behavior, and physical facilities present at each project. We calculated both yearling and subyearling survival rates through spillways, turbines, and juvenile bypass systems for each project using data presented in the NMFS (2000a). The monthly survival values used in this analysis for both yearling and subyearling life-history patterns are shown in Table III.A.7 and Table III.A.8. It should be noted that the survival values do not include the mortality component associated with juvenile passage through reservoirs.

Dam Survival (Adults)

Adult chinook survival past each mainstem dam was assumed to average 93 percent under the Current Potential. Thus, total adult survival through mainstem river reaches is highly dependent on the number of dams each adult must pass. For example, adult chinook returning to the Salmon River would have to pass eight mainstem dams, and thus their overall survival rate would be 60 percent (0.98⁸ = 60 percent). In contrast, the survival rate for adults returning to the John Day River would be approximately 80 percent because they must migrate past only three mainstem dams.

Under the Historic Potential, adult chinook survival through the mainstem Columbia and Snake Rivers was assumed to average 92 percent.

In-river Survival (Juveniles)

The survival rates used for modeling the Current Potential for subyearling and yearling juveniles migrating in-river through the hydroelectric complex were based on the range of values presented in recently published scientific literature.

Data presented by NMFS (2000a) show that from 1993-1999 yearling survival from Lower Granite Reservoir to the tailrace of Bonneville Dam ranged from about 31 percent to 51 percent. This equates to a project survival rate of approximately 86 percent to 92 percent. For modeling the Current Potential, we assumed that yearling survival past eight hydroelectric projects averages 36 percent (88 percent per project).

For subyearling chinook we assumed that in-river survival from the head of Lower Granite Reservoir to the tailrace of Bonneville Dam was 29 percent. This equates to a project survival rate of ~85 percent. The survival value only applies to active migrants. For inactive migrants, or life history trajectories that spend more time in the reservoirs (rearing stage), mortality increases in proportion to the time spent in the reservoirs. Thus, overall survival varies dependent on the trajectory examined. This approach is consistent with the data presented in a recent NMFS document (NMFS 2000a). NMFS scientists reported that subyearling survival varied dramatically (13-51 percent) in tests conducted in the Snake River from 1995-1999. However, these survival estimates included mortality from parr to the active migrant stage.

The juvenile survival rates presented above formed the basis for model calibration with regard to overall survival through the mainstem Columbia and Snake Rivers. Because the dam survival values were fixed, the overall survival targets for both life histories required that juvenile survival rates through the reservoirs be adjusted as needed, which we achieved by modifying the habitat quality attributes for each reservoir during the key juvenile migration periods (see Juvenile Migration Timing). Resulting reservoir survival values for the Current Potential are presented in Table III.A.9 and Table III.A.10 for yearling and subyearling chinook, respectively. It should be noted that juvenile survival through the reservoirs is affected by the amount of time the juvenile spends in the reservoir and the benchmark survival value for the specific life stage (subyearling, yearling, etc.).

We set the survival benchmarks for yearling and subyearling chinook at 97.5 percent and 35 percent, respectively. These benchmark survival values were based on the assumption that yearlings require 14 days, and subyearling 56 days, to migrate from natal streams to the estuary under ideal environmental conditions. This equates to a daily survival rate of 99.8 percent (97.5^1/14) for yearlings and 98.1 percent (0.35^1/56) for subyearlings^.

For each reservoir, we calculated the daily survival rate for juvenile chinook using the following formulas:

Daily Yearling Survival Rate = (B^1/14*RSR^1/30)

Daily Subyearling Survival = (B^1/56*RSR^1/30)

Where-          

B= benchmark survival rate

RSR = Reservoir survival rate by month

 

Yearling and subyearling juvenile chinook survival rates used for modeling the Historic Potential are presented in Table III.A.11 and Table III.A.12. We calculated the survival values based on mainstem habitat quality, juvenile travel time through each reach, and the benchmark survival values used for each life stage.

Combining the survival data in Table III.A.11 and Table III.A.12 results in the survival estimates presented in Table III.A.13 for yearling and subyearling chinook migrating from above either Lower Granite or Wells dam to the tailrace of Bonneville Dam.

Fish Transportation (Juveniles)

Survival associated with juvenile fish transportation is presented below for Current Potential only; juvenile transport does not occur under the Historic Potential.

The percent of the yearlings and subyearling collected at each of the four lower Snake River and McNary Dam facilities is presented in Table III.A.14. The values in Table III.A.14 represent the percent of the juvenile population arriving at each facility that is collected and transported to the tailrace of Bonneville Dam.

We assumed 98 percent of the transported juveniles survive to the point of release (NMFS 2000b). We also assumed survival rates of transported Snake River yearling and subyearling chinook once released from the barges are 50 percent and 35 percent that of juveniles migrating in-river, respectively. We selected these values based on a review of recent literature estimating the differential post-Bonneville Dam survival for in-river and transported juvenile salmonids. The 50 percent value we used for yearling chinook was based on data presented in Bouwes et al. (1999). The subyearling value (35 percent) was based on data presented in PATH (1999). We increased the transport survival rate for subyearlings transported from McNary Dam to 60 percent to maintain a transport survival benefit for subyearling chinook migrating from the mid-Columbia River.

Marine

We present the information on the three components of the marine environment listed below:

1.           Estuary

2.           Nearshore

3.           Ocean

The nearshore area was used to describe the early ocean life of juvenile salmonids (period from ocean entry to December 31).

Because biological rules were not developed for these areas, we used data from the literature and professional expertise to determine juvenile survival in each component of the marine environment. These survival rates were applied to each of the 74 salmon stocks analyzed.

For the estuary, biologists determined impacts to salmonids by developing ratings for a subset of the biological performance attributes. The ratings were based on USGS river flow data, river temperature information, the results of bird predation studies conducted near the mouth of the Columbia River (Roby et al. 1998) and marine mammal predation studies (reviewed in Park 1993). Ratings for juveniles and adults are summarized in Table III.A.15a, Table III.A.15b, and Table III.A.15c.

Chinook ocean survival rates used for modeling purposes beginning with the first full year in the ocean were the same as those used by the Pacific Salmon Commission Chinook Technical Committee. The derivation of these rates is undocumented but are used by the CTC for chinook cohort analysis, thus are consistent with their ocean modeling exercises. The rates are summarized by age (shown are ages for ocean type life history) in Table III.A.16.

We developed the nearshore/early marine survival rates based on the unexplained residual from known estuary and marine survival rates of the expected SAR rates for natural Columbia River chinook. Subyearling and yearling early marine rates were 55 percent and 40 percent, respectively.

Coarse Screening of Environmental Attributes

As noted previously, we derived the environmental attributes from easily obtainable data for each of the five provinces. The quality of these data varied by attribute, 6-HUC, and province; and it was determined early in the process that the quality was sufficient for conducting an analysis at the basin and province scales only. To help improve the quality of this data set, fisheries biologists familiar with the stream habitat present in each province reviewed a subset of the derived environmental attribute data set. Specific environmental attributes reviewed included:

1.           Fine Sediment

2.           Bed Scour

3.           Low Flow

4.           Riparian Function

5.           Maximum Temperature

We selected these attributes for review based on past EDT analyses, where ratings for these attributes have shown a relatively large effect on salmon survival and resulting salmon performance. The biologists reviewed each of the five attributes for the 822 6-HUCs that are used by chinook salmon. Each biologist examined the data set and made changes to the attributes based on available data and professional opinion. We incorporated all of the changes in the environmental attributes proposed by the review biologists into the final analysis; they are, therefore, reflected in the modeling results.

Development of Hatchery Attributes

The EDT approach addresses the hatchery environment and hatchery reared populations the same way it addresses natural habitat and wild populations. Survival conditions in the hatchery environment are captured in the form of biological performance attributes, and hatchery populations are analyzed based upon the hatchery and natural environments available to them. Obtaining input data from individual hatchery facilities was beyond the scope of this study and we, therefore, focused on the performance of hatchery populations after their release.

Survival of Hatchery Fish

Where studies of direct comparisons between hatchery and natural populations in the natural environment are sparse, it is generally assumed that post release survival of hatchery fish is less than that for wild fish over the same life stages. This difference can be attributed to both first generation (non-genetic) and trans-generation (genetic) factors. Our assumptions here relied largely on the approach and conclusions presented in RASP (1992). In Table III.A.17, the genetic and non-genetic factors have been combined into one survival multiplier for post release survival. In the population analysis, we apply this additional rate of mortality from the time of release until the end of the first year in the ocean.

Based on experimental results with new culture practices intended to improve survival of hatchery fish we assumed that future supplementation programs would achieve somewhat higher survival.

Effects of Hatchery Fish on Wild Populations

In the EDT analysis, hatchery fish can affect wild/natural populations through ecological or genetic interactions. Ecological interactions involve competition for food and space, predation (directly or indirectly by affecting behavior of predators), and ecological function. Genetic interactions result from hatchery fish interbreeding with wild fish in the natural environment.

We estimated competition effects due to hatchery fish based on estimated densities of hatchery juveniles by stream reach over time and on maximum densities drawn from the literature. We computed the density of hatchery fish from time and rate of release of hatchery fish at each facility and estimated rates of downstream movement of those fish. Using the Beverton-Holt survival function and benchmark maximum density parameters, we estimated the survival impacts on wild fish for every stream reach and time period. We did not include direct and indirect effects of predation in this analysis. We did include ecological effects due to nutrient enhancement from carcasses (positive increase in survival) and due to pathogens associated with hatchery programs as direct, site-specific inputs.

Hatchery fish access natural spawning grounds inadvertently through straying or as a result of supplementation with the intent to augment natural spawning. We relied on RASP (1992) for estimates of the survival (fitness) effect on natural populations of hatchery introgression as a function of the hatchery-natural composition of the spawning population (see Table III.A.17). In order to calculate the ratio of hatchery to wild and compute the demographic contribution of hatchery spawners to the subsequent generation, we somewhat arbitrarily assumed that the total escapement (hatchery plus natural) to the spawning grounds would not exceed the natural spawner capacity.

Hatchery Production

The total number of hatchery fish by species released in each alternative and condition is presented in Table III.A.18a, Table III.A.18b, and Table III.A.18c.

Supplementation

The future alternatives all assume that some of the returning hatchery fish will spawn with naturally produced fish in the wild (Figure III.A.3).

Development of the Harvest Attributes

We obtained the data used in this analysis to determine the rate and location of adult harvest from the following sources:

·        Fisheries Regulatory Assessment Model (FRAM)

·        Chinook Technical Committee, Pacific Salmon Commission

·        Status Report, Columbia River Fish Runs and Fisheries, 1938-1997. WDFW/ODFW

·        1996 All Species Review, Columbia River Fish Management Plan. US V. Oregon, Technical Advisory Committee, 1997.

·        Biological Assessment, Technical Advisory Committee. 1998

For this analysis, we defined the harvest rate base period to be 1992-1996 developed harvest rates for both ocean and mainstem Columbia River fisheries (Zones 1-6). We based the harvest rates used in this analysis on published rates for ten Columbia River Harvest indicator stocks (Table III.A.18a, Table III.A.18b, Table III.A.18c). The data presented in Table III.A.19 show the specific indicator stock that we used in setting harvest rates for each of the 74 fish populations examined in this analysis. The analysis does not include estimates of sport or commercial harvest in the tributaries. Thus, the adult run sizes reported for each province are based on the number of fish entering each tributary.

Population Structure

The EDT analytical model is based on the analysis of life history pathways through the environment. The analytical model includes a Trajectory Generator module that generates multiple pathways, referred to as trajectories, through space and time. Each trajectory may vary in the duration, rate of travel, and timing of life stages (Figure III.A.4).

We use the term life history pattern to mean a collection of similar trajectories (Figure III.A.5). These trajectories share life history behaviors, such as ocean entry timing (e.g., age at ocean entry or seasonal timing) or migration pattern during freshwater residence (e.g., freshwater residence in natal stream or redistribution to non-natal stream for “overwintering”).

Finally, the uppermost level of organization of biological performance is the population. Trajectories are grouped into loosely defined populations based on common geographic area and common life history pattern (i.e., spring chinook in the Upper Yakima basin). We describe populations based on available documentation for a basin (status reports, harvest management units, etc).

General Chinook Life Histories

Chinook salmon exhibit a wide variety of life history patterns (Reimers 1971; reviewed by Healey 1991). At the most basic level, chinook salmon life histories are defined by stream type and ocean type behavioral patterns (Healey 1991; first described in Gilbert 1913). Stream type chinook remain in freshwater for one year before migrating to sea, which is typical of northern populations and headwater tributaries of southern rivers. Ocean type chinook migrate to sea during their first year of life, which is more common in coastal streams and rivers south of 56°N.

Taylor (1990a) hypothesized that age of seaward migration (stream vs. ocean type) is environmentally modulated (temperature and photoperiod) and shows an inheritable component to differences in growth rate and agonistic behavior between stream and ocean type chinook (Taylor, 1990b). Clarke et al. (1992) demonstrated an inheritable response to photoperiod and resulting saltwater tolerance. We conclude that, at the population level, the proportion of stream and ocean type life history patterns should be specified in the EDT model; the environment as described in the model is insufficient to define the frequency of these types.

Additional variation within these life history types is common for chinook. Reimers (1971) described four patterns for ocean type chinook in the Sixes River. Healey (1991) expanded on the general life history types by recognizing a tactical component to the life history model, defining additional variation within these types as adaptations to uncertainties in juvenile survival and productivity among habitat types. Under this hypothesis, expression of multiple behavioral patterns is a function of the environment. Genetically identical ocean type populations may have a different suite of expressed life history patterns depending upon the environment they encounter.

The EDT analysis includes variation in life history patterns within a life history type in our analysis of chinook performance. We hypothesize that restoration of lost life history patterns is largely a function of reestablishing connectivity of the habitat, where fragmentation of the habitat may be the result of physical (e.g., dams) or biological (e.g., temperature affected) constraints on migration or utilization.

Application to the Columbia River

Application of variation in life history patterns within the EDT analytical model is largely a function of identifying life stage durations, migration travel speeds, and timing windows. This section addresses the input parameters used in the EDT model to define a set of chinook patterns. Parameters are intended to be broad so that a pattern can be applied to multiple locations within a basin (e.g., Hanford Reach or Snake River fall chinook populations).

When a range of input values is applied, the Trajectory Generator module randomly selects a value within the range. Distribution within the range is assumed to be uniform for life stage duration and timing windows. Thus, when river entry timing is said to extend from March to May, trajectories will be generated across all dates. Migration rates (travel speed) use a non-uniform distribution. We assume that a majority (~75 percent) of the trajectories will travel in the lower 25 percent of the range (Figure III.A.6).

Stream Type Patterns

The suite of available patterns presented in the previous section can be summarized as differences in duration and migration speeds, at key life stages. For example, fry colonization (2-week period immediately following emergence) can occur very quickly and with movement downstream of less than 10 meters. In contrast, the life stage may extend beyond two weeks (but less than 3 ½ weeks), and fry may colonize locations several kilometers downstream of emergence.

We included four basic patterns for generating stream type trajectories (Figure III.A.7):

·        Resident—remain in natal stream throughout freshwater residence.

·        Spring Dispersal—dispersal to downstream habitat during fry colonization followed by summer and winter rearing at the same location.

·        Fall Redistribution—remain in natal stream through summer followed by downstream redistribution in fall; movement can vary from <1 kilometer to ~10 kilometers).

·        Spring-Fall Dispersal—dispersal to downstream habitat during fry colonization; summer rearing followed by downstream redistribution in fall).

Ocean Type Patterns

The range of patterns reviewed from the Sixes River (Reimers 1971) can be summarized as differences in spring and early summer rearing (the range is from high rate of movement to a resident type pattern of summer rearing). Differences in estuarine residence can be described as rate of travel within this habitat type. Dispersal during fry colonization is assumed to occur for all ocean type patterns (Figure III.A.7).

The EDT analysis was limited to populations upstream of Bonneville Dam. The distance of these populations from the estuary lead us to conclude that rearing of subyearling chinook was largely riverine; time spent in the estuary was constrained to 2-4 weeks. Reimers (1971) described patterns of extended estuarine rearing (6-10 weeks), which are more common in coastal rivers. The exception was subyearling chinook collected early in the transportation system; these trajectories were constrained to spend a longer period of time (4-6 weeks) in the lower river-estuary.

Chinook Population Descriptions

We included 66 chinook populations in the analysis. Several of these existed only in the historic conditions as they are blocked by dams or inundated by reservoirs. Populations were not described for upstream of Chief Joseph or Hells Canyon. Assumptions of juvenile age (life history type), adult river timing, and spawning timing differed among populations. Each population and key assumptions are described in Table III.A.20.

Analysis Method

The modeling component of EDT was used to produce estimates of chinook productivity, capacity, and diversity for both the Historic Potential and Current Potential (Lestelle et al., 1996). A more detailed description of the EDT Model including formulas, assumptions, and inherent workings is included in Appendix A.

Uncertainty and World View Assumptions

The major assumptions used in modeling the three worldviews (Technology Pessimistic, Moderate, and Technology Optimistic) are shown in Table III.A.21 and Table III.A.22 below. Model runs based on these worldviews were completed for the Historic and Current Potentials, as well as for Alternatives 2, 5, and 6. A more detailed discussion of the assumptions used for the Technology Pessimistic and Technology Optimistic worldviews is offered below. Assumptions used under the Moderate scenario are included in the tables for reader convenience. The Moderate assumptions were discussed earlier in this section and are, therefore, not repeated here. Dam survivals by Alternative and worldview are shown in Appendix E.

World View Assumptions

Technology Pessimistic

The Technology Pessimistic worldview will be discussed for In-river Transport Survival, Hatchery Fish Parameters, Habitat, and Marine.

In-river and Transport Survival

The in-river and transport survival rates used for modeling subyearling performance were based on data developed by PATH and NMFS for fall chinook (PATH 1999, NMFS 2000b). Based on our review of these analyses, we set typical subyearling in-river and transport survival for Snake River stocks at 27 percent and nine percent, respectively. Transport survival at McNary Dam was set at 40 percent in order to more closely match the expected in-river survival rate for both subyearlings and yearlings that migrate from this facility to the tailrace of Bonneville Dam. Under the Technology Pessimistic worldview, it is assumed that in-river survival is at the low end of recent estimates and that transport is ineffective.

Hatchery Fish Parameters

The post-release survival rate of hatchery fish under this worldview was set at 10 percent and 15 percent that of naturally produced fish for juveniles reared under conventional and supplementation type facilities, respectively. It was also assumed that as hatchery fish abundance increases on the spawning grounds, wild fish fitness decreases (Table III.A.22). The assumption under the Technology Pessimistic worldview being that hatchery fish have low survival and negatively affect the fitness of wild stocks.

Habitat

The biological rules that translate environmental attributes into survival parameters produce an estimate of relative productivity (based on Moderate assumptions) for each 6-HUC, for each month, and for each life stage. The assumed range of uncertainty of this estimate is shown in Figure III.A.8. The Technology Pessimistic worldview, assumes the lower values in this range, reflecting a greater sensitivity to habitat conditions that deviate from the optimal for each life stage.

Marine

Estuary survival rates for juvenile chinook were altered as described under the habitat section above.

The nearshore survival values used for modeling subyearling and yearling survival through this area were increased to 86 percent and 74 percent, respectively. Nearshore survival rates were changed to increase the number of adults returning to the Columbia River under the Technology Pessimistic worldview so that they were similar to those produced under the Moderate and Technology Optimistic worldviews. This step was needed to meet the assumption that each of the worldviews produces similar numbers of fish under the Current Potential. Therefore, the worldviews agree on how many fish are being produced, but disagree on why (i.e., inherent assumptions vary by worldview). In the Technology Pessimistic worldview, it is assumed that nearshore ocean conditions have less an effect on adult run sizes than factors such as hydro development and habitat degradation.

Ocean survival rates used for modeling this worldview were identical to those described for the Moderate worldview.

Technology Optimistic

The Technology Optimistic worldview will be discussed for In-River and Transport Survival, Hatchery Fish Parameters, Hatchery, and Marine.

In-river and Transport Survival

The yearling in-river and transport survival values used for modeling the Technology Optimistic worldview was obtained from values developed by the NMFS (NMFS 2000b). NMFS scientists reported that in-river survival for yearling chinook migrating from the Lower Granite Reservoir to the tailrace of Bonneville Dam averaged approximately 51 percent for migration years 1997 to 1999. NMFS also reported that the “differential post-Bonneville Dam survival”, or so-called D-value, ranged from about 78 percent to 83 percent. In this worldview it was assumed that the post-release survival rate of transported yearling chinook is 80 percent.

Subyearling chinook in-river and transport survival rates were set for Snake River stocks at 35 percent and 60 percent, respectively. These values were deemed to represent the high end of recent estimates for these parameters. The McNary transport survival rate was set at 80 percent to match the values used for yearling chinook and to ensure a transport survival benefit for all stocks.

Hatchery Fish Parameters

Under the Technology Optimistic worldview, the survival rates for hatchery fish reared using conventional or supplementation type rearing practices were set at 50 percent and 60 percent. The Technology Optimistic worldview also assumes that hatchery fish impacts to wild populations are less severe than assumed under the Moderate worldview.

Habitat

The Technology Optimistic worldview, assumes the higher values in this range, reflecting a lesser sensitivity to habitat conditions that deviate from the optimal for each life stage (Figure III.A.8).

Marine

Estuary survival rates for juvenile chinook were altered as described under the habitat section above.

The nearshore survival values used for modeling subyearling and yearling survival through this area were reduced to 35 percent and 20 percent, respectively. Nearshore survival rates were lowered to decrease the number of adults returning to the Columbia River under the Technology Optimistic worldview so that they were similar to those produced under the Moderate and Technology Pessimistic worldviews. This step was needed to meet the assumption that each of the worldviews produces similar numbers of fish under the Current Potential. Thus, the worldviews agree on how many fish are being produced, but disagree on why (i.e., inherent assumptions vary by worldview). A major assumption under the Technology Optimistic worldview is that poor nearshore ocean conditions are a major factor responsible for the low adult returns observed in recent years.

Ocean survival rates were not changed under this worldview and are therefore the same as those used for modeling the Moderate worldview.

Data Quality

The amount of effort required to accurately rate habitat quality for over 259,000 miles of terrestrial and aquatic habitat is indeed daunting. The task becomes even more difficult given that the data needed to rate all 45 of the environmental attributes do not exist for some areas, are of poor quality in others, were collected over varying time frames, and assembled by multiple agencies using various methodologies. In short, there is considerable uncertainty in the quality of the data used in this analysis.

This problem was recognized at the start of the project but after reviewing the available data, we deemed it sufficient for conducting an analysis at the basin and province levels. However, data resolution is insufficient to draw inferences about habitat and salmon performance at the subbasin, watershed or reach level. The Council envisions that data quality problems will be corrected to the extent possible during the assessment phase of the Council’s program.

The data and analysis are sufficient to achieve the goal of providing planners and biologists the tools needed to identify, analyze, and prioritize actions to recover salmonid populations in their respective basins. Work products or tools resulting from this process include; 1) a future vision for the basin, 2) a set of Scientific Principles to guide recovery actions, 3) a Conceptual Framework, 4) analysis methodology, and 5) a draft data set for review and refinement for each subbasin of interest in the Columbia River Basin.

(Back to Table of Contents) 

---