U.S. Dept Commerce/NOAA/NMFS/NWFSC/Publications
Tech Memo-22: Status Review for Mid-Columbia River Summer Chinook Salmon

Summary of Biological Information

Environmental Features

The Columbia River is the third largest river in North America and drains an area of approximately 668,000 km2. This area includes British Columbia, Idaho, Washington, Montana, and Oregon, and smaller sections of Wyoming, Nevada, and Utah. It flows through or borders on three physiographic regions of the Pacific Northwest: the Rocky Mountain System, the Intermontane Plateau, and the Pacific Mountain System (Scott et al. 1989). Originating in Lake Windermere, B.C. in the northern Rocky Mountains, the Columbia River flows through British Columbia for over 1,000 km before entering the United States via the Okanogan Highlands. The river flows westerly until it turns south, approximately at its confluence with the Okanogan River (RKm 859), where it forms a boundary between two distinct ecoregions: the Columbia River Basin (part of the Intermontane Plateau) on the east, and the Cascade Mountains (part of the Pacific Mountain System) on the west. Approximately 25 km below its confluence with the Snake River (RKm 522), the Columbia River turns westerly towards the Pacific Ocean. For the purposes of this Biological Status Review, the MCR is defined as the mainstem river and tributaries between McNary Dam (Rkm 470) and Chief Joseph Dam (Rkm 878; Fig. 1). This definition was previously used by Mullan et al. (1992b) and Chapman et al. (1994).

The western side of the MCR is generally mesic, alpine habitat. Rivers originating there drain the eastern slopes of the Cascade Mountains as relatively short streams that begin precipitously and make a transition to low gradient streams in the lower reaches. These rivers receive the majority of their runoff from snowmelt in the spring and early summer. The five major east-slope rivers of the Cascade Mountains are the Yakima, Wenatchee, Entiat, Methow, and Okanogan (Fig. 1). General habitat features of these rivers are presented in Table 1 (see also Chapman et al. 1994).

Figure 1
Figure 1. Map of mid-Columbia River Basin, showing principal tributaries and hydroelectric facilities.

The eastern side of the MCR is a basaltic plateau that reaches 763 m in elevation and is principally xeric, sagebrush- grassland habitat. Between the Grand Coulee Dam and the Snake River, there are no chinook salmon streams entering the MCR from the eastern side.

In general, the five major MCR tributaries are not biologically productive. At certain size-at-age classes, resident trout populations in the Methow, Entiat, and Wenatchee Rivers have standing crop levels among the lowest ever reported (summarized in Mullan et al. 1992b). The habitat quality index score for these three rivers was 47 on a scale of 11 to 113, indicating low overall potential for salmonid spawning and rearing (Mullan et al. 1992b). Among MCR tributaries, the Okanogan River, which flows through four mainstem lakes, is somewhat more fertile. The Yakima River is presently enriched due to agricultural runoff and reservoir storage. Overall water quality in the other petitioned streams is excellent (Mullan et al. 1992b).

Table 1. Habitat characteristics of the mid-Columbia River Basin and tributaries inhabited by ocean-type chinook salmon (from Bryant and Parkhurst 1950; Davidson 1953; CBIC 1957; Mullan et al. 1992b; Chapman et al. 1994).

River: Okanogan Methow Chelan Entiat Wenatchee Yakima Columbia
859 843 810 779 754 540 470-
Gradient low steep/
low steep/
Terrain grass alpine/
grass alpine/
Runoff snow/
reservoir snow/
reservoir reservoir
--a 13.9 n/a 3.5 19.8 36.8 2405b
Climate arid wet/
arid wet/
Productivity med. low low low low high low/
Temperature low/
med. low/

a Dash indicates data not available.
b At Rock Island Dam.
By the turn of the century, sawmill, hydroelectric, and irrigation dams had already decimated salmon populations in this area. Since then, the condition of spawning habitat has improved greatly (Mullan et al. 1992b). The small dams on tributaries have been removed, irrigation diversions have been screened, and riprap has been placed over eroded stream banks, providing critical summer and winter habitat for juveniles (Chapman 1989). Mullan et al. (1992b) concluded that, with the exception of the Yakima River, degradation of habitat in MCR tributaries does not appear to be a significant cause of run depression. In fact, these authors consider that the area covered by the petition is currently at or near maximum historical smolt production for chinook salmon.

The Yakima River differs somewhat from other MCR tributaries. At its headwaters, the streams are steep and drain mountains about 1,500 to 3,000 m in elevation. The river valleys then flatten out and meander down gentle slopes into the Columbia River. As previously mentioned, the overall bioproductivity of the Yakima River is higher than that of other MCR streams because of agricultural runoff and lake and reservoir storage. However, these factors may also lead to increased water temperature, which inhibits salmonid migration and spawning (Mullan et al. 1992b).

Production of late-run chinook salmon in the Yakima River has also been severely curtailed by unrestrained irrigation practices. These practices include the use of unscreened irrigation diversions and exploitation of river water to an extent that produces low flows, which diminish summer habitat (Robison 1957). Although some of these irrigation practices have been corrected, no "summer" chinook salmon (i.e., early part of late run) have been observed in the Yakima River Basin since the 1970s, most likely due to the presence of inhospitable thermal conditions for adult chinook in the lower river (Busack 1990).

Life History Characteristics

Detailed life history data (age at spawning, sex ratios, etc.) are plentiful for many hatchery populations of MCR ocean- type chinook salmon, but data are limited and inconsistent for wild populations. Life history characteristics were specifically identified as "critical data gaps" for most subbasins in the production plans of the Columbia Basin Fish and Wildlife Authority (CBFWA 1990). Howell et al. (1985, p. 449) summarized the situation for Columbia River summer-run fish: "Basic juvenile and adult life history information is almost completely unknown for naturally produced summer chinook."

Considering the long history of salmon management by various fisheries entities, the paucity of basic biological information is both surprising and discouraging. In the context of an ESA biological review, this lack of information hampers the identification of distinct population segments or ESUs. Efforts to gather detailed life history information have only recently been initiated.

Juvenile Life History Characters

Chinook salmon populations have been separated into two basic types based on juvenile life history characteristics: those whose juveniles migrate to sea as subyearlings, known as "ocean- type" populations; and those whose juveniles migrate to sea as yearlings, designated as "stream-type" populations (Gilbert 1912, Taylor 1990, Healey 1991). Ocean-type chinook salmon in the MCR Basin spend most of their ocean life in coastal waters, returning to fresh water a few months prior to spawning. Stream-type fish, on the other hand, perform extensive offshore migrations, returning to fresh water many months prior to spawning (Healey 1991).

A strong tendency toward one or the other of these types is also found within most chinook salmon populations outside the MCR. Ocean-type populations dominate the southern range of the species from California through the coastal streams of Oregon and Washington, and stream-type fish dominate the range from approximately 56° N in British Columbia through Alaska (Taylor 1989, Healey 1991). However, in the southern portion of its range, stream-type chinook salmon are relatively common in upstream areas of most large rivers, while small rivers contain primarily ocean-type fish. Stream-type populations also appear to predominate in Asian representatives of the species (Healey 1991).

Variations in stream temperature regimes due to latitude or altitude appear to be the major factor controlling the general distribution of the two types. However, Healey (1983) suggested that other factors, such as distance of the spawning migration, annual river-discharge cycles, and ocean migration patterns may also be important.

In North America, the Columbia River is located near the middle latitudes of the chinook salmon range. The Columbia River is inhabited by populations with high diversity in juvenile migrational behavior and timing: both stream- and ocean-type populations inhabit the basin. As in areas outside the Columbia River, stream temperatures, which vary with elevation, appear to control the distribution of the two types. Mainstem areas and lower tributary streams of the Columbia and Snake Rivers produce only ocean-type juveniles, and upper tributaries of the Columbia and Snake Rivers produce only stream-type juveniles.

However, some tributaries, including the MCR streams listed in the petition, produce both types. In both the mid-Columbia and Snake Rivers, spring-run chinook salmon produce stream-type juveniles, and fall-run chinook salmon produce ocean-type juveniles. However, the so-called "summer-run" adults produce ocean-type juveniles in the MCR above McNary Dam and stream-type juveniles in the Snake River.

In summary, available life history information indicates a strong affinity between fish designated as summer- and fall-run in the MCR, and between spring- and summer-run fish in the Snake River (Matthews and Waples 1991). For example, ocean-type chinook salmon in the mainstem Yakima River exhibit life-history and spawning characteristics similar to those of ocean-type fish in the Hanford Reach of the Columbia River. Genetic data (discussed below) also support the hypothesis that these affinities correspond to ancestral relationships.

Run and Spawn Timing

The temporal distribution of adults as they enter fresh water to spawn is referred to as run timing. Historically, chinook salmon have entered the mouth of the Columbia River almost continuously. Commencing in February, the run peaked in mid-June and ended in late November (Thompson 1951, Mullan 1987). In general, early-returning fish were stream-types destined for upper tributary areas, while late-returning fish were ocean-types destined for mainstem areas (Fulton 1968).

However, during the peak of chinook salmon migration, a large number of both run-types migrated upstream, filling a large portion of the spawning habitat. Thus, the so-called "summer- run" was a mix of chinook salmon composed of late-migrating stream-type fish destined for upper tributaries of the Snake and Columbia Rivers and early-migrating ocean-type fish destined for lower tributary and mainstem areas of the middle Columbia and Snake Rivers.

Since the turn of the century, human activities (i.e., overfishing, dam building, etc.) have severely fragmented or dislocated portions of the ancestral continuum of migrating chinook salmon in the Columbia River, leaving what now appears as noncontinuous or discrete populations. The middle portion was depleted by early commercial harvests, leaving the early (spring- run) and late (fall-run) portions separated as semidiscrete run groups (Thompson 1951, Beiningen 1976).

Because chinook salmon spawning coincides with a declining temperature cycle (Miller and Brannon 1982), temperature variation, controlled primarily by elevation, is thought to be the key factor influencing the run and spawn timing of ocean- and stream-type populations. In most cases, stream-type chinook salmon spawn earlier, at higher elevations, and further upstream than ocean-type chinook salmon.

Spawning fish of both types use the upstream portions of their respective spawning areas first and the downstream portions last, thus providing opportunities for mixing among groups of fish whose spawning activities overlap spatially and temporally (i.e., spring- and summer-run fish in the Snake River and summer- and fall-run fish in the Columbia River). This phenomenon in the MCR is succinctly described by Mullan (1987, p. 3): "This time- space dimension was originally filled by successive waves of chinook salmon spawners."

In the Columbia River, adult chinook salmon migrating upstream past Bonneville Dam from March through May, June through July, and August through October have been categorized as spring-, summer-, and fall-run fish, respectively (Burner 1951). However, run-partitioning dates are progressively later at each dam encountered as adult fish migrate upstream. While annual run-partitioning dates remain static at all dams, adult migration timing varies annually, with water temperature as the primary controlling factor. Moreover, to some degree, the middle portion of the run (i.e., summer-run chinook salmon) is overlapped early in the migration by the spring-run and later by the fall-run.

Therefore, the separation of Columbia River chinook salmon into three races, based principally on adult run-timing at dams, is an arbitrary distinction (Fulton 1968, Chapman et al. 1982). Unfortunately, use of this distinction in accounting methods has often resulted in large census errors and considerable confusion regarding the ancestral relationships among chinook salmon populations in the Columbia River Basin.

Ocean-type chinook salmon in the Columbia River exist in two basic forms: "upriver brights" and "tules." Upriver brights enter the river first, mature slowly, and retain their silvery oceanic coloration well into the freshwater migration. This run of chinook salmon spawns from somewhere above the site of Grand Coulee Dam downstream to an area near the present site of The Dalles Dam. Spawning occurred both in the mainstem Columbia River and in the lower sections of tributaries (Fulton 1968, Dauble and Watson 1990) but was probably limited below the mouth of the Umatilla River (Bryant and Parkhurst 1950). The Snake River portion of the ocean-type run, historically spawning from Shoshone Falls downstream to the confluence of the Snake and Columbia Rivers, was listed as a threatened species under the ESA (Waples et al. 1991).

Tules are the last chinook salmon to enter the river; they are sexually mature upon entry and spawn in lower mainstem and tributary areas primarily below The Dalles Dam (Fulton 1968), that is, outside the petitioned area.

Currently, ocean-type chinook salmon pass Bonneville Dam between late May/early June (upriver brights) and late September/ early October (tules) (Howell et al. 1985). The early portion of the upriver bright run passes Priest Rapids Dam between mid-June and early August and spawns primarily in the lower reaches of the petitioned tributary streams above Rock Island Dam from late September through early November. In the area above Rock Island Dam, summer- and fall-run adults intermingle and spawn at the same time (Edson 1958, Mullan 1987, Craig and Suomela 1941). For example, Meekin (1967) and Meekin et al.(1966) could not distinguish mainstem spawners (presumably fall-run) from tributary spawners (presumably summer-run) based on time of passage over Wells Dam.

Naturally produced, summer-run chinook salmon introduced into the Wells summer chinook spawning channel have been observed spawning as late as mid-December (Allen et al. 1971), at least a month later than so-called summer chinook salmon are said to spawn (NEDC et al. 1993). Conversely, chinook salmon designated as fall-run according to their run timing have been observed spawning in the Priest Rapids fall chinook spawning channel as early as mid-September (Allen 1966, 1967). This spawn timing is more typical of summer- or even spring-run chinook salmon, according to some criteria (NEDC et al. 1993).

Recently, radio-tag data evaluating segregation of chinook salmon populations by run timing at dams showed that a significant portion of summer-run adults spawned in the mainstem Columbia River, while a significant portion of fall-run adults spawned in the Okanogan River system (L. Stuehrenberg, unpubl. data, NMFS). The petitioners claimed that the Okanogan River contains summer-run fish only (NEDC et al. 1993). However, it is apparent that the run times used by fishery managers to partition and allocate Columbia River chinook salmon are not necessarily recognized by the fish themselves. As Mullan (1987, p. 57) stated, "There are no clear differences between summer-run and fall-run chinook salmon in the mid-Columbia."

Fish from later in the run spawn primarily in the Hanford Reach below Priest Rapids Dam from late October to late December. A small number of fish not passing Priest Rapids Dam spawn in lower areas of the Yakima River in October and November (CBFWA 1990, WDF et al. 1993). Below McNary Dam, another small run of upriver brights enters the Deschutes River over a protracted period from late June to October, spawning in mainstem areas below Pelton Reregulating Dam from October through December (CBFWA 1990). According to Howell et al. (1985) the John Day River supports a "negligible" run of upriver brights; however, no data have been obtained for these fish.

Age at Spawning, Sex Ratio, and Fecundity

Area-specific data for age at spawning, adult sex ratios, and fecundity are generally lacking for wild populations of MCR chinook salmon. A life history characteristic that appears to differ among the three alleged forms of Columbia River chinook salmon is the abundance of early-maturing males, known as jacks (Howell et al. 1985). Generally, jacks are least abundant in stream-type chinook salmon populations and become progressively more abundant over the duration of the ocean-type run (Mullan 1987, Healey 1991, Mullan et al. 1992a).

However, because so-called summer- and fall-run fish were not observed in natural spawning areas, the difference noted between arbitrary groupings of ocean-type fish at dams is not persuasive. Based on an exhaustive summary of adult salmon counts at Rock Island Dam from 1933 through 1985, Mullan (1987) noted that jacks occurred less frequently in the early summer portion of the run (24%) than in the fall portion (48%). However, as noted earlier, attempts to partition population characteristics into run-specific clusters based on dam counts are tenuous due to inter- and intraseasonal variability in run timing of Columbia River chinook salmon.

Some of these apparently large differences in jack counts between early- and late-run ocean-type fish could simply be an artifact of the census location. For example, during the spring- run of Columbia River stream-type chinook salmon, jacks tend to migrate later and are nearly absent during the first third of the run. This is readily apparent because they are the first run of the season and, thus, are not overlapped early by another run. If this pattern holds for runs of ocean-type fish (see the section on straying below), then many jacks counted as fall-run chinook salmon may actually be destined for locations generally associated with early ocean-type, or summer-run fish. The accuracy of dam counts is also compromised because they include hatchery fish whose life history traits may have been altered (e.g., Mullan et al. 1992b).

The percentages of return by age for hatchery upriver brights have been listed by Howell et al. (1985). For the 1962-79 broods, 2-year-olds (jacks) comprised 34.1% of the total return, 3-year-olds 23.8%, 4-year-olds 34.6%, and 5-year-olds 7.5%. Six-year-old fish were rare, comprising less than 0.5% of the return (Mullan 1987). The overall age values were similar to area-specific values documented for wild upriver brights in the Deschutes (CBFWA 1990) and Snake Rivers (Chapman et al. 1991) and in the Hanford Reach of the Columbia River (Dauble and Watson 1990). Regarding age structure, mainstem Yakima River ocean-type chinook salmon were considered typical of chinook salmon in the Hanford Reach of the Columbia River (Busack 1990).

In upriver bright populations, all 2-year-old and most 3-year-old fish were males, whereas females predominated the older age classes (Howell et al. 1985, Dauble and Watson 1990). Overall, males slightly outnumbered females.

Fecundity data are not available for wild summer-run fish. At Wells Dam Hatchery (summer-run fish), fecundity averaged 4,935 eggs per female between 1967 and 1970, and at Priest Rapids Hatchery (fall-run fish), fecundity averaged 4,704 eggs from 1978 to 1992 (Howell et al. 1985). For fall-run chinook salmon utilizing artificial spawning channels in the MCR, Mathews and Meekin (1971) observed a mean fecundity of 5,015 eggs.

Ocean Distribution

Information on the ocean distribution of wild chinook salmon populations from the Columbia River Basin is limited (Waples et al. 1991, Matthews and Waples 1991). However, hatchery fish have received coded wire tags for over two decades, and catches of these fish provide some general insight into oceanic migratory patterns. Seven consecutive broods of Snake River ocean-type chinook salmon consistently displayed a more southerly oceanic catch distribution than MCR "fall-run" salmon (Waples et al. 1991). On the other hand, similar, but more limited, data showed virtually no difference in the oceanic distributions of MCR ocean-type fish released from Wells Dam ("summer-run") and Priest Rapids ("fall-run") hatcheries (Howell et al. 1985).

Juvenile Behavior

Timing of fry emergence has not been well documented for naturally produced ocean-type fish. In the MCR, fry emerge primarily in April and May (Chapman et al. 1994). At the Wells Dam spawning channel (summer-run fish), fry emerged from January through April during 1968-71 (Howell et al. 1985). For the 1963-67 broods at the Priest Rapids Hatchery spawning channel, emergence occurred primarily in late April and early May (Howell et al. 1985).

Typically, chinook salmon fry move downstream after emergence. For many populations of ocean-type fish, fry may continue migrating to the estuary or take up residence in the river for a few weeks to a year or more before entering the ocean (Healey 1991). In the Columbia River Basin and its tributaries, all ocean-type fry leave redd areas a few days to weeks after emergence (Chapman et al. 1994). Some fry rear only a short distance from nursery areas before migrating, while others may migrate downstream a considerable distance to rear. Although the exact mechanisms controlling dispersal behavior are largely unknown, they are probably related to a variety of factors such as inter- and intraspecific social interactions (Reimers 1968, Taylor 1988), habitat availability (Lister and Walker 1966), and river discharge (Healey 1991). Chapman et al. (1994) summarized results of several recent studies that suggest fish size or growth may be important variables regulating downstream movements.

Subyearling, ocean-type chinook salmon in the Columbia River tend to migrate downstream slowly, foraging and growing as they move seaward. These fish move out of rearing areas in late spring or early summer, with the majority passing downstream through McNary Dam from mid-July through mid-September. Fish originating from upstream areas migrate about 2-3 weeks later than those from downstream areas (Chapman et al. 1994). Impoundment of the river has likely shifted the migrational timings of these fish later than during predevelopment times (Park 1969). This later passage has apparently increased the proportion of fish that remain in the Columbia River over winter (Chapman et al. 1994).

Subyearling chinook salmon migrants use estuaries extensively for rearing prior to ocean entry (Healey 1991). In the Columbia River, estuarine residence times vary greatly from a few days to several months or longer. Rich (1920) recorded subyearling chinook in the estuary in all months, with some staying over winter.

"June hogs"

Many residents of the Pacific Northwest are aware of stories alleging that a specific run of particularly large chinook salmon, the so-called "June hogs," once migrated up the Columbia River (e.g., Seufert 1980). These fish, said to have averaged 18-45 kg in weight, supposedly predominated the middle portion of the run passing through the lower river and migrated to spawn somewhere in the Columbia River Basin. Most assumed that June hogs were summer-run fish.

However, Seufert (1980, p. 9), referred to them as "huge spring chinook," which would indicate stream-type chinook salmon. In addition, early settlers observed chinook salmon spawning as early as August in the upper Columbia River in British Columbia (Bryant and Parkhurst 1950), a life history pattern characteristic of stream-type chinook salmon in the Columbia River Basin. By comparison, all observed populations of upper Fraser River chinook salmon have stream-type juvenile life histories (Taylor 1989).

Regarding latitude, altitude, climate, and geography, the upper Columbia River is similar to the upper Fraser River and is more distant from the ocean. Therefore, it is logical that the life histories of chinook salmon populations in the two systems would be similar as well. In fact, after deglaciation, the Columbia River appears to have been the principal source for the repopulation of Fraser River fish fauna in general (McPhail and Lindsey 1986) and chinook salmon in particular (Utter et al. 1989).

We found no empirical evidence indicating that a unique population of massive fish ever existed in the Columbia River. Historical accounts from the early 1800s suggested chinook salmon caught by aboriginal people in the upper Columbia River at Kettle Falls averaged about 7.0 kg in weight (Mullan et al. 1992b). Wild adult ocean-type fish sampled at Rock Island Dam in 1940 weighed about 8.0 kg (Fulton and Pearson 1981). From historic catch records, Beiningen (1976) estimated a mean weight for "summer-run" chinook salmon of 8.5 kg, while Chapman (1986) used a mean weight of 10.5 kg in his estimates of population abundance for the late-1800s. On the other hand, early settlers of the upper Columbia River were said to have witnessed "summer-run" adult fish averaging 18.0 kg (Bryant and Parkhurst 1950). These anecdotes must be considered with caution, since no weights were actually reported and scale analysis was not available to determine juvenile life histories.

Considering that the size of the ancestral Columbia River chinook salmon population has been estimated at 2-4 million fish (Ebel et al. 1989), very large chinook salmon were undoubtedly common in the past. A few are occasionally observed today. As noted earlier, the ancestral chinook salmon run peaked in the lower river in early summer and was represented by many spawning populations of both stream- and ocean-type fish from upper tributaries in the Columbia and Snake Rivers. Therefore, it is possible that June hogs were simply the largest members of many different spawning populations. By the early 1900s, overfishing had largely extirpated the majority of Columbia River chinook salmon, particularly the largest individuals (Thompson 1951, Beiningen 1976). Some commercial fishing methods have been shown to dramatically reduce the mean size and age of chinook salmon populations (Ricker 1981).


Adult anadromous salmonids that spawn in areas other than their natal stream or hatchery are known as strays. However, some strays may actually be wanderers, as described by Chapman et al. (1991). Wandering fish enter nonnatal streams or areas and eventually depart to spawn elsewhere. Tagging data indicate that wandering may be a relatively common behavior in anadromous salmonids (Meekin 1967, Bjornn et al. 1992), especially in areas where hatchery releases occur close to spawning areas (Chapman et al. 1994). Unnatural obstacles (dams, weirs, traps, etc.) may partially or totally prohibit corrections or adjustments by these fish. In situations such as these, where voluntary egress is prevented, wandering fish may be falsely classified as strays.

Homing is well developed in anadromous salmonids, with olfactory cues providing the primary mechanism for river, tributary, and possibly even riffle selection (Groves et al. 1968, Hasler and Scholz 1983). Homing to specific natal environments has undoubtedly influenced the genetic interaction among neighboring populations, and in general, there is a decreasing likelihood of gene flow between salmon populations as geographic distance between them increases (see Quinn 1993, Utter et al. 1989, Shaklee et al. 1991). For example, it can be safely assumed that Alaskan stream-type chinook salmon do not stray into the Columbia River system.

While mixing between the same types of geographically- proximal chinook salmon stocks is undoubtedly greater, the extent to which it occurs naturally is not well understood. It is becoming increasingly apparent, however, that vacant habitat can be recolonized relatively quickly by salmonids from nearby populations (Milner and Bailey 1989).

Accounts of straying by Columbia River chinook salmon populations are confusing and have focused primarily on hatchery fish. Chapman et al. (1991, 1994) concluded that stream-type chinook salmon stray less than ocean-types. However, Rich and Holmes (1928) concluded the opposite.

Tule chinook salmon from the Washington Department of Fisheries Cowlitz Hatchery exhibited an average home-stream fidelity of 98.6% for four brood years (Quinn and Fresh 1984). Older fish tended to stray the most and jacks, returning later in the year, strayed the least. Straying also appeared to be related to brood-year success, with higher straying rates occurring when survival was low.

McIssac and Quinn (1988) reported 99% homing accuracy for upriver brights released from Priest Rapids Hatchery. These authors reported that homing appeared to be somewhat under genetic control. If this is true, then the large assortment of recent stock relocations, primarily for various hatchery or enhancement purposes, may have increased straying and therefore the mixing of Columbia River salmon. In fact, the petitioned populations of MCR chinook salmon were founded with many individuals originally from regions hundreds of kilometers upstream from the tributaries they now inhabit.

Portions of MCR late-run chinook salmon have been mixed considerably over the past two to three decades. This mixing was due to the variety of methods employed to collect broodstock at dams, hatcheries, or other areas and as a result of juvenile outplantings into various areas, including the petitioned streams (reviewed in Chapman et al. 1994). Since 1967, as many as 20% of summer-run chinook salmon broodstock for Wells Hatchery operations have been collected from the late component (so called fall-run) of ocean-type fish passing over Wells Dam after the nominal cutoff date (28 August) between summer- and fall-run groups (Allen 1966, 1967; Allen et al. 1971; Chapman et al. 1994). Moreover, recoveries of coded-wire-tagged adults from various juvenile releases in the late 1970s and 1980s indicated that wild and hatchery summer-run fish originating above Rock Island Dam have spawned extensively with designated fall-run fish originating in the Hanford Reach and Priest Rapids Hatchery (Chapman et al. 1994).

Conversely, about 15% of the so-called fall chinook salmon emigrating from spawning beds below Priest Rapids Dam have returned to spawn in the Columbia River system above Wells Dam (Chapman et al. 1994). The possibility of substantial genetic exchange between chinook salmon populations above and below Rock Island Dam was noted almost half a century ago (Fish and Hanavan 1948). Attempts to maintain discrete hatchery stocks have only recently been initiated.

The Yakima River has been heavily planted with Bonneville Hatchery ocean-type chinook salmon (Table 2), which are said to stray at substantial rates (Busack 1990, WDF et al. 1993).

Stock Histories

Since settlement of the Columbia River Basin in the mid- nineteenth century, a variety of activities associated with development and commerce have had negative consequences for Columbia River salmonid populations. The list of harmful activities includes prodigious overharvest of salmon; destructive or unregulated land management practices, including timber harvest, mining, livestock grazing, and irrigation; and construction of hydroelectric facilities with absent or inadequate adult and juvenile fish passage facilities.

High harvest rates continue today, primarily by fisheries in the northern ocean ranges of Columbia River salmonids (Howell et al. 1985). According to estimates, it may take scores of years for riparian habitat destroyed by logging to recover (Sedell and Swanson 1984). Livestock grazing continues to degrade stream habitats (Platts 1991). Recent examinations of screens at Columbia River irrigation diversions reveal that many screening devices need modernization (WDF et al. 1990).

Yet, in spite of historic and contemporary human activities adversely affecting Columbia River salmonids, the number of ocean-type chinook salmon returning to mid-Columbia River spawning areas has increased substantially since the construction of Grand Coulee Dam (discussed below). Three factors appear to be primarily responsible for the increase: 1) the number of ocean-type chinook salmon in the MCR in the early 1940s was extremely low, so any improvements to the system would result in an increase; 2) at least partial success of the Grand Coulee Fish Maintenance Project, which corrected some practices harmful to salmon and began supplementation efforts in the MCR Basin (Fish and Hanavan 1948), and 3) displacement of upriver stocks into the Hanford Reach.

Link to Table 2

Grand Coulee Fish Maintenance Project, 1939-43

The single most important event affecting the distribution of ocean-type chinook salmon in the middle and upper Columbia River was the construction of the Grand Coulee Dam (RKm 959) in 1939, which completely eliminated passage of anadromous salmon above that point. To compensate for the loss of habitat, the federal government initiated the Grand Coulee Fish Maintenance Project (GCFMP). The GCFMP sought to maintain fish runs in the Columbia River above Rock Island Dam (RKm 730) by two means: 1) improving salmonid habitat; and 2) establishing hatchery operations (Fish and Hanavan 1948).

The primary method of habitat improvement during the GCFMP was the obligatory installation of screens on irrigation diversions in tributaries of the mid-Columbia River. The screens prevented juvenile chinook salmon from being drawn into irrigation systems and presumably made a major contribution to the increase in MCR populations since the 1940s (discussed below). In contrast, chinook salmon populations in the Yakima River did not recover during the same period, as screening of irrigation diversions was not obligatory in that system (Robison 1957).

All adult fish passing Rock Island Dam from 1939 to 1943 were taken either to USFWS hatcheries on the Wenatchee or Methow Rivers for artificial spawning or to fenced reaches of the Wenatchee or Entiat River for natural spawning. Juveniles derived from adults of mixed-stock origin crossing Rock Island Dam were reared at USFWS hatcheries and transplanted into the Wenatchee, Methow, and Entiat Rivers.

Fish trapping operations began in May 1939 and continued through late fall each year until 1943. A total of five complete brood years were affected. Early-run fish (stream type) were treated separately from late-run fish (ocean type), and few distinctions were made regarding either "summer" or "fall" components of the late run, as all late-run fish were captured.

Fish and Hanavan (1948) estimated that 20 to 34% (mean = 27%) of the fish passing Rock Island Dam each year were, by present day standards, "fall-run" chinook salmon (i.e., crossed Rock Island Dam after 20 August). Because the GCFMP continued for 5 years and used all late-run fish, including those destined for now-inaccessible spawning areas in British Columbia, all present day ocean-type chinook salmon above Rock Island Dam are the progeny of the mixture of chinook salmon collected at Rock Island Dam from 1939 to 1943.

The only MCR tributary that did not receive spawning adults or mixed-stock hatchery juveniles during the 5-year GCFMP was the Okanogan River (Fish and Hanavan 1948, Mullan et al. 1992b). All chinook salmon adults destined for the Okanogan River from 1939 to 1943 were intercepted and either forced to spawn in other tributaries or in one of the USFWS hatcheries.

As none of the progeny of GCFMP fish were planted in the Okanogan River during this 5-year period, the native population of ocean-type chinook salmon in the Okanogan River was virtually eliminated. It is possible that 6-year old adults returned in 1944, thereby escaping the effects of GCFMP; but the low frequency of this age class (<1%) in Columbia River chinook salmon populations makes this an unlikely event. The ocean-type chinook salmon now observed in the Okanogan River are likely the progeny of mixed-stock strays from other tributaries or from the mainstem Columbia River. These strays must have repopulated the Okanogan River after termination of the GCFMP.

Artificial Propagation

Hatchery efforts with ocean-type chinook salmon in the MCR have been continuous and intensive since the implementation of the GCFMP. Three USFWS hatcheries were established during the GCFMP, and six Washington Department of Fisheries facilities have been constructed since then (Table 3). Currently, only the Leavenworth and Entiat facilities are not rearing ocean-type chinook salmon. As noted above, by 1944, all MCR chinook salmon were essentially the progeny of relocated stock (Fish and Hanavan 1948). Since 1941, over 200 million ocean-type chinook salmon have been released into the MCR Basin as either 0-age or yearling fish (Table 2). Only approximately 6.2 million (~3.0%) of these fish were from stocks outside the MCR Basin.

Table 3. Rearing facilities for chinook salmon in the mid-Columbia River operated by the U.S. Fish and Wildlife Service (USFWS) and Washington Department of Fisheries (WDF). Modified from Chapman et al. (1994).

Facility Agency Years* River
Leavenworth NFH USFWS 1941-present Wenatchee
Entiat NFH USFWS 1941-1964 Entiat
Winthrop NFH USFWS 1942-1983 Methow
Priest Rapids WDF 1963-present mid-Columbia
Rocky Reach WDF 1974-present mid-Columbia
Wells WDF 1967-present mid-Columbia
Eastbank WDF 1990-pesent Wenatchee
  acclimation ponds WDF 1990-present Methow
  acclimation ponds WDF 1990-present Okanogan

* Years when ocean-type chinook salmon were reared.
NFH = National Fish Hatchery.

Phenotypic Characteristics

Variation in phenotypic characteristics of chinook salmon in the Columbia River Basin was evaluated in a multivariate study by Schreck et al. (1986), who examined meristic and morphometric variation in 56 samples of ocean- and stream-type chinook salmon. Previous analysis of 20 trusslike morphometric characters had shown that fish with similar life histories tended to have similar body shapes (Winans 1984). Schreck et al. (1986) also found that ocean-type fish from the MCR grouped together (both "fall-run" and "summer-run"), and stream-type fish from the Snake River in Idaho grouped together (both "spring-run" and "summer- run").

In the same study, a multivariate analysis of nine meristic characters revealed two major groups in the Columbia River Basin. The first group contained stream-type chinook salmon from west of the Cascade Mountains and all ocean-type chinook salmon from the MCR. The other group contained stream-type chinook salmon from east of the Cascade Mountains and stream-type chinook salmon from Idaho. For all but one of the meristic characters analyzed, results indicated that between-year variability did not account for substantial levels of variation among stocks.

Genetic Characteristics

To evaluate chinook salmon populations in the Snake River, Matthews and Waples (1991) compiled a 21-locus allozymic (protein electrophoretic) data set for 44 samples within the Columbia River Basin. This analysis used relatively few loci to allow inclusion of samples dating back to 1980. Three major groups were identified: 1) fall-run and MCR summer-run fish; 2) Willamette River populations; and 3) spring-run and Snake River summer-run fish. Because summer- and spring-run fish from the Snake River were genetically similar and shared similar life histories, they were considered a single species as defined by the ESA.

Marshall (1993, 1994a and b) examined a 36-locus data set for chinook salmon which included data for several new populations in the MCR and which used only recent samples (Appendix). This data set provided a more comprehensive view of the pattern of genetic differentiation among MCR populations than the data set described by Matthews and Waples (1991). Marshall (1994a and b) provided dendrograms based on cluster analyses of two common genetic distance measures.

The dendrogram based on Nei's (1978) genetic distance produced results similar to those found by Matthews and Waples (1991). A clear separation between ocean- and stream-type fish in the Columbia River Basin was observed (Fig. 2). Notably, after correcting for sampling error, genetic distances between fall- and summer-run fish from the MCR were essentially zero.

Figure 2
Figure 2. Genetic relationships among selected samples of chinook salmon in the Columbia River Basin based on Nei's unbiased genetic distance (1978) and unweighted, pair-group method and arithmetic averages (UPGMA; Sneath and Sokal, 1963) for 36 loci (Marshall 1994a and b). Loci used and sample information are described in the Appendix. SP = spring-run, SU = summer-run, and FA = fall-run.

The same general pattern was seen in the dendrogram based on Cavalli-Sforza and Edwards (1967) chord distance (Fig. 3): the clearest distinction was between the stream-type and ocean-type fish. This analysis differs from Nei's (1978) in that Yakima River ocean-type fish aligned more closely to Hanford Reach/Priest Rapids stocks than to Bonneville Hatchery fish, and Deschutes River fish aligned with Lyons Ferry Hatchery/Marion Drain fish rather than with fish from the Little White Salmon Hatchery. Nei's unbiased distance measure (Nei 1978) adjusts for increased genetic similarity due to small sample sizes, whereas Cavalli-Sforza and Edwards' (1967) distance metric does not. However, as in Nei's dendrogram, MCR samples formed a distinct group in which summer- and fall-run samples clustered with one another. In these dendrograms, we included a sample from the Sandy River, which feeds into the lower Columbia River below Bonneville Dam, to provide an indication of the level of divergence between fall chinook salmon from the lower and middle reaches of the Columbia River.

The close genetic relationship among the late-run MCR samples was also seen in pair-wise G-tests comparing frequencies of polymorphic loci among samples. Marshall (1993) reported that all the "summer-run" samples shared a high degree of genetic similarity (i.e., samples from the Wenatchee River, Wells Hatchery, and the Similkameen River were not significantly different (P > 0.05)). Further, the Priest Rapids Hatchery sample of fall-run fish was not significantly different from the Hanford Reach (fall-run), Wells Hatchery (summer-run), or Similkameen River (summer-run) samples, and it differed only slightly from the Wenatchee River (summer-run) sample (0.01 < P < 0.05).

Figure 3
Figure 3. Genetic relationships among selected samples of chinook salmon in the Columbia River Basin based on Cavalli-Sforza and Edwards (1967) chord distance and UPGMA clustering for 36 loci (Marshall 1994a and b). Same loci and samples as Figure 2. SP = spring-run, SU = summer-run, and FA = fall-run

These same patterns of genetic variation among populations of chinook salmon in the basin have also been reported by other investigators. Using similar clustering analyses of allozyme data sets based on fewer loci, Schreck et al. (1986), Utter et al. (1987), and Hershberger et al. (1988) all found a close relationship between summer- and fall-runs in the MCR. Utter et al. (1989) also showed a high similarity between "fall-run" and "summer-run" samples in the MCR using multidimensional analysis of allozyme data.
Table of Contents