A summary of representative estimates of gene diversity statistics appears in Table 2 for chinook salmon and other species of salmon and sea run trout. The geographical areas covered in the studies listed in the table are similar, except for the studies of coho salmon (Wehrhahn and Powell 1987, Reisenbichler and Phelps 1987), which were conducted over smaller areas. Genetic subdivision among populations within drainages or among drainages (or adult run type) was estimated with GST=HS/HT, where HS is the average within-population gene diversity and HT is the total gene diversity, disregarding genetic subdivision. The percentage of gene diversity contained within populations, on average over loci, ranges from about 80% to about 98% in species of salmon and anadromous trouts. Chinook salmon in the Pacific Northwest tend to show greater levels of genetic subdivision among populations (GST 11-18%) than do chum, coho, pink salmon (GST 2-9%), and steelhead (GST 1.7%) in many of the same areas. Like chinook salmon, sockeye salmon (O. nerka) tend to show a greater degree of genetic subdivision among populations (GST 18%) than do other species of salmon. Chinook salmon populations in Alaska tend to show less genetic differentiation (GST 5.9%) than do southern populations in British Columbia, Washington, Oregon, and California.
To examine evidence for reproductively isolated populations or groups of populations, we analyzed allelic frequencies collected over 15 years by geneticists at NMFS, University of California at Davis, Washington Department of Fish and Wildlife, and the Alaska Department of Fish and Game. This set of data included both published and unpublished allelic frequencies collected with standardized laboratory procedures and compiled for use by participating fishery management agencies. Complete sets of data were available for 31 polymorphic loci: mAAT-1*, sAAT-1,2*, sAAT-3*, sAAT-4*, ADA-1*, ADA-2*, mAH-4*, sAH*, GPI-A*, GR*, HAGH*, mIDHP-2*, sIDHP-1*, sIDHP-2*, LDH-B2*, LDH-C*, mMDH-2*, sMDH-A1,2*, sMDH-B1,2*, sMEP-1*, MPI*, PEPA*, PEPB-1*, PEPD-2*, PEPLT*, PGDH*, PGK-2*, PGM-1*, PGM-2*, sSOD-1*, TPI-4*. Two loci, mAH-4* and GR*, were not available for Alaska chinook salmon samples, so analyses including these samples were based on only 29 loci. For populations sampled more than 1 year--some as many as 3 or 4 years--allelic frequencies for each locus were combined, and the pooled frequencies were used to represent the population frequencies. In several instances, allelic frequencies for neighboring populations were also combined, if the sum of the individual G-tests of frequencies between samples, divided by the sum of the degrees of freedom was not significant. (This data set also serves as a population baseline for estimating the stock contributions of chinook salmon to mixed-population ocean or river-mouth harvests, chiefly along the coasts of Washington and Oregon.) A total of 193 populations extending from Alaska to California were included in the present analyses (Table 3 and Fig. 18). We calculated Rogers' (1972), Nei's unbiased (1978), and Cavalli-Sforza and Edwards' (1967) chord distances between samples, and searched for genetically-discrete geographical groups with multidimensional scaling in three dimensions and with the UPGMA tree algorithm.
All 193 population units were included in the first analysis to examine large-scale geographical patterns of genetic structure among chinook salmon populations from Alaska to California. A major feature of the UPGMA tree and MDS analysis (Fig. 19) of these samples was a clear genetic separation between populations with stream-type life histories and those with ocean-type life histories. Stream-type populations extend from Alaska, through northern British Columbia, into the upper Fraser River, and into the mid- and upper Columbia River Basin. Ocean-type populations, and populations showing both ocean- and stream-type juvenile migration (mixed-type populations), extend from central British Columbia to the Sacramento-San Joaquin River drainage in California. The transition zone from ocean- and mixed-type populations in the south to only stream-type populations in the north occurs along the central coast of British Columbia. In this zone, populations such as those in the Kitimat, Atnarko, and Wannock Rivers were intermediate in the MDS diagram between the two larger clusters representing ocean- and stream-type populations. Samples from populations in the lower and South Thompson River, a Fraser River tributary, also clustered in an intermediate position.
Several subclusters appeared within stream-type chinook salmon. Six samples from south-central and northwestern Alaska were genetically distinct from all other samples. These Alaskan samples showed surprisingly little genetic differentiation from each other, even though they were collected over an area extending from Bristol Bay to south-central Alaska. The amount of genetic diversity among these populations was considerably less than that among populations extending over comparable areas in British Columbia, Washington, Oregon, and California. Two samples from southeastern Alaska clustered with samples from northern British Columbia. Geographical patterns were also apparent among the remaining stream-type samples. Stream-type populations in the Columbia River Basin were genetically distinct from stream-type populations in the upper Fraser, Skeena, Nass, and Stikine Rivers in British Columbia.
Several distinct subclusters also appeared among ocean-type samples of chinook salmon. Samples from southern British Columbia and from Puget Sound rivers fell into a large subcluster. Another subcluster contained samples from the coastal rivers of Washington, Oregon, and California. Samples from the upper Klamath River were genetically distinct from other samples of ocean-type populations and clustered near the convergence of the two life-history groups. Other distinct subclusters of ocean-type fish included samples from the Columbia River Basin and those from the Sacramento-San Joaquin River drainage. The following analyses of subsets of these samples examine these groups in more detail.
A subset including samples from 83 ocean-type populations in southern British Columbia, Washington, Oregon, and California was analyzed with both the UPGMA (Fig. 20) and MDS (Fig. 21) clustering methods. Since the purpose of analyzing this subset of samples was to discern relationships among coastal populations, Columbia River and upper Klamath River populations were not included because they were genetically very different from coastal populations. In the subset of 83 samples, 5 clusters of more or less genetically distinct samples appeared in both analyses. All the samples from British Columbia, including samples from the lower Fraser River, Vancouver Island, and southern British Columbia mainland clustered together in the MDS diagram. A large distinct cluster of British Columbia populations was also apparent in the UPGMA tree. However, two samples from the lower British Columbia mainland grouped separately. In both the MDS and UPGMA clustering methods, geographically nearby samples were more similar to each other than were more distantly separated samples. British Columbia samples, as a group, were most closely related to samples from populations in Puget Sound.
A second large cluster included samples from populations of chinook salmon in rivers draining into Puget Sound. Four groupings within this cluster were apparent in the UPGMA tree: 1) the Elwha River populations, 2) the Nooksack River populations, 3) populations from the Skagit and Stilliguamish Rivers, and 4) south Puget Sound populations and Skagit Hatchery fall-run and summer-run populations. In the three-dimensional MDS diagram, the samples from the Elwha River were intermediate between the Puget Sound samples and samples from the coast of Washington.
A third large UPGMA cluster included all samples from the coast of Washington. In the UPGMA tree, the cluster of samples from rivers along the Washington coast joined with a cluster of samples from north Oregon coastal rivers. In the MDS diagram, however, Washington coastal river samples were situated between Puget Sound river samples and Oregon coastal river samples. The Washington coastal clusters in both clustering methods contained a sample from the Hoko River, which drains into the Strait of Juan de Fuca west of the Elwha River. In the UPGMA tree, samples from the Quinault, Queets, and Hoh Rivers formed a subcluster separate from other samples from Washington outer-coastal rivers.
In both the MDS diagram and the UPGMA tree, a fourth cluster included samples from northern and mid-Oregon coastal rivers as far south as Euchre Creek. One exception was the sample of spring-run chinook salmon from the Rock Creek Hatchery on the Umpqua River, which was more closely related to samples from southern Oregon coastal rivers than to samples from mid-Oregon. Northern and mid-Oregon coastal river samples, as a group, appeared to be more closely related to Washington coastal river samples than to samples from rivers in southern Oregon and northern California.
A fifth cluster included samples from southern Oregon coastal rivers, the lower Klamath River, and coastal rivers in northern California. Two distinct subclusters of samples appeared within this cluster. One contained samples from populations in the lower Klamath River and coastal rivers to the north. This subcluster also contained the spring-run sample from the Rock Creek Hatchery as mentioned above. The second subcluster contained samples from coastal rivers south of the Klamath River. The sample from Omagar Creek, located in the lower Klamath River, did not appear in either of these two subclusters.
We analyzed a set of allelic frequencies for 31 loci in 55 samples from the Columbia and Snake Rivers to depict population structure among populations in these drainages. An MDS diagram of Cavalli-Sforza and Edwards' chord genetic distance best illustrated the major features of this analysis (Fig. 22). Samples in this analysis were separated into two distinct clusters: ocean-type populations and stream-type populations; except for a sample of spring-run chinook salmon from the Klickitat River, which was genetically intermediate between the two clusters.
Additional genetic population structure was apparent within these two life-history types. Within ocean-type chinook salmon, samples of spring- and fall-run chinook salmon from the lower Columbia River were distinct from all inland samples. The lower Columbia River group included naturally spawning fish from the Lewis and Sandy Rivers and from hatchery brood stock derived from populations west of the Cascade Mountain Range. Four samples, three from Willamette River hatcheries and one from the North Fork Clackamas River, were genetically distinct from other ocean-type chinook salmon in the Columbia River drainage.
Samples of ocean-type fish from localities east of the Cascade Crest included fish from both "bright" fall- and summer-run populations, including fall-run populations at the Bonneville and Little White Salmon hatcheries and in the Klickitat River. Although these populations are located on the west side of the Cascade Crest, brood stocks used in the hatchery programs in these rivers were derived from upriver populations of ocean-type chinook salmon. The Klickitat River summer-run population, which was introduced from upriver sources, appeared in the MDS diagram in an intermediate position between inland and lower Columbia River ocean-type populations.
The arrangement of samples of stream-type chinook salmon in the MDS diagram (Fig. 22) is largely consistent with geographical relationships among populations, except for a few notable samples. Samples of ocean-type fish (lefthand side of Figure) were clearly separated from stream type fish (righthand side of Figure). A genetically diverse group of samples of stream-type fish (squares) from the Klickitat, John Day, Deschutes, and Yakima Rivers of the mid Columbia River were positioned between the extremes of ocean-type and stream-type fish. A second group of stream-type fish (inverted triangles plus samples 90 and 91) were positioned between mid-Columbia River spring-run fish and fish from spring- and summer-run populations in the Snake River. This group included geographically diverse samples from the Wenatchee and Methow Rivers in the upper Columbia River, as well as two samples (90, 91) from the Grande Ronde River, a tributary of the Snake River. The inclusion of samples from the Wenatchee, Methow, and Grand Ronde River tributaries in this group may be due to a long history of introducing Carson Hatchery fish, or fish derived from Carson Hatchery fish, into upper Columbia River tributaries. Carson Hatchery was initially stocked with fish from the Snake River, and introductions followed by hybridization may have produced the similarity of upper Columbia River spring-run fish to Snake River fish. The third cluster of stream-type chinook salmon was most distantly related to ocean-type chinook salmon and included samples from Snake River populations in the Salmon and Imnaha Rivers, and Rapid River, and Lookingglass Hatcheries.
The genetic groupings of chinook salmon appearing in our analyses of the coast-wide set of allelic frequencies were largely consistent with those described in previous studies of chinook salmon. Our results for populations in Alaska agreed with those of Gharrett et al. (1987), who also found that chinook salmon populations in south-central and northwestern Alaska showed less inter-population genetic diversity than did populations in other regions, and that south-central and northwestern Alaska populations were genetically distinct from populations in southeastern Alaska. Populations in southeastern Alaska appear to be genetically most similar to stream-type populations in northern British Columbia. Our analysis and that of Utter et al. (1989) indicated that stream-type populations in the upper Fraser River were closely allied with stream-type populations in northern British Columbia.
Ocean-type chinook salmon populations in Vancouver Island rivers, in the lower Fraser River, and in rivers in southern British Columbia form a genetically distinct, though diverse, group of populations. Utter et al. (1989) proposed a similar grouping of populations, but placed a single sample from west Vancouver Island with coastal populations to the south. Puget Sound populations of chinook salmon appear to constitute a genetically distinct group, a conclusion that is consistent with the results of Utter et al. (1989) and Marshall et al. (1995). In our analyses, Washington coastal populations appeared to form a genetically distinct group that was most similar to, but still distinct from, Oregon coastal populations. The Washington coastal group included the Hoko River population in the western part of the Strait of Juan de Fuca. Chinook salmon in the Elwha River, which also drains into the Strait of Juan de Fuca, were genetically intermediate between Puget Sound and Washington coastal populations. Marshall et al. (1995) grouped this and other Strait of Juan de Fuca populations with Washington coastal populations.
Chinook salmon populations in the Columbia and Snake Rivers appear to be separated into two large genetic groups: those producing ocean-type juvenile outmigrants and those producing stream-type outmigrants. The subdivision of Columbia River Basin populations into two major genetic units is consistent with Waples et al. (1991a) and Marshall et al. (1995). The first group includes populations in lower Columbia River tributaries, with both spring-run and fall-run "tule" life histories. These ocean-type populations exhibit a range of juvenile life-history patterns that appear to depend on local environmental conditions. The Willamette River hatchery populations form a distinct subgroup within the lower Columbia River group. Ocean-type chinook salmon populations east of the Cascade Range Crest include both summer- and fall-run "bright" populations, and are genetically distinct from lower Columbia River ocean-type populations. Fall-run populations in the Snake River, Deschutes River, and Marion Drain (Yakima River) form a distinct subgroup. These genetic groupings are also consistent with the analyses of Waples et al. (1991a) and Marshall et al. (1995).
The second major group of chinook salmon in the Columbia and Snake River drainage consists of spring- or summer-run fish. Three relatively distinct subgroups appeared within these stream-type populations. One subgroup includes populations in the Klickitat, John Day, Deschutes, and Yakima Rivers of the mid Columbia River. A second subgroup includes upper Columbia River spring-run chinook salmon in the Wenatchee and Methow Rivers, but also spring-run fish in the Grande Ronde River and Carson Hatchery. A third subgroup consists of Snake River spring- and summer-run populations in the Imnaha and Salmon Rivers, and in the Rapid River and Lookingglass Hatcheries. These groupings are consistent with those found by Waples et al. (1991a). However, Marshall et al. (1995), who examined only populations in Washington State for genetic variability, identified three groups of stream-type chinook salmon 1) Yakima River, 2) Wenatchee and Methow Rivers, and 3) a Snake River spring-run population (Tucannon River). The Klickitat River spring-run population appears to be genetically intermediate between upper and lower Columbia River groups, a conclusion consistent with that of Marshall et al. (1995).
All populations of chinook salmon south of the Columbia River drainage appear to consist of ocean-type fish. Populations along the north coast of Oregon form a genetically distinct group, consisting of populations north of and including the Elk River, except for the Rock Creek Hatchery spring-run population, which shows greater genetic affinity to southern Oregon coastal populations. A southern coastal group includes populations south of the Elk River to and including populations in the lower Klamath River in northern California. However, Euchre Creek, located near the Rogue River, has been stocked extensively with Elk River stock and clustered with populations north of Cape Blanco. A California coastal group consists of populations south of the Klamath River. These genetic groups are consistent with Bartley et al. (1992). Upper Klamath River populations of chinook salmon are genetically distinct from other northern California populations. The results of Bartley and Gall (1990) and Bartley et al. (1992) are consistent with these groupings of northern California and southern Oregon populations.
Sacramento and San Joaquin River populations are genetically distinct from northern California coastal and Klamath River populations. Previous studies grouped populations in the Sacramento River and with those in the San Joaquin River (Utter et al. 1989, Bartley and Gall 1990, Bartley et al. 1992). However, Hedgecock et al. (1995), Banks (1996), and Nielsen (1995, 1997) surveyed DNA markers and these results indicate that the winter, spring, fall, and late-fall runs are genetically distinct from one another.
Most of the ESUs described below include multiple spawning populations of chinook salmon, and most also extend over a considerable geographic area. This result is consistent with NMFS' species definition paper, which states that, in general, "ESUs should correspond to more comprehensive units unless there is clear evidence that evolutionarily important differences exist between smaller population segments" (Waples 1991b, p. 20). However, considerable diversity in genetic or life-history traits or habitat features exists within most ESUs, and maintaining this diversity is critical to their overall health. The descriptions below briefly summarize some of the notable types of diversity within each ESU, and this diversity is considered in the next section in evaluating risk to the ESU as a whole.
According to NMFS policy, populations of Pacific salmon will be considered "distinct" (and hence "species" as defined by the ESA) if they represent evolutionarily significant units of the biological species. A variety of factors are considered in evaluating the two criteria for salmon populations or groups of populations to be considered ESUs: reproductive isolation and substantial contribution to ecological/genetic diversity of the species as a whole.
Previous status reviews conducted by NMFS have identified three ESUs of chinook salmon in the Columbia River: Snake River fall (Waples et al. 1991b), Snake River spring and summer (Matthews and Waples 1991), and mid-Columbia River summer-run chinook salmon (Waknitz et al. 1995). In addition, prior to development of the ESU policy, NMFS recognized Sacramento River winter chinook salmon as a "distinct population segment" under the ESA (NMFS 1987). In reviewing the biological and ecological information concerning west coast chinook salmon, the Biological Review Team identified 11 additional ESUs for chinook salmon from Washington, Oregon, and California. Genetic data (from protein electrophoresis and DNA analysis) and tagging information were key factors considered for the reproductive isolation criterion, supplemented by inferences about barriers to migration created by natural features. A number of factors were considered to be important in evaluations of ecological/genetic diversity. Data on life-history characteristics (especially age at smoltification, ocean distribution, time of freshwater entry, and age at maturation) and geographic, hydrological, and environmental characteristics were the most informative.
The predominant differentiation in chinook salmon life-history types is between ocean- and stream-type chinook salmon. Gilbert (1912) initially defined ocean- and stream-type life-history types to discriminate between fish that emigrated to saltwater as subyearlings (ocean-type) and those that emigrated at one or more years of age (stream-type). Healey (1983, 1991) utilized a number of additional life-history traits to expand this process to describe two races of chinook salmon. In Healey's scheme, ocean-type populations typically migrate to seawater in their first year of life and spend most of their oceanic life in coastal waters, whereas stream-type populations migrate to sea as yearlings and often make extensive oceanic migrations. Stream-type fish spawn in the upper Fraser River and Columbia River Basins, as well as coastal areas north of about latitude 55N (Healey 1983). Ocean-type chinook salmon spawn in the Sacramento River and the mainstem and lower tributaries of the Columbia, Snake, and Fraser River Basins, and throughout western North American coastal drainages to approximately 55N. In this review, we have followed Healey's scheme, which focuses on populations rather than individual fish, and focuses on a suite of genetic and life-history traits rather than just age at juvenile outmigration.
In some areas within the Columbia River Basin, stream- and ocean-type chinook salmon stocks spawn in relatively close proximity to one another but are separated by run timing. Stream-type chinook salmon include spring-run populations in the Columbia River and its tributaries east of the Cascade Crest, and spring- and summer-run fish in the Snake River and its tributaries; ocean-type chinook salmon include fall-run chinook salmon in both the Columbia and Snake River Basins, summer-run chinook salmon from the Columbia River, and spring-run fish from the lower Columbia River. Although it has also been known for some time that there are substantial genetic differences between stream- and ocean-type chinook salmon in both the Fraser and Columbia River Basins, the genetic analyses in this status review show clearly for the first time that the two life-history forms represent two major (and presumably monophyletic) evolutionary lineages. Genetic differences between the two forms, as measured by variation in allozymes, are of the same order of magnitude as the differences found between the inland and coastal subspecies of steelhead (O. mykiss) and between even- and odd-year pink salmon (O. gorbuscha).
Adult run time has also long been used to identify different temporal "races" of chinook salmon. In cases where the run-time differences correspond to differences between stream- and ocean-type fish (e.g. in the Columbia and Fraser River Basins), relatively large genetic differences (as well as ecological and life-history differences) can be found between the different runs. In most coastal areas, however, life-history and genetic differences between the runs are relatively modest. Although many populations have some fraction of yearling migrants, all the coastal populations are part of the ocean-type lineage, and spring- and fall-run fish are very similar in ocean distribution patterns and genetic characteristics.
Among basins supporting only ocean-type chinook salmon, the Sacramento River system is somewhat unusual in that its large size and ecological diversity historically allowed for substantial spatial as well as temporal separation of different runs. Genetic and life-history data both suggest that considerable differentiation among the runs has occurred in this basin. The Klamath River Basin shares some features of coastal rivers but historically also provided an opportunity for substantial spatial separation of different temporal runs. As discussed below, the BRT found that the diversity in run timing made identifying ESUs difficult in the Klamath and Sacramento River Basins.
The ecological importance and underlying genetic basis of specific life-history traits has been discussed in a previous section. The BRT considered differences in life-history traits as a possible indicator of adaptation to different environmental regimes and resource partitioning within those regimes.
Based on preliminary information indicating substantial ecological, geographic, and genetic differences among chinook salmon from the Columbia and Sacramento Rivers and coastal drainages, the BRT considered the following three geographic areas separately in making ESU determinations: California Central Valley, coastal basins and Puget Sound, and Columbia River. Some of the factors considered important in defining ESUs within each area are briefly discussed here, followed by more detailed descriptions of each of the proposed ESUs.
The Sacramento River winter chinook salmon was designated as a distinct population segment (NMFS 1987) almost entirely on its unique life-history features. No genetic data for the population were available at the time of the listing determination, and the NMFS species policy had not been formulated. Recent DNA data show substantial differences between the winter run and all other runs in the basin. The BRT concluded that the life-history and genetic data collectively support designation of the winter run as an ESU. The DNA data also show significant differences between spring-run fish and the fall and late-fall runs. Ecological data show strong evidence for historic spatial and temporal isolation of the spring run, and the BRT also concluded that this run represents an ESU. The majority of the BRT felt that differences between fall and late-fall runs were consistent with diversity within a single ESU and did not warrant the creation of separate ESUs for these runs.
All populations of chinook salmon in Puget Sound and coastal drainages of Washington, Oregon, and California are considered ocean type. In these areas, life-history differences exist between spring- and fall-run fish, but not to the same extent as is observed in larger inland basins, and genetic data indicate the two run types are polyphyletic in coastal drainages. Utter et al. (1989) identified three genetic groups of chinook salmon in this geographic region: Puget Sound, upper Klamath River Basin, and other coastal streams from the Olympic Peninsula to northern California. Recent genetic data indicate the presence of more geographically clustered groups along the coast. Based primarily on genetic data, geographic and environmental features, and life-history traits, the BRT identified five ESUs in this area: Puget Sound, Washington Coast, Oregon Coast, Southern Oregon and California Coast, and Upper Klamath and Trinity Rivers. A minority of the BRT proposed that the Southern Oregon and California Coast ESU should be split into two ESUs, with a boundary south of the Klamath River.
As noted above, a major phylogenetic break occurs between stream- and ocean-type chinook salmon in the Columbia River. Populations from both types were included in ESUs defined in previous status reviews. Groups whose ESU status had not been determined previously include ocean-type fish below McNary Dam, stream-type fish from outside the Snake River Basin, and spring-run chinook salmon in the upper Willamette River. Willamette River spring-run fish are isolated from, and genetically quite distinct from, all other Columbia River chinook salmon, and the BRT agreed that they represent an ESU. The BRT also concluded that ocean-type fish spawning below the Cascade Crest, including both spring and fall chinook salmon, were part of a single ESU. This ESU includes the "tule" fall runs, which return in an advanced stage of maturation and exhibit distinct secondary maturation characteristics: darkened skin, resorbed scales, and pronounced kype. These are distinguishable from "upriver brights", which return to spawning sites above the Cascade Crest and enter freshwater at a less advanced stage of maturation.
Four geographic/genetic groups of stream-type chinook salmon can be identified in the Columbia River: Snake River, Columbia River tributaries from Bonneville Dam to the Snake River, Yakima River Basin, and upper Columbia River (tributaries upstream of the Yakima River). The latter group includes all populations affected by the Grand Coulee Fish Maintenance Project. The majority of the BRT concluded that there are three ESUs in this area: Snake River, upper Columbia River, and mid-Columbia River (Bonneville Dam to Yakima River, inclusive). Scenarios favored by minorities of the BRT included a single ESU encompassing all stream-type chinook salmon, two ESUs (Snake River and Columbia River), and four ESUs (each of the abovementioned groups).
The BRT also considered several populations of "upriver bright" ocean-type chinook salmon whose ESU status had not been resolved in previous status reviews. Excluded from discussion were several upriver bright chinook salmon populations in the Wind, White and Little White Salmon, and Klickitat Rivers; historical records (e.g., Fulton 1968) do not document native populations in these areas, and current populations are believed to be the result of stock transfers. Native fall-run populations in the John Day, Umatilla, and Walla Walla Rivers have been extirpated (Kostow 1995), and populations that are presently found in these systems are also considered to be the result of introductions. Of particular interest are populations in the Deschutes River and Marion Drain in the Yakima River drainage that have shown a genetic affinity with Snake River fall chinook salmon (Waples et al. 1991b, WDF et al. 1993). A minority of the BRT felt that the Marion Drain population should be considered part of the Snake River ESU, but the majority felt that the origin of this population is too uncertain to determine its ESU status. A majority of the BRT concluded that the Deschutes River population should be considered part of the Snake River ESU, whereas a minority felt that this population was historically part of a separate ESU that included populations from the John Day, Umatilla, and Walla Walla Rivers. All members felt it was important to develop more definitive information about the Deschutes River population and its possible link to Snake River fish.