U.S. Dept Commerce/NOAA/NMFS/NWFSC/Publications
NOAA F/NWC-201 Status Review for Snake River Fall Chinook Salmon


Return to previous section of document.

Ocean Distribution

No direct information is available regarding the ocean distribution of wild Snake River fall chinook salmon, and efforts to study wild upper Columbia River fish have only recently been initiated. However, CWTs have been used on hatchery fish from both rivers since the late 1970s, and catches of tagged fish in ocean and river fisheries provide some insight into migratory patterns. Complete data for year classes from 1978 to 1984 are available (Busack 1991c), and Figure 4 shows the proportion of adult recoveries from different geographic areas. To avoid possible biases from comparison of releases at different ages, only zero+ age (subyearling) releases were considered.

Although there are inherent difficulties in making statistical comparisons of CWT recoveries for different populations (Busack 1991c), a clear difference was evident in the ocean distribution of Snake and upper Columbia River fall chinook salmon, and the patterns were consistent over the duration of the study. In the years studied, Snake River fish had a more southerly distribution, with a significant proportion (about 20-50%) of recoveries taken in Washington, Oregon, and California and very few (<5%) in Alaska. The converse was true of upper Columbia River fish; there were substantial catches in Alaska (20-35%) and few in southern areas (<10%). McNeil (1991) summarized the CWT data using a slightly different format and showed that for brood years 1978-85, the proportion of all CWT recoveries occurring in California and Oregon was much higher for Snake River fish (24.5%) than for upper Columbia River fish (6%). According to data presented by Howell et al. (1985), the proportion of CWT recoveries taken off California and Oregon is also very low for fall chinook salmon from the lower Columbia River (about 1-6% for wild fish from the Lewis and Willamette Rivers and hatchery fish from Bonneville pool and lower river facilities). Fall chinook salmon from the Sacramento River "migrate north along the California and Oregon coast with numbers decreasing rapidly along the Washington coast" (Van Hyning 1973, p. 73).

Phenotypic Characteristics

Utter et al. (1982) compared available data for adult fall chinook salmon females and found that fish from the upper Columbia River were significantly larger than those from the Snake River. Mean length in three collections (1977-79) of fish from the upper Columbia River ranged from 86 to 88.6 cm, compared to a range of 75.2 to 83.8 cm in seven collections (1957-79) from the Snake River. Caution should be used in interpreting these results because 1979 was the only year both populations were sampled. Nevertheless, Mains and Smith (1964) found a similar difference in juvenile size, with migrant chinook salmon fry in the Columbia River being larger than those in the Snake River. Because adults in both rivers spawn at about the same time, Utter et al. (1982) concluded that the differences in fry size reflect environmental differences between the upper Columbia and Snake Rivers and/or genetic differences between the two populations.

Environmental Features

Geological, topographical, and hydrological features of the Snake River Basin are unique in the Pacific Northwest (Chapman et al. 1991). The basin extends into five states (Idaho, Oregon, Washington, Wyoming, and Nevada), drains an area of approximately 267,000 km2, and incorporates a range of vegetative life zones, climatic regions, and geological formations, including the deepest canyon (Hells Canyon) in North America.

Utter et al. (1982) presented data documenting substantial differences in water temperature between the upper Columbia and Snake Rivers. Over a 2-year period in the 1960s, mean monthly summer water temperature in the Snake River (Weiser, Idaho) was 6-8°C higher than at Rock Island Dam in the upper Columbia River (Fig. 5). The annual temperature range was also considerably greater in the Snake River. Sylvester (1959) and Chapman et al. (1991) present data indicating that a similar pattern has been found at other sites in the two rivers, both more recently and in the past. The high summer water temperatures apparently prevent juvenile fall chinook salmon from rearing in the main-stem Snake River after July. In contrast, in the upper Columbia River, some fall chinook salmon may rear in the river into August, reaching lengths of 110 to 130 mm before migrating to the ocean (Allen and Meekin 1973).

The two rivers also differ in other water characteristics. In a 4-year study, Sylvester (1959) found monthly means for pH and total alkalinity of 8.2 and 99 ppm, respectively, at the mouth of the Snake River, and 7.8 and 64 ppm in the Columbia River at Pasco, Washington, just upstream from the confluence with the Snake River. In addition, the Snake River is typically much more turbid than the Columbia River.

Genetics

Recent reports (Utter et al. 1982; Schreck et al. 1986; Utter et al. 1989; Waples et al. 1991) summarizing electrophoretic information for Columbia River Basin chinook salmon establish the following genetic relationships:

1) On a broad scale [Nei's (1978) genetic distance4 > 0.02], populations can be grouped into three clusters (Fig. 6): a) spring- and summer-run fish from the Snake River and spring-run fish from mid- to upper-Columbia River; b) spring chinook salmon from the Willamette River; and c) fall chinook salmon. The third cluster also includes some hatchery stocks of spring chinook salmon from the lower Columbia River and some upper Columbia River summer-run fish with life history patterns similar to fall-run fish.

2) Substantial genetic differences also exist between lower Columbia River ("tule") fall chinook salmon and "brights" from the upper Columbia and Snake Rivers (genetic distance > 0.01).

3) Upriver bright fall chinook salmon can be further divided into upper Columbia and Snake River components (separated by a genetic distance of about 0.005). The two forms differ by about 10-20% in frequencies of alleles at several gene loci, and these differences were relatively constant across several years of sampling in the late 1970s and early 1980s. An upriver bright run also occurs in the Deschutes River, a mid-Columbia River tributary, and samples from there were genetically more similar to Snake River than to upper Columbia River samples. An upriver bright sample from a small irrigation ditch (Marion Drain) in the Yakima River Basin also showed a greater genetic affinity to Snake River samples than to upper Columbia River samples from the Hanford Reach area.

4) More recent (1985-90) samples of Snake River fall chinook salmon from Lyons Ferry Hatchery suggest that some mixing with upper Columbia River fish has occurred. At several gene loci (sAH*, sIDHP-1,2*, PEP-LT*, and sSOD-1*), allele frequency differences between the two populations are currently much smaller than they were a decade ago (Fig. 7). This result is consistent with reports (see above) that, in recent years, a sizeable fraction of the Lyons Ferry brood stock has been stray hatchery fish of upper Columbia River stock.

Figure 7 also shows that measures taken to reduce the genetic impacts of straying into Lyons Ferry Hatchery can be effective. Two data points are shown for Snake River fall chinook salmon in 1990; the open circle represents a "random" sample from all untagged fish (group C above), and the open triangle represents returning Lyons Ferry Hatchery adults identified by CWTs (group A above). Allele frequencies for group C have clearly converged toward profiles typical of upper Columbia River fish, whereas allele frequencies in group A are more typical of pre-1980 Snake River wild fish. This latter result reflects the fact that tagged Lyons Ferry Hatchery adults returning in 1990 were primarily from the 1986 brood year, which predated the most extensive straying events.

Although the electrophoretic data are extensive both in space and time, they have a major limitation-little direct information is available regarding the genetic makeup of wild Snake River fall chinook salmon. All of the electrophoretic data for Snake River fall chinook salmon collected after 1981 are for fish taken at Lyons Ferry Hatchery. An attempt in 1990 by WDF to collect a sample of wild fish for electrophoretic analysis was not successful. Early (1977-81) samples of adults trapped at Ice Harbor Dam presumably represent wild fish; however, the location of Ice Harbor Dam (on the lower Snake River just 15 km above the confluence with the Columbia River) raises some questions regarding samples collected there. First, if discrete populations of fall chinook salmon occurred historically in the Snake River [as suggested by ODFW (1991a)], the Ice Harbor Dam samples may have included mixtures of fish from different gene pools. Second, if some of the fish trapped at Ice Harbor Dam were actually upper Columbia River fish, the true differences between native Columbia River and Snake River fall chinook salmon may have been greater than indicated by the early Ice Harbor collections.

The latter concern is based on the observation that chinook salmon, as do other Pacific salmon species, occasionally wander into nearby rivers before eventually making their way to their natal spawning area (Chapman et al. 1991); collecting adults prior to spawning (as occurs at Ice Harbor Dam) can "create" strays by preventing this behavior. However, the tagging study of McIsaac and Quinn (1988) and the CWT data for Lyons Ferry Hatchery suggest that homing of wild upper Columbia River fall chinook salmon is very precise, so this may not have been a substantial factor in the early Ice Harbor collections.

Concerns about a possible mixture of Snake River populations in the early Ice Harbor collections are difficult to evaluate. Data demonstrating the existence of multiple populations of Snake River fall chinook salmon are lacking. Smouse et al. (1990) reported allele frequency differences between two temporally-spaced samples taken in 1981 at Ice Harbor Dam, but the differences were relatively small and not statistically significant.

DISCUSSION AND CONCLUSIONS

Differences in Run-timing

Results reported by Schreck et al. (1986) and Utter et al. (1989) suggest that neither spring-, summer-, nor fall-run chinook salmon represent monophyletic lineages in the Pacific Northwest. Both authors found that, in general, geographic proximity was more important than run-timing in predicting similarities between stocks. Thus, fish with different run-times from the same area often were more similar than were fish from different areas with the same run-timing. This pattern suggests that run-time differences may have evolved independently a number of times following colonization of a new area by one form. Foote et al. (1989) concluded that a similar phenomenon-derivation of freshwater kokanee from anadromous sockeye salmon-occurred numerous times within the species O. nerka.

However, in spite of this general pattern, in some cases substantial differences are found between populations from the same geographic area having different run-times. Striking examples of this are the pronounced genetic and life history differences between fall chinook salmon and spring/summer chinook salmon in the Snake River. In that drainage, fall chinook salmon spawn in lower elevations, generally main-stem areas and migrate to sea as subyearlings, whereas spring and summer chinook salmon spawn in smaller, higher elevation tributaries and outmigrate as yearlings. Several studies have also shown large allele frequency differences between spring/summer- and fall-run fish in the Snake River. Therefore, because of compelling evidence that fall chinook salmon in the Snake River are reproductively isolated from spring and summer chinook salmon, Snake River fall chinook salmon are being considered separately from the other two forms in this ESA evaluation (see also NMFS Status Review for Snake River Spring and Summer Chinook Salmon; Matthews and Waples 1991).

Distinct Population Segments

We next address the question whether Snake River fall chinook salmon constitute one or more ESUs. If they are not an ESU, then presumably they are part of a larger ESU that would have to be identified and defined. To be considered an ESU, and hence a "species" under the ESA, a population must satisfy two criteria: it must be reproductively isolated, and it must contribute substantially to the ecological/genetic diversity of the biological species.

Reproductive isolation

Historically, the primary spawning areas for Snake River fall chinook salmon were geographically well separated from fall chinook salmon habitat in the upper Columbia River. Since about 1960, impassable dams in the Hells Canyon complex have restricted Snake River fall chinook salmon to the lower 397 km of the river, closer to upper Columbia River spawning grounds. However, four dams on the lower Snake River also flooded spawning habitat there, thus increasing possibilities for isolation of the remaining population from Columbia River populations (ODFW 1991a). In addition, tagging data suggest at most a low level of straying of upper Columbia River fish into the Snake River through the mid-1980s. Furthermore, protein electrophoretic data gathered over several years in the late 1970s and early 1980s showed consistent genetic differences between Snake River and upper Columbia River fall chinook salmon that would not be expected unless reproductive isolation between the two forms had been strong for a substantial time.

The reason for the genetic similarity of samples from the Deschutes River and Marion Drain to Snake River fall chinook salmon is not clear at this time. The Marion Drain is a channel that facilitates return to the Yakima River of water used for irrigation. It was dug earlier in this century and, for reasons that are not well understood, has attracted spawning populations of salmon and steelhead. Another sample from the nearby Yakima River is genetically more similar to other upper Columbia River fall chinook salmon than to the Marion Drain sample (Busack 1991b). It is possible that fish displaced by destruction of spawning habitat in the Snake River have colonized the drainage ditch in recent years. The Deschutes River appears to have sustained a native run of "fall" chinook salmon, although tagging studies have shown that fish spawning there in the fall cross Bonneville Dam during the June-July period designated for summer-run fish (Howell et al. 1985). WDF plans to take new samples for genetic analysis from the current fall run in the Deschutes River (C. Busack5), and results may help to better define the relationship between these fish and the Snake River population.

There is genetic and tagging evidence that, beginning in the mid-1980s, Columbia River fall chinook salmon of hatchery origin have strayed into the Snake River and have been used for brood stock at Lyons Ferry Hatchery. The effects of this straying are considered in the next section.

Ecological/genetic diversity

Genetic differences detected by protein electrophoresis between fall chinook salmon and spring/summer chinook salmon in the Snake River are quite substantial and clearly reflect independent evolutionary lineages. As a group, the "upriver bright" fall chinook salmon are also clearly distinguished genetically from lower Columbia River "tule" fall chinook salmon. Within the "upriver bright" group, consistent differences indicating reproductive isolation are also found between populations in the Snake and upper Columbia Rivers. These latter differences, however, are not large quantitatively. Thus, the genetic data are consistent with the existence of adaptive differences between the two "upriver bright" populations, but the data do not in themselves provide strong evidence for such adaptations. In evaluating the contribution of Snake River fall chinook salmon to ecological/genetic diversity of the species, therefore, attention should focus on other factors.

As the largest historic producer of chinook salmon in the world, the Columbia River Basin clearly plays an integral role in maintaining the long-term health and viability of the species. In turn, the Snake River, which is the largest tributary of the Columbia River, contributes substantially to the ecological diversity and productivity of the Basin. Prior to construction of the Hell's Canyon complex of dams, the Snake River was the most important natural production area in the Basin for fall chinook salmon (Fulton 1968).

Among North American chinook salmon populations with the "ocean" type juvenile life history pattern, the fall run in the Snake River historically migrated the farthest from the ocean (over 1500 km). Fall chinook salmon in the upper Columbia River also undergo a lengthy freshwater migration. In contrast, "ocean" type chinook salmon in other major North American river systems (Sacramento, Klamath, Fraser) migrate no more than a few hundred kilometers into fresh water. Although the Snake River population is currently restricted to habitat in the lower river, genes associated with the more lengthy migration may still reside in the population. In general, longer freshwater migrations in chinook salmon are associated with more extensive oceanic migrations (Healey 1983). Thus, maintaining populations occupying habitat well inland can be important in maintaining diversity in the marine ecosystem as well.

Habitat characteristics and adult ocean distribution provide the strongest evidence for adaptive differences between fall chinook salmon in the Snake and upper Columbia Rivers. The best-documented environmental difference between the Snake and upper Columbia Rivers is water temperature. In the summer months, water temperatures in the Snake River can exceed 25°C, and during this period average monthly temperatures are often 6-8°C higher than the upper Columbia River. These temperature regimes may exclude juvenile fall chinook salmon from rearing in the main-stem Snake River during this period, thus encouraging the evolution of behavioral mechanisms to avoid the warm water. Summer water temperatures in fall chinook salmon habitat in the lower Columbia River are also typically much lower than the Snake River (Van Hyning 1973).

The high summer water temperatures in the Snake River suggest that, if the present populations were lost, other chinook salmon might have difficulty successfully colonizing this area. Even if exogenous adults can move into the area (as has occurred recently with stray hatchery fish of upper Columbia River origin), their progeny may have poor survival in the new environment. Of course, this is difficult to demonstrate experimentally without placing the native population at risk. However, numerous other studies show that, in general, transfers of Pacific salmon within the historic range of the species have not been successful (e.g., Withler 1982). This is particularly true of attempts to establish lower river fish in upriver areas. Presumably, this failure reflects the importance of local adaptation and the inability of transplanted fish to home accurately.

Substantially higher pH, alkalinity, and turbidity (relative to the upper Columbia River) also characterize Snake River water. Although the effects of these factors on salmonids are not as well understood as are thermal effects, these environmental differences also may lead to local adaptations.

Tagging studies demonstrate substantial differences in ocean distribution between fish from the Snake and upper Columbia Rivers, and the patterns are consistent over several years. Snake River fish have a more southerly distribution, with a significant proportion (about 20-50%) of recoveries taken in southern Oregon and California and very few (<5%) in Alaska. The converse is true of upper Columbia River fish; substantial catches occur in Alaska (20-35%) and few in southern areas (<10%). These differences indicate that the two populations utilize the marine habitat in different ways. It is important to the long-term health of the species to maintain such interpopulation differences because 1) this diversity allows the species as a whole to more effectively utilize available habitat, and 2) loss of this diversity would place the species at greater risk from unpredictable changes in the environment.

Species determination

Available evidence indicates that historically, fall chinook salmon in the Snake River were substantially isolated from other chinook salmon populations. The importance of the Columbia River Basin to the long-term health of the biological species is clear. Among chinook salmon populations with "ocean" type life-history strategies, Snake River fall-run fish are exceptional in the length of their freshwater migration. High summer temperatures require special adaptations in juvenile behavior and reduce the possibility that other populations could rapidly colonize the habitat. Together with the distinctive ocean distribution of Snake River fall chinook salmon, these factors argue for the important role the Snake River population plays in contributing to the ecological/genetic diversity of the species. We therefore conclude that historically, Snake River fall chinook salmon were an ESU of the biological species O. tshawytscha. Although hydropower development has drastically reduced available habitat for this population, evidence indicates that it remained distinct at least through the early 1980s. This is the same conclusion reached by Utter et al. (1982), who reviewed evidence for distinct population segments of fall chinook salmon in conjunction with an ESA evaluation a decade ago. As noted above, further work is necessary to establish the relationship between the Snake River population and fall chinook salmon in the Deschutes River and the Marion Drain.

Evidence for the existence of multiple distinct populations of fall chinook salmon within the Snake River Basin is scant. Some population subdivision may occur (or may have in the past), but the inability to clearly identify and sample discrete spawning populations has precluded a more definitive study of this possibility.

Information that has become available only within the last year raises some questions regarding the present status of the Snake River population. In recent years, strays from hatcheries producing upper Columbia River fall chinook salmon have appeared in the Snake River in increasing numbers and have been incorporated into brood stock at Lyons Ferry Hatchery. According to data collected in 1990, a high percentage of adults taken at Lower Granite Dam and on fall chinook salmon spawning grounds are estimated to be of hatchery origin, including strays from the Columbia River. Protein electrophoretic data confirm that introgression of upper Columbia River genes into Lyons Ferry Hatchery brood stock has occurred, but there is no direct information about the genetic effects of hatchery strays on the wild Snake River population.

Although the NMFS Biological Review Team (BRT) concluded that, historically, Snake River fall chinook salmon were an ESU, it is not so clear whether this is still the case. One viewpoint is that introgression from Columbia River hatchery strays has caused the Snake River population to lose the qualities that made it "distinct" for ESA purposes. Evidence in support of this viewpoint includes genetic and tagging data documenting effects of straying on Lyons Ferry Hatchery brood stock, estimates that in 1990 a high proportion of fish passing Lower Granite Dam and found on nearby spawning grounds were hatchery strays, and the lack of any positive information documenting the continued existence of "pure" wild fish. However, given that 1) an ESU was present until at least the early 1980s, 2) substantial straying of upper Columbia River hatchery fish has occurred only within the last generation, and 3) no direct evidence exists for genetic change to wild fall chinook salmon in the Snake River, the BRT felt it would be premature to conclude that the ESU no longer exists.

Status of the ESU

The BRT evaluated a number of factors in considering if Snake River fall chinook salmon are threatened or endangered. The current population occupies a fraction of its former range, the remaining (and, historically, the most productive) habitat having been inundated by reservoirs or blocked by dams. Although historical abundance of fall chinook salmon in the Snake River is difficult to estimate, adult returns have declined by about three orders of magnitude since the 1940s, and perhaps by another order of magnitude from pristine levels.

Relatively precise estimates of the number of fall chinook salmon entering the Snake River have been available only since the completion of Ice Harbor Dam. The estimated numbers of wild spawners in 1987, 1989, and 1990 are the second, fourth, and first lowest on record, respectively. In 1990, just 78 wild fish are estimated to have passed Lower Granite Dam, only 31% of the number in the next lowest year (1987).

We applied the model of Dennis et al. (1991) to time series of adult counts at Ice Harbor Dam (1960-present) and Lower Granite Dam (1980-present). Inputs for the model were 5-year running sums of the number of adults at the uppermost dam on the lower Snake River (this running sum is termed the "index"). For the more recent time series, estimated numbers of wild adults at Lower Granite Dam (Table 5) were used. Two time series were considered because predictions of the model can vary substantially depending on the time series chosen. In particular, a time series that spans a fundamental change in parameters affecting the population can give misleading results. The first time series provides information for the period following construction of the first of the four dams on the lower Snake River, and the second shows trends for the period following construction of the last dam (1980 is the first year for which almost all returning adults had to outmigrate through Lower Granite Dam as juveniles). Although the first series is longer (and, other things being equal, would be preferred), the second series may more accurately reflect the population's response to conditions during the past decade. A test described by Dennis et al. (1991) verified that a statistically significant change in growth rate parameters of the population occurred in 1980.

Based on the time series beginning in 1960, the Dennis model indicates that extinction of the ESU is a virtual certainty in the absence of any changes (probability of the index dropping below one fish within 100 years > 99.9%). Predictions are not quite so bleak for the more recent time series, but even so, the model estimated the probability of extinction within 100 years as 10.8%.

In light of the above factors, and further considering that a) draught conditions have likely adversely affected juvenile survival in several recent years, reducing the prospects for recovery in the near future as these year classes return as adults, and b) there is clear evidence that stray hatchery fish of non-Snake River origin pose a serious threat to the genetic integrity of the wild population, the BRT concluded that Snake River fall chinook salmon face a substantial risk of extinction if present conditions continue.


Go to Citations

Table of Contents