U.S. Dept Commerce/NOAA/NMFS/NWFSC/Publications

NOAA-NWFSC Tech Memo-24: Status Review of Coho Salmon
Coho Salmon Life History

Several types of biological evidence were considered in evaluating the contribution of west coast coho salmon to ecological/genetic diversity of the biological species under the ESA. Life history traits examined for naturally spawning coho salmon populations included smolt size and outmigration timing, age and size at spawning, river entry timing, spawn timing, fecundity, and ocean migration patterns based on marine code-wire-tag (CWT) recoveries. The primary objective of this examination was to determine regional patterns in these traits that might indicate stock groupings, and to identify geographic areas where patterns change. Because these traits are believed to have both genetic and environmental bases, similarities among populations could indicate either shared genetic heritage or similar responses to shared environmental conditions.

Compilation and comparison of life history trait information on a regional scale is confounded by several factors. First is the high spatial and temporal variability of these traits, which is presumably in large part a reflection of high environmental variability. Fish examined in different years or from different locations or habitats within a basin may display different life history characteristics, making it difficult to estimate values that characterize historic or basinwide populations. This variability creates considerable noise, which may be as large as differences between geographically distant populations, and may mask subtle regional patterns. High interannual variability also means that results of studies may be sensitive to the time period over which they were conducted. For example, measurements of many life history traits for Oregon coho salmon during the 1983 El Niño were very different than those in the years preceding and succeeding that event (Johnson 1988).

A second factor which has confounded data compilation is lack of information on life history traits, especially the lack of long-term data sets, from most naturally spawning populations. Very little information on any life history trait is available for many central California populations, and the available information on some life history traits of naturally spawning populations in the better-studied areas (southern British Columbia and Washington) is also lacking or limited. There appear to be population or regional differences in some traits, such as spawner size, relative fecundity, body morphology, and egg size, in coho salmon (Beacham 1982, Hjort and Schreck 1982, Taylor and McPhail 1985, Swain and Holtby 1989, Fleming and Gross 1990, Murray et al. 1990) and other salmon species (Riddell and Leggett 1981, Beacham and Murray 1987, Jonsson and L'Abée-Lund 1993), but not enough information is available on these traits to examine coastwide patterns.

A third confounding complication is that anthropogenic activities such as land-use practices (Hartman et al. 1984, Holtby 1987) and artificial propagation (Steward and Bjornn 1990, Flagg et al. 1995) may alter life history traits. To help limit this bias, life history trait comparisons in this status review have focused on naturally spawning populations. However, because of the widespread practice of off-station plants of hatchery fry and smolts, many studies of naturally spawning populations may have included first- or second-generation hatchery fish. Life history trait information from hatchery populations was used only when insufficient information from naturally spawning populations was available, as in the case of ocean migration patterns. In this case, comparisons of ocean migration patterns in paired natural and hatchery populations were made to confirm that patterns exhibited by hatchery populations could serve as a surrogate for those of naturally spawning populations. As with environmental variability, the effects of anthropogenic activities may confuse the determination of average life history traits yet are difficult to factor out.

Because of these potential sources of variability, we felt that statistical analyses of life history traits would not be particularly informative. Instead, data were collected from as many sources as possible from each system to give some indication of the average results, and older data sets were especially sought to indicate coho salmon population traits prior to the proliferation of hatchery programs which produced fish with relatively high survival rates.


From central British Columbia south, the vast majority of coho salmon adults are 3-year-olds, having spent approximately 18 months in fresh water and 18 months in salt water (Gilbert 1912, Pritchard 1940, Marr 1943, Briggs 1953, Shapovalov and Taft 1954, Foerster 1955, Milne 1957, Salo and Bayliff 1958, Loeffel and Wendler 1968, Wright 1970). The primary exception to this pattern are jacks, sexually mature males that return to freshwater to spawn after only 5-7 months in the ocean. However, in southeast and central Alaska, the majority of coho salmon adults are 4-year-olds, having spent an additional year in fresh water before going to sea (Godfrey et al. 1975, Crone and Bond 1976). The Keogh River at the north end of Vancouver Island produces relatively low (8-11% of total outmigrating smolts) but consistent numbers of 4-year-old adults (Irvine and Ward 1989), suggesting that the transition zone between predominantly 3-year-old and 4-year-old adults occurs somewhere between central British Columbia and southeast Alaska.

Trends in Jacks--Drucker (1972) suggested that there is a latitudinal cline in the proportion of jacks in a coho salmon population, with populations in California having more jacks and those in British Columbia having almost none. Although the production of jacks is a heritable trait in coho salmon (Iwamoto et al. 1984), it is also strongly influenced by environmental factors (Shapovalov and Taft 1954, Silverstein and Hershberger 1992). The proportion of jacks in a given coho salmon population appears to be highly variable and may range from less than 6% to over 43% over 9-35 years of monitoring (Shapovalov and Taft 1954, Fraser et al. 1983, Cramer and Cramer 1994).

Some systems have also shown long-term changes in the proportion of jacks produced. The Tenmile Lakes system (Oregon) historically produced large numbers of jacks (Morgan and Henry 1959) but no longer does (Ursitti 1989), presumably because of altered freshwater predation pressures (Reimers et al. 1993). Because of this high level of variability in the relative production of jacks in a population, the proportion of jacks appeared to be a poor indicator of regional patterns and was not pursued further.


Because larger females have higher fecundity than smaller females, any comparison of fecundity between populations is confounded by differences in female size (Rounsefell 1957). Consequently, comparisons of fecundity should be adjusted for size (Beacham 1982), which requires measurements of both size and fecundity from the same individuals. Available information that provides these measurements for naturally spawning coho salmon populations was insufficient to adequately evaluate patterns of relative fecundity in west coast coho salmon.

However, two analyses of fecundity of coho salmon provide some insight. Beacham (1982) found differences in relative fecundity between coho salmon populations in California, Washington, British Columbia, and Alaska that he attributed to river size, but only weak increases in fecundity with increasing north latitude. In contrast, Fleming and Gross (1990) found significant increases in fecundity with increasing north latitude. The fact that separate researchers reached different conclusions about fecundity over approximately the same area suggests that the observed relationships are strongly influenced by both the data used and temporal and spatial variability in fecundity. This degree of variability may interfere with the ability to detect differences between areas. Other researchers have reported that fecundity can be effectively used to differentiate other salmon populations (Gard et al. 1987), while fecundity of Clupeidae showed strong latitudinal trends in eastern North America (Jessop 1993).

Smolt Size and Outmigration Timing

There does not appear to be any clear, regional pattern for either smolt outmigration timing (Fig. 10, Appendix Table C-1) or smolt size (Fig. 11, Appendix Table C-2) in west coast coho salmon. Regardless of the area of origin, peak outmigration timing generally occurs in May, with some runs earlier or later, and with most smolts measuring 90-115 mm fork length. Smolts from southwest Washington and the Klamath River Basin (northern California) tend to be relatively large, but this is possibly due to influences of off-station hatchery plants. Large smolts observed in Tenmile Lakes were thought to have resulted from a productive lake-rearing environment (McGie 1970).

Smolt outmigration timing and smolt size appear to respond to small-scale habitat variability. Smolts residing in ponds or lakes often have different outmigration timing and are a different size than smolts residing in streams within the same basin (Swales et al. 1988, Irvine and Ward 1989, Rodgers et al. 1993, Nielsen 1994). Both smolt outmigration timing and size exhibit considerable interannual variation; mean smolt sizes from a single system can vary by over 15 mm between years (Blankenship et al. 1983; Fraser et al. 1983; Lenzi 1983, 1985, 1987), while peak outmigration timing can vary by several weeks to a month (Shapovalov and Taft 1954; Salo and Bayliff 1958; Blankenship and Tivel 1980; Seiler et al. 1981, 1984; Blankenship et al. 1983; Fraser et al. 1983; Lenzi 1983, 1985, 1987).

Because of their responses to small-scale habitat variability, smolt size and outmigration timing have also been shown to be affected by anthropogenic activities, including habitat degradation (Moring and Lantz 1975, Scrivener and Andersen 1984, Holtby and Scrivener 1989), habitat restoration (Johnson et al. 1993, Rodgers et al. 1993), and flow control (Fraser et al. 1983). These factors thoroughly complicate the assessment of any regional pattern that may exist for either trait, since these activities have occurred throughout the range of coho salmon. Sampling design may also influence reported smolt sizes and outmigration timing.

Despite these confounding problems, it appears that regional patterns for some aspects of smolt outmigration do exist. Spence (1994 App.) conducted a detailed evaluation of coho salmon smolt outmigration timing and the factors that appear to influence it. Using only long-term data sets to minimize interannual variability, he found that the duration and between- year variation of smolt outmigration timing exhibited distinctive patterns between areas. These areas were identified as the Columbia River and south, Puget Sound/Strait of Georgia, and central/north British Columbia and Alaska. Spence concluded that these patterns were likely driven by differences in the predictability of nearshore ocean conditions.

Adult Run Timing

In general, river entry and spawn timing showed considerable spatial and temporal variability. Despite this high variability, some regional patterns were observed. Most west coast coho salmon enter rivers in October (Fig. 12, Appendix Table C-3) and spawn from November to December and occasionally into January (Fig. 13, Appendix Table C-4). However, coho salmon from central California enter rivers much later, in late December or January, and spawn immediately afterwards, probably in response to late peak river flows of limited duration. Consequently, central California fish spend little time between river entry and spawning, while northern stocks may spend 1 or 2 months in fresh water before spawning (Flint and Zillges 1980, Fraser et al. 1983). Stocks from British Columbia, Washington, and the Columbia River often have very early (entering rivers in July or August) or late (spawning into March) runs in addition to normally timed runs.

Coho salmon river entry timing is influenced by many factors; one of the most important appears to be river flow (Shapovalov and Taft 1954, Salo and Bayliff 1958, Sumner 1953, Eames et al. 1981, Lister et al. 1981). Coho salmon wait for freshets before entering rivers, so a delay in fall rains delays river entry and, potentially, spawn timing as well. Delays in river entry of over a month are not unusual (Salo and Bayliff 1958, Eames et al. 1981). Many small California systems have sandbars which block their mouths for most of the year except during winter. In these systems, coho salmon and other salmon species are unable to enter the rivers until sufficiently strong freshets break the sandbars (Sandercock 1991).

There is also considerable temporal variability in river entry and spawn timing, especially in large river systems. For example, the Skagit (northern Washington), Chehalis (southwest Washington), Columbia, and Klamath Rivers have coho salmon which enter freshwater over a broad period from August until December (WDF 1951, Leidy and Leidy 1984, WDF et al. 1993, J. Polos 1994 App.). In general, earlier migrating fish spawn farther upstream within a basin than later migrating fish, which enter rivers in a more advanced state of sexual maturity (Sandercock 1991).

On a smaller scale, Lister et al. (1981) found that spawn timing of coho salmon in tributaries of the Cowichan River (British Columbia) was strongly correlated to tributary water temperature: coho salmon spawning in warmer tributaries spawned later than those spawning in colder tributaries. All these factors make determinations and comparisons of average or peak river entry and spawn timing difficult because of the high spatial and temporal variability exhibited within basins. Compared to normal run times, river entry of some coho salmon stocks are exceptionally early or late; these stocks are often referred to as summer or winter runs, respectively (Godfrey 1965), and are thought to have evolved in response to particular flow conditions (Sandercock 1991). The relationship between populations with unusually timed runs and normally timed runs within the same basin is not well understood. For example, in some cases, such as the Soleduck (Washington coast) and Clackamas (Willamette River) Rivers, differently-timed, sympatric runs are thought to be largely reproductively isolated from each other (Houston 1983, Cramer and Cramer 1994), while in the Grays Harbor Basin, there is believed to be reproductive overlap (WDF et al. 1993). Exceptionally timed runs are found in numerous geographic areas. However, because there is no evidence to suggest that all runs of a certain type are closely related, we considered differently-timed runs to be a component of overall life history diversity within each area.

Spawner Size

Regional variation--Like the other life history traits discussed above, adult spawner size in naturally spawning populations shows considerable spatial and temporal variability which may obscure regional patterns of variation. Except for the tendency of some populations of Puget Sound/Strait of Georgia coho salmon to be slightly smaller, there did not appear to be obvious patterns for adult spawner size ( Fig. 14, Appendix Table C-5). Similarly, Sandercock (1991) observed no obvious patterns of spawner size across the range of the species.

Variability in spawner size results from numerous factors and occurs both over the course of a run and between years (Chapman 1940, Salo and Bayliff 1958, Shapovalov and Taft 1954, Fraser et al. 1983). Spawner size is affected by migration patterns (Allen 1959), genetic heritage (Hershberger et al. 1990), and conditions experienced during the last year of growth (van den Berghe and Gross 1989), especially during anomalous ocean events such as El Niños (Johnson 1988). In addition, runs that enter freshwater later tend to have larger spawners than those entering earlier (Sandercock 1991), and coho salmon that spawn in mainstem areas may be larger than those spawning in tributaries (Lister et al. 1981).

Decrease in spawner size--One factor which has thoroughly confounded comparisons of spawner size is that coho salmon, throughout their range, are declining in size over time, and the rates of decrease are population- or area-specific (Ricker 1981, Bigler and Helle 1994). Decreases in size for other salmon species have also been observed (Ricker 1981, Healey 1986, Ishida et al. 1993, Bigler and Helle 1994).

Table 1 shows statistics of size regressed on time for coho salmon collected from various fisheries or locations in British Columbia, Washington, Oregon and California. Although the data sets used for these regressions include measurements made on both hatchery and naturally spawning fish, they provide the most reasonable proxies for specific information on size of naturally spawning coho salmon, which are largely unavailable. In most cases, the slope of the relationship was negative, although not always statistically different than zero (P < 0.05) Table 1). Differences in the rate of decline in adult size among areas, such as those indicated by the varied regression slopes in Table 1, make regional patterns in adult size difficult to interpret. In addition, long-term data sets on size of all commercially-caught and troll- caught coho salmon in Washington State indicated declines in size that began in the mid-1950s (Fig. 15) (Wright 1970, WDF 1981, Hoines 1994). This suggests that declines in adult size in other areas, such as Puget Sound, may have begun earlier than the available data sets indicate. However, no other evidence exists that indicates earlier declines in size.

Table 1. Results of regression analysis of coho salmon size (fork length (cm) or total weight (kg)) over time for selected populations and fisheries in British Columbia, Washington (WA), Oregon (OR), and California. Slopes which are statistically significant at P < 0.05 are indicated in bold.





P value
Big Qualicum (Vancouver Is.)spawnerslength 59-72 0.144-3.570.181 a
WA total commercial catch all weight 35-91 0.669-0.030.000 b
All WA commercial troll troll weight 54-92 0.718-0.040.000 c
Apple Cove Point test fishery purse seine length 85-940.299-0.53 0.102 d
Apple Cove Point test fishery purse seine weight 85-940.468-0.10 0.029 d
All Puget Sound net weight 68-91 0.587 -0.060.000 b
Big Beef Creek (Hood Canal) spawnerlength 78-91 0.081-0.430.325 e
Nooksack in river weight 77-93 0.299 0.03 0.023 f
Skagit in river weight 78-93 0.581 -0.060.001 f
Stillaguamish in river weight 80-93 0.082 -0.020.320 f
Duwamish/Green in river weight 72-93 0.567 -0.090.000 f
Puyallup in river weight 72-93 0.714 -0.090.000 f
Nisqually in river weight 72-93 0.663 -0.080.000 f
Deschutes River spawnersa length 78-92 0.514 -0.960.003 e
Skokomish in river weight 79-90 0.207 -0.040.138 f
Dungeness in river weight 75-83 0.170 -0.060.271 f
Elwha in river weight 77-93 0.471 -0.080.002 f
Sooes/Waatch in river weight 72-93 0.095 -0.020.163 f
Quillayute in river weight 72-93 0.560 -0.050.000 f
Hoh in river weight 72-93 0.258 -0.030.016 f
Queets in river weight 77-93 0.280 -0.030.011 f
Quinault in river weight 77-93 0.432 -0.030.001 f
Chehalis in river weight 77-93 0.021 0.01 0.584 f
Bingham Creek (Chehalis R.)spawnersalength 83-91 0.300-0.740.127 e
Columbia total catchin riverweight 54-92 0.400 -0.030.000 g
Columbia in river length 76-93 0.079 -1.290.258 h
All Oregon troll troll weight 52-90 0.150 -0.030.015 i
Oregon troll, Augusttroll weight 52-90 0.117 -0.040.033 i
Nehalem spawners length 78-92 0.053 -1.950.408 j
Alsea spawners length 78-92 0.050 -1.730.422 j
Coquille spawners length 78-92 0.012 1.38 0.695 j
Rogue spawners length 76-86 0.007 -1.500.794 k
All CA troll troll weight 52-90 0.323 -0.050.000 h
CA troll, Septembertroll weight 52-90 0.258 -0.050.001 h
Klamath estuary test fishery beach seine length 81-910.0560.280.509 l

aMales and females combined.
Sources: a: Fraser et al. 1983; b: WDF 1981, Hoines 1994; c: Wright 1970, WDF 1981, Hoines 1994; d: Anderson and Milward 1992, S. Boessow 1994 App.; e: Seiler et al. 1981, C. Knudsen 1995 App.; f: WDFW 1994a; g: ODFW and WDF 1993; h: S. King 1994 App.; i: PFMC 1993b; j: S. Jacobs 1994a App.; k: ODFW 1989; l: Adair et al. 1982- 1985, Tuss et al. 1989, Kisanuki et al. 1991, Rueth et al. 1992.

The size of coho salmon adults in Puget Sound/Strait of Georgia is declining at a much faster rate than in other areas (Table 1). Coho salmon caught in in-river fisheries in Puget Sound decreased in weight by about 50% between 1972 and 1993, from average weights of approximately 4 kg to about 2 kg (Fig. 16). Whether the size of naturally spawning coho salmon in Puget Sound is also declining is largely unknown. Coho salmon weight in the Skagit River, a river managed for natural production (WDF et al. 1993), declined from about 3.5 kg in 1978 to about 2.5 kg in 1992 (Fig. 16), showing a clear, statistically significant downward trend (Table 1).

Big Beef Creek (Hood Canal) and Deschutes River (south Puget Sound) populations are the only naturally spawning Puget Sound coho salmon populations for which there are long- term (14-15 years) size data. These data show that average spawner length decreased between 1978 and 1991/92 from about 64 and 60 cm fork length (FL) for Deschutes River and Big Beef Creek populations, respectively, to approximately 53 cm FL for both populations (Fig. 17) (Seiler et al. 1981, Knudsen 1995 App.). The Deschutes River regression of length over time was statistically significant (P < 0.05), while that for Big Beef Creek was not (Table 1).

Because measurements from Big Beef Creek and Deschutes River were taken for length and those from the Skagit River were taken for weight, the two analyses are not directly comparable. To compare declines in size between these naturally spawning populations (Big Beef Creek, Deschutes and Skagit Rivers), length data were converted to weight data using the length- weight equation described by Holtby and Healey (1986). This equation was calculated from coho salmon returning to Rosewall Creek on Vancouver Island and was used to estimate weight (log10weight (g) = 3.3183 [log10fork length (mm)]-5.843).

The regression of estimated weight over time for Deschutes River coho salmon was statistically significant (P = 0.003), and the regression statistics (slope = -0.07, r2 = 0.510) were quite similar to those from the Skagit River (P = 0.001, slope = -0.06, r2 = 0.581) (Table 1). In contrast, estimated weight regressed over time for Big Beef Creek fish was not statistically significant (P = 0.315, slope = -0.02, r2 = 0.084), and it exhibited about one-third of the rate of decline in size as either Deschutes or Skagit River coho salmon. This comparison suggests that rates of decrease in size over time of Skagit and Deschutes River coho salmon are roughly comparable, and they exceed the rate of decline of the naturally spawning population at Big Beef Creek. Whether other naturally spawning Puget Sound populations are declining in size at similar rates remains to be determined.

The average size of coho salmon caught in the 1994 Washington Department of Fish and Wildlife (WDFW) Apple Cove Point test fishery (2.2 kg, 55.6 cm FL) was larger than in the previous 2-3 years (1.4-2.1 kg, 47.9-52.5 cm FL in 1991-93) (Fig. 18) (Anderson and Milward 1992, S. Boessow 1994 App.). Two possible explanations for this increase are the near-absence of ocean harvest in 1994 and the low abundance of coho salmon, which led to the fishing restriction. Many salmon harvest methods are size-selective for larger fish (Ricker 1981, Healey 1986), and 1994 was the first year on record when almost all ocean harvest for coho salmon was halted (PFMC 1993a). The 1994 fishing restriction was implemented because the year class was expected to be extremely weak; however, this may have allowed fish that did survive to attain greater size since adult size is often inversely correlated with year class strength (Ishida et al. 1993).

Even including the 1994 data, the sizes of fish caught in the test fishery still show a decline over the period, and the decline in weight is statistically significant (Table 1). Whether continued relaxation of ocean fishing pressure and weak year classes would allow Puget Sound coho salmon to return to their previous size is not known. In any case, 1994 sizes are still smaller than those observed in 1985, and considerably smaller than the 3.6 kg reported for Puget Sound-caught coho salmon during 1915-26 (WDFG 1928).

It is not clear whether the dramatic size reductions observed in Puget Sound/Strait of Georgia coho salmon are due to harvest practices, effects of fish culture, declining ocean productivity, density-dependent effects in the marine and freshwater environments attributable to large numbers of hatchery releases, or a combination of these factors. Similarly, it is not known whether there have been permanent genetic changes related to size changes in these populations. Regardless of its cause or genetic basis, reduced adult size in itself poses a number of serious risks to natural populations of coho salmon, and could be a sign of other factors placing the population at risk.

Declines in adult size can have direct implications for individual reproductive success and population viability. As is the case in other salmon species, coho salmon fecundity is a nonlinear function of size (Fleming and Gross 1989), such that a small reduction in size can lead to a substantial reduction in fecundity. For example, using the length-fecundity relationship given by Shapovalov and Taft (1954, Fecundity = 0.01153 FL (cm)2.9403), a 17% decrease in spawner size (from 60 to 50 cm FL) results in a 42% reduction in fecundity (from 1,950 to 1,141 eggs). Knudsen (1995 App.) estimated that as female sizes decreased between 1960 and 1992 at four Washington state hatcheries (Skykomish, Simpson, George Adams, and Puyallup), coho salmon fecundity decreased by one-third to one-half.

Smaller coho salmon females also dig fewer and shallower redds than do larger females (van den Berghe and Gross 1984). This subjects the redds of smaller individuals to greater risk of destruction by superimposition of redds of larger individuals or by scouring from floods. Flooding frequency has increased throughout much of Puget Sound because of habitat degradation (Booth 1991), further decreasing the survival potential of redds created by small females. Smaller coho salmon may also be unable to consume prey items available to larger individuals, or may be more susceptible to some forms of predation (Holtby et al. 1990). There also appears to be some size advantage for anadromous fishes making long or strenuous migrations (Bernatchez and Dodson 1987, L'Abe-Lund 1991). All these factors suggest that smaller adults may be less able to reach spawning grounds and successfully spawn than larger adults, and this can directly affect population survival rates.

Ocean Migration

Coded-wire tag studies--Ocean distribution of coho salmon, inferred from marine recoveries of coded-wire-tagged fish, showed distinctive differences between regions. Coded-wire tags (CWTs) are primarily recovered in salt or fresh water as the salmon return to their natal streams after overwintering in the ocean. Consequently, CWT recovery patterns only indicate ocean migration patterns during the last few months of a 1 ½-year long migration. Although patterns of movement during earlier stages of ocean migration have been studied (e.g., Loeffel and Forster 1970, French et al. 1975, Hartt 1980, Miller et al. 1983, Hartt and Dell 1986, Pearcy and Fisher 1988), the studies are insufficiently broad in scope to adequately compare early migration patterns for coho salmon released from different areas. However, the extremely large number of CWTs released and recovered for coho salmon provides a detailed picture of the later stages of ocean migration.

Ocean distribution patterns based on CWT marine recovery patterns were determined from CWT recovery data for 66 North American hatcheries (Appendix Table C-6) from the Pacific States Marine Fisheries Commission's (PSMFC 1994) Regional Mark Information System. Marine (as defined in the database) CWT recoveries of adults and jacks were expanded for sampling but not for unmarked fish and were summed over all years for each hatchery by state or province of landing. These tag recoveries represent 1,892,270 coho salmon released between 1972 and 1991 and recovered between 1973 and 1992. Recoveries were made during an average of 10 years for each facility, with an average of 28,671 tags recovered per facility; only Warm Springs had less than 1,000 total recoveries, while six facilities had over 100,000 total recoveries (Appendix Table C-6).

The patterns of recoveries showed marked differences between areas, with extremely limited transition zones between areas (Fig. 19). Eight general CWT recovery patterns were identified, which can be grouped by releases from the following areas: 1) northern California and the Oregon coast south of Cape Blanco, 2) the Oregon coast north of Cape Blanco, 3) Columbia River, 4) Washington coast, 5) Puget Sound, Hood Canal, and Strait of Juan de Fuca, 6) southern British Columbia, 7) northern British Columbia, and 8) Alaska. Patterns observed in each of these areas are discussed below.

1) Northern California and Oregon south of Cape Blanco. Coho salmon released from the southernmost facilities (those south of Cape Blanco) had the most southerly recovery patterns: these fish were recovered primarily in California (65-92%), with some recoveries in Oregon (7-34%) and almost none (<1%) in Washington or British Columbia (Fig. 19). The recovery pattern of coho salmon released from the southernmost hatchery, Warm Springs (Russian River), had a much higher proportion of California recoveries (92%) than the other California and southern Oregon facilities. Whether this represents a unique recovery pattern, or results from the southerly location of the hatchery, is not known. No hatcheries in central California release or recover sufficient numbers of coho salmon tags to be used for comparison.

2) Oregon coast north of Cape Blanco. Tagged coho salmon from Oregon coast hatcheries north of Cape Blanco have a more northern distribution than those released farther south. The majority of Oregon coast coho salmon are recovered in Oregon (57-60%), followed by California (27-39%), Washington (2-9%), British Columbia (2-6%), and Alaska (<1%) (Fig. 19). The Butte Falls Hatchery is located on the Rogue River (south of Cape Blanco) but rears Umpqua River fish and releases them into the Umpqua River. The recovery pattern of these fish is most similar to the Oregon coast pattern, rather than nearby Cole Rivers Hatchery, indicating that ocean distribution is more heavily influenced by stock history and release location than it is by rearing location.

3) Columbia River. Coho salmon released from Columbia River hatcheries are recovered primarily in Oregon (36-67%) and Washington (22-54%), with lower but consistent recoveries from British Columbia (2-16%) and California (1-15%) (Fig. 19). Compared to Oregon coast coho salmon, Columbia River fish are recovered less frequently in California and more frequently in Washington. Although they share the same general recovery pattern, coho salmon from Washington-side Columbia River hatcheries are caught more frequently in Washington and British Columbia and less frequently in Oregon than those from Oregon-side hatcheries. This is presumably the result of a successful program aimed at increasing the Washington catch of Washington-produced Columbia River coho salmon (Hopley undated).

4) Washington coast. Coho salmon released from these coastal hatcheries are recovered primarily in British Columbia (37-74%) and Washington (18-53%), with few recoveries from Oregon (3-16%) and almost none (<1%) from California or Alaska (Fig. 19). Compared to Columbia River fish, Washington coastal hatchery coho salmon have much higher recovery rates from British Columbia and much lower recovery rates from Oregon and California. The Makah National Fish Hatchery produces coho salmon with exceptionally high recoveries in British Columbia. As this facility is closest to Canadian waters, its fish may be more susceptible to Canadian fisheries than other coastal stocks. Tagged coho salmon from the Simpson Hatchery (Chehalis River) have very low Oregon recoveries and relatively high Washington recoveries, perhaps reflecting high terminal marine fisheries.

5) Puget Sound, Hood Canal and Strait of Juan de Fuca. Coho salmon released from Puget Sound, Hood Canal, and Strait of Juan de Fuca hatcheries have approximately equal marine recoveries from Washington (23-72%) and British Columbia (27-74%), with few recoveries from Oregon (0-3%), and essentially none from Alaska or California (Fig. 19). Recovery patterns from this group are similar to those of fish released from the Washington coast, except that the Oregon catch is much smaller and the Washington catch tends to be higher. Lower Elwha River Hatchery fish have a recovery pattern which is intermediate between that of fish from hatcheries to the east and west; their high British Columbia recoveries are similar to recoveries of fish from the Makah National Fish Hatchery. These fish are seldom recovered in Oregon, however, which is the pattern typical of coho salmon from Puget Sound/Hood Canal and other Strait of Juan de Fuca hatcheries.

The proportion of Washington recoveries for coho salmon released from Puget Sound hatcheries generally increases from north to south, presumably because fish returning to south Puget Sound facilities spend more time in Washington waters. Removing Puget Sound recoveries from Washington recoveries to correct for marine fisheries that other populations are not subjected to, results in this group having recovery patterns intermediate between those from the Washington coast and British Columbia (Fig. 20). The corrected marine recoveries are highest in British Columbia (65-86%), followed by recoveries in Washington waters outside of Puget Sound (14-32%), with few recoveries in Oregon and California (<3% combined) (J. DeLong 1994 App.).

6) Southern British Columbia. Coho salmon released from Vancouver Island and south mainland British Columbia facilities are recovered primarily from British Columbia (90-99%) and Washington (0-9%), with few recoveries from Alaska (0-1%), Oregon (<1%), or California (0%) (Fig. 19).

7) Northern British Columbia. Marine CWT recovery patterns for fish released from British Columbian facilities north of Vancouver Island are intermediate between those from Alaska and Vancouver Island/south British Columbia mainland, with the majority of recoveries from British Columbia (61-85%), and the remainder from Alaska.

8) Alaska. Tagged coho salmon released from Alaskan facilities were overwhelmingly recovered in Alaskan waters (>98%), with the remainder captured in British Columbia (Fig. 19).

The methodology used in this analysis did not address several sources of variability that may have altered the observed patterns. For example, experimental release groups may have had unique migration patterns compared to nonexperimental production releases from the same facility, and observed migration patterns were probably also affected by differences in the number of tags released and recovered each year, and by interannual variation in migration patterns. However, addressing these factors was beyond the scope of our status review. Because the observed differences in recovery patterns between areas were large and often represented presence or absence of recoveries by state or province rather than differences of a few percentage points, manipulations to correct for sources of variability would be expected to clarify, rather than cloud, recovery patterns. For example, Puget Sound coho salmon are not recovered in California, and no amount of data manipulation is likely to change that fact.

Although there appear to be differences in CWT recovery patterns between hatchery populations from the eight areas, it is reasonable to ask whether hatchery migratory patterns are similar to those of nearby naturally spawning populations and can therefore be used as a surrogate for naturally spawning populations. In order to address this uncertainty, CWT recovery patterns of the few naturally spawning populations that have been tagged were compared to those of nearby hatchery populations. In most cases, recovery patterns of the two groups were quite similar to each other (Fig. 21) and to the regional pattern (Garrison and Carmichael 1982, Garrison 1985, Cramer and Cramer 1994, Lestelle and Weller 1994, PSMFC 1994). The two populations (from the Hoko and Skokomish Rivers) which were less similar to nearby hatcheries are purposely avoided by terminal fisheries targeting the hatchery runs (WDF et al. 1992).

Other methods--An assessment of differences in ocean migration patterns independent of CWT recovery patterns was also made by considering changes in adult size from different areas during anomalous years. Interannual variation in adult size is largely caused by variation in growth rates during the last year in the ocean (van den Berghe and Gross 1989). Assuming that variation in ocean productivity is area-specific, differences in ocean migration patterns could cause differences in growth rates and therefore in adult size, in addition to other factors which may also influence adult size. Distinctive differences in adult size between areas were apparent during anomalous years. For example, adult coho salmon from the Columbia River and Oregon coast north of Cape Blanco experienced a dramatic decrease in size in 1983 during the strong El Nio (Johnson 1988), and underwent smaller decreases in size in 1989 and 1992 (Fig. 22) (S. Jacobs 1994a App., S. King 1994 App., S. Markey 1994 App.). Coho salmon from other areas did not exhibit the marked size decrease in 1983 but showed decreases during other years (Fig. 22-23). For example, Rogue River fish declined in size in 1979 and 1982 (ODFW 1989), Washington coast fish size declined slightly in 1976, 1989 and 1992 (WDFW 1994a), and Puget Sound coho salmon size declined in 1976, 1984, and 1993 (WDFW 1994a).

These patterns suggest that the ocean environment experienced by these groups were different, at least during the anomalous years; this difference would occur if ocean migration patterns were also different. The CWT recovery patterns (Fig. 19, Appendix Table C-6) generally agree with groupings based on observed patterns in adult size: Rogue River (Cole Rivers Hatchery) fish are predominately recovered in California; both Oregon coast and Columbia River coho salmon have high recovery rates from Oregon, and low rates from Washington and British Columbia; and Washington coast and Puget Sound/Hood Canal coho salmon have high recovery rates from Washington and British Columbia, and low recovery rates from Oregon. Although the true causes of unusually small adult sizes from some areas and not others is not known, it is likely that ocean migration patterns are a factor.

Ocean migration patterns and genetic heritage--This discussion assumes that differences in migratory patterns between areas, as inferred from CWT recovery patterns or changes in adult size, reflect differences in the genetic heritage of those groups. Several lines of evidence support the notion that ocean migration patterns have some genetic basis. For example, CWT recovery patterns of local and transplanted coho salmon released from the same general area are often different. Oregon Aqua Foods (Yaquina Bay) and Anadromous Inc. (Coos Bay) began coho salmon production using primarily Puget Sound stocks (Wagoner et al. 1990, Borgenson et al. 1991). The CWT recovery patterns from these fish are much more northerly than those of other Oregon coast stocks (Table 2) (PSMFC 1994), despite the high Oregon recoveries, presumably from terminal fisheries, which target these two stocks. Similarly, Alsea and Klaskanine coho salmon were released from California hatcheries (Jensen 1971). Klaskanine recovery patterns were much more northerly than patterns of local stocks, although Alsea patterns were not (Table 2). The WDFW has also based much of its Columbia River coho salmon production on migration patterns of different stocks (Hopley undated). The agency has concentrated production on Type N (north-turning) stocks because they are caught more frequently by Washington fishers than Type S (south-turning) stocks. Other studies using different salmon species have indicated that ocean migration patterns are a heritable characteristic (Nicholas and Hankin 1988).

Table 2. Comparison of ocean recovery patterns of marked local and exotic coho salmon stocks released from the same general areas (PSMFC 1994 and Jensen 1971).
Local/ Recovery (%) by state/provincea
exotic Stock Release location Years BC WA OR CA
Exotic Puget Sound Anadromous Inc., Coos Bay75, 77-8716 9 71 5
Exotic Puget Sound Oregon Aqua Foods, Yaquina Bay74-89 13 12 69 5
Local Oregon coast Oregon coast north of Cape Blanco73-89 2-6 2-9 57-6027-39
Exotic Klaskanine R. Mad River 57, 61 42b 59
Local Pudding Cr. Pudding Creekc 57, 61 3b 98
Local Mad River Mad River Hatchery75, 78-79, 84-86 0 1 20 79
Exotic Alsea River Noyo River 61-62 9b 91
Local Pudding Cr. Pudding Creekc 61-62 11b 90

aPercentages may not add to 100% because of rounding.
bIncludes marine recoveries from both Washington and Oregon.
cPudding Creek is approximately 165 km south of the mouth of the Mad River and 3 km north of the mouth of the Noyo River.

In conclusion, it appears that at least some portion of ocean migration patterns are genetically based. Given the similarity of recovery patterns for hatchery and nearby naturally spawning populations, hatchery and naturally spawning coho salmon from the eight different regions have distinctive ocean migration patterns, suggesting similar genetic heritages.


Disease resistance is listed as one of several phenotypic traits to consider when determining the ecological/genetic importance of salmon populations under the ESA (Waples 1991b, p. 14). Using this guideline, the resistance to Ceratomyxa shasta of most Columbia River coho salmon was one of many factors contributing to the conclusion that lower Columbia River coho salmon were a historical ESU (Johnson et al. 1991). It was recently suggested that coho salmon from the Nehalem River, Oregon, qualify as their own ESU because they are the only Oregon coast stock which is resistant to C. shasta (Cramer 1994). Aside from the fact that numerous factors in concert are used to determine ESU boundaries, several key questions remain to be adequately answered concerning resistance to, and the historical distribution of, C. shasta in the Nehalem River.

At present, there is considerable confusion surrounding the historical resistance of populations to C. shasta, likelihood of C. shasta resistance detection, and the rate at which populations may acquire resistance (Zinn et al. 1977). The documented distribution of C. shasta within the Pacific Northwest has been expanding since its discovery in 1948 (Hoffmaster et al. 1988, Bartholomew et al. 1989). Whether this increase reflects a true spread of the disease or improved detection methods remains unclear (Bartholomew et al. 1989). However, it appears that the abundance of C. shasta, at least in the Columbia River Basin, really is increasing (Ratliff 1983), and may be spreading to other areas. Accordingly, the historical presence and abundance of C. shasta within other river basins, such as the Nehalem, is unknown.

Conflicting reports about C. shasta resistance of several coho salmon populations, such as those of the Alsea River (Schafer 1968, Udey et al. 1975, Zinn et al. 1977) and Columbia River (Conrad and Decew 1966, Hemmingsen et al. 1986), have also confused the interpretation of C. shasta resistance. Ceratomyxa shasta-resistance has also been identified in populations which are not thought to have been exposed to the parasite, and not all populations that are expected to be resistant because of exposure are in fact resistant (Zinn et al. 1977). In addition, there is also some concern that commonly used methods of detecting infection in fish are inadequate (Yasutake et al. 1986, Bartholomew et al. 1992), while methods of detecting the parasite in open waters are hampered by parasite concentrations which have high spatial and temporal variability (Sanders et al. 1970, Hoffmaster et al. 1988). These factors make it difficult to interpret resistance to C. shasta in terms of ESU determinations.

Clackamas River Late-Run Coho Salmon

One population that warrants specific discussion because of its complex history is late- run Clackamas River coho salmon. The Clackamas River, a tributary of the Willamette River, was excluded from the petition for lower Columbia River coho salmon considered by NMFS in 1991 (Johnson et al. 1991), but it is within the area under consideration for this status review. Cramer and Cramer (1994) suggested that this population is the last remaining viable wild coho salmon population in the Columbia River Basin. This section briefly reviews information relevant to coho salmon from the Clackamas River; unless noted, the following information comes from Cramer and Cramer (1994).

The Clackamas River historically had runs of coho salmon and other anadromous species. However, the river also has a long history of obstructions to fish passage by dams. Cazadero Dam (1905, River Kilometer (RKm) 47) and River Mill Dam (1911, RKm 38) were the first large dams to completely block river flow. Both dams were equipped with fish passage facilities, which were often blocked for egg taking. In 1917, the fish ladder at Cazadero Dam washed out, and for 22 years, until the fish ladder was finally restored in 1939, coho salmon were unable to access the upper Clackamas River.

Subsequently, the upper river was repopulated by natural immigration and, possibly, unrecorded releases. Because of the relatively low success of hatcheries at producing adult coho salmon at that time (Hopley undated, Lichatowich and Nicholas in press), the immigrants were most likely natural coho salmon from either the Clackamas River below RKm 47, the Willamette River, or elsewhere in the lower Columbia River. In 1958, North Fork Dam was built at RKm 50. This dam was built with an extensive fish passage facility that has allowed enumeration of salmon entering and leaving the upper Clackamas River.

The history of coho salmon production and runs in the Clackamas River is also complex. Delph Creek Hatchery, located on Eagle Creek (Clackamas River) first raised coho salmon between 1945 and 1948, using Stubbe Creek (a Clackamas River mainstem tributary) as the source population. After the closure of Delph Creek Hatchery in 1954, the Eagle Creek Hatchery began operation in 1956. Initially, Eagle Creek Hatchery used two stocks of coho salmon: an early-run stock transplanted from the Sandy and Toutle Rivers, and a late-run stock which was present at the site and was possibly the progeny of Delph Creek Hatchery stocks. Eagle Creek Hatchery managers recognized that the two stocks had different run timings--the earlier Sandy/Toutle stock spawning peak occurred in November, and the later natural stock peak occurred in January--and made attempts to avoid spawning the two stocks together. However, the criteria by which the two groups were differentiated is not known, so it is possible that some mixing of the two stocks occurred. In 1967, production of the late-run stock at the Eagle Creek Hatchery was terminated, confining coho salmon production to the early-run Sandy/Toutle-derived stock.

At present, the distribution of coho salmon passing the North Fork Dam is bimodal (Fig. 24), with a peak of early-run fish passing the dam in September, and a late-run peak passing the dam in January/February. Early and late-run populations also spawn in different areas of the basin, with the earlier fish spawning higher in the basin than later fish. It has been suggested that differences in run timing between the early and late populations, in addition to spatially-segregated spawning areas, have kept the two populations reproductively isolated. However, the early and late runs are much earlier and later, respectively, than they formerly were.

Beginning with the first North Fork Dam counts until about 1980, the distribution of coho salmon over the dam was unimodal, with a single peak in late November/early December, which presumably included both early and late runs (Fig. 24). Cramer and Cramer (1994) argued that intensive fishing pressure during the middle of this peak, targeted on Cowlitz River coho salmon, caused the changes in run timing. According to this hypothesis, this severe harvest pressure selected against the intermediate run timing and forced the two tails of the Clackamas River run to diverge, thus producing the current bimodal distribution of unusually early- and unusually late-running coho salmon.

Since the run timing over the North Fork Dam of early- and late-run Clackamas coho salmon overlapped extensively prior to about 1980, the spawn timings of the two populations may have also overlapped. Spawning areas currently available to late-run coho salmon are thought to be limited by the cold water temperatures they encounter because of their late run timing. Prior to the shift to even later run timings, late-run coho salmon may have been able to use more of the upper basin, potentially overlapping areas used by early-run fish. Although early- and late-run Clackamas coho salmon currently appear to be reproductively isolated spatially and temporally, they may have been less so previously, and this would have allowed mixing of the two populations either naturally in the river or in the Eagle Creek Hatchery, where both populations were maintained between 1956 and 1967.

Cramer and Cramer (1994) suggested that late-run Clackamas coho salmon are distinctive from other lower Columbia River coho salmon because of their late run timing, ocean migration pattern, large adult size, high fecundity, and small egg size. Current timing of late-run Clackamas River coho salmon is extreme, with spawning occurring from February to March. However, the historical timing of late-run Clackamas coho salmon is thought to have occurred in December or January, closer to other native, lower Columbia River coho salmon populations. The ocean distribution of late-run Clackamas coho salmon, as inferred from marine CWT recoveries, includes fewer Washington recoveries than other Oregon-side, lower Columbia River early-run hatchery populations but is otherwise similar to these (Fig. 25), and other Columbia River populations, as discussed earlier (Fig. 19).

Adult late-run Clackamas coho salmon are large compared to other lower Columbia River and west coast coho salmon (Fig. 14, Appendix Table C-5). However, this is not surprising because they reside longer in the ocean than other coho salmon, especially during the late summer and early fall when growth is rapid (Allen 1959). Other late run coho salmon, such as those from the Satsop River (Chehalis River Basin), are also large (WDF 1966).

Based on a limited comparison with two Columbia River hatchery stocks, Cramer and Cramer (1994) suggested that Clackamas River coho salmon have high fecundity and small egg size. However, compared to other west coast coho salmon, late-run Clackamas coho salmon do not have exceptionally high fecundity given their body size (Fig. 26) (Crone and Bond 1976, Beacham 1982, Cramer and Cramer 1994), nor is their egg size, expressed as egg weight, unusually small (Fig. 27) (Fleming and Gross 1989, Cramer and Cramer 1994). If anything, the hatchery populations used by Cramer and Cramer (1994) for comparison purposes were unusual, having large egg size and relatively low fecundity given their size. Although Clackamas River coho salmon do have a late run timing and are large, neither of these traits are unusual within the Columbia River Basin, nor was their fecundity or egg size exceptional. Consequently, we found no characteristics which would clearly distinguish late-run Clackamas River coho salmon from other Columbia River stocks.

Table of Contents