Chum salmon in different runs may swim at different speeds during their return migrations from oceanic feeding grounds (Salo 1991). However, an insufficient number of studies have been conducted to determine run-specific travel times. For example, Lyamin (1949) found that chum salmon tagged at different locations migrated between 300 km in 15 days (20 km/day) and 1,200 km in 15 days (80 km/day), whereas Shmidt (1947, cited in Lyamin 1949) reported maximal speeds of 43-63 km/day. Fish tagged in British Columbia traveled at a rate of 14 km/day from Johnstone Strait to the mouth of Fraser River (Anderson and Beacham 1983).
An additional complication in calculating stock-specific timing of freshwater entry is "milling" (Hunter 1959, Koski 1975): All species of anadromous salmonids may delay their entry into freshwater or into terminal spawning areas as they approach the mouths of their natal rivers at the end of the marine phase of their life cycle. For example, some tagged chum salmon in Skagit Bay remained in the estuary for up to 21 days (Eames et al. 1981). Also, as noted above, the average time between tagging at the mouth of the Stillaguamish River and recovery of marked carcasses on the spawning grounds was 31 days (Hiss et al. 1982a, b). Some of these tagged fish spent more than 6 weeks in freshwater before spawning (Hiss et al. 1982a, b).
Chum salmon are particularly vulnerable to fisheries and natural predation during this period of milling. For example, Evenson and Calambokidis (1993) found that the number of harbor seals at Dosewallips State Park in Hood Canal, Washington, was highest when adult chum salmon were present. Fisheries aimed at other species or more plentiful hatchery stocks of chum salmon (as described by Tynan (1992) and Cook-Tabor (1995) for Hood Canal) can also incidentally harvest fish milling in Hood Canal. Adult chum salmon concentrate in such large numbers in estuaries and off the mouths of small streams that their dorsal fins break the water's surface. The cause of milling is unclear, although maturation may play a role, but often fish will move into a river only after a period of rain, when water flow increases. For whatever reason, the period of milling becomes shorter as the spawning season progresses (Salo 1991).
In the past, observations of chum salmon behavior have suggested to some that the species may have a greater tendency to stray than other species of Oncorhynchus (reviewed in Lister et al. 1981). There are a number of reasons why this perception could have developed. 1) O. keta spawn near the mouths of streams, and their young do not conduct the long, downstream, freshwater migrations that are common in some salmonid species. It has been hypothesized that juvenile salmonids who do conduct long, freshwater migrations may sequentially imprint on a chain of migratory cues that assist them as adults in returning to their natal streams (Lister et al. 1981). 2) Observations of the reluctance of adult chum salmon to surmount small falls or rapids have suggested to some that they may go upstream as far as they can toward natal areas, but once they reach a barrier, they spawn. 3) Adult chum salmon also are more sexually mature when they enter freshwater than most species of anadromous salmonids and may not be able to endure delays in reaching their natal areas; if delayed, they may be forced to spawn at the first available location. 4) It has been observed (McNeil 1969, Lister et al. 1981) that, when spawning densities of chum salmon become high in some rivers (especially those with hatchery runs), straying to nearby streams may increase.
Only a few experimental studies have directly addressed this issue (Lister et al. 1981; Quinn 1984, 1993; Salo 1991; Altukhov and Salmenkova 1994; Tallman and Healy 1994), and these studies have concluded that under normal circumstances, straying in chum salmon is no greater than in any other species of Oncorhynchus. Lister et al. (1981) reviewed experimental studies on straying in all species of Pacific salmon, but found only a few unpublished reports that used chum salmon as the study species (Table 8). In a review of the life history of chum salmon, Salo (1991) found that the only information relevant to straying in chum salmon was collected incidentally from studies designed for other purposes. Genetic studies of straying for all species of Pacific salmon are limited (reviewed in Quinn 1984, 1993, 1997), and only one experimental study of chum salmon straying has been published (Smoker and Thrower 1995).
Mark-release and recapture studies--The first reported experimental mark-release study to document straying in chum salmon was conducted in Japan by Sakano (1960, cited in Okazaki 1982a). In this study, over 2 million native, fin-clipped chum salmon fry were released into the Tokoro River on northern Hokkaido Island and into the Chitose River on the southern part of the island. Both the Tokoro and Chitose Rivers are large rivers that empty directly into the sea, and "straying" in this study is defined as the portion of fish that were captured in a river system other than their native Tokoro or Chitose Rivers. The straying rate was estimated to be only 2% for fish released into the Tokoro River, with most of the fish recovered in their natal stream. However, some strays were recovered as far as 350 km away in the Teshio River on western Hokkaido Island. Fish released into the Chitose River strayed at a rate as high as 10%, with marked fish recovered as far as 2,000 km away on Honshu Island. However, this rate of straying among fish released into the Teshio River has perhaps been underestimated, due to an inadequate geographic sampling range; most of the rivers monitored were on Hokkaido Island, while many strays were incidentally recovered far to the south on Honshu Island (Okazaki 1982a).
Salo (1991) collated information on straying and homing in chum salmon from studies designed to gather other kinds of information. For example, Salo and Noble (1952a, b; 1953) marked juvenile (dorsal fin marks) and adult chum salmon (fin clips) at the mouth of Minter Creek, Washington for 2 years and found no marked adults in nearby streams. In another study cited by Salo (1991), Wolcott (1978) released fry from the Quilcene National Fish Hatchery into Walcott Slough, near Brinnon on Hood Canal, and found no strays in nearby streams. The authors concluded that adults returned "unerringly" to traps on the slough, even though a natural run did not exist there. In his own studies, Salo (1991) observed that adult chum salmon returned to a weir trap set at the outlet of a small stream from which the fish had emigrated as fry and not to a nearby trap set on the mainstream. These results were taken by Salo (1991) as support for strong homing behavior in chum salmon.
In their review of straying in Pacific salmon, Lister et al. (1981) identified one unpublished mark-and-release study on chum salmon in Alaska and two studies from the Fraser River in Canada. These studies were summarized in a common format and published as appendices by Lister et al. (1981). In the Alaskan study, Freitag and Ward (reported in Lister et al. 1981) (Table 8) designed a study to evaluate imprinting in off-site rearing facilities: They reared chum salmon to the fry stage in two hatcheries, then released some into their natal stream and others from a hatchery. Straying was defined as occurring when fish were captured at sites other than where they had been released as juveniles. None of the fish released into their natal stream were found in nearby streams. However, these results were inconclusive, because only 17 of 100,000 marked fish returned over the 2 years of the study. An 8.8% straying rate, based on returns of 56 fish (0.12% of the total release), was reported for the fish released from the hatchery.
The two studies identified by Lister et al. (1981) on straying in chum salmon in the lower Fraser River did not involve transfers of fish from one area to another. Straying was also defined in these studies as occurring when fish were captured in streams other than where they had been released as juveniles. In the first study, fish from Inches-Barnes Creek, a tributary to the Fraser River, were spawned and eggs reared in a hatchery on the creek (Foye, reported in Lister et al. 1981). Juveniles were fin-clipped as fed fry and released into the creek. The returning adults strayed at a rate of 7.2% to streams within 2 km of the natal stream. In another study, adults from Blaney Creek, a tributary to the North Alouette River on the lower Fraser River, were spawned and their eggs incubated at a facility on-site (Harding, reported in Lister et al. 1981). Juveniles were fin-clipped and released into Blaney Creek (252,900 were released in 1974 and 190,033 in 1975). A large difference in adult straying was found between the 2 years, with 45.7% straying in 1977 but only 9.3% in 1978 (Harding 1981, cited in Lister et al. 1981). Harding hypothesized that the high straying rate in 1977 resulted from a large escapement that saturated available spawning sites in Blaney Creek, forcing many fish to seek other areas to spawn. Blaney Creek had a spawning density of 1.7 females/m2 in 1977, but only 0.2 females/m2 in 1978 (Banford and Bailey 1979). While spawner densities vary greatly among spawning areas, the density in 1977 was more than twice the optimal density for chum salmon suggested by McNeil (1969).
Lister et al. (1981) reviewed almost 400 studies on straying in Pacific salmon, including the ones above on chum salmon, and found a small but measurable straying rate in almost every study. They concluded that the average straying rate of adults released as smolts from the hatchery where they had been reared was 2.6% across species. However, straying rates varied widely under different conditions (such as site of release or life-history stage of fish at release). Lister et al. (1981:39) concluded that "relatively large adult returns (of chum salmon) to sites with hatchery facilities could result in increased straying to nearby streams. This pattern was evident from chum salmon marking studies conducted to evaluate hatchery operations."
In Washington, Eames et al. (1981, 1983) tagged adult chum salmon in 1976 and 1977 in northern Puget Sound near Port Gardner, Bellingham Bay, and Skagit Bay. However, the primary purpose of these studies was to estimate run-size, not straying. In this study, "strays" were defined as "tagged fish which moved to other river systems or which were taken in the various [out of basin or nonreported] fisheries" (Eames et al. 1981:147). However, the fish tagged in this study were caught in the estuary, and it is unlikely that all the fish tagged were native to the drainage. With these caveats, the researchers estimated straying rates between 6.9% and 63.2% for the fish in their study.
Also in Washington, observations by WDFW biologists (Fuss and Hopley 1991), during a study of five consecutive broods of coded-wire tagged chum salmon released from the Hood Canal Hatchery, revealed very few stray fish in streams located in Hood Canal.
Other observations have shown that chum salmon periodically appeared in some southern Oregon and California streams, with few if any known chum salmon spawning at locations nearby (Moyle et al. in press). It therefore appears that chum salmon either stray thousands of kilometers from the nearest spawning areas in some years, or they maintain spawning populations in California and southern Oregon in areas that have been poorly surveyed.
Genetic studies on straying--Bams (1976) published a study on pink salmon that suggested impaired homing in transplanted fish and in hybrids between populations. This hypothesis was tested with chum salmon by Smoker and Thrower (1995). For two years they crossed fish from two populations of chum salmon, tagged the offspring with coded microwire tags, and released about 20,000 age-0 fry per year from control and treatment groups up to 65 km from their natal streams (Table 8). Recoveries were monitored for 3 years per brood year. None of the 611 tagged fish that were recovered strayed from the native (non-hybridized) group, and only one stray was recovered from the hybridized group. Although the number of fish released was small, the results support the hypothesis that straying rates for native and transplanted hatchery chum salmon are not necessarily great.
However, even small straying rates from large hatchery releases may result in a high proportion of hatchery fish in small, natural populations. This can substantially affect natural populations, particularly if the stray fish are of non-native origin. (Jacobs 1988, Thorpe 1994). Many small populations of summer-run chum salmon in southeast Alaska have escapements of less than 5,000 adults, and hatchery programs in some areas (see "Artificial Propagation") have released more than 100 million fry per year, producing returns approaching 1 million fish per year. In recent years, otolith marking has been used on many of these hatchery releases, with extensive monitoring of local wild populations for hatchery-marked fish. While no results have yet been published, early returns indicate that only a small proportion of the total return of hatchery fish is straying into local streams, although strays may represent up to 50% of the fish in some of these streams (Thrower19).
In freshwater, a variety of population-specific reproductive behaviors have been described for chum salmon in Asia (e.g., Sano and Nagasawa 1958) and North America (e.g., Johnson et al. 1971; Tautz and Groot 1975; Duker 1977, 1982; Helle 1981; and Schroder 1973, 1982) that may act to isolate populations. For example, Duker (1982) described a model for the pre-spawning phase of courtship that involved mate recognition based on a variety of auditory, tactile, and visual clues (e.g., species-specific body coloration, the black and white pigmentation pattern inside the mouth) (Schroder 1981, Duker 1982). However, studies to determine whether these behaviors or characteristics act to reproductively isolate particular populations or runs of chum salmon from one another have not yet been made.
There is extensive literature on selection of spawning sites and redd characteristics for chum salmon (reviewed in Bakkala 1970, Smirnov 1975, Salo 1991), which indicates that under specific circumstances chum salmon spawn in a wide variety of locations. In general, chum salmon are reported to spawn in shallower, slower-running streams and side channels more frequently than do other salmonid species, perhaps to avoid competition with pink salmon (Bakkala 1971, Smirnov 1975, Salo 1991). In Asia, there are also extensive differences reported between seasonal run types, with summer chum salmon reported to spawn in deeper waters and higher velocities than fall chum salmon (Sano and Nagasawa 1958, Soin 1954, Smirnov 1975, Salo 1991), even though Smirnov (1975:50) reported that the "autumn chum is larger than the summer one and its redds are also larger."
Smirnov (1975) suggests that the differences in physical parameters between the two run times of chum salmon in the Russian Federation may be caused by interactions with pink salmon. Fall chum salmon migrate farther inland than pink salmon, but summer chum and pink salmon spawn in similar areas. Both species spawn within about 100 km of seawater, although in most years, the spawning grounds of the two are widely separated: Summer chum salmon spawn in the lower and middle reaches of rivers, whereas pink salmon usually migrate into the upper reaches. But even in years when both pink and summer chum salmon are abundant, and their spawning grounds are close together in the middle reaches, the two species maintain separation by choosing different locations for their redds. Smirnov reported that in these circumstances, summer chum, unlike pink salmon, spawn in deep, lower-velocity pools, away from riffles and closer to river banks.
The velocity of water in spawning areas has been a widely studied area of research. In the Amur River Basin, water velocities of 10-80 cm/sec were measured over summer chum salmon spawning sites, and velocities of 10-30 cm/sec in riffles over fall chum salmon spawning grounds. However, fall chum salmon also spawned in pools in this region where the velocity was reported to be quite insignificant (Soin 1954, Smirnov 1975). On Hokkaido Island, Sano and Nagasawa (1958) also found that fall chum salmon selected spawning areas with lower water velocities (10-20 cm/sec) than did summer chum salmon in the Amur River area. These differences in the physical characteristics of spawning areas may act to isolate populations or runs in the same river (Salo 1991).
In Washington, Johnson et al. (1971) measured water velocities near 1,000 chum salmon redds and found that velocities where fish spawned varied from 0.0 to 167.6 cm/sec and that over 80% of the fish spawned in velocities between 21.3 and 83.8 cm/sec. This range is similar to that found in other species of salmon. For example, velocities of streams where chinook salmon spawn are reported to range from 10 to 150 cm/sec. Johnson et al. (1971) also attempted to correlate abundance indices of chum salmon in Washington with environmental variables such as stream discharge, velocity, and surface water temperatures, but found no relationship between run size and these variables. He concluded that he was unable to measure or to isolate the critical areas in which environmental factors influence run size.
Subgravel flow (upwelled groundwater) may also be important in the choice of redd sites by chum salmon. Salo (1991:240) reported that "chum salmon prefer to spawn immediately above turbulent areas or where there was upwelling." Sano (1966:46), in a summary of available information on Far Eastern chum salmon, reported that throughout the Russian Federation and on Hokkaido Island in Japan, autumn chum salmon "utilize mostly spring areas of upper tributaries, [as] damage by freezing and other severe winter conditions is relatively minor in most years." However, Sano also notes (p. 46), based on studies by Smirnov in the 1940s, that "summer chum salmon spawn earlier in the season, and they do not particularly choose spring areas."
Smirnov (1975) noted that "the summer spawning chum from Kamchatka gravitates towards the places of emergence of ground water" (p. 50) and that in the redds "of the summer Amur and Sakhalin chum the eggs are mainly flushed by the so-called subterranean water, replenished by the infiltrating streams water" (p. 49). Smirnov further noted that the summer chum salmon in the Amur River area began to spawn from August to September during the warmest time of the year when water temperatures fluctuated from 9.8 to 13.6°C. He reported that incubation remained above 3°C through October and then dropped to zero.
Smirnov (1975:50) also noted that in many areas of the Russian Federation and on the Islands of Hokkaido and Honshu in Japan, fall chum salmon reproduced in localities supplied with groundwater even when temperatures did not go below freezing: "In limnocrenes, or spring-fed spawning creeks on the basin of the Amur, the summer temperatures do not exceed 11-12°C; in the winter they fluctuate within the limit 2.5-5°C (on Hokkaido spawning grounds, sometimes higher)."
Biologists at WDFW reported that chum salmon in Washington do not preferentially choose areas of upwelling groundwater for redd construction; rather they suggest that chum salmon in Washington "most commonly" use "areas at the head of riffles" (Crawford 1997:4). As reported in Turner (1995) and repeated in Crawford (1997:4):
We [WDFW] are unaware of any evidence that Washington chum salmon specifically select spawning sites with upwelling ground water. . .Washington chum salmon would not seem to need this particular adaptation. Upwelling ground water would be an advantage for summer chum . . .but most of the streams involved do not seem to match the definition of 'streams with cool, upwelling ground water.' The summer chum streams of the Strait of Juan de Fuca and the Kitsap Peninsula are characterized by low summer/fall flows and likely experience elevated stream temperatures during the summer chum spawning period. In fact, a lack of ground water influence may pose a particular problem for summer chum during periods of summer drought as has occurred in western Washington in recent years. A more likely reason that summer chum spawn where they do is the low flow condition of spawning streams at the time of return, confining these fish to the lower reaches of the streams.
Fecundity and egg size of chum salmon have been extensively reported in the literature (reviewed by Bakkala 1970 and Salo 1991); however, in most cases, comparative regional or run-type information by age, size, or relative survival rates are lacking. Salo (1991:244) considered fecundity data unreliable for comparison among regions and among runs, because it was "not certain how representative the samples [were] for the reported geographical regions and rivers of origin."
Nevertheless, some latitudinal and run-type trends were evident for absolute fecundities (number of eggs/female) and relative fecundities (number of eggs/cm of length) (Salo 1991). One pronounced trend was that the ranges of absolute fecundity for both individual and annual means were higher and larger among Asian chum salmon runs than among North American runs. For example, individual fecundities from numerous studies, summarized by Salo (1991), varied from about 900 to 8,000 eggs per female in Asian chum salmon, but only from 2,000 to 4,000 eggs per female in North American chum salmon. The annual mean of these fecundities ranged from about 1,800 to 4,000 eggs per female in Asian chum salmon and from 2,000 to 3,600 eggs per female in North American chum salmon. Differences also existed among northern and southern populations in the two regions. Rivers in northern Asia had generally higher relative fecundities than rivers in southern Asia. However, in North America the opposite was true: Fall-run fish from southern rivers tended to have higher relative fecundities than fall-run fish from northern rivers. The different regional trends are difficult to interpret because the various studies were not always comparable. However, Salo (1991) suggested that differences may be related to decreasing survival rates from south to north in Asia, and from north to south in North America.
Fecundity differences between run times--Differences in both relative and absolute fecundities have been extensively documented in fish with different run times (summer- and fall-run chum salmon) in the Amur River (Lovetskaya 1948; Birman 1951, 1956; Svetovidova 1961; Sano 1966; Kulikova 1972), and to a lesser extent in the Yukon River (Andersen 1983, Trasky 1974) and in Hood Canal (Koski 1975). Summer-run chum salmon generally spawn within 100 km of the mouths of both the Amur and Yukon Rivers, whereas some stocks of fall-run fish historically migrated hundreds of kilometers upriver to spawn. Interestingly, summer-run fish in the Amur River have higher fecundities on average than do fall-run fish spawning in the lower river, although in the Yukon River the opposite trend appears. However, the difference between runs in the Yukon River is not large, and few data are available to compare fecundities between the two run times (45.5 eggs/cm, N=23 for summer-run fish; and 41.2 eggs/cm, N=24 for fall-run fish).
Fecundity and egg size for summer- and fall-run chum salmon in Hood Canal were measured in Big Beef Creek by Koski (1975) in 1967-69. He found that summer- or early-run females were smaller for a given age than in late-run fish (males were of similar size), but early-run fish had slightly larger fecundities per body length and per weight. Early-run fish had on average 50 eggs per cm of body length compared to 46 eggs/cm in fall-run fish. Early-run fish also averaged about 526 g less than later-returning fish, but had about 100 more eggs. Koski also found that early-run fish had larger eggs than late-returning fish of the same body size.
The rates of chum salmon embryonic and juvenile development tend to decline at high latitudes in both Asia and North America, but vary among populations within an area, apparently because of adaptation to local environmental conditions (e.g., summer-, fall-, and winter-run chum salmon in southern Puget Sound or Hood Canal) (Bakkala 1970, Salo 1991). One of the earliest detectable differences between chum salmon in different areas is the time of hatching of eggs and the emergence of alevins from gravel. Differences between areas are caused by physical factors such as stream flow, water temperature, dissolved oxygen, and gravel composition, and by such biotic factors as genetics, spawning time, and spawning density, all of which can affect survival (reviewed in Bakkala 1970, Salo 1991).
The rate of embryonic development in chum salmon is influenced most by water temperature (reviewed in Bakkala 1970, Koski 1975, Salo 1991). The amount of heat, measured in thermal units (TUs),20 required by fertilized chum salmon eggs to develop and hatch is about 400-600 TUs, and the heat required to complete yolk absorption is about 700-1,000 TUs. Lower water temperatures can prolong the time required from fertilization to hatching by 1.5-4.5 months. For example, fertilized eggs hatch in about 100-150 days (400-600 TUs) at 4°C, but hatch in only 26-40 days at 15°C.
The time to hatching also varies among populations and among individuals within a population (Salo 1991). Koski (1975) found differences in the time to hatching between early- and late-returning chum salmon at Big Beef Creek, a tributary to Hood Canal. For 2 years (1968-1969 and 1969-1970), early-returning (peak September) and late-returning (peak late November or December) fish spawned and their offspring were reared in spawning channels in the creek. Fry emerged from February to June, but the timing of fry emergence differed between early- and late-returning fish by an average of 35 days each year. Early-run fish took longer to hatch, and this difference between the two runs was consistent from year to year. However, the longer hatching time of early-returning spawners led to fry with lower average weight and less lipid content than fry of late-returning spawners. Lower weight and fewer food reserves in early-return fry may decrease their chances of survival during early life history. The difference in incubation times for eggs from these early- and late-returning fish suggested a genetic difference between the two runs, and Koski (1975) concluded that natural selection apparently acted on hatching times: Fry tended to emerge when they had their best chances of surviving in streams and estuaries.
Changes in hatching times due to adaptation to cold water have also been found for chum salmon in the Susitna River, Alaska (Wangaard and Burger 1983) and in the Amur River in Asia (Disler 1954, cited in Bakkala 1970). In these populations, low incubation temperatures resulted in faster embryonic development than for embryos in other populations at the same temperature. In Canada, however, Beacham and Murray (1986) failed to find differences in hatching times among eggs from adults with early, middle, and late spawning times that had been incubated at constant temperatures of 4, 8, and 12°C. Nevertheless, the time of emergence in that study depended on the timing of spawning: Earlier-spawning fish laid larger eggs that took longer to develop than did smaller eggs from later-spawning fish.
Other factors, such as dissolved oxygen, gravel size, salinity, nutritional condition, and even the behavior of alevins in the gravel, can also influence the time to hatching, emergence from the gravel, or both (reviewed in Bakkala 1970, Schroder et al. 1974, Schroder 1977, Salo 1991). For example, Fast and Stober (1984) found that developing chum salmon embryos in small coastal streams required less oxygen than had been reported for either coho salmon (O. kisutch) or steelhead (O. mykiss), but it is unknown to what extent chum salmon in different areas vary in their oxygen requirements. The relative importance of various factors influencing early development in different populations has not been evaluated.
However, despite a large amount of variability in incubation environments, even over short distances, chum salmon display a variety of developmental responses that result in similar emergence and outmigration times among fry within an area. Variability in some of these responses appears to reflect differences among individual fish, but it also reflects differences among populations in adult run and spawning times, egg size, and temperature-development requirements.
Observations of chum salmon fry are often more difficult to make than are observations of juveniles of other salmonids because chum salmon outmigrants 1) are smaller than outmigrants of other salmonids, 2) migrate at night, 3) usually have shorter distances to migrate to reach saltwater than do other species, and 4) do not school as tightly as some other salmonids (e.g., pink and sockeye fry) (Salo and Bayliff 1958, Beall 1972, Koski 1975, Seiler et al. 1981, and reviewed in Salo 1991). Moreover, some chum salmon fry outmigrate in conditions less conducive to scientific observation. For example, observation of outmigrating chum salmon fry in northern Russian rivers draining into the Arctic Ocean is obscured by ice on the rivers at that time of year (Sano 1966, 1967).
Nonetheless, several key facets of fry outmigration are known (Table 9). Downstream migration may take only a few hours or days in rivers where spawning sites are close to the mouth of the river, or it may take several months, as in the Yukon and Amur Rivers, where spawning sites are located hundreds of kilometers upriver. The timing of outmigration is usually associated with increasing day length, warming of estuarine waters, and high densities of plankton (Walters et al. 1978). Juvenile chum salmon at southern localities, such as those in Washington and southern British Columbia, migrate downstream earlier (late January through May) than do fry in northern British Columbia and southeastern Alaska (April to June) (Table 9).
In the Yukon and Noatak Rivers in northern Alaska, chum salmon fry migrate downstream beginning in late May, with the breakup of river ice. Outmigration continues until fall, with peak movement in June and July (Martin et al. 1986) (Table 9). However, several exceptions to this general pattern can be found. For example, Koerner (1993) reported that in Fish Creek, a tributary of the Salmon River near Hyder, Alaska, chum salmon fry outmigrated over an extended period from late February through May. Chum salmon in this creek also spawned over an extended period from mid-June through October, which may have contributed to the long period of emergence from the gravel and outmigration (Helle, footnote 14).
Several cues influence the timing of downstream migration, resulting in considerable variability in migration timing. These cues include time of adult spawning, stream temperatures during egg incubation and after hatching, fry size and nutritional condition, population density, food availability, stream discharge volume and turbidity, physiological changes in the fry, tidal cycles, and day length (Simenstad et al. 1982, Salo 1991). In the Russian Federation, Soldatov (1912, cited in Smirnov 1975) found that chum salmon outmigrations did not always immediately follow emergence; juveniles in many rivers remained up to 3-4 months in the river and grew to a considerable size before outmigration (Kostarev 1970, as cited in Salo 1991). In Washington, chum may reside in freshwater for as long as a month (Salo and Noble 1953, Bostick 1955, Beall 1972). Juvenile residence times in freshwater longer than a month have also been reported in the mainstems of the Skagit (Dames and Moore 1976) and Nooksack (Tyler 1964) Rivers.