Previous reviews of the status of chum salmon populations have been conducted (Nehlsen et al. 1991, WDF et al. 1993, Nickelson et al. 1992, Kostow 1995). These reviews used a variety of methods and criteria for evaluating the status of salmon stocks. Nehlsen et al. (1991) considered the status of populations coastwide and evaluated their risk of extinction. They reported only the status of populations they considered to be at risk of extinction, categorizing populations as possibly extinct, at high risk of extinction, at moderate risk of extinction, or of special concern.
Nehlsen et al. (1991) considered populations "at high risk of extinction" to have likely reached the threshold for classification as endangered under the ESA. Stocks were placed in this category if they had declined from historical levels, were continuing to decline, or had spawning escapements of less than 200. Populations were classified as "at moderate risk of extinction" if they had declined from historic levels but presently appear to be stable at a level above 200 spawners. Nehlsen et al. (1991) felt that populations in this category had reached the threshold for classification as threatened under the ESA. Populations were classified as "of special concern" if a relatively minor disturbance could threaten them, insufficient data were available for them, they were influenced by large releases of hatchery fish, or they possessed some unique character. Nehlsen et al. (1991) also included a partial list of populations that they believed to be extinct. The other reviews are limited to individual states and are thus more limited in area, but are intended as inventories of populations and are more complete in terms of coverage within the areas they cover.
WDF et al. (1993) classified stocks by origin (native, non-native, mixed, or unknown), production (wild, composite, or unknown), and status (healthy, depressed, critical, or unknown). However, the stock status designations in SASSI (WDF et al. 1993) are not considered a "risk assessment" by these authors because SASSI is a survey that did not address threats to the future of the populations listed (Ames footnote 15). Stock status was classified as healthy if recent production was consistent with current habitat conditions. However, WDF et al. (1993) combined hatchery production with natural production if a hatchery was located on a stream that supported natural spawning, and, because the report considered only recent production status, it did not consider possible negative impacts of hatchery production on natural populations. WDF et al. (1993) recognized 72 chum salmon stocks in Washington. Of these, 48 were rated healthy, 3 were rated depressed, 2 were critical, 18 were of unknown status, and 1 (Chambers Creek summer-run chum salmon) was considered extinct (Table 18). Only one stock is classified as extinct because the authors considered only recent and not historical extinctions. Half of the stocks of unknown status were from the Strait of Juan de Fuca and the west side of the Olympic Peninsula. The classification of stock status by WDF et al. (1993) differed substantially from that of Nehlsen et al. (1991) mostly because of differences in the intent of the two classifications.
Nehlsen et al. (1991) considered the Duwamish/Green River fall chum salmon to be at high risk of extinction because of habitat loss and degradation. WDF et al. (1993) divided chum salmon in this watershed into a possible remnant native stock (Duwamish/Green River) whose status was unknown and into a hatchery run (Crisp or Keta Creek) introduced from releases of Quilcene and Hood Canal hatchery stocks. The Crisp Creek status was listed as healthy, although all population-assessment data available are from hatchery rack counts and hatchery planting records. As reported in the South Puget Sound SASSI Appendix (WDF et al. 1993, Appendix 1--South Puget Sound, p. 191):
Currently, most of the fish are seen between Burns and Crisp Creeks. Natural spawning does occur in this reach, but many of these fish may be hatchery fish headed for the Keta Creek facility. In fact, there is some doubt that any native fish exist . . . Efforts are underway to determine if any of the native fish remain.
Nehlsen et al. (1991) considered the Elwha and Ozette Rivers' fall chum salmon runs to be possibly extinct, whereas WDF et al. (1993) listed the status of these runs as unknown. Nehlsen et al. (1991) listed Nisqually and Walla Walla Rivers' fall chum salmon as extinct and the Washougal River fall chum salmon as possibly extinct; these stocks were not mentioned by WDF et al. (1993), because SASSI addressed only recent extinctions (Ames footnote 15). Nehlsen et al. (1991) considered Hood Canal and Chambers Creek summer chum salmon to be at moderate risk, whereas WDF et al. (1993) considered Chambers Creek summer chum salmon as extinct and Hood Canal summer chum salmon as critical. Finally, Nehlsen et al. (1991) considered lower Columbia River chum salmon to be at moderate risk; WDF et al. (1993) divided these fish into three separate stocks, two of which were classified as depressed and one as healthy.
Within Oregon, Nehlsen et al. (1991) (Table 19) listed 11 chum salmon populations, in addition to the Lower Columbia River chum salmon population, which also extended into Washington (Table 18). They considered the Umatilla River population to be extinct, and populations in Tillamook Bay, Netarts Bay, and Nestucca Bay at moderate risk of extinction. They also identified seven stocks from Siletz Bay south which they considered to be at high risk of extinction.
Chilcote et al. (1992) listed an inventory of chum salmon runs in Oregon and evaluated them under the Oregon Wild Fish Policy (Chilcote et al. 1992) (Table 19). This policy has two compliance criteria: a hatchery criterion that requires naturally spawning populations to have no more than 10% strays from a genetically-dissimilar hatchery stock or 50% strays from a genetically-similar hatchery stock, and a numerical criterion that requires at least 300 average spawners. Kostow (1995) is a revision of Chilcote et al. (1992), with newer information on stock presence or absence. Chilcote et al. (1992) considered the percentages of hatchery strays and their genetic constitution in all chum salmon runs in Oregon in compliance with the hatchery criteria. Of 50 populations of chum salmon identified in Oregon, they considered 4 to be in compliance with the numerical criterion and 4 out of compliance. The remaining 42 populations were of unknown status.
Nickelson et al. (1992) evaluated the status of coastal populations of chum salmon in Oregon. They classified populations as healthy if available spawning habitat was fully seeded and abundance trends were stable or increasing over the last 20 years. They classified populations as "of special concern" if they were believed to be composed of fewer than 300 spawners, or had a naturally spawning population that consisted of more than 50% strays from a genetically-similar hatchery stock or 10% strays from a dissimilar hatchery stock. Populations were classified as depressed if available spawning habitat was not fully seeded, abundance trends were declining over the last 20 years, or abundance trends in recent years were below the 20-year average. Nickelson et al. (1992) classified the status of 26 stocks; of these, 10 were considered healthy, 12 "of special concern," and 4 of unknown status due to insufficient data.
The 1994 biennial report on wild fish status in Oregon (Kostow 1995) considered chum salmon populations in the Columbia River to be very depressed to extinct. Kostow (1995) also described the Nehalem River as having a population of several hundred adults, and the Necanicum River population as very small, unstable, and vulnerable. Chum salmon populations in Tillamook Bay, Netarts Bay, and the Nestucca River were described as the most substantial populations in Oregon, with Tillamook Bay having estimated escapements of 10,500 adults in 1992 and 7,500 adults in 1993. Chum salmon populations south of the Nestucca River were described as very depressed or extinct, with remnant populations in the Salmon, Alsea, Yaquina, Siletz, and Coos Bay River systems, and scattered adults occasionally seen in other basins (Kostow 1995).
Nehlsen et al. (1991) listed chum salmon runs in the Sacramento River and Klamath River as extinct.
Quantitative evaluations of data included comparisons of current and historical abundance of chum salmon, and calculation of recent trends in escapement. Historical abundance information for these ESUs is largely anecdotal. Time-series data were available for many populations, but the amount and quality of the data varied among ESUs. We compiled and analyzed this information to provide several summary statistics of the abundance of naturally spawning populations, including (where available) recent total spawning run size and escapement, percent annual change in total escapement, and recent naturally-produced spawning run size and escapement.
Although this evaluation used the best data available, they have several limitations, and not all summary statistics were available for all populations. For example, spawner abundance varies by state and region, and in some areas, (particularly where chum salmon are not presently abundant) abundance was generally not measured directly. Instead, abundance was often estimated from catch (which itself may not always have been measured accurately) or from limited survey data. In many cases, limited data were also used to separate hatchery production from natural production.
Information on stock abundance was compiled from records in a variety of state, federal, and tribal agencies. We believe this information to be complete in terms of long-term adult abundance records for chum salmon in the regions included in this review. Principal data sources were run reconstruction and fishery statistics from commercial, tribal, and recreational fisheries, and escapement estimates from stream surveys of spawning escapement. However, although the above provide the "best" estimates of chum salmon production, actual run-size may vary from these estimates. Specific problems are discussed below for each data type.
Information on the abundance of natural chum salmon populations in Puget Sound includes fishery landings data and spawning escapement surveys. While chum salmon fisheries occur in several Puget Sound rivers, most chum salmon are harvested in saltwater, as fish return to different spawning areas. The relative run size in terminal areas and genetic mixed-stock analysis (MSA) indicate that various stocks are included in these mixed-stock fisheries (Graves 1989).
In the northern portion of Puget Sound, chum salmon escapement is estimated for river basins by scaling base-year mark-recapture estimates (Eames et al. 1981) by the ratio of total spawners in index areas to the index values in the base years. Estimates for individual stocks at a finer level are usually not available. In general, index counts are available, but the base-year data necessary to scale them at the stock level are lacking for some areas (Hendrick32). Data for southern Puget Sound are primarily from spawning ground surveys (where all, or nearly all, of the spawning habitat is surveyed) and from hatchery rack counts. Prior to 1976, spawning ground surveys were used primarily to define peak counts of live fish, and since then most estimates have been based on area under-the-curve (AUC) calculations derived from weekly spawning ground surveys. WDFW used the AUC method, after it was introduced in 1976, to estimate peak abundances for some locations in southern Puget Sound and Hood Canal with data reaching back to 1968 (Uehara33).
Hood Canal summer- and fall-run chum salmon estimates of escapement from 1968 to the present are based on the AUC methodology. Survey frequency was low prior to the mid-1970s, and peak live spawner abundance and the full extent of spawning were often not reflected in the curves. This means that early calculations underestimate abundances.
Chum salmon escapement has not been monitored by tribal biologists on the west side of the Olympic Peninsula, and the only stock status data available were from landings. Chum salmon are generally not targeted in western Olympic Peninsula or tribal fisheries, and since the fisheries tend to occur near river mouths, it is not known to what extent the fish in the landings in a river originate from that river. In Grays Harbor, index counts of peak spawners were expanded to estimate total escapement, and in Willapa Bay, WDFW estimated escapement using AUC methodology from spawning ground surveys of adults (Brix34).
According to Washington fisheries co-managers (WDF et al. 1993), Columbia River chum salmon populations on the Washington side of the river were limited to Grays River, Hardy Creek, and Hamilton Creek. However, biologists from WDFW have also observed chum salmon in the Lewis, Cowlitz, and Kalama Rivers (Ames footnote 15). Fish spawn in limited areas of these streams, and spawning grounds have been surveyed annually since 1976. These surveys cover most of the spawning habitat in the streams surveyed, and counts at Bonneville Dam monitor chum salmon ascending the Columbia River beyond these streams. WDF et al. (1993) estimated that recent spawning escapement of chum salmon in the Columbia River was in the range of "a few thousand up to ten thousand."
In Oregon, chum salmon spawning escapement has been monitored by ODFW using peak counts of live and dead fish in standard surveys in the Tillamook District since 1948 at Moss Creek, Clear Creek (Kilchis River), and the Little North Fork of the Wilson River. An additional survey site (Nestucca River) was added in 1950. The intention of ODFW was to monitor trends in escapement, and they have not attempted to quantify total escapement (Cooney and Jacobs 1994).
From 1957 through 1972, an impassible culvert was present near the mouth of Moss Creek, and from 1960 through 1982 one was present on Clear Creek. Also in 1960, logging and road slides clouded the water and hampered viewing in the Wilson River. In an attempt to improve the census of spawning fish, ODFW surveyed 11 more areas, and in 1976 reclassified 4 of these supplemental surveys as standard surveys. The degree to which the surveys tracked spawner abundance is questionable, because the amount of habitat has not been well monitored and there have been unrecorded changes in quality and quantity of the survey areas. Nevertheless, Nickelson et al. (1992) used the relationship between the survey data and commercial landings to estimate total escapement to Tillamook Bay. In 1992, ODFW added 21 supplemental surveys in addition to its 8 standard surveys (Cooney and Jacobs 1994). Supplemental surveys were located from the North Coast District as far south as the Yaquina River. These supplemental surveys have been monitored since then (Klumph and Braun 1995). In addition, a trap constructed by Oregon State University on Whiskey Creek in Netarts Bay has been in operation since 1969 (Nickelson et al. 1992).
In California, chum salmon spawning escapement has not been monitored.
In Washington, run size in Puget Sound was calculated by run reconstruction. Interceptions by Canadian fisheries were considered relatively minor and were not incorporated in the run reconstructions. Landings within Puget Sound were attributed to production units on the basis of geographic locations of the fisheries and spawning grounds, and tagging studies. Run reconstruction was also used to calculate run size in Grays Harbor and Willapa Bay. In other areas, run-size estimates were not available. Run size can be estimated by summing terminal catch and spawning escapement, because ocean interceptions are assumed to be negligible. However, in coastal rivers on the Olympic Peninsula, the Columbia River Basin (exclusive of the Washington side), and coastal Oregon streams, spawning escapement has not been estimated.
Fishery landings were estimated more accurately than spawning escapement. However, landings contain less information on the abundance of individual stocks. While they are informative in monitoring the overall abundance of chum salmon in broad geographic regions within Puget Sound, the bulk of commercial fishing occurs on mixed stocks, and the current data have not as yet been resolved into component stocks. For coastal rivers on the west side of the Olympic Peninsula, these are the only data available and more accurately reflect landings from individual rivers.
To represent current run size or escapement where recent data were available, we computed the geometric mean of the most recent 5 years reported (or fewer years if the data series was shorter than 5 years). We used only estimates that reflected the total abundance for an entire river basin or tributary, avoiding index counts or dam counts that represent only a small portion of available habitat. Where adequate data were available, trends in total escapement (or run size if escapement data were not available) were calculated for all data sets with more than 5 years of data, based on total escapement or an escapement index (such as fish per mile from a stream survey).
As an indication of overall trends in chum salmon populations in individual streams, we calculated average (over the available data series) percentage annual change in adult spawner indices within each river basin. Trends were calculated as the slope (a) of the regression of ln(abundance) against years corresponding to the biological model N(t) = beat. Slopes significantly different from 0 (P<0.05) were noted. The regressions provided direct estimates of mean instantaneous rates of population change (a); these values were subsequently converted to percentage annual change, calculated as 100(ea - 1). No attempt was made to account for the influence of hatchery-produced fish on these estimates, so the estimated trends included any supplementation effect of hatchery fish. Trend analysis can also be influenced by climate regime shifts and other factors.
The Puget Sound/Strait of Georgia ESU of chum salmon encompasses much diversity in
life history, and includes summer, fall, and winter runs of chum salmon. WDF et al. (1993)
identified 38 stocks with sufficient data to calculate trends in escapement within the area
encompassed by this ESU: 10 had negative trends and 23 had positive trends (Table 20). All of
the statistically significant trends (P< 0.05) were positive and the slopes of many negative trends
were close to zero. One stock, the Chambers Creek population of summer chum salmon, met the
WDF et al. (1993) definition of extinct, a stock currently being tracked that is believed to have
been extirpated in its original range. The sum of the recent 5-year geometric means of these
escapement trends, which are not exhaustive, indicate a recent average escapement of more than
300,000 natural spawners for the Puget Sound/Strait of Georgia ESU as a whole.
Commercial harvest of chum salmon has been increasing since the early 1970s throughout the state (Fig. 24) and the majority of this harvest has been from the Puget Sound/Strait of Georgia ESU. The recent average chum salmon harvest from Puget Sound (1988-1992) was 1.185 million fish (WDFW 1995). This suggests a total abundance of about 1.5 million adult chum salmon. This increasing harvest, coupled with generally increasing trends in spawning escapement, provides compelling evidence that chum salmon are abundant and have been increasing in abundance in recent years within this ESU.
While most populations in this ESU appear to be healthy and increasing in abundance, there appears to be a potential for loss of genetic diversity within this ESU, especially in populations that display the most distinctive life histories. For example, four summer-run stocks from southern Puget Sound were identified by WDF et al. (1993). Of these four, one was classified as extinct, two were of mixed production, and all were relatively small. Of the three extant stocks, Blackjack Creek has a 5-year geometric mean spawning escapement of 524, Case Inlet has 4,570, and Hammersley Inlet has 7,728, with about 40,000 total summer chum salmon spawners in southern Puget Sound estimated in 1994. The latter two stocks had hatchery supplementation programs that were believed to be major contributors to the runs until they were discontinued in 1992 (WDF et al. 1993). The last brood year produced by these hatchery programs (1991 brood year) returned as adults at age-4 in 1995 and age-5 in 1996. While all three summer-run populations are apparently stable or increasing, they represent a small fraction of the ESU. The winter-run life history is represented by only two stocks. The Chambers Creek stock is increasing in abundance, and the Nisqually River stock is a relatively large run with a 5-year geometric mean escapement of more than 16,000 spawners. Both stocks are classified as wild production.
Analysis of biological information for the Hood Canal summer-run chum salmon ESU is more extensive than that for other ESUs. This extended analysis reflects the deliberations of the BRT in considering the dynamic changes in summer-run chum salmon abundance that have occurred in this ESU over the past several years.
Although summer chum salmon in this ESU have experienced a continuing decline over the past 30 years, escapement in 1995-96 increased dramatically in some streams (Fig. 25). Spawning escapement of summer chum salmon in Hood Canal (excluding the Union River) numbered over 40,000 fish in 1968, but was reduced to only 173 fish in 1989 (WDF et al. 1993). In 1991, only 7 of 12 streams that historically contained spawning runs of summer chum salmon still had escapements (Cook-Tabor 1994, WDFW 1996). Then in 1995, escapement increased to more than 21,000 fish in northern Hood Canal, the largest return in more than 20 years (WDFW 1996). These increases in escapement were observed primarily in rivers on the west side of Hood Canal (Fig. 26), with the largest increase in the Big Quilcene River where the USFWS has been conducting an enhancement program starting with the 1992 brood year (Tables 21 and 22). Streams on the east side of Hood Canal continued either to have no returning adults (Big Beef Creek, Anderson Creek, and the Dewatto River) or no increases in escapement (Tahuya and Union Rivers) (Fig. 26).
Summer runs of chum salmon in the Strait of Juan de Fuca (Snow and Salmon Creeks in Discovery Bay and Jimmycomelately Creek in Sequim Bay) are also part of this ESU. While these populations have not demonstrated the marked declining trend that has characterized the summer-run populations in Hood Canal in recent years, they are at very low population levels (Fig. 27). Further, though escapement of summer-run chum salmon to Salmon Creek increased in 1996, the other two populations in the Strait of Juan de Fuca did not show similar increases (Fig. 27), and the overall trend in the Strait populations was one of continued decline. WDF et al. (1993) considered the Discovery Bay population to be critical and the Sequim Bay population to be depressed (Table 18).
Several factors may have contributed to the dramatic increase in abundance seen in some Hood Canal streams in 1995 and 1996. These include hatchery supplementation, reduction in harvest rate, increase in marine survival, and improvements in freshwater habitat. Information relevant to these factors is discussed below.
USFWS began hatchery production of summer chum salmon in Quilcene Bay with the 1992 brood year (Table 21). In brood year 1992, 27.95% of the fry released from Quilcene Hatchery had their adipose fins removed, and most of these were tagged with micro- (half-size) coded wire tags (CWTs); in subsequent years, all releases have been unmarked. Marked fish returned as age-3 spawners in 1995 and as age-4 spawners in 1996. Examination of ratios of marked and unmarked fish provides some indication of the role of hatchery production in the population increases seen in some areas in 1995 and 1996.
Table 22 shows the number of adults surveyed in several areas in 1995 and 1996 and the number of those fish that were fin-clipped. These data were provided by WDFW and USFWS. Based on the proportion of marked and unmarked fish in the fishery in Quilcene Bay, returns to the hatchery, and returns to Big Quilcene River (considering only age-3 fish in 1995-96 and only age-4 fish in 1996), USFWS estimated that the survival of marked fish was only 21.4% that of unmarked fish in 1995 and 39.9% of unmarked fish in 1996. USFWS assumed in making these estimates that all fish sampled in Quilcene Bay and the Big Quilcene River originated from hatcheries. A major weakness with the USFWS method is that there was apparently no independent way to confirm this assumption. This method overestimates tagging mortality if unmarked fish of non-hatchery origin are included in the samples. While this error almost certainly occurred, we do not have a quantitative estimate of its magnitude. Bailey (1995) estimated the survival rate of adipose-clipped and micro-coded wire-tagged chum salmon fry from the Nitinat Hatchery released into Barclay Sound on the west coast of Vancouver Island was approximately 50% that of unmarked chum salmon. Given these assumptions, the estimated hatchery contributions to escapement to the Big Quilcene River were 32% of the 3-year-olds in 1995 and 63% of the 4-year-olds in 1996.
Table 22 presents 95% confidence limits on the largest number of marked fish that could have been present in each stream given the estimated spawning escapement, the number of carcasses examined, and the number of marks detected in 1996. The tabulated maximal number of marks represents the maximal number of marked fish that could have been present in the spawning escapement with less than 0.05 probability of observing no marked fish in the carcass surveys. These limits were calculated from a hypergeometric distribution, with the assumption that the carcasses in the surveys were drawn randomly from a population the size of the estimated escapement without replacement.
Table 22 also shows an estimate of the maximal number of hatchery fish that could have been present in those streams in 1996. These estimates were obtained by expanding the maximal number of marked fish by the effective mark rate. The latter value was obtained by adjusting the actual mark rate (28%) for relative marking survival (0.34 in 1996); the resulting estimate of effective marking rate is 0.11. The total estimate was obtained by pooling all samples, and placed an upper bound on the total number of possible strays assuming that all streams were equally likely to receive strays. The numbers of potential strays from the USFWS enhancement program was clearly too small to account for the observed increases in escapement to the Dosewallips, Hamma Hamma, and Duckabush Rivers.
Collectively, these results suggest the following conclusions about the effects of hatchery supplementation on adult returns in 1995 and 1996.
WDFW divides Hood Canal into five regions for fishery management and assessment (Fig. 3). Spawning populations of summer chum salmon are restricted to four of these regions: Area 12A, Dabob Bay and Quilcene Bays, includes marine waters to which Big Quilcene and Little Quilcene Rivers are tributaries; Area 12B, Central Hood Canal, includes the marine waters adjacent to other east and westside rivers such as Anderson Creek and the Dosewallips, Hamma Hamma, and Duckabush Rivers; Area 12C, South Hood Canal, includes the population in marine waters adjacent to Lilliwaup Creek; and Area 12D, Southeast Hood Canal, includes the marine waters adjacent to Union River. Population responses in these four areas have been quite different, with increases in returning spawners apparent only in Areas 12A and 12B (Fig. 26) and Area 12D (Union) being stable.
Historically, summer chum salmon have not been a primary fishery target in Hood Canal, since harvests have focused on chinook, coho, and fall chum salmon. Summer chum salmon have a run timing that overlaps those of chinook and coho salmon, and they have been incidentally harvested in fisheries directed at those species (Tynan 1992). Prior to 1974, Hood Canal was designated a commercial salmon fishing preserve, with the only net fisheries in Hood Canal occurring on the Skokomish Reservation (WDF et al. 1973). In 1974, commercial fisheries were opened in Hood Canal and incidental harvest rates on summer chum salmon began to increase rapidly. By the late 1970s, incidental harvest rates had increased to 50-80% in most of Hood Canal and exceeded 90% in Area 12A during the 1980s (Fig 25B). In 1991, coho salmon fishing in the main part of Hood Canal was closed to protect depressed natural coho salmon runs. Commercial fisheries, targeting hatchery-produced coho salmon, continued in Quilcene Bay. Beginning in 1992, fishing practices in this fishery, including changes in gear, seasons, and fishing locations, were modified to protect summer chum salmon (WDFW 1996). Since then, the tribal and nontribal harvests of coho salmon during the summer chum migration have been by beach seine with the requirement that summer chum salmon be released or surrendered to the USFWS for broodstock in the interagency enhancement program at Quilcene National Fish Hatchery.
Exploitation rates on summer-run chum salmon in Hood Canal have been greatly reduced since 1991 as a result of closures of the coho salmon fishery and efforts to reduce the harvest of summer chum salmon (WDFW 1996). Between 1991 and 1996, harvests removed an average of 2.5% of the summer-run chum salmon returning to Hood Canal, compared with an average of 71% in the period from 1980 to 1989.
These harvest rates, and the reconstructed run sizes on which they are based, are imprecise and are probably overestimated in recent years, when summer-run chum salmon abundance has been depressed. Much of the imprecision and bias stems from the practice of using a fixed cutoff date to attribute chum salmon in fishery landings to the summer or fall run (Lampsakis footnote 19). Because a fixed cutoff date is used to assign chum salmon to the summer or fall run, and the runs overlap, some fish will always be misclassified. With year-to-year change in run timing, the fraction of fish misclassified can vary considerably. Even if the fraction of each misclassified run remains constant, as abundance changes, the number of fish misclassified changes. Because the summer run has declined relative to the fall run, the number of fall-run chum salmon misclassified as summer run in recent years has probably been greater than the number of summer-run chum salmon misclassified as fall run. Consequently, the cutoff date used to distinguish summer run from fall run is presently being reevaluated by state and tribal biologists. This may have led to an overestimation of harvest rates calculated with WDFW adult accounting periods. Tribal biologists advocate using an earlier cutoff date to delineate the summer run from the fall run to compensate for the change in relative abundance of the two runs in Hood Canal. In spite of these caveats, the reconstructed run sizes and harvest rates based on them are the best data available to evaluate the effects of harvest and changes in population productivity.
While changes in harvest rates coincide to some degree with changes in run size, there are some important differences. In general, run sizes declined in the 1970s and 1980s as harvest rates increased to relatively high levels, rebounding in 1995 and 1996 after harvest was largely curtailed (Fig. 28). Reductions in harvests alone are insufficient to account for the population rebounds in Dabob Bay and Central Hood Canal, and no populations in south Hood Canal and Southeast Hood Canal have rebounded.
Summer-run chum salmon are still harvested incidentally in British Columbia in pink and sockeye salmon fisheries in the Strait of Juan de Fuca (Area 20) and Johnstone and Georgia Straits (LeClair 1995, 1996; PSMFC data 1997, Tynan 1995, 1997). Summer-run chum salmon are also taken in troll fisheries off the west coast of Vancouver Island (PSMFC data 1995). Net and troll fisheries in these areas target Fraser River sockeye and coho salmon but incidentally harvest chum salmon. Bycatch of chum salmon in Canadian Area 20 in the period from 1968 to 1995 has estimated 2,803 fish (Tynan 1995, 1997). These harvests have traditionally been allocated between U.S. and British Columbia populations using the proportions determined from genetic mixed-stock analysis (MSA) estimates in samples of fall chum salmon caught in later fisheries that were directed at chum salmon (PSC Joint Chum Technical Committee 1996).
Recently, fishery managers have begun to suspect that Hood Canal and Strait of Juan de Fuca summer-run chum salmon may be the majority of chum salmon migrating through Area 20 in August and early September when Area 20 fisheries for sockeye and pink salmon occur (WDFW 1996). Genetic MSA was used to estimate the proportion of Hood Canal summer chum salmon in the Area 20 catch (LeClair 1995, 1996). Estimates indicated that Hood Canal and Strait of Juan de Fuca summer-run chum salmon accounted for 31% of the Area 20 catch in 1995 and 68% of the catch in 1996 (WDFW 1996). This corresponded to estimated harvest rates on Hood Canal fish of ~3% in 1995 and ~1.5% in 1996, and on Strait of Juan de Fuca fish of ~17% in 1995 and ~2% in 1996.
Changes in survival, whether in the marine or freshwater life-history phases, would be reflected in changes in the number of spawners or potential spawners produced per parent (cohort replacement rate). The predominant age at which individual broods of Hood Canal summer chum salmon mature is variable, ranging from 92% age-3 to 83% age-4 for the 1968 to 1985 brood years (Tynan 1992). However, the mean age of maturation during this period was 3.6 years, so we have used spawning escapement 4 years earlier as the measure of parental abundance for a given year's run. This erroneously attributes some production to adjacent year's escapement, but differences in year to year escapement are relatively small compared to the total contrast in escapement levels.
Using the WDFW run reconstruction database (Big Eagle et al. 1995), updates from WDFW (Haynes35), and the escapement numbers for 1996, it is possible to reconstruct the size of terminal runs and the exploitation rate (Fig. 28) of Hood Canal summer chum salmon through 1996. Reconstructed terminal runs represent the best available data on stock productivity. Although the scale of run reconstruction is coarse and there are potential errors and biases in attributing catch to runs in management sub-areas (discussed in preceding sections on harvest), changes in harvest rates have been so large that these biases are relatively small.
A plot of spawners-per-spawner 4 years earlier (Fig. 29) suggests that substantial increases in the cohort replacement rates did not occur until the 1990s. However, if we consider the number of recruits-per-spawner (Fig. 30), which accounts for the effects of incidental harvest, it appears that there were increases in productivity in the early 1980s. These increases coincided with a climatic regime shift in the north Pacific Ocean during the late 1970s that has generally been viewed as beneficial to chum salmon (Francis and Mantua In press). However, apparent productivity did not increase uniformly. The largest increases in productivity were apparent only for the populations in Dabob Bay, with populations in the central portion of Hood Canal showing little increase in productivity until after 1990. The timing and magnitude of apparent productivity changes, and the inconsistencies between different areas within Hood Canal, suggest the unlikelihood that changes in marine productivity contributed substantially either to the decline in Hood Canal in the late 1970s or to the recent rebound in some of the populations.
Record numbers of chum salmon returned to many areas of Alaska and the Pacific Northwest in 1995 and 1996. In this ESU, some streams have also had large returns of summer chum salmon, while other streams have not. Possible explanations are that 1) any changes in marine survival for this ESU are being overwhelmed either by differences in survival during freshwater or by estuarine life-history phases, or 2) summer chum salmon in Hood Canal have ocean migration patterns different than other chum salmon, or 3) other factors such as susceptibility to diseases.