Review of Artificial Production of Anadromous and Resident Fish in the Columbia River Basin

part 2 of 3

III. Scientific Foundation

All salmon management programs are derived from a scientific foundation-a set of assumptions, theories and principles that describe how the salmon ecosystem functions (ISG 1996). Science deals with the biological and ecological criteria that are integrated to form a conceptual foundation around the process. The foundation is a powerful part of any management program. It is used to interpret information, identify problems (impediments to achieving objectives) and select restoration strategies. Unfortunately the conceptual foundation is rarely explicitly stated or evaluated, and as a consequence programs can suffer from errors in concept. When limited scientific inquiry and false assumptions are a part of the process, the program derived from them will have a high likelihood of failure.

The conceptual foundation of the Columbia River hatchery program has never been specified or examined in detail. In this section, we attempt to describe the set of assumptions, upon which, we believe the hatchery program was based. Since it has never been explicitly stated the conceptual foundation described here had to be derived from our review of the program-its apparent objectives; assumptions stated by practitioners and its measures of performance. The conceptual foundation we present is thus qualified as our interpretation of the historical record, and accounts for the period ending in the 1960s; the point at which this assessment (second phase) will begin.

A. The Early Conceptual Foundation of Hatcheries The early hatchery program was consistent with the over arching assumption that salmonid production systems could be simplified, controlled, and made more productive. Hatchery technology not only simplified and controlled production, it circumvented the need for natural ecological processes and freshwater habitat. The program intention was simply to increase catch by protecting the eggs, maximizing the number of fry released, and harvesting the returns from the sea. Given the hypothetical fecundity of 3000 eggs, a spawning pair may successfully produce something in the neighborhood 500 fry to emergence under natural stream conditions. Under the same scenario, artificially spawning and incubating those 3000 eggs would result in about 2500 fry to emergence under the hatchery scenario, or a five fold increase over natural incubation because of the protection against predation, disease, poor incubation conditions and scouring floods. So the rationale of the early practitioners was not an unreasonable expectation of the advantage hatchery fry production could bestow. Moreover, it was a concept that when properly employed has brought substantive results, as demonstrated in an example that will be discussed in the next section (B). The problem in the beginning was one of dimension. Even with a five fold improvement in egg survival, the number of females intercepted was insignificant compared to the number spawning naturally, even when the run was seriously depressed. The primary problem, however, was that fry were distributed to a variety of streams with little or no information about the suitability of habitat or risk for young fish.

It was the natural extension of the concept that if protecting the incubating eggs from such harm would result in a five fold improvement of fry production, and hence the extrapolation to a five fold improvement in adult returns, then why not control the rest of freshwater rearing to reduce losses from predation, disease, starvation, and environmental alterations in the natural stream? Therefore, taking the simple equation one step further, of the 500 wild fry emerging naturally, 45 might be expected to reach the smolt stage and enter marine waters, from which 2 to 5 adults would return. However, extrapolating the hatchery survival advantage to the next life history stage, if the now 2500 fry successfully incubated from 3000 eggs in the hatchery were reared and protected through the succeeding freshwater rearing period, 2000 fingerlings could be produced to the smolt stage, equating to a total hatchery production benefit nearly 44 times greater than natural production of the original 3000 eggs. Rather than 2 to 5 adults returning per pair of natural spawners, given marine survival equal to natural fry, the hatchery benefit would equate to over 100 returning adults from the same pair of spawners. The simple extrapolation of hatchery survival to return success was the presumptive expectation of the hatchery enthusiasts, and the basis for the expansion of the hatchery building program that has spanned a half century to the present distribution of artificial production throughout the Basin (Figure 5).

Experience has demonstrated, however, that successful production of juveniles in hatcheries is not so simple and that hatchery production by itself can not guarantee a sustained increase in catch, or even an increase in catch for that matter. However, the point in laboring the expectation that ushered in the development of hatcheries is that the fundamental premise is very similar to the basic assumption inherent in the subsequent development of Pacific salmon hatcheries throughout the Pacific Northwest. That presumptive view has not changed substantially, and production augmentation is presently being executed in at least the Columbia River Fishery Development Program, but with a more conservative expectation of benefit.

Part of the problem is that early salmon managers viewed rivers as agri-ecosystems capable of being simplified, controlled and through cultivation (artificial propagation) brought to higher levels of production (Bottom 1997; Lichatowich et al. 1996). The agricultural approach to management led to an emphasis on single species production objectives that separated the development of fisheries science from the major developments in ecology. Fisheries adopted agricultural objectives and supporting science instead of the holistic approach advocated by early fisheries workers such as Forbes (McIntosh 1985; Bottom 1997). Viewing rivers as farms, led to the belief that individual enterprise alone could overcome any natural limits to production (OSBFC 1890). As late as 1960, the Washington Department of Fisheries still believed that fish farming was closely linked to farming on land and shared the same principles and rewards (WDF 1960).

Figure 5. Columbia River Basin State, Tribal and Federal hatchery locations.

An agricultural model for salmon production was expressed by several early salmon managers. The following is a sample of their statements:

"Professor Baird often said 'one acre of water was worth seven acres of land, if properly cultivated, ' but I am convinced that the Professor erred only in this, that I believe one acre of the waters of any salmon stream in Oregon, if judiciously cultivated under favorable circumstances, and if not paralyzed by ignorant vicious legislation, is worth more as a medium for the product of a food supply than forty acres of the best land in the State."(Hume 1893)
"It has been the habit to cultivate the land and neglect the water?. We have tilled the ground four thousand years; we have just begun to till the water?. Less care and labor are needed to raise fish than to raise other animals, or even to raise vegetables." (Oregon State Board of Fish Commissioners 1890)
"Modern incubation equipment for fish propagation compares with greenhouse methods to increase the survival of plants?. As man makes ready the soil for growing of better crops, so may he improve the water for the growing of fish. The steps to be taken in the harvest of surplus seed, the surplus crops, the preparation of land or water follows the same fundamental requirements." (Washington Department of Fisheries 1960)

Commercial aquaculture, or fish husbandry for commercial markets with other agriculture commodities in the Pacific Northwest, has demonstrated production capabilities even better than the original hatchery practitioners envisioned, because the fish farmers control the entire life cycle from spawning to adult harvest and realize the equivalent of 1800 marketable adult size fish per spawning pair. However, while the application of agricultural principles has been beneficial in some aquacultural enterprises, when applied to anadromous salmonids released to experience over three-quarters of their life in the natural environment, it has generally failed.

In retrospect, when we look back to the era of "farming nature", in light of the major leaps that agriculture has made and continues to make in animal husbandry, the assumption that watersheds could be treated as farms and managed like agricultural enterprises was understandable. This logic led to the belief that natural limits on production could be ignored, and through fish culture, levels of production greatly increased. Initially production from natural populations was assumed to be limited by spawning success, and production of the ocean relatively unlimited. Consequently, the belief that increased survival of fry and fingerlings in the hatchery would translate proportionately to increased adult return is epitomized in the following excerpts.

"It is imperative, therefore, that some means be adopted to counteract the depletions arising from this source (habitat degradation); but the most important reason for the artificial propagation is the fact that the natural method is extremely wasteful, which is not true of the artificial method." (Smith 1919 p. 6)
"In my opinion, if the salmon runs of this state are to be maintained and increased, it is going to be necessary to constantly construct new hatcheries. The much greater effectiveness of hatchery operations, as compared with natural propagation, has in my judgment been so effectively proven as to no longer permit discussions among those who are acquainted with the situation." (WDFG 1921 p. 17)
"There can be no doubt in the mind of anyone who has studied the question, that the future prosperity of our salmon fisheries depend largely upon artificial propagation... I am convinced that not more than 10 percent of the ova spawned in the open streams are hatched, owing principally to spawn-eating fish that prey on them... while from artificial propagation 90 percent are successfully hatched. What more need be said in favor of fish culture?" (Oregon State Fish and Game Protector 1896 p. 33)
"Nature ... produces great quantities of seed that nature does not utilize or need. It looks like a vast store that has been provided for nature, to hold in reserve against the time when the increased population of the earth should need it and the sagacity of man should utilize it. At all events nature has never utilized this reserve, and man finds it already here to meet his wants." (Stone 1884 p. 21)

The assumptions that watersheds could be made more productive through agricultural practices and that natural limits on production could be circumvented were the foundation upon which the hatchery program was constructed. Moreover, hatchery production was assumed to be additive to natural production, with no interaction or impact on natural populations. Given the expected translation of hatchery survival to adult returns, practitioners also assumed that the principle measure of success for a production hatchery should be the numbers of juveniles released. Obviously, there would be an associated expectation that harvest level should also increase, but accounting for catch over many fisheries and jurisdictions was much more difficult and less practical than simply monitoring numbers of juveniles produced.

In summary, the fundamental assumptions governing the development of the Columbia River hatchery program before 1960, and the genesis of the early conceptual foundation of hatchery production, was centered on six general assertions:

  • It was not only possible but also desirable to simplify and control production of anadromous salmonids to increase their abundance.
  • Anadromous salmonids could be effectively managed through the application of agricultural practices and science.
  • Production limitations during freshwater life stages could be circumvented by hatcheries and the capacity of the ocean was relatively unlimited.
  • Artificially propagated fish released to the rivers added to production from natural populations. There were no negative interactions.
  • The probability of success was so high that evaluation of adult returns was not necessary.

B. Basic Derivations in the Hatchery Framework Development of a conceptual foundation applicable to Columbia Basin hatchery programs has to be consistent to what is known about salmonid life history and ecological processes. Any fisheries management effort that does not integrate the management criteria around the inherent life history strategies that have evolved among the specific salmonid species, including stock specific differences, will fail. Pacific salmonids have evolved specific characteristics and population structures in synchrony with their native habitat (Brannon, in press), and ignorance, or disregard, of that synchrony will weigh heavily against any management attempts to sustain or build wild fish populations. In essence, the conceptual foundation must be flexible enough to accommodate derivations in life histories among salmonid species, including those differences within the mixture of stocks representing the species.

The pervasiveness of genetic characteristics in the life history of salmonids, and hence the importance of organismic synchrony with the spatial and temporal environmental variables defining their habitat, cannot be overstated. As part of the freshwater ecological system, all of the salmonids show temporal and spatial specificity within their respective population structure. That timing of adult return and spawning is controlled by the genetic predisposition of the fish. As most apparent in chinook salmon (Figure 6), adult timing has evolved in response to mean incubation temperatures associated with the natal stream, and thus specific to each population across their entire range (Brannon in press). Because of the adaptive advantage endowed through selection, changes in temperature brought about by human perturbation, or natural phenomena such volcanism or fire, will result in asynchrony of temporal specificity and fitness will fall. If such changes exceed the rate of genetic compensation, it would lead to extirpation.

Figure 6. Relationship between mean incubation temperature and adult return time for chinook salmon.

Mean incubation temperatures dictate the temporal pattern in parental spawning because selection has timed emergence to occur in the optimum spring period for subsequent growth and survival. Since temperature controls rate of incubation, to achieve such temporal synchrony, spawning must advance progressively from early to later timing when mean incubation temperatures advance respectively from cool to warmer incubation environments (Figure 7).

Figure 7. Relationship between temperature (oC) and number of days of incubation to alevin yolk absorption.

Temperature is the single most important environmental factor in the adaptive evolution of salmonids, and its importance in life history strategy has been a basic oversight in artificial propagation. Hatchery management characteristically has moved stocks of fish throughout the Basin, or intercepted fish destined upstream for propagation in the lower Columbia. Such practices have major impacts on the ability of the fish to survive. Unless the fish are maintained as a hatchery stock for fishery augmentation, in which case the native traits are displaced by the control that hatcheries exert on temperature, feed, and release date, those fish will not perform well in the natural environment. Fish expected to spawn naturally from such origins are out of synchrony with their new environment and any production will perform poorly compared to their wild counterparts. If mean incubation temperature is different from their native waters by only 1? C, emergence timing will change by four weeks, markedly reducing their ability to compete under the new regime. The width of the population spawning curve is representative of the temporal tolerance around the optimum, and when that curve is very narrow, the tolerance of the population to temporal perturbation is diminished (Brannon in press). The single most critical factor responsible for limited success establishing natural runs with hatchery fish is the incongruity in temporal paragons.

Perhaps the best example of genetic specificity is rheotaxis and orientation of sockeye fry migrating from their stream incubation site to their nursery lake (Brannon 1972). The emerging fry are naive to any experience that would assist in that journey and must totally depend on innate responses for guidance. Millions of small fry are involved in the migratory process, and timeliness under river conditions of limited food resources for that number of fry is essential to their future success. Some must only migrate downstream to reach the nursery lake, but others have to swim upstream, and still others swim down one stream and up another in almost a mechanical rheotactic drive to their goal.

The genetic role in rheotaxis is demonstrated by the example of sockeye populations in the Fraser River in British Columbia (Brannon 1972). Chilko Lake has a major outlet spawning population in Chilko River, and the fry must swim upstream to reach the lake. Stellako River sockeye, on the other hand, spawn above Fraser Lake, and the emerging fry migrate downstream to reach the nursery area. Emerging fry from these populations were tested in an artificial stream and each showed the appropriate rheotactic response necessary to reach their respective lakes (Table 4), even when they were incubated under laboratory conditions over a hundred miles away from their native systems. The strong genetic control of the behavior was demonstrated definitively by the hybrid cross between populations showing an intermediate response to that of the parental populations.

Table 4. Rheotactic response of emerging sockeye fry and hybrid crosses from Chilko and Stellako river incubation areas under laboratory conditions (Brannon 1972).


Percent Response

Hybrid cross

Predetermined or preferred directional orientation was also demonstrated among these populations after they enter the nursery lake system. Sockeye fry follow a migratory pattern that distributes the population throughout the system, presumably to optimize food resources. In a test maze where the fish could select any direction, Chilko fry preferred a SE direction, while Stellako fry showed a NE preference (Figure 8), which corresponded to the initial direction the fry would negotiate in distributing down the axis of their lake. Quinn (1985) reconfirmed the genetic basis of the behavior and demonstrated that the innate orientation would shift corresponding with an artificially induced electromagnetic field when tested under laboratory conditions, indicating that juvenile sockeye used magnetic fields to orient along genetically predetermined pathways.

Figure 8. Directional preference of post-emergent fry from Chilko and Fraser lakes, in British Columbia, when tested in orientation arena in the absence of velocity (Brannon 1972).

The point of the above discussion is to demonstrate that natural populations of salmonids are genetically programmed to optimize survival, and execute temporal and spatial patterns of behavior most favorable to maximum fitness. Disconnecting the organismic and environmental linkages effectively disrupts the synchrony and reduces fitness back to the level of a founding population. Survival success returns to the odds of happenstance, and adaptive evolution must start over again. Typical central hatchery programs that follow such management plans, and repeatedly distribute fish around the watershed to encourage the development of natural runs, are doing a disservice to both the resource and to the hatchery system they represent. These fish will have little contribution value to natural production, and by continually or even intermittently spreading stocks around the system, they keep the fish perpetually biologically incompetent for those environments.

The challenge in developing the conceptual foundation for hatcheries is to re-prioritize production and operation goals to address the biological needs of the stock being propagated. Hatcheries have to eliminate the Johnny Appleseed approach, and concentrate on understanding the organism, life history strategy they espouse, and the habitat limitations of the streams they contribute to.

C. The Conceptual Foundation as an Adaptive Process In the complexity of freshwater life history strategies among anadromous salmonids, chinook are at one end of the extreme and pink salmon are at the other, with coho, steelhead, sockeye, and chum salmon in between, in that order. Stream dwelling species, such as chinook, coho and steelhead, are limited most often by the rearing capacity of their stream. Generally factors associated with spatial and nutritional requirements of stream dwelling salmonids determine the upper limit of population biomass that can be sustained within the stream, and strategies to maximize productivity around those parameters evolve to define the population. Sockeye, chum, and pink salmon use freshwater streams only for spawning, with the juveniles immediately migrating to their nursery environments in lake (sockeye) or marine (chum and pink) waters for rearing. Only the spawning area of the stream generally limits these species, since the productivity of their nursery environment most often exceeds the capacity of the spawning grounds available.

In development of the conceptual foundation of hatchery programs, the process must allow for differences inherent in the fish targeted. Successful applications of the hatchery concept are those cases that do not deviate significantly from the biological repertoire of the fish, and were successful in addressing the limiting factors in the natural life history of the species. The Prince William Sound (PWS) pink salmon hatchery program is a good example (Linley in press). In the early 1970s the commercial fishery on pink salmon was threatened by the low return of fish into the Sound, and hence they believed the relatively small numbers of fry naturally produced were insufficient to rebuild the run. The non-profit hatchery program was started, involving the artificial spawning and incubation of fry for release into PWS. Fry releases were synchronized with the beginning of the spring plankton bloom, which was the biological optimum for rapid growth. Their success was unprecedented (Figure 9). Adult returns improved four fold over the previous ten-year average of 5 million adults, and has reached numbers as high as 45 million returning fish. Percent survival of fry released to achieve those levels of return success ranged from 0.9% to 13.0% (Figure 10) at the Armin F. Koernig hatchery (Linley in press), far exceeding the survival performance of any fingerling or smolt production hatchery on the Columbia. The survival variability was attributed to variations in marine productivity, temperatures, and predation, based on annual monitoring of those conditions in the Sound (Willette 1992). Success in the PWS hatchery program was experienced by working within the life history definition of the species, and has succeeded for 20 generations.

Similar success addressing production restraints from loss of habitat was experienced with sockeye returning to Weaver Creek on the Fraser River (IPSFC and PSC annual reports). Logging had caused high variability in flows, and the loss of redds and low returns were threatening the viability of the run. The Salmon Commission built an artificial spawning channel on the stream in which flow was controlled and much of the silt and fine material prevented from infiltrating the graded spawning substrate. Natural spawners used the channel with egg to fry survival rates averaging well over 60%, or about 10 fold better than survival in the adjacent stream. Adult returns showed a marked improvement, amounting to an average of about 250,000 fish annually (Figure 11).

Figure 9. Annual run size of pink salmon returning to hatchery and natural production streams in Prince William Sound, Alaska.

Figure 10. Percent survival of pink salmon fry released from Armin F. Koernig hatchery in Prince William Sound, Alaska.

Figure 11. Annual run size of sockeye salmon returning to Weaver Creek in British Columbia (IPSFC/PSC Rept).

The Weaver Creek channel (hatchery) concept succeeded because the operation was complementary to the biology of the species, and addressed only that portion of the life history that was limiting the population. In both PWS pink salmon hatchery program and the Weaver Creek sockeye salmon spawning channel, the conceptual foundation was consistent with the species life history and integrated the solution to the production problem effectively. However, these species present a different kind of challenge than that facing the Columbia Basin hatcheries. Sockeye and pink salmon are normally limited by freshwater spawning area, and the hatchery approaches used in both cases addressed that limitation with relatively minimum intrusion in the ecological system. The stream dwelling species (chinook, coho, and steelhead) create a different problem when limited rearing habitat is the primary source of population decline. Hatchery rearing programs have a more difficult task of integrating cultured fish into the natural system because, unlike artificial incubation programs, under present hatchery rearing environments the fish are removed from everything that would resemble or prepare them for the natural stream environment they must compete in once they are released. However, even under these conditions, hatchery programs have shown success in increasing production. The Makah Nation Fish Hatchery is a good example.

In the late 1970s, the Makah Indian Nation sought to increase the production of anadromous salmonids associated with the streams on their reservation. The Sooes River chinook population was being seriously threatened by clear-cut watershed instability, runoff from log yards, and over-fishing by the coastal and Canadian fisheries. Fewer than 100 fish were reaching the spawning grounds on some years. In cooperation with the USFWS, the Makah National Fish Hatchery was built on the Sooes River, entering the Pacific Ocean just south of Cape Flattery. Plans were initiated to introduce chinook from other hatcheries, but the Makahs insisted that only Sooes chinook be propagated, even if the hatchery was not fully utilized in the first few years. They felt Sooes River fall chinook were uniquely adapted to that coastal system, with large eggs and an early migration timing to marine waters. Therefore, the hatchery program was to enhance the Sooes River chinook population, and a breeding plan was followed to maintain the diversity present. Fish excess to hatchery needs were permitted to spawn naturally, and in theory both the hatchery population and the naturally spawning fish commingled as a single population. Age-3 returns from hatchery propagation started in 1984, and by 1988 hatchery contributions were a significant share of the total return (Figure 12). By the late 1990s well over 2000 fish were returning from both the hatchery and the natural production.

Figure 12. Chinook salmon annual return to Sooes River, Washington, from hatchery and natural production.

The Sooes River chinook salmon hatchery program success is attributed in part to the emphasis on the native stock. The selective advantage of the adaptive traits manifest in the physical and behavioral characteristics of the stock were not compromised by introductions of other chinook that would have suffered from incongruity with that coastal system. Also attributing to their success is the proximity to the marine environment. Naturally produced fish have a relatively brief period of freshwater residence, and the hatchery fish can be in brackish water within an hour after release from the hatchery.

These examples of pink, sockeye, and chinook hatchery programs that have had good success in reaching their production objectives demonstrate that the conceptual framework of such measures is critically important to the development of functional enhancement systems. Admittedly, none of the above examples are subject to the severely anomalous conditions facing Columbia River salmon and steelhead. The point in fact, however, is that if Columbia Basin hatcheries are to have success in enhancing natural production and restore some of the runs to self sustaining populations, the conceptual foundation has to be that much more specific to the task. To integrate the hatchery complex into the Columbia Basin ecosystem, and still reach the commercial, tribal, and public fishery objectives, the model has to be rigorously defined and the biology of the component species well understood, to meet the challenge.

IV. Organization and Classification of Artificial Production

We have stated that implicit in the artificial production of salmon, and the fundamental premise behind development of salmon hatcheries in the Basin, was the belief that increases in the number of juvenile salmon produced and released from hatcheries would result in a proportional increase of harvestable adults. Although expectations of artificial production have matured to something more qualified by experience, that basic premise has continued to be a strong impetus behind hatchery substitution for habitat loss and reduced access to historical spawning grounds. New hatcheries are being constructed in anticipation of markedly increased adult returns resulting from such operations. How these new hatchery complexes integrate into the Basin ecosystem, will be defined by how management applies the conceptual framework to meet the objectives they have for the fishery.

The application of the hatchery model in the management of salmon fisheries, and hence the basis on which performance of such hatcheries must be judged, depends entirely on the objectives or strategies being addressed (Table 5). With the possible exception of hatcheries that are used solely to restore specific populations nearing extirpation, all hatcheries are intended to provide fish for harvest. Management strategies fall under two categories of purpose, one to augment natural production for harvest, and the other to mitigate for the loss of harvest as a result of the diminution or elimination of salmon producing habitat, and excluding their access to that habitat. It is instructive, therefore, to define more precisely the nature of augmentation and mitigation in the Columbia Basin because of their application in mandates of Congress to enhance production or compensate for its loss as the river has developed around other societal needs. It is also essential to understand the classification of hatcheries in this document if assessment of past performance and current status is to provide the intended framework on which future management decisions and policies will be based.

Table 5. Organization and classification of artificial production.

Augmentation Mitigation
Fishery Maintenance Recovery Preservation Restoration

A. Harvest Augmentation Early in the development of mid-nineteenth century salmon fisheries, and as commercial harvests of Columbia River chinook salmon were doubling every season, artificial production was given serious consideration as a means to augment the harvest of salmon beyond that which could be sustained by natural production. Freshwater production of young salmon in natural river systems was correctly assumed to be limited by spawning success and habitat, and hatcheries were conceived as a means to overcome such constraints on natural production. The fact that egg-to-fry survival could be increased as much as ten-fold through the process of artificially spawning and incubation in hatcheries was the general motivation behind construction of the first Columbia Basin hatchery in 1876, located on the Clackamus River. The expectation followed that adult returns would materialize from such technological interventions, reminiscent of philosophical deductions from technological advancements in agriculture and animal husbandry. Anadromous salmonid population reduction occurred so extensively in the Columbia that augmentation was used simply to compensate for overfishing, and was never able to be applied in that system for harvests expanded beyond what occurred historically from natural production.

Although attempts to assess hatchery contribution to the harvest did not occur until more recent times, and in spite of divided opinion within the scientific community about hatchery success (Lythe 1948), the belief that artificial production contributed to the fishery has been responsible for development of substantial hatchery effort. There were three fundamental assumptions associated with the use of hatcheries for the purpose of harvest augmentation. (1) The freshwater environment limits natural production, (2) ocean carrying capacity exceeds natural production potential, and (3) hatchery production will not negatively impact natural populations. These assumptions still prevail, and are criteria that need to be carefully assessed in applications of harvest augmentation programs to justify use of such technology for that objective in the Columbia River.

The first and second assumptions have credence, but the lower end of the productivity threshold in the marine environment is a very powerful limiting force on production, regardless of the magnitude of production in freshwater. Augmentation of harvest through hatchery production has been demonstrated most recently with pink salmon in Prince William Sound as seen in Figure 9, and highly correlated with marine conditions (Willette 1992). Several hatchery programs in Alaska demonstrate very positive augmentation success, routinely above 10% survival of fingerling sockeye, and higher than 20% among some groups on fingerling coho (Marianne McNair ADF&G personal communication).

Successful augmentation hatchery programs are not rare in Washington and Oregon either. The old Washington Department of Fisheries was formed to manage marine fisheries in the state specifically for commercial harvest, and augmentation was the objective of Washington State hatcheries. Hood Canal chum salmon hatchery production is a good example (Fuss 1998). The size of the chum salmon run in Hood Canal has been directly related to the level of hatchery fry releases (Figure 13). Similarly, coho

Figure 13. A comparison of Hood Canal chum salmon releases and subsequent run size. (Fuss 1989)

production in Puget Sound shows a strong relationship between hatchery production and return run size. Fuss (1998) points out however, that regardless of hatchery contributions, if the environmental restraints are limiting the carrying capacity, production levels off or declines to whatever the environment will support (Figure 14).

Figure 14. A comparison of hatchery releases of Puget Sound 1+ coho with subsequent run size. (Fuss 1998)

Commercial ocean ranching is another hatchery program that has demonstrated variability in return associated with marine productivity. McNeil (1991) reported the very positive influence of ocean ranching hatchery production on commercial landing of coho in the Oregon Production Index (OPI) area during the 1970s (Figure 15). With the

Figure 15. Five-year running average of the total coho salmon harvest in the Oregon Production Index area. (McNeil 1991)

expansion of ocean ranching production and favorable marine conditions, coho catch in the OPI reached unprecedented high levels in the 1960s to late 1970s, exceeding previous natural production by 60%, with 26% of the entire coho catch in the OPI attributed to hatchery production from one ocean ranching facility. Performance in production success varied along the same pattern as the natural coho success, in a cyclic manner associated with ocean conditions, but representing a hatchery smolt survival rate ranging from 1.3% to over 20%.

In the context of ecosystem management, the second and third assumptions listed above create major problems in attempts to accommodate harvest augmentation objectives. Ecosystem management and harvest augmentation are basically conflicting strategies that must be resolved consistent with the long-range goals for the fishery. The real question is not whether hatcheries are able to successfully produce salmon and steelhead artificially; that has been demonstrated many times. The deciding issue is whether hatchery production can integrate within the ecological framework on which future salmon management is proposed to operate. It follows, therefore, that before resolution can be addressed on the use of augmentation strategy in the Columbia River, careful assessment of harvest augmentation success through application of hatcheries outside the Basin, and the measured ecological impacts, should be undertaken.

B. Mitigation With the development of water resources in the Columbia River, nearly half of the accessible river system is deprived of salmon, and much of the remaining habitat has been significantly compromised for incubation and rearing to some degree. Mitigation for these losses has been through the development of hatcheries, and major hatchery programs now prevail in the Columbia River system, and presently represent a significant and continuing investment. Conceptually, mitigation hatcheries are meant to replace harvest potentially lost as a result of habitat alteration associated with the various projects on the river. These losses, related to dams, water diversions, and habitat degradation, have been justified or made "socially acceptable" (Christie et al. 1987) by the precept that the resulting losses in natural production of salmon would be compensated for via hatchery production. Consequently, with the extensive development of the Columbia River, most of the 93 artificial production facilities (hatcheries, ponds, and release sites) in the river system are presently operated for mitigation purposes.

It is not without concern that these major program developments, like augmentation, have progressed extensively without careful assessment of their effectiveness in meeting their primary objectives. The problem in making such assessments of mitigation hatcheries on the Columbia, however, is their application has been somewhat equivocal, with some taking on a distinct augmentation role to increase harvest, while others have been applied in supplementation to strengthen the numerical base of wild populations. With the decline of naturally reproducing stocks of salmon in the Columbia River, and the contemplated further use of hatcheries to overcome these losses, assessment of their effectiveness, limitations, and application must be made. Mitigation must also be viewed in the broader perspective of its present use in the Basin, including measures to stem the risk of extinction. Classification of mitigation hatcheries, therefore, fall within four different categories associated with degrees of salmon extirpation, including maintenance, recovery, preservation, and restoration.

(1) Maintenance is consistent with the original objective of mitigation as a mechanism to maintain those runs of salmon that would otherwise be reduced or extirpated by river developments resulting from habitat degradation or migratory impasse. For example, with the construction of dams on the river, especially those without fish passage, the risk of partial or total loss of the run was mitigated by replacement with hatchery fish. The objective is maintenance of the pre-existing run of salmon at or near its previous abundance. Maintenance hatcheries may substitute or circumvent the need for natural habitat, characterized by attempts to mitigate development of the hydro-system in the upper Columbia and Snake rivers, or they can supplement the number of naturally spawning salmon affected by development.

With the present emphasis on sustaining natural runs of salmon, supplementation has taken a much greater role in maintenance conservation. Conceptually, supplementation is meant to reinforce populations without loss of the genetic structure. Supplementation, therefore, is employed to enhance the native stocks of salmon and steelhead by increasing their reproductive base through artificial propagation, using only the native gene pool in the process. Maintenance, in its most basic rendition, is to maintain contribution of salmon and steelhead approximate to those levels immediately preceding developments affecting their productivity.

(2) Recovery has become an increasing responsibility of mitigation. Compelled by the decline of salmon and steelhead in the Columbia system, major efforts are being expended on rebuilding runs to levels that are considered sustaining under the stress imposed on these populations in the migratory corridor of the mainstem river, and the condition of their endemic habitat. In the context of mitigation with emphasis on native populations, supplementation is by definition the rebuilding of the native population of anadromous salmonids. Application of artificial propagation in rebuilding populations has been thwarted by the disregard of population genetics and careful breeding programs (Ryman and St?l, 1980; Allendorf and Utter, 1979; Cross and King, 1983), as well as poor conditioning of fish while in the hatchery environment (Swain and Riddell, 1990). Salmonids have evolved in synchrony with their environments, and each population, therefore, has adapted to the specific characteristics of their respective habitat. Spawning time, emergence timing, juvenile distribution, marine orientation and distribution are not random, but occur in specific patterns of time and space for each population (Brannon, 1984). In the technical sense, therefore, enhancement of specific wild salmonids must observe these synchronies between the native stocks and their environments, and this perspective is the central theme of mitigation in recovery.

(3) Preservation is the most extreme of measures in mitigation to retain representation of stocks at risk of extinction, and characteristically has been implemented when numbers have degenerated to such low levels that risks associated with emigration and marine life phases threaten extinction. Preservation is approached along two different avenues. The first is to increase the numerical base in captivity from which to rehabilitate a population through maintenance of captive broodstock. Maximizing reproductive potential under captive breeding over two generations can multiply the numerical base from which reintroductions can take place by several hundred fold, and provide the numerical advantage and genetic predisposition necessary for recovery. Such a preservation approach is meant to be short-term, involving only a limited number of generations. However, when a major cause of the decline persists, such as the problems with the migratory corridor on the Snake and Columbia rivers, such preservation programs may have to continue until conditions favor natural recovery.

The second avenue in preservation is to provide repositories of genetic diversity for future introduction and recovery. Captive brood can be applied in such approaches, but germ plasm repositories are the most feasible, inexpensive, long-term approach. Rather than the "choice of last resort" germ plasm preservation should be included in routine population recover measures. Healthy populations need to be the target for gamete cryopreservation to assure that repositories contain representative genetic diversity, and from which domestication and inbreeding can be avoided in mitigation hatcheries. Both avenues are meant to preserve genetic diversity or to keep stocks from demographic extinction, and assist in recovery when habitat and migratory passage are restored.

(4) Restoration is the re-establishment of a salmon or steelhead run in the place of an extirpated natural population. Understandably, establishing a successfully reproducing run requires sufficient similarity between the introduced fish and the extirpated population to facilitate synchrony with controlling environment phenomena. Matching genetic predispositions to optimize the likelihood of success is key to restoration strategy. Important among the environmental factors are winter stream temperatures and length of the freshwater migratory pathway. These features determine timing and distribution patterns of native stocks. The optimum strategy is to use these features to select candidates for introduction most like those demonstrated by the native phenotype.

Restoration mitigation is a difficult task, and necessarily of greater duration to realize functional re-establishment of a run because of the generation time required for the adaptive evolution or re-creation of the appropriate form. The critical measure of success is not the number of returning fish to the hatchery. Hatchery environments are secure and forgiving of timing asynchronies that can easily be amended by feeding programs that exaggerate size at time of release. Restoration criteria must target only the naturally reproducing segment of the run, and hatchery programming should be altered to accommodate the spawning, incubation and migratory timing patterns evolving among those fish. Differentiation between what is observed among hatchery contributions and returns from natural reproduction is a difficult and long-term process, but restoration cannot be accomplished with anything less. To have successful restoration is to have established a self-perpetuating wild run, free of hatchery dependence.

C. Determinants of Performance In determining the performance of augmentation and mitigation hatcheries, it is apparent that the objective identifies the determinant criteria. Moreover, the criteria is only satisfied in terms of the adult return response, as measured in the harvest fishery or the return destination. Augmentation has the objective of increased harvest, or contribution of returning adults to the fishery. Mitigation has the objectives associated with maintenance, recovery, preservation, or restoration measured as contribution of reproductive adults in the target population. In both augmentation and mitigation hatchery programs, genetic and demographic concerns must be addressed. In the former, if genetic compatibility is not a management concern, then isolation of the returning fish from neighboring native stocks must be at least be assured or the level of straying non-consequential. In the latter, genetic identity and diversity are basic to the objectives sought in each of the mitigation functions. In this particular document, the key assessment criteria are listed below, and apply to both augmentation and mitigation programs.

1) Has the hatchery achieved it objective?
2) Has the hatchery incurred costs to natural production?
3) Are there genetic impacts associated with the hatchery production?
4) Is the benefit greater than the cost?

These criteria are relatively simple and straightforward. However, their resolution has an uncertain complexity because of the overriding influence of marine conditions, the effects of mixed stock fisheries, interaction among runs of fish, and the influences of the dynamic intercourse within ecological communities on the ultimate return success of a run. Therefore, in as much as it is possible, the performance measures involved in the SRT assessment will be qualified based on relative information on annual variations in marine productivity, temperature trends, and associated predator occurrence, distance up the freshwater migratory corridor, and other controlling influences unrelated to the actual hatchery variables involved.

V. Synthesis of Artificial Production Reviews

Points of view on the value and importance of artificial production are not lacking in fisheries science. Hatchery production has been the center of controversy with regard to the long-term benefits to the health of the resource as long as hatcheries have existed on the Pacific Coast. Both the ecological and economic points of view have been debated without resolution because the conclusions usually reflect the preconceived perspective of the reviewers. One side of the issue is dominated by practitioners that base their point of view on the evidence of hatchery returns, but tend to ignore the ecological implications of hatchery fish on endemic stocks or the larger biological community. The other side is dominated by scientists who base their point of view on theory and ecological principles, in spite of societal benefits of a propagated fishery. As general background on the topic, it is informative to examine the reviews on the subject and get a better appreciation of the issues confronting the use of artificial production. It is important to keep in mind, however, that artificial production in these assessments is narrowly defined around the standard production hatchery where tray incubators and concrete raceways provide the artificial incubation and rearing habitat. Other forms of artificial production were not included.

A. Early Hatchery Evaluations While it would appear that use of a major program such as hatchery production to augment and mitigate for loss of legendary fisheries would be evaluated to determine if it is achieving its objectives, that did not occur in the Columbia River hatchery program. Part of the explanation for this failure comes from the ideological rather than scientific roots of the programs (see Historical Overview of Artificial Production). A major shortcoming of ideological driven technology is that it is not allowed to fail. Its success is assured by ignoring the signs of failure so by the time the failure is recognized great damage has usually already occurred (Dyson 1997). This observation clearly describes the Columbia River hatchery program prior to 1960, and to a lesser extent after 1960 as well.

During their first 80 years of operation, claims of success for the hatchery program were based on short-term correlations; evidence that was weak at best, or on no evidence at all. Extravagant and undocumented claims of hatchery effectiveness characterized the early history of the program. For example, in 1883, George Brown Goode of the U.S. Fish Commission told the International Fisheries Exhibition in London, England that the Pacific salmon fisheries in the Sacramento and Columbia rivers were under the complete control of fish culture (Maitland 1884). When Goode made that claim, the only hatchery on the Columbia River had been closed for two years (Cobb 1930). This again illustrates the disconnect between science and hatcheries in its early developmental period.

Perhaps the first serious evaluation of the hatchery program came from Marshall McDonald, who succeeded Spencer Baird. He concluded

". . . we have relied too exclusively upon artificial propagation as a sole and adequate means for maintenance of our fisheries. The artificial impregnation and hatching of fish ova and the planting of fry have been conducted on a stupendous scale. We have been disposed to measure results by quantity rather than quality, to estimate our triumphs by volume rather than potentiality. We have paid too little attention to the necessary conditions to be fulfilled in order to give the largest return for a given expenditure of effort and money." (McDonald, 1894, p.15).

McDonald raised three important concerns regarding the use of hatcheries including:

1) a warning regarding an over dependence on hatchery production as a substitute for stewardship;
2) a criticism of hatchery performance based on the quantity of juveniles released rather than the quality of the adult populations; and
3) a recommendation to evaluate the quality of the receiving waters in watersheds to be stocked with hatchery fish.

To varying degrees all of these concerns are still valid today.

The assertion that scientific evaluations did not exist in the early decades of the hatchery program, has been challenged by state salmon managers pointing specifically to a marking experiment carried out from 1895-1900 (Dehart 1997). In this experiment, 5000 chinook salmon eggs were transferred from the Sacramento River and incubated at the Clackamas Hatchery in the Columbia Basin. The fry were marked by removing the adipose fin and released, and for the next several years cannery men recorded the appearance of these fish in their facilities. Sex and weight were determined for some of the fish. However, to label this experiment scientifically valid, the following would have to be accepted:

1) That 5,000 chinook salmon eggs transferred from the Sacramento River and released as marked fry in the Clackamas River achieved a minimum 10% return as adults just to the canneries.
2) That the majority of adults returned in their third year, a year earlier than average, and they were 5 pounds heavier than the average for the Columbia River-one supposed 3-year old weighed 57 pounds.
3) That the cannery operators reliably identified the marked salmon and accurately recorded their weights. The fish commissioner apparently did not personally inspect the fish that the cannery operators claimed to be marked.

The validity of the experiment is questionable and the results were questioned by at least one contemporary biologist (Gilbert 1913).

Other experiments relied on short-term correlations. The common practice before 1910 was to release juvenile salmon shortly after hatching and before they started to feed. In 1911, hatchery managers held a group of chinook salmon and fed them for several months before release. The catch increased in 1914, the year managers expected the first returns from their experiment. After five successive years of improved catches in the Columbia River, the Oregon Fish and Game Commission announced the success of their experiments:

"...this new method has now passed the experimental stage, and ...the Columbia River as a salmon producer has 'come back.' By following the present system, and adding to the capacity of our hatcheries, thereby increasing the output of young fish, there is no reason to doubt but that the annual pack can in time be built up to greater numbers than ever before known in the history of the industry..." (Oregon Fish and Game Commission 1919).

Subsequent review indicated that the claims of hatchery success were premature and the increased catch was not caused by the new methodology (Johnson 1984) and probably had little to do with artificial propagation. Instead, the increase in harvest from 1914 to 1920 was consistent with the pattern of variation in harvest for the previous 20 years (Figure 3) and probably resulted from favorable environmental conditions. For example, the 1914 chinook salmon run into the Umatilla River, which had no hatchery, also increased dramatically (Van Cleve and Ting 1960), supporting the suggestion that the increase in harvest was a response to natural climatic fluctuations.

In 1914, Willis Rich initiated studies of the life history of chinook salmon which had two practical purposes: 1) to determine the value of hatchery work; and 2) to understand the differences in early life history between spring and fall chinook (Rich 1920). Rich also initiated several marking experiments at hatcheries in the Basin to test the efficiency of hatchery practices and to test the homing ability of chinook salmon (Rich and Holmes 1929). The marking experiments were a major improvement over earlier "evaluations", but they did not come close to the standards of experimental design used in later evaluations.

Based on his observations on the timing of the migration of juvenile chinook salmon, Rich (1920) concluded that the release of sack fry should be terminated. He recommended that fry be held in the hatchery and released during the natural migration. He also recommended that juveniles be allowed to migrate out of the hatchery ponds on their own volition.

Nationally, by the 1920s, biologists were beginning to question the efficacy of fish culture during its first 50 years and as a result hatchery programs came under increasing criticism (Wood 1953). Rich (1922) completed a statistical study of the Columbia River Hatchery Program discussed in the previous section, but that study was never published. The lack of rigorous, scientific evaluation of the hatchery programs for Pacific salmon led Cobb (1930) to conclude that artificial propagation could become a threat to the Pacific salmon fishery. Cobb was not opposed to artificial propagation, but he believed that managers had to put aside their optimism and stop relying on hatcheries alone to increase or maintain the fishery.

By the 1940s, individual hatcheries were fin-clipping juvenile salmon in order to evaluate returns to the hatchery from routine production or to evaluate experimental hatchery practices. Often the experiments had too few recoveries to be conclusive. Wallis (1964) summarizes the results of many of those studies.

Extended rearing in the hatcheries prompted research into the nutritional requirements of juvenile salmon and the prevention and treatment of diseases. Through the 1950s, the development of new feeds, better prevention and treatment of diseases, and improved hatchery practices such as the optimal size and time of release (Hagger and Noble 1976) started to produce tangible results. By the 1960s smolt to adult survival had increased significantly.

In the early 1960s, Congress placed a moratorium on new hatcheries until their effectiveness was evaluated. In response, the National Marine Fisheries Service (NMFS) conducted a series of large scale evaluations of the contribution of chinook and coho salmon from Columbia River hatcheries to various fisheries in the Northeast Pacific. The 1961 through 1964 broods of juvenile fall chinook from 13 hatcheries in the Columbia Basin were fin clipped before release so their contribution to the sport and commercial fisheries could be estimated. Results of the evaluation were positive. The benefit cost ratio for all hatcheries combined for each of the brood years was 1961-3.7:1; 1962-2.0:1; 1963-7.2:1; and 1964-3.8:1. The potential catch per 1,000 fish released was 1961-6.7; 1962-3.1; 1963-10.0; and 1964-6.5. Average survival for all hatcheries combined was 0.7%. Overall, an estimated 14% of the fall chinook salmon caught in the sport and commercial fisheries from southeast Alaska to northern California originated from the Columbia River hatcheries (Wahle and Vreeland, 1978).

The NMFS repeated the fall chinook evaluation with the 1978 to the 1982 broods. Total survival for all four brood years and all facilities was 0.33% or about half the survival of the earlier study, however the benefit-cost ratio was still positive at 5.7:1. The overall contribution to the fishery was 1.9 adults for each 1,000 juveniles released (Vreeland 1989). The NMFS used a similar approach to evaluate the contribution made to the west coast fisheries by the 1965 and 1966 broods of coho salmon. Juvenile coho salmon from 20 hatcheries in the Columbia Basin were marked for the study. Recoveries were monitored from British Columbia to California. Coho salmon from Columbia River hatcheries made up about 16% of the total catch in the sampling area (Wahle et al. 1974). These evaluations were well designed and executed, but they only addressed the first question listed among the four criteria on determents of performance.

B. Recent Review Summaries of Independent Panels Three independent scientific panels recently reviewed the use of hatcheries in Pacific salmon management, including the Northwest Power Planning Council's Independent Scientific Group (ISG), 1996; the National Research Council (NRC), 1996; and the National Fish Hatchery Review Panel (NFHRP), 1994. The three panels were in general agreement on three important points: (1) In spite of some success, hatcheries generally failed to meet their objectives, (2) hatcheries have contributed to the decline of wild salmon, and (3) the region's salmon managers have failed to conduct adequate monitoring and evaluation to determine if the hatchery objectives were achieved. These reviews conclude that over the last century, massive funding for hatcheries not only failed to achieve their objectives, but more importantly the lack of monitoring and evaluation meant that the region passed up the opportunity to learn adaptively about artificial propagation of Pacific salmon (NRC 1996).

The individual reviews are summarized below.

ISG - Return to the River - The ISG concluded that artificial production has been institutionalized in the Columbia River Basin. Today 80% of the salmon and steelhead in the Basin were hatched and reared in hatcheries. From 1981-1991 expenditures on hatcheries accounted for 40% of the budget for salmon restoration. Fifty percent of the increase in salmon production of salmon from the NPPC's program is expected to come from artificial production. The historical assumption by management institutions was that artificial production could compensate for habitat destruction, which led to less emphasis on habitat protection and more emphasis on hatchery construction. More recently hatchery programs have been intended to augment declining natural production due in large part to habitat degradation throughout the Basin and to maintain a supply of salmon for the fishing industry.

In the context of the entire history of the hatchery program and salmon management in the Columbia River Basin, ISG concluded that artificial production has failed to replace or mitigate lost natural production of salmonids due to habitat degradation. Since 1960, total releases from hatcheries have increased substantially but the number of adult salmon entering the river has not increased. Furthermore, hatchery-reared fish have become the dominant portion of the run.

It was determined that artificial production can have adverse effects on wild fish including increased mortality in mixed stock fisheries, genetic interactions that can cause reduced fitness of wild populations and loss of population genetic variability, spread of disease, and increased competition with wild fish. The ISG recommended that hatchery populations should be evaluated for evidence of selection, and changes in fitness or genetic diversity associated with residence in the hatchery environment.

The ISG felt that new roles for artificial production need to be defined. Artificial production should likely have a more limited role than at present. The use and role of artificial production needs to be coordinated with the overall Columbia River Basin restoration goal, as well as with subbasin-specific goals. Hatcheries may need to serve as temporary refuges for endangered or critically depressed stocks until factors limiting their abundance can be corrected. Ideally supplementation should be viewed as a small scale and temporary strategy to boost natural production. New supplementation projects should follow the guidelines developed by the Regional Assessment of Supplementation Program (RASP). Supplementation should be used in conjunction with, but not in place of, habitat restoration and modification of downstream mortality factors. Supplementation should be approached cautiously in an experimental framework that relies on careful design, rigorous evaluation, and incorporates adaptive management.

It was concluded that the role of artificial production in salmon restoration has to be redefined. Hatcheries should have a more limited role in salmon production and restoration, and should be integrated into strategies that focus on habitat restoration, reduction of human-induced mortality, and conservation of existing genetic and life history diversity in natural populations. Hatcheries could have a useful role as temporary refuges for dwindling populations while causes of natural mortality are alleviated or a temporary role in rebuilding depressed populations through supplementation.

A comprehensive evaluation of hatchery programs in the Columbia Rive Basin has never been conducted. The ISG believes an evaluation should be undertaken and should address the following questions: 1) Do salmon and steelhead of hatchery origin contribute to the fisheries and/or escapement and is the economic value of that contribution greater than the cost to produce it? 2) Is the level of contribution consistent with the purpose or objective of the hatchery? For example if a hatchery is intended to replace natural production lost due to habitat degradation, this question asks did the hatchery, in fact, replace the lost production? 3) Do artificial produced fish add to existing natural production or do they replace it, i.e., does the hatchery operation generate a cost to natural production through mixed stock fisheries, domestication, and genetic introgression?

NRC - Upstream - The national debate on the use of hatcheries has gone on for most of this century, but with the serious decline of anadromous salmonids across the nation, and hatcheries being proposed as part of the recovery plan, the NRC launched a review of hatchery performance, and made sweeping determinations on how hatcheries should be employed. They concluded that management of hatcheries has had adverse effects on natural salmon populations. Hatcheries can be useful as part of an integrated, comprehensive approach to restoring sustainable runs of salmon, but by themselves they are not an effective technical solution to the salmon problem. Hatcheries are not a proven technology for achieving sustained increases in adult production. Indeed, their use often has contributed to damage of wild runs. In many areas, there is reason to question whether hatcheries can sustain long-term yield because they can lead to loss of population and genetic diversity. It is unlikely that hatcheries can make up for declines in abundance caused by fishing, habitat loss, etc., over the long term. Hatcheries might be useful as short-term aids to a population in immediate trouble while long-term, sustainable solutions are being developed. Such a new mission for hatcheries - as a temporary aid in rehabilitating natural populations - could be important in reversing past damage from hatcheries as well as from other causes.

The NRC proposed that the intent of hatchery operations should be changed from that of making up for losses of juvenile fish production and for increasing catches of adults. They should be viewed instead as part of a bioregional plan for protecting or rebuilding salmon populations and should be used only when they will not cause harm to natural populations. Hatcheries should be considered an experimental treatment in an integrated, regional rebuilding program and they should be evaluated accordingly. Great care should be taken to minimize their known and potential adverse effects on genetic structure of metapopulations and on the ecological capacities of streams and the ocean. Special care needs to be taken to avoid transplanting hatchery fish to regions in which naturally spawning fish are genetically different. The aim of hatcheries should be to assist recovery and opportunity for genetic expression of wild populations, not to maximize catch in the near term. Only when it is clear that hatchery production does not harm wild fish should the use of hatcheries be considered for augmenting catches. Hatcheries should be audited rigorously. Any hatchery that "mines" broodstock from mixed wild and natural escapements should be a candidate for immediate closure. It is useful for all hatchery fish to be identifiable. Marking hatchery fish externally is particularly important when fishers and managers need to distinguish between hatchery and wild fish.

It was concluded that current hatchery practices do not operate within a coherent strategy based on the genetic structure of salmon populations. A number of hatcheries operate without appropriate genetic guidance from an explicit conservation policy. Consistency and coordination of practices across hatcheries that affect the same or interacting demes and metapopulations is generally lacking. All hatchery programs should adopt a genetic conservation goal of maintaining genetic diversity among and within both hatchery and naturally spawning populations. Hatchery practices that affect straying - genetic interaction between local wild fish and hatchery-produced fish - should be closely examined for consistency with regional efforts.

The NRC recommended that hatcheries should be dismantled, revised, or reprogrammed if they interfere with a comprehensive rehabilitation strategy designed to rebuild natural populations of anadromous salmon sustainability. Hatcheries should be tested for their ability to rehabilitate populations whose natural regenerative potential is constrained severely by both short- and long-term limitations on rehabilitation of freshwater habitats. Hatcheries should be excluded or phased out from regions where the prognosis for freshwater habitat rehabilitation is much higher.

They also recommended that decision-making about uses of hatcheries should occur within the larger context of the region where the watersheds are located and should include a focus on the whole watershed, rather than only on the fish. Coordination should be improved among all hatcheries - release timing, scale of releases, operating practices, and monitoring and evaluation of individual and cumulative hatchery effects, including a coast-wide database and wild fish proportions and numbers. Hatcheries should be part of an experimental treatment within an adaptively managed program in some regions but not in others.

NFHRP-- The Director of the US Fish and Wildlife Service (USFWS) asked the National Fish and Wildlife Foundation to conduct a review and assessment of the USFWS federal fish hatchery program and make recommendations for the future role of the National Fish Hatchery Program in ecosystem management of fisheries resources. The National Fish and Wildlife Foundation (through a contract to the Conservation Fund) convened a panel of 16 fisheries and conservation authorities (NFHRP) to conduct the review.

The Panel felt the National Fish Hatchery Program needed a fundamental redirection of programs, personnel and facilities toward supporting ecosystem management whether it relates to restoring depleted anadromous populations or the recovery of ESA-listed stocks. A well-defined national fisheries program with definite goals, objectives, implementation and evaluation strategies did not exist.

The Panel identified habitat alteration or destruction as the primary causes of decline and noted that resource managers have responded to declines in returning salmon by requesting hatcheries to produce more fish for release, with very little assessment or evaluation work being conducted. The assumption that more fish would solve the problem of decline had very little evaluation to verify the approach.

Mitigation based solely on hatchery production (involving 38 of the 78 USFWS hatcheries) has failed to halt population declines, therefore, as a better alternative, habitat protection and restoration were believed to be the key to survival of native fish stocks.

The Panel concluded its report by proposing a new role for hatcheries and a new approach to resource management in which, hatcheries would serve a support function to managers, producing only those species, stocks, strains, races and numbers that were compatible with ecosystem management plans and specifically identified in those plans. Fisheries management plans should include genetic and ecological assessments of native stocks and strains in any ecosystem subject to new fishery resource projects for restoration or enhancement or for the stocking of newly created waters. This should be followed by careful risk assessment. Restoration of sport fishing in altered or newly created waters should involve the use of propagated fish of the most similar native stock known to inhabit the same type of habitats. Before any hatchery fish are planted, a comprehensive assessment, analysis, and a fisheries management plan should have been completed to address concerns about native stocks. Similarly, in efforts to restore depleted populations or to re-establish new populations, resource managers should avoid stocking any non-native strains or species.

C. Relevance of Past Assessments to the Present Task As general background on the how hatcheries got started on the Pacific Coast, and the Columbia River in particular, the origin and the evaluations in the preceding sections are most worthwhile. It was made obvious that to proceed with artificial production "as usual" is poorly advised, and even the assumptions basic to the hatchery program that have carried over from the early years need to be corrected in light of what is known about specific life history requirements of the different salmonid species that are managed. The most compelling point, however, is the change in the general philosophy on resource management that hatchery programs must now address. The human influence on the environment is so pervasive and domineering that resources no longer can demonstrate the resiliency and forgiveness of abuse that was so common in past exploitation. The ecosystem approach to fisheries management is not so much a new paradigm as it is a necessity for the preservation of the fisheries resources. Fish species and their component populations cannot sustain themselves apart from the habitat they evolved with. Ecosystem management is not a revolutionary approach, it is the exercise in common sense to curb the loss of natural productivity and to maintain the health of fisheries resources for public use under the concept of the "normative ecosystem" (Williams et al. in press).

Regarding the three recent independent reviews of hatcheries by the ISG, NRC, and NFHRP, it is noteworthy that apart from primary agreement among reviews that artificial production had (1) generally failed to meet its objective, (2) imparted adverse effects on natural populations, and (3) failed to evaluate hatchery programs, there was further significant consensus on seven other issues. (4) There was agreement that past programs were based on untested assumptions. (5) They felt there was a need to link supplementation with habitat improvements. (6) A need to include genetic considerations and (7) eliminate stock transfers and introductions of non-native species. A need to (8) develop a new role for artificial production, using (9) more experimental approaches, and (10) using hatcheries as temporary refuges, rather than in long-term production management. These efforts provided insights that need to be accommodated in hatchery management. They were comprehensive enough that retracing that ground by the SRT would only be repetitive and add no further resolution to the problems that were identified. It is important to point out that the reviews were not a referendum against hatcheries, but rather a very creditable assessment of hatchery success in reaching their objectives and how programs should change.

We must also recognize that the practitioners' view was not represented on the three panels, nor was the view of commercial harvesters, or that of the angling public, all of which are pertinent to decision making about hatchery application. University scientists dominated or were well represented on the review panels. The NRC for example was made up of 15 participants, of which 12 were associated with a university. There were no members experienced in hatchery production or aquaculture on the NRC panel. Even the NFHRP panel, charged to assess USFWS hatcheries, did not have equitable representation from hatchery production management. Moreover, the reviews were largely based on ecological theory, biological principles, and some empirical evidence, but little rigorous analysis of actual data was undertaken. This is not a criticism of the process, because it is critical that the understanding and implications of the hatchery production be grounded in the basic science relevant to the subject. This is necessary regardless of how successful hatchery programs are or can become. To adequately manage the resource on a sustained basis, there can be no compromise with the requirements of biological processes. Whether society decides that other priorities supersede the need to maintain a specific population or a habitat, is another issue, but if fisheries management is serious about building naturally sustained production, the science must be the basis of any approach.