Natural Climate Insurance for Pacific Northwest Salmon and Salmon...

American Fisheries Society Symposium 43:xx–xx, 2004© 2004 by the American Fisheries Society

121

Natural Climate Insurance for PacificNorthwest Salmon and Salmon Fisheries:

Finding Our Way through the Entangled Bank

NATHAN MANTUA1

University of WashingtonJoint Institute for the Study of the Atmosphere and Oceans/

School of Marine Affairs Climate Impacts Group, Box 354235Seattle, Washington 98195-4235, USA

ROBERT C. FRANCIS2

University of Washington, School of Aquatic and Fisheries Sciences,Box 355020, Seattle, Washington 8195-5020, USA

Abstract.—This essay focuses on the linkages between climate (variability and change) andsustainable salmon management policies. We show the importance of climate in its effects onsalmon production as well as how unpredictable these effects are. Our assessment leads us toconclude that the treatment of environmental uncertainty poses a fundamental conflict betweenthe kind of policies that have been traditionally used in fishery management—basicallycommand and control policies that assume predictability and assert engineering solutions to justone or a few aspects of highly complicated problems—and what the environmental variabil-ity dictates—policies that embrace environmental variability and uncertainty and acknowl-edge a lack of predictability for salmon ecosystems. In this regard, we conclude that threethings need to happen in order to integrate climate information into sustainable salmon man-agement policies:

1. De-emphasize the role of preseason run-size predictions in management activities.2. Emphasize preseason and in-season monitoring of both the resource and its environment.3. Focus on strategies that minimize the importance of uncertain climate variability and

change scenarios to increase the resilience of short and long-term planning decisions.

Our bottom line is that sustainable salmon fisheries cannot be engineered with technologicalfixes and prediction programs, but that climate insurance for Pacific Northwest salmon can beenhanced by restoring and maintaining healthy, complex, and connected freshwater and estu-arine habitat and ensuring adequate spawner escapements. If we are interested in purchasinglong-term climate insurance for wild salmon so they can better cope with changing ocean con-ditions, we will likely get the best return on investments aimed at restoring the health andintegrity of our beleaguered watersheds. We also believe that the health of northwest salmonresources is inherently dependent upon social, economic, and political pressures in this worldof multi-objective resource conflict. Because of the human dimensions of salmon fisheries, weneed salmon fisheries if we hope to sustain wild salmon, and vice versa.

1 E-mail: [email protected] E-mail: [email protected]

knudsen.qxp 5/27/2004 2:11 PM 21

122 MANTUA AND FRANCIS

IntroductionPacific Northwest salmon and salmon fisheries arein crisis. Federal (U.S.) Endangered Species Actlistings of wild stocks have caused many regionalfisheries to be virtually shut down. In recent years,most of the salmon harvested from California toWashington were spawned in hatcheries. At thesame time, the rapid expansion of a global salmonaquaculture industry has drastically reduced mar-ket prices for salmon, greatly diminishing the prof-itability for most salmon fishers. Salmon manage-ment policies have, for decades, been focused onachieving better scientific understanding of fac-tors underlying salmon productivity, then usingthis knowledge to either forecast yield or engineerthe fish and their habitats to realize consistenthigh levels of production. Clearly, these policieshave failed to sustain fisheries and to sustain wildsalmon populations.

This essay focuses on the role of climate vari-ations in salmon fisheries. We discuss the impor-tance of climate in its effects on salmon produc-tion as well as how unpredictable these effects are.Our assessment leads us to conclude that the treat-ment of environmental uncertainty poses a funda-mental conflict between the kind of policies thathave been traditionally used in fishery manage-ment—command and control policies—and what theenvironmental variability dictates—what Hollingand Meffe (1995) call “Golden Rule” policies thatacknowledge a lack of control and predictabilityand aim to facilitate the natural processes thoughtto provide ecosystem resilience.

To develop our argument, we provide a briefreview of state-of-the-art prediction efforts in bothsalmon fisheries and climate science, discuss recentresults from selected studies of environmentalimpacts on Pacific salmon, describe the nature ofhuman-induced climate change and its potentialimpact on Pacific Northwest salmon managementpolicies, and relate how salmon have evolved diverselife histories and populations to deal with environ-mental variability and uncertainty. We concludethe essay by urging a fundamental shift in salmonresource conservation and management efforts thatinvolves new linkages between science, econom-ics, and policy.

Prediction andSalmon Management

For the past few decades, many salmon fishery man-agement agencies have relied on preseason run-size predictions to determine harvest rates and allo-cations between various user groups. Such forecastsrely on a number of explicit assumptions aboutsalmon population dynamics. For example, in manystreams, empirically determined spawner–recruitrelationships have been used to set escapement goalsat the maximum sustained yield (MSY) populationpoint (see Ricker 1958). Implicit in deterministicstock–recruit concepts like MSY are assumptionsthat environmental change is (a) relatively unim-portant, (b) unpredictable, and/or (c) difficult totranslate into the biophysical impacts that ultimatelyalter the spawner–recruit relationship of interest.

Yet fisheries scientists have long recognizedthat some of the errors in deterministic (spawner-,cohort-, and/or smolt-based) run-size predictionsmay be explained by environmental changes, andmany have attempted to incorporate environmen-tal data into such fish prediction models. For exam-ple, in the case of Oregon coho, a number ofresearchers have developed statistical models cor-relating various indices of the marine environmentwith smolt-to-adult return rates (e.g., Nickelson1986; Lawson 1997; Coronado and Hilborn 1998;Ryding and Skalski 1999; Cole 2000, Hobday andBoehlert 2001; Koslow et al. 2002; Logerwell etal. 2003). There is a growing body of evidence thatsupports the use of such models, as recent syn-theses of field studies in the northeast Pacific Oceanhave confirmed strong food-web responses to cli-mate-related changes in nearshore habitat. Thesebiophysical interactions include changes in upperocean stratification, nutrient concentrations, phy-toplankton production, zooplankton productionand community structure, forage fish abundance,as well as the distribution of highly migratorypelagics (see Chavez et al. 2002; Mackas andGalbraith 2002; Pearcy 2002; Peterson et al. 2002;and Chavez et al. 2003; for recent perspectives).On the terrestrial side, scientists have long knownthat temperature and stream flow extremes canhave a strong bearing on freshwater productivityfor salmonids (e.g., Spence 1995; Bradford and

knudsen.qxp 5/27/2004 2:11 PM Page 122

NATURAL CLIMATE INSURANCE FOR PACIFIC NORTHWEST SALMON AND SALMON FISHERIES 123

Irvine 2000). For example, researchers at Wash-ington’s Department of Fish and Wildlife havedeveloped empirically based regression models thatuse stream flow indices as predictors for wild cohosmolt production for several index stocks (Seiler etal. 2003).

The whole issue of deterministic prediction isa part of what Holling and Meffe (1995) term “com-mand and control” management. The intent is todefine a problem, bound it, and develop a solutionfor its control. In order to achieve this, policy mustbe linked to what Holling (1993) calls “first stream”science—essentially a science that assumes that anatural resource system is both knowable and pre-dictable (see Table 1 for a fuller description of thislinkage of science and policy).

Climate PredictionWhile it is trivial to predict that winter tempera-tures will be “cold” relative to those in summer inplaces like the Pacific Northwest, predicting thewithin and between winter climate variations ismuch more difficult. In some cases, year-to-yearvariations on the seasonal rhythms of climate arequite large, though rarely as large as the typical sea-sonal changes. When poor ocean conditions areblamed for causing problems for salmonids, the tar-get of that blame is never the strong and predictableseasonal climate change but instead it is focused onthe unexpected variations superimposed on the reg-ular seasonal rhythms. Since the late 20th cen-tury, governments of many nations have investedlarge efforts in predicting this type of climate vari-ation, with a special focus on predicting season-to-season and year-to-year climate changes (NRC 1996).

Much of the climate prediction effort has beenmotivated by the promise of societal benefits withan improved forewarning of droughts, floods, orclimate-related changes in natural resources likesalmonids or other valuable fish stocks. The ideais simple: better predictions of future climate shouldlead to better predictions and improved steward-ship of affected resources. This notion fits very wellwith some of the annual activities commonly foundin fishery management agencies. As noted above,in some regions preseason, run-size forecasts, usu-ally based on the number of parent spawners, smolts,

and/or cohort returns, are issued prior to harvestseasons in order to set harvest rates and determineallocations. Errors in preseason, run-size forecastshave been partially attributed to climate and habi-tat changes, for instance varying ocean condi-tions and ocean carrying capacity like thosedescribed by Pearcy (1992). Thus, better climatepredictions offer the promise of reducing some ofthe preseason run-size forecast errors, at least incases where the links between climate, environ-ment, and fishery productivity have been quanti-fied.

The state-of-the-art skill in the science of cli-mate prediction rests largely on a demonstratedability to monitor and predict the status of thetropical El Niño-Southern Oscillation (hereaftersimply El Niño). Cane et al. (1986) initiated themodern era of climate forecasting with a boldmodel-based forecast for an El Niño event in1986–1987 that was essentially correct. Since 1986,there has been some skill in predicting El Niño atlead times of one to a few seasons into the future.Once an El Niño forecast is made, predictions forclimate conditions outside the tropics can easily bemade, though forecast accuracy is always imper-fect (e.g., Bengston et al. 1993). During the 1990s,there were a few notable successes and a few equallynotable prediction failures (see below).

El Niño-based climate predictions are nowroutinely generated from a variety of sources, rang-ing from past climate data to the outputs gener-ated with sophisticated computer models. In theearly stages of the 1997–1998 El Niño—as earlyas June 1997—both approaches were used to makeremarkably accurate forecasts for the 1998 win-ter and spring climate of North America (see Barn-ston et al. 1999 and Mason et al. 1999 for reviewsof climate predictions associated with the extremelystrong 1997/1998 El Niño event). During El Niñoand La Niña periods since 1998, climate predic-tions for North America were often skillful, thoughnot as spectacularly successful, and in some casesquite inaccurate (for examples, see the NOAA Cli-mate Prediction Center’s web page athttp://www.cpc.ncep.noaa.gov). While existing fore-cast models have shown impressive skill in predicting theonset and demise of recent El Niño events, none havedemonstrated great skill in predicting the magnitude and



reach of tropical El Niño events. However, the com-bination of a comprehensive El Niño monitoringsystem—consisting of real-time buoy, satellite,island, and ship observations—along with the 8–14-month evolution of most events, means that real-time monitoring provides continuous updates oneach El Niño event’s progress in ways that can beused to minimize forecast errors over the course ofeach event. The multiseason evolution and sophis-ticated real-time monitoring system for El Niñoallows resource managers to update their expec-tations for “El Niño impacts” as the event evolves,so early forecast errors can be corrected.

In contrast to the relatively skillful short-termclimate predictions described above, climate sci-entists have demonstrated no skill in climatepredictions at lead times longer than 1 year intothe future. In spite of this situation, skillful pre-dictions at lead times from a few years to a fewdecades into the future may be possible if scien-tists decipher the now mysterious processes thatgive rise to multi-year climate variations like thoseassociated with the Pacific Decadal Oscillation(PDO) (Mantua and Hare 2002). If this happens,appropriate monitoring systems can be designedand deployed, and the necessary prediction mod-els can be developed (NRC 1998).

An essential point here is that today’s best cli-mate forecasts are probabilistic in nature. Unlikethe case of deterministic salmon run forecasts,deterministic climate predictions are simply notbelieved to be possible. To the best of our currentknowledge, future climate is subject to highlyunpredictable changes because of random eventsthat we simply cannot foresee. Thus, climate fore-casts are always presented as changes in the oddsfor certain events, making climate forecasting akinto riverboat gambling. Playing roulette, successcomes with correctly guessing “red” or “black”more often than not. Like playing the colors on aroulette wheel, the science of climate forecastinguses an approach that assigns odds for relativelycrude outcomes like “above average temperature”or “near normal precipitation.” Long-term fore-casts that detail the exact amount of snowfall thata specific ski resort will get in the coming winterare more likely based on someone’s wishes than onmodern scientific methods.

Climate and SalmonBecause of the never-ending pressures of naturalselection, wild animals have evolved behaviors thatallow them to “fit” into their seasonally chang-ing habitats. For northwest salmonids, aspects ofthis evolved behavior include the distinct seasonalruns of various stocks of the same and differentspecies. Thus, the strong seasonal rhythms in thelife history of salmonids can more likely be explainedas a consequence of evolution in an environmentwith strong seasonal rhythms than as a result of“climate prediction” by fish.

The results of a study by Logerwell et al. (2003)highlights a few important aspects of the pre-dictability of ocean climate variations thought tobe important for Oregon hatchery coho salmonOncorhynchus kisutch. Oregon hatchery coho have arelatively simple life history, at least in compari-son to other species and stocks of Pacific salmon.In the Pacific Northwest, most hatchery coho adultsspawn during fall or early winter months. Afterincubation, the eggs hatch into fry that develop asfreshwater fish for the next year or so. During theirsecond spring, hatchery juveniles undergo the smolt-ing process at which time they are released fromthe hatchery and migrate rapidly to sea. Typically,Oregon hatchery coho spend about 18 months atsea before returning to their natal rivers and/orhatcheries to spawn as mature 3 year olds.

There is abundant evidence that Pacific salmon,both of hatchery and natural origins, experiencelarge year-to-year and decade-to-decade changes inproductivity. A 30-year database for Oregon hatch-ery coho shows that smolt-to-adult survival ratesin the period 1969–1977 ranged from 3% to 12%,while in the period 1991–1998, those rates wereconsistently below 1% (see Figure 1). In periodslike 1969–1977 and 1984–1991, the year-to-yearchanges were also large, ranging from ~2% to 6%.The extremely low return rates in the 1990s indi-cated in Figure 1 were also observed for many otherstocks of wild and hatchery salmonids in the north-west. Figure 2 shows observed smolt-to-adult sur-vival rates for five wild and seven hatchery cohostocks tracked by Washington’s Department of Fishand Wildlife (Seiler et al. 2003). While there isclearly a large degree of between-stock variability



at year-to-year time scales, a common feature ineight of these records is decreasing survival trendsfrom the mid-1980s to the early 1990s and sus-tained low survival rates for the 1991–1998 period.Hobday and Boehlert (2001) examined smolt-to-adult return records for dozens of coho hatcheriesfrom Vancouver Island to California and the inlandwaters of the Georgia Strait and Puget Sound. Forthese regions, they found evidence for regionallycoherent very low coho survival rates for 1991–1998smolt releases. In summary, there is compellingevidence for regionally coherent low smolt-to-adultsurvival rates, presumably due to inhospitable oceanconditions, for many northwest salmon stocks dur-ing this dismal production period.

Logerwell et al. (2003) developed a rela-tively simple model for using physical environ-mental data to explain past smolt-to-adult survivalrates by synthesizing key results of many earlierstudies of coho marine survival and ocean environ-mental change (e.g., Nickelson 1986; Pearcy 1992;Lawson 1997; Coronado and Hilborn 1998; Ryd-ing and Skalski 1999; Cole 2000; Koslow et al.2002). The model is based on four environmentalindices: (1) the coastal ocean surface temperaturein the winter prior to smolt migration; (2) the dateof the Spring Transition, the date at whichupwelling-favorable winds are initiated; (3) coastalsea-level during the smolts’ first spring at sea (aproxy for coastal upwelling and alongshore trans-

Figure 1. Comparison of modeled and observed Oregon Production Index coho smolt-to-adult survival. This sim-ple environmental model uses observed relationships between smolt-to-adult survival and coastal ocean temper-atures, coastal sea level, and winds from 1969 to 1998 to translate subsequent environmental conditions into smolt-to-adult survival rate estimates. Estimates for ocean entry years 1999, 2000, and 2001 indicate improved oceanconditions over those observed between 1989 and 1998. See Logerwell et al. 2003 for more details.



port); and (4) the coastal ocean surface temperatureduring the maturing coho’s first and only winterat sea. Each of the four environmental indices isbelieved to capture different influences on the qual-ity of the near-surface coastal ocean habitat wherejuvenile and maturing coho are found. An impor-tant finding in this work is that the four differ-ent environmental indices are essentially uncor-related with each other. Yet, considering these fourenvironmental indices in sequence yields an envi-ronmental explanation for most of the ups anddowns in Oregon hatchery coho smolt-to-adult sur-vival rates for the past three decades.

This model was developed using data from1969 to 1998 and, with observed environmentaldata, has provided estimates for smolt-to-adultcoho survival for smolts entering the ocean in 1999,2000, and 2001 (see the *s in Figure 1). Based onthis model, coastal ocean conditions were muchimproved for Oregon coho smolts in 1999, 2000,and 2001, a result that was generally consistentwith substantial increases in observed smolt-to-adult coho survival rates. Compared to conditionsin 1991–1998, this improvement in smolt-to-adult return rates is believed to be related tofood–web changes prompted by significantly cooler

Figure 2. Observed smolt-to-adult survival rates (SARs) for 12 Washington coho stocks (by ocean entry year). Bluelines indicate observed SARs for the stocks listed in each plot, (W) indicates a wild stock, while (H) indicates ahatchery stock. The red lines (labeled GAM) depict the predicted OPI coho survival rates based on Logerwell etal.’s (2003) simple model that includes environmental conditions in the coastal ocean (repeated from Figure 1).



wintertime coastal ocean temperatures, an earlieronset of springtime upwelling winds, and moreupwelling and stronger equator-ward surfacecurrents in the months of April, May, and June(e.g., see Peterson et al. 2002).

We interpret the results of this modeling workas evidence that the “ocean conditions” importantfor Oregon coho are the net result of sequential tra-jectories of virtually independent climatic processes. Large-scale climate patterns like El Niño and the PDO aremodestly correlated with Oregon’s coastal winter-time SST, yet very poorly correlated with the springtransition date and springtime coastal sea levels.

This means that knowing El Niño was under-way in the summer and fall of 2002 provided someconfidence that coastal ocean temperatures duringthe subsequent winter (2002/2003) would likelybe warmer than average. Yet it provided no cluesabout how early, how strong, or how often thespringtime upwelling winds would blow or whatthe coastal sea level would be in spring 2003, norwould it provide skill for predicting ocean tem-peratures in the winter of 2003/2004.

If this biophysical model is valid, predictabil-ity for ocean conditions important for Oregon cohois severely limited. Viewed from another perspec-tive, it suggests that Oregon coho face a high degreeof environmental uncertainty every year, at least interms of ocean conditions (but surely also in termsof stream and estuary conditions).

Climate Change and SalmonThere is a growing recognition that, at time framesof a few to many decades into the future, human-caused climate change scenarios may provideuseful insights into the potential for protecting andrestoring depleted salmon populations. The cul-prit behind human-caused climate change has beenidentified: emissions from burning fossil fuels anddeforestation have substantially increased the atmos-pheric concentrations of radiatively important tracegases, the “greenhouse gases.” While it is impos-sible to predict future greenhouse gas emissions,there is little doubt that greenhouse gas concen-trations will continue to rise for at least the nextfew decades and that stabilizing late 21st centurygreenhouse gas concentrations at levels two or eventhree times preindustrial levels will require major

changes in global energy use patterns and technol-ogy (IPCC 2001). The consequences of the human-caused enhancement of Earth’s natural greenhouseeffect are largely unknown; yet, future climate sce-narios developed by the Intergovernmental Panelon Climate Change (IPCC) all point to a high like-lihood that climate in the 21st century and beyondwill be substantially warmer, and significantly dif-ferent, than any experienced in the past millen-nium (IPCC 2001; NAS 2001).

While there has been some speculation aboutclimate change impacts on fisheries, the transla-tion of climate change scenarios into fishery impactscenarios poses a range of difficult problems. Animpacts assessment methodology in use today isrelatively straightforward. Typically, output isobtained from a climate simulation model; selectedclimate model outputs are then prescribed for abiophysical response model. For example, Welchet al. (1998a, 1998b) hypothesized that marinehabitat for sockeye salmon Oncorhynchus nerka andsteelhead O. mykiss is defined in part by sharp “ther-mal limits” and that warmer upper ocean temper-atures in the late 21st century (around the time ofCO2 doubling) will significantly shrink marinehabitat for these species, as well as displace themnorthward into reduced areas in the Bering andChukchi seas. Likewise, terrestrial temperaturechange scenarios from a suite of climate changesimulations have been used to estimate changes instream habitat for salmonids. In one recent study,projected peak stream temperature changes andknown freshwater thermal limits for salmonidswere used to estimate stream habitat losses in thewestern continental United States due to globalwarming (DFW/NRDC 2002).

Even after such methodologies are applied, thetranslation of climate change information intoimpacts scenarios carry large uncertainties: in addi-tion to generally poorly known transfer functionsand/or parameterizations in the biophysical mod-els, there are dozens of combinations of emissionsscenarios and climate models to choose from, withlittle guidance as to which climate scenarios aremost likely to be realized (Mantua and Mote 2002).An important message for fishery scientists, managers,and policy-makers is the near consensus in the climateresearch community that the climate of the 20th cen-tury is not likely to provide a good guide for climate of



the near and distant future. A baseline assumptionfor us, then, is that environmental uncertainty forPacific salmon will remain large, if not growing,in the remainder of the 21st century.

Life withEnvironmental Uncertainty

Many scientists have postulated that a diversity ofbehaviors and environmental sensitivities serve asevolutionary responses to successful adaptation inuncertain environments (e.g., see ISG 2000). Forexample, coastal rockfish have evolved a protractedage structure that allows them to weather years ofenvironmental conditions unfavorable to high pro-duction. Longevity is their adaptive strategy forperpetuation. Pacific hake have evolved the abil-ity to undertake long seasonal migrations, everyyear feeding in one large marine ecosystem (PacificNorthwest coastal ocean) and reproducing in another(the Southern California Bight). Western Alaskasockeye salmon have evolved very complex popu-lation structures, what Hilborn et al. (2003) callbiocomplexity, to maintain their resilience to envi-ronmental variability and change.

At the metapopulation level, each species ofPacific salmon exhibits many such risk-spreadingbehaviors through a broad diversity of time-space habitat use by different stocks and substocksof the same species. Lichatowich (1999) put it thisway: “life history diversity is the salmon’s responseto the old adage of not putting all one’s eggs in thesame basket.” In the words of Hilborn et al. “bio-complexity of fish [western Alaska sockeye salmon]stocks is critical for maintaining their resilience toenvironmental change.”

Empirical evidence for diverse life historybehaviors and habitat use by salmonids, and theirclose relationship to the physical environment, isfound nearly everywhere researchers have looked.For instance, studies of natural coho smolt produc-tion in western Washington yield evidence for awide variety of stream flow sensitivities in nearbywatersheds (Seiler et al. 2003). In some systems,wild coho smolt production is limited by high win-ter flows that scour nests and damage incubatingeggs. In other streams, the main limiting factor islow summer flows that reduce rearing habitat. Inother streams, high fall flows allow spawners to

access otherwise unreachable tributary spawningbeds. Seasonal climate and short-term (day-to-day)weather variations cause these limiting flow fac-tors to vary sometimes within and sometimesbetween streams from year-to-year as well. The bot-tom line is that the complex landscape and varietyof watersheds in western Washington provide adiversity of habitats with different environmen-tal sensitivities. Because coho salmon occupy eachof the different habitats, the species as a whole car-ries a diverse portfolio of climate and environmen-tal sensitivity, what we like to think of as an evolvedexpression of natural climate insurance.

Recent studies have attempted to better under-stand the temporal and spatial characteristics ofmarine salmon habitat as well. Weitkamp and Neely(2002) document the fact that different coho stocksutilize distinctly different areas in the coastal watersof the northeast Pacific. Mueter et al. (2002) demon-strate that year-to-year coastal SST and upwellingwind variations covary at spatial scales of just a fewhundred kilometers in spring, summer, and fall,but at much larger spatial scales (1,000–2,500 km)in winter. Taken together, these studies suggest thatdifferent stocks of the same species (in this case,coho) also experience a diversity of marine habitatat the same time in different places, and at differ-ent times in the same places. Just like the case offreshwater habitat uncertainty and complexity,prominent characteristics of marine habitat forsalmon are diversity, variability, and unpredictablechanges at a broad spectrum of time and space scales.

Given the complexity of freshwater salmonhabitat in western North America and marine habi-tat in the Pacific Ocean, it seems likely that theseexamples of complex habitat, variability, environ-mental sensitivity, and complex salmon stock struc-ture are not unique to western Washington coho orwestern Alaska sockeye salmon. Instead, biocom-plexity in wild Pacific salmon populations appearsto be more likely the rule than the exception.

What Does It All Meanfor Salmon Policy?

How might an explicit consideration of climateand environmental uncertainty aid salmon manage-ment efforts? We first review some key characteris-tics of the inseparable linkages between science and



natural resource policy, and then show that, in ourview, the issue of saving wild salmon is insepara-bly linked to the issue of saving salmon fisheries.Finally, we offer some specific recommendations ofwhat might be done to resolve some existing cli-mate-related conflicts between salmon ecology andsalmon management.

Although at times they may seem independ-ent, in our view, natural resource science and pol-icy are inextricably linked. In fact, this is essentialif policy is to be based on scientific assessments andanalyses. Holling and Meffe (1995) and Holling(1993) characterize two distinct forms of this link-age, which we summarize in Table 1. The first links“command and control” management with “firststream” science. The key underlying assumptionsof both are that relevant aspects of the naturalresource system are knowable and predictable andthat variability can be controlled. Examples of thisare salmon policies that try to maintain a constantabsolute harvest or constant escapement, or usehatcheries to boost fishery production at times oflow natural production. The second links “goldenrule” management with “second stream” science.The assumption here is that relevant aspects of thesystem are inherently unknowable and unpredictableand that variability is to be celebrated and pre-served. The golden rule management goal is tofacilitate existing natural processes rather than con-trol them. For example, to maintain and restorediverse and connected freshwater and estuarinehabitat, to restrict harvest rates to very low levelsto allow as wide a diversity of populations to attainadequate spawning escapement, and to time hatch-ery releases to have minimal direct interactionswith wild populations are all examples of goldenrule management.

It is our contention that sustainable wild salmonpopulations and sustainable salmon fisheries aretightly linked. In our view, one necessary elementfor sustainable salmon fisheries is sustainable wildsalmon populations. While salmon hatcheries nowsupport the bulk of salmon harvests in the PacificNorthwest, history tells us that hatchery produc-tion of salmonids is not a sustainable fishery policy(Lichatowich 1999; Taylor 1999). From an ecolog-ical perspective, the genetic and behavioral diver-sity found in wild salmon populations and absentin most hatchery populations may be critical for the

long-term survival of salmon in the face of environ-mental change (Hilborn et al. 2003). To sustainwild salmon in our world of natural resource con-flict is to sustain the direct interaction betweenhumans and salmon. For example, in the 1950s,hydropower interests lobbied for the constructionof Moran Dam, which would have risen 220 m fromthe bottom of the Fraser River Canyon and blockedaccess to 44% of the upriver sockeye spawninggrounds. This proposal created a Columbia River-style tradeoff between the choice of industrial devel-opment over wild salmon habitat and productivity.Fishing interests led fierce political opposition thateventually halted this project (Meggs 1991). With-out viable fisheries, many Pacific Northwest salmonstocks would likely have gone extinct long ago.Fisheries generate the social and economic incen-tives that build the political clout needed to pre-serve the source of their sustenance.

With these management considerations inmind, what should we do? It seems to us that peo-ple now know enough about salmon ecology toprovide a list of needs for policy makers to developeffective action plans to save threatened wild salmonpopulations. Key pieces of an effective restora-tion plan should include the following (ISG 2000):

• Restrict harvests so adequate numbers of adultscan spawn, and avoid harvest policies thatmight alter the life history and/or genetic diver-sity of the spawning population.

• Restore diversity by reforming and/or closinghatcheries to significantly reduce negativeimpacts on wild stocks.

• Restore and protect habitat, and where neces-sary, remove barriers to fish passage.

• Align economic incentives for salmon fisherieswith conservation incentives for preservingwild salmon populations.

• Accept variability, environmental uncertainty,and acknowledge a lack of predictability (cf.ISG 2000).

In contrast, current efforts to sustain salmonfisheries are clearly a matter of often poorly coor-dinated politics, economics, law, and ecology. Aswild salmon numbers have declined, a combina-tion of scientific, political, and socioeconomicpressures have led to a slow withering of harvest



opportunities, an increased dependence on hatch-ery production, and seasons that are ratcheted downto smaller and smaller windows of time to protectendangered wild stocks while still allowing forharvest opportunities. Policies have generallyfocused on single-species biomass or numbers ofsalmon (e.g., OPI hatchery coho for harvest), ratherthan considering the ecosystems of which the tar-get fish populations are a part. Salmon restorationefforts have concentrated on tweaking the statusquo, and this has been especially true with effortsto improve fish passage around dams and to reformhatcheries. Management agencies have attemptedto reduce and/or eliminate variability by usingsalmon hatcheries to divorce fish production fromhabitat. Globalized economics have forced har-vested wild salmon to be sucked into an interna-tional commodity market that is flooded with verylow-price, high volumes of aquaculture salmon, amarket reality that most Pacific Northwest salmonfishers are struggling to compete in. Wild salmonmarkets must be restructured if they are to sur-vive, perhaps in ways that focus on relativelylow volumes of high-valued premium quality fishfor domestic and foreign consumption. Suchspecialty markets are already developing with brandnames such as “Copper River Sockeye” and “YukonRiver Kings.” Finally, deterministic and static con-cepts like maximum sustained yield have beenused to set harvest goals, pressuring fishery man-

agers to develop better predictive models so har-vests can be maximized while still offering pro-tection for spawning stocks, a scenario under whichmany wild populations exhibit chronically lowproduction (Knudsen 2000).

In summary, we have seen scientific manage-ment that forces complex and highly variable bio-physical systems to be balanced on a razor’s edgeto satisfy social demands for both extraction and pro-tection of the same resource along with alternativeuses for critical habitat. As a result, we have devel-oped conflicts between those who want to sustainwild salmon populations and those who want tosustain salmon fisheries. Clearly, this has becomea “lose–lose” situation.

From this comparison, the main climate com-ponent of the conflict between protecting and restor-ing wild salmon populations and protecting salmonfisheries lies in the treatment of environmentalvariability, and it is in this realm where the poten-tial for policy reform looks greatest. While socioe-conomic pressures produce political pressures toreduce variability and/or increase predictability sothat resource use can be maximized, salmon habi-tat and ecosystems contain fundamentally unpre-dictable dynamics (cf. Holling and Meffe 1995)(Table 1). Climate enters this picture through itsrole in providing strong limitations on ecosys-tem predictability, a situation that is not likely tochange in the future.

Table 1. Key characteristics of linkages between science and natural resource policy, after Holling and Meffe (1995).

First stream Second stream

Science • System knowable and • Ecosystem evolving, haspredictable inherent unknowability and

unpredictability

• Science of parts and • Science of integrationdisciplines

• Seek prediction • Seek understanding

Command and control Golden rule

Policy • Problem perceived, • Retain or restore criticalbounded, and solution types and ranges of naturalfor control developed variations

• Objective: reduce • Facilitate existingvariability and make processes and variabilitysystem more predictable



So how might we better manage our salmonresources if changes in climate, ocean conditions,and salmon populations are so difficult to predict?We have three suggestions:

• De-emphasize preseason run-size predictions.• Emphasize monitoring, including in-season

run-size assessments to guide short-term har-vest decisions and the development of appro-priate real-time environmental measurementsystems to gain insights into environmentalimpacts on stock productivity.

• Focus on strategies that minimize the impor-tance of uncertain climate variability andchange scenarios.

Now consider these suggestions one at a time.First, if predicting the future presents such a dif-ficult challenge, it would be wise to distance man-agement performance from prediction accuracy. Foryear-to-year management needs, monitoring targetstocks and their habitat offers a much more fail-safeapproach than does a continued reliance on highlyuncertain predictions. Simple environmental mod-els like the one used to estimate coho smolt-to-adultsurvival provide one means for translating the prod-ucts of environmental monitoring into an estimatefor difficult to measure coho survival rates. An evenbetter approach would rest on directly monitoringthe target stocks, a daunting task, yet one thathas received considerable attention in recent researchprojects along the Pacific Coast. If social and polit-ical pressures demand preseason predictions, suchmonitoring-based “forecasts” (using any combi-nation of cohort, smolt, or environmental models)should explicitly acknowledge the large uncertain-ties that exist. Thus, when preseason run-size fore-casts are made, stakeholders and managers shouldbe presented with a range of possible populationsto work with, along with some estimate for theodds of actually witnessing the high, low, and mid-dle parts of the predicted range. Alaska’s Depart-ment of Fish and Game manages Bristol Bay sock-eye salmon fisheries in precisely this way, annuallydeveloping preseason run-size forecasts with explicit(and often large) uncertainties, along with in-sea-son run-size monitoring and real-time terminal-area harvest decisions to ensure escapement goalsare met. On the other hand, in California, Oregon,

and Washington, where many wild salmon pop-ulations are severely depleted, harvest managersrely on deterministic and scientifically question-able preseason run-size forecasts, difficult to ver-ify assumptions about at-sea distributions of manywild and hatchery stocks, and mixed-stock oceanfisheries (SRSRP 2001). It is clear that Alaska’semphasis on monitoring rather than determinis-tic predictions has contributed to the sustainabil-ity of Alaska’s wild sockeye salmon fisheries. Webelieve the Alaska approach provides an attractivealternative to the northwest salmon management’scontinued reliance on deterministic preseason run-size forecasts.

A second critical step is to do enough moni-toring so that changes in freshwater and marineproductivity can be tracked and discriminated.Today, only a small number of streams are moni-tored for the full life cycle of salmonids (e.g., seeSeiler et al. 2003). Keeping track of adult spawn-ers and estimating harvests allows for a gross esti-mate of productivity. Because marine variabilityhas strong influences on salmon productivity, it iscritically important that harvest rates be reducedin periods of low ocean productivity to allow foradequate spawner escapements (Knudsen 2002).A better understanding of changes in marine andfreshwater productivity rests on the establishmentof long-term, continuous records of the age-struc-ture of spawners, parr, and smolt production, inaddition to the more commonly collected indexcounts used to estimate total spawners. A betterunderstanding for existing bottlenecks in wildsalmon productivity is also necessary to maximizethe “bang for each buck” spent on salmon restora-tion projects. We realize that monitoring programsare expensive and difficult to maintain. However,we suggest that at least as much money and effortshould be spent on monitoring as is already spenton artificial enhancement and short-term engineer-ing fixes (i.e., salmon hatcheries).

Finally, for dealing with the uncertaintiesposed by natural climate variability as well aslong-term anthropogenic climate change, the mostconservative and effective long-term strategy isone that minimizes the importance of highlyuncertain climate change scenarios and their atten-dant impact scenarios for wild salmon. In thisrealm, the critical step is to place a much higher



priority on restoring the natural climate insur-ance that wild salmon populations must haveevolved to survive and thrive in the face of pastenvironmental changes. We believe that this insur-ance is intimately associated with a diversity oflife history behaviors that, in turn, are directlylinked to the availability of healthy, complex, andconnected freshwater habitat. A diversity of fresh-water habitat leads to a wide range of seasonalrunoff patterns, as well as a wide range of short-term runoff responses to short-term weatherand geologic events. Such environmental variabil-ity effectively forces substocks of the same speciesinto different niches and different behaviorsthrough the never-ending process of natural selec-tion. If we are interested in purchasing long-termclimate insurance for wild salmon so they can bet-ter cope with changing ocean conditions, we willlikely get the best return on investments aimedat restoring the health and integrity of our belea-guered watersheds. If we do this effectively, wemight also bring back the resource base requiredfor viable and sustainable salmon fisheries.

AcknowledgmentsWhile this essay was inspired by a variety of insight-ful commentary on the Pacific Northwest salmoncrisis, Jim Lichatowich’s Salmon Without Rivers, Joe-seph Taylor’s Making Salmon, and Arthur McEvoy’sThe Fisherman’s Problem were especially provocativebooks that prompted us to consider the broaderaspects of salmon fisheries than those related to thebiophysical interactions of climate, environment,and salmon. Taylor’s (2001) unpublished essay“Seeking the Entangled Bank: Ecologically andEconomically Sustainable Salmon Recovery” pushedus to further grapple with the human side of thenorthwest salmon equation. We thank Ed Miles,Amy Snover, Bill Pearcy, and three anonymousreviewers for their constructive comments on draftversions of this essay. This essay was supportedby collaborations in the University of Washing-ton’s Climate Impacts Group, an interdisciplinaryeffort within the Joint Institute for the Study ofthe Atmosphere and Oceans and School of MarineAffairs and was funded under NOAA’s cooperativeagreement #NAI17RJ1232 and The Hayes Cen-ter. This is JISAO contribution #884.

ReferencesBarnston, A. G., A. Leetmaa, V. E. Kousky, R. E.

Livezey, E. A. O’Lenic, H. Van den Dool, A. J.Wagner, and D. A. Unger. 1999. NCEP fore-casts of the El Niño of 1997–98 and its U.S.Impacts. Bulletin of the American Meteoro-logical Society 80:1829–1852.

Bengston, L., U. Schlese, E. Roeckner, M. Latif, T. P.Barnett, and N. E. Graham. 1993. A two-tieredapproach to long range climate forecasting. Sci-ence 261:1026–1029.

Bradford, M. J., and J. R. Irvine. 2000. Land use, cli-mate change, and the decline of Thompson River,British Columbia, coho salmon. Canadian Jour-nal of Fisheries and Aquatic Sciences 57:13–16.

Cane, M. A., S. E. Zebiak, and S. C. Dolan. 1986.Experimental forecasts of El Niño. Nature (Lon-don) 321:827–832.

Chavez, F. P., J. T. Pennington, C. G. Castro, J. P.Ryan, R. P. Michisaki, B. Schlining, P. Walz, K.R. Buck, A. McFadyen, and C. A. Collins. 2002.Biological and chemical consequences of the1997–1998 El Niño in central California waters.Progress in Oceanography 54:205–232.

Chavez, F. P., J. Ryan, S. E. Lluch-Cota, and M. NiquenC. 2003. From anchovies to sardines and back:multidecadal change in the Pacific Ocean. Sci-ence 299:217–221.

Cole, J. 2000. Coastal sea surface temperature andcoho salmon production off the north-west UnitedStates. Fisheries Oceanography 9:1–16.

Coronado, C., and R. Hilborn. 1998. Spatial and tem-poral factors affecting survival in coho salmon(Oncorhynchus kisutch) in the Pacific Northwest.Canadian Journal of Fisheries and Aquatic Sci-ences 55:2067–2077.

DFW/NRDC (Defenders of Wildlife/NationalResources Defense Council). 2002. Effects ofglobal warming on trout and salmon habitat inU.S. streams. Available from the Defenders ofWildlife, 1101 Fourteenth Street, NW, Suite1400, Washington, D.C. 20005. Available at:http://www.defenders.org.

Hilborn, R., T. P. Quinn, D. E. Schindler, D. E.Rogers. 2003. Biocomplexity and fisheries sus-tainability. Proceedings of the National Acad-emy of Sciences 100:6564–6568.

Hobday, A. J., and G. W. Boehlert. 2001. The roleof coastal ocean variation in spatial and tempo-ral patterns in survival and size of coho salmon



(Oncorhunchus kisutch). Canadian Journal of Fish-eries and Aquatic Sciences 58:2021–2036.

Holling, C. S. 1993. Investing in research for sustain-ability. Ecological Applications 3:552–555.

Holling, C. S., and G. K. Meffe. 1995. Commandand control and the pathology of naturalresources management. Conservation Biology10:328–337.

IPCC (Intergovernmental Panel on Climate Change).2001. Summary for policymakers, a report ofworking group I. Available at: http://www.ipcc.ch

Independent Scientific Group (ISG). 2000. ScientificIssues in the restoration of salmonid fishes in theColumbia River. Fisheries 24(3):10–19.

Knudsen, E. E. 2000. Managing Pacific salmon escape-ments: the gaps between theory and reality. Pages237–272 in E. E. Knudsen, C. S. Steward, D. D.MacDonald, J. E. Williams, and D. W. Reiser,editors. Sustainable fisheries management: Pacificsalmon. Lewis, New York.

Knudsen, E. E. 2002. Ecological perspectives onPacific salmon: can we sustain biodiversity andfisheries? Pages 227–319 14 in K. D. Lynch, M.L. Jones, and W. W. Taylor, editors. SustainingNorth American salmon: perspectives acrossregions and disciplines. American Fisheries Soci-ety, Bethesda, Maryland.

Koslow, J. A., A. Hobday, and G. W. Boehlert. 2002.Climate variability and marine survival of cohosalmon (Oncorhynchus kisutch) off the coast of Cal-ifornia, Oregon and Washington. FisheriesOceanography 11(2):65–77.

Lawson, P. 1997. Interannual variability in growthand survival of chinook and coho salmon. Pages81–92 in R. Emmet and M. Schiewe, editors.Estuarine and ocean survival of northeasternPacific salmon: proceedings of the workshop.U.S. Department of Commerce. NOAA Techni-cal Memo., NMFS-NWFSC-29.

Lichatowich, J. 1999. Salmon without rivers. IslandPress, Washington, D.C.

Logerwell, E., N. Mantua, P. Lawson, R. Francis, andV. Agostini. 2003. Tracking environmental bot-tlenecks in the coastal zone for understandingand predicting Oregon coho marine survival.Fisheries Oceanography 12(3):1–15.

Mackas, D. L., and M. Galbraith. 2002. Zooplank-ton community composition along the innerportion of Line P during the 1997–1998 ElNiño event. Progress in Oceanography54:423–437.

Mantua, N. J. and S. R. Hare. 2002. The PacificDecadal Oscillation. Journal of Oceanography58(1):35–44.

Mantua, N. J., and P. M. Mote. 2002. Uncertaintyin scenarios of human-caused climate changescenarios. Pages 263–272 in N. A. McGinn,editor. Fisheries in a changing climate. Amer-ican Fisheries Society, Symposium 32, Bethesda,Maryland.

Mason, S. J., S. M. Goddard, N. E. Graham, E.Yulaeva, L. Sun, and P. A. Arkin. 1999. The IRIseasonal climate prediction system and the1997/98 El Niño event. Bulletin of the Ameri-can Meteorological Society 80:1853–1873.

Meggs, G. 1991. Salmon: the decline of the BritishColumbia fishery. Douglas and McIntyre, Van-couver.

Mueter, F. J., D. M. Ware, and R. M. Peterman. 2002.Spatial correlation patterns in coastal environ-mental variables and survival rates of Pacificsalmon in the northeast Pacific Ocean. FisheriesOceanography 11(4):205–218.

NAS (National Academy of Sciences). 2001. Climatechange science: an analysis of some key questions.National Academy Press, Washington, D.C.

NRC (National Research Council). 1996. Learningto predict climate variations associated with ElNiño and the Southern Oscillation: accomplish-ments and legacies of the TOGA Program.National Academy Press, Washington, D.C.

NRC (National Research Council). 1998. Decade-to-century scale climate variability and change: ascience strategy. National Academy Press, Wash-ington, D.C.

Nickelson, T. 1986. Influences of upwelling, oceantemperature, and smolt abundance on marinesurvival of coho salmon (Oncorhynchus kisutch)in the Oregon Production Area. Canadian Jour-nal of Fisheries and Aquatic Sciences43:527–535.

Pearcy, W. G. 1992. Ocean ecology of north Pacificsalmonids. University of Washington Press,Seattle.

Pearcy, W. G. 2002. Marine nekton off Oregon andthe 1997–98 El Niño. Progress in Oceanogra-phy 54:399–403.

Peterson, W. T., J. E. Keister, L. R. Feinberg. 2002.The effects of the 1997–99 El Niño/La Niñaevents on hydrography and zooplankton off thecentral Oregon coast. Progress in Oceanography54:381–398.

knudsen.qxp 5/27/2004 2:12 PM 3


Ricker, W. E. 1958. Handbook of computations forbiological statistics of fish populations. FisheriesResearch Board of Canada 119.

Ryding, K., and J. Skalski. 1999. Multivariate regres-sion relationships between ocean conditions andearly marine survival of coho salmon (Oncorhynchuskisutch). Canadian Journal of Fisheries and AquaticSciences 56:2374–2384.

SRSRP (Salmon Recovery Scientific Review Panel).2001. Report for the meeting held August 27–29,2001. National Marine Fisheries Service, North-west Fisheries Science Center, Seattle. Availableat: http://research.nwfsc.noaa.gov/cbd/trt/rsrp.htm

Seiler, D., G. Volkhardt, S. Neuhauser, P. Hanratty, L.Kishimoto, P. Topping, and M. Ackley. 2003. 2003wild coho forecasts for Puget Sound and Washing-ton coastal systems. Washington Department ofFish and Wildlife Science Division, Olympia.

Spence, B. 1995. Geographic variation in timing offry emergence and smolt migration in cohosalmon (Oncorhynchus kisutch). Doctoral disserta-tion. Oregon State University, Corvallis.

Taylor, J. E., III. 1999. Making salmon. Universityof Washington Press, Seattle.

Taylor, J. E., III. 2001. Seeking the entangled bank:ecologically and economically sustainable salmonrecovery. Unpublished manuscript. Presentationto Northwest Salmon Crisis Lecture Series, Uni-versity of Washington, Seattle.

Weitkamp, L., and K. Neely. 2002. Coho salmon(Oncorhynchus kisutch) ocean migration patterns:insight from marine coded-wire tag recoveries.Canadian Journal of Fisheries and Aquatic Sci-ences 59:1100–1115.

Welch, D. W., Y. Ishida, and K. Nagasawa. 1998a.Thermal limits and ocean migrations of sockeyesalmon (Oncorhynchus nerka): Long-term conse-quences of global warming. Canadian Journal ofFisheries and Aquatic Sciences 55:937–948.

Welch, D. W., Y. Ishida, K. Nagasawa, and J. P. Eve-son. 1998b. Thermal limits on the ocean distri-bution of steelhead trout (Oncorhynchus mykiss).North Pacific Anadromous Fish CommissionBulletin 1:396–404.


Natural Climate Insurance for Pacific Northwest Salmon and Salmon...

Documents

Transcript of Natural Climate Insurance for Pacific Northwest Salmon and Salmon...