October 2002 Vol 52 No 10 BioScience 917

Articles

Salmon runs in the Pacific Northwest have been declining for decades so much so that many runs are

threatened or endangered others have been completely ex-tirpated (Nehlsen et al 1991) This ldquosalmon crisisrdquo loomslarge in the public eye because it has serious and wide-ranging economic cultural and ecological repercussionsBillions of dollars have gone into industrial and agriculturalprojects that alter regional rivers in ways that often unin-tentionally make them inaccessible or unsuitable for salmonRecently billions more have been spent in largely unsuc-cessful attempts to restore the languishing salmon runs(Lichatowich 1999) Moreover enormous nonmonetary re-sources have been expended in assigning and denying re-sponsibility for failed runs and debating the possible efficacyof various remedies

As resources that are devoted to reversing declining runsof salmon have increased scientists and resource managershave been expanding our understanding of the ecological roleof salmon and other anadromous fishes which return fromthe sea to spawn in fresh water We have known for years thatspawning salmon serve as a food resource for wildlife species(eg Shuman 1950) and when they die after spawning (asmost Pacific salmon do) their carcasses provide nutrients (egcarbon [C] nitrogen [N] phosphorus [P]) to freshwatersystems (eg Juday et al 1932) More recently scientists havedocumented that these ldquosalmon-derived nutrientrdquo subsidiesmay have significant impacts on both freshwater and ripar-ian communities and on the life histories of organisms thatlive there (Willson et al 1998 Cederholm et al 1999)

Because of the burgeoning interest in salmon growingindications of their ecological importance and recent calls formanagement to consider the role of salmon in aquatic and ter-restrial ecosystems (eg Larkin and Slaney 1997) we take thisopportunity to review what is understood about the functionof salmon as key elements of ecological systems Our objec-tives are twofold First we expand on previous reviews of

salmon (Willson et al 1998 Cederholm et al 1999) to includerecent research that has amplified and modified earlier ideasabout the contribution of salmon to ecosystem processes Indoing so we describe the composition magnitude and dis-tribution of marine inputs to freshwater and terrestrial sys-tems via salmon We use an expanding group of studies per-taining to stream nutrient budgets and salmon physiology toconstruct a schematic that illustrates salmon-derived prod-ucts and the pathways by which they enter and are retainedin aquatic and terrestrial food webs We then consider the eco-logical variation associated with salmonid ecosystems and howthis may influence the ecological response to the salmon in-put Second we consider how this variation in ecosystem re-sponse may influence management and conservation efforts

Scott M Gende (e-mail smgendefsfedus) is an ecologist and Richard TEdwards is a research aquatic ecologist in the Aquatic and Land InteractionsProgram with the Pacific Northwest Research Station 2770 Sherwood LaneSuite 2A Juneau AK 99801 Mary F Willson is a research faculty memberat the University of AlaskandashFairbanks 5230 Terrace Place Juneau AK99801 Mark S Wipfli is a research aquatic ecologist in the Aquatic and LandInteractions Program Pacific Northwest Research Station 1133 N WesternAvenue Wenatchee WA 98801 copy 2002 American Institute of BiologicalSciences

Pacific Salmon in Aquaticand Terrestrial Ecosystems

SCOTT M GENDE RICHARD T EDWARDS MARY F WILLSON AND MARK S WIPFLI

PACIFIC SALMON SUBSIDIZE FRESHWATER

AND TERRESTRIAL ECOSYSTEMS THROUGH

SEVERAL PATHWAYS WHICH GENERATES

UNIQUE MANAGEMENT AND CONSERVA-

TION ISSUES BUT ALSO PROVIDES

VALUABLE RESEARCH OPPORTUNITIES

918 BioScience October 2002 Vol 52 No 10

Articles

We conclude by suggesting new research directions to help fillthe gaps in our current understanding of salmonid ecosystems

The salmon input We focus on five species of Pacific salmon that spawn infreshwater systems of North America chinook (Oncorhynchustshawytscha) sockeye (O nerka) pink (O gorbuscha) chum(O keta) and coho (O kisutch) All five species share a gen-eral life history Adults return to freshwater usually in late sum-mer and fall where they cease feeding spawn and die Aftersome months young emerge from the gravel in early springand depending on the species spend up to 2 years in fresh-water habitats before migrating out to sea The fish remain atsea for 1 to 7 years feeding and gaining over 90 of their bio-mass before returning to fresh water to complete the cycle(Groot and Margolis 1991)

Composition When salmon enter fresh water they havestored nearly all of the energy necessary for upstream mi-gration and reproduction but the magnitude of resource re-serves varies greatly among populations often in accord withthe length of upstream migration (Brett 1995 Hendry andBerg 1999) For example chum salmon in the Yukon Rivermay spawn near the mouth or nearly 2000 kilometers (km)upriver The difference in chemical composition of ldquoup-riverrdquo versus ldquodownriverrdquo spawning salmon is so large that itwas suggested as a metric to separate the stocks that are har-vested in a mixed-stock fishery on the coast (Brett 1995)

As salmon migrate and approach spawning areas theirbody composition changes dramatically Both males and fe-males store much of their lipid throughout the visceral andsoma tissue particularly in the muscle skin and skeletal tis-sue (Hendry and Berg 1999 Gende 2002) Stored lipids fuelmigration to spawning grounds which results in loss of someof the energy to metabolic heat en route and produces car-bon dioxide and water as waste products Females use somestored lipid for production of eggs which can constitutemore than 20 of body mass whereas males tap both lipidand protein reserves to develop secondary sexual character-istics (Hendry and Berg 1999) Once on the spawning groundsfish use most of the remaining lipid to fuel spawning activi-ties such as excavating and defending redds (females) andfighting for access to females (males) Body tissue proteins alsoare used as an energy source resulting in endogenous nitro-gen excretion in the form of ammonia (and some urea) pri-marily across the gill membrane (Wood 1995) The length oftime fish live on the spawning grounds varies dependingupon the population (generally less than 3 weeks) Althougha large fraction of the lipid and protein has been metabolizedor deposited in the gravel as eggs carcasses may contain upto 16 protein and 35 kilojoules per gram wet mass of en-ergy (Hendry and Berg 1999 Gende 2002) Other forms ofN such as collagen in the skin may not be depleted and re-main within the carcass

Salmon also contain macroelements that is mineralsfound in large amounts (Robbins 1993) such as potassium

and calcium (Ca) Although salmon lose calcium phosphateduring migration which allows bone tissue to turn into car-tilage necessary for the formation of secondary sexual char-acteristics total body Ca and P do not decline precipitouslyduring freshwater migration and spawning (at least for pinkand chum salmon populations that spawn in small coastalstreams Gende 2002) Phosphorus stored chiefly in the bonemuscle and male gonads makes up less than 05 of the bodymass but may be an important input when considering thelarge numbers of fish entering streams and lakes (Donaldson1967) Few studies have quantified the macroelement contentof the fish but clearly it may play a role in the nutritional qual-ity of salmon for consumers for example some mineralsare necessary in small amounts for health and growth (Rob-bins 1993)

Magnitude The flux of salmon biomass entering freshwater from the ocean can be massive A large run of 20 mil-lion sockeye (to the Bristol Bay region for example) canyield as much as 54 x 107 kilograms (kg) of biomass whichequates to 24 x 104 kg of P 18 x 105 kg of N 27 x 105 kg ofCa plus other macroelements Fish commonly migrate uplarge rivers and disperse into tributaries Thus the density offish in tiny coastal streams with small numbers of spawnersmay equal or exceed that in tributaries of major rivers that hostlarger absolute numbers of fish

The number of fish entering a system also varies tempo-rally at several scales Over past centuries salmon abundancehas varied dramatically in relation to geological changes in theland and changes in ocean conditions (Francis and Hare1994) Ocean conditions that influence primary productiv-ity in the North Pacific oscillate over a multidecadal cycle(Mantua et al 1997) and probably affect the growth and sur-vival rates of salmon (Gargett 1997) Natural variations alsooccur over shorter time intervals of a few years in responseto differences in precipitation and stream flow diseases andpopulation feedback mechanisms Since the onset of indus-trial fishing in the 1800s the number of fish returning to a sys-tem is also heavily contingent upon harvest levels (Finney etal 2000) with some stocks suffering 90 mortality and oth-ers with 5 or no mortality (Templin et al 1996)

Distribution Historically streams on both sides of theNorth Atlantic (including the Mediterranean) and the NorthPacific supported strong runs of anadromous salmon Climateand landscape changes through geological time wroughthabitat changes that periodically extinguished local popula-tions but residual populations recolonized habitable streamsor established populations in newly accessible rivers More re-cently anthropogenic changes have destroyed many of the runs(Lichatowich 1999) and decreased spawning area by chang-ing stream habitats As a result these changes reduced the in-flux of salmon to streams and by reducing the source pop-ulations diminished the chances of recolonization

In natural conditions the enormous load of salmon nu-trients (C N P etc) is distributed upstream as far as suitable

habitat is accessible Thus stream systems serve as conduits forthe input of ocean-derived materials to freshwater and ter-restrial systems In a large watershed salmon enter smaller andsmaller tributaries until they are dispersed throughout the wa-tershed sometimes into tiny headwater streams far into theinterior (figure 1a) For example fish entering the ColumbiaRiver historically dispersed as far as Redfish Lake Idaho a mi-gration of over 1000 km with an elevation gain of over 2000meters (m) (Groot and Margolis 1991 Gross et al 1998)Smaller coastal watersheds usually receive fewer fish butthere are thousands of smaller streams throughout the land-scape (figure 2) Although most nutrients are deposited nearthe stream most mobile consumers are close to salmonstreams For example the Tongass National Forest encom-passing almost all of southeastern Alaska contains nearly5000 salmon-supporting streams (Halupka et al 2000) con-sequently 47 of the forested area within the Tongass fallswithin 05 km of a salmon stream and over 90 within 5 km(Willson et al forthcoming) The influx of anadromous fisheffectively extends the interface between ocean and landthereby expanding the surface area over which ecologicalexchanges take place

Once salmon arrive at spawning streams their nutrients arespread still more widely overthe landscape by the activitiesof terrestrial consumers andwater movements For ex-ample bears congregate atstreams to catch salmon andoften drag the carcasses intothe riparian forest where theyare partially consumed(Gende et al 2001a) Bearsforaging at streams in BritishColumbia move 58 to 90of all salmon biomass to landsometimes hundreds of me-ters from the stream (Reim-chen 2000) and further dis-tribute the minerals andnutrients in the form of urineand feces as they movethroughout the riparian andupland forests (Hilderbrandet al 1999a) Stream insectsfeeding on salmon carcassesoften have aerial adult phasesduring which they can fly farfrom the natal streamsAvianscavengers remove chunks ofsalmon tissue and carry themonto land and also leave theirexcretory products across thelandscape In addition theporous gravels and mobilechannel beds characteristic

of spawning streams provide areas beneath and beside the sur-face channel where water flows back into the subsurface sat-urated zone (ie the hyporheic zone Edwards 1999) Prod-ucts of salmon decomposition move with this watertransferring large amounts of salmon-derived N and P to ad-jacent riparian zones 70 m or more from the spawning streamchannel (OrsquoKeefe and Edwards forthcoming)

After spawning an unknown proportion of the salmon in-put is exported and remobilized as stream currents contin-ually carry carcasses and decomposition products back down-stream toward the ocean A fraction of the input is also lostas outmigrating smolts (eg Gross et al 1998 Lyle and Elliott1998) many of which die while at sea Therefore mechanismsof salmon-nutrient retention become important by retard-ing that loss and making nutrients available to other organ-isms over a longer time span For instance many plant speciesgrow rapidly in the spring and early summer whereas the bulkof salmon runs occur in late summer to early fall Withoutmechanisms to store nutrients over winter there would be lit-tle stimulation of total annual primary productivity bysalmon-derived inputs

Retention mechanisms vary with latitude climate ani-mal populations vegetation cover and stream geomorphol-

October 2002 Vol 52 No 10 BioScience 919

Articles

Figure 1 (a) Predevelopment pattern illustrating how migrating and spawning salmon effec-tively extend the influence of oceanic productivity to freshwater systems by migrating far intothe continental interior through a large river system As fish move through the mainstem tospawning grounds their biomass is widely distributed throughout the landscape and ecosys-tem nutrient inputs (per square meter) may actually increase toward the low-order (head-water) reaches (b) Truncation of marine-derived nutrient distribution caused by dams andhabitat destruction

a b

ogy In forested regions fallen trees in streams create physi-cal barriers that retain carcasses (Cederholm and Peterson1985) and pools where carcasses accrue and decompose or be-come buried in the stream substrate Direct consumption bypredators and scavengers also stores the biomass as con-sumer tissue Within northern streams freezing may also bea significant retention mechanism by locking carcasses in

the ice and snow where they can be an important food sourcefor scavengers during winter or the following spring (egHansen 1987)

Biofilms on sediment surfaces are another potential site forstorage of salmon-derived nutrients Inorganic forms of N andP and dissolved organic matter (DOM) are rapidly taken upinto the matrix of algae bacteria fungi protozoans andnonliving organic matter that make up biofilms (Freeman andLock 1995) DOM leaching from salmon tissue is rapidlysorbed onto stream sediments (Bilby et al 1996) Salmon-derived ammonium (NH4

+) and phosphorous moving intoa hyporheic zone within a stream in southwestern Alaskawere removed within the first few meters of subsurface flow(OrsquoKeefe and Edwards forthcoming) presumably withinbiofilms on the sediment surface Storage within biofilmsfor weeks or months followed by mineralization and rein-troduction into the flowstream are potentially importantmechanisms by which marine-derived N and P could becomeavailable to surface algae during the following growth season

Hyporheic flows extending several hundred meters into ri-parian floodplain forests have been documented in salmonstreams which creates an enormous potential storage volume(Clinton et al 2002) Hyporheic zones contain much greaterepilithic surface area than surface benthos (Edwards 1999) andexist largely below the flood-scour depth Thus hyporheic stor-age is probably a large although poorly quantified storage areaof salmon-derived nutrients

In small coastal streams many carcasses may be flushedback into the ocean or spawning may occur in the intertidalzone Estuarine algae can take up the salmon-derived nutri-ents however thereby feeding copepods that are in turn fedupon by juvenile salmon all of which serves as a positive feed-back mechanism for salmon production (Fujiwara and High-smith 1997)

Dispersal pathwaysAlthough it is common to refer to salmon-derived nutrientsas if they were a uniform pool the ecosystem effects of ma-terials derived from salmon vary greatly with their chemicalform relative to various consumerrsquos needs Confusion over theldquoimportancerdquo of salmon to lakes streams and forests hasarisen in part because of the failure to distinguish the twobroad types of pathways by which salmon tissue is incorpo-rated into terrestrial and freshwater ecosystems (1) direct con-sumption of salmon as food by which ldquoinputrdquo passes up thefood chain in fairly predictable steps and (2) recycling of theproducts of decomposition leaching and excretion whichmove through a variety of less well-studied pathways Figure3 illustrates those consumption and recycling pathwaysschematically over the consumption and decay sequences Thevertical axis represents time starting with the entry of salmoninto spawning areas (early)progressing to more spawning andresidence time on beds (mid) continuing to the end of spawn-ing when dead and dying fish dominate (late) and finally end-ing with the postspawning period when remaining carcassesare processed The progression of the fish along the time

920 BioScience October 2002 Vol 52 No 10

Articles

Figure 2 Penetration of salmon-bearing coastal streamsinto the continental margin within a section of TongassNational Forest in southeastern Alaska Althoughstreams are shorter and runs are smaller (in absolutenumbers) than under historic conditions in large interiorrivers the high drainage density ensures that marine in-puts are distributed widely within much of the coastalforested regions Southeast Alaska has over 5000 salmonstreams initial estimates have found that over 90 ofthis forested area falls within 5 kilometers (km) of asalmon stream Inset Representation of the spatial pat-tern of reduction of marine inputs to the terrestrialecosystem caused by different anthropogenic effects (a) intact stream ecosystem (b) normal salmon escape-ment but impaired transfer caused by elimination ofbears or other predators or reduction in hydrologic cou-pling (c) intact transfer mechanisms but reduced run sizecaused by obstructions habitat degradation overfishingand so on (d) loss of transfer mechanisms and run reduc-tions and (e) run extirpated with or without impairmentof transfer mechanisms

a

d

b

c

e

SoutheastAlaska

series varies with species location and physical factors suchas hydrology

The left side of the diagram represents recycling pathwaysthat are dominated by excretion and decomposition and me-diated by invertebrates fungi bacteria and physical processesThese processes embody what are commonly termed ldquobottom-uprdquo effects influencing ecosystem processes via plant orbiofilm production The right side of the diagram representsconsumption pathways by which salmon biomass is incor-porated into trophic webs directly via feeding Salmon nu-trients can enter the food webs at many trophic levels becauseof the omnivorous nature of many stream and terrestrialbiota Consequently the right side cannot be easily charac-terized by a directional flow of salmon biomass (eg top-downor bottom-up) confounding any simplistic view that these sys-tems are regulated by one or the other (see also Power 1992)

The exact nature of the salmon-derived material enteringthe food web varies at different stages in the decompositioncycle of the salmon For example on the recycling side of the

diagram the first inorganic nutrient supplied to stream wa-ter by salmon is NH4

+ excreted by living fish before spawn-ing mortality begins (OrsquoKeefe and Edwards forthcoming) Aweek or two later levels of ammonium and soluble reactivephosphorus (SRP) in stream water further increase proba-bly leached from carcasses or gametes released during spawn-ing activitiesAs salmon continue to die and biomass from thecarcasses is processed by consumer or microbial activity SRPand NH4

+ increase but then decrease as the number of fishin the stream declines (eg Brickell and Goering 1970 Sugaiand Burrell 1984) Finally when only the skeletal tissue re-mains P and Ca in the bones are the primary nutrients leftOn the consumption side of the diagram predators such asbear or otter (or other vertebrates large enough to capture liveripe adult salmon) feed on lipid-rich living fish in the earlystages of the spawning cycle but as spawning progressesfeeding by scavengers on eggs carcasses and ldquoleftoversrdquo in-creases As the spawning run progresses and most fish havedepleted much of their energy the average energetic ldquore-

October 2002 Vol 52 No 10 BioScience 921

Articles

Figure 3 Schematic of major dispersal pathways for salmon-derived materials during the course of spawning On the left isthe ldquorecyclingrdquo pathway and on the right is the ldquodirect consumptionrdquo pathway Boxes beneath the derived products of feedingpathways suggest changes in the importance of various components as spawning progresses and tissue chemistry and con-sumers change Relative proportions are not to scale they are simply suggestive of trends

wardrdquo of tissue consumption decreases (Gende 2002) In-sects and fungi may take increasing amounts of salmon tis-sue as the number of carcasses increase (eg Reimchen 2000)

The distinction between the two dispersal pathways is par-ticularly important with reference to the techniques commonlyused to infer the importance of salmon-derived inputs atpopulation and ecosystem levels For example a common ap-proach has been to use stable isotope signatures to quantifytransfer of salmon nutrients to various consumers and biofilms(eg Kline et al 1997) The large difference in the heavy iso-tope (15N and 13C) composition of salmon tissue relative tofreshwater or terrestrial values has been used to estimate theproportion of salmon-derived N or C in animal tissues in-vertebrates and biofilms Nitrogen flow within the trophicstructure as indicated by 15N composition of consumers isassumed by some to also provide information about flows ofsalmon-derived P When organisms eat salmon tissue (con-sumption pathway) the ratio of C N and P in fish tissue maybe relatively well preserved making stable isotopes a usefultracer method In contrast the original marine elementalsignature is not preserved in the excretion and decompositionpathways because C N and P are physically and chemicallydecoupled and subsequently processed by widely divergentbiogeochemical processes For example N is subject to sev-eral microbially mediated processes (nitrification denitrifi-cation etc) that can dramatically alter its absolute concen-tration and isotopic composition (Kline et al 1997) Incontrast P (which has no stable isotopes) is not subject to lossby conversion to gas but is strongly sorbed to inorganic min-erals or precipitated out of solution under some conditionsTherefore the use of stable isotopes to infer the magnitudeof transfers within processes represented on the left side of fig-ure 3 although increasingly used is poorly documented andhighly speculative compared with the consumption path-way By extension it should not be assumed that the impor-tance of salmon biomass as food is directly correlated with theimportance of inorganic nutrients to bottom-up pathways

Given such widely divergent pathways products and con-sequences the term marine-derived nutrients is so impreciseas to be useless except when referring to the general phe-nomenon of the large influx of marine-originated biomass (inthis case salmon but see also Polis and Hurd 1996) It is im-portant that terms distinguish the specific salmon productbecause the importance of different salmon contributions rel-ative to other input sources varies the mechanisms control-ling their uptake and retention are distinctly different and thetype of nutrient that is limiting will vary Improved under-standing would be promoted by more specific terms such asldquosalmon-derived nitrogenrdquo or ldquosalmon-derived lipidsrdquo

Ecological consequences of the inputTo assess the biological importance of salmon-derived nu-trients we must know the magnitude composition and vari-ability of the input as well as the specific attributes of the wa-tershed receiving them For example within streams thepotential importance of salmon-derived N versus P in sup-

porting primary production varies with the magnitude ofother sources Historically streams in the Pacific Northwestwere considered nitrogen limited because of low nitrogen in-puts and the dominance of phosphorus-rich bedrock How-ever more recent research has shown that inorganic N con-centrations in streams vary widely from 10 to 20 micrograms(microg) N per liter to over 1600 microg N per liter (OrsquoKeefe and Edwards forthcoming) a concentration at which N would notbe expected to limit photosynthesis What is more high ni-trate concentrations are often associated with alder a nitrogen-fixing tree common in the Pacific Northwest the distributionof which has been expanded by long-term climatic changes(Hu et al 2001) and logging practices (Ruth and Harris1979) Hence the importance of salmon-derived N may beless than commonly assumed and varies with natural vege-tation patterns and human management activities

In contrast P concentrations tend to be uniformly low ex-cept in areas with P-rich sedimentary bedrock (Ashley andSlaney 1997) and recent work has highlighted P as the dom-inant limiting nutrient (Bothwell 1989 Ashley and Slaney1997) Nutrient patterns in Idaho streams suggest that one-quarter to one-half are nutrient limited and that half of thoseare P limited (Thomas et al forthcoming) Salmon molar NPratios range from 121 to 151 (Ashley and Slaney 1997)making them relatively phosphorus rich Within Lynx CreekAlaska phosphorus that is imported by spawning sockeyesalmon at average run sizes constitutes a large proportion ofthe P available to epilithon (organic matter attached to rocksurfaces) on an annual basis (OrsquoKeefe and Edwards forth-coming) and may be the most important marine product ofthe recycling pathway In some systems however light ratherthan nutrients limits primary production Thus a pulse ofsalmon-derived nutrients may have little or no effect on pri-mary productivity (Rand et al 1992) although salmon maystill be an important resource for stream or terrestrial biotavia the consumption pathway

Given the heterogeneity in habitats and limiting factors itfollows that the ecological consequences of inputs vary amonghabitats and with dispersal pathways Epilithic chlorophyllstanding stocks increased following salmon spawning insome studies (eg Richey et al 1975 Wipfli et al 1998) wereunaffected in other studies (Minshall et al 1991) and decreasedafter spawning episodes in still other streams (Minikawa1997) Increases in salmon-derived inorganic nitrogen andphosphorus compounds have been documented from severalstreams (Brickell and Goering 1970 Schuldt and Hershey1995 Minikawa 1997) confirming that excretion and min-eralization of carcasses does increase the inorganic nutrientcapital of some streams Furthermore artificially increasinginputs of inorganic P and N in streams in British Columbiaincreased chlorophyll accrual rates benthic insect density andthe growth rates and size of fish (Perrin et al 1987 Johnstonet al 1990 Mundie et al 1991) Therefore by inference it isassumed that nutrients released by spawning salmon alsohave the same effect Increased ecosystem primary produc-tivity in streams as a result of salmon nutrient inputs remains

922 BioScience October 2002 Vol 52 No 10

Articles

an interesting hypothesis that has not been confirmed how-ever particularly across a broad range of stream types andspawner densities

Increased lake productivity that is caused by salmon-nu-trient inputs is better documented Returning sockeye salmoncan contribute a large proportion of available P and N de-pending upon the size of the salmon run (Hartman andBurgner 1972 Stockner and Shortreed 1975 Mathisen et al1988) which may elevate phytoplankton and zooplanktondensities and increase juvenile salmon production (Narver1967) As in stream systems experimental nutrient inputs tolakes primarily P increased lake productivity (Hyatt andStockner 1985 Stockner and MacIsaac 1996) in contrast nu-trient budgets within Redfish Lake Idaho suggest that stim-ulation of lake production by spawning runs of salmon hasalways been small thus indicating that lake responses also varywith geography (Gross et al 1998)

There is abundant evidence that salmon availability influ-ences population dynamics of consumers via the consump-tion pathway More carcasses generally translate into higherdensities and elevated growth rates of invertebrates and ju-venile salmonids may grow faster by directly consumingsalmon tissue or consuming invertebrates that have beenscavenging salmon carcasses (Johnson and Ringler 1979Bilby et al 1996 1998 Wipfli et al 1998 1999 Chaloner andWipfli 2002) Furthermore juvenile coho salmon had higherlevels of whole-body lipids and a higher proportion of lipidallocated to energy reserves (triacylglycerol) when reared inthe presence of salmon carcasses (Ron Heintz National Ma-rine Fisheries Service Auke Bay AK personal communica-tion 22 September 2002) which may lead to higher survival(Wipfli et al forthcoming) Additionally marine isotopes ofN and C are higher in several trophic levels in salmon versusnonsalmon streams (Kline et al 1997) illustrating that thesefreshwater biota are sequestering marine nutrients into bodytissues presumably by direct consumption of salmon tissue(Bilby et al 1996)

Similar responses may occur in terrestrial communitiesLower trophic levels such as invertebrate scavengers (egdipterans) utilize the available salmon biomass (eg Reim-chen 2000) and reproduce thereby increasing local densitiesby conversion of salmon biomass into invertebrate tissue Den-sities of many vertebrates increase locally presumably bymoving from surrounding areas to feed on salmon Duringthe breeding season insectivorous riparian passerines arefound in greater densities on salmon streams than on otherstreams suggesting that bird communities may be respond-ing to the pulse of invertebrates produced by the availabilityof salmon (Gende and Willson 2001) Across the landscapethe carrying capacity of bears increases where salmon are avail-able with populations up to 80 times denser in coastal areaswhere salmon are abundant than in interior areas (Miller etal 1997 Hilderbrand et al 1999b) Fitness-related variablesincluding growth rates litter sizes and reproductive successhave been attributed to salmon availability for salmon con-sumers such as eagles bears and mustelids which suggests

that salmon play an important role in the population dynamicsof these terrestrial consumers (Hansen 1987 Ben-David1997 Hilderbrand et al 1999c)

Terrestrial vegetation also may respond to the presence ofsalmon The 15N signal presumed to be salmon-derived hasbeen detected in riparian shrubs and trees up to 500 m or morefrom streams (Bilby et al 1996 Ben-David et al 1998 Hilder-brand et al 1999a) in Washington and Alaska Marine sig-natures are higher in areas where bears feed on salmon (Ben-David et al 1998 Hilderbrand et al 1999a) which suggeststhat the foraging activities of bears play an important role inmaking salmon-derived nutrients available to terrestrial veg-etation There is also some indication that growth of ripar-ian trees may increase where salmon-derived nutrient inputoccurs particularly in areas with bear foraging activity(Helfield and Naiman 2001) although the spatial extent of thisphenomenon is unknown Increased growth of riparian veg-etation caused by salmon inputs if it occurs widely could haveramifications for riparian systems by changing litter largewoody debris and the amount of light reaching streambedsas well as by altering terrestrial vegetation structure (Helfieldand Naiman 2001) food for herbivorous insects and coverfor nesting birds

In addition to the direct effects of salmon subsidies thereare several indirect ecological ramifications of these subsidiesWe note three possible examples (1) Salmon are a majorsource of food for bears but bears also consume fleshy fruitsand thus serve as important seed dispersal agents for nu-merous plant species in coastal forests (eg Willson 1993)Without salmon bear densities would be lower and seed dis-persal patterns could be altered with unknown consequences(2) Fertilization of plants commonly leads to higher nutrientcontent and enhanced growth and some herbivorous in-sects attack fertilized plants at high rates (Price 1991) Birdsfeed on many herbivorous insects and in some circum-stances are capable of reducing the herbivore load on plantsthus fostering better plant growth (Marquis and Whelan1994) Higher densities of insectivorous breeding birds alongsalmon streams in spring (Gende and Willson 2001) mightmean that natural control of herbivorous insects is better insalmon-subsidized forests (3) Because salmon subsidies canlead to enhanced growth and survival of stream-resident fish(Bilby et al 1998) life-history strategies that are dependenton juvenile growth rates may change For instance the tim-ing and even the probability of migration from fresh water tothe sea may vary with juvenile growth rates (Healy and Heard1984) which in turn affects body size patterns of spawningcompetition and fecundity with ramifications for populationproductivity and thus for consumers and commercial harvests

Conservation and managementIt is clear from the growing body of literature that salmon mayinfluence the food webs trophic structure nutrient budgetsand possibly the productivity of freshwater and terrestrial sys-tems although the effect varies widely between systems andis contingent upon timing scale retention mechanisms al-

October 2002 Vol 52 No 10 BioScience 923

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

918 BioScience October 2002 Vol 52 No 10

Articles

We conclude by suggesting new research directions to help fillthe gaps in our current understanding of salmonid ecosystems

The salmon input We focus on five species of Pacific salmon that spawn infreshwater systems of North America chinook (Oncorhynchustshawytscha) sockeye (O nerka) pink (O gorbuscha) chum(O keta) and coho (O kisutch) All five species share a gen-eral life history Adults return to freshwater usually in late sum-mer and fall where they cease feeding spawn and die Aftersome months young emerge from the gravel in early springand depending on the species spend up to 2 years in fresh-water habitats before migrating out to sea The fish remain atsea for 1 to 7 years feeding and gaining over 90 of their bio-mass before returning to fresh water to complete the cycle(Groot and Margolis 1991)

Composition When salmon enter fresh water they havestored nearly all of the energy necessary for upstream mi-gration and reproduction but the magnitude of resource re-serves varies greatly among populations often in accord withthe length of upstream migration (Brett 1995 Hendry andBerg 1999) For example chum salmon in the Yukon Rivermay spawn near the mouth or nearly 2000 kilometers (km)upriver The difference in chemical composition of ldquoup-riverrdquo versus ldquodownriverrdquo spawning salmon is so large that itwas suggested as a metric to separate the stocks that are har-vested in a mixed-stock fishery on the coast (Brett 1995)

As salmon migrate and approach spawning areas theirbody composition changes dramatically Both males and fe-males store much of their lipid throughout the visceral andsoma tissue particularly in the muscle skin and skeletal tis-sue (Hendry and Berg 1999 Gende 2002) Stored lipids fuelmigration to spawning grounds which results in loss of someof the energy to metabolic heat en route and produces car-bon dioxide and water as waste products Females use somestored lipid for production of eggs which can constitutemore than 20 of body mass whereas males tap both lipidand protein reserves to develop secondary sexual character-istics (Hendry and Berg 1999) Once on the spawning groundsfish use most of the remaining lipid to fuel spawning activi-ties such as excavating and defending redds (females) andfighting for access to females (males) Body tissue proteins alsoare used as an energy source resulting in endogenous nitro-gen excretion in the form of ammonia (and some urea) pri-marily across the gill membrane (Wood 1995) The length oftime fish live on the spawning grounds varies dependingupon the population (generally less than 3 weeks) Althougha large fraction of the lipid and protein has been metabolizedor deposited in the gravel as eggs carcasses may contain upto 16 protein and 35 kilojoules per gram wet mass of en-ergy (Hendry and Berg 1999 Gende 2002) Other forms ofN such as collagen in the skin may not be depleted and re-main within the carcass

Salmon also contain macroelements that is mineralsfound in large amounts (Robbins 1993) such as potassium

and calcium (Ca) Although salmon lose calcium phosphateduring migration which allows bone tissue to turn into car-tilage necessary for the formation of secondary sexual char-acteristics total body Ca and P do not decline precipitouslyduring freshwater migration and spawning (at least for pinkand chum salmon populations that spawn in small coastalstreams Gende 2002) Phosphorus stored chiefly in the bonemuscle and male gonads makes up less than 05 of the bodymass but may be an important input when considering thelarge numbers of fish entering streams and lakes (Donaldson1967) Few studies have quantified the macroelement contentof the fish but clearly it may play a role in the nutritional qual-ity of salmon for consumers for example some mineralsare necessary in small amounts for health and growth (Rob-bins 1993)

Magnitude The flux of salmon biomass entering freshwater from the ocean can be massive A large run of 20 mil-lion sockeye (to the Bristol Bay region for example) canyield as much as 54 x 107 kilograms (kg) of biomass whichequates to 24 x 104 kg of P 18 x 105 kg of N 27 x 105 kg ofCa plus other macroelements Fish commonly migrate uplarge rivers and disperse into tributaries Thus the density offish in tiny coastal streams with small numbers of spawnersmay equal or exceed that in tributaries of major rivers that hostlarger absolute numbers of fish

The number of fish entering a system also varies tempo-rally at several scales Over past centuries salmon abundancehas varied dramatically in relation to geological changes in theland and changes in ocean conditions (Francis and Hare1994) Ocean conditions that influence primary productiv-ity in the North Pacific oscillate over a multidecadal cycle(Mantua et al 1997) and probably affect the growth and sur-vival rates of salmon (Gargett 1997) Natural variations alsooccur over shorter time intervals of a few years in responseto differences in precipitation and stream flow diseases andpopulation feedback mechanisms Since the onset of indus-trial fishing in the 1800s the number of fish returning to a sys-tem is also heavily contingent upon harvest levels (Finney etal 2000) with some stocks suffering 90 mortality and oth-ers with 5 or no mortality (Templin et al 1996)

Distribution Historically streams on both sides of theNorth Atlantic (including the Mediterranean) and the NorthPacific supported strong runs of anadromous salmon Climateand landscape changes through geological time wroughthabitat changes that periodically extinguished local popula-tions but residual populations recolonized habitable streamsor established populations in newly accessible rivers More re-cently anthropogenic changes have destroyed many of the runs(Lichatowich 1999) and decreased spawning area by chang-ing stream habitats As a result these changes reduced the in-flux of salmon to streams and by reducing the source pop-ulations diminished the chances of recolonization

In natural conditions the enormous load of salmon nu-trients (C N P etc) is distributed upstream as far as suitable

habitat is accessible Thus stream systems serve as conduits forthe input of ocean-derived materials to freshwater and ter-restrial systems In a large watershed salmon enter smaller andsmaller tributaries until they are dispersed throughout the wa-tershed sometimes into tiny headwater streams far into theinterior (figure 1a) For example fish entering the ColumbiaRiver historically dispersed as far as Redfish Lake Idaho a mi-gration of over 1000 km with an elevation gain of over 2000meters (m) (Groot and Margolis 1991 Gross et al 1998)Smaller coastal watersheds usually receive fewer fish butthere are thousands of smaller streams throughout the land-scape (figure 2) Although most nutrients are deposited nearthe stream most mobile consumers are close to salmonstreams For example the Tongass National Forest encom-passing almost all of southeastern Alaska contains nearly5000 salmon-supporting streams (Halupka et al 2000) con-sequently 47 of the forested area within the Tongass fallswithin 05 km of a salmon stream and over 90 within 5 km(Willson et al forthcoming) The influx of anadromous fisheffectively extends the interface between ocean and landthereby expanding the surface area over which ecologicalexchanges take place

Once salmon arrive at spawning streams their nutrients arespread still more widely overthe landscape by the activitiesof terrestrial consumers andwater movements For ex-ample bears congregate atstreams to catch salmon andoften drag the carcasses intothe riparian forest where theyare partially consumed(Gende et al 2001a) Bearsforaging at streams in BritishColumbia move 58 to 90of all salmon biomass to landsometimes hundreds of me-ters from the stream (Reim-chen 2000) and further dis-tribute the minerals andnutrients in the form of urineand feces as they movethroughout the riparian andupland forests (Hilderbrandet al 1999a) Stream insectsfeeding on salmon carcassesoften have aerial adult phasesduring which they can fly farfrom the natal streamsAvianscavengers remove chunks ofsalmon tissue and carry themonto land and also leave theirexcretory products across thelandscape In addition theporous gravels and mobilechannel beds characteristic

of spawning streams provide areas beneath and beside the sur-face channel where water flows back into the subsurface sat-urated zone (ie the hyporheic zone Edwards 1999) Prod-ucts of salmon decomposition move with this watertransferring large amounts of salmon-derived N and P to ad-jacent riparian zones 70 m or more from the spawning streamchannel (OrsquoKeefe and Edwards forthcoming)

After spawning an unknown proportion of the salmon in-put is exported and remobilized as stream currents contin-ually carry carcasses and decomposition products back down-stream toward the ocean A fraction of the input is also lostas outmigrating smolts (eg Gross et al 1998 Lyle and Elliott1998) many of which die while at sea Therefore mechanismsof salmon-nutrient retention become important by retard-ing that loss and making nutrients available to other organ-isms over a longer time span For instance many plant speciesgrow rapidly in the spring and early summer whereas the bulkof salmon runs occur in late summer to early fall Withoutmechanisms to store nutrients over winter there would be lit-tle stimulation of total annual primary productivity bysalmon-derived inputs

Retention mechanisms vary with latitude climate ani-mal populations vegetation cover and stream geomorphol-

October 2002 Vol 52 No 10 BioScience 919

Articles

Figure 1 (a) Predevelopment pattern illustrating how migrating and spawning salmon effec-tively extend the influence of oceanic productivity to freshwater systems by migrating far intothe continental interior through a large river system As fish move through the mainstem tospawning grounds their biomass is widely distributed throughout the landscape and ecosys-tem nutrient inputs (per square meter) may actually increase toward the low-order (head-water) reaches (b) Truncation of marine-derived nutrient distribution caused by dams andhabitat destruction

a b

ogy In forested regions fallen trees in streams create physi-cal barriers that retain carcasses (Cederholm and Peterson1985) and pools where carcasses accrue and decompose or be-come buried in the stream substrate Direct consumption bypredators and scavengers also stores the biomass as con-sumer tissue Within northern streams freezing may also bea significant retention mechanism by locking carcasses in

the ice and snow where they can be an important food sourcefor scavengers during winter or the following spring (egHansen 1987)

Biofilms on sediment surfaces are another potential site forstorage of salmon-derived nutrients Inorganic forms of N andP and dissolved organic matter (DOM) are rapidly taken upinto the matrix of algae bacteria fungi protozoans andnonliving organic matter that make up biofilms (Freeman andLock 1995) DOM leaching from salmon tissue is rapidlysorbed onto stream sediments (Bilby et al 1996) Salmon-derived ammonium (NH4

+) and phosphorous moving intoa hyporheic zone within a stream in southwestern Alaskawere removed within the first few meters of subsurface flow(OrsquoKeefe and Edwards forthcoming) presumably withinbiofilms on the sediment surface Storage within biofilmsfor weeks or months followed by mineralization and rein-troduction into the flowstream are potentially importantmechanisms by which marine-derived N and P could becomeavailable to surface algae during the following growth season

Hyporheic flows extending several hundred meters into ri-parian floodplain forests have been documented in salmonstreams which creates an enormous potential storage volume(Clinton et al 2002) Hyporheic zones contain much greaterepilithic surface area than surface benthos (Edwards 1999) andexist largely below the flood-scour depth Thus hyporheic stor-age is probably a large although poorly quantified storage areaof salmon-derived nutrients

In small coastal streams many carcasses may be flushedback into the ocean or spawning may occur in the intertidalzone Estuarine algae can take up the salmon-derived nutri-ents however thereby feeding copepods that are in turn fedupon by juvenile salmon all of which serves as a positive feed-back mechanism for salmon production (Fujiwara and High-smith 1997)

Dispersal pathwaysAlthough it is common to refer to salmon-derived nutrientsas if they were a uniform pool the ecosystem effects of ma-terials derived from salmon vary greatly with their chemicalform relative to various consumerrsquos needs Confusion over theldquoimportancerdquo of salmon to lakes streams and forests hasarisen in part because of the failure to distinguish the twobroad types of pathways by which salmon tissue is incorpo-rated into terrestrial and freshwater ecosystems (1) direct con-sumption of salmon as food by which ldquoinputrdquo passes up thefood chain in fairly predictable steps and (2) recycling of theproducts of decomposition leaching and excretion whichmove through a variety of less well-studied pathways Figure3 illustrates those consumption and recycling pathwaysschematically over the consumption and decay sequences Thevertical axis represents time starting with the entry of salmoninto spawning areas (early)progressing to more spawning andresidence time on beds (mid) continuing to the end of spawn-ing when dead and dying fish dominate (late) and finally end-ing with the postspawning period when remaining carcassesare processed The progression of the fish along the time

920 BioScience October 2002 Vol 52 No 10

Articles

Figure 2 Penetration of salmon-bearing coastal streamsinto the continental margin within a section of TongassNational Forest in southeastern Alaska Althoughstreams are shorter and runs are smaller (in absolutenumbers) than under historic conditions in large interiorrivers the high drainage density ensures that marine in-puts are distributed widely within much of the coastalforested regions Southeast Alaska has over 5000 salmonstreams initial estimates have found that over 90 ofthis forested area falls within 5 kilometers (km) of asalmon stream Inset Representation of the spatial pat-tern of reduction of marine inputs to the terrestrialecosystem caused by different anthropogenic effects (a) intact stream ecosystem (b) normal salmon escape-ment but impaired transfer caused by elimination ofbears or other predators or reduction in hydrologic cou-pling (c) intact transfer mechanisms but reduced run sizecaused by obstructions habitat degradation overfishingand so on (d) loss of transfer mechanisms and run reduc-tions and (e) run extirpated with or without impairmentof transfer mechanisms

a

d

b

c

e

SoutheastAlaska

series varies with species location and physical factors suchas hydrology

The left side of the diagram represents recycling pathwaysthat are dominated by excretion and decomposition and me-diated by invertebrates fungi bacteria and physical processesThese processes embody what are commonly termed ldquobottom-uprdquo effects influencing ecosystem processes via plant orbiofilm production The right side of the diagram representsconsumption pathways by which salmon biomass is incor-porated into trophic webs directly via feeding Salmon nu-trients can enter the food webs at many trophic levels becauseof the omnivorous nature of many stream and terrestrialbiota Consequently the right side cannot be easily charac-terized by a directional flow of salmon biomass (eg top-downor bottom-up) confounding any simplistic view that these sys-tems are regulated by one or the other (see also Power 1992)

The exact nature of the salmon-derived material enteringthe food web varies at different stages in the decompositioncycle of the salmon For example on the recycling side of the

diagram the first inorganic nutrient supplied to stream wa-ter by salmon is NH4

+ excreted by living fish before spawn-ing mortality begins (OrsquoKeefe and Edwards forthcoming) Aweek or two later levels of ammonium and soluble reactivephosphorus (SRP) in stream water further increase proba-bly leached from carcasses or gametes released during spawn-ing activitiesAs salmon continue to die and biomass from thecarcasses is processed by consumer or microbial activity SRPand NH4

+ increase but then decrease as the number of fishin the stream declines (eg Brickell and Goering 1970 Sugaiand Burrell 1984) Finally when only the skeletal tissue re-mains P and Ca in the bones are the primary nutrients leftOn the consumption side of the diagram predators such asbear or otter (or other vertebrates large enough to capture liveripe adult salmon) feed on lipid-rich living fish in the earlystages of the spawning cycle but as spawning progressesfeeding by scavengers on eggs carcasses and ldquoleftoversrdquo in-creases As the spawning run progresses and most fish havedepleted much of their energy the average energetic ldquore-

October 2002 Vol 52 No 10 BioScience 921

Articles

Figure 3 Schematic of major dispersal pathways for salmon-derived materials during the course of spawning On the left isthe ldquorecyclingrdquo pathway and on the right is the ldquodirect consumptionrdquo pathway Boxes beneath the derived products of feedingpathways suggest changes in the importance of various components as spawning progresses and tissue chemistry and con-sumers change Relative proportions are not to scale they are simply suggestive of trends

wardrdquo of tissue consumption decreases (Gende 2002) In-sects and fungi may take increasing amounts of salmon tis-sue as the number of carcasses increase (eg Reimchen 2000)

The distinction between the two dispersal pathways is par-ticularly important with reference to the techniques commonlyused to infer the importance of salmon-derived inputs atpopulation and ecosystem levels For example a common ap-proach has been to use stable isotope signatures to quantifytransfer of salmon nutrients to various consumers and biofilms(eg Kline et al 1997) The large difference in the heavy iso-tope (15N and 13C) composition of salmon tissue relative tofreshwater or terrestrial values has been used to estimate theproportion of salmon-derived N or C in animal tissues in-vertebrates and biofilms Nitrogen flow within the trophicstructure as indicated by 15N composition of consumers isassumed by some to also provide information about flows ofsalmon-derived P When organisms eat salmon tissue (con-sumption pathway) the ratio of C N and P in fish tissue maybe relatively well preserved making stable isotopes a usefultracer method In contrast the original marine elementalsignature is not preserved in the excretion and decompositionpathways because C N and P are physically and chemicallydecoupled and subsequently processed by widely divergentbiogeochemical processes For example N is subject to sev-eral microbially mediated processes (nitrification denitrifi-cation etc) that can dramatically alter its absolute concen-tration and isotopic composition (Kline et al 1997) Incontrast P (which has no stable isotopes) is not subject to lossby conversion to gas but is strongly sorbed to inorganic min-erals or precipitated out of solution under some conditionsTherefore the use of stable isotopes to infer the magnitudeof transfers within processes represented on the left side of fig-ure 3 although increasingly used is poorly documented andhighly speculative compared with the consumption path-way By extension it should not be assumed that the impor-tance of salmon biomass as food is directly correlated with theimportance of inorganic nutrients to bottom-up pathways

Given such widely divergent pathways products and con-sequences the term marine-derived nutrients is so impreciseas to be useless except when referring to the general phe-nomenon of the large influx of marine-originated biomass (inthis case salmon but see also Polis and Hurd 1996) It is im-portant that terms distinguish the specific salmon productbecause the importance of different salmon contributions rel-ative to other input sources varies the mechanisms control-ling their uptake and retention are distinctly different and thetype of nutrient that is limiting will vary Improved under-standing would be promoted by more specific terms such asldquosalmon-derived nitrogenrdquo or ldquosalmon-derived lipidsrdquo

Ecological consequences of the inputTo assess the biological importance of salmon-derived nu-trients we must know the magnitude composition and vari-ability of the input as well as the specific attributes of the wa-tershed receiving them For example within streams thepotential importance of salmon-derived N versus P in sup-

porting primary production varies with the magnitude ofother sources Historically streams in the Pacific Northwestwere considered nitrogen limited because of low nitrogen in-puts and the dominance of phosphorus-rich bedrock How-ever more recent research has shown that inorganic N con-centrations in streams vary widely from 10 to 20 micrograms(microg) N per liter to over 1600 microg N per liter (OrsquoKeefe and Edwards forthcoming) a concentration at which N would notbe expected to limit photosynthesis What is more high ni-trate concentrations are often associated with alder a nitrogen-fixing tree common in the Pacific Northwest the distributionof which has been expanded by long-term climatic changes(Hu et al 2001) and logging practices (Ruth and Harris1979) Hence the importance of salmon-derived N may beless than commonly assumed and varies with natural vege-tation patterns and human management activities

In contrast P concentrations tend to be uniformly low ex-cept in areas with P-rich sedimentary bedrock (Ashley andSlaney 1997) and recent work has highlighted P as the dom-inant limiting nutrient (Bothwell 1989 Ashley and Slaney1997) Nutrient patterns in Idaho streams suggest that one-quarter to one-half are nutrient limited and that half of thoseare P limited (Thomas et al forthcoming) Salmon molar NPratios range from 121 to 151 (Ashley and Slaney 1997)making them relatively phosphorus rich Within Lynx CreekAlaska phosphorus that is imported by spawning sockeyesalmon at average run sizes constitutes a large proportion ofthe P available to epilithon (organic matter attached to rocksurfaces) on an annual basis (OrsquoKeefe and Edwards forth-coming) and may be the most important marine product ofthe recycling pathway In some systems however light ratherthan nutrients limits primary production Thus a pulse ofsalmon-derived nutrients may have little or no effect on pri-mary productivity (Rand et al 1992) although salmon maystill be an important resource for stream or terrestrial biotavia the consumption pathway

Given the heterogeneity in habitats and limiting factors itfollows that the ecological consequences of inputs vary amonghabitats and with dispersal pathways Epilithic chlorophyllstanding stocks increased following salmon spawning insome studies (eg Richey et al 1975 Wipfli et al 1998) wereunaffected in other studies (Minshall et al 1991) and decreasedafter spawning episodes in still other streams (Minikawa1997) Increases in salmon-derived inorganic nitrogen andphosphorus compounds have been documented from severalstreams (Brickell and Goering 1970 Schuldt and Hershey1995 Minikawa 1997) confirming that excretion and min-eralization of carcasses does increase the inorganic nutrientcapital of some streams Furthermore artificially increasinginputs of inorganic P and N in streams in British Columbiaincreased chlorophyll accrual rates benthic insect density andthe growth rates and size of fish (Perrin et al 1987 Johnstonet al 1990 Mundie et al 1991) Therefore by inference it isassumed that nutrients released by spawning salmon alsohave the same effect Increased ecosystem primary produc-tivity in streams as a result of salmon nutrient inputs remains

922 BioScience October 2002 Vol 52 No 10

Articles

an interesting hypothesis that has not been confirmed how-ever particularly across a broad range of stream types andspawner densities

Increased lake productivity that is caused by salmon-nu-trient inputs is better documented Returning sockeye salmoncan contribute a large proportion of available P and N de-pending upon the size of the salmon run (Hartman andBurgner 1972 Stockner and Shortreed 1975 Mathisen et al1988) which may elevate phytoplankton and zooplanktondensities and increase juvenile salmon production (Narver1967) As in stream systems experimental nutrient inputs tolakes primarily P increased lake productivity (Hyatt andStockner 1985 Stockner and MacIsaac 1996) in contrast nu-trient budgets within Redfish Lake Idaho suggest that stim-ulation of lake production by spawning runs of salmon hasalways been small thus indicating that lake responses also varywith geography (Gross et al 1998)

There is abundant evidence that salmon availability influ-ences population dynamics of consumers via the consump-tion pathway More carcasses generally translate into higherdensities and elevated growth rates of invertebrates and ju-venile salmonids may grow faster by directly consumingsalmon tissue or consuming invertebrates that have beenscavenging salmon carcasses (Johnson and Ringler 1979Bilby et al 1996 1998 Wipfli et al 1998 1999 Chaloner andWipfli 2002) Furthermore juvenile coho salmon had higherlevels of whole-body lipids and a higher proportion of lipidallocated to energy reserves (triacylglycerol) when reared inthe presence of salmon carcasses (Ron Heintz National Ma-rine Fisheries Service Auke Bay AK personal communica-tion 22 September 2002) which may lead to higher survival(Wipfli et al forthcoming) Additionally marine isotopes ofN and C are higher in several trophic levels in salmon versusnonsalmon streams (Kline et al 1997) illustrating that thesefreshwater biota are sequestering marine nutrients into bodytissues presumably by direct consumption of salmon tissue(Bilby et al 1996)

Similar responses may occur in terrestrial communitiesLower trophic levels such as invertebrate scavengers (egdipterans) utilize the available salmon biomass (eg Reim-chen 2000) and reproduce thereby increasing local densitiesby conversion of salmon biomass into invertebrate tissue Den-sities of many vertebrates increase locally presumably bymoving from surrounding areas to feed on salmon Duringthe breeding season insectivorous riparian passerines arefound in greater densities on salmon streams than on otherstreams suggesting that bird communities may be respond-ing to the pulse of invertebrates produced by the availabilityof salmon (Gende and Willson 2001) Across the landscapethe carrying capacity of bears increases where salmon are avail-able with populations up to 80 times denser in coastal areaswhere salmon are abundant than in interior areas (Miller etal 1997 Hilderbrand et al 1999b) Fitness-related variablesincluding growth rates litter sizes and reproductive successhave been attributed to salmon availability for salmon con-sumers such as eagles bears and mustelids which suggests

that salmon play an important role in the population dynamicsof these terrestrial consumers (Hansen 1987 Ben-David1997 Hilderbrand et al 1999c)

Terrestrial vegetation also may respond to the presence ofsalmon The 15N signal presumed to be salmon-derived hasbeen detected in riparian shrubs and trees up to 500 m or morefrom streams (Bilby et al 1996 Ben-David et al 1998 Hilder-brand et al 1999a) in Washington and Alaska Marine sig-natures are higher in areas where bears feed on salmon (Ben-David et al 1998 Hilderbrand et al 1999a) which suggeststhat the foraging activities of bears play an important role inmaking salmon-derived nutrients available to terrestrial veg-etation There is also some indication that growth of ripar-ian trees may increase where salmon-derived nutrient inputoccurs particularly in areas with bear foraging activity(Helfield and Naiman 2001) although the spatial extent of thisphenomenon is unknown Increased growth of riparian veg-etation caused by salmon inputs if it occurs widely could haveramifications for riparian systems by changing litter largewoody debris and the amount of light reaching streambedsas well as by altering terrestrial vegetation structure (Helfieldand Naiman 2001) food for herbivorous insects and coverfor nesting birds

In addition to the direct effects of salmon subsidies thereare several indirect ecological ramifications of these subsidiesWe note three possible examples (1) Salmon are a majorsource of food for bears but bears also consume fleshy fruitsand thus serve as important seed dispersal agents for nu-merous plant species in coastal forests (eg Willson 1993)Without salmon bear densities would be lower and seed dis-persal patterns could be altered with unknown consequences(2) Fertilization of plants commonly leads to higher nutrientcontent and enhanced growth and some herbivorous in-sects attack fertilized plants at high rates (Price 1991) Birdsfeed on many herbivorous insects and in some circum-stances are capable of reducing the herbivore load on plantsthus fostering better plant growth (Marquis and Whelan1994) Higher densities of insectivorous breeding birds alongsalmon streams in spring (Gende and Willson 2001) mightmean that natural control of herbivorous insects is better insalmon-subsidized forests (3) Because salmon subsidies canlead to enhanced growth and survival of stream-resident fish(Bilby et al 1998) life-history strategies that are dependenton juvenile growth rates may change For instance the tim-ing and even the probability of migration from fresh water tothe sea may vary with juvenile growth rates (Healy and Heard1984) which in turn affects body size patterns of spawningcompetition and fecundity with ramifications for populationproductivity and thus for consumers and commercial harvests

Conservation and managementIt is clear from the growing body of literature that salmon mayinfluence the food webs trophic structure nutrient budgetsand possibly the productivity of freshwater and terrestrial sys-tems although the effect varies widely between systems andis contingent upon timing scale retention mechanisms al-

October 2002 Vol 52 No 10 BioScience 923

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

habitat is accessible Thus stream systems serve as conduits forthe input of ocean-derived materials to freshwater and ter-restrial systems In a large watershed salmon enter smaller andsmaller tributaries until they are dispersed throughout the wa-tershed sometimes into tiny headwater streams far into theinterior (figure 1a) For example fish entering the ColumbiaRiver historically dispersed as far as Redfish Lake Idaho a mi-gration of over 1000 km with an elevation gain of over 2000meters (m) (Groot and Margolis 1991 Gross et al 1998)Smaller coastal watersheds usually receive fewer fish butthere are thousands of smaller streams throughout the land-scape (figure 2) Although most nutrients are deposited nearthe stream most mobile consumers are close to salmonstreams For example the Tongass National Forest encom-passing almost all of southeastern Alaska contains nearly5000 salmon-supporting streams (Halupka et al 2000) con-sequently 47 of the forested area within the Tongass fallswithin 05 km of a salmon stream and over 90 within 5 km(Willson et al forthcoming) The influx of anadromous fisheffectively extends the interface between ocean and landthereby expanding the surface area over which ecologicalexchanges take place

Once salmon arrive at spawning streams their nutrients arespread still more widely overthe landscape by the activitiesof terrestrial consumers andwater movements For ex-ample bears congregate atstreams to catch salmon andoften drag the carcasses intothe riparian forest where theyare partially consumed(Gende et al 2001a) Bearsforaging at streams in BritishColumbia move 58 to 90of all salmon biomass to landsometimes hundreds of me-ters from the stream (Reim-chen 2000) and further dis-tribute the minerals andnutrients in the form of urineand feces as they movethroughout the riparian andupland forests (Hilderbrandet al 1999a) Stream insectsfeeding on salmon carcassesoften have aerial adult phasesduring which they can fly farfrom the natal streamsAvianscavengers remove chunks ofsalmon tissue and carry themonto land and also leave theirexcretory products across thelandscape In addition theporous gravels and mobilechannel beds characteristic

of spawning streams provide areas beneath and beside the sur-face channel where water flows back into the subsurface sat-urated zone (ie the hyporheic zone Edwards 1999) Prod-ucts of salmon decomposition move with this watertransferring large amounts of salmon-derived N and P to ad-jacent riparian zones 70 m or more from the spawning streamchannel (OrsquoKeefe and Edwards forthcoming)

After spawning an unknown proportion of the salmon in-put is exported and remobilized as stream currents contin-ually carry carcasses and decomposition products back down-stream toward the ocean A fraction of the input is also lostas outmigrating smolts (eg Gross et al 1998 Lyle and Elliott1998) many of which die while at sea Therefore mechanismsof salmon-nutrient retention become important by retard-ing that loss and making nutrients available to other organ-isms over a longer time span For instance many plant speciesgrow rapidly in the spring and early summer whereas the bulkof salmon runs occur in late summer to early fall Withoutmechanisms to store nutrients over winter there would be lit-tle stimulation of total annual primary productivity bysalmon-derived inputs

Retention mechanisms vary with latitude climate ani-mal populations vegetation cover and stream geomorphol-

October 2002 Vol 52 No 10 BioScience 919

Articles

Figure 1 (a) Predevelopment pattern illustrating how migrating and spawning salmon effec-tively extend the influence of oceanic productivity to freshwater systems by migrating far intothe continental interior through a large river system As fish move through the mainstem tospawning grounds their biomass is widely distributed throughout the landscape and ecosys-tem nutrient inputs (per square meter) may actually increase toward the low-order (head-water) reaches (b) Truncation of marine-derived nutrient distribution caused by dams andhabitat destruction

a b

ogy In forested regions fallen trees in streams create physi-cal barriers that retain carcasses (Cederholm and Peterson1985) and pools where carcasses accrue and decompose or be-come buried in the stream substrate Direct consumption bypredators and scavengers also stores the biomass as con-sumer tissue Within northern streams freezing may also bea significant retention mechanism by locking carcasses in

the ice and snow where they can be an important food sourcefor scavengers during winter or the following spring (egHansen 1987)

Biofilms on sediment surfaces are another potential site forstorage of salmon-derived nutrients Inorganic forms of N andP and dissolved organic matter (DOM) are rapidly taken upinto the matrix of algae bacteria fungi protozoans andnonliving organic matter that make up biofilms (Freeman andLock 1995) DOM leaching from salmon tissue is rapidlysorbed onto stream sediments (Bilby et al 1996) Salmon-derived ammonium (NH4

+) and phosphorous moving intoa hyporheic zone within a stream in southwestern Alaskawere removed within the first few meters of subsurface flow(OrsquoKeefe and Edwards forthcoming) presumably withinbiofilms on the sediment surface Storage within biofilmsfor weeks or months followed by mineralization and rein-troduction into the flowstream are potentially importantmechanisms by which marine-derived N and P could becomeavailable to surface algae during the following growth season

Hyporheic flows extending several hundred meters into ri-parian floodplain forests have been documented in salmonstreams which creates an enormous potential storage volume(Clinton et al 2002) Hyporheic zones contain much greaterepilithic surface area than surface benthos (Edwards 1999) andexist largely below the flood-scour depth Thus hyporheic stor-age is probably a large although poorly quantified storage areaof salmon-derived nutrients

In small coastal streams many carcasses may be flushedback into the ocean or spawning may occur in the intertidalzone Estuarine algae can take up the salmon-derived nutri-ents however thereby feeding copepods that are in turn fedupon by juvenile salmon all of which serves as a positive feed-back mechanism for salmon production (Fujiwara and High-smith 1997)

Dispersal pathwaysAlthough it is common to refer to salmon-derived nutrientsas if they were a uniform pool the ecosystem effects of ma-terials derived from salmon vary greatly with their chemicalform relative to various consumerrsquos needs Confusion over theldquoimportancerdquo of salmon to lakes streams and forests hasarisen in part because of the failure to distinguish the twobroad types of pathways by which salmon tissue is incorpo-rated into terrestrial and freshwater ecosystems (1) direct con-sumption of salmon as food by which ldquoinputrdquo passes up thefood chain in fairly predictable steps and (2) recycling of theproducts of decomposition leaching and excretion whichmove through a variety of less well-studied pathways Figure3 illustrates those consumption and recycling pathwaysschematically over the consumption and decay sequences Thevertical axis represents time starting with the entry of salmoninto spawning areas (early)progressing to more spawning andresidence time on beds (mid) continuing to the end of spawn-ing when dead and dying fish dominate (late) and finally end-ing with the postspawning period when remaining carcassesare processed The progression of the fish along the time

920 BioScience October 2002 Vol 52 No 10

Articles

Figure 2 Penetration of salmon-bearing coastal streamsinto the continental margin within a section of TongassNational Forest in southeastern Alaska Althoughstreams are shorter and runs are smaller (in absolutenumbers) than under historic conditions in large interiorrivers the high drainage density ensures that marine in-puts are distributed widely within much of the coastalforested regions Southeast Alaska has over 5000 salmonstreams initial estimates have found that over 90 ofthis forested area falls within 5 kilometers (km) of asalmon stream Inset Representation of the spatial pat-tern of reduction of marine inputs to the terrestrialecosystem caused by different anthropogenic effects (a) intact stream ecosystem (b) normal salmon escape-ment but impaired transfer caused by elimination ofbears or other predators or reduction in hydrologic cou-pling (c) intact transfer mechanisms but reduced run sizecaused by obstructions habitat degradation overfishingand so on (d) loss of transfer mechanisms and run reduc-tions and (e) run extirpated with or without impairmentof transfer mechanisms

a

d

b

c

e

SoutheastAlaska

series varies with species location and physical factors suchas hydrology

The left side of the diagram represents recycling pathwaysthat are dominated by excretion and decomposition and me-diated by invertebrates fungi bacteria and physical processesThese processes embody what are commonly termed ldquobottom-uprdquo effects influencing ecosystem processes via plant orbiofilm production The right side of the diagram representsconsumption pathways by which salmon biomass is incor-porated into trophic webs directly via feeding Salmon nu-trients can enter the food webs at many trophic levels becauseof the omnivorous nature of many stream and terrestrialbiota Consequently the right side cannot be easily charac-terized by a directional flow of salmon biomass (eg top-downor bottom-up) confounding any simplistic view that these sys-tems are regulated by one or the other (see also Power 1992)

The exact nature of the salmon-derived material enteringthe food web varies at different stages in the decompositioncycle of the salmon For example on the recycling side of the

diagram the first inorganic nutrient supplied to stream wa-ter by salmon is NH4

+ excreted by living fish before spawn-ing mortality begins (OrsquoKeefe and Edwards forthcoming) Aweek or two later levels of ammonium and soluble reactivephosphorus (SRP) in stream water further increase proba-bly leached from carcasses or gametes released during spawn-ing activitiesAs salmon continue to die and biomass from thecarcasses is processed by consumer or microbial activity SRPand NH4

+ increase but then decrease as the number of fishin the stream declines (eg Brickell and Goering 1970 Sugaiand Burrell 1984) Finally when only the skeletal tissue re-mains P and Ca in the bones are the primary nutrients leftOn the consumption side of the diagram predators such asbear or otter (or other vertebrates large enough to capture liveripe adult salmon) feed on lipid-rich living fish in the earlystages of the spawning cycle but as spawning progressesfeeding by scavengers on eggs carcasses and ldquoleftoversrdquo in-creases As the spawning run progresses and most fish havedepleted much of their energy the average energetic ldquore-

October 2002 Vol 52 No 10 BioScience 921

Articles

Figure 3 Schematic of major dispersal pathways for salmon-derived materials during the course of spawning On the left isthe ldquorecyclingrdquo pathway and on the right is the ldquodirect consumptionrdquo pathway Boxes beneath the derived products of feedingpathways suggest changes in the importance of various components as spawning progresses and tissue chemistry and con-sumers change Relative proportions are not to scale they are simply suggestive of trends

wardrdquo of tissue consumption decreases (Gende 2002) In-sects and fungi may take increasing amounts of salmon tis-sue as the number of carcasses increase (eg Reimchen 2000)

The distinction between the two dispersal pathways is par-ticularly important with reference to the techniques commonlyused to infer the importance of salmon-derived inputs atpopulation and ecosystem levels For example a common ap-proach has been to use stable isotope signatures to quantifytransfer of salmon nutrients to various consumers and biofilms(eg Kline et al 1997) The large difference in the heavy iso-tope (15N and 13C) composition of salmon tissue relative tofreshwater or terrestrial values has been used to estimate theproportion of salmon-derived N or C in animal tissues in-vertebrates and biofilms Nitrogen flow within the trophicstructure as indicated by 15N composition of consumers isassumed by some to also provide information about flows ofsalmon-derived P When organisms eat salmon tissue (con-sumption pathway) the ratio of C N and P in fish tissue maybe relatively well preserved making stable isotopes a usefultracer method In contrast the original marine elementalsignature is not preserved in the excretion and decompositionpathways because C N and P are physically and chemicallydecoupled and subsequently processed by widely divergentbiogeochemical processes For example N is subject to sev-eral microbially mediated processes (nitrification denitrifi-cation etc) that can dramatically alter its absolute concen-tration and isotopic composition (Kline et al 1997) Incontrast P (which has no stable isotopes) is not subject to lossby conversion to gas but is strongly sorbed to inorganic min-erals or precipitated out of solution under some conditionsTherefore the use of stable isotopes to infer the magnitudeof transfers within processes represented on the left side of fig-ure 3 although increasingly used is poorly documented andhighly speculative compared with the consumption path-way By extension it should not be assumed that the impor-tance of salmon biomass as food is directly correlated with theimportance of inorganic nutrients to bottom-up pathways

Given such widely divergent pathways products and con-sequences the term marine-derived nutrients is so impreciseas to be useless except when referring to the general phe-nomenon of the large influx of marine-originated biomass (inthis case salmon but see also Polis and Hurd 1996) It is im-portant that terms distinguish the specific salmon productbecause the importance of different salmon contributions rel-ative to other input sources varies the mechanisms control-ling their uptake and retention are distinctly different and thetype of nutrient that is limiting will vary Improved under-standing would be promoted by more specific terms such asldquosalmon-derived nitrogenrdquo or ldquosalmon-derived lipidsrdquo

Ecological consequences of the inputTo assess the biological importance of salmon-derived nu-trients we must know the magnitude composition and vari-ability of the input as well as the specific attributes of the wa-tershed receiving them For example within streams thepotential importance of salmon-derived N versus P in sup-

porting primary production varies with the magnitude ofother sources Historically streams in the Pacific Northwestwere considered nitrogen limited because of low nitrogen in-puts and the dominance of phosphorus-rich bedrock How-ever more recent research has shown that inorganic N con-centrations in streams vary widely from 10 to 20 micrograms(microg) N per liter to over 1600 microg N per liter (OrsquoKeefe and Edwards forthcoming) a concentration at which N would notbe expected to limit photosynthesis What is more high ni-trate concentrations are often associated with alder a nitrogen-fixing tree common in the Pacific Northwest the distributionof which has been expanded by long-term climatic changes(Hu et al 2001) and logging practices (Ruth and Harris1979) Hence the importance of salmon-derived N may beless than commonly assumed and varies with natural vege-tation patterns and human management activities

In contrast P concentrations tend to be uniformly low ex-cept in areas with P-rich sedimentary bedrock (Ashley andSlaney 1997) and recent work has highlighted P as the dom-inant limiting nutrient (Bothwell 1989 Ashley and Slaney1997) Nutrient patterns in Idaho streams suggest that one-quarter to one-half are nutrient limited and that half of thoseare P limited (Thomas et al forthcoming) Salmon molar NPratios range from 121 to 151 (Ashley and Slaney 1997)making them relatively phosphorus rich Within Lynx CreekAlaska phosphorus that is imported by spawning sockeyesalmon at average run sizes constitutes a large proportion ofthe P available to epilithon (organic matter attached to rocksurfaces) on an annual basis (OrsquoKeefe and Edwards forth-coming) and may be the most important marine product ofthe recycling pathway In some systems however light ratherthan nutrients limits primary production Thus a pulse ofsalmon-derived nutrients may have little or no effect on pri-mary productivity (Rand et al 1992) although salmon maystill be an important resource for stream or terrestrial biotavia the consumption pathway

Given the heterogeneity in habitats and limiting factors itfollows that the ecological consequences of inputs vary amonghabitats and with dispersal pathways Epilithic chlorophyllstanding stocks increased following salmon spawning insome studies (eg Richey et al 1975 Wipfli et al 1998) wereunaffected in other studies (Minshall et al 1991) and decreasedafter spawning episodes in still other streams (Minikawa1997) Increases in salmon-derived inorganic nitrogen andphosphorus compounds have been documented from severalstreams (Brickell and Goering 1970 Schuldt and Hershey1995 Minikawa 1997) confirming that excretion and min-eralization of carcasses does increase the inorganic nutrientcapital of some streams Furthermore artificially increasinginputs of inorganic P and N in streams in British Columbiaincreased chlorophyll accrual rates benthic insect density andthe growth rates and size of fish (Perrin et al 1987 Johnstonet al 1990 Mundie et al 1991) Therefore by inference it isassumed that nutrients released by spawning salmon alsohave the same effect Increased ecosystem primary produc-tivity in streams as a result of salmon nutrient inputs remains

922 BioScience October 2002 Vol 52 No 10

Articles

an interesting hypothesis that has not been confirmed how-ever particularly across a broad range of stream types andspawner densities

Increased lake productivity that is caused by salmon-nu-trient inputs is better documented Returning sockeye salmoncan contribute a large proportion of available P and N de-pending upon the size of the salmon run (Hartman andBurgner 1972 Stockner and Shortreed 1975 Mathisen et al1988) which may elevate phytoplankton and zooplanktondensities and increase juvenile salmon production (Narver1967) As in stream systems experimental nutrient inputs tolakes primarily P increased lake productivity (Hyatt andStockner 1985 Stockner and MacIsaac 1996) in contrast nu-trient budgets within Redfish Lake Idaho suggest that stim-ulation of lake production by spawning runs of salmon hasalways been small thus indicating that lake responses also varywith geography (Gross et al 1998)

There is abundant evidence that salmon availability influ-ences population dynamics of consumers via the consump-tion pathway More carcasses generally translate into higherdensities and elevated growth rates of invertebrates and ju-venile salmonids may grow faster by directly consumingsalmon tissue or consuming invertebrates that have beenscavenging salmon carcasses (Johnson and Ringler 1979Bilby et al 1996 1998 Wipfli et al 1998 1999 Chaloner andWipfli 2002) Furthermore juvenile coho salmon had higherlevels of whole-body lipids and a higher proportion of lipidallocated to energy reserves (triacylglycerol) when reared inthe presence of salmon carcasses (Ron Heintz National Ma-rine Fisheries Service Auke Bay AK personal communica-tion 22 September 2002) which may lead to higher survival(Wipfli et al forthcoming) Additionally marine isotopes ofN and C are higher in several trophic levels in salmon versusnonsalmon streams (Kline et al 1997) illustrating that thesefreshwater biota are sequestering marine nutrients into bodytissues presumably by direct consumption of salmon tissue(Bilby et al 1996)

Similar responses may occur in terrestrial communitiesLower trophic levels such as invertebrate scavengers (egdipterans) utilize the available salmon biomass (eg Reim-chen 2000) and reproduce thereby increasing local densitiesby conversion of salmon biomass into invertebrate tissue Den-sities of many vertebrates increase locally presumably bymoving from surrounding areas to feed on salmon Duringthe breeding season insectivorous riparian passerines arefound in greater densities on salmon streams than on otherstreams suggesting that bird communities may be respond-ing to the pulse of invertebrates produced by the availabilityof salmon (Gende and Willson 2001) Across the landscapethe carrying capacity of bears increases where salmon are avail-able with populations up to 80 times denser in coastal areaswhere salmon are abundant than in interior areas (Miller etal 1997 Hilderbrand et al 1999b) Fitness-related variablesincluding growth rates litter sizes and reproductive successhave been attributed to salmon availability for salmon con-sumers such as eagles bears and mustelids which suggests

that salmon play an important role in the population dynamicsof these terrestrial consumers (Hansen 1987 Ben-David1997 Hilderbrand et al 1999c)

Terrestrial vegetation also may respond to the presence ofsalmon The 15N signal presumed to be salmon-derived hasbeen detected in riparian shrubs and trees up to 500 m or morefrom streams (Bilby et al 1996 Ben-David et al 1998 Hilder-brand et al 1999a) in Washington and Alaska Marine sig-natures are higher in areas where bears feed on salmon (Ben-David et al 1998 Hilderbrand et al 1999a) which suggeststhat the foraging activities of bears play an important role inmaking salmon-derived nutrients available to terrestrial veg-etation There is also some indication that growth of ripar-ian trees may increase where salmon-derived nutrient inputoccurs particularly in areas with bear foraging activity(Helfield and Naiman 2001) although the spatial extent of thisphenomenon is unknown Increased growth of riparian veg-etation caused by salmon inputs if it occurs widely could haveramifications for riparian systems by changing litter largewoody debris and the amount of light reaching streambedsas well as by altering terrestrial vegetation structure (Helfieldand Naiman 2001) food for herbivorous insects and coverfor nesting birds

In addition to the direct effects of salmon subsidies thereare several indirect ecological ramifications of these subsidiesWe note three possible examples (1) Salmon are a majorsource of food for bears but bears also consume fleshy fruitsand thus serve as important seed dispersal agents for nu-merous plant species in coastal forests (eg Willson 1993)Without salmon bear densities would be lower and seed dis-persal patterns could be altered with unknown consequences(2) Fertilization of plants commonly leads to higher nutrientcontent and enhanced growth and some herbivorous in-sects attack fertilized plants at high rates (Price 1991) Birdsfeed on many herbivorous insects and in some circum-stances are capable of reducing the herbivore load on plantsthus fostering better plant growth (Marquis and Whelan1994) Higher densities of insectivorous breeding birds alongsalmon streams in spring (Gende and Willson 2001) mightmean that natural control of herbivorous insects is better insalmon-subsidized forests (3) Because salmon subsidies canlead to enhanced growth and survival of stream-resident fish(Bilby et al 1998) life-history strategies that are dependenton juvenile growth rates may change For instance the tim-ing and even the probability of migration from fresh water tothe sea may vary with juvenile growth rates (Healy and Heard1984) which in turn affects body size patterns of spawningcompetition and fecundity with ramifications for populationproductivity and thus for consumers and commercial harvests

Conservation and managementIt is clear from the growing body of literature that salmon mayinfluence the food webs trophic structure nutrient budgetsand possibly the productivity of freshwater and terrestrial sys-tems although the effect varies widely between systems andis contingent upon timing scale retention mechanisms al-

October 2002 Vol 52 No 10 BioScience 923

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

ogy In forested regions fallen trees in streams create physi-cal barriers that retain carcasses (Cederholm and Peterson1985) and pools where carcasses accrue and decompose or be-come buried in the stream substrate Direct consumption bypredators and scavengers also stores the biomass as con-sumer tissue Within northern streams freezing may also bea significant retention mechanism by locking carcasses in

the ice and snow where they can be an important food sourcefor scavengers during winter or the following spring (egHansen 1987)

Biofilms on sediment surfaces are another potential site forstorage of salmon-derived nutrients Inorganic forms of N andP and dissolved organic matter (DOM) are rapidly taken upinto the matrix of algae bacteria fungi protozoans andnonliving organic matter that make up biofilms (Freeman andLock 1995) DOM leaching from salmon tissue is rapidlysorbed onto stream sediments (Bilby et al 1996) Salmon-derived ammonium (NH4

+) and phosphorous moving intoa hyporheic zone within a stream in southwestern Alaskawere removed within the first few meters of subsurface flow(OrsquoKeefe and Edwards forthcoming) presumably withinbiofilms on the sediment surface Storage within biofilmsfor weeks or months followed by mineralization and rein-troduction into the flowstream are potentially importantmechanisms by which marine-derived N and P could becomeavailable to surface algae during the following growth season

Hyporheic flows extending several hundred meters into ri-parian floodplain forests have been documented in salmonstreams which creates an enormous potential storage volume(Clinton et al 2002) Hyporheic zones contain much greaterepilithic surface area than surface benthos (Edwards 1999) andexist largely below the flood-scour depth Thus hyporheic stor-age is probably a large although poorly quantified storage areaof salmon-derived nutrients

In small coastal streams many carcasses may be flushedback into the ocean or spawning may occur in the intertidalzone Estuarine algae can take up the salmon-derived nutri-ents however thereby feeding copepods that are in turn fedupon by juvenile salmon all of which serves as a positive feed-back mechanism for salmon production (Fujiwara and High-smith 1997)

Dispersal pathwaysAlthough it is common to refer to salmon-derived nutrientsas if they were a uniform pool the ecosystem effects of ma-terials derived from salmon vary greatly with their chemicalform relative to various consumerrsquos needs Confusion over theldquoimportancerdquo of salmon to lakes streams and forests hasarisen in part because of the failure to distinguish the twobroad types of pathways by which salmon tissue is incorpo-rated into terrestrial and freshwater ecosystems (1) direct con-sumption of salmon as food by which ldquoinputrdquo passes up thefood chain in fairly predictable steps and (2) recycling of theproducts of decomposition leaching and excretion whichmove through a variety of less well-studied pathways Figure3 illustrates those consumption and recycling pathwaysschematically over the consumption and decay sequences Thevertical axis represents time starting with the entry of salmoninto spawning areas (early)progressing to more spawning andresidence time on beds (mid) continuing to the end of spawn-ing when dead and dying fish dominate (late) and finally end-ing with the postspawning period when remaining carcassesare processed The progression of the fish along the time

920 BioScience October 2002 Vol 52 No 10

Articles

Figure 2 Penetration of salmon-bearing coastal streamsinto the continental margin within a section of TongassNational Forest in southeastern Alaska Althoughstreams are shorter and runs are smaller (in absolutenumbers) than under historic conditions in large interiorrivers the high drainage density ensures that marine in-puts are distributed widely within much of the coastalforested regions Southeast Alaska has over 5000 salmonstreams initial estimates have found that over 90 ofthis forested area falls within 5 kilometers (km) of asalmon stream Inset Representation of the spatial pat-tern of reduction of marine inputs to the terrestrialecosystem caused by different anthropogenic effects (a) intact stream ecosystem (b) normal salmon escape-ment but impaired transfer caused by elimination ofbears or other predators or reduction in hydrologic cou-pling (c) intact transfer mechanisms but reduced run sizecaused by obstructions habitat degradation overfishingand so on (d) loss of transfer mechanisms and run reduc-tions and (e) run extirpated with or without impairmentof transfer mechanisms

a

d

b

c

e

SoutheastAlaska

series varies with species location and physical factors suchas hydrology

The left side of the diagram represents recycling pathwaysthat are dominated by excretion and decomposition and me-diated by invertebrates fungi bacteria and physical processesThese processes embody what are commonly termed ldquobottom-uprdquo effects influencing ecosystem processes via plant orbiofilm production The right side of the diagram representsconsumption pathways by which salmon biomass is incor-porated into trophic webs directly via feeding Salmon nu-trients can enter the food webs at many trophic levels becauseof the omnivorous nature of many stream and terrestrialbiota Consequently the right side cannot be easily charac-terized by a directional flow of salmon biomass (eg top-downor bottom-up) confounding any simplistic view that these sys-tems are regulated by one or the other (see also Power 1992)

The exact nature of the salmon-derived material enteringthe food web varies at different stages in the decompositioncycle of the salmon For example on the recycling side of the

diagram the first inorganic nutrient supplied to stream wa-ter by salmon is NH4

+ excreted by living fish before spawn-ing mortality begins (OrsquoKeefe and Edwards forthcoming) Aweek or two later levels of ammonium and soluble reactivephosphorus (SRP) in stream water further increase proba-bly leached from carcasses or gametes released during spawn-ing activitiesAs salmon continue to die and biomass from thecarcasses is processed by consumer or microbial activity SRPand NH4

+ increase but then decrease as the number of fishin the stream declines (eg Brickell and Goering 1970 Sugaiand Burrell 1984) Finally when only the skeletal tissue re-mains P and Ca in the bones are the primary nutrients leftOn the consumption side of the diagram predators such asbear or otter (or other vertebrates large enough to capture liveripe adult salmon) feed on lipid-rich living fish in the earlystages of the spawning cycle but as spawning progressesfeeding by scavengers on eggs carcasses and ldquoleftoversrdquo in-creases As the spawning run progresses and most fish havedepleted much of their energy the average energetic ldquore-

October 2002 Vol 52 No 10 BioScience 921

Articles

Figure 3 Schematic of major dispersal pathways for salmon-derived materials during the course of spawning On the left isthe ldquorecyclingrdquo pathway and on the right is the ldquodirect consumptionrdquo pathway Boxes beneath the derived products of feedingpathways suggest changes in the importance of various components as spawning progresses and tissue chemistry and con-sumers change Relative proportions are not to scale they are simply suggestive of trends

wardrdquo of tissue consumption decreases (Gende 2002) In-sects and fungi may take increasing amounts of salmon tis-sue as the number of carcasses increase (eg Reimchen 2000)

The distinction between the two dispersal pathways is par-ticularly important with reference to the techniques commonlyused to infer the importance of salmon-derived inputs atpopulation and ecosystem levels For example a common ap-proach has been to use stable isotope signatures to quantifytransfer of salmon nutrients to various consumers and biofilms(eg Kline et al 1997) The large difference in the heavy iso-tope (15N and 13C) composition of salmon tissue relative tofreshwater or terrestrial values has been used to estimate theproportion of salmon-derived N or C in animal tissues in-vertebrates and biofilms Nitrogen flow within the trophicstructure as indicated by 15N composition of consumers isassumed by some to also provide information about flows ofsalmon-derived P When organisms eat salmon tissue (con-sumption pathway) the ratio of C N and P in fish tissue maybe relatively well preserved making stable isotopes a usefultracer method In contrast the original marine elementalsignature is not preserved in the excretion and decompositionpathways because C N and P are physically and chemicallydecoupled and subsequently processed by widely divergentbiogeochemical processes For example N is subject to sev-eral microbially mediated processes (nitrification denitrifi-cation etc) that can dramatically alter its absolute concen-tration and isotopic composition (Kline et al 1997) Incontrast P (which has no stable isotopes) is not subject to lossby conversion to gas but is strongly sorbed to inorganic min-erals or precipitated out of solution under some conditionsTherefore the use of stable isotopes to infer the magnitudeof transfers within processes represented on the left side of fig-ure 3 although increasingly used is poorly documented andhighly speculative compared with the consumption path-way By extension it should not be assumed that the impor-tance of salmon biomass as food is directly correlated with theimportance of inorganic nutrients to bottom-up pathways

Given such widely divergent pathways products and con-sequences the term marine-derived nutrients is so impreciseas to be useless except when referring to the general phe-nomenon of the large influx of marine-originated biomass (inthis case salmon but see also Polis and Hurd 1996) It is im-portant that terms distinguish the specific salmon productbecause the importance of different salmon contributions rel-ative to other input sources varies the mechanisms control-ling their uptake and retention are distinctly different and thetype of nutrient that is limiting will vary Improved under-standing would be promoted by more specific terms such asldquosalmon-derived nitrogenrdquo or ldquosalmon-derived lipidsrdquo

Ecological consequences of the inputTo assess the biological importance of salmon-derived nu-trients we must know the magnitude composition and vari-ability of the input as well as the specific attributes of the wa-tershed receiving them For example within streams thepotential importance of salmon-derived N versus P in sup-

porting primary production varies with the magnitude ofother sources Historically streams in the Pacific Northwestwere considered nitrogen limited because of low nitrogen in-puts and the dominance of phosphorus-rich bedrock How-ever more recent research has shown that inorganic N con-centrations in streams vary widely from 10 to 20 micrograms(microg) N per liter to over 1600 microg N per liter (OrsquoKeefe and Edwards forthcoming) a concentration at which N would notbe expected to limit photosynthesis What is more high ni-trate concentrations are often associated with alder a nitrogen-fixing tree common in the Pacific Northwest the distributionof which has been expanded by long-term climatic changes(Hu et al 2001) and logging practices (Ruth and Harris1979) Hence the importance of salmon-derived N may beless than commonly assumed and varies with natural vege-tation patterns and human management activities

In contrast P concentrations tend to be uniformly low ex-cept in areas with P-rich sedimentary bedrock (Ashley andSlaney 1997) and recent work has highlighted P as the dom-inant limiting nutrient (Bothwell 1989 Ashley and Slaney1997) Nutrient patterns in Idaho streams suggest that one-quarter to one-half are nutrient limited and that half of thoseare P limited (Thomas et al forthcoming) Salmon molar NPratios range from 121 to 151 (Ashley and Slaney 1997)making them relatively phosphorus rich Within Lynx CreekAlaska phosphorus that is imported by spawning sockeyesalmon at average run sizes constitutes a large proportion ofthe P available to epilithon (organic matter attached to rocksurfaces) on an annual basis (OrsquoKeefe and Edwards forth-coming) and may be the most important marine product ofthe recycling pathway In some systems however light ratherthan nutrients limits primary production Thus a pulse ofsalmon-derived nutrients may have little or no effect on pri-mary productivity (Rand et al 1992) although salmon maystill be an important resource for stream or terrestrial biotavia the consumption pathway

Given the heterogeneity in habitats and limiting factors itfollows that the ecological consequences of inputs vary amonghabitats and with dispersal pathways Epilithic chlorophyllstanding stocks increased following salmon spawning insome studies (eg Richey et al 1975 Wipfli et al 1998) wereunaffected in other studies (Minshall et al 1991) and decreasedafter spawning episodes in still other streams (Minikawa1997) Increases in salmon-derived inorganic nitrogen andphosphorus compounds have been documented from severalstreams (Brickell and Goering 1970 Schuldt and Hershey1995 Minikawa 1997) confirming that excretion and min-eralization of carcasses does increase the inorganic nutrientcapital of some streams Furthermore artificially increasinginputs of inorganic P and N in streams in British Columbiaincreased chlorophyll accrual rates benthic insect density andthe growth rates and size of fish (Perrin et al 1987 Johnstonet al 1990 Mundie et al 1991) Therefore by inference it isassumed that nutrients released by spawning salmon alsohave the same effect Increased ecosystem primary produc-tivity in streams as a result of salmon nutrient inputs remains

922 BioScience October 2002 Vol 52 No 10

Articles

an interesting hypothesis that has not been confirmed how-ever particularly across a broad range of stream types andspawner densities

Increased lake productivity that is caused by salmon-nu-trient inputs is better documented Returning sockeye salmoncan contribute a large proportion of available P and N de-pending upon the size of the salmon run (Hartman andBurgner 1972 Stockner and Shortreed 1975 Mathisen et al1988) which may elevate phytoplankton and zooplanktondensities and increase juvenile salmon production (Narver1967) As in stream systems experimental nutrient inputs tolakes primarily P increased lake productivity (Hyatt andStockner 1985 Stockner and MacIsaac 1996) in contrast nu-trient budgets within Redfish Lake Idaho suggest that stim-ulation of lake production by spawning runs of salmon hasalways been small thus indicating that lake responses also varywith geography (Gross et al 1998)

There is abundant evidence that salmon availability influ-ences population dynamics of consumers via the consump-tion pathway More carcasses generally translate into higherdensities and elevated growth rates of invertebrates and ju-venile salmonids may grow faster by directly consumingsalmon tissue or consuming invertebrates that have beenscavenging salmon carcasses (Johnson and Ringler 1979Bilby et al 1996 1998 Wipfli et al 1998 1999 Chaloner andWipfli 2002) Furthermore juvenile coho salmon had higherlevels of whole-body lipids and a higher proportion of lipidallocated to energy reserves (triacylglycerol) when reared inthe presence of salmon carcasses (Ron Heintz National Ma-rine Fisheries Service Auke Bay AK personal communica-tion 22 September 2002) which may lead to higher survival(Wipfli et al forthcoming) Additionally marine isotopes ofN and C are higher in several trophic levels in salmon versusnonsalmon streams (Kline et al 1997) illustrating that thesefreshwater biota are sequestering marine nutrients into bodytissues presumably by direct consumption of salmon tissue(Bilby et al 1996)

Similar responses may occur in terrestrial communitiesLower trophic levels such as invertebrate scavengers (egdipterans) utilize the available salmon biomass (eg Reim-chen 2000) and reproduce thereby increasing local densitiesby conversion of salmon biomass into invertebrate tissue Den-sities of many vertebrates increase locally presumably bymoving from surrounding areas to feed on salmon Duringthe breeding season insectivorous riparian passerines arefound in greater densities on salmon streams than on otherstreams suggesting that bird communities may be respond-ing to the pulse of invertebrates produced by the availabilityof salmon (Gende and Willson 2001) Across the landscapethe carrying capacity of bears increases where salmon are avail-able with populations up to 80 times denser in coastal areaswhere salmon are abundant than in interior areas (Miller etal 1997 Hilderbrand et al 1999b) Fitness-related variablesincluding growth rates litter sizes and reproductive successhave been attributed to salmon availability for salmon con-sumers such as eagles bears and mustelids which suggests

that salmon play an important role in the population dynamicsof these terrestrial consumers (Hansen 1987 Ben-David1997 Hilderbrand et al 1999c)

Terrestrial vegetation also may respond to the presence ofsalmon The 15N signal presumed to be salmon-derived hasbeen detected in riparian shrubs and trees up to 500 m or morefrom streams (Bilby et al 1996 Ben-David et al 1998 Hilder-brand et al 1999a) in Washington and Alaska Marine sig-natures are higher in areas where bears feed on salmon (Ben-David et al 1998 Hilderbrand et al 1999a) which suggeststhat the foraging activities of bears play an important role inmaking salmon-derived nutrients available to terrestrial veg-etation There is also some indication that growth of ripar-ian trees may increase where salmon-derived nutrient inputoccurs particularly in areas with bear foraging activity(Helfield and Naiman 2001) although the spatial extent of thisphenomenon is unknown Increased growth of riparian veg-etation caused by salmon inputs if it occurs widely could haveramifications for riparian systems by changing litter largewoody debris and the amount of light reaching streambedsas well as by altering terrestrial vegetation structure (Helfieldand Naiman 2001) food for herbivorous insects and coverfor nesting birds

In addition to the direct effects of salmon subsidies thereare several indirect ecological ramifications of these subsidiesWe note three possible examples (1) Salmon are a majorsource of food for bears but bears also consume fleshy fruitsand thus serve as important seed dispersal agents for nu-merous plant species in coastal forests (eg Willson 1993)Without salmon bear densities would be lower and seed dis-persal patterns could be altered with unknown consequences(2) Fertilization of plants commonly leads to higher nutrientcontent and enhanced growth and some herbivorous in-sects attack fertilized plants at high rates (Price 1991) Birdsfeed on many herbivorous insects and in some circum-stances are capable of reducing the herbivore load on plantsthus fostering better plant growth (Marquis and Whelan1994) Higher densities of insectivorous breeding birds alongsalmon streams in spring (Gende and Willson 2001) mightmean that natural control of herbivorous insects is better insalmon-subsidized forests (3) Because salmon subsidies canlead to enhanced growth and survival of stream-resident fish(Bilby et al 1998) life-history strategies that are dependenton juvenile growth rates may change For instance the tim-ing and even the probability of migration from fresh water tothe sea may vary with juvenile growth rates (Healy and Heard1984) which in turn affects body size patterns of spawningcompetition and fecundity with ramifications for populationproductivity and thus for consumers and commercial harvests

Conservation and managementIt is clear from the growing body of literature that salmon mayinfluence the food webs trophic structure nutrient budgetsand possibly the productivity of freshwater and terrestrial sys-tems although the effect varies widely between systems andis contingent upon timing scale retention mechanisms al-

October 2002 Vol 52 No 10 BioScience 923

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

series varies with species location and physical factors suchas hydrology

The left side of the diagram represents recycling pathwaysthat are dominated by excretion and decomposition and me-diated by invertebrates fungi bacteria and physical processesThese processes embody what are commonly termed ldquobottom-uprdquo effects influencing ecosystem processes via plant orbiofilm production The right side of the diagram representsconsumption pathways by which salmon biomass is incor-porated into trophic webs directly via feeding Salmon nu-trients can enter the food webs at many trophic levels becauseof the omnivorous nature of many stream and terrestrialbiota Consequently the right side cannot be easily charac-terized by a directional flow of salmon biomass (eg top-downor bottom-up) confounding any simplistic view that these sys-tems are regulated by one or the other (see also Power 1992)

The exact nature of the salmon-derived material enteringthe food web varies at different stages in the decompositioncycle of the salmon For example on the recycling side of the

diagram the first inorganic nutrient supplied to stream wa-ter by salmon is NH4

+ excreted by living fish before spawn-ing mortality begins (OrsquoKeefe and Edwards forthcoming) Aweek or two later levels of ammonium and soluble reactivephosphorus (SRP) in stream water further increase proba-bly leached from carcasses or gametes released during spawn-ing activitiesAs salmon continue to die and biomass from thecarcasses is processed by consumer or microbial activity SRPand NH4

+ increase but then decrease as the number of fishin the stream declines (eg Brickell and Goering 1970 Sugaiand Burrell 1984) Finally when only the skeletal tissue re-mains P and Ca in the bones are the primary nutrients leftOn the consumption side of the diagram predators such asbear or otter (or other vertebrates large enough to capture liveripe adult salmon) feed on lipid-rich living fish in the earlystages of the spawning cycle but as spawning progressesfeeding by scavengers on eggs carcasses and ldquoleftoversrdquo in-creases As the spawning run progresses and most fish havedepleted much of their energy the average energetic ldquore-

October 2002 Vol 52 No 10 BioScience 921

Articles

Figure 3 Schematic of major dispersal pathways for salmon-derived materials during the course of spawning On the left isthe ldquorecyclingrdquo pathway and on the right is the ldquodirect consumptionrdquo pathway Boxes beneath the derived products of feedingpathways suggest changes in the importance of various components as spawning progresses and tissue chemistry and con-sumers change Relative proportions are not to scale they are simply suggestive of trends

wardrdquo of tissue consumption decreases (Gende 2002) In-sects and fungi may take increasing amounts of salmon tis-sue as the number of carcasses increase (eg Reimchen 2000)

The distinction between the two dispersal pathways is par-ticularly important with reference to the techniques commonlyused to infer the importance of salmon-derived inputs atpopulation and ecosystem levels For example a common ap-proach has been to use stable isotope signatures to quantifytransfer of salmon nutrients to various consumers and biofilms(eg Kline et al 1997) The large difference in the heavy iso-tope (15N and 13C) composition of salmon tissue relative tofreshwater or terrestrial values has been used to estimate theproportion of salmon-derived N or C in animal tissues in-vertebrates and biofilms Nitrogen flow within the trophicstructure as indicated by 15N composition of consumers isassumed by some to also provide information about flows ofsalmon-derived P When organisms eat salmon tissue (con-sumption pathway) the ratio of C N and P in fish tissue maybe relatively well preserved making stable isotopes a usefultracer method In contrast the original marine elementalsignature is not preserved in the excretion and decompositionpathways because C N and P are physically and chemicallydecoupled and subsequently processed by widely divergentbiogeochemical processes For example N is subject to sev-eral microbially mediated processes (nitrification denitrifi-cation etc) that can dramatically alter its absolute concen-tration and isotopic composition (Kline et al 1997) Incontrast P (which has no stable isotopes) is not subject to lossby conversion to gas but is strongly sorbed to inorganic min-erals or precipitated out of solution under some conditionsTherefore the use of stable isotopes to infer the magnitudeof transfers within processes represented on the left side of fig-ure 3 although increasingly used is poorly documented andhighly speculative compared with the consumption path-way By extension it should not be assumed that the impor-tance of salmon biomass as food is directly correlated with theimportance of inorganic nutrients to bottom-up pathways

Given such widely divergent pathways products and con-sequences the term marine-derived nutrients is so impreciseas to be useless except when referring to the general phe-nomenon of the large influx of marine-originated biomass (inthis case salmon but see also Polis and Hurd 1996) It is im-portant that terms distinguish the specific salmon productbecause the importance of different salmon contributions rel-ative to other input sources varies the mechanisms control-ling their uptake and retention are distinctly different and thetype of nutrient that is limiting will vary Improved under-standing would be promoted by more specific terms such asldquosalmon-derived nitrogenrdquo or ldquosalmon-derived lipidsrdquo

Ecological consequences of the inputTo assess the biological importance of salmon-derived nu-trients we must know the magnitude composition and vari-ability of the input as well as the specific attributes of the wa-tershed receiving them For example within streams thepotential importance of salmon-derived N versus P in sup-

porting primary production varies with the magnitude ofother sources Historically streams in the Pacific Northwestwere considered nitrogen limited because of low nitrogen in-puts and the dominance of phosphorus-rich bedrock How-ever more recent research has shown that inorganic N con-centrations in streams vary widely from 10 to 20 micrograms(microg) N per liter to over 1600 microg N per liter (OrsquoKeefe and Edwards forthcoming) a concentration at which N would notbe expected to limit photosynthesis What is more high ni-trate concentrations are often associated with alder a nitrogen-fixing tree common in the Pacific Northwest the distributionof which has been expanded by long-term climatic changes(Hu et al 2001) and logging practices (Ruth and Harris1979) Hence the importance of salmon-derived N may beless than commonly assumed and varies with natural vege-tation patterns and human management activities

In contrast P concentrations tend to be uniformly low ex-cept in areas with P-rich sedimentary bedrock (Ashley andSlaney 1997) and recent work has highlighted P as the dom-inant limiting nutrient (Bothwell 1989 Ashley and Slaney1997) Nutrient patterns in Idaho streams suggest that one-quarter to one-half are nutrient limited and that half of thoseare P limited (Thomas et al forthcoming) Salmon molar NPratios range from 121 to 151 (Ashley and Slaney 1997)making them relatively phosphorus rich Within Lynx CreekAlaska phosphorus that is imported by spawning sockeyesalmon at average run sizes constitutes a large proportion ofthe P available to epilithon (organic matter attached to rocksurfaces) on an annual basis (OrsquoKeefe and Edwards forth-coming) and may be the most important marine product ofthe recycling pathway In some systems however light ratherthan nutrients limits primary production Thus a pulse ofsalmon-derived nutrients may have little or no effect on pri-mary productivity (Rand et al 1992) although salmon maystill be an important resource for stream or terrestrial biotavia the consumption pathway

Given the heterogeneity in habitats and limiting factors itfollows that the ecological consequences of inputs vary amonghabitats and with dispersal pathways Epilithic chlorophyllstanding stocks increased following salmon spawning insome studies (eg Richey et al 1975 Wipfli et al 1998) wereunaffected in other studies (Minshall et al 1991) and decreasedafter spawning episodes in still other streams (Minikawa1997) Increases in salmon-derived inorganic nitrogen andphosphorus compounds have been documented from severalstreams (Brickell and Goering 1970 Schuldt and Hershey1995 Minikawa 1997) confirming that excretion and min-eralization of carcasses does increase the inorganic nutrientcapital of some streams Furthermore artificially increasinginputs of inorganic P and N in streams in British Columbiaincreased chlorophyll accrual rates benthic insect density andthe growth rates and size of fish (Perrin et al 1987 Johnstonet al 1990 Mundie et al 1991) Therefore by inference it isassumed that nutrients released by spawning salmon alsohave the same effect Increased ecosystem primary produc-tivity in streams as a result of salmon nutrient inputs remains

922 BioScience October 2002 Vol 52 No 10

Articles

an interesting hypothesis that has not been confirmed how-ever particularly across a broad range of stream types andspawner densities

Increased lake productivity that is caused by salmon-nu-trient inputs is better documented Returning sockeye salmoncan contribute a large proportion of available P and N de-pending upon the size of the salmon run (Hartman andBurgner 1972 Stockner and Shortreed 1975 Mathisen et al1988) which may elevate phytoplankton and zooplanktondensities and increase juvenile salmon production (Narver1967) As in stream systems experimental nutrient inputs tolakes primarily P increased lake productivity (Hyatt andStockner 1985 Stockner and MacIsaac 1996) in contrast nu-trient budgets within Redfish Lake Idaho suggest that stim-ulation of lake production by spawning runs of salmon hasalways been small thus indicating that lake responses also varywith geography (Gross et al 1998)

There is abundant evidence that salmon availability influ-ences population dynamics of consumers via the consump-tion pathway More carcasses generally translate into higherdensities and elevated growth rates of invertebrates and ju-venile salmonids may grow faster by directly consumingsalmon tissue or consuming invertebrates that have beenscavenging salmon carcasses (Johnson and Ringler 1979Bilby et al 1996 1998 Wipfli et al 1998 1999 Chaloner andWipfli 2002) Furthermore juvenile coho salmon had higherlevels of whole-body lipids and a higher proportion of lipidallocated to energy reserves (triacylglycerol) when reared inthe presence of salmon carcasses (Ron Heintz National Ma-rine Fisheries Service Auke Bay AK personal communica-tion 22 September 2002) which may lead to higher survival(Wipfli et al forthcoming) Additionally marine isotopes ofN and C are higher in several trophic levels in salmon versusnonsalmon streams (Kline et al 1997) illustrating that thesefreshwater biota are sequestering marine nutrients into bodytissues presumably by direct consumption of salmon tissue(Bilby et al 1996)

Similar responses may occur in terrestrial communitiesLower trophic levels such as invertebrate scavengers (egdipterans) utilize the available salmon biomass (eg Reim-chen 2000) and reproduce thereby increasing local densitiesby conversion of salmon biomass into invertebrate tissue Den-sities of many vertebrates increase locally presumably bymoving from surrounding areas to feed on salmon Duringthe breeding season insectivorous riparian passerines arefound in greater densities on salmon streams than on otherstreams suggesting that bird communities may be respond-ing to the pulse of invertebrates produced by the availabilityof salmon (Gende and Willson 2001) Across the landscapethe carrying capacity of bears increases where salmon are avail-able with populations up to 80 times denser in coastal areaswhere salmon are abundant than in interior areas (Miller etal 1997 Hilderbrand et al 1999b) Fitness-related variablesincluding growth rates litter sizes and reproductive successhave been attributed to salmon availability for salmon con-sumers such as eagles bears and mustelids which suggests

that salmon play an important role in the population dynamicsof these terrestrial consumers (Hansen 1987 Ben-David1997 Hilderbrand et al 1999c)

Terrestrial vegetation also may respond to the presence ofsalmon The 15N signal presumed to be salmon-derived hasbeen detected in riparian shrubs and trees up to 500 m or morefrom streams (Bilby et al 1996 Ben-David et al 1998 Hilder-brand et al 1999a) in Washington and Alaska Marine sig-natures are higher in areas where bears feed on salmon (Ben-David et al 1998 Hilderbrand et al 1999a) which suggeststhat the foraging activities of bears play an important role inmaking salmon-derived nutrients available to terrestrial veg-etation There is also some indication that growth of ripar-ian trees may increase where salmon-derived nutrient inputoccurs particularly in areas with bear foraging activity(Helfield and Naiman 2001) although the spatial extent of thisphenomenon is unknown Increased growth of riparian veg-etation caused by salmon inputs if it occurs widely could haveramifications for riparian systems by changing litter largewoody debris and the amount of light reaching streambedsas well as by altering terrestrial vegetation structure (Helfieldand Naiman 2001) food for herbivorous insects and coverfor nesting birds

In addition to the direct effects of salmon subsidies thereare several indirect ecological ramifications of these subsidiesWe note three possible examples (1) Salmon are a majorsource of food for bears but bears also consume fleshy fruitsand thus serve as important seed dispersal agents for nu-merous plant species in coastal forests (eg Willson 1993)Without salmon bear densities would be lower and seed dis-persal patterns could be altered with unknown consequences(2) Fertilization of plants commonly leads to higher nutrientcontent and enhanced growth and some herbivorous in-sects attack fertilized plants at high rates (Price 1991) Birdsfeed on many herbivorous insects and in some circum-stances are capable of reducing the herbivore load on plantsthus fostering better plant growth (Marquis and Whelan1994) Higher densities of insectivorous breeding birds alongsalmon streams in spring (Gende and Willson 2001) mightmean that natural control of herbivorous insects is better insalmon-subsidized forests (3) Because salmon subsidies canlead to enhanced growth and survival of stream-resident fish(Bilby et al 1998) life-history strategies that are dependenton juvenile growth rates may change For instance the tim-ing and even the probability of migration from fresh water tothe sea may vary with juvenile growth rates (Healy and Heard1984) which in turn affects body size patterns of spawningcompetition and fecundity with ramifications for populationproductivity and thus for consumers and commercial harvests

Conservation and managementIt is clear from the growing body of literature that salmon mayinfluence the food webs trophic structure nutrient budgetsand possibly the productivity of freshwater and terrestrial sys-tems although the effect varies widely between systems andis contingent upon timing scale retention mechanisms al-

October 2002 Vol 52 No 10 BioScience 923

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

wardrdquo of tissue consumption decreases (Gende 2002) In-sects and fungi may take increasing amounts of salmon tis-sue as the number of carcasses increase (eg Reimchen 2000)

The distinction between the two dispersal pathways is par-ticularly important with reference to the techniques commonlyused to infer the importance of salmon-derived inputs atpopulation and ecosystem levels For example a common ap-proach has been to use stable isotope signatures to quantifytransfer of salmon nutrients to various consumers and biofilms(eg Kline et al 1997) The large difference in the heavy iso-tope (15N and 13C) composition of salmon tissue relative tofreshwater or terrestrial values has been used to estimate theproportion of salmon-derived N or C in animal tissues in-vertebrates and biofilms Nitrogen flow within the trophicstructure as indicated by 15N composition of consumers isassumed by some to also provide information about flows ofsalmon-derived P When organisms eat salmon tissue (con-sumption pathway) the ratio of C N and P in fish tissue maybe relatively well preserved making stable isotopes a usefultracer method In contrast the original marine elementalsignature is not preserved in the excretion and decompositionpathways because C N and P are physically and chemicallydecoupled and subsequently processed by widely divergentbiogeochemical processes For example N is subject to sev-eral microbially mediated processes (nitrification denitrifi-cation etc) that can dramatically alter its absolute concen-tration and isotopic composition (Kline et al 1997) Incontrast P (which has no stable isotopes) is not subject to lossby conversion to gas but is strongly sorbed to inorganic min-erals or precipitated out of solution under some conditionsTherefore the use of stable isotopes to infer the magnitudeof transfers within processes represented on the left side of fig-ure 3 although increasingly used is poorly documented andhighly speculative compared with the consumption path-way By extension it should not be assumed that the impor-tance of salmon biomass as food is directly correlated with theimportance of inorganic nutrients to bottom-up pathways

Given such widely divergent pathways products and con-sequences the term marine-derived nutrients is so impreciseas to be useless except when referring to the general phe-nomenon of the large influx of marine-originated biomass (inthis case salmon but see also Polis and Hurd 1996) It is im-portant that terms distinguish the specific salmon productbecause the importance of different salmon contributions rel-ative to other input sources varies the mechanisms control-ling their uptake and retention are distinctly different and thetype of nutrient that is limiting will vary Improved under-standing would be promoted by more specific terms such asldquosalmon-derived nitrogenrdquo or ldquosalmon-derived lipidsrdquo

Ecological consequences of the inputTo assess the biological importance of salmon-derived nu-trients we must know the magnitude composition and vari-ability of the input as well as the specific attributes of the wa-tershed receiving them For example within streams thepotential importance of salmon-derived N versus P in sup-

porting primary production varies with the magnitude ofother sources Historically streams in the Pacific Northwestwere considered nitrogen limited because of low nitrogen in-puts and the dominance of phosphorus-rich bedrock How-ever more recent research has shown that inorganic N con-centrations in streams vary widely from 10 to 20 micrograms(microg) N per liter to over 1600 microg N per liter (OrsquoKeefe and Edwards forthcoming) a concentration at which N would notbe expected to limit photosynthesis What is more high ni-trate concentrations are often associated with alder a nitrogen-fixing tree common in the Pacific Northwest the distributionof which has been expanded by long-term climatic changes(Hu et al 2001) and logging practices (Ruth and Harris1979) Hence the importance of salmon-derived N may beless than commonly assumed and varies with natural vege-tation patterns and human management activities

In contrast P concentrations tend to be uniformly low ex-cept in areas with P-rich sedimentary bedrock (Ashley andSlaney 1997) and recent work has highlighted P as the dom-inant limiting nutrient (Bothwell 1989 Ashley and Slaney1997) Nutrient patterns in Idaho streams suggest that one-quarter to one-half are nutrient limited and that half of thoseare P limited (Thomas et al forthcoming) Salmon molar NPratios range from 121 to 151 (Ashley and Slaney 1997)making them relatively phosphorus rich Within Lynx CreekAlaska phosphorus that is imported by spawning sockeyesalmon at average run sizes constitutes a large proportion ofthe P available to epilithon (organic matter attached to rocksurfaces) on an annual basis (OrsquoKeefe and Edwards forth-coming) and may be the most important marine product ofthe recycling pathway In some systems however light ratherthan nutrients limits primary production Thus a pulse ofsalmon-derived nutrients may have little or no effect on pri-mary productivity (Rand et al 1992) although salmon maystill be an important resource for stream or terrestrial biotavia the consumption pathway

Given the heterogeneity in habitats and limiting factors itfollows that the ecological consequences of inputs vary amonghabitats and with dispersal pathways Epilithic chlorophyllstanding stocks increased following salmon spawning insome studies (eg Richey et al 1975 Wipfli et al 1998) wereunaffected in other studies (Minshall et al 1991) and decreasedafter spawning episodes in still other streams (Minikawa1997) Increases in salmon-derived inorganic nitrogen andphosphorus compounds have been documented from severalstreams (Brickell and Goering 1970 Schuldt and Hershey1995 Minikawa 1997) confirming that excretion and min-eralization of carcasses does increase the inorganic nutrientcapital of some streams Furthermore artificially increasinginputs of inorganic P and N in streams in British Columbiaincreased chlorophyll accrual rates benthic insect density andthe growth rates and size of fish (Perrin et al 1987 Johnstonet al 1990 Mundie et al 1991) Therefore by inference it isassumed that nutrients released by spawning salmon alsohave the same effect Increased ecosystem primary produc-tivity in streams as a result of salmon nutrient inputs remains

922 BioScience October 2002 Vol 52 No 10

Articles

an interesting hypothesis that has not been confirmed how-ever particularly across a broad range of stream types andspawner densities

Increased lake productivity that is caused by salmon-nu-trient inputs is better documented Returning sockeye salmoncan contribute a large proportion of available P and N de-pending upon the size of the salmon run (Hartman andBurgner 1972 Stockner and Shortreed 1975 Mathisen et al1988) which may elevate phytoplankton and zooplanktondensities and increase juvenile salmon production (Narver1967) As in stream systems experimental nutrient inputs tolakes primarily P increased lake productivity (Hyatt andStockner 1985 Stockner and MacIsaac 1996) in contrast nu-trient budgets within Redfish Lake Idaho suggest that stim-ulation of lake production by spawning runs of salmon hasalways been small thus indicating that lake responses also varywith geography (Gross et al 1998)

There is abundant evidence that salmon availability influ-ences population dynamics of consumers via the consump-tion pathway More carcasses generally translate into higherdensities and elevated growth rates of invertebrates and ju-venile salmonids may grow faster by directly consumingsalmon tissue or consuming invertebrates that have beenscavenging salmon carcasses (Johnson and Ringler 1979Bilby et al 1996 1998 Wipfli et al 1998 1999 Chaloner andWipfli 2002) Furthermore juvenile coho salmon had higherlevels of whole-body lipids and a higher proportion of lipidallocated to energy reserves (triacylglycerol) when reared inthe presence of salmon carcasses (Ron Heintz National Ma-rine Fisheries Service Auke Bay AK personal communica-tion 22 September 2002) which may lead to higher survival(Wipfli et al forthcoming) Additionally marine isotopes ofN and C are higher in several trophic levels in salmon versusnonsalmon streams (Kline et al 1997) illustrating that thesefreshwater biota are sequestering marine nutrients into bodytissues presumably by direct consumption of salmon tissue(Bilby et al 1996)

Similar responses may occur in terrestrial communitiesLower trophic levels such as invertebrate scavengers (egdipterans) utilize the available salmon biomass (eg Reim-chen 2000) and reproduce thereby increasing local densitiesby conversion of salmon biomass into invertebrate tissue Den-sities of many vertebrates increase locally presumably bymoving from surrounding areas to feed on salmon Duringthe breeding season insectivorous riparian passerines arefound in greater densities on salmon streams than on otherstreams suggesting that bird communities may be respond-ing to the pulse of invertebrates produced by the availabilityof salmon (Gende and Willson 2001) Across the landscapethe carrying capacity of bears increases where salmon are avail-able with populations up to 80 times denser in coastal areaswhere salmon are abundant than in interior areas (Miller etal 1997 Hilderbrand et al 1999b) Fitness-related variablesincluding growth rates litter sizes and reproductive successhave been attributed to salmon availability for salmon con-sumers such as eagles bears and mustelids which suggests

that salmon play an important role in the population dynamicsof these terrestrial consumers (Hansen 1987 Ben-David1997 Hilderbrand et al 1999c)

Terrestrial vegetation also may respond to the presence ofsalmon The 15N signal presumed to be salmon-derived hasbeen detected in riparian shrubs and trees up to 500 m or morefrom streams (Bilby et al 1996 Ben-David et al 1998 Hilder-brand et al 1999a) in Washington and Alaska Marine sig-natures are higher in areas where bears feed on salmon (Ben-David et al 1998 Hilderbrand et al 1999a) which suggeststhat the foraging activities of bears play an important role inmaking salmon-derived nutrients available to terrestrial veg-etation There is also some indication that growth of ripar-ian trees may increase where salmon-derived nutrient inputoccurs particularly in areas with bear foraging activity(Helfield and Naiman 2001) although the spatial extent of thisphenomenon is unknown Increased growth of riparian veg-etation caused by salmon inputs if it occurs widely could haveramifications for riparian systems by changing litter largewoody debris and the amount of light reaching streambedsas well as by altering terrestrial vegetation structure (Helfieldand Naiman 2001) food for herbivorous insects and coverfor nesting birds

In addition to the direct effects of salmon subsidies thereare several indirect ecological ramifications of these subsidiesWe note three possible examples (1) Salmon are a majorsource of food for bears but bears also consume fleshy fruitsand thus serve as important seed dispersal agents for nu-merous plant species in coastal forests (eg Willson 1993)Without salmon bear densities would be lower and seed dis-persal patterns could be altered with unknown consequences(2) Fertilization of plants commonly leads to higher nutrientcontent and enhanced growth and some herbivorous in-sects attack fertilized plants at high rates (Price 1991) Birdsfeed on many herbivorous insects and in some circum-stances are capable of reducing the herbivore load on plantsthus fostering better plant growth (Marquis and Whelan1994) Higher densities of insectivorous breeding birds alongsalmon streams in spring (Gende and Willson 2001) mightmean that natural control of herbivorous insects is better insalmon-subsidized forests (3) Because salmon subsidies canlead to enhanced growth and survival of stream-resident fish(Bilby et al 1998) life-history strategies that are dependenton juvenile growth rates may change For instance the tim-ing and even the probability of migration from fresh water tothe sea may vary with juvenile growth rates (Healy and Heard1984) which in turn affects body size patterns of spawningcompetition and fecundity with ramifications for populationproductivity and thus for consumers and commercial harvests

Conservation and managementIt is clear from the growing body of literature that salmon mayinfluence the food webs trophic structure nutrient budgetsand possibly the productivity of freshwater and terrestrial sys-tems although the effect varies widely between systems andis contingent upon timing scale retention mechanisms al-

October 2002 Vol 52 No 10 BioScience 923

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

an interesting hypothesis that has not been confirmed how-ever particularly across a broad range of stream types andspawner densities

Increased lake productivity that is caused by salmon-nu-trient inputs is better documented Returning sockeye salmoncan contribute a large proportion of available P and N de-pending upon the size of the salmon run (Hartman andBurgner 1972 Stockner and Shortreed 1975 Mathisen et al1988) which may elevate phytoplankton and zooplanktondensities and increase juvenile salmon production (Narver1967) As in stream systems experimental nutrient inputs tolakes primarily P increased lake productivity (Hyatt andStockner 1985 Stockner and MacIsaac 1996) in contrast nu-trient budgets within Redfish Lake Idaho suggest that stim-ulation of lake production by spawning runs of salmon hasalways been small thus indicating that lake responses also varywith geography (Gross et al 1998)

There is abundant evidence that salmon availability influ-ences population dynamics of consumers via the consump-tion pathway More carcasses generally translate into higherdensities and elevated growth rates of invertebrates and ju-venile salmonids may grow faster by directly consumingsalmon tissue or consuming invertebrates that have beenscavenging salmon carcasses (Johnson and Ringler 1979Bilby et al 1996 1998 Wipfli et al 1998 1999 Chaloner andWipfli 2002) Furthermore juvenile coho salmon had higherlevels of whole-body lipids and a higher proportion of lipidallocated to energy reserves (triacylglycerol) when reared inthe presence of salmon carcasses (Ron Heintz National Ma-rine Fisheries Service Auke Bay AK personal communica-tion 22 September 2002) which may lead to higher survival(Wipfli et al forthcoming) Additionally marine isotopes ofN and C are higher in several trophic levels in salmon versusnonsalmon streams (Kline et al 1997) illustrating that thesefreshwater biota are sequestering marine nutrients into bodytissues presumably by direct consumption of salmon tissue(Bilby et al 1996)

Similar responses may occur in terrestrial communitiesLower trophic levels such as invertebrate scavengers (egdipterans) utilize the available salmon biomass (eg Reim-chen 2000) and reproduce thereby increasing local densitiesby conversion of salmon biomass into invertebrate tissue Den-sities of many vertebrates increase locally presumably bymoving from surrounding areas to feed on salmon Duringthe breeding season insectivorous riparian passerines arefound in greater densities on salmon streams than on otherstreams suggesting that bird communities may be respond-ing to the pulse of invertebrates produced by the availabilityof salmon (Gende and Willson 2001) Across the landscapethe carrying capacity of bears increases where salmon are avail-able with populations up to 80 times denser in coastal areaswhere salmon are abundant than in interior areas (Miller etal 1997 Hilderbrand et al 1999b) Fitness-related variablesincluding growth rates litter sizes and reproductive successhave been attributed to salmon availability for salmon con-sumers such as eagles bears and mustelids which suggests

that salmon play an important role in the population dynamicsof these terrestrial consumers (Hansen 1987 Ben-David1997 Hilderbrand et al 1999c)

Terrestrial vegetation also may respond to the presence ofsalmon The 15N signal presumed to be salmon-derived hasbeen detected in riparian shrubs and trees up to 500 m or morefrom streams (Bilby et al 1996 Ben-David et al 1998 Hilder-brand et al 1999a) in Washington and Alaska Marine sig-natures are higher in areas where bears feed on salmon (Ben-David et al 1998 Hilderbrand et al 1999a) which suggeststhat the foraging activities of bears play an important role inmaking salmon-derived nutrients available to terrestrial veg-etation There is also some indication that growth of ripar-ian trees may increase where salmon-derived nutrient inputoccurs particularly in areas with bear foraging activity(Helfield and Naiman 2001) although the spatial extent of thisphenomenon is unknown Increased growth of riparian veg-etation caused by salmon inputs if it occurs widely could haveramifications for riparian systems by changing litter largewoody debris and the amount of light reaching streambedsas well as by altering terrestrial vegetation structure (Helfieldand Naiman 2001) food for herbivorous insects and coverfor nesting birds

In addition to the direct effects of salmon subsidies thereare several indirect ecological ramifications of these subsidiesWe note three possible examples (1) Salmon are a majorsource of food for bears but bears also consume fleshy fruitsand thus serve as important seed dispersal agents for nu-merous plant species in coastal forests (eg Willson 1993)Without salmon bear densities would be lower and seed dis-persal patterns could be altered with unknown consequences(2) Fertilization of plants commonly leads to higher nutrientcontent and enhanced growth and some herbivorous in-sects attack fertilized plants at high rates (Price 1991) Birdsfeed on many herbivorous insects and in some circum-stances are capable of reducing the herbivore load on plantsthus fostering better plant growth (Marquis and Whelan1994) Higher densities of insectivorous breeding birds alongsalmon streams in spring (Gende and Willson 2001) mightmean that natural control of herbivorous insects is better insalmon-subsidized forests (3) Because salmon subsidies canlead to enhanced growth and survival of stream-resident fish(Bilby et al 1998) life-history strategies that are dependenton juvenile growth rates may change For instance the tim-ing and even the probability of migration from fresh water tothe sea may vary with juvenile growth rates (Healy and Heard1984) which in turn affects body size patterns of spawningcompetition and fecundity with ramifications for populationproductivity and thus for consumers and commercial harvests

Conservation and managementIt is clear from the growing body of literature that salmon mayinfluence the food webs trophic structure nutrient budgetsand possibly the productivity of freshwater and terrestrial sys-tems although the effect varies widely between systems andis contingent upon timing scale retention mechanisms al-

October 2002 Vol 52 No 10 BioScience 923

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

ternative nutrient sources and baseline limiting factors Howmight these results influence resource managers and conser-vation practitioners

The emerging picture of the ecological importance ofsalmon subsidies to freshwater and terrestrial ecosystemsforcefully emphasizes the importance of a broad holisticperspective The links between ocean and land mean thatmanagement of an ocean fishery can have far-reaching effectson distant ecosystems and vice versa Any management ac-tivity that reduces the numbers of salmon returning to spawn-ing grounds may influence processes that are driven by thesalmon input (figures 1b 2) Furthermore those links ne-cessitate cross-disciplinary research and applications of knowl-edge (Willson et al 1998) The big picture must be viewed withcare however Regional and temporal variation in inputsand outcomes means that results from a single study cannotbe assumed to be universally applicable Factors limiting pro-ductivity differ among locations Moreover the ways in whichnutrients are spread and the degree to which they are spreadacross the landscape vary even in natural systems Attemptsto reintroduce some single component (eg C or N) to ahighly complex failing system run the risk of all simplistic ap-proaches in that they neglect the inherent complexity of thesystem

Furthermore managers considering the role of salmonshould recognize that the ecological response to the salmonsubsidy is species specific Artificially placing a few salmon car-casses on stream banks (or in streams) may locally increaseinvertebrate populations by several orders of magnitude asthey colonize and reproduce within hours or days of theavailability of the carcasses The same number of carcasseshowever would not permit a population response by largervertebrates such as eagles bears or mustelids (which in turnserve as important nutrient-dispersal agents) because thatwould require thousands of kilograms of salmon distributedacross many streams over a long time period

Emerging management techniques are primarily designedto manipulate processes via the recycling pathway without ex-plicitly considering biomass-related ecosystem flows (via theconsumption pathway) For example in an effort to replacethe inorganic minerals reduced by the depletion of salmonruns companies have produced slow-release fertilizer bri-quettes that are designed to increase inorganic P concentra-tions in selected streams (Sterling et al 2000) Fertilization ofstreams and lakes with inorganic P and N has successfully in-creased algal standing stocks salmonid fry weights and pro-duction (Stockner and MacIsaac 1996 Ashley and Slaney1997) However if a significant proportion of the increasedsalmon numbers produced by such augmentation is not al-lowed to return die and be consumed in the natal waters thetransfer of salmon-derived products to aquatic and partic-ularly to terrestrial food webs will remain truncated even ifstream and lake productivity is enhanced If managers see nu-trient augmentation solely as a way to increase salmon har-vests rather than as a stopgap measure to enhance runs un-til densities of adult fish return to a sustainable level (Gresh

et al 2000) then the decoupling of subsidies to the terrestrialecosystem will remain In such a scenario the failure to dis-tinguish salmon as food from salmon as inorganic nutrientscould result in unbalanced management practices

A broad perspective carries an ecological message for fish-eries management which is driven chiefly by commercialconsiderations and the goal of harvesting as many fish aspossible What is ldquopossiblerdquo has been altered during the courseof commercial exploitation from ldquotaking everything thatcould be caughtrdquo to the concept of ldquomaximum sustainableyieldrdquo (Smith 1994) Even that concept has been questionedhowever because uncertainty in estimating fish populationdynamics makes prediction of sustainable harvest levels verydifficult (Hilborn and Walters 1992) Most recently the eco-logical value of salmon subsidies has attracted managementattention chiefly for enhancing fish production Yet to beachieved is inclusion of the wider ecological significance ofsalmon for the landscape Small stocks are ecologically im-portant as sources for colonizers food for wildlife and nu-trients for freshwater and terrestrial systems but they arerarely counted and sustain unknown levels of harvest be-cause of mixed-stock fisheries Small stocks also increase ge-netic variation which is important in maintaining evolu-tionarily significant units

A primary goal of conservation and restoration is obviouslythe conservation or restoration of the salmon runs themselvesbecause without them none of the related processes operateTo this end efforts at enhancing stream productivity waterquality natural water flow patterns and stream accessibilityand suitability are all appropriate In addition the broad per-spective provides new goals for conservation and restora-tion efforts by drawing attention to the ecological roles ofsalmon subsidies Restoration of fully functional systemsclearly depends on inclusion of the means of spreading thesalmon subsidies across the landscape via surface and hy-porheic flows and populations of consumers (figure 2b) Insome regions full restoration is clearly impossible becausechanges in animal habitats hydrology and stream geomorph-ology have permanently altered ecosystem function In thesecases the systems may suffer from one (or more) of theldquoratchetsrdquo described by Pitcher (2001) for example human-caused species extinction that prevent systems from return-ing to their natural state In other regions less altered by hu-man activity full maintenance of natural ecosystem functionmay be possible Of particular importance is the preservationand understanding of the processes in relatively pristine sys-tems so that they can provide a baseline goal toward whichrestoration efforts can be aimed In so doing another ratchetcan be avoided for example a sliding scale of what is perceivedas natural caused by a lack of true reference systems (Pitcher2001)

Future researchIt is clear that salmon-derived nutrient subsidies can play asignificant role in the ecology of aquatic and terrestrial ecosys-tems but site- and taxon-specific variability influence the

924 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

magnitude of the response For fisheries managers to acceptthe concept that salmon escapements should be managed tomaximize ecosystem productivity and then to translate thatconcept into improved management researchers must firstprovide some estimates of the relationship between the num-ber of fish allowed to escape commercial harvest and returnto spawn and basin-specific intrinsic factors and productiv-ity More research is required on dosendashresponse relation-ships where varying spawning densities lead to predictableecosystem responses For example there is evidence that theecological response varies with the density of available car-casses In Alaska NH4

+ and SRP concentrations varied pre-dictably with run size (OrsquoKeefe and Edwards forthcoming)and biofilm and benthic macroinvertebrate standing stocksappeared to reach an asymptote at intermediate levels of car-cass availability (Wipfli et al 1998 1999) Presumably abovesome level of food availability or mineral input other processeslimited production

To our knowledge the only attempt to consider the eco-logical effects of salmon spawners while establishing escape-ment goals was by Bilby and colleagues (2001) who proposedusing the stable isotope signature of stream fish They observedthat tissue 15N values reached an asymptote in juvenile cohosalmon as escapement levels increased suggesting that cohofry might be used to define the point at which the (freshwa-ter) ecosystem is saturated by spawners Although the use-fulness of their approach has not been broadly confirmed itchallenges other scientists to develop additional approachesto translate our emerging understanding of the ecological ef-fects of salmon-derived nutrients into practical managementtechniques

The role of salmon in influencing ecosystem productivityneeds clarification Despite the conventional wisdom thatspawning salmon increase aquatic ecosystem productivity onlyone published study has quantified primary productivity (g per m2 per day) in response to the presence of spawnersThe sole reference documenting an increase in primary pro-ductivity in streams (Richey et al 1975) was for landlockedkokanee salmon and results from that study also showed thatthe response varied widely with run size and stream flow Weknow of no study in which secondary production by streaminvertebrates has been quantified Most publications reportindirect responses such as increases in density standingstocks or individual growth rates Although such surrogatesmay be correlated to ecosystem productivity the relationshipis not necessarily direct or consistent throughout the rangeof salmon or over time Although it is intuitively appealingto assume that such evidence suggests that productivity is en-hanced studies confirming such productivity have not beendone

The influence of salmon-derived inputs on biodiversity islargely unknown because baseline levels of productivity andthe relationship between biodiversity and productivity mayvary among sites For example the relationship between di-versity and productivity was initially thought to be hump-shaped Diversity increases with productivity at lower levels

of productivity but decreases as productivity continues to in-crease (eg VanderMeulen et al 2001) If that relationship ap-plies to systems subsidized by salmon then the outcome fordiversity clearly depends on the initial levels of productivityAn increase of diversity would be predicted only if the initiallevels of productivity were relatively low However recentresearch has suggested that this relationship varies among sitesand taxa (Mittelbach et al 2001 Schmid 2002) An addi-tional consideration is that increased productivity may per-mit increases of population size thus lowering the risk of ex-tinction and buffering biodiversity through time Althoughthere is some evidence of salmon influencing biodiversity orcommunity structure of invertebrates (Piorkowski 1995)the relationship between biodiversity and productivity insalmon-subsidized systems has yet to be established in acomprehensive manner

The validity of using stable isotope techniques to tracksalmon biomass throughout receiving ecosystems requiresconfirmation Carbon and nitrogen isotopes have been usedextensively to study the role of salmon in aquatic and terres-trial ecosystems these heavy isotopes can be traced only bymaking many assumptions about competing processes andalternative isotopic pool signatures however (Kline et al1997) As yet little work has tested the validity of those as-sumptions or how other factors may influence isotope sig-natures including fractionation rates among trophic levelsvegetation patterns (eg the role of nitrogen-fixing alder) andthe changes in isotopic signatures of salmon (Doucett et al1999) Recent work in southwestern Alaska has shown thatdenitrification potentials are greater in spawning streamsthan in reference streams without salmon (Gilles Pinay Uni-versiteacute de Rennes I Rennes France personal communication15 November 2001) which suggests that there are systematicviolations of some key assumptions in using 15N values to tracemarine N Even where underlying assumptions are valid theecological relevance is not clear when for example streambiofilm has 45 of its N derived from salmon especially inP-limited ecosystems

Long-term whole-system manipulations are necessary toquantify dosendashresponse relationships and to avoid experi-mental design flaws in current approaches (see also Schindleret al 2000) Published research is largely descriptive not ex-perimental and tracking the fate of salmon biomass withinecosystems is difficult because of uncontrolled and poorlyquantified confounding factors The value of nonsalmonreference streams as controls is weakened by potential sys-tematic bias and the difference in dispersal pathways be-tween nonliving and living fish limits the generality of small-scale fish-addition experiments Detailed study of systemswhere escapements are dramatically altered either by re-ducing existing runs for prolonged periods or by studying runintroductions in areas where fish passes have been con-structed would assist the pursuit of the previously suggestedresearch elements and would clarify our interpretation ofexisting data

October 2002 Vol 52 No 10 BioScience 925

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

Finally we have focused on the role of Pacific salmon in thePacific Northwest because most of the information on salmoninputs to freshwater and terrestrial ecosystems comes fromwork on Pacific salmon But anadromy is not unique to thePacific Northwest nor to salmon other species with anadro-mous life histories include smelt sturgeon noodlefish andlamprey Anadromy is widespread in the temperate and bo-real regions of the Northern Hemisphere (McDowall 1988)and there are some reports that consumers respond to the sub-sidies provided by some of these species (eg Gende et al2001b Marston et al 2002) Thus the ecological roles and pop-ulation sizes of other anadromous fishes both past and pre-sent need to be addressed

ConclusionIn A Sand County Almanac Aldo Leopold (1949) describedthe incremental movement of atom X from headwaters toocean driven by the forces of gravity and discharge to its ul-timate ldquoprisonrdquoin the sea Understanding the implications andcontrols of ldquonutrient spiralingrdquo as this phenomenon has beentermed has driven much of recent stream ecosystem research(eg Peterson et al 2001) Our current understanding of thephenomenon of salmon-derived nutrient input clearly showsthat a small but important proportion of those atoms escapetheir ldquoprisonrdquo to return in the bodies of ocean-dwelling or-ganisms whose behavior drives them back against gravity andstream discharge to penetrate the continent Quantifying theecological effects of this phenomenon and translating that un-derstanding into useful conceptual and practical tools to bet-ter manage oceanic freshwater and terrestrial ecosystemsmdashwithout reference to the jurisdictional organizational andconceptual boundaries that currently inhibit usmdashremains achallenge for scientists and managers alike

AcknowledgmentsThis manuscript benefited greatly from reviews by LisaThompson Tom Quinn Tom OrsquoKeefe Wayne Minshall andDominic Chaloner Stacey Poulson was instrumental in cre-ating the figures Funding was provided by the USDA ForestService Aquatic and Land Interactions Program PacificNorthwest Research Station Juneau Alaska

References citedAshley KI Slaney PA 1997 Accelerating recovery of stream river and pond

productivity by low-level nutrient replacement Chapter 13 in Slaney PAZaldokas D eds Fish Habitat Rehabilitation Procedures VancouverBritish Columbia (Canada) Ministry of Environment Lands and Parks

Ben-David M 1997 Timing of reproduction in wild mink The influence ofspawning Pacific salmon Canadian Journal of Zoology 75 376ndash382

Ben-David M Hanley TA Schell DM 1998 Fertilization of terrestrial veg-etation by spawning Pacific salmon The role of flooding and predatoractivity Oikos 83 47ndash55

Bilby RE Fransen BR Bisson PA 1996 Incorporation of nitrogen and car-bon from spawning coho salmon into the trophic system of smallstreams Evidence from stable isotopes Canadian Journal of Fisheries andAquatic Sciences 53 164ndash173

Bilby RE Fransen BR Bisson PA Walter JK 1998 Response of juvenilecoho salmon (Oncorhynchus kisutch) and steelhead (Oncorhynchus

mykiss) to the addition of salmon carcasses to two streams in southwesternWashington USA Canadian Journal of Fisheries and Aquatic Sciences55 1909ndash1918

Bilby RE Fransen BR Walter JK Cederholm CJ Scarlett WJ 2001 Prelim-inary evaluation of the use of nitrogen stable isotope ratios to establishescapement levels for Pacific salmon Fisheries 26 6ndash13

Bothwell ML 1989 Phosphorus-limited growth dynamics of lotic peri-phytic diatom communities Areal biomass and cellular growth rate re-sponses Canadian Journal of Fisheries and Aquatic Sciences 461293ndash1301

Brett JR 1995 Energetics Pages 3ndash68 in Groot C Margolis L Clarke WCeds Physiological Ecology of Pacific Salmon Vancouver British Co-lumbia (Canada) UBC Press

Brickell DC Goering JJ 1970 Chemical effects of salmon decomposition onaquatic ecosystems Pages 125ndash138 in Murphy RS Nyquist D eds Wa-ter Pollution Control in Cold Climates Washington (DC) Environ-mental Protection Agency

Cederholm CJ Peterson NP 1985 The retention of coho salmon (On-corhynchus kisutch) carcasses by organic debris in small streams Cana-dian Journal of Fisheries and Aquatic Sciences 42 1222ndash1225

Cederholm CJ Kunze MD Murota T Sibatani A 1999 Pacific salmon car-casses Essential contributions of nutrients and energy for aquatic andterrestrial ecosystems Fisheries 24 6ndash15

Chaloner DT Wipfli MS 2002 Influence of decomposing Pacific salmon car-casses on macroinvertebrate growth and standing stock in southeasternAlaska streams Journal of the North American Benthological Society 21430ndash442

Clinton SM Edwards RT Naiman RJ 2002 ForestndashRiver Interactions In-fluence on hyporheic dissolved organic carbon concentrations in afloodplain terrace Journal of the American Water Resources Association38 619ndash632

Donaldson JR 1967 The phosphorous budget of Iliamna Lake Alaska as re-lated to the cyclic abundance of sockeye salmon PhD dissertation Uni-versity of Washington Seattle

Doucett RR Booth RK Power G McKinley RS 1999 Effects of the spawn-ing migration on the nutritional status of anadromous Atlantic salmon(Salmo salar) Insights from stable-isotope analysis Canadian Journal ofFisheries and Aquatic Sciences 56 2172ndash2180

Edwards RT 1999 The hyporheic zone Pages 399ndash429 in Naiman RJ BilbyRE eds River Ecology and Management New York Springer-Verlag

Finney BP Gregory-Eaves I Sweetman J Douglas MSV Smol JP 2000 Im-pacts of climatic change and fishing on Pacific salmon abundance overthe past 300 years Science 290 795ndash799

Francis RC Hare SR 1994 Decadal-scale regime shifts in the large marineecosystems of the Northeast Pacific A case for historical science Fish-eries Oceanography 3 279ndash291

Freeman C Lock MA 1995 The biofilm polysaccharide matrix A bufferagainst changing organic substrate supply Limnology and Oceanogra-phy 40 273ndash278

Fujiwara M Highsmith RC 1997 Harpacticoid copepods Potential link be-tween inbound adult salmon and outbound juvenile salmon MarineEcology Progress Series 158 205ndash213

Gargett AE 1997 The optimal stability lsquowindowrsquo A mechanism underlyingdecadal fluctuations in North Pacific salmon stocks Fisheries Oceanog-raphy 6 109ndash117

Gende SM 2002 Foraging behavior of bears at salmon streams Intake choiceand the role of salmon life history PhD dissertation University of Wash-ington Seattle

Gende SM Willson MF 2001 Passerine densities in riparian forests ofsoutheast Alaska Potential role of anadromous spawning salmon Con-dor 103 624ndash629

Gende SM Quinn TPWillson MF 2001a Consumption choice by bears feed-ing on salmon Oecologia 127 372ndash382

Gende SM Womble JN Willson MF Marston BH 2001b Cooperative for-aging by Steller sea lions (Eumetopias jubatus) Canadian Field-Naturalist115 355ndash356

926 BioScience October 2002 Vol 52 No 10

Articles

Gresh T Lichatowich J Schoonmaker P 2000 An estimation of historicand current levels of salmon production in the northeast Pacific ecosys-tem Fisheries 25 15ndash21

Groot C Margolis L 1991 Pacific Salmon Life Histories Vancouver BritishColumbia (Canada) UBC Press

Gross HP Wurtsbaugh WA Luecke C 1998 The role of anadromous sock-eye salmon in the nutrient loading and productivity of Redfish LakeIdaho Transactions of the American Fisheries Society 127 1ndash18

Halupka KC Bryant MD Willson MF Everest FH 2000 Biological Char-acteristics and Population Status of Anadromous Salmon in SoutheastAlaska Portland (OR) US Department of Agriculture General TechnicalReport PNW-GTR-468

Hansen AJ 1987 Regulation of bald eagle reproductive rates in southeastAlaska Ecology 68 1387ndash1392

Hartman WL Burgner RL 1972 Limnology and fish ecology of sockeyesalmon nursery lakes of the world Journal of the Fisheries Research Boardof Canada 29 699ndash715

Healy MC Heard WR 1984 Inter- and intrapopulation variation in the fe-cundity of chinook salmon (Oncorhynchus tshawytscha) and its relevanceto life history theory Canadian Journal of Fisheries and Aquatic Sciences41 476ndash483

Helfield JM Naiman RJ 2001 Effects of salmon-derived nitrogen on riparianforest growth and implications for stream productivity Ecology 822403ndash2409

Hendry AP Berg OK 1999 Secondary sexual characters energy use senes-cence and the cost of reproduction in sockeye salmon Canadian Jour-nal of Zoology 77 1663ndash1675

Hilborn R Walters CJ 1992 Quantitative fisheries stock assessment Choicedynamics and uncertainty New York Chapman and Hall

Hilderbrand GV Hanley TA Robbins CT Schwartz CC 1999a Role ofbrown bears (Ursus arctos) in the flow of marine nitrogen into a terrestrialecosystem Oecologia 121 546ndash550

Hilderbrand GV Schwartz CC Robbins CT Jacoby ME Hanley TA ArthurSM Servheen C 1999b The importance of meat particularly salmonto body size population productivity and conservation of North Amer-ican brown bears Canadian Journal of Zoology 77 132ndash138

Hilderbrand GV Jenkins SG Schwartz CC Hanley TA Robbins CT 1999cEffect of seasonal differences in dietary meat intake on changes in bodymass and composition in wild and captive brown bears Canadian Jour-nal of Zoology 77 1623ndash1630

Hu F Finney BP Brubaker LB 2001 Effects of Holocene Alnus expansionon aquatic productivity nitrogen cycling and soil development in south-western Alaska Ecosystems 4 358ndash368

Hyatt KD Stockner JG 1985 Responses of sockeye salmon (Oncorhyncusnerka) to fertilization of British Columbia coastal lakes Canadian Jour-nal of Fisheries and Aquatic Sciences 42 320ndash331

Johnson JH Ringler NH 1979 The occurrence of blow fly larvae (DipteraCalliphoridae) on salmon carcasses and their utilization as food by ju-venile salmon and trout Great Lakes Entomologist 12 137ndash140

Johnston NT Perrin CJ Slaney PA Ward BR 1990 Increased juvenilesalmonid growth by whole-river fertilization Canadian Journal of Fish-eries and Aquatic Sciences 47 862ndash872

Juday C Rich WH Kemmerer GI Mann A1932 Limnological studies of Kar-luck Lake Alaska 1926ndash1930 Bulletin of the Bureau of Fisheries 47407ndash434

Kline TC Jr Goering JJ Piorkowski RJ 1997 The effect of salmon carcasseson Alaskan freshwaters Pages 179ndash204 in Milner AM Oswood MW edsFreshwaters of Alaska Ecological Syntheses New York Springer

Larkin GA Slaney PA 1997 Implications of trends in marine-derived nu-trient influx to south coastal British Columbia salmonid production Fish-eries 22 16ndash24

Leopold A1949A Sand County Almanac New York Oxford University PressLichatowich J 1999 Salmon without Rivers A History of the Pacific Salmon

Crisis New York Island PressLyle AA Elliott JM 1998 Migratory salmonids as vectors of carbon nitro-

gen and phosphorous between marine and freshwater environments innorth-east England Science of the Total Environment 210211 457ndash468

Mantua NJ Hare SR Zhang Y Wallace JM Francis RC 1997 A Pacific in-terdecadal climate oscillation with impacts on salmon production Bul-letin of the Meteorological Society 78 1069ndash1079

Marquis RJ Whelan CJ 1994 Insectivorous birds increase growth of whiteoak through consumption of leaf-chewing insects Ecology 75 2007ndash2014

Marston BH Willson MF Gende SM 2002 Predator aggregations at a eu-lachon spawning run Marine Ecology Progress Series 231 229ndash236

Mathisen OA Parker PL Goering JJ Kline TC Poe PH Scalan RS 1988 Re-cycling of marine elements transported into freshwater systems byanadromous salmonVereinigung fuumlr Theoretische und Angewandte Lim-nologie 23 2249ndash2258

McDowall RM 1988 Diadromy in Fishes Migrations between Freshwaterand Marine Environments London Croom Helm

Miller SD et al 1997 Brown and black bear density estimation in Alaska us-ing radiotelemetry and replicated mark-resight techniques WildlifeMonographs 133 1ndash55

Minakawa N 1997 The dynamics of aquatic insect communities associatedwith salmon spawning PhD dissertation University of WashingtonSeattle

Minshall GW Hitchcock E Barnes JR 1991 Decomposition of rainbow trout(Oncorhynchus mykiss) carcasses in a forest stream ecosystem inhabitedonly by nonanadromous fish populations Canadian Journal of Fisheriesand Aquatic Sciences 48 191ndash195

Mittelbach GG Steiner CF Scheiner SM Gross KL Reynolds HL Waide RBWillig MR Dodson SI Gough L 2001 What is the observed relation-ship between species richness and productivity Ecology 82 2381ndash2396

Mundie JH Simpson KS Perrin CJ 1991 Responses of stream periphytonand benthic insects to increases in dissolved inorganic phosphorous ina mesocosm Canadian Journal of Fisheries and Aquatic Sciences 482061ndash2072

Narver DW 1967 Primary productivity in the Babine Lake system BritishColumbia Journal of the Fisheries Research Board of Canada 242045ndash2052

Nehlsen W Williams JE Lichatowich JA 1991 Pacific salmon at the cross-roads Stocks at risk from California Oregon Idaho and WashingtonFisheries 16 4ndash21

OrsquoKeefe TC Edwards RT Evidence for hyporheic transfer and removal of ma-rine-derived nutrients in sockeye streams in southeast Alaska AmericanFisheries Society Symposium 33 Forthcoming

Perrin CJ Bothwell ML Slaney PA 1987 Experimental enrichment of a coastalstream in British Columbia Effects of organic and inorganic additionson autotrophic periphyton production Canadian Journal of Fisheriesand Aquatic Sciences 44 1247ndash1256

Peterson BJ et al 2001 Control of nitrogen export from watersheds byheadwater streams Science 292 86ndash90

Piorkowski RJ 1995 Ecological effects of spawning salmon on several south-central Alaskan streams PhD dissertation University of Alaska Fairbanks

Pitcher TJ 2001 Fisheries managed to rebuild ecosystems Reconstructingthe past to salvage the future Ecological Applications 11 601ndash617

Polis GA Hurd SD 1996 Linking marine and terrestrial food webs Al-lochthonous input from the ocean supports high secondary productiv-ity on small islands and coastal land communities American Natural-ist 147 396ndash423

Power ME 1992 Top-down and bottom-up forces in food webs Do plantshave primacy Ecology 73 733ndash746

Price PW 1991 The plant vigor hypothesis and herbivore attack Oikos 62244ndash251

Rand PS Hall AS McDowell WH Ringler NH Kennen JG 1992 Factors lim-iting primary productivity in Lake Ontario tributaries receiving salmonmigrations Canadian Journal of Fisheries and Aquatic Sciences 492377ndash2385

Reimchen TE 2000 Some ecological and evolutionary aspects of bearndashsalmoninteractions in coastal British Columbia Canadian Journal of Zoology78 448ndash457

Richey JE Perkins MA Goldman CR 1975 Effects of Kokanee salmon(Oncorhynchus nerka) decomposition on the ecology of a subalpinestream Journal of the Fisheries Research Board of Canada 32 817ndash820

October 2002 Vol 52 No 10 BioScience 927

Articles

Robbins CT 1993 Wildlife Feeding and Nutrition San Diego Academic PressRuth RH Harris AS 1979 Management of Western Hemlock-Sitka Spruce

Forests for Timber Production Portland (OR) USDA Forest Service Gen-eral Technical Report PNW-88

Schindler DE Herwig BR Carpenter SR 2000 Biotic manipulations ofaquatic ecosystems Pages 308ndash317 in Sala OE Jackson RB MooneyRB Howarth RW eds Methods in Ecosystem Science New York Springer

Schmid B 2002 The species richness-productivity controversy Trends in Ecol-ogy and Evolution 17 113ndash114

Schuldt JA Hershey AE 1995 Effect of salmon carcass decomposition on LakeSuperior tributary streams Journal of the North American Bentholog-ical Society 14 259ndash268

Shuman RF 1950 Bear depredations on red salmon spawning populationsin the Karluk River system 1947 Journal of Wildlife Management 14 1ndash9

Smith TD 1994 Scaling Fisheries New York Cambridge University PressSterling MS Ashley KI Bautista AB 2000 Slow-release fertilizer for reha-

bilitating oligotrophic streams A physical characterization Water Qual-ity Research Journal of Canada 35 73ndash94

Stockner JG MacIsaac EA 1996 British Columbia lake enrichment pro-gramme Two decades of habitat enhancement for sockeye salmon Reg-ulated Rivers Research and Management 12 547ndash561

Stockner JG Shortreed KRS 1975 Phytoplankton succession and primaryproduction in Babine Lake British Columbia Journal of the Fisheries Re-search Board of Canada 32 2413ndash2427

Sugai SF Burrell DC 1984 Transport of dissolved organic carbon nutrientsand trace metals from the Wilson and Blossom Rivers to Smeaton Baysoutheast Alaska Canadian Journal of Fisheries and Aquatic Sciences 41180ndash190

Templin WD Collie JS Quinn TJ 1996 Run reconstruction of the wild pinksalmon fishery in Prince William Sound 1990ndash1991 American FisheriesSociety Symposium 18 499ndash508

Thomas SA Royer TV Minshall GW Snyder EB Current nutrient conditions

and the contribution of marine derived nutrients in Idaho streams

Proceedings of the Mercury Deposition Network Workshop American

Fisheries Society Forthcoming

VanderMeulen MA Hudson AJ Scheiner SM 2001 Three evolutionary hy-

potheses for the hump-shaped productivity-diversity curve Evolution-

ary Ecology Research 3 379ndash392

Willson MF 1993 Mammals as seed-dispersal mutualists in North America

Oikos 67 159ndash176

Willson MF Gende SM Marston BH 1998 Fishes and the forest Expand-

ing perspectives on fishndashwildlife interactions BioScience 48 455ndash462

Willson MF Gende SM Bisson P Anadromous fishes as ecological links be-

tween ocean fresh water and land In Polis G Power M Huxel G eds

Foodwebs at the Landscape Level Chicago University of Chicago Press

Forthcoming

Wipfli MS Hudson J Caouette J 1998 Influence of salmon carcasses on

stream productivity Response of biofilm and benthic macroinverte-

brates in southeastern Alaska USA Canadian Journal of Fisheries and

Aquatic Sciences 55 1503ndash1511

Wipfli MS Hudson JP Chaloner DT Caouette JP 1999 Influence of salmon

spawner densities on stream productivity in southeast Alaska Canadian

Journal of Fisheries and Aquatic Sciences 56 1600ndash1611

Wipfli MS Hudson JP Caouette JP Chaloner DT Marine subsidies in fresh-

waters Salmon carcasses increase the growth rates of stream-resident

salmonids Transactions of the American Fisheries Society Forthcom-

ing

Wood CM 1995 Excretion Pages 381ndash438 in Groot C Margolis L Clarke

WC eds Physiological Ecology of Pacific SalmonVancouver British Co-

lumbia (Canada) UBC Press

928 BioScience October 2002 Vol 52 No 10

Articles