Resource depression on the Northwest Coast ofNorth America

VIRGINIA L. BUTLER *

The mammal and fish faunal record from eight sites on the Columbia River (Oregon,USA) dating to the last 2200 years is examined to study subsistence change before andafter European contact. Results show an increased use of low-ronked resources before

contact and increased use of high-ranked resources after contact, trends that arepredicted from changing demography and human predation pressure.

Key-words: North America, zooarchaeology. prehistoric subsistence, prey choice model, resource depression

IntroductionA variety of evidence is accumulating fromvarious parts of the world that suggests pasthuman foragers greatly affected the animal andplant populations they were exploiting (Botkin1980; Broughton 1997; Holdaway &. Jacomb2000; Nagaoka 1998; Steadman 1995; Stiner eta1. 1999). Besides identifying prey response tohuman harvesting pressure, such studies alsotrack ways human predators adjusted to reducedprey abundance, pointing out the dynamicnature of predator-prey interactions. Many stud-ies have been conducted in a theoretical frame-work derived from evolutionary ecology,particularly foraging theory, and have demon-strated enormous explanatory power in account-ing for subsistence change in human economies(e.g. Christenson 1980; Grayson &. Cannon 1999;Gremillion 1996; O'Connell &. Hawkes 1981).

With some exceptions (Cohen 1981; Croes1995) researchers reviewing economies in Pa-cific Northwest North America have not con-sidered whether human foragers altered preypopulations. Some workers (e.g. Matson 1992)have expressed the view that key resources, likesalmonids, were difficult for humans to 'over-use'. Moreover, the Northwest Coast region haslong been viewed as a 'Garden of Eden', wherethe richness of the resource base was in somemeasure responsible for the 'exceptional'hunter-gatherers that lived there, known fortheir relatively high population density. socialorganization which included slavery. elaborate

art style and sedentary or semi-sedentary set-tlement patterns - attributes generally linkedto agriculture-based societies.

Several conditions suggest it would be use-ful to re-examine these views. Abundantarchaeological evidence shows that humanpopulations were growing and becoming in-creasingly sedentary, circumscribed and terri-torial during the late Holocene (Ames &:.Maschner 1999; Matson &:. Coupland 1995).Certainly in light of results from other parts ofthe world, it is reasonable to hypothesize thatgrowing populations of relatively sedentaryforagers depleted local food resources. Further-more, human population size was drasticallyreduced with the introduction of disease atEuropean contact. Such a decline should re-sult in reduced foraging pressure; prey popu-lations would have the opportunity to reboundand, in turn, human foragers could take advan-tage of increased prey abundance and shift. re-source selection.

This paper examines questions of late Holo-cene subsistence change, particularly the casefor human-induced resource depression in thePacific Northwest. I draw on the prey choicemodel from foraging theory to derive expecta-tions about resource selection and subsistencechange that would result from changes in for-aging pressure. I then test this model using themammal and fish faunal record from severalsites on the Lower Columbia River dating tothe last 2200 years. Results show an increased

. Department of Anthropology. Portland State University. Portland OR 97207. USA. virginiaOch2.ch.pdx.edu

Received 24 August 1999. accepted 31 October 1999, revised 7 January 2000, revised 5 June 2000.

ANTIQUITY 74 (2000): 64~1

VIRGJNIA L. Btm.ER850

High

~~

.ow

TimeFIGURE 1. Relative change in contribution of highand low ranked vertebrate resources before andafter population collapse.

use of low-ranked resources before Europeancontact and greater use of high-ranked resourcesafter contact. which are predicted from changesin human demography. The results are provoca-tive, suggesting that historic descriptions ofexceptional resource abundance in part mayreflect drastically reduced Native populationsize and that ethnographic records of subsist-ence practices may not reflect long-term trends.

The modelThe prey choice model provides a set of pre-dictions on which prey items a predator shouldchoose when an array of potential choices areencountered (Stephens &: Krebs 1986). Themodel holds that a predator's most efficientstrategy is to take the highest-ranked prey whenit is encountered. Whether lower-ranked preyis taken depends on the encounter rate withthe higher-ranked resources. Encounter ratesare based mainly on prey densities; low-rankedprey will enter the diet more frequently as theabundance of higher-ranked prey declines. Hmobility is limited or if the predator popula-tion increases, or if both occur, then resourcedepression of the high-ranked prey is likely tooccur. Basically, the prey choice model suggeststhat human foragers will pursue resources thatprovide the highest energetic returns becausenatural selection favours predators that oper-ate like this, or predators have sufficiently so-phisticated and flexible decision-mAking skills,or both.

Using the prey choice model requires thatfood resources be ranked according to theirprofitability, or their post-encounter return rates.

Based on theoretical modelling and empiricaltesting of prey selection by human and non-human foragers. a number of studies show thatbody size is a good proxy measure for prey rank(Broughton 1994). Up to a certain size. the largerthe animal. the higher the rank. Madsen i:Schmitt (1998) argue that while this relation-ship may hold when individual organisms areprocured. when multiple individuals are cap-tured at one time. as with mass harvesting (e.g.netting). the biomass of the entire group of in-dividuals is the appropriate unit of compari-son. Their work highlights the need to considertechnology in evaluating resource selectionmodels. which I return to below.

In considering the prey choice model andconditions specific to human foragers in par-ticular. I developed a number of predictionsregarding changes in human subsistence for theLower Columbia in the Late Holocene (FIGURE1). During initial phases of human occupation.human foragers exploit the highest-ranked re-sources. As human population size increases.predation pressure increases and abundanceof locally available. high-ranked resources de-clines. Foragers could travel to more distantlocations to procure higher-ranked resources,but several factors make this unlikely. Withoverall declines in mobility and territorial cir-cumscription. foragers may not have access toresources outside their territory. If populationgrowth occurs at a regional scale. high-rankedresources in other areas would be depleted aswell. Finally. even if higher-ranked resourceswere available in distant areas and foragers hadaccess to such resources. costs of transportingresources back to a central base would have tobe weighed against using local. lower-rankedresources. It is predicted that as encounters withhigh-ranked resources decline. human forag-ers would shift to taking lower and lower rankedresources (FIGURE 1). With European contact.infectious diseases are introduced. to whichindigenous people have little resistance andpopulations are drastically reduced. With re-duced human predation pressure. density ofhigher-ranked prey would increase leading toincreased encounter rates with higher-rankedresources and human foragers would begin tomake greater use of these resources.

Given the relationship between body sizeand resource ranking. changes in the contribu-tion of high -and low-ranked resources in hu-

RESOURCE DEPRESSION ON THE NORTHWEST roAST OF NORTH AMERICA 651

FIGURE 2. Location ofsites included instudy, LowerColumbia River, OR(Site 1 :35 CO 3; Site2: 35 CO 5; Site 3:35CO 7; Site 4: 35 MU 1,Site 5: 35 MU 6; Site6:35 MU 9; Site 7:35MU 105; Site 8: 35MU 112).

man diet can be estimated using archaeologi-cal faunal assemblages from dated contexts.Thus the predictions outlined in the model canbe tested. The model is by design very simple,suggesting that prey abundance is controlledby the size of the predator population, withindependent environmental variation playinga limited role in setting prey abundance andencounter rates.

Faunal dataFaunal data used to test the model are fromeight sites located along sloughs and backwa-ter channels adjacent to the Lower ColumbiaRiver in the vicinity of modem-day Portland(OR) (FIGURE 2). Here. the Columbia flowsthrough a low-lying alluvial bottomland; ex-tensive wetlands were located across the val-

ley floor before the region was modified byurban/industrial activities (Ames et a1. 1992).Above the broad valley bottom are numerousridges and dissected plateaux. All of the sitesincluded in the study are located on the valleybottom in a roughly linear pattern over a straight-line distance of about 25 kIn (FIGURE 2). Habi-tats in the area included cold clear rivers,shallow lakes, freshwater marsh, oak woodlands,grasslands and conifer forests that supported anumber of species of migratory and residentfreshwater fish. terrestrial and aquatic mam-mals and waterfowl (Saleeby 1983; Ames eta/.1996).

I chose this area to test the prey choice modelfor several reasons. First, estimates of aborigi-nal population density of the Lower ColumbiaValley are among the highest in North America

652 VIRGINIA L. BUTLER

MerrybellMUg

Odocoileus sp. 2Cervus elaphusOdocoileus/ CervusUrsus american usLynx rofus 2Castor canadensis 4 10 1 3Procyon Jotor 6 32 5 .5Canis sp. 1 4 36 1 3Erithizon dorsatum 1Enhydra lutris 1Phoca vitulina 1 2Vulpes vulpes 3Martes americana 1Lutra canadensis 2 4Ondatra zibethicus 5 g 2 23Mustela vison 3 2 2Sylvilagus bachmani 2ApJodontia rota 2indeterminate large mammal 12 66 212 85 48 343 23indeterminate medium mammal 7 12 1 37indeterminate small mammal 7 2 2 5Ptychocheilus oregonensis 36 5 33 23Mylocheilus caurinus 12 6 2 21 1Acrocheilus alutaceus 3Gila bicolor 2small Cyprinidae 59 4indeterminate Cyprinidae 2 3 16 5Catostomus macrocheilus 72 31 75 4 40 2Cyprinidae/Catostomidae 2 54 23 45 39 8 50 3Thaleichthys pacificus 1 99 21 1Gasterasteus aculeatus 24 147Acipenser sp. 7 12 57 6 113 116 29 77 37 35Oncorhynchussp. 10 7 150 22 127 32 452 3 61

total NISP 34 159 963 55 511 521 257 1436 111 244

1 Only a very small number of mammal remains could be identified below class and the remains are not considered further.

taxon MuJffiS

188721714

9121019

Multnomah I Multnomah 2CO3 MU1 MUl12' CO5 CO7 MU105'

35 160 50 5310 40 14 16 7 "1 3 5 a

202

132

292

i814

TABLE 1. Number of Identified Specimens (NISP) by s{te and component, mammals and fish.

(e.g. Kroeber 1939; Boyd &: Hajda 1987). If de-mographic pressure and human foraging prac-tices caused resource depletion in the PacificNorthwest. it should be evident here. Also. asa consequence of introduced disease. indigenouspeople in the valley suffered enormous popu-lation loss (Boyd 1985). Such a reduction inhuman foraging pressure should allow me todetect rebound in prey population levels. Inaddition. the reported sites are located in verysimilar environments. Thus the kinds of re-sources available to prehistoric foragers wouldhave been comparable and changes in faunalrepresentation should not reflect ecologicaldifferences. Finally. since most of the sites were

excavated and analysed as part of one project,differences seen across assemblages should notreflect archaeological practices.

The faunal data are from two sources. Thefirst is from six sites on or adjacent to SauvieIsland that were tested in the early 1970s m8inlyto produce a cultural chronology (Pettigrew1981) (FIGURE 2). Based on house features,modified objects and faunal indicators, all ofthe sites likely represent year-round residen-tial occupation. Faunal remains were recoveredthrough screening sediments through quarter-inch (6.4-mm) mesh and analysed by Saleeby(1983). The second set of faunal data is fromtwo sites (35 MU 105 and 35 MU 112) located

ltnomMU6

22172

1

ah3007

2164

151

RESOURCE DEPRESSION ON THE NORTHWEST COAST OF NORTH AMERICA 653

site! Menoybell Multnomah 1 Multnomah 2 Mulmomah 3component 600 BC-AD 200 AD 200-1250 AD 1250-1750 AD 1750-1835

MUg XCO3MUIMU 112CO 5 (Strat 4-5)CO 7 {Strat 3-4}MU 105CO 5 (Strat 1-3)MU 6 (Strat 1-7. unit A)

(Strat 1-3, unit B)CO 7 (Strat 1-2}

xxx

xxx

xx

TABLE 2.Chronologicalassignment of sitesand components. x

goods (see also Saleeby 1983). I assigned thetwo Columbia Slough sites to the appropriatephase using radiocarbon dates (TABLE 2).

It is critical for testing the resource depres-sion model to include faunal assemblages dat-ing after population decline, which followedintroduction of European diseases. There is somecontroversy in the region over whether thisoccurred with historically documented face-to-face contact (c. 1750-1775) or centuries be-fore, as a result of pan-hemispheric epidemicsassociated with the earliest entry of Europe-ans into the New World (Boyd 1985; Campbell1989; Dobyns 1987; Ramenofsky 1987). Unfor-tunately. developing a more refined temporalorder for the faunal assemblages that wouldallow one to test for pre-1750 population lossis not feasible for this project. Here. I assumethat Multnomah 3 phase deposits (which aredistinguished from the earlier phase by thepresence of European trade goods) marks thetime of initial disease and appreciate the limi-tations this may impose on testing the model.

Change in resource selectionThere are a number of ways to track changesin faunal resource selection. Drawing on thework of Bayham (1979) and Broughton (1994;1997), I have created indices that calculate aratio of high- to low-ranked resources, with rankassigned based on estimates of body size. Torank the fish taxa, I examined body sizes ofLower Columbia fish identified in the faunalassemblages (TABLE 3) and divided the set into'large' and 'small' based on rather striking dif-ferences in body size. Large taxa includeAcipenser sp. (sturgeon) that routinely attainlengths of several metres and weights of over100 kg and Oncorhynchus sp. (salmon and trout)that are typically longer than 500 mm and weigh

on Columbia Slough. just south of Sauvie Is-land. near the confluence of the Wll1ametle andColumbia Rivers (FIGURE 2). These sites prob-ably represent short-term seasonal camps basedon artefact and feature character (Ellis 1996;1998). Faunal remains. mainly fish. were re-covered from a series of bulk. samples. proc-essed through 1.0-mm mesh. and identified byButler (1996; 1998).

I focused study on the mammal and fish as-semblages. which represent the bulk. of thecollections. TABLE 1 summarizes the taxonomicfrequencies by site and component for sites withmore than one time period represented. A com-mon counting unit was available for all thecollections. number of identified specimens(NISP); given this and the fact that NISP andanother common counting unit. minimumnumber of individuals (MNI). have been shownto provide similar measure of taxonomic fre-quency (Grayson 1984). NISP was used in myanalysis.

Use of quarter-inch mesh at the Sauvie Is-land sites very likely resulted in the loss of smallspecimens. including remains of small-bodiedfishes; thus in comparing taxonomic represen-tation between the Sauvie Island data with thatfrom Columbia Slough. differences in screensize used will have to be considered. On theother hand. at least for the Sauvie Island data.given that the bias was consistent. recovery perS8 would not explain differences seen amongthose site assemblages.

To examine change in resource selection. thefaunal assemblages must be temporally ordered.Here. I assigned the faunal assemblages fromthe Sauvie Island sites (or site components) toone of the four phases constructed by Pettigrew(1981) based on radiocarbon dates. historicaltypes and presence of Euro-American trade

654 VIRGINIA L. BU1tBR

length (mm) phase mammal/small fish index

.88

.69

.36

.79

fish index

MerrybellMultnomah 1Multnomah 2Multnomah 3

800-34001300-2130

750-800480-800450-610610-710250-750300-485

.89

.45

.45

.74

TABLE 4. Indices determined in aggregate by phase.

mal taxa only represent about 2% of the mam-mal assemblage. this should not affect the com-parisons. Remains of small, burrowing rodentsand mole (Scapanus) are likely to be intrusiveand are not included in comparisons.

The first index. referred to here as the mam-mal/small fish index, provides a ratio of largeprey (mAmmals) to mAmmAl and small-bodiedprey (small fishes); the larger the ratio, the greaterthe contribution of large, higher-ranked prey:

210-300160-205150-200305-356200-300125-170

30-75

fish taxon

'large' taxaAcipenser tronsmontanus*A. medirostris*Oncorhynchus tsha~schao. ketoo. kisutchO. nerka*O. mykiss **o. clarki**

'small'taxaPtychocheilus oregonensisMylocheilus caurinusAcrocheilus alutaceusGila bicolorCatostomus macrocheilusThaleichthys pacificusGasterosteus aculeatus**. refers to fork length.. refers to total length

1: NISP mammalTABLE 3. Approximate adult body size of LowerColumbia fish (measure is standard length. unlessotherwise indicated {data from Lee et al. 1980).

I NISP mammal + I NISP small fish

This index was calculated in the aggregate (COIn-bining all faunal assemblages for each phase)as well as for each site or site component in-cluded in the study (TABLE 4, TABLE 5).

To e~ne changes in ~ of fish usedspecifically, I calculated the fish index, whichis the ratio of large fish to large fish and smallfish taxa; as before, the ~the ratio, the greaterthe contribution of large, higher-ranked fish.The ratio

-1: ~-~p l~e fish

1: NISP large fish + 1: NISP small fish

also was calculated in the aggregate (combin-ing all faunal assemblages for each phase) aswell as for each site or site component includedin the study (TABLE 4, TABLE 5).

To repeat expectations, if late Holocene hu-man foragers caused resource depression of high-ranked taxa, we should see a progressive declinein frequency of high-ranked mammals (relativeto small-bodied fish) and a decline in large-bodied, high-ranked fish (relative to lower-ranked fish) until European contact. With contactand decline in human population levels, therecord should show an increased use of high-ranked resources (mammals and large fish) anddecline in use of small, lower-ranked fish.

several kilograms. Historically. o. tshawytscha(chinook salmon) was far and away the mostabundant salmon species in the Columbia Ba-sin (Fulton 1968); this salmonid is much largerthan other species with typical weights over10 kg and some as heavy as 45 kg (Netboy 1980).Because the Oncorhynchus archaeological speci-mens could not be identified to species. theextent to which small forms of Oncorhynchus(e.g. non-migratory forms of O. mykiss. clarki)were used cannot be known from the faunal dataitsel£ However. given the overwhelming predomi-nance ofiarge. migratory salmonids in the region.Native people most likely relied on them. Thelength of 'small' fishes is generally much shorterthan 400 mm (TABLE 3).1\-..'0 species (Catostomusmacrocheilus; Ptyr;hocheilus oregonensis) some-times attain weights of 2-3 kg and lengths of halfa metre (Wydoski &; Whitney 1979). but are usu-ally much smaller than this.

Almost all of the mammal taxa present inthe archaeological assemblages reach muchgreater size than the small fish taxa. Except forthree taxa (see below), mammals typically weighmore than 3 kg and most attain much largersize than this. Only three taxa (Martesamericana. Mustela rison, Ondatro zibethicus)generally weigh less than 1 kg and overlap insize with the small fish. Since these small mam-

RESOURCE DEPRESSION ON THE NORTHWEST COAST OF NORTH AMERICA 655

mammal/small fish index

fish index salmon index sturgeon indexphase

MerrybellMU 9 .88 .89

Multnomah 1CO 3 1.0 1.0MU 1 .69 .47 .40 .20MU 112 -. .11 .11

Multnomah 2CO 5 .45 -42 .10 -37CO 7 .45 .81 .45 -43MU 105 -. .24 .14 -13

Multnomah 3CO 5 .84 -78 -75 -35MU 6 .91 .87 .33 .86CO 7 .41 .52 -41 .28

* index not calculated given extremely small number of mammal remains recovered

TABLE 5. Indices bysite and componentand phase.

0.8

0.6>(u

~.5

0.4

02

FIGURE 3. Plot ofindices, based onaggregate NISP byphase. Solid line:mammal/mammal+small fishindex; broken line:large fish/largefish+small fish.

0

M-ny~ Mllt M1lt2 M1lt3

Were the expectations met? Based on aggre-gated data, trends in the mammal/small fishindex (FIGURE 3) match the expectations, show-ing a progressive decline in use of mammalsin the earliest three phases and then, in thepost-contact phase, an increase in mammal use.The aggregate fish index declines after the ini-tial phase and increases after contact, whichsupports the predictions (FIGURE 3). However,the fish index is similar during the two mid-dle, pre-contact phases, which suggests that afteran initial decline about AD 200, large fish popu-lation size remained relatively consistent un-til European contact.

To examine variation in prey selectionacross sites and time period, the indices werecalculated and plotted for each faunal assem-blage. As shown in FIGURE 4 for the mam-mal/small fish index. most assemblages followthe predicted pattern. with the exception ofone assemblage in the post-contact phase (35CO 7). This site shows the continuing impor-tance of small fish. when the expectation isfor such resources to be dropped from use. Itested to insure that the ratios obtained forthe assemblages were not affected by samplesize; the two variables are not related(r,= -.452, .5<p>.20).

656 VIRGINIA L. BUTLER

..

..xQ)

"0C

.r.I/)il:

"'(ijE(/)-."'(ijEEco~

AA

.4 .

.2

FIGURE 4. Plot ofmammal/mammal+small fishindices by site andphase.

0.0

Merrybell Mult 1 Mult 2 Mult 3

xQ)

"Cc:

..c:C/)

i.i:

FIGURE 5. Plot oflarge fish/largefish+small fishindices by site andphase.Mult 2 Mult 3Merrybell Mutt 1

RESOURCE DEPRESSION ON niB NORTHWEST COAST OF NORTH AMERICA 657

1.0 I~j;~:'!"; ~. Sturgeon Index

* Salmon Index.81

*

,6~xQ)

"Cc:

¥...4 *

J.

-*:

.2~ .i*.

FIGURE 5 plots the fish index for each faunalassemblage and, again, the predicted trends areexpressed, although variation in resource useamong sites in a given time period is some.-times quite high. For example, duringMultnomah 1, indices range from 1.0, at 35 CO3, where large fish were exclusively used, to.11 at 35 MU 112, where small fish absolutelydominate the fish assemblage. Some of the vari-ation in Multnomah 1 and 2 is explained prob-ably by recovery methods. The lowest fishindices (that would indicate greatest represen-tation of small fish) were obtained for sites 35MU 112 and 35 MU 105. These were the onlysites in the study where 1.o-rom mesh was used,mesh size that would favour recovery and rep-resentation of small-bodied fishes. Assemblagesample size and fish index value are not re-lated (r. = -395, .5<p>.20).

Overall, when Acipenser and Oncorhynchusremains are aggregated as 'large' fish, their taxo-nomic frequency increases after contact (FIG-URES 3 8r: 5), which supports the notion thatthese fish populations were rebounding follow-ing human population decline and loweredpredation pressure. However, the picture be-comes more complex when one considers tem-poral variation in faunal abundance of each largetaxon relative to small fish frequency. 1 calcu-lated the Salmon index (1: NISP Oncorhynchus /1: NISP Oncorhynchus + 1: NISP small fish) andthe Sturgeon index (1: NISP Acipenser/1: NISPAcipenser + 1: NISP small fish) for each siteand site component (TABLE 5) to examine inmore detail whether the two taxa behave insimilar ways and according to the model pre-dictions. (I excluded MU 9 from the MerrybellPhase and CO 3 from Multnomah 1 from thecomparison, given very small sample sizes.) Thepre-contact assemblages from Multnomah 1 and2 are dominated by small-bodied fish (FIGURE6 and TABLE 5), which follows the predictedpattern. On the other hand, during Multnomah3 evidence for rebound in the form of high stur-geon and salmon indices are not consistentacross site assemblages. Only at CO 5 is thesalmon index high; only MU 6 provides a highsturgeon index, and at CO 7, as noted above,small fish continue to dominate the assemblage(TABLE 5). These data suggest site-specific func-tional variation in resource use that is not ac-counted for by the simple model considered

0.0

~

,

Mult 1 Mult 2 Mult 3

FIGURE 6. Plot of sturgeon/sturgeon+small fishand salmon/salmon+small fish indices by siteand phase.

here. That said. it should be noted that the in-dices from the post-contact assemblages CO 5and MU 6 are the highest of the sequence, whichis predicted from the model, and suggests hu-man foragers began to target large-bodied, higher-ranked fish in response to changing prey density.

Given the coarse-grained approach taken, thecloseness of fit between the model and the re-sults is remarkable. Unfortunately. the statisticalsignificance of these trends cannot be evalu-ated given the small number of data points andthus the trend cannot be evaluated in a rigor-ous sense. Nonetheless, the results are certainlyprovocative and suggest the need to test withadditional samples.

Other considerationsWhile the results are generally consistent withthe prediction that human exploitation pres-sure caused the decline in abundance of high-ranked prey, prey abundance could also declinebecause of climatic or other environmentalchange independent of cultural factors. Whileclimatic variation associated with the MedievalClimatic Anomaly or Uttle Ice Age may haveoccurred, detailed local records are lacking and

658 VIRGINIA L. BlJ11.ER

from the archaeological record. Evidence formass harvesting. inferred from so-called netweights - girdled. notched or perforated peb-bles and cobbles - is present from sites as oldas 7000-10,000 years in the Columbia Basin(Johnston 1987). Closer to the study area. netweights were recovered from several of the sitesincluded in this study but, unfortunately, thesmall sample size recovered prevents trackingchanges in capture technology used in the re-gion. Overall however, there is no evidence fromthe study area or Columbia Basin for innova-tions in fishing technology in the last 2000 years.

DiscussionMy analysis shows some striking changes inhuman subsistence in the Lower ColumbiaValley in the late prehistoric and early historicperiod. Using the prey choice model, .I havesuggested that decline in use of high-rankedresources like mammals and large fish - stur-geon and salmon - reflect human-caused re-source depression of these prey. Researchersin the Pacific Northwest have rarely consid-ered the possibility that indigenous peopleoverexploited prey populations. One exceptionto this is Hewes (1947; 1973) who asserted thatNative people over-fished salmon stocks. Hebased his thesis on comparative analysis of 19th-century commercial fisheries data (see also Craig&: Hacker 1940). Hewes suggested that excep-tionally large commercial catches on the Co-lumbia in the 1860s were possible becauseaboriginal fishing pressure was much reducedin the early 1800s, as a result of populationcollapse due to introduced European disease.Thus according to Hewes, in the early 1800s.salmonid stocks were in a 'resting period'. (Ina similar way. the effects of over-fishing areinferred from comparing ocean fishery yieldsbefore and after World Wars I and ll. Prior toboth wars, fishery yields were limited and duringthe wars. fishing itself was curtailed. In the threeyears following wartime. fishery yields weresignificantly higher than the pre-war condition,suggesting the abundance was the 'surplus'resulting from reduced wartime fishery(Beverton &: Holt 1957; Cushing 1975).) Fish-ery scientists Craig &. Hacker (1940) also specu-lated that aboriginal fishing methods such asweirs had the potential to reduce severely oreliminate salmon runs. They note ethnohistoricsources for the Columbia Basin, which describe

it is not possible to determine whether mam-mals or large fish would have been differen-tially affected by such changes.

It is also important to consider whether sub-sistence changes are linked to technologicalchange in prey capture. Use of body size as ameasure of prey rank presumes that individu-als are the target rather than a mass of indi-viduals (Madsen & Schmitt 1998). Calling onthe prey choice model, I have suggested thatthe shift from large (high-ranked) to small (low-ranked) prey occurred during late prehistorictimes because of reduced encounters with largerprey, caused by human foraging pressure. How-ever, if technological change occurred, such asmass harvesting of small fish with nets or weirs,then the rank of small prey could be elevatedrelative to larger fish. In other words, humansmay have increased their use of small fish notbecause large fish became scarcer (owing toforaging pressure) but because small fish cap-tured in mass provided higher energetic returns(relative to large fish caught individually).

This explanation is unlikely for several rea-sons. First, it is not consistent with the increaseduse of large fish after European contact. If smallfish taken with nets were targeted because theyprovided greater energetic returns (than largefish taken singly) in late prehistoric times, thenwhy did use of small fish (captured with netsor otherwise) in several instances decline afterEuropean contact as indicated from the archaeo-faunal record? Further, large fish, particularlysalmon and trout, were probably captured atleast some of the time using mass harvestinggear. Ethnohistoric sources emphasize thatColumbia Basin Indians were 'well equippedto catch large quantities of fish' (Craig & Hacker1940: 142), using seines, weirs and other meth-ods. While several fishes were taken using masscapture, most descriptions mention the takingof salmon (Craig & Hacker 1940). By all accounts,indigenous people took advantage of salmonand trout life history, through mass capture offishes that were part of large runs that weremigrating to the spawning ground (Rostlund1952).

The question of technological change in fishcapture could also be examined using the arte-fact record. While it is difficult to link specificprey to technology used to capture them, onecan estimate change in capture technology atsome scale, by examining the modified objects~

RESOURCE DEPRESSION ON THE NORTHWEST COAST OF NORTH AMERICA 859

Emeryville Shell Midden. located near Oakland(CA), he documented a decline in abundanceof large fish, particularly sturgeon. and an in-crease in smaller-bodied fish over time (1997).In high elevation sites in the western Great Basin,Grayson (in press) found a decline in moun-tain sheep and increase in marmot over time,which, he argued reflects depletion of the largermammal by prehistoric hunters.

Results presented here suggest a similar pat-tern of human-induced resource depression forthe Lower Columbia. A number of additionaltasks should be carried out to substantiate thisclaim. Analysis of assemblages from other sitesin the region is needed. particularly those dat-ing between 3000-2000 years ago, for whichthere is very little faunal material to date. Itwill also be necessary to directly measure ex-ploitation pressure, through study of demo-graphic structure of prey populations. Abundantresearch has demonstrated that human preda-tors can alter prey population dynamics, caus-ing reduction in maximum and average size ofprey (e.g. Beverton &: Holt 1957; Moreno et aJ.1984). Measuring skeletal elements (Broughton1997; Butler in press; Casteel 1976) from well-dated contexts provides an indication of changesin demographic structure of exploited populationsthat can be linked to predation pressure.

ConclusionsCalling on the prey choice model, this studypredicted changes in human subsistence pat-terns associated with demographic change inthe Lower Columbia, which were confirmedthrough study of m!L1!Ll!1al and fish remains fromdated contexts. These results are provocative- and certainly suggest that other regional se-

quences be examined for similar patterns. Thisstudy also provides empirical support for theview that Native subsistence systems under-went significant change as a result of Europeancontact. In a very real sense, ethnographic orearly historic descriptions of Native Americanresource use may not reflect practices of dec-ades or centuries before, given the profoundconsequences of depopulation and subsistencereorganization (see also Ames 1991; Broughton1997; Campbell 1989; Cannon 1995; Ford 1989).

This conclusion challenges some commonconceptions about the role of salmon in North-west economies. Ethnohistoric descriptionsoften emphasize the abundance of salmonids

the construction of weirs across tributarystreams; upstream-swimming fish were divertedinto baskets or underwater pens where theycould be easily netted or speared. Such meth-ods could effectively block most of a run on itsway to the spawning ground. They further notethat migrating adult chinook (Oncorhynchustshawytscha) and sockeye (0. nerka) salmonmainly return to their natal spawning ground(there is limited amount of 'wandering' or "stray-ing' of adults); thus it might take many yearsfor new populations to become established inthe event of a spawning population being se-verely depleted.

Schalk (1986; see also Rostlund 1952) haschallenged the notion of the 'resting period',pointing out that reduced predation pressure(of the early 1800s) would actually lead to re-duction in salmonid stock size, not a surplus.Schalk considers that with reduced post-Con-tact aboriginal fishing, more adults would re-turn to the spawning ground than couldsuccessfully spawn; in turn, excessive compe-tition among developing young could lead tofewer individuals actually migrating to sea, andoverall decline in salmonid abundance.Rostlund dismissed out-of-hand Hewes' viewthat aboriginal fishers overexploited salmon bynoting 'there is no evidence' (1952: 17) for suchaction. Importantly, it is hard to imagine thekind of evidence that would have satisfiedRostlund: given significantly reduced post-Contact aboriginal population levels, Nativepeoples would not likely overexploit salmonidstocks, so evidence would not be expected fromethnographic or historic documents. Further-more, evidence from archaeofaunal assemblageswould not be available, given the lack of sys-tematic concern for recovery and analysis offaunal remains half a century ago, whenRostlund carried out his work.

Clearly a primary way explicitly to test thethesis that native populations over-exploitedprey uses the archaeological faunal record. Ina series of faunal studies of coastal and inlandCalifornia sites, Broughton (1994; 1997) hasdocumented striking patterns in subsistencechanges in the late Holocene which, he argues,reflect human-induced resource depression. Atseveral sites along the Sacramento River, hefound an increase in use of small-bodied mam-mals and small, resident freshwater fish overtime (1994). In his faunal analysis of the

VIRGINIA L. Btm.ER660

Plateau, Campbell suggests 'the nearly univer-sal economic specialization on fish emphasizedin descriptions of Plateau subsistence and ad-aptation may be partially an artifact of drasti-cally lowered population levels in the 18th and19th centuries' (1989: 189). She offers that thepre-decline populations would likely have beenmore dispersed across the landscape, utilizinga greater range of habitats and resources, andwere less reliant on salmonids than post-de-cline populations. Her suggestion is certainlysupported here.

This study more generally illustrates theutility of foraging models to understandingsubsistence change at the regional scale. Thelast 25 years has seen a tremendous growth inzooarchaeology and, in many regions of theworld, detailed faunal records are beginningto accumulate from sites excavated for a vari-ety of research or heritage management goals.Notwithstanding issues of assemblage compa-rability, such samples have tremendous valuefor addressing questions of subsistence changeand variation and are particularly useful whencoupled with simple but powerful models fromforaging theory.

and the importance of such fish to NativeAmerican diet (Gibbs 1877). Earlyanthropolo-gists in many areas of the Northwest also viewedsalmon as a major staple - and that countlessaspects of lifeways (complex social organiza-tion, art, over-wintering strategy) were madepossible by the catching, drying and storing ofsalmon (e.g. Post 1938). Archaeologists, in turn,have often presumed that salmon was the main-stay in prehistoric times, often in the absenceof empirical evidence of salmon use (Monks1987). Matson noted, 'given that salmon is themost important resource on the Northwest Coast,the outstanding question is: How did this cometo pass?' (1992: 371). And without agreeing withhis colleagues, Ames noted the general opinionthat 'virtually all workers accept the crucial roleof salmon intensification in socioeconomic evo-lution on the coast' (1994: 211).

Undoubtedly salmonids were important foodresources to Northwest peoples for millenniabefore contact. My study shows, however, thaton the Columbia River, what has often beendescribed as the premier salmon-producing riverin the world, lower-ranked resources (small fish)were becoming an increasingly important re-source during the late Holocene. Further, itsuggests that the ethnohistoric record thatemphasizes salmonids may be a product of re-duced foraging pressure as a result of depopu-lation. Reviewing subsistence issues on the

Acknowledgements. I thank K. Ames. D. Grayson. G. Monksand an anonymous reviewer for their very helpful com-ments on the manuscript and D. Ellis for the opportunityto analyse faunal remains from Columbia Slough.

BoYD. R. T. &. Y.P. HA/DA. 1987. Seasonal population movementalong the Lower Columbia River: the social and ecologi-cal context. American Ethnologist 14: 309-26.

BROUGIfI'ON. J.M. 1994. Late Holocene resource intensificationin the Sacramento Valley. California: the vertebrate evi-dence. Journal of Archaeological Science 21: 501-14.

1997. Widening diet breadth. declining foraging efficiency. andprehistoric harvest p~ure: ichthyofaunal evidence fromthe Emeryville Shellmound, Califumia. Antiquity 71: 84~2.

Blm.ER. V.L. 1996. Fish remains from 35 MU 105. AppendixB. in Data Recovery Excavations at the Columbia SloughSite (35 MU 105). Multnomah County. Oregon. Archaeo-logical Investigations Northwest. Inc.. Report 121.

1998. Fish remains from 35 MU 112. Appendix c. in Cul-tural Resource Survey and Archaeological Test Excava-tions for the Proposed Pacific Gateway Storm SewerExtension Project. Portland. Oregon. Archaeological In-vestigations Northwest. Inc.. Report 159.

In press. Changing fish use on Mangaia. Southern CookIs-lands: resource depression and the prey choice model.International Journal af Osteoarchaeology.

CAMPBELL. S.l(. 1989. Postcolumbian culture history in theNorthern Columbia Plateau: AD 1506-1900. Ph.D disser-tation. University of Washington. Seattle (WA).

CANNON. A. 1995. The rattish and marine resource deficien-cies on the Northwest Coast. Canadian Journal of Archae-ology 19: 49-60.

ReferencesAMES. K.M. 1991. The archaeology of the longue duree: temporal

and spatial scale in the evolution of social complexity onthe southern Northwest Coest. Antjquity65: 935-45.

1994. The Northwest Coest: complex hunter-gatherers. ecol-ogy. and social evolution. Annual Reviews in Anthropology23: 209-29.

AMES. K.M.. W.L. CoRNETT & S.G. HAMILTOO. 1996. Archaeologi-cal Investigations (1991-1995) at 45CL1 (Cathlopotle): Clark

County Washington. Wapato Valley AlCbaeological ProjectReport Numbe!' 6, Dep8rtlnent of Anthropology, PortlandState University and US Fish and Wlldlife SeNice.

AMES. K.M. & H.G. MASCHNER. 1999. Peoples of the NorthwestCoast: their archaeology and prehistory. London: Thamesand Hudson.

AMES. K.M., D.F. RAKl'Z. S. HAMILTON .. C. McAFEE. 1992. House-

hold archaeology of a southern northwest coast plankhouse. Journal of Field Archaeology 19: 275-90.

BAYHAM. F.E. 1979. Factors influencing the archaic pattern ofanimal utilization. Kiva 44: 219-35.

BEVERTON. R.J.H. & S.J. HOLT. 1957. On the dynamics of ex-ploited fish populations. London: HMSO.

BoTKIN, S. 1980. Effects of human exploitation on shellfishpopulations at Malibu Creek. California. in Earle &Christenson (ed.): 121-39.

BoYD. R. T. 1985. The introduction of infectious d~ amongthe Indians of the Pacific Northwest. 1774-1874. Ph.Ddissertation, University of Washington. Seattle (WA).

RESOURCE DEPRESSION ON THE NORTHWEST COAST OF NORTH AMERICA 861

~. R. T. 1987. An:h88O1ogical evidence of fishing in theSouthern Plateau. a cultural area of the Columbia Pla-teau. Matera th..ia. Univenity of Idaho. Moscow (W).

KI(DIa. A.L 1939. CuJtum1 and natwuI - of Natiw NorthAmerica. Berkeley (CA): Unlvenity of California Pr.-.Publications in American An:haeology . Ethnolosy 38.

LIB. D.S.. C.R. GILBERT. C.H. HOCUTT. R.E. JINKINS. D.E.McAWSTER . J.R. STAUPPD. JI. 1980. Atl06 of NorthAmerican freshwuw fisha. Raleigh (NC): North Car0-lina Biological Surrey. Publication 1980-12.

MADSEN. D.B. . D.N. SaDArrr. 1998. Mass collecting and thediet breadth model: A Great Basin example. Journal ofArchaeoloBical Science 25: 445-55.

MA~. R.G. 1992. The nolution of northwM cout sub8iat-8OC8. in R_n:h in Economic Anthropology. Suppl..ment B: 367-428. Gmenwich (CT): JAI Preu.

MATSON. R.G. . G. CoUPLAND. 1995. The prehi6tory of theNorth-.t Coa6t. New York (NY): Academic Pr.-.

MOND. G.G. 1987. Prey u bait: the Deep Bay example. CQno-dian Journal of An:hoeology 11: 11&-42.

~. CA.. J.P. S\muLApC). H.F. JARA. 1984. Man 88a predatmin the intertidal mna of southern Chile. Oikos 42: 15~.

NAGAOKA. L 1998. ReIOun:e depletion. extinction. and sub-siatence chaop in southern New Zeeland. Poster p..seated at the 63rd Annual Meeting of the Society forAmerican Archaeology Seattle.

NETBOY. A. 1960. The Columbio River salmon and stBBlheadtrout: their fight for .wvivoJ. Seattle (WA): University ofWuhlngton Pr.-.

O'CONNELl.. J.F. A K. HAWKES. 1961. Alyawara plant use andoptimal foraging theory. in B. Wlnterhalder. E.A. Smith(ed.). Hunter-gath_r foroging 6trotegJes: ethnogrophicand archeological anolyses: 99-125. Chicago (IL): Unl-venlty of ChiC880 Pr8a.

PETTIGREW. R.M. 1981. A prehi6toric culture sequence in thePortland Basin of the Lower Columbia Valley. Eugene (OR):Univenlty of 0f8I0D. Anthropological Paper 22.

PmT. R. 1938. The subliatence quat. in L Spier (ed.). TheSiDkaJetk or 6Outhem Okanogan of Wa8hington: 11-34.Menuha (WI): Gearse Banta Publishlns Co.

RAMENOFSKY. A.F. 1967. Vectors of death: The archoeology ofEuropean contact. AlbuqueRIue (NM): University of NewMexIco Pr8S8.

RosTLUND. E. 1952. F,...hwuter fiM and fishing in native NorthAnlerico. Berkeley (CA): Unlvenity of Callfomia Press.

SALEEBY. B. 1963. Prehiatoric settlement patterns in the Port.!aDd Buin of the Lower Colwobla: ethnohi8toric. an:haeo-l~ and bi~ per8p.:;tjvs. Ph.D ~oo.Unlvenlty of Oreson. Eugene (OR).

Schalk. R.F. 1986. Estimating salmon and steelhead u8888 inthe Columbia Buln before 1850: the anthropologicalperspective. North-.t Environmental Joumal 2: 1-29.

~. D. W. 1995. Prehiatoric extinctions of Pacific lalandbirda: blodivenity meets ZOO8rcha8O1ogy. Science 267:1123-31.

STIPHINS. D.W. AlA KREBS. 1986. Foroging theory. Princeton(NJ): Princeton University Pr-..

Sm8. M.C. N.D.~. T.A. 5umYEIJ.. E. TatDt«)Y A O. BAa-Y~. 1999. PaIeoliIbk: populatkIn powd1 PU!- evid8lcedby small animal populations. Science 283: 190-94.

WYDOSKI. R.S. . R.R. WInTNEY. 1979. Inland fi.h.. of Wash-inItOII. Sawe (WA): University of Washington Pr8a.

CASTEEl.. R.W. 1918. FI.h remain. in archaeology. New York(NY): Academic Preu.

QIRISTINS(»I. A.L 1980. Change in the human food niche innepo- to population growth. in Earle. (1Irlsteoson:31-72.

CoHEN. M.N. 1981. PlClfic cout forasers: amuent or over-crowded? in S. Koyama. D.H. Thomu (ed.). Affluentfonlptl: Pacific CoaIt8 Ea.t and West: 215-95. Osaka:na1iooal M- ofEthDO1osy. Seori Ethno~ Studiee9.

CRAx;. J.A.. R.L HAcxn. 1940. The history and developmentof the fishedel of the Columbia River. Bulletin US Bu-reau of Fi.heri.. 49: 133-218.

CIoEs. D.L 1995. The Hoko BiRr an:hoeolO8icaJ 8iie complex.Pullman (WA): Wuhlngton State University Prwa.

CUSHING. D.H. 1915. Marine ecology and fi.heri... Cambridge:Cambridge University Prwa.

()JBYNS. H. 1983. Their numben became thinned. Knoxville(TN): Univssity of T- Prwa.

EARLE. T.L . A.L 0IRIsTENS0N (ed.). 1980. Modelins chon,.in prehistoric IUbli6tence economie.. New York (NY):Academic Prwa.

Ews. D. V. 1998. Data r-=overy excavatioDl at the ColwnbiaSloup Site (35 MU 105). MultDomah County. Oregon.An:baeologicailuVeltiptioD8 Northwest. Inc.. Repon 121.

1998. Cultural Resource Survey and Archaeological TestExcavations for the ~ Pacific c.teway Stonn SewerExt8DIlon Project. Portland. arwgon. Archaeological 10-vwtiptions Northwest. Inc.. Report 159.

FORD. P.J. 1989. Archaeological and ethnographic correlal.. ofHelonailly: problema and solutions on Ihe NorthweslCo8It. Canadian Journal af Archaeology 13: 133-50.

FULTt:»f. LA. 1968. Spawning.,... and abundance ofChiDookSalmon (Oncorhynchus tshalOYf8cha) 10 the ColumbiaRiver Buio - pul and present. US Fish and WildlifeService Special Scientific Report. Fisheries No 511.

GIBBS. G. 1811. TribM of western Washington and Northw..tanson. Contributionl to North American Erhnalogy 1(2):151-381.

GRAYSON. D.K. 1984. Quantitotive zooarchoeo/ogy. New York(NY): Academic Preu.

In Preel. The A1ta Toquima vertebrale fauna. in D.H. Tho-DIU (ed.). The archaeology of Monitor Valley 4: AltoToquima and the Mount Jefferson Complex. New York(NY): American Mu.eum of Natural Hillory. Anthropo-logical Papers.

GRAYSON. D.L . M.D. CANN(»I. 1999. Human pa18OKO10lYand forasiD8 theory in the Greal Buin. In C. Beck (ed.).Model. for the millennium: Great Ba8in anthropologytoday: 141-51. SaIl Lake City (trr): Un1versity of Utah.

GREMILLION. K.J. 1998. Diffusion and adoption of crops In evo-lutioaery perspective. /oomaI of AnthropolO8icol ArclIM-ology 15: 183-204.

HIWBS.G.W.1941.AborIsInaIUMofftahery~inNorth-western North America. Ph.D. dissertation. Universllyof California. Berkeley (CA).

1913. indian fish.rt. productivity In pM-CODtact Iim8Inthe PKific Salmon Area. Northwest AnthropologicolR~ Not.. 1:133-55.

HOLDAWAY. R.N. . c. JAroMa. 2000. Repid extinction of theMou (Av..:DInomithiform..): model. 1..1. and impllca-tiODl. Science 281: 2250-53.