In, Harmes Rockshelter:
A Final Report on 11,000
gul~~l.
usetorent
A. H3.CkS, ea.);
pp. 320-337,
state Univ. Press, Pullman.
Years of
Washington
II
FISH REMAINS
Virginia L. Butler
current study,manyof the bulk sampleswere
sorted(GustafsonandWegener1998)and found
to include fish remainsalongwith other cultural
material.
Stratigraphicunit information was usedin
this analysiswhereavailable. In manycases,
however,field assignments
were not made.
Using depth,spatialcoordinatesandredrawn
stratigraphicprofiles for reference,Hicks
assignedthesematerialsto stratigraphicunits
and suggeststheserepresentonly provisional
recordsof vertical positionbecausethey arenot
basedon field observations(seeChapterTwo).
Remainswereidentified to the finest
taxonomiclevel using Butler's comparative
collectionsfrom the ColumbiaBasin and
adjacentregionsin westernNorth America.
Becauseof their very small size,most of the
floodplain faunalremainswereidentified using
low power magnification(10-2Ox).FlShremains
were quantifiedusingNumberof Identified
Specimens(NISP) (Grayson1984). For
Cypriniformes(minnow and sucker)vertebrae,
the first two and last vertebraeon the column
were distinguished;definition of abdominaland
caudalvertebrae,which representmost vertebrae
on the column,follows Wheelerand Jones
(1989). Salmonidvertebraewere assignedto
four categoriesbasedon morphological
differenceassociatedwith position on column
(Butler 1993).
To estimatevariationin body sizeof fishes
represented.the width (measurelabeledas
diameter,shownin Casteel(1976:84)was
measuredusingdigital calipersor a micrometer
within the microscope.
All of the specimenswereexaminedfor
evidenceof burning. For severalof the
specimensfrom the floodplain, it wasdifficult to
determinewhethera dark color reflectedburning
or stainingthat might result from absorptionof
mineralsin the surroundingmatrix. The decision
to call a specimenburnedwasbasedon
conservativecriteria: only thosespecimens
which wereuniformly black or calcined
(white/blueishcast)werecalledburned.
Severalregionalsyntheses(e.g.,Ames et a1.
1998;Schalkand Cleveland1983)havenoted
the presenceof fish remainsat Marmes
Rockshelter.However,until the currentproject,
fish bonesfrom the site had not beenstudiedin
any detail. Given that many questionsabout
regionalhumansubsistencechangeturn on
establishingthe relative importanceof fish vs.
terrestrialmammals,the changingrole of
salmon,and whetherHoloceneenvironmental
changeaffectedsalmonabundanceand
distribution,studyingthe fish remainsfrom this
early site is extremelyworthwhile. The recovery
of fish remainsfrom the MarmesRockshelter
and adjacentfloodplain providesan opportunity
to exploresomeof thesequestions.The remains
describedrepresentsomeof the earliest,well
documentedarchaeologicalfish remainsfrom the
ColumbiaPlateauand as such,provide a very
importantrecordof fishesin regional streamsas
well ashumanresourceusepatterns.
Methods and Materials
The fish remainscollectedfrom the rockshelter
andfloodplain areasof the site were sampledin
very different ways. Remainsfrom the
rockshelterwererecoveredduring all four of the
1960sexcavationfield seasons.Excavated
matrix wassieved'dry, throughlA-inchmesh
screens. However,during 1968,sedimentsin
the upper stratawere quickly removedandnot
screenedandfaunal recoverywaslimited. Given
theserecoveryconcernsas well astime and
budgetaryconstraints,only a sampleof the fish
remainsfrom the rockshelterwere selectedfor
study(seeChapterTwo, Samplingfor
descriptionof the sample).
At the floodplain locale, detailedexcavation
focusedonly on the Late Pleistocene/early
Holocenesediments.Larger artifactsandfaunal
itemsrecoveredby the excavatorswerebagged
separately,but most of the excavatedsediment
waswater screenedthrough I-rom mesh. The
screenresidueretainedfrom eachunit level was
dried andbaggedasa bulk sample. For the
319
Descriptive Summary of Fish Remains
speciesbearsa seriesof horizontalstruts,rather
thanthe fenestrationscharacteristicof
Oncorhynchus.A secondspeciesof Prosopium
known for the Columbiabasin,P. coulteri
(Pygmy whitefish), reachesmuch smalleradult
sizethan P. williamsoni. The archaeological
specimenwasassignedto P. williamsoni based
on its large size.
Two vertebrae(inventory numbers18987
and 18953)of very similar shapeand sizecould
not be identified below the family level. They
arefrom a small salmonid (vertebradiameteris
about 1.5 mm). As an indication of body size,a
modemfish of Salvelinusmallna (Dolly Varden)
measuring216 mm in total length hasvertebra
type-3 with widths that rangebetween2.5 and
3.0 mm, suggestingthat the archaeological
samplesare from fish smallerthan200 mm.
Both vertebraehave very largeopeningsfor the
notochord(diameterof openingis about.5 mm),
which is not characteristicof Oncorhynchusor
Prosopium. The specimensmostclosely match
Salvelinus,which includesthe Bull trout, S.
confluentus,andDolly Varden,S. mallna.
However,the notochordC?pening
for the
archaeologicalspecimensis largerthan that
found on availablecomparativespecimens,and
thus the materialswere assignedto the family
level.
Historically, the SnakeRiver and several
tnoutary rivers and streamsprovidedextensive
spawninghabitatfor migratory speciesof
Oncorhynchus,particularly O. tshawytscha
(chinook), o. nerka (sockeye),and o. mykiss
(steelheador rainbow trout) (parkhurst1950).
Vast numbersof spring and summerruns of
chinook usedthe SnakeRiver upriver from the
PalouseRiver as a passag~way to spawning
groundsin tributary rivers and streams;the fall
run chinook usedextensivespawninghabitatin
the main stemSnakeRiver betweenthe Palouse
River and Hells Canyon(Fulton 1968).
According to fishery documents,the Palouse
River itself, however,was, at leasthistorically,
not usedby spawningsalmon(Fulton 1968). In
his fisheriessurveyand historic overview of
SnakeRiver Basin anadromousfishes,Parkhurst
(1950) did not surveythe PalouseRiver. He
notesthat a high falls (PalouseFalls) located
about 10 kID abovethe mouth "rendersthe
StreaDlinaccessibleto migratory fish" (1950:5).
Ethnographicrecordsshowthat salmon
fishing was extremelyproductiveat the
confluenceof the Palouseand SnakeRivers (Ray
1975). Lewis and Clark andlater explorers
describea very large village at the mouth of the
The following information describesthe criteria
usedin assigningthe specimensto taxonomic
categoryaswell asinformation on ecologyand
habitats. Descriptionsareprovided for each
order of fishesidentified. Appendix N presents
the analysisdata.
Class Osteichthyes. Bony Fishes.
Order Acinenseriformes
Family Acipenseridae- sturgeons
Acipensersp.- sturgeon
Materials: Rockshelter: 1 scote.
Remarks: Two speciesof sturgeonare known
for westernNorth America.A. medirostris(green
sturgeon)andA transmontanus(white
sturgeon). Greensturgeonare known today only
from brackishwatersof the lower 40 miles of the
Columbiaandfully marineenvironments.while
white sturgeonare documentedthroughoutthe
river system.including the upper Columbiaand
SnakeRivers. Basedon habitatpreferenceand
moderndistribution,the scute(or bony body
scale)is mostprobablyfrom the white sturgeon.
Order Salmoniformes
Family Salmonidae-Salmons,Trouts,and
Whitefish
Materials: Floodplain: 1 indeterminatevertebra
type, 2 vertebratype-3.
Oncorhynchussp.- Salmonand Trout
Materials: Rockshelter: I basipterygium.2
vertebratype-lor atlas,13 vertebratype-2, 10
vertebratype-3,2 vertebratype-4, I
indetemlinatevertebratype, 65 vertebra
fragments: 94 specimens.
Floodplain: 2 gill rakers,10 vertebrafragments:
12 specimens
Prosopiumwilliamsoni-Mountain Whitefish
Materials:Floodplain: 1 vertebratype-3
Remarks: Thereareten speciesof salmon.trout
and whitefish with recordsfor the Columbia
basinupriver of the estuary. The genus
Oncorhynchusis representedby six speciesof
anadromousandresidentforms. The
Oncorhynchusspecimensfrom the site were
assignedbasedon their large size,or, for the
vertebrae,distinctive shapeand fenestration
pattern. A singlevertebrawas assignedto
ProsopiumwiUiamsoni;the centrumof this
320
PalouseRjver. RossCox. who spenttime in the
village around 1812,notedthat in early August,
peoplethere wereengagedin catchingand
drying salmonin largenumbers(Ray 1975).
Historic documentsalso show fishing campsand
villages along the PalouseRiver itself (Ray
1975). Ray refers to two suchlocalesin
particular: A'patap, which was locatedat the
foot of PalouseFalls, and C/axo'pa,aboutfour
miles abovethe mouth. Unfortunately,the
documentsdo not indicatewhetherthe Native
American fishery alongthe PalouseRjver
targetedresidentfreshwaterfishesor
anadromoussalmonandtrout.
The falls havebeena barrier to migratory
salmonidsat leastsincethe late Pleistoceneand
Holoceneperiods,thusit is clear that the Palouse
Rjver itself was not a passagewayfor fish
migrating to headwaterareasto spawn. It is
possibleof coursethat the PalouseRjver channel
below the falls wasusedby salmonfor spawning
and thus a fishery might have developedto target
suchfishes. One would needto evaluatethe
potential streamconditions(bedmorphology,
temperature,flow patterns)to establishwhether
the lower Palouseonceprovided suitable
spawninghabitat. For now, the pre-damrecords
for fish distribution leadto the conclusionthat
the PalouseRjver neversupportedlarge
saImonidpopulations.
Floodplain: 5 dentaries,3 pharyngea1s:8
specimens.
Acrocheilusalutaceus--Chiselmouth
Materials: Rockshelter: 1 dentary,2
hyomandibolae,1 prootic.
Floodplain: 1 dentary,1 pharyngeal.
Mylocheiluscaurinus-Peamouth
Materials: Floodplain: 5 pharyngeals.
Richardsoniusbalteatus-Redside shiner
Materials: Floodplain: 1 pharyngeal.
Rhinichthyssp.-Dace
Materials: Floodplain: 2 PharyngeaIs.
Rhinic hthy sIRichardsonius
Materials: Floodplain: 2 pharyngeals.
PamilyCatostomidae
Cawstomussp.-suckers
Materials: Rockshelter:1 basioccipital, 1
exoccipital. 1 hyomandibula.3 endopterygoids,3
metapterygoids,1 palatine,1 parasphenoid,1
parietal. 1 preopercle,2 prootics, 1 sphenotic,1
urohyal, 1 vomer, 9 cleithra,2 coracoids,5
Pharyngeals,3 lit vertebrae,6 2ndvertebrae:42
specimens.
Floodplain: 3 basioccipitals,3 ceratohyals,2
dentaries,1 epihyal, 1 hyomandibula.1
metapterygoid,2 opercles,1 palatine,1 quadrate,
1 urohyal, 1 vomer, 8 cleithra. 15 pharyngeals
andpharyngealteeth. 15 111
vertebrae,S28d
vertebrae: 60 specimens.
Order Cvorinifonnes
Family Cyprinidae-Minnows
Materials: Rockshelter:1 basioccipital,1
basiphenoid,1 exoccipital, 1 interopercle,4
endopterygoids,3 metapterygoids,3 opercles,3
parasphenoids,3 preopercles,1 prootic, 1
pterosphenoid,6 cleithra, 1 coracoid,3
C. macrocheilus-Largescalesucker
Materials: Rockshelter: 2 dentaries,1
hyomandibula,2 maxillae, 3 palatines,1
quadrate:9 specimens.
Floodplain: 1 maxilla.
basiptergia,
11pl1aryngeals,
5 1It vertebrae
(atlas),4 ZIK1
vertebrae, 4 abdominal vertebrae:
56 specimens.
Floodplain: 3 articulars,Z basioccipitals,Z
ceratohyais.Z epihyals,1 hyomandibula,4
opercles,1 parasphenoid,1 pterotic, Z quadrates,
1 urohyals,6 cleithra,97 pharyngealsand
pharyngealteeth. 17 lit vertebrae(atlas),9 ZIK1
vertebrae,11 1- or ZIK1
vertebrae,Z abdominal
vertebra: 161 specimens.
C. columbianus-Bridgelip
Materials: Floodplain: 1 maxilla.
sucker
Cyp rini daeJCato sto mi dae
Materials: Rockshelter: 105 abdominal
vertebrae, 66 caudal vertebrae, 3 indetenninate
vertebra type, 11 vertebra fragments: 185
specimens.
Floodplain: 2 hyomandibulae, 2 scapulae, 5
pharyngeals and pharyngeal teeth, 5 111or 21M!
vertebrae, 418 abdominal vertebrae, 374 caudal
vertebrae, 14 ultimate vertebrae, 24
indeterminate vertebrae type, 659 vertebra
fragments: 1,503 specimens.
Ptychocheilusoregonensis-Northem Pike
Minnow
Materials: Rockshelter:2 basioccitaIs.4
ceratohya1s.1 dentary.8 hyomandibulae.2
maxillae. 1 palatine. 1 urohyal. 1 vomer.9
Pharyngeals:29 specimens.
321
occursin shallowwateralong the downstream
endsof pools.
Remarks: Nine species of Cyprinidae are known
historically for the Columbia Basin (Lee et al.
1980). Several elements of the jaw, pharyngeal
arch (toothed bone at the rear of the tnouth) and
lateral face are extremely distinctive and species
or generic assignments of sufficiently complete
specimens was possible. Six species of
Catostomus are known for the Columbia-Snake
River Basin (C. macrocheilus, C. columbianus,
C. platyrhynchus, C. catostO11ULS',
C. discobolus,
C. ardens). The specimens assigned to species
very closely matched the reference material
However, reference material was lacking for C.
discobolus (known in the Snake River only
above Great Falls) and C. ardens (found in the
Snake River above Shoshone Falls) so the
assignment of remains to C. macrocheilus and C.
columbianus is somewhat provisional.
Except for the first and second vertebra on
the column. which can be distinguished as
Cyprinidae or Catostomus, vertebrae from these
taxonomic groups cannot be distinguished, so the
joint category, Cyprinidae/Catostomidae. was
used. In a few cases. abdominal vertebrae were
fused to the second vertebra and these abdominal
vertebrae could be identified to a finer taxonomic
level Finally. several extremely eroded cranial
and postcranial specimens could not be identified
more precisely than Cyprinidae/Catostomidae.
Fishes in the Cypriniformes order occupy a
wide range of habitats but in general terms tend
to occupy more slowly moving. warmer waters
than saImonids. Of the cyprinids identified at
Marmes. Ptychocheilus oregonensis (northern
pike minnow) favors slow to moderate currents
in streams and prefers the warmest temperatures
in the waters it occupies (Wydoski and Whitney
1979). Acroclzeilus alutaceus (chiselmouth) also
prefers warmer areas of streams in moderately
fast to fast moving waters (Wydoski and
Whitney 1979). Most studies of Mylocheilus
caurinus (peamouth) have focused on lake
populations. In Lake Washington (Seattle, W A).
peamouth tend to occupy the warmest water.
favoring deep water during the winter and
moving inshore during spring and summer
(Wydoski and Whitney 1979).
Catostomus species are bottom fishes.
feeding on algae or bottom dwelling
invertebrates. Both C. macrocheilus and C.
columbianus occupy quiet areas in the
backwaters or edges of the main current of
streams (Wydoski and Whitney 1979). During
spawning seasonof largescale sucker. which
occurs usually in April or May. large schools are
found along river edges. Spawning usually
Results
Before reviewing resultsof analysis,it is
importantto considerseveralaspectsof
archaeologicalrecoveryand curationthat have
biasedthe fish assemblage
availablefor this
study. Early reportson excavationprocedures
note that lA-inchmeshscreenswereusedduring
eachof the four field seasonsat the site
(Gustafson1972). The effectsof scretn size on
fish faunal recoveryin the rocksheltercannotbe
directly documentedgiven that bulk samplesare
not availablefor analysis. However,the
presenceof very smaUfish remainsfrom the
floodplain bulk samples(seebelow) and
abundantresearchelsewhere(Butler 1987a;
Casteel1972;Gordon1993),suggeststhat useof
large meshscreensin the rockshelterhasbiased
the sample'infavor of relatively large-bodied
fishes. Besidesscreensizebias, however,which
canbe estimatedto someextent,other recovery
practicessuggestthe collectedsamplemay bear
a very poor relationshipto the targetpopulation
of fish remainsin the rockshelterdeposit. An
early documentdescribingfield proceduresnotes
that "not all bonefragmentswere savedduring
the early seasonsof excavation"(Gustafson
1972:54). Unfortunately,field recordsare not
availablethat indicatethe extentof this practice
or which excavatedareasareparticularly
affected,thusit is difficult to control for this
problem.
Also, therearediscrepanciesbetweenthis
author'srecordsof fish taxa and relative
abun~ce and thoseof previousresearchers
(Gustafson1972),which suggestthat the sample
availablefor this studymay be different from
that originally excavated.Gustafsonnotes
" Salmonidvertebraeand other fish remains
sometimesareabundant(particularly in the
storagepit areas-Units VI and Vll)" (1972:
106). As notedbelow, salmonidremainsin the
recently analyzedsamplearemost commonin
StratumV, wherethey representover 60% of the
fish fauna;in StratumVI, a single salmonid
specimenwasidentified and in Stratumvll,
about 15%of the fish remainsare from
salmonids(emphasismine). In addition, thereis
a discrepancyin the speciesreportedin
Gustafson(1972) andidentified in the current
study sample. Gustafsonsenta sample(from
unknownprovenience)of fish remainsto
322
field recoverylossor curatoriallossacrossthe
site deposits,interpretationsaboutchanging
frequenciesof fish taxa within the site will be
tentative,sincethe changescould result from
samplingproblems.
William Taylor (U.S. National Museum)who
identified two speciespositively: PtyChocheilus
oregonense(= oregonensis)andMylocheilus
caurinus. This authoridentified Ptychocheilus
in the Rocksheltersampleand otba cyprinids,
but not Mylocheilus. Mylocheilusbasextremely
distinctivejaw and pharyngealmorphology;it is
unlikely that the discrepancyreflects analyst
identificationem>r. Given all of the other
documentedlosses-both of specimensand
provenience-associatedwith the rockshelter
assemblage
(Hicks andMoura 1998).it appears
likely that the differencesin reporting result from
the currentstudy's samplenot incl00ing all of
the fish remainsthat Taylor and Gustafson
examined.It appearsthat parts of the recovered
fish assemblage
havebecomemisplacedor lost
over the 40+ yearssincethe inceptionof the
project.
In sum.given the d~nted
and indirectly
suggestedbiasesassociatedwith the fish bone
samplefor this study. it would be unwiseto
ass~ the materialsreportedon hereare
representativeof the fish bonesin the rockshelter
deposits.In termsof quantification.it is most
appropriateto treat the fish assemblageat the
nominalscaleonly ratherthan ordinal or ranked
scale. That is. it is bestto view the fish recordas
a list of taxapresentratherthan useNISP values
to examinerelative taxonomicimportance.
Similarly. given that thereis no control over
Rockshelter Fish Remains
A total of 688 fish remainswereidentified from
the rockshelter(Table 11.1). About 60% of
these(420 specimens)could be identified to a
taxonomiclevel below fish andeight taxa are
represented.Freshwaterminnowsand suckers
(Cyprinidae,Catostomidae)dominatethe
assemblage.Ptychocheilusoregonensis
(northernpike minnow) is the dominant
freshwaterspeciesrepresented;Catostomus
macrocheilus(largescalesucker)is the sole
speciesof suckerpresent Large-bodied
salmonids(Oncorhynchus),alIOOst
certainly
from migratory runs,represent13.7%of the
collection and a singlespecimenfrom sturgeon
is present As notedabove,it is problematicto
ass~ the relative abundanceof fish taxa in the
rockshelteris representativeof the target
populationof fish remainsin the site deposits.
For now and until bettercontrol of the biasesis
achieved,it is bestto treatthesedataat the
nominal scale.
Table 11.1Frequency of fish remains by taxon, rockshelter.
Taxon
Rockshelter
Acipenser sp.
Oncorhynchus sp.
Cyprinidae
Ptychocheilus oregonensis
Acrocheilus alutaceus
Catostomus sp.
Catostomus macrocheilus
CyprinidaelCat 0stomidae
Unidentifiable
Total
323
J
to arguethat cultural agentsburnedthe bonein
question. It is of coursepossiblethat evenif
cultural practiceis responsiblefor the burning,
the burningpostdatesthe naturaldepositionof
faunal remains. Theseconcernsare moot for the
Rockshelterfish remains;only one specimen
from the rockshelter(InventoryNo. 9199,in
StratumI) wasburned,sothis in itself doesnot
provide particular supporteitherway.
Another potential way to examinefish bone
origin and siteformation processeswould
examinespatialdistribution of fish bonesrelative
to other classesof animal remainsand clearly
modified objectsor featurecontext. If fish bone
abundancevaried in concertwith abundanceor
distribution of modified objectsor feature
context,then it would be possibleto arguefor a
commonsource,humans. Unfortunately,given
the variation in collection proceduresacross
excavationunits and strata,it will be difficult to
usefish boneabundanceanddistribution to
evaluatetaphonomicquestions.
In sum,while a definitive conclusionabout
the origin of the fish bonesin the rockshelteris
not possible,severalfactorsand conditions
supportthe humanrole in fonning the fish
deposit,including the lack of carnivoredamage
on the sampleexamined,the lack of discussion
of carnivoredamagein previousfaunal reports
(Gustafson1972;GustafsonandWegener1998),
and the presenceof fish taxa that were likely not
availablein the nearbyriver.
Taphonomy and Origin of Fish Remains. In
evaluatingthe significanceand meaningof a
faunal assemblage
in cultural terms,it is first
necessaryto establishthat the remainsin fact
result from humanactivity (Grayson 1991;
Lyman 1994). Geologicalstudy of the site
matrix suggeststhat the sourceof the site
depositsis primarily eolian, with endogenous
roof fall andhumanoccupationaldebris
contributingmatrix as well (seeChapterFive).
Thereis no evidencefor fluvial depositionin
StratumI andabove,so the possibility that fish
remainsrepresentflood rafted carcassescan be
eliminated. Potentiallynonhumanterrestrialor
avian scavengers
or predatorsof fish could have
broughtwhole fish or fish parts to the
rockshelter. It is unlikely that the salmon
(Oncorhynchussp.) remainsenteredthe
rockshelterthis way, however. As noted
previously,the PalouseRiver probably never
supportedaoadromoussalmonruns; the nearest
sourceof thesefish would be over a mile away
in the SnakeRiver. It is unlikely that nonhuman
scavengerswould havetransportedthe salmonid
carcasses
or partsthis distance. Perhaps
significantly, Gustafson(1972) recorded
specimensfrom Canis_cf.latrans (probably
coyote)in eachof the rockshelterstrata
(1972:Table5.1), which suggeststhat at least
one carnivorethat is known to eat fish (Butler
and Schroeder1998)lived in the vicinity of the
rockshelter. However,Gustafsondoesnot note
any evidencefor carnivoredamageon the large
mammalbones(Gustafson1972;Gustafsonand
Wegener1998),although,the absenceof this
observationmay indicatethis surfaceattribute
wasnot examined,ratherthan a real absenceof
carnivoredamB.ge.Time and budgetary
constraintspreventedcarrying out a detailed
study of surfacemodification of fish bonefor the
currentstudy. However,a small grab sampleof
fish remains(about 15 specimens)wasexamined
under 10-20power magnification anddid not
reveal any patternsindicative of digestive
process(Butler and Schroeder
1998-dark
staining,rounding,erosion,vertebra
compression).One specimenshowed'signof
rodent gnawingbut noneexhibited sign of
carnivoreprocessing.
Sometimesburning is usedto link a faunal
assemblage
to humanactivity (Balme 1980;
Butler 1990). Burnedbonecanresult from
naturalfires, but if widespreadburning is not
indicated(multiple artifactsand sedimentsdo not
showevidenceof burning), then it is reasonable
Intra-site variation. Table 11.2showsthe
frequencyof taxa acrossthe 14 strata(and
aggregatestrataas assignedin the field and
laboratory)in the rockshelter. Table 11.3shows
the freqlIencyof taxa for the threestratumwits
that containedmost of the fish remains-V, VI.
and vll. For temporalreference,M3Z~~ tephra
underliesStratumV, which is a mixed deposit
containingsomeM3Z3matephraalong with
other windblown sedimentsandprovided a
single radiocarbondate(4250:f:300 B. P.). A
numberof radiocarbondateswere obtainedfrom
StrataVI andvll, providing radiocarbonages
rangingbetween1940and 660 B. P. (Sheppard
et al. 1987). In generalterms,the earliest
StratumV likely representsa time spanfrom
about 6,800yearsagoto about4,000 yearsago.
StratumVI likely spansthe time period from
1300to 1940B. P. and StratumVII representsan
ageof between660 and 1600B. P. (Sheppardet
al. 1987;GustafsonandWegener1998).
324
As shownin Table 11.3,thereis some
striking variationin taxonomicrepresentation
acrossthe threestrati~hic units, particularly in
the relative frequencyof Oncorhynchussp. Unit
V is dominatedby Oncorhynchus,which
representsover 60% of the fish remainsin the
unit. In Unit VI, Oncorhynchusis represented
by a singlespecimenandresidentfreshwater
fishescomprisemost of the collection. In
Stratumvn. freshwaterfishes still dominate
(with about85%),but the abundanceof
Oncorhynchusis greaterthan in StratumVI. A
Chi Squarecontingencytest shows-that these
differences are significant (Xl = 129.70,P <
.0001);all of the residentfreshwatertaxa in each
stratumwere aggregatedto compensatefor small
samplesizes. Thesedifferencesmight reflect
site basedor region-widechangesin the
organizationof subsistenceand settlement
activities. It is widely acceptedthat sometime
after 5,000yearsago,peoplemadeincreasing
useof salmonas a storedresource. Perhapsthe
higherfrequencyof salmonin Unit V signalsthe
storeduseof this resource. The presenceof
storagefeaturesin Units V and abovesuggests
that the rockshelterwas servingas a place where
resourceswerecached. If it is acceptedthat the
PalouseRiver neversupportedmuch of an
anadromoussalmonpopulation,then the
prominenceof salmonremainsmay reflect the
captureof salmonin the main stemSnakeRiver
andtransportof dried fish to the Rockshelter.
This reasoningfollows Chatters(1987) useof
"geographicdisplacement"as an indicator of
food storage.When remainsof organismsare
recoveredsomedistancefrom their known
habitator distribution,their presenceindicates
resourcetransport. Given the costsassociated
with food transWrt, efforts to reduceweight,
throughbulk processingand drying, would be
promoted. Interpretingthe changingtaxonomic
frequencyin this way, however,could be in
error, given the samplingproblemsdiscussed
earlier,andthe suggestionremainstentative.
regressionanalysis,but rathersimple
comparisonsof vertebrameasuresof fishesof
known length to identify changein body size
represented.
For Oncorhynchus;only eight vertebrae
werecompleteenoughto measure.Valuesrange
between6.2 mmto 10.7mm(Table 11.4). The
small samplesizepreventsexaminingany
temporalchangesin body sizesrepresented.
Basedon comparisonof vertebraefrom fish of
known length (Table 11.5),the rockshelter
samplescome from a rangeof body sizes,
rangingbetweenabout350 and 800 mm in
standardlength (the distancefrom the tip of the
snoutto the baseof the tail or endof hypural).
Over 100 vertebraefrom minnowsand
suckerswere completeenoughto measure.As
seenin Table 11.6,meandiametervaries little
(between5.4 and 7.1 mm) from the lower strata
to the upper units, suggestingthereis little
variation in the sizesof freshwaterfish that were
depositedthroughoutthe rockshelter. Taking
into accountthe standarddeviation,most of the
vertebraerangebetween4 and 8 mm in size.
Basedon comparisonwith modemfish body size
and vertebrasize(Table 11.7),vertebrameasures
this size are from fish that rangein sizebetween
about250 and 500 mm standardlength.
Summary. The rockshelterfish faunashows
that a range of fish taxawere usedby people
occupyingthe rockshelter.Fish remainswere
identified in eachof the stratigraphicunits
spanningthe Holocenerecordof occupation,
indicating that fish playedsomepart in
subsistenceactivities for the 10,000yearsthe site
wasused. Both residentfreshwaterand
anadromousfishesarepresent.suggestingthat
pastpeopJeweregeneralizedin their fishing
practices. The residentfreshwaterfish would
havebeenavailablein the river adjacentto the
site. The PalouseRiver, however,probably did
not supporta migratorysalmonrun throughout
the Holocene. The presenceof large-bodied
salmon,likely representingmigratory fish in the
site, suggeststhat peopletraveledto the main
stemSnakeRiver for this resource.
There hasbeenlittle studyof the life history
and habitatrequirementsof the resident
freshwaterfish speciesin Plateaurivers and
streamsthat areprominentin the rockshelter.
Body size. Vertebraeweremeasuredto roughly
estimatethe sizesof fishespresentand whether
therewereany changesin the body sizeof fish
over time. Casteel(1976) showedthat vertebral
sizeandlinear dimensionslike length were
highly couelated. This study did not rely on
325
Table 11.2 Frequency (NISP) offish taxa by strata/unir, rocbhelter.
Acipenser sp.
Oncorhynchus
3
6
sp.
8
1
Cyprinidae
Ptychocheilus
oregonensis
Acrocheilus
62
5
1
3
4
1
2
1
3
3
alutaceus
1
2
Catostomus sp,
.5
Catostomus
macrocheilus
Cyprinidael
1
16 1
3
3
1
1
13
26
16
17
4:
3
3
17
10
2
1
66
44
7
,
27
Table 11.3Frequency (NISP) offish taxa ac~
Oncorhynchussp.
Cyprinidae
Ptychocheilusoregonensis
Acrocheilusalutaceus
Catostomussp.
Catostomusmacrocheilus
Cyprinidae/Catostomidae
.
Total
VB
VI
'II
NISP
62
5
1
1
1
63.9
5.2
1.0
1.0
1.0
27
91
27.8
100.0
,11
three strata/units, rockshelter.
v
Taxon
1
1
NISP
1
26
16
1
17
5
66
112
'It
O.7
19.7
12.1
0.7
12.9
3.8
SO.O
100.0'
'II
NISP
~
17
4
~8
19.3
4.5
10
11.4
44
88
SO.O
100.0
Table 11.4Mean width of Oncorhynchusvertebrae acrossstrata/unit, rocbhelter.
Strata
Mean Vertebra
Width (mm)
N
Std. Deviation
1.8
10.1
1
1
-
V
m
9.6
6.2
3
1
1.4
--
I
7.8
2
4.2
VB-VDI
VB
Note: somestratumassignmentswere madeafter fieldwork wascomple~d
provisional
326
shouldbe considered
.s
Table 11.5 Standard leogth1and vertebra widths of selectedspeciesof Oncorhynchus.
Taxon
~
Vertebra
Body Size
Width(mm)2
Standard
Length(mm)
11.5
825
10.1
630
8.7
575
8.2
6703
Oncorhynchustshawytscha
(VLB 92-6-8)
o. tshawytscha
(VLB 86-20-4)
o. mutch
(VLB86-4-1 )
o. mykiss
(VLB 86-6-12)
O. clarki
6.1
~91-10-1)
-
365
-
lengthfromtip of snoutto endof hypuralbone
2measureis an averagewidth of six vertebraearbitrarily selectedfrom eachskeleton
3 measureis fork length
Thus it is difficult to gleandetailedinsight on the
paleoenvironmentalsignificanceof the
archaeologicalremains. The speciesidentified
occupya rangeof river, streamandlake habitats,
including slow and fast moving water,deep
pools and shallows;their preferenceis relatively
warm water. Perhapsthe most that can be said is
that the presenceof severalspeciesof freshwater
fishesthroughoutthe Holocenesequenceof
humanoccupationat the site suggeststhe
adjacentriver provided adequatehabitatfor these
fish throughoutthe Holocene. Given that the
record for region-wideand local enviromrental
changeis clear,it may be significantto note that
the river wascapableof supportingfish
populationsfor the IO,OOO-year
period.
Floodplain Fish Remains
As notedpreviously,the fish remainscomefrom
a largenumberof bulk samplesthat were
retainedin I-mm meshin the field andsorted
undera controlledsettingin the laboratory. Use
of fine meshandcontrolledsortinginsures
minim~110ss
of small bonespecimens,andin
twn, minima] loss of remainsof small-bodied
fishesif they arepresentin the site deposits.
Significantly,during the excavationandfield
screeningprocess,someitems,including faunal
remains,wereremovedfrom the screenand
baggedseparately.Unfortunately,if this
includedfish remains,noneof this materialwas
locatedduring this study,nor were any records
Table 11.6 Mean width ofcyprinidae/catostomidae vertebra~ acrossstrata/unit, rockshelter.
Strata
Mean Vertebra
Width (mm)
N
Std. Deviation
vm
vn
VI
V
ill
I-ll
I
5.5
5.7
6.4
5.9
7.1
5.9
S.4
6
33
36
14
1
1
2
1.5
1.6
1.6
1.1
j0.1.
Note: somestrammu~ts
consideredprovisional
wa'e madeafter fieldwork wasro"Inpletedand shouldbe
327
~
Table 11.7 Standard length and vertebra widths of selectedCypriniJormes.
Taxon
Catostomusmacrocheilus
(VLB 92-5-6)
Catostomuscolumbianus
(VLB92-7-10)
Catostomus
fumeiventris
(UMMZ181667)
Catostomusmacrocheilus
(VLB 92-5-9)
Mylocheiluscaurinus
(VLB87-10-3)
Gila bicolor
(VLB89-1-25)
Catostomusplatyrhynchus
(VLB92-10-7)
Rhinichthyscataracte
(VLB92-7-5)
Vertebra
Width (mmY
7.4
Body Size
StandardLength (mm)
425
5.1
315
4.2
260
3.8
239
2.6
192
2.7
160
2.5
147
0.9
65
,
-me.as--ure
is an averagewidth of three abdominaland threecaudalvertebraearbittarily selectedfrom each
skeleton
found which in~cate the extentof this practice.
Presumably,iffish remainswere selectively
baggedduring field screeningand ultimately
lost, they would havebeenrelatively large
specimens.In short,thereexiststhe real
possibility that the floodplain fish sample
availablefor this study is biasedagainst largebodiedfishes and infavor of small-bodiedfishes.
Obviously,theseproblemsintroducedifficulties
in interpretingthe fish faunalrecord in the
deposit. As with the rockshelterfish sample,
samplingand curationproblemsmeanthat the
fish remainsprovide a nominal record of fish
taxaand body sizespresent. As relevant,the
following discussionreviewsthe extentto which
thesebiasesaffect this study's interpretationsof
the spatialand vertical patterningin fish remains.
Taphonomy and Fish Bone Origin. How did
the fISh bones come to be in the floodplain
deposit? Do they represent the remains of past
human subsistence or could they reflect natural
deposition of fish by floodwaters or nonhuman
scavengers or carnivores? As noted previously,
sorting out the agents responsible for a faunal
deposit is essential before one can identify
cultural activity patterns. Given the very old age
of the Marmes floodplain deposits as well as the
potential and realized significance of the site to
our understanding of human occupation of the
region (e.g., Ames et al. 1998), it is necessary to
carefully review the agents responsible for the
fish remains found there.
Geoarchaeological analysis of the early
Holocene sediments that were the focus of study
shows that they are horizontally stratified fluvial
deposits left by low energy flooding (see Chapter
Five). Radiocarbon dates from the deposits
suggest they were deposited about the same time
as Unit I and n in the Rockshelter (between
10,000 and 8500 radiocarbon years ago).
Unequivocal human use of the riverside location
is indicated by stone tools, several features, and
human remains found in the stratified deposits.
Results. As shown in Table 11.8.2.481 fish
remains were recovered from the floodplain bulk
samples. About 70% of these (1.762 specimens)
could be identified below class fish. Fourteen
taxa were identified; resident freshwater fishes
absolutely dominate the assemblage. with seven
different species represented. Remains of
cyprinids (identified to family and species) are
more abundant than catostomids (suckers). The
salmonids represent a very small fraction (less
than 1%) of the assemblage.
328
.
Table 11.8 Frequency (NISP) of fish taxa, floodplain.
Taxon
F100dplain
NISP
%
Salmonidae
Oncorhynchus sp.
Prosopium williamsoni
3
12
1
161
Cyprinidae
Ptychocheilus oregonensis
Acrocheilus alutaceus
Mylocheilus caurinus
Richardsonius balteatus
Rhinichthys sp.
RichardsoniuslRhinic hthy s
Catostomus sp.
Catostomus macrocheilus
Catostomus columbianus
Cyprini dae/Catosto mi dae
8
2
5
1
2
2
60
1
1
1,503
719
Unidentifiable
Total
Overbankdepositionwasepisodic,asthe
sedimentaryrecord showsa seriesof weakly
developedA horizons(seeChapterFive). A
very generalsite formation model suggeststhat
humansengagedin a variety of ~vities in the
areabelow the rockshelterand next to the river,
organichorizonsformed, andthenlow-energy
flood waterperiodically over toppedthe bank
allowing fine sedimentsto settleout of
suspensionand bury remainson the surface.
Given sucha 1OOdel,
it is conceivablethat fish
were sweptby floodwatersonto the bank where
they becametrappedandultimately buried.
SeveralfKtors pnd lines of evidencesuggestthis
is not the case,however.
Onecha1Ienge
to the fluvial origin considers
the likely seasonof the floods,the body sizeof
the fish in the depositandthe life history and
demographyof the sourcefish populationsthat
would generatesucha deposit. FIrSt,river
flooding along streamsandrivers in the Plateau
is associatedwith the ~lting snowpack. which
occursin the spring and early summer. The
sourcepopulationfor a naturalfluvial fish
depositwould be fish in the river during spring
andearly summer. Importantly, Kross western
North America,the numerousspeciesof
cyprinids and catostomidsspawnduring this
time of year (Moyle 1976;Siglei'and Siglei'
1987;Wydoski and Whitney 1979). While
specificmatingbehaviorand habitatpreference
0.1
0.5
0.0
6.5
0.3
0.1
0.2
0.0
0.1
0.1
2.4
0.0
0.0
60.6
29.0
varies from species to species. for all taxa.
spawning entails reproductively mature adults
aggregating for days or weeks at a time. in pools
or shallows in the river. I suggest that the source
population for a natural flood-rafted deposit of
freshwater fish should include adults representing a range in age or size classes. The
body size and age of reproductively viable
individuals as well as the lnaxin1um size aM age
reached is highly variable -=ross species; several
small-bodied, short-lived cyprinids Rhinichthys,
Richardsonius. beco~ mature when they are
less than 100 mm and then rarely reach over 100
mm in size. Several cyprinids and catostomids
(Ptychocheilus oregonenis, Mylocheilus
caurinw, Catostomus macrocheilus) beco~
sexually mature when they are over 100 mm in
length and often attain lengths of over 300 mm
and ages of over 15 years. The Marmes
Rockshelter vertebra measurementsprovide
independent evidence that relatively large
cyprinids and catostomids were present in the
early Holocene in the site vicinity. As noted
previously. vertebra from cyprinid/catostomid
fishes ovez 300 mm in standard length are most
commonly represented.
In short. if the floodplain fish remains are
the rem~ants of flood-rafted carcasses.a range in
body size should be present in the fish
assemblage.reflecting the range in body size of a
329
spawningpopulation.As shownin Table 11.9,
about 400 vertebraefrom the floodplain were
measured.with samplesfrom eachhorizon
documented.The meansizeis very small (1.9
mm) andremarkablyconsistentfor eachhorizon
or aggregatehorizon.suggestingfirst that the
body sizerepresentedis quite small and second
overall uniformity in sizeclassrepresented.
Comparingthesevertebrasizesto modemfishes
(Table 11.7)suggeststhat the meansizeof fish is
considerablysmallerthan 200 mm in standard
length. It might be suggestedthat the small
vertebrasizeresultsfrom the deposition
primarily of the small-bodiedgenerasuchas
RichardsoniusandRhinichthys. While these
generaare notedin the deposit,their remainsare
much lessfrequentthanthe larger-bodied
cyprinids (Table 11.8). Overall, the scarcityof
large fishesin the floodplain depositdoesnot
match the expectationfor a fluvial depositof
spawningadults,which shouldinclude a rangein
body sizes,including relatively large individuals.
The bestexplanationfor this narrow and
relatively ~mallbody sizeis somekind of
selectivemortality, notably humanselectionof
small fishes.
Of coursethis explanationpresumesthat the
scarcityof "large" vertebraein the depositsis
real andnot a samplingproblem (e.g.,selective
field baggingand subsequentloss of such
vertebrae). Given that it is not possibleto rule
out this bias,statementsaboutbody size
distribution and frequencymust remaintentative.
It would be usefulto havesupporting
evidencefor a humanrole in the form of cut
marks or patternedburning. Cut markswere not
seenon any of the specimens.Importantly, cut
marks haveonly rarely beennoted on
archaeologicalfish remains(cf. Barrett 1997),
and noneto this analyst'sknowledgehavebeen
reportedon fish remainsin Plateauor coastal
archaeologicalsites. Thus,the absenceof
evidencefor cut marksshouldnot be usedto
underminethe argumentfor a humansourcefor
the fish bones.
As notedpreviously,presenceor absenceof
burning can be usefulin sortingout taphonomic
agentsresponsiblefor a bonedeposit. In the
caseof the floodplain faunalremains,however,
bwning is not helpful largely becausethe
incidenceof burning acrossthe site and
vertebrateclassesis so high. Reportingon the
nonfish assemblage,
Gustafsonand Wegener
(seeChapterTen) note that most of the bone~
330
showssign of burning. This includesan
estimated60,000fragmentsfrom unidentified
taxa as well as bones from animals that are likely
noncultural in origin (i.e. surface dwelling
snakes,burrowing rodents). Someof this
burning may result from humanfires (cooking or
trashfires), but the ubiquity of the burning
acrossmost of the vertebrateclasses(including
taxa likely of nonculturalorigin) makesit
difficult to useburningper seasgood evidence
for a human source.
Regardingthe fish remains,Over30% of the
remainsaltogethershow clear sign of burning
(Table 11.10). This proportion is much lower
thanthat notedfor other vertebrateclassesby
Gustafsonand Wegener,which quite likely
reflects different criteria usedin identifying
burning. Table 11.10also showsthe frequency
of burning acrossthe horizonsin the floodplain.
For the threehorizonswith sizeablecounts,
Harrison,Marmes,andjoint Marmes/Harrison,
the frequencyof burning is very similar,
suggestingthat whateveragentsareresponsible
for the burning, they actedconsistentlyover the
courseof site formation. FIShbody part
representationalsomight be usefulin sorting out
taphonomicorigins (Butler 1990,1996). In
particular, if the fish representnatural,floodrafted carcasses,the entire skeletonshouldbe
represented.Further,if burial was swift,
specimenfragmentationshouldbe minimal and
the specimensshouldbe in goodcondition. The
first expectationis met, but not the second. A
review of skeletalelementrepresentation(Table
11.11)showsthat numerouselementsof the
head,fins and vertebralcolumn arepresent.
Given that manyelementsof the skeletoncould
not be identified below the combinedfamily
taxo~ Cyprinidae-catostomidae,this review
considersall of the elementsfrom these
freshwaterfishestogether. [The quantity listed
in Table 11.11is the minimum numberof
elements(MNE; Bunn 1982)which selectsthe
bestrepresentedSectionof eachelementand
countsthe numberof timesit occurs. Using this
quantity rather thanNISP controlsfor the
problem introducedby Specimenfragmentation.
With :MNE, a singleskeletalelementwill be
countedone time, no matterhow many
fragmentsit may havebeenbrokeninto; with
NISP, a single elementcould be brokeninto
severalfragmentsandif all of the fragments
wererecognizable,all of them would be
counted]. One elementfrom the head~
Table 11.9 Mean width of CyprinidaelCatostomidaevertebrae acrosshorizon, floodplain.
Horizon
MeanVertebra
Widdl (mID)
1.9
28
Std.
Deviation
.6
Mixed Marmes.
Harrison
2.0
SI
.4
Harrison
1.9
264
.5
40
.4
Marmcs
BeneathHamson
Unassigned
N
4.6
1.9
.S
384
1.91
--Note: somestratumassignmentsweremadeafter fieldwork was completedand shC)uld
be
consideredprovisional
Total
Table 11.10Frequency of burned specimensacrossthe floodplain horizons!
Horizon
M~
bmued
%burned2
47
33.8
Mixed Mamx:s-Harrison
47
Hatrison
616
BeneathHarrison
0
Unassigned
73
33.2
785
31.6
Total
1--.
somestratumassignmentswere madeafter fieldwork wascompletedand shouldbe considered
..
nal
r.:;:...ts the percent of bones in each horizon or horizOn aggregate that are burned
indicated,thereis evidencefrom thesedata as
well for significant specimenbreakageand
~ntatioD,
especiallyof the beadele~nts.
Exceptfor the pharyngeal,all of the head
elementsarerepresentedby veryfew speci~.
This patternm:>stlikely is explainedby bone
destructionthat hasrenderedthe remains
unidentifiableor so small that they passed
throughthe I-mm ~
While degreeof
specimenfragmentationwasnot recoldedduring
analysis,the author'simpressionof the
assemblagewas that exceptfor vertebralcentra,
specimenstendedto be fragmentary.The very
low frequencyof mostbeadelements(Table
11.11)certain1
y supportsthis notion.
(pharyngeal)is representedby 39 specimens;
given that two pharyngealsare in an individual
fish. this indicatesa minimum of 18 individuals
arerepresentedin the deposit. The minimum
numberof fish representedby the abdominal
vertebraeis 22 (basedon an averageof 19
abdominalvertebraeper individual) andthe
minimum numberof fish representedby the
caudalvertebraeis 21 (basedon an averageof 18
caudalvertebraeper individual). Thesevery
similar valuesof both headand trunk elements
(controlling for the numberof times the elements
occurin the skeleton)strongly indicatethat the
entire skeletonwas initially depositedon the
floodplain. While whole body depositionis
331
Table 11.11Frequency of skeletal elementswith landmarks (MNE) from Cyprinidae and
Catostomidae,floodplain.
MNE
Element
Cranial
articular
basioccipital
ceratohyal
dentary
epihyal
hyomandibula
maxilla
metapterygoid
opercle
palatine
parasphenoid
pharyngeal
pterotic
quadrate
urohyal
vomer
PairedFins
cleithrum
scapula
Vertebral Column
lit vertebra
2Mvertebra
lit or 2M vertebra
abdominalvertebra
caudalvertebra
ultimate vertebra
Total
The fragmentarycondition of the fish
remainsis comparableto that describedfor much
of the nonfish fauna(seeChapterTen).
Moreover,patterningin fragmentationacross
vertebrategroupsmay in fact offer clues asto the
sourceof the destruction. Gustafsonand
Wegenerpoint out that mostof the
approximately1,300nonfish specimensthat they
were ableto identify to sometaxonomiclevel
wererodentsandreptiles,taxa which they also
note, are leastlikely to reflect cultural use and
deposition. Further,it wasthesespecimensthat
showedthe highestdegreeof integrity. In other
words,fragmentationis not evenly distributed
acrossvertebrateclasses,which implies that the
sourceof fragmentationis not a ubiquitousforce
(like sedimentchemistryor post-depositional
~
3
4
4
8
3
3
2
1
6
1
1
39
1
3
2
1
0.3
0.4
0.4
0.8
0.3
0.3
0.2
0.1
0.6
0.1
0.1
4.0
0.1
0.3
0.2
0.1
13
2
1.3
0.2
30
11
12
420
374
14
3.1
1.1
1.2
42.8
38.1
1.4
trampling),but a more discriminatingforce that
was targetingparticular animalgroups. Humans
would seemto be a likely candidatefor this, in
the form of cooking and butcheringactivities.
It is also importantto considerwhether
nonhumancarnivoreor scavengingactivities
havedepositedor modified the fish assemblage
from the floodplain. As notedfor the rockshelter
fauna,GustafsonandWegener(ChapterTen) do
not discussany sign of carnivoregnawingor
other markingson the nonfishassemblageon the
floodplain. Budgetaryandtime constraints
preventedundertakinga detailedstudyof surface
morphologythat might be usedto identify
digestiveprocess. The generalimpressionis that
characteristicpatternsassociatedwith digestive
process(rounding,pitting, andvertebra
332
deformation)were not present. Most of the
specimenswereeither darkly stainedor burned.
Basedon availableinformation, the faunal
assemblage
doesnot appearto havebeen
ravagedby carnivores.
In sum. severalfactors suggestthat humans
are DX>st
likely responsiblefor the fish deposit.
If it is assumedthat there was minima1removal
of largevertebraeduring field screening,the
narrowrangeand small body sizepresent
suggestsa form of selectivemortality was
operatingand humansare the most likely
candidatefor this. Further,the fish remainsare
fragmentarywhich would be expectedif the fish
were subjectedto variouscooking and
processingactivities. Certainly nonhuman
agentscan causebone destruction,but the fact
that fragmentationis unevenlydistributedacross
the vertebrateassemblagesuggestsindirectly that
the agentof modification wasselectiveand
again,humansare a likely agentfor this
selection. Additional supportfor the humanrole
in generatingthe fish depositwould examine
whetherthe intra-sitefrequencyof fish remains
is correlatedwith other clearly cultural items
(e.g.,lithics); asproject reportsfor thesedataare
generatedit is recommendedthat thesestudiesbe
conducted. As well, it would be usefulto carry
out sometest excavations .off-site" to evaluate
the likelihood for suchnaturalaccumulationsof
freshwaterfish remainsto occur.
containedfish bonesin the two main horizons
(Marmesand Harrison).but the overall
frequencyof fish bonein the two horizonsis
very different The 34 sampleswith fish bonesin
the Harrisonunit providedover 1.800fish
remains.while the 38 samplesin MarInes
provided only 139fish specimens.Differences
in volumesof excavatedor sortedmatrix from
the two horizonsdo not explain this striking
pattern. Rather.thesedatasuggestthat during
the accumulationof the HarrisonHorizon. there
was greateruseand depositionof fish thanin the
MarInesHorizon.
While thesedatasuggestthat fish useper se
may havevaried over time. thereis remarkable
consistencyin the body sizeof fish present
(Table 11.9). As notedin the taphonomy
section.vertebrasizeis uniformly small across
horizons.suggestinga consistencyin fish
captureandfishing activities (capturemethods.
and usepatterns)over time.
Summary. The floodplain fish assemblage
provides a number of striking patterns that
should be reviewed. First. the record shows a
dominance of freshwater fishes and extreme
scarcity of salmon remains. All of the faunal
remains show a high degree of fragmentation.
and it is reasonable to be concerned that the
scarcity of salmon bone is in part due to the high
degree of bone destruction (Butler 1987b).
Salmon bone is not as "densewin terms of
mineral content per volume as minnow and
sucker bone (Butler 1996; Butl~ and Schroeder
1998). If bony tissues of both salmon and
minnows/suckers were subjected to similar
destructive process, then salmon bone would
degrade more readily and less likely be part of
the faunal ~rd
in the site. In the case of the
floodplain assemblage,this scenario is not
supported. All of the matrix was processed
through 1-mm mesh, which is an extremely
small mesh size. Fragmentary remains of
salmon vertebrae or other durable cranial
elements (e.g., gill rakers, teeth) should have
been recovered in this fine screen matrix in
greater numbers if, in fact, they had been present
in higher numbers in the deposits. The virtual
absence of salmon bones and teeth in the deposit
suggestsin fact that salmon were not used much
by people that occupied the floodplain.
Intra-site variation. Table 11.12showsthe
distribution of identified taxaacrosshorizonsin
the floodplain deposits. As shown.the bulk of
the assemblage
(over 75%) is from the Harrison
Horizon. Given that freshwaterfishesare the
overwhelmingdominanttaxain the locale,they
of coursedominateeachof the horizons.
~tion
of fish distn"butionacrossbulk
samplesshowsthat the distribution of fish bone
is extremelyuneven. Of the 240 buJksamples
processedonly 84 (35%) had fish bone.
Furthemlore,73% of the fish bonerecovered
wasfrom just six samples(Table 11.13). These
datasuggestthat fish bone wasdepositedin
discreteareasratherthan widely dispersedacross
the sitedeposits. The patternof clumping can be
seenin the distribution of fish boneacross
horizonsas well. As shownin Table 11.13,
roughly similar numbersof bulk samples
333
~
=
.;
-a.
~
=
=
=
~
~
=
~
"'c
=
!oj
'C
=
.:
£
=
~
.s
.:
'"
C
=
~
~
z
'-'
;0.,
C.I
C
Q;
=
='
Q;
I..
~
~
...
...
Q;
:c
~
]
101)
rn
~
~
~
;
~
=
~
~
~
t/)
z
~
~
~
~
f/)
z
*
~
'a
~
:I:
.~ ~
]~
z~
CI)
~
Z
~
rIJ
~
Z
~
~~
=cId
~ ~ ~
~.~ ~
~:c~f;}
~
.~
:I:
~ ~
~ .~
~:z:
=
0
~
E--
/
-
-~
~
NII')- N
~r-:~~If}~l"'!~~~..~~l"'!~
QeoO\ooooooff'leo~8
~N-
~\QII')
-.."';0
r---ff'lN
NO--ff'I
~
...
..
~"'ff'I
'
I
~
-
j
8
N'
0 ~i
VOl r---'
- _I
r---o
00
r---
...~
--
~8
~O
<'I
oi
-0
00
.~
u
.E
"0
~~
-- I]"0u
-~
~8
- ->
~
-
I~~
~.
ff'I
-
Nr--r---~
NON
~
0
~
,
~~
~"!a:
..
~
-~
ff'l0
..
~
'c
0
i
]
.£
I~
0
u
cn
=
~
~
~
:9
u
'-=
~
u
1
~
rJ
~
~
Q
cn
~
~
~
~
1'+0
:s ~
~
- _
00
88
-11')11')
c=
:-a
s '=
~
~
--
'5
e]i s
-~
..
S S S 0=
E~"O
10;
E
~
--
10; '5
~
.8
cn
~8
- "3
00~00~8
-N
'ii~
~
~
r \
I
-
-
-
~
r---
--r
~
~
oooiooo
..
~
~"':::
~
~-~II')-II')
NN~--~--~
-
..
't;
~
~""
~::
'-
~~:~
~
~ ~
~
r~
o~
r~
~
r~
~~-ei -~~
c~
o.~
~~ ~ 8 ,-~ ~:-- ~ E 8 ~r~
"';::
~~
~~
= "E ~ -=
o
e
c 2 -c
~~E~°'ii~~§l-§~~~]
."0
~-="O'5--
~~p.~~_'::~'::s~~
"; ~ 'r~
Q ~ ~ -~
CI)",Q.,~Q.,~~~~~"""",~~
334
Table 11.13Frequency of fish remains by horizon, floodplain.
Horizon
#Bulk
Samples
withFJSh
FISh
count
2
0
38
2
2
34
6
Unassigned
Above MarDles
MarDles
~
139
5.6
Mixed Harrison& Marmes
148
6.0
BetweenHaITison& MarInes
3
0.1
Harrison
74.9
1,858
BeneathHaITison
11
0.4
Total
Note: somestratumassignmentsweremadeafter fieldwork was completedand shouldbe
consideredprovisional
Therearetwo obviousreasonswhy salmon
may not havebeenused. First. the site is located
over a mile and a half from the closestsourceof
migratory fish, in the main stemSnakeRiver.
Perhapsschedulingof other economicactivities
meantthat local residentswere only ableto take
advantageof very local fishery resourcesthat
werein the adjacentriver. It is alsopossiblethat
the lack of salmonidremainsreflectsthe season
of site occupation. Migratory salmonidswould
havebeenavailablechiefly in the late spring,
summerand fall. H the occupationof the
floodplain did not coincide with the timing of
migratory ruDS-perhapSpeopleoccupiedthe
site in the winter and spring-then the absence
of salmonwould be explained.
A third factor could be environmental,
suggestingthat salmonidswere simply not very
abundantin regional streams,becauseof poor
spawninghabitator perhapsoceanicconditions.
Chattersand others(Chatterset al. 1995)have
suggestedthat conditionslater in the Holocene
would havebeenpoor for salmon,dueto Wamler
watertemperaturein spawninghabitatand
overall poor streamconditions. To properly
evaluatethis explanationfor the early Holocene
recordon the SnakeRiver, we needmorefinescaleenvironmentaldata aswell asstudiesof
contemporaryfish assemblages
from othersites
in the region (especiallyon the SnakeRiver).
Notably, given the Marmessite'slocation on a
non-salmonproducingriver, it will be difficult to
usethe site'sfish recordper seto identify the
role of environmentalchangein affecting
sa1monidabundance.
The secondmajorpatternregardingthe fish
remainsis the small body sizepresent. FISh
body size estimationhasnot beencarried out for
other Plateauarchaeologicalsitesfor any time
period, so we lack a comparativebasisfor
examiningthe pattern. Additionally, there is
little detailedinformationon the life history,
seasonalmovements,and schoolingbehavior for
the many freshwaterspecieson the Plateauthat
could be usedto modelhumanfishing strategies.
Prehistoricfishersmay haveusedspears,hook
and line, or massharvestingsuchasnets,traps,
or poison. Given the narrow sizerange,a form
of selectivemortality is indicated,particularly
the useof massharvestingthat wastargetinga
particular
size classof fish (Butler 1996;
I
Greenspan1998). Thereareno bonepoints or
net weightsin the site'sdepositsthat might be
usedasindependentevidencefor fishing
methodsused. To developa more
comprehensiveunderstandingof Plateaufishing
strategies,especialIyfor freshwaterfishesfor
which we have little knowledge,the kinds of
strategiesusedto catchcertainspeciesand body
sizes,and the moregeneralfactorsthat affect
decisionsaboutfishing or technologyselection,
it would be very usefulto carry out fishing
experimentsin local environments(e.g.,Kirch
and Dye 1979;Raymondand Sobel 1990).
335
vertebraefrom the floodplain provide a mean
width of 1.9 mID(sd=O.5).Grantingthe very
small samplesizefrom the rockshelter,the
vertebraesizesrepresentedin the two localesare
extremelydifferent Recoverypractices,
particularly the useof lA-inchmeshscreenscan
be usedto explainthe lack of small vertebraein
d1erockshelter. Possibleselectiveremoval (and
subsequentloss) of "large", vertebraeduring
screeningof floodplain sedimentsmight help
explain the scarcityof suchvertebraein d1e
faunal assemblage.On the other hand,if the
dominanceof small vertebrae(and scarcityof
large vertebrae)in the floodplain depositsis real,
that large vertebraearetmly scarcein the
floodplain deposits,then it is reasonableto
examinewhy the two localeshavesuchdifferent
representationof fish body sizes.
One explanationwould suggestthat the
differencereflectsdifferencesin cultural
processingbetweenthe two areas. Perhapsthe
samesizedresidentfreshwaterfish were
capturedanddepositedin the two areasandthat
differencesin butcheringandcookingpatterns
betweenthe areasled to higher ratesof bone
destructionin the floodplain. Gustafsonand
Wegener(seeChapterTen) point out that the
characterof the mammalianfaunais very
Comparison of Rockshelter and Floodplain
Fish Fauna
Table 11.14showsthe frequencyof fish remains
identified in the floodplain and the lowestunits
of the rockshelter,which arecontemporaneous.
The recordsshowa numberof differences. FIrSt,
the floodplain hasa much richer fish faunawith
14 taxa, while the rockshelteronly hasfour.
This differenceat leastin part must be linked to
the major differencein archaeologicalrecovery
betweenlocales. Use of I-mm meshscreenin
the floodplain led to the recoveryof several
small-bodiedtaxa,remainsof which would not
be caughtin 1/4"(6.4 mm) mesh. The useof finer
meshalso generatedan overall larger sample
size,which in iiself would tend to produce
higher richnessvalues(Grayson1984;Gordon
1993). Both showthe presenceof anadromous
andresidentfreshwaterfishes. Comparisonof
taxonomicidentificationsdoesnot revealany
striking differencethat could not be explainedby
recoverypractices.
Onecan also comparethe two localesbased
on vertebrasize. Only threevertebraefrom the
lower units of the rockshelterwere complete
enoughto measure(Table 11.6); thesesuggesta
meanwidth of 5.6 mm. The almost 400
Table 11.14 Frequency (NISP) of fish taxa in the floodplain and
contemporary units of the rockshelter.
Salmonidae
Oncorhynchus sp.
Prosopium williamsoni
Cyprinidae
Ptychocheilus oregonensis
Acrocheilus alutaceus
Mylocheilus caurinus
Richardsonius balteatus
Rhinichthys sp.
Ric hardsoniuslRhinic hthys
Catostomus sp.
Catostomus macrocheilus
Catostomus columbianus
Cyp rini dae/Cato stomi dae
3
1
2
1
16
1
1
3
4
Unidentifiable
S
4
Total
21
Note: somestratumassignmentswere madeafter fieldwork was completedand shouldbe considered
provisional
336
different in the two areas. The rockshelter
mammalbonestend to be much morecomplete
(andunburned)than the floodplain mammal
remainswhich are extremelyfragmentedand
burned. They suggestthat cultural processing
differencesmay help explain this.
This explanationdoesnot accountfor the
fish pattern,however. The first and second
vertebraeon the vertebralcolumn of the
minnowsandsuckersare amongthe densestin
the skeleton(Butler 1996). Evenif the largerbodiedfisheshad beenmore aggressively
processed(which would lead to a higher
incidenceof bonedestruction),the first and
secondvertebraeshouldbe presentin some
numbers,if suchfish werecaughtanddeposited
on the floodplain. Thesevertebraefrom the
floodplain fish fauna wereincluded with other
vertebrameasuresand the data showoverall that
mainly small individuals arerepresented.In
short,failure to recordmorelarge vertebraein
the floodplain is not easily explainedby
increasedfragmentationof bonesof largerbodied fishes.
The small samplesizeof the rockshelterfish
faunain the earlieststratigraphicunits (I, ll) will
makeit difficult to isolatethe particular cultural
or naturalmechanismsthat accountfor the
differencesbetweenlocales. Also, becausethe
faunal recordsfrom the two areaswere sampled
in suchdifferent ways,it may be difficult to sort
how much of the .patterning. is simply a product
of our sampling. Additional comparativestudy
of the featureandartifact recordbetweenthe two
localeswill perhapshelp shedadditionallight on
the cultural and naturalfactorsthat accountfor
the striking differencesin faunalrecords.
337
MARMES ROCKSHELTERAPPENDICES
AV~~ABLE ON-LINE AT WSU MUSEUM
OF AN1HR! 0 POLOG Y WEB SITE:
WWW.ARCHAEOLOGY.WSU
.EDU
The appendiceslisted in the Tableof Contents,pagesvi-vii, are not includedin this
volume.They are availablefor viewing on the Web site, Museumof Anthropology,
WashingtonStateUniversity: www.archaeology.wsu.edu.
ockshe-ter
cY
.I
Editedby
Bren t A. Hicks
ConfederatedTribes of the Colville Reservation
History/Archaeology Department
With Contributions by
Virginia L. Buder, John L. Fagan,PamelaJ. Ford,
ShawnGibson, Carl E. Gustafson,Brent A. Hicks,
Gary Huckleberry,Joy Mastrogiuseppe,Guy Moura, Terry L. Ozbun,
Matthew J. Root, Danid O. Stueber,Roben M. Wegener,
PeterE. WIgand, and Maureen Zehendner
2004.
Washington
~--
state
University
Press,
,
Pullman
Acknowledgement of FederalAssistance
and Publication Disclaimer
This study of the MarmesSitewasperformedfor the ConfederatedTribesof the
ColvUleReservation.
The ConfederatedTribesof the Colville Reservationconducted
and funded the study through a contractwith the United StatesArmy Corpsof Engineers.Additionally,publicationwasassistedby a grant &om the Departmentof the
Interior, National ParkService.
Any opinions,fmdings,and conclusionsor recommendationsexpressed
in this
materialarethoseof the author/editorand do not necessarily
reflectthe viewsof the
Departmentof the Interior, the Army Corpsof Engineers,or the ConfederatedTribes
of the Colville Reservation.