AN ESTIMATE OF MORTALITY OF CHINOOK SALMON IN THECOLUMBIA RIVER NEAR BONNEVILLE DAM DURING THESUMMER RUN OF 1955' . .

BY THEODORE R. MERRELL, JR., I. MELVIN D. COLLlNS,2 AND ,JOSEEH W. GREENOUGH8BUREAU OF COMMERCIAL FISflERIt:S BlOLOGlCAL LABORATORY

AUKE BAY, ALASKA 99821

ABSTRACT

In 1955 the Oregon Fish Commission estimated thenumbers of dead chinook salmon, Oncorhynch~

tshawytscha, near Bonneville Dam and studied t!:teprobable causes of death.

The estimates of numbers of dead fish were madefrom ratios of tagged to untagged floating carcassesbelow the dam. Tagged s!llmon carcasses were releasedat the dam, and the river below the dam was· searchedsystematically to recover tagged and untagged car­casses. The introduce!! tagged carcasses aii~ the un­tagged carcasses of ~sh that died in the river wereassumed to have equal chances of recov~ry, providedthey were not too severely mutilated to be recoverable.This assumption was verified experiment!,l.ly,

On June 30 and July 1, 1955, when riverflows wererelatively high, 1,169 tagg~d chinook salmon carcasseswere released at Bonneville Dam. Thirty-one tagged and117 untagged carcasses were recovered in searches down­stream from the release point. On the basis of theserecoveries, an estimated ~,412 summer-run chinooksalmon died near the dam between June 21 and July10. On the basis of this estimate, 16.8 percent of thetotal chinook salmon run died at Bonneville ~m Inthis period.

The numbers of floating carcasses in 1954 and 1955were directly related to spillway dischal'ge; great~st

The problem of facilitating passage of anad­romous fish over dams and evaluating the effectsof dams on the fish has become a matter of in­creasing importance and concern becau~e Qf thesteady increase in the number of dams in th~' past2 decades. Each dam is an impediment to ~igra-

tory fish. .The Columbia River, one of the greatest rivers

in the world for chinook salmon, Oncorhynchustshawytscha, now has 11 dams across the ma~n

stem and numerous dams across tributaries.

Published February 1971.

FISHERY BULLETIN: VOL. 68, NO.3

numbers of f10atlqg dead fish coincided with ColumbiaRiver flows In e~ce88'of7,100c.m.s. At Bo~nevilleDamfall chln~ok salmon runs have never been subjectedto flows above 1,100 c:m.s. (kiIllng flows); spring runsare exposed to such f1o}Vs in some ye~rs; and summerruns nearly always encounter SJJch flows. Water tem­perat~re, turbidity, dls~as;, and' Injuries froin 'gill nets. ... .did not affect the number' of' c!lrcasses. Althouah thespecific causes of death and the pr~ciie areas at Bonne­ville Dam I where death" OCcurred were riot'determinedIn our study, the majpr .s~uice'~of chinook salmonmQrtality was associB"teci'wlth the spillway during highflows. Other Investiglitors subsequently demonstratedthat during high· f10~ theC61umbia River· that hasplunged over .dam spi'lw~y~ Is. sup,¢rsaturated 'withatmospheric pitrogeq, "This' supersaturation may beone of ~he principal' caulleS of..dea~h 9f fish at main­stem dams.Bonnevill~Dam is only a~out 18.3 m. high, and hun­

dreds. of thousands of !Ial!Jlon l!uccessfully negotiatethefishways .each year; yet many salmon are kiIledduring periods of. high flow. '(~9mplacency a"out theefficien,cy of salmon passage over' large dams Is, there­fore, unwa.-ranted, even .when:,elaborate well-designedpassage .facilities are present" alid few dead or injuredfish are notice~.· .

Bonneville Dam, the first dam on the lowerColumbia River, w~ cOJ;pPleted in 1938; it is 227km~ from the river" mo~th. B.efore its construction,an~dromo~s fi~h n}jgrated with little difficultyinto the upper Qoh,lmbia' a,nd Snake Rivers andtheir tributaries' to spawn:. The lower river hadtwo ·natural barl'iers-O~adeRapids aJ).d Celilo

L .

I Theodore R. Merrell, Jr., Fishery'"Biologist, Bureau of CommercialFisheries Biological Laboratory, Auke Bay,Alaska 99821. He was AquaticBiologist with .tli~.Oregon ~~ 'Commission at the time of the stUdy.

• -Melvin D. Colljns;:Aq,,':tic Biologil!t, Fish CommiBBion of OregonResearch·LaboratorY,"Clackamas. Oreg. 91015. .

• JOseph.,W. !i~i1o!.'gh;: Bi'1m~0~ian, Bureau of Commercial FisheriesBiological Laboratory; Au~:~ay.A!as~ 99821.

461

Falls-but fish could readily pass these except inperiods of extremely low riverflow. CascadeRapids was inundated by Bonneville Dam in1938; Celilo Falls by The Dalles Dam in 1957.

Each year Pacific salmon, Oncorhynchus spp.,and steelhead trout, Salmo gairdneri, that spawnabove Bonneville Dam yield several million dollarsto commercial and sport fisheries in the river andocean. Because of the great value of these fish, itis important to ensure that they pass over the9am with a minimum of loss and delay.

Hanson, Zimmer, and Donaldson (1950)counted many dead fish in the river below Bonne­ville Dam in various seasons and years, and fisher- .men in boats below the dam often observedfloating dead fish or dying fish on the surface.Chinook salmon were reported most frequently,but other species were also noted: sockeye orColumbia River blueback salmon, O. nerka; steel­head trout; American shad, Alosa sapidissima;

white sturgeon, Acipenser transmontanus; carp,Cyprinus carpio; and Pacific lamprey, Ento­sphenus tridentatus. Biologists and fishermen sus­pected that many of these fish died from attempt­ing to find a route over the dam or from beingswept downstream through the spillway afterascending a series of gravity-flow fish ladders,the principal means for fish to migrate over thedam. Although many of the fish seemed to nego­tiate the ladders satisfactorily, Schoning andJohnson (1956) estimated that chinook salmonwere delayed an average of 2.6 to 3.0 days intheir migration.

This paper reports a study of the magnitudeand possible causes of the mortality of fish atBonneville Dam, with particular reference to thesummer run of chinook salmon, which is believedto be more adversely affected than other runs.

Bonneville Dam (fig. 1) consists of two con­crete sections-the spillway and the powerhouse--

FIGURE I.-Bonneville Dam, showing the spillway and powerhouse sections and the difference in turbulence betweenthe two sections. Photo taken April 29, 1958, when spillway flow was 4,000 c.m.s. and powerhouse flow, 2,700c.m.s. (Photo courtesy of U.S. Army Corps of Engineers.)

462 U.S. FISH AND WILDLIFE SERVICE

which are separated by Bradford Island. Thespillway spans the Columbia River north of theisland, and the powerhouse spans the southchannel. The spillway has eighteen 15.2:-m.-widegates that release water not needed for powergeneration. To attract fish to the adjacent ladderentrances. the gate at each end of the spillwaydischarges water (except during very high or verylow flows). The gates are the vertical-lift type,and water is released under them at a maximumforebay depth of about 17.8 m. (fig. 2). The spill­way extends downstream from the gates to formtwo rows of concrete baffles that partially dissi­pate the energy of the water rushing under thegates. The flow through the spillway varies froma few cubic meters per second to several thousand.

Bonneville Dam has little water storage capac­ity. Differences in the water level between theforebay and the tail water range from 12.1 to18.3 m., depending on riverflow and power genera­tion requirements. The powerhouse contains 10Kaplan turbines. each with a maximum dischargecapacity of 410 c.m.s. Flow through the power­house is nearly constant throughout the year,but flow through the spillway varies greatly.

Facilities for passage of adult salmon upstreamover the dam are extensive and complex. Mostfish go up two gravity-flow ladders. which haveslopes of 0.3 m. in 6 m. The Washington shore(north) ladder has a single entrance adjacent tothe north end of the spillway and exits into theforebay about 120 m. upstream from the dam.The Bradford Island (south) ladder has threeentrances and a single exit into the forebay about120 m. above the powerhouse on the south side ofBradford Island. One entrance of the BradfordIsland ladder is adjacent to the south end of thespillway; another is at the north end of the power­house; and the third. consisting of many smallentrances. is across the powerhouse above thedraft tubes.

Information on dead fish below Bonneville Damcomes from several sources. The principal sourcesare Federal and State agencies: the BCF (Bureauof Commercial Fisheries), U.S. Army Corps ofEngineers, Oregon Fish Commission, and Wash­ington Department of Fisheries. Some informationis also provided by commercial fishermen, whospend much time on the river during the majorfish migrations; salmon sport fishermen; and

CARCASS RELEASE POINT-1954

CARCASS RELEASE POINT -1955

~~~/7t-=-----.. ~~TAIL WATER

• FLOW DIRECTION

~---~~~~-'--

FIGURE 2.-Bonneville Dam spillway construction and flow cross-sectional diagram (from U.S. Army Corps of Engineersdrawing). Elevations are in meters above mean sea level.

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 463

sturgeon sport fishermen, who seek salmon car­casses for bait.

Vntil our study, only one extensive effort hadbeen made to evaluate the significance of deadfish near the dam. In 1946 FWS (V.S. Fish andWildlife Service) and the V.S. Army Corps ofEngineers investigated the causes of injuries anddeaths of fish (Hanson et al., 1950). The majoremphasis of that study was on counting, fromboats, the floating dead fish near the dam duringthe main part of the chinook salmon fall migration.The study did not provide the information neededto determine the relation between numbers offloating dead salmon, numbers of dead salmonnot observed, and numbers of salmon surviving tocontinue their upstream spawning migration.

Sporadic observations in other years by biolo­gists and fishermen also failed to provide informa­tion that could be used to evaluate the significanceof floating dead salmon. To illustrate the difficultyof interpreting such information, we cite two ex­amples of observations made at a time when un­usually large numbers of floating dead salmonwere in the Columbia River.

The first observation was on September 9,1943. Arnie J. Suomela, Fishery Biologist withthe Washington State Department of Fisheries,spent 8 hours searching from a boat for deadsalmon between Bonneville Dam and MultnomahFalls. He found 146 floating chinook salmon, ninesteelhead trout, and one sockeye salmon; thefollowing statement is from his report. "CheckedOregon shore and river on trip down to Mult­nomah Falls. Only two salmon were found on thispart of trip. The first floating dead fish were foundat Butler's Eddy, at 3:05 p.m.; and floaters werefound from that point to below the spillway atthe dam. This observation definitely traces themortality to the north spillway channel and it isreasonable to believe that the mortality is occur­ring at the spillway."4

The second observation was in the spring of1952. The Oregonian newspaper for June 12, 1952,reported: "A commercial fisherman . . . recently. . . found 'thousands' of dead chinook salmonbetween Martins Slough [102 km. downstreamfrom ·Bonneville Dam] and the mouth of theLewis River. [He] said the beaches were litteredwith spring chinook. 'You could find more as youget closer to the Dam.' "

'Unpublished field note•• Washington State Department of Fisheries.

464

The inadequacy of the simple observationalmethod in assessing the true magnitude of mortal­ity is exemplified further in a classic study by theInternational Pacific Salmon Fisheries Commis­sion. At Hell's Gate on the Fraser River, salmonwere blocked by turbulent water at certain river­flows, and an annual loss of thousands (evenmillions) of salmon there has been well docu­mented (Thompson, 1945; Talbot, 1950; Jackson,1950). Although a great mortality was suspected,only a relatively small number of moribund ordead fish were sighted on extensive searches down­stream from Hell's ·Gate over a period of manyyears. Thompson (1945: 96) described the situa­tion at Hell's Gate in 1941: ·'It could be said thatnumbers of them [sockeye salmon] were observedapproaching death, having reached a conditionwhich obviously precluded their passage throughany difficult currents; yet simple observationcould not prove that death actually occurred northat the percentage dying was very high. To findeven hundreds of fish near death along the rifflesin the river, or in the creeks did not necessarilyprove that a great part of the run perished below.Some form of evidence more conclusive wasnecessary."

Thus, on neither the Columbia nor FraserRivers did simple observations of dead fish pro­vide a basis for estimating the true mortality.For this reason, we devised a different method.

Our aims were (1) to estimate the mortality ofadult chinook salmon near Bonneville Dam duringa period when large numbers of salmon werepassing the dam and (2) to evaluate factors con­tributing to or associated with these deaths, suchas streamflow, temperature, turbidity, commercialfishing, fish passage facilities, and disease.Throughout this paper, data on counts of fish atBonneville Dam and on flow, turbidity, andtemperature of the Columbia· River are from U.S.Army ·Corps of Engineers (1943-56).

We first examined records of observations ofdead fish and counts of chinook salmon throughthe ladders at the dam to determine when maxi­mum mortality had occurred. Biologists, com­mercial fishermen, and boat moorage operatorsnear the dam generally reported that the numberof floating dead salmon was greatest in the spring,coincident with high riverflows and large numbersof chinook salmon in the river. We, therefore,selected the spring period of high riverflows for

U.S. FISH AND WILDLIFE SERVICE

carcasses. Search time was recorded in hours andminutes. Table 1 shows the numbers of floaters(by species).

TABLE l.-Floaters observed during three search periods onColumbia River downstream from Bonneville Dam, July ISto September IS, 1954, and numbers of chinook salmon float­ers observed per hour of search

The searches were all made relatively close tothe dam. During the first search period (July 12to August 5), observers in the five boats searchedmidstream and shoreline sections from the damto the mouth of the Willamette. River. In theearly part of this period an additional observerwas stationed on shore on a high dock at Ells­worth, Wash., where he could scan the main riverchannel with binoculars. The second search period(August 9-20) included the same areas. Duringthe final search period (September 4-12), effortwas reduced to a roving search from two boatsbetween Ellsworth and the dam and observationsfrom the Ellsworth dock. One of the two boatspatrolled the 31 km. of river from Ellsworth toCape Horn, and the other the 23 km. of river fromCape Horn to the dam (fig. 3).

1955

Searches by boat for floaters in 1955 werelimited mainly to the main current in midstreamon transects across the river channel at eightlocations. The boats and equipment used byobservers were the same as those used in 1954,and search time was again recorded in hours andminutes. The eight stations and their locations(fig. 3) were (1) St. Helens-93 km. below thedam, and 1.6 km. upstream from St. Helens,Oreg.; (2) Willamette-70 km. below the dam,immediately upstream from the confluence ofthe Willamette and Columbia Rivers; (3) Ells­worth-53 km. below the dam at Ellsworth,Wash.; (4) Reed Island-35 km. below the dam,immediately downstream from Reed Island; (5)Cape Horn-23 km. below the dam; (6) McGowan

Number1.10.20.8

N/£mbrr114

Number812

Floaters observed Chinook----------- floaters

Steel- LTniden- observedChinook head Sockeye titled per hours:llmoll trout ulmon salmonids of search

Period ofsearch

Number NumberJuly 12 to Aug. L_____ 30 7Aug. 9-20______________ 6 14Sept. 4-12______________ 99 54

1954

In 1954 we searched for floaters downstreamfrom Bonneville Dam during three periods (table1). Five boats were used; four were 4.9 to 5.5-m.skiffs propelled by outboard motors, and the fifthwas a 7.9-m. inboard Columbia River gill-netter.The skiffs were usually manned by one personand the large boat by two. Efficiency of observa­tions was assumed to be equal for all boats. Eachobserver was equipped with Petersen disk tags,Polaroid5 glasses, and a dip net to recover floating

• Trade Dame referred to in this publication does not imply endorsementof commercial product.

SEARCHES FROM BOATS

intensive investigation. (We subsequently foundthat summer is usually the period of maximummortality.)

The method we used to estimate the totalnumbers of dead fish was based on a mark-and­recovery technique. Preparatory to making theestimates, we determined the relative floatingqualities of tagged experimental (killed, frozen,and thawed) and "natural" river-killed salmon,located points where salmon lodge downstreamfrom the dam before they float, and made system­atic surveys from boats and airplanes to deter­mine the best sites for observing dead fish nearthe dam. Searchers in boats below BonnevilleDam recovered all the tagged and imtaggedfloating' salmon used for the population estimate.

BCF contracted with the Oregon Fish Commis­sion to make this study.

ABUNDANCE AND LOCATION OFFLOATING SALMON CARCASSES

Dead salmon in the Columbia River are mostevident as "floaters" (partially decomposed float­ing carcasses); hereafter in this paper this termrefers to any floating dead salmonid. Floaters arecarried downstream by the current and can usuallybe seen from a considerable distance. We searchedsystematically from boats and aircraft in 1954 and1955 and also tagged and recovered floaters tolearn how they disperse in the Columbia River.We also introduced tagged dead chinook salmoninto the river at Bonneville Dam in both yearsto provide a basis for estimating total numbers ofdead fish; in 1955 the observations from boatswere used to estimate the mortality of summerchinook salmon.

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 465

WASHINGTON

.'" .' ::',. 'CA'S'C'ADE RAPiD~r ..... '.."......:~ ..STATION B

. THE DALLES

BONNEVILLE DAMSTATION 7MOFFETT CREEK

STATION 6MCGOWAN

STATION 3ELLSWORTH

STATION 4REED ISLAND

STATION 5CAPE HOR~ : .'

40I

STATION 2WILLAMETTE

STATION IST, HELENS

20,

KILOMETERS

OREGON

o,

:."

::'.,;:

'.'..

z«wuouu..U~

·' '. '. COLl!~BIA Rill'" ' ' ' ,.t::R

FIGURE 3.-Locations of search stations on Columbia River near Bonneville Dam where floaters were observed in 1954and 1955,

-7 km, below the dam; (7) Moffett Creek­3 km, below the dam opposite the mouth ofMoffett Creek; and (8) The Dalles-72 km. abovethe dam, 0,8 km, downstream from The Dalles,Oreg,

At each station, an observer in a boat rovedback and forth across the river 8 hours a day on atransect perpendicular to the riverflow, All ob­servations were recorded in one of three cate­gories: river site-midstream search at a station;river vicinity-midstream search at other thanan established station (usually en route to andfrom a station at the beginning and end of theday); and shore-search on foot alongshore(table 2), Eighty-two percent of the total searchtime was at river sites, 15 percent at river vicini­ties, and 3 percent on shore. Stations 6 and 7were alternately manned by the same boat crewat different times, depending on visibility. Ob­servations began on April 4 at stations 2, 4, 6,and 7 and on May 3 at station 8. The observerat station 1 was moved to station 5 at the end ofJune, and on July 12 the observer at station 4was moved to station 3, A total of 1,666 hours

and 50 minutes were spent on all searches in1955 (table 2),

By the time the observations began in April,appreciable numbers of spring chinook salmonhad passed Bonneville Dam. The numbers passingthrough the ladders increased rapidly in late Apriland reached a peak on May 2, when 13,763 fishwere recorded, Only four chinook salmon floaters,or 0.02 per hour of search, were found in Aprilat search stations, and only eight-again 0,02 perhour-were found in May (table 3), Thus, inApril and May, when a record run of 170,205spring chinook salmon passed the dam, the ob­servers found only a few chinook salmon floatersat the four stations between St, Helens and thedam,

During the first half of June, only 11,551chinook salmon passed the dam, but 13 chinooksalmon floaters (0,07 per hour) were found down­stream from the dam. In the second half of June,most of the summer chinook salmon run (33, 951)passed the dam, and searchers found an. increas­ing number as floaters below the dam--o,23 perhour of search (table 3),

466 U,S, FISH AND WILDLIFE SERVICE

TABLE 2.-Time obBe",ers spent searching in boats for floatingsalmon carcasses at eight stations on the Columbia Riflernear BonnelJille Dam, April 4 to July 22,1955

[See text lor dellnltlon of search areas)

Salmon countcd Floaters found I Floaters per hour

Chinook Sockeye Chinook Sockeye Chinook Sockeyesalmon salmon salmon salmon salmon salmonTime period

TABLE a.-Chinook and sockeye salmon counted over Bonne­ville Dam, April 1 to July 15, 1955, and number found asfloaters per search hour at stations below dam, April 4 toJuly 22, 1955

found per hour of search were: May, 0.0; June,0.04; and July, 0.17.

Our experience in 1954 and 1955 demonstratedthat floaters could be effectively counted byobservers in boats.

Total

12 3511 10

ShoreRivervicinity

12 3511 10

Riversite

Hr. Min. Hr. Min. Hr. Min. Hr. Min.

Station and month(station numberIn parentheses)

The Dalles (8)May ._____ 72 10 1 40 35 74 25June •• ._____ 65 15 ••• __ 1 45 57 0July •••_.__ ._____ 88 15 •__._. .______ 88 15

TotaL__ ._•• -2-15-40---1-40---2...:....::.20=:..:-2..:.19=--4~0~

Moffett Creek (7)AprlL • __June • _

Num~r Number0.02.02 _

.07 •

.23 •

.40 0.12

Numberoooo

53

Num~r48

1345

182

Numbero

7768

21,855199,095

Num~rAprlll-30_______ 84,436May 1-31.______ 85,769June 1-15_______ 11,551June 16-30______ 33,951July 1-15________ 23,034

I Floaters found and search time spent on shore search area (table 2) notIncluded.

23 4523 45Total _

McGowan (6)ApriL • 17 40 2125 25

5-----1----1-0--- 30 5May • 53 50 80 5

June___________________ 44 20 26 50 1 50 73 0July .. • 59 30 23 50 24 15 107 35

TotaL .-1--75-20----88--10--27-....:15--290=-- 45

85 30 85 30

1~ ~ ----T-ao-----------45-- Ig~ t&92 35 2 35 -----7- 30 102 40

The greatest numbers of floaters were foundbelow the dam in July: between the dam andstation 2, 182 chinook and 53 sockeye salmonfloaters were recovered (0.40 chinook and 0.12sockeye salmon per hour of search). BetweenJuly 1 and 15, 23,034 chinook and 199,095 sockeyesalmon were counted over the dam. Thus, al­though counts of chinook salmon were declining,the rate of recovery of floaters in early July wasmuch greater than in earlier periods-nearlydouble the next highest rate in June. Counts ofsalmon at the dam and floaters in the river wereboth declining by July 22 when the study ended.

Searches above the dam were limited to station8 from May 3 to July 22. Chinook salmon floaters

Grand totaL 1,361 30 245 35 59 45 1,666 50

SEARCHES FROM AIRCRAFT IN 1954 AND 1955

Synoptic observations of floaters in the Colum­bia River were needed to determine if at any giventime the numbers of floaters were greater belowdams on the river than above. The aerial surveymethod was chosen because Merrell had seencarcasses of chinook salmon on low-altitudespawning survey flights on the Columbia Riverand the upper Snake River in Idaho-a distancetoo great to cover with boat searches. At thetime of our study the Columbia River had onlytwo dams, Bonneville and McNary (237 km. upthe Columbia River from Bonneville Dam).

A chartered two-place single-engine, high-wingmonoplane with a cruising speed of about 120 km.per hour was used for all flights. The pilot sat inthe rear seat, and a biologist in the front seat, fromwhich point the visibility was excellent. A blindspot directly below the aircraft when it was inlevel flight did not significantly hamper observa­tions. Merrell and CoIlins made all observations.Altitude varied between 15 and 100 m. The riverwas searched on both the upstream and down­stream flights. Over sections where the river waswide, observations were confined to one side ofthe river on the first trip and to the opposite sideon the return trip; less than half the total riversurface could be observed in these sections. Innarrower sections, partiCularly between TheDalles and McNary Dams, a greater portion ofthe river surface, and at times the entire width of

Ii

M 4078 30

163 10

387S 15II

19 50

9 010 50

75 4067 40

St. Helens (I)May •June _

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 467

the river, could be seen clearly. Because all flightswere made in the same manner, it is not importantwhether we sawall floaters; the significant pointis that we determined the relative abundance offloaters.

Floaters could be seen easily at the low altitudeat which observations were made. Althoughchinook salmon could usually be distinguishedfrom other species by their relatively large size,all floaters were combined in interpreting the ob­servations. Adverse light conditions and wavesoccasionally reduced the efficiency of the observa­tions. To compensate for the glare on the watersurface, the plane was flown on the side of theriver toward the sun and the observers worePolaroid glasses. Waves and whitecaps were themost serious problems because they reduced thedistance at which carcasses could be seen. For­tunately, visibility was generally good on allflights. We recorded all floaters except in the rareinstance when on the return trip we definitelyrecognized a carcass that we had counted before.We may have counted some twice but believe thisseldom occurred.

We assumed that we saw a nearly constantproportion of the floaters present on each flight.Granting this assumption, the variations in num­bers and distribution of floaters on different flightsreflect real differences. In 1954, between July 16and September 17, floaters were counted on 10flights over the Columbia River between Long­view, Wash., and the McNary Dam (320 km.).Figure 4 shows the numbers of floaters in 8-km.sections of the river on each flight. In 1955, be­tween May 6 and September 13, floaters werecounted on seven flights from the mouth of theColumbia River at Astoria, Oreg., to its con­fluence with the Snake River and up the SnakeRiver to Lewiston, Idaho-a total distance of720 km. Figure 5 shows the numbers of floatersin 8-km. sections of the area surveyed. The surveybetween Astoria, Oreg., and Lewiston, Idaho,extended over 2 days, July 14 and 15.

Floaters in both years were not uniformlydistributed throughout the river but were con­$istently concentrated at certain locations. Thegreatest density was downstream from Bonne­ville Dam when riverflows were high or whenrelatively large numbers of migrating salmon werepresent. Floaters were present, but at much lowerdensities, throughout the river between Long-

468

view and McNary Dam in July of both years­probably as a result of rapid downstream dis­persion from points of occurrence of high mortalityduring this high-flow period.

Most flights were at times when no experi­mental dead chinook salmon (released by us)were in the river. Only on the flights of Septem­ber 9, 1954, and July 9, 1955, could experimentalfish have been present; six were seen below Bonne­ville Dam on the first date and four on the second.

The observations on the flights of July 14 and15, 1955 (fig. 5), were of particular significancebecause they included the greatest length ofriver of any surveys and were made during theperiod of our experiment to estimate chinooksalmon mortality. Visibility was good on bothdays, and we believed our observations revealedtypical distribution of floaters at high river­flows. The fact that floaters were most numerousbelow Bonneville Dam suggests that dead fishwere originating near the dam. In the lowest 225km. of the Snake River, where there were nodams, only three floaters were seen, despite thepresence of large numbers of live sockeye andchinook salmon that were migrating up the river.

The aerial surveys were useful in showing thatfloaters were usually more numerous below Bonne­ville Dam than in other areas and in indicatingthe times when the greatest numbers of floaterswere present.

MARK-AND-RECOVERY EXPERIMENTSTO ESTIMATE ABUNDANCE OFDEAD CHINOOK SALMON

In the mark-and-recovery technique we usedto measure mortality of chinook salmon nearBonneville Dam during periods of high flow,carcasses were tagged and introduced into theriver at the dam in 1954 and 1955. The recoverysample consisted of the floaters found on boatsearches downstream from the dam at the searchstations shown in figure 3. The sample containedthe tagged carcasses as well as the carcasses ofchinook salmon that died naturally in the river.Only the 1955 experiment had enough recoveriesfor us to estimate the numbers of dead chinooksalmon.

Our experiments differed from an experimentthat would lead to a standard Petersen-type ofpopulation estimate in one important respect:Instead of removing carcasses from the river,

U.S. FISH AND WILDLIFE SERVICE

tagging them, and returning them to the river,we acquired carcasses from other sources, tagged

them, and added them to the population of car­casses already in the river.

oLU>It:IIJenCDoenIt:IIJ!C(g~

enIIJ en..J

en :J:IIJ ...I IIJ..J~ Z ... ...1 ...I ..Jer

~IIJ a:er ..J~IIJ ...I 00 > ero u )0.

s: ~ ... IIJ 0 a:IIJ z 0= ...I crCl %: (I);:)Z2 IIJ..J er Z2Z -10

9 ..,; ...I a: ocr :J:I&J ...I ucren LU~ CDO ... U CD 20

.JULY 164

100 0 100 200a DOWNSTREAM I UPSTREAM

DISTANCE FROM BONNEVILLE DAM (KILOMETERS)

FIGURE 4.-Distribution of floaters observed in 8-km. sections of Co­lumbia River between Longview, Wash., and McNary Dam (320 km.),July 16 to September 17, 1954.

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 469

'"ZIIJ.JIIJ::z:.-:'"

I&J.J.J:>I&Jzz20<1:IDO

'"",.JIIJ.J.J<I:.JLLCl:o0.J1IJ::i::Z:I&JI-U

~

U

SCl:.JID

4

o

o

4

o

MAY 6

JUNE 14

--JUNE 27

- •JULY 9

I- _.. I • . L II.. • ..JULY 14,15

• .. ... • I • .1 • . I • -SEPT. 13

.. J - .ZOO 100 100 ZOO 300 400 500

DOWNSTREAM I UPSTREAMDISTANCE FROM BONNEVILLE DAM (KILOMETERS)

FIGURE 5.-Distribution of floaters observed in 8-km. sections from the mouth of Columbia River, Astoria,Oreg., to its confluence with the Snake River and up the Snake River to Lewiston, Idaho (720 km.),May 6 to September 13, 1955.

It is a simple matter to modify the standardPetersen-type of estimate to fit the circumstancesof our experiment, as is shown below. But first,it is instructive to identify our experiment as atypical example of a much larger class of CIR(change-in-ratio) experimental techniques (Paulikand Robson, 1969).

CIR experiments are designed to estirpatepopulation characteristi~s, s1.1.ch as abundance,productivity, rate of expl9itation, and survivaland mortality rates. Any CIa estimate is basedon the observed differences in the relative num­bers of two distinguishable types of individualsat two points in time when the population isobserved. In our experiment the two types ofcarcasses are those that are marked (x-type indi­viduals) and those that are unmarked (y-typeindividuals); the first observation of the popula­tion is made immediately before tagging (time 1),and the second observation is mage at time ofrecovery (time 2). ...

470

Throughout the following analysis an importantdistinction is made bet~een carcasses (bothtagged and untagged) that are potentially re­coverable at the time of death and carcassesthat are unrecoverable at the time of death.These two classes of carcasses are mutually ex­clusive. A potentially recoverable carcass is amore or less unmutilated carcass of a fish thatsinks to the bottom of the river after it dies.After several days, it mayor may not float tothe surface and drift through the area in whichrecovery crews are searching. Some recoverablec~rcasses do not float because they are buried orwedged on the bottom, eaten by scavengers,stranded on the bank, or otherwise preventedfrom floating (these carcasses are still consideredrecoverable). An unrecoverable carcass is thecarcass of a fish so severely mutilated at the timeof death that it cannot float to the surface and berecovered. Thus, a recoverable carcass has afixed chance of floating to the surface where it

U.S. FISH AND WILDLIFE SERVICE

(3)

(4)

can be recovered, whereas" an unrecoverable car­cass has no chance of floating to the surface.(All floating salmonid carcasses are consideredhere as floaters.)

We follow Paulik and Robson (1969) in intro­ducing the following notation:

Y 1 = number of untagged recoverable car­casses at time 1. (We wish to estimateY 1. In our estimate based on 1955 d!lta,this quantity is interpreted as thenumber of recoverable carcasses belowBonneville Dam resulting from earliermortalities that could possibly be re­covered during our sampling period.)

N 1 = number of tagged and untagged re­coverable carcasses combined at time 1.

PI = fraction of marked recoverable car­casses at time 1.

P2 = fraction of marked recoverable car­casses at time 2.

R,. = change in population of marked re­coverable carcasses between time 1and time 2 (i.e., the number of taggedcarcasses introduced into the river be­cause it is reasonably assumed thatall marked carcasses are recoverable).

R = change in total population of markedand. unmarked recoverable carcassesbetween time 1 and time 2.

n2 = number of carcasses observed in asample at time 2.

X2 = number of marked carcasses observedin a .sample at time 2.

P2 = Xdn2 = estimate of P2.The fraction of marked recoverable carcasses at

time 2 can be written as the ratio of the numberof these carc~s at time 1 corrected ·for changesthat have taken place between times 1 and 2 tothe total number. of tagged and untagged re­coverable carcasses at time 2. That is,

A P1N 1 + R,. (1)P2 = N 1 + R

Solving (1) for N 1 gives

N1~ R,. - P2R (2)

P2 - PI

Because in our experiment there was no changein the population of unmarked recoverable car­casses, R = Rx, and because the population con­sisted entirely of unmarked recoverable carcasses

at time 1, N 1 = Y1 and PI = O. Substituting thesequantities and replacing unknown quantities withtheir estimates, we arrive at an equation for esti­mating the number of recoverable chinook salmoncarcasses that were killed near Bonneville Dam:

Yl = R~~2 - 1)If all recoverable carcasses, marked and un­

marked, are equally likely to be included in therecovery sample and if the recovery sampling iswith replacement (as it was), then it is appro­priate to make use of binqmial sampling theoryto set a confidence interval about Y1. To do this,upper and lower limits are set on P2 (the propor­tion of marked carcasses in the recovery sample(Pearson and Hartley, 1966», and these limits arethen converted to upper and lower limits for t"1

by substituting in equation (3).As was noted above, this same result can be

developed from a Petersen-type estimate bynoting that the total number of recoverable car­casses at time of release is t 1 + Rx ; the numbertagged is Rx ; the size of the recovery sample isn2; and the number of recoveries is X2. Hence,

A + R _ R,.n2Yl ,. - X2

When solved for "th this yields expression (3)above.

The validity of this method for estimating thenumber of recoverable carcasses in the river be­low Bonneville Dam depends on several basicassumptions. First, we assume that carcasses ofuntagged chinook salmon float and are similarin all other significant respects to the taggedcarcasses we introduced into the river. With re­gard to floating qualities, we demonstrated thevalidity of this assumption experimentally byusing fresh and frozen chinook salmon carcasses·.(These experiments are described in the. appendix.)Differences in floating characteristics were insig­nificant at water temperatures similar to thoseat the time of our 1955 experiment~ ,

Broadly interpreted, the above assumpt~on im­plies that all of the untagged carcasses were offish that had died near the dam; our extensiveobservations substantiate this implication. ·In thefall of 1954; when riverflow was low and carcasseswere unlikely to be swept far downstream aft~

death before bec~ming sufficiently buoyant to

CHINOOK SALMON ~ORTALITY IN COLUMBIA RIVER NEAR BONNEVILl"E DAM 471

For these data, x2 = 15.35 with 1 d.f., and thehypothesis of homogeneity is strongly rejected.(The observed frequency for recoveries from spill­way releases is smaller than is generally consid­ered desirable for this statistical treatment, butin this experiment the conclusion is so clear-cutthat we need not be greatly concerned over thisfact.) The conclusion is that carcasses released inthe spillway are less likely to be recovered thancarcasses released at the powerhouse.

The small number of carcasses recovered fromthe spillway release indicated that many of themhad probably disintegrated in the extremelyturbulent flow of the spillway rollback (fig. 2).Such carcasses would have a reduced chance offloating and being recovered. The less turbulentflow of the powerhouse discharge probably con­tributed to the higher recovery rate of carcassesreleased there.

Although the 1954 experiment did not produceadequate data to estimate precisely the number

charge at the downstream face of the powerhouse.Finally, 50 were dropped into the river in Sep­tember from the Bridge of The Gods, 8 km. abovethe dam, and 20 were released in August from aboat at Oneonta. 11 km. below the dam. .

Less than 1 percent of the 1,095 carcasses wererecovered. Only one of the 280 carcasses releasedin August was recovered; it was found at Oneontaon August 10 at the same location where it hadbeen released on August 5. Eight carcasses wererecovered from the 815 released in September­six from the powerhouse and two from the spillwayreleases. No carcasses were recovered from thegroup dropped from the Bridge of The Gods.

The significance of the apparent difference inthe rates of recovery of the carcasses released atthe two sites in September can be evaluated by achi-square test for homogeneity. The null hy­pothesis is that the probability of a carcass beingrecovered does not depend on release location.To test this hypothesis, we constructed the follow­ing fourfold contingency table: '

Powerhouse Spillwayrelease site release site Total

float, most carcasses were found within a fewmiles downstream from the dam. In the summer of1955, many fresh dead salmon were found on thebottom in shallow water on gravel bars immedi­ately below the dam. Finally, very few floatingcarcasses were observed during aerial surveys ofthe forebay above the dam. (See also the data onp. 15, last paragraph.) .

Another assumption is that no tags were lostfrom recoverable carcasses. This assumption issupported by the fact that there was no evidenceof missing tags on untagged chinook salmonfloaters. The tags were fastened to the jaw in 1954and to the caudal peduncle in 1955; both locationsare exceptionally secure anchoring points fortags on dead salmon.

1954 EXPERIMENT

To secure carcasses for releases in 1954, 1,095chinook salmon were placed in frozen storage inthe summer and fall of 1953. All were ice glazedto retard dehydration and oxidation. About 245of these fish were spring chinook salmon of bothsexes that had been killed during construction ofLookout Point Dam on the Middle Fork of theWillamette River; the rest were fall male chinooksalmon from Bonneville and Oxbow Hatcheries onthe Columbia River.

Frozen chinook salmon float when placed inwater; to ensure that they would sink like a natu­rally killed salmon, the frozen carcasses werethawed in air for 24 hours before being tagged andreleased. The tags were sequentially numbered,bright-colored nylon or plastic ribbons fastenedto the jaw. The tagged carcasses were releasedduring two periods: The first group (280 carcasses)was released August 4 to 6 when spillway flowsaveraged 3,300 c.m.s. and powerhouse flows4,100 c.m.s.; the second group (815 carcasses) wasreleased September 2 to 3 when spillway flowsaveraged 1,300 c.m.s. and powerhouse flows· 3,700c.m.s.

Earlier evidence suggested that the spillwaychannel was the source.of most dead salmon nearthe dam. Therefore, most of the tagged carcasses(240 in August and 635 in September) were re­leased into the spillway channel either by droppingthem from the dam immediately above the gatesor into the rollback below the gates (fig. 2).Another 150 carcasses (20 in August and 130 inSeptember) were dropped into the draft tube di,;-

Recovered •• _Unrecovered. _Total releases _

6124130

2633

. 635

8767765

472 U.S. FISH AND WILDLIFE SERVICE

of recoverable carcasses, we can, if we assumethat all carcasses released in the powerhousechannel were recoverable, make a crude estimateof the fraction of" recoverable carcasses from thespillway releases. "

If none of the carcasses released at the spillwayhad been rendered unrecoverable, the proportionof recoveries from the spillway releases shouldhave equaled the proportion of recoveries fromthe powerhouse releases. On this assumption, theexpected number of recoveries from spillway re-

. (635)(6)leases IS 130 = 29.3. Because only two car-

casses were actually recovered, we can estimatethat the proportion of recoverable carcasses forfish dying by being swept over the spillway is only

~ = 0.0683 and that the proportion of unre­29.3coverable carcasses is 0.9317. Because this esti­mate is based on small numbers of recoveries,however, sampling error could be large.

1955 EXPERIMENT

In 1955, 1,169 carcasses were released. Theywere all male fall chinook salmon collected in1954 from Bonneville, OiXBow, and Spring CreekHatcheries. They were killed by a blow on thehead and placed in frozen storage a few hourslater; as in 1953, the carcasses were ice glazed toretard dehydration and oxidation during over­winter storage.

When the carcasses were removed from storagefor the 1955 experiment, they were handled inthe same manner as the carcasses for the 1954experiment except that they were tagged withsequentially numbered cellulose-acetate Petersendisks fastened by nickel pins through the caudalpeduncle.. Figure 6 demonstrates the rationale on whichour entire 1955 study was based: the number ofchinook salmon floaters is an indication of thetotal number of dead chinook salmon in the river.Few floaters were recovered until late June, when

-UIozcrII);:)o%I-

10­:=Iig~

~4ooI&JI­Z;:)

o5 u

zo:E..J

~:ll:

g~J:U

L..r--_........IO15 25

MAY

r"v-,."\ i \

(' '""", " TOTAL RIVER FLOW

I \

! "'-I ''4''~.i FLOATERS \• RECOVERED \! PER HOUR \

I 'I \

I \~\ IJI"', \,

/', I "

......' \,)

CHINOOKSALMONCOUNTEDAT DAM

52515APRIL

5

2

13

15

14

~ 12z

~ "~

o~IO

II) 9:Eu 8~

9 7II.

ffi 6>a::..a~~

-a:ILlIII:=Ii;:)ZILl

~a:ILl

~~8

5%a:ILlQ.

e6a:

~uILla:II)

15 4

~II.Zo32crII)

~~Iu 0 L-__...L--r--,.--,....u.,-'i......---II~r-T"'-,-....----'.,.......,.

5 15 25 5 15 25JUNE JULY

FIGURE 6.-Daily counts of chinook salmon and total riverftow at Bonneville Dam and number of untaggedchinook salmon floaters recovered per hour, Bonneville Dam to St. Helens, 1955.

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 473

[Only data actually used In computations are included]

TABLE 4.-Data for computing mortality of the sum·merchinook salmon run near Bonneville Dam, 1955

Another minor problem was the removal ofuntagged carcasses (either floating or stranded inshallow water alongshore) from the river by

Tagged floatersChinook salmon recovered from

counted at Tagged releases of UntaggedBonneville Dam carcasses floaters

Date ladders relessed June 30 July 1 recovered

Numlnr Numlnr Number Number Number

1171813

338 _

1,169

3,5674,7984,9123,8903,8682,2202,3431,6801,2911,185

48,766TotsL __

day at each of the seven stations below the dam(fig. 3). Theonly exception was July 16, when nosearches were made. This was the 16th day afterthe releases on June 30 and the 15th day after thereleases on July 1; for this reason estimates offractions of recoverable carcasses available forrecovery a given number of days after release(discussed on p. 16) are slightly biased.

The absence of recovery effort on July 16 alsoaffects the estimate of Y1, the number of recover­able untagged carcasses in the river. However,trial calculations made by using plausible valuesfor the numbers of tagged and untagged carcassesthat might have been recovered on July 16 sug­gest that the likely error is slight. For example, ifwe assume that one tagged carcass and six un­tagged carcasses would have been recovered-aprobable event in the light of the data of table4-our estimate of Y1 (given on p. 17) wouldchange by only 1.8 percent.

June:21. _22 _23 _24 _25 _26 _27 _28 _29 _30 _

July:1._________ 1,411 831 _2__________ 1,713 _3__________ 1,667 _

L::::::: ~:~ :::::::::::-----0----------ii::::::::::::::::6__________ 2,576 0 0 _7__________ 1,940 0 0 88__________ 1,339 3 4 159__________ 1,241 5 7 13

10__________ 962 0 0 711._______________________________________ 1 0 1312________________________________________ 4 4 1913________________________________________ 0 0 1514________________________________________ 0 1 1615________________________________________ 0 1 916 _

17________________________________________ 0 1 218________ ___ __ 0 0 _19________ 0 0 _20___________ _ ____ _ 0 0 _21._______________ _ _ 0 0 _22_______________ _ __ 0 0 _

the rate of recovery rose rapidly. We then re­leased our entire supply of carcasses over a 2-dayperiod, June 30 and July 1, so that our estimateof mortality would include a period during whichlarge numbers of salmon were dying. The countsof chinook salmon migrating over the dam andthe riverflows were also near their summer maxi­mums. Thus, large numbers of live fish were pres­ent, coinciding with inimical high flows.

Because of our experience in 1954 with disinte­gration of carcasses in the extremely turbulentflow in the spillway rollback, we released thetagged carcasses in 1955 into the spillway channelfrom a boat downstream from the rollback. Inthat area high velocities and extreme turbulencesshunt migrating salmon toward the fish ladderentrances at either end of the spillway (figs. 1and 2). At the release points, which were dis­tributed across the river, one man kept the boatin position under power while a second pitchedthe tagged carcasses overboard-1,068 were re­leased in the spillway channel (183 m. below thedam) and 101 in the powerhouse channel (64 m.below the dam).

The 2-day release period coincided with a tem­porary slackening of chinook salmon passage overthe dam; the count dropped from a peak of 4,912fish on June 23 to slightly more than 1,000 perday near the end of June. At this time, manylive fish were below the dam and were liable to bekilled, as shown by a second peak count of 3,434on July 4 and large numbers of chinook salmonfloaters through the first half of July (fig. 6).

The search by boats for floaters downstreamfrom Bonneville Dam was not begun until July 5because we knew from previous experiments onthe floating characteristics of chinook salmon car­casses that the tagged carcasses would not floatin fewer than 5 days at the existing water tem­perature of 13.9° to 14.4° C. (see appendix).Daily searches for floaters continued until July 22.Table 4 gives the release and recovery data onwhich we base our estimate of the number ofrecoverable chinook salmon carcasses from mor­talities at the dam from June 21 through July 10,1955. No tagged carcasses were recovered fromJuly 18 through July 22. Table 4 includes onlydata actually used in computing the estimate.

The level of recovery effort throughout therecovery period was essentially constant-an ob­server in a boat recovered floaters for 8 hours a

474 U.S. FISH AND WILDLIFE SERVICE

sturgeon fishermen. Putrefied salmon flesh isregarded by many sturgeon fishermen as superiorbait, and one of the most productive locationson the Columbia River for sturgeon fishing andfor finding putrefied salmon is just below Bonne­ville Dam.

Beacon Rock Moorage, 6.4 km. below the dam,is the principal base of operations for sturgeonfishermen in the area. Lee Motley, a moorageoperator, recorded carcasses found by his em·ployees and customers near the dam during thespring and summer of 1955. His data are as fol­lows:

Of the carcasses found by sturgeon fishermenin July, a relatively large number were foundduring the sampling period, July 7 to 17 (23untagged and three tagged chinook salmon and24 sockeye salmon). This record probably includesmost of the carcasses removed by fishermen inthe study area.

The numbers of carcasses found by sturgeonfishermen increased steadily during the summer.The relatively large number of carcasses of bothchinook and sockeye salmon in July is of particularsignificance in relation to mortality at the dam.Many of these were not floating but were freshlykilled fish that had collected on a shallow sub­merged bar near the upper end of Hamilton Island,close to the spillway. Each year Lee Motley re­trieves many fish on this bar.

Some carcasses recovered by sturgeon fisher­men probably would have been found at our regu­lar search stations had they not been intercepted.The fishermen returned the three tagged carcassesto the river because they knew of our study anddid not want to interfere with it; two taggedchinook salmon floaters were subsequently re­covered at station 6 (fig. 3). The removal of 23untagged chinook salmon carcasses by mooragepersonnel during the period of the experimentresults in a slightly smaller estimate of mortalitythan would have been obtained if a few of theremoved untagged carcasses had been recovered

Sockeye SteelheadUntaggl.'d Tagged salmon trout

Chinook salmon

at search stations. If these were typical recover­able carcasses, a correction for the resulting biascould readily be made by the method suggestedby Paulik and Robson (1969). However, carcassesremoved by fishermen probably have a higherprobability of floating than typical recoverablecarcasses, so that a bias correction is not possible.In any event, the bias is very slight and does notsignificantly affect our conclusions.

The removal of carcasses by scavengers wasalso considered. Scavengers near Bonneville Damthat feed on dead salmon both before and afterthey float include sturgeon, squawfish, gulls,crows, raccoons, and skunks. Sometimes almostevery floater was accompanied by one gull or more,but at other times, only a few were accompaniedby gulls. The effect (if any) of gull scavenging onthe length of time carcasses remain at the surfacewas not determined. Floaters that go agroundprobably disappear more quickly than those thatremain afloat because they are more accessibleto terrestrial or avian scavengers. Gulls and crowswere frequently seen feeding on dead salmon thathad drifted ashore. Nocturnal feeding by raccoonsand skunks, which are numerous in the area, maybe even more important in the rate of disappear­ance of carcasses along the riverbanks.

.Hence, scavengers reduce to an unknown extentthe number of dead salmon available to be ob­served. Whatever effect scavengers might havehad, there was no evidence. that it was differentfor tagged and untagged carcasses in the searcharea. Therefore; scavenging was assumed to haveno significant effect on the estimate of mortality.

Floaters originating above the dam could alsoaffect our mortality estimate, but we believe thatfew, if any, such chinook salmon floaters driftedinto the recovery area during the estimate period.The best and most direct way to evaluate thepossibility of recruitment of floaters originatingabove the dam into the search area below thedam would be to observe the river from above thedam. We looked for floaters on two dates from theBridge of The Gods, which spans the river 4.8 km.above the dam, where the entire surface can easilybe seen. No chinook salmon floaters were ob­served passing under the bridge on July 10. 1955,whereas seven were sighted below the dam. Again,on July 3, 1956, observers were stationed at sta­tion 5 (Cape Horn), and on the Bridge of TheGods. In 6 hours and 35 minutes no chinook

4.33 39

13

2948

AprIL _May • _June _July _

Search period

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 475

Tagged floaters Total tagged floaters recoveredDays after release recovered on each day after rele8S6

, Two recoveries on July 9 were arbitrarily assigned to releases of J one 30and July 1 because the release datee were unknown.

floaters rapidly passed through the entire recoveryarea within a few hours during high riverflows.

TABLE 5.-Nllmber and percentage of 81 tagged chinooksalmon floaters recovered beloUl Bonneville Dam on eachday after release, 1955

Ptrc~nt .mootMdbV tllr~~.

21.9919.8715.7110.469.43

10.466.282.102.101.57

Percent

12.9032.2616.130.00

16.1312.903.233.230.003.23

Numbu N'Umb~r

7___________ 48___________ '109___________ ' 5

10___________ 011___________ 512___________ 413___________ 114___________ 115___________ 016___________ 1

Knowledge of the estimated percentage of re­coverable carcasses that have a chance of float­ing and becoming available for recovery anygiven number of days after death (table 5)enables us to estimate the percentage of recover­able carcasses of fish dying on any given datethat have a chance of flqating and becomingavailable for recovery at some time during theJuly 7 to 17 recovery period. The method forcalculation of these estimates is shown in figure 7which, together with table 4, "indicates the struc­ture of the 1955 experiment. The figure showsthat, of the recoverable carcasses of chinooksalmon dying on June 21, an estimated 1.57percent were recoverable 16 days later on July 7,the first day of the recovery period. Similarly,considering recoverable carcasses from mortalitieson June 22, 2.10 percent were recoverable 15days later on July 7 and 1.57 percent 16 dayslater on July 8; consequently, a total of 3.67percent of the carcasses from June 22 mortalitieswere recoverable during the recovery period.Computations for other days through July 10follow the same pattern.

We are now ready to estimate Y11 the numberof untagged recoverable carcasses in the riverduring the July 7 to 17 recovery period. As hasbeen noted, of 1,169 carcasses tagged and re­leased on June 30 and July 1, 31 were recoveredfloating during the recovery period. At the sametime, 117 untagged floating chinook salmon car­casses were observed (table 4). Assuming thatall tagged carcassp.'3 are recoverable, Rx = 1,169,

U.S. FISH AND WILDLIFE SERVICE476

salmon floaters were seen from the bridge, butfive were found in 1 hour and 50 minutes at sta­tion 5 below the dam. Furthermore, during theperiod of our study in both 1954 and 1955,greater numbers of floaters were usually found inthe 48 km. of river below Bonneville Dam thanin any other section of river. No adjustment in thecalculated mortality appears necessary for re­cruitment of untagged floaters into the samplearea from above the dam.

The 31 tagged floaters used in the mortalityestimates were recovered by our search crewsbelow the dam from July 8 to 17 (table 4). Thefirst tagged floater was found by a fisherman onJuly 7, the 7th day after the first date carcasseswere released (June 30); but it, as well as threeother carcasses found by fishermen, was not usedin computing the mortality estimate. Our search­ers actually recovered 32 tagged floaters, but onewas eliminated from the computations becauseit was on shore.

No tagged floaters were recovered by searchcrews after July 17, although sampling continuedthrough July 22 (table 4). Because no taggedfish were recovered before July 7 or after July 17,we assume that tagged carcasses were availablefor recovery only during the period July 7 to 17.During this 1i-day period, 117 untagged chinooksalmon floaters were recovered at search stations.

The "tagged chinook salmon floaters were re­covered from 7 to 16 days after they were released(table 5). Because tagged and untagged carcassesbehave in a similar fashion, the smoothed dailypercentages of total tags recovered given in table5 can be interpreted to mean that 21.99 percentof the recoverable carcasses of chinook salmondying on a given day will have a chance of float­ing and becoming available for recovery on the7th day after death (of this 21.99 percent, somewill actually float and others will not); 19.87 per­cent will have a chance of becoming availablefor recovery on the 8th day; 15.71 percent on the9th day; and so forth. Note that an additionalassumption is being made here: An individualfloating carcass is available for recovery only onthe day that it floats because it drifts out of therecovery area in less than 1 day. This assumptionis supported by experiments in which chinooksalmon floaters were tagged and later recovered(see appendix). These experiments showed that

I DAYS CARCASSES ON BOTTOM I R:C~~~~~~ALGl 00: if~c-tt~SO~~HI DAY OF DEATH I I ,r--16TH DAY AFTER DEATH--"7

I -.L., 'OjOj ",'I. 'I.' ",ro ~ ",ro '" ° ° ",I'" ,~. ~. ~. ,0' Oj~ ,0' ro'!' ","i ","i ,":>I JULY 10 -------------9' ' , , , ' , ' , 21.99

J,ULY 9 ------------- I 41.86JULY,8 _--- t I '57.57

JULY 7 t----------------t I' '68.03~JULY 6 --+- t I' 77.46 ,,-<i>0

.JULY 5 ---1------- - tit '87.92 <S',<-q,.:~JULY4 ----i------- ,r '94.20 q,.,<-O,,-qj!-0

JULY 3 ------i----- '96.30 ~r:. ...'1.I. ~{!-

JULY2-------i--- , '98.40 Q'IJ(..O.Ji.

-=======::z::::z:::::z:::::z~I' ",'<- q,.'<;JULY 1 -----------t--- . 99.97 ..'b~I f<,""~JUNE 30 ~ 2 iii 99.97 0.Ji. ..~

I ",Or:.JUNE 29 4 i i 77.98 .."" '*'1!C;i ","

JUNE 28 ---------- I 58.11 "..... " ..tJ ..

JUNE 27___________ I '42.40 (..~..r:.'Ci

JUNE 26_---. ' I 31.94 ~O~c.'<-~ IJUNE 25 , I 22.51 ,.-1:-'11'...'1.'<' I

~" ~JUNE24____________ I 12.05 '1.'<;!<Q'" I'" 0JUNE23 , I 5.77 ~~~... I

JUNE 22 ----------- I 3.67 ~.;:,..)''' IJUNE21_---_______, , , , , , I 1.57 tJ

~~~~~~~~~~ I"," ,OJ' ,.,. ,0' OJ' ,0' roo "'. "'. ,.

j--".DAY RECOVERY PERIOD--l

21 22 23 24 25 26 27 28 29 30 I 2 3 4 5 6 7 B 9 10 II 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27JUNE JULY

FIGURE 7.-Availability of carcasses for recovery. When chinook salmon die, the carcasses sink to the bottom. Recoverablecarcasses mayor may not float to the surface between 7 and 16 days after death. When a carcass floats, it can be re­covered only on the day it floats. Percentages of recoverable carcasses that become recoverable on the 7th through 16thdays after death are given at the top of the figure, and the sums of these percentages falling within the ll-day recoveryperiod are given on the right diagonal edge for each date of mortality.

X2 = 31, n2 = 117 + 31 = 148, and P2 = 31/148 =0.2095. Substituting in equation (3) gives anestimate of Y1:

y1= RX(~2 - 1) = l,169(0.2~05 - 1) = 4,412

untagged recoverable carcasses. An approximate95-percent confidence interval on P2 is 0.12 :$0.2095 :$ 0.33. The ~ corresponding 95-percentconfidence interval on Y 1 is 2,373 :$ 4,412 :$ 8,572untagged recoverable carcasses.

It is clear from the preceding discussion andfrom figure 7 that this estimate of 4,412 untaggedrecoverable chinook salmon carcasses represents:

(1.57 percent of the recoverable carcasses fromJune 21 mortality)

+ (3.67 percent of the recoverable carcassesfrom June 22 mortality)

+ ... + (21.99 percent of the recoverable car­casses from July 10 mortality).

PROPORTION OF CHINOOK SALMONRUN KILLED .

To 'estimate the proportion of the chinooksalmon run killed near Bonneville Dam duringour experiment in 1955, we relate our estimate ofthe number of recoverable carcasses in the riverbelow the dam (i.e., Y1 = 4,412 recoverable car­casses) to the size of the run producing these car­casses.

Two additional factors should be taken intoaccount. In our 1955 experiment, all tagged car­casses were released immediately below the dam.This situation would correspond to one in whichall chinook salmon mortality at the dam occursbefore the fish are counted over the dam. However,

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 477

earlier tagging experiments in which salmon weretagged as they emerged from the ladders abovethe dam showed that some salmon are swept downover the dam and are caught by fishermen belowor counted as they reascend the ladders (Schoningand Johnson, 1956). Some of the salmon that areswept over the dam probably do not survive,although we have no direct evidence of what por­tion is killed. The mortality model that we havedeveloped provides for the possibility that someof the salmon counted over the dam are subse­quently killed by being swept back over the dam.

We have evidence from our 1954 experiments(previously described) that carcasses of fishkilled by being swept over the spillway may beso severely mutilated by the extreme turbulencethat they are rendered unrecoverable. Thus, thesecond additional factor taken into account inour mortality model is the possible presence ofunrecoverable carcasses from mortality occurringafter counting.

To derive the mortality model, some additionalsymbolism is required. In these symbols, the sub­script i refers to days on which mOltality occursthat could possibly produce floating carcasses forrecovery during the recovery period. (Recall thatrecoverable carcasses may float and be availablefor recovery 7, 8, ... , or 16 days after death.)Thus, i = 1 corresponds to June 21 because thisis the first day which could have produced float­ing carcasses for recovery during the recoveryperiod. (The first day of the recovery period,July 7, is the 16th day after June 21.) Similarly,i = 20 corresponds to July 10, the last day whichcould have prod!lced floating carcasses for re­covery during the recovery period. (The last dayof the recovery period, July 17, is the 7th dayafter July 10.) Reference to figure 7 will help toclarify this subscripting scheme.

M = proportion of the chinook salmon rundying near Bonneville Dam. (M is thequantity to be estimated. If the propor­tion of the run dying at BonnevilleDam remained constant from June 21through July 10, then M is this quan­tity. However, if the proportion dyingvaried during this period, M can bethought of as a weighted average ofthe daily proportions dying, with theweighting being related to the extent towhich carcasses from a given day's

478

mortality become available for re­covel"y during the July 7 to 17 recoveryperiod.)

Ci = count over fish ladders on day i.D i = mortalities on day i below the dam.qi = proportion of total mortality on day i

producing carcasses which, if recover­able, could be recovered during theJuly 7 to 17 recovery period.

Note that Ci + D i is the total run on day i andthat M(C i + Di)qi is the total number of mortal­ities on day i producing carcasses, which, if re­coverable. could be recovered during the recoveryperiod. Therefore,

~ (~h )~ M(C i + Di)qi = M ~ Ciqi + ~ Diq; (5)

is an expression for the total number of mortalitiesproducing carcasses (both recoverable and unre­coverable) which, if recoverable, could be re­covered during the recovery period. For 1955, the

~ .term L: Ciqi can be estimated from our experl-

;=1

mental data and the observed fish ladder countsat Bonneville Dam. Table 6 shows this calculation.

20

TABLE G.-Estimation of L: C;qi fOT 1955 experiment ati==l

Bonneville Dam

CI q;ClqlDate (see table 4) (see table 5)

June:3,567 0.0157 56.021_________

122_________2 4,798 .0367 176.1

23_________ 3 4,912 .0577 283.424_________ 4 3,890 .1205 468.725_________

5 3,868 .2251 870.726_________6 2.220 .3194 709.127_________7 2,343 .4240 993.4

28_________ 8 1,680 .58ll 976.229_________9 1,291 .7798 1,006.730_________ 10 1,185 .9997 1,184.6

July:11 1,4ll .9997 1,410.61_________

2_________12 1,713 .9840 1,685.6

3_________ 13 1,667 .9630 1,605.34_________ 14 3,434 .9420 3,234.85_________

15 2,729 .8792 2,399.36_________16 2,576 .7746 1,995.47_________17 1,940 .6803 1,319.88_________18 1,339 .5757 770.9

9_________ 19 1,241 .4186 519.510_________20 962 .2199 211.5

20Estimate of L: Ciql = 21,877.6

i=1

We now define the quantities required to takeaccount of (1) deaths that occur after countingand (2) the possibility that some of the carcasses

U.S. FISH AND WILDLIFE SERVICE

(11)20 C· YL iqi + 1

i =1

M

M=(20 )() 20i~ Ciqi + Y1 1 - fa + farai~ Cjqi

(10)

If all mortality occurs below the dam (fa = 0),then (10) becomes

20

(8) is solved for L D iq i, and the result is sub-i=1

stituted into (7). An expression for M is thenderived by solving (7). The final result is:

Yl

This situation corresponds to the manner in which'we performed our experiment ip. 1955 when wereleased all tagged carcasses below the dam.Recalling that t i = 4,412 untagged recoverable

20

carcasses and that :E Ciqi was estimated to bei=1

21,877.8 fish, we use (11) to calculate l\?r = 0.1678.In other words, on the assumptipn that all mortal­ity occurs below the dam, we estimate that 16.78percent of the chinook salmon run was killed nearBonneville Dam in 1955 at the time of our experi-ment. ,.

It would be desirable to set a confidence intervalabout the estimate of mortality level given byequation (11). Unfortunately, this does not seemto be possible. Ail quantities appearing in equa­tion (11) are subject to sampling error. Thevariances of the qj'S and the variance of Y1 couldbe approximately estimated from our experi­mental data for 1955, but no estimates are avail­able for the variances of the C;'s-the dailychinook salmon counts over the dam. These countsare known to be inexact and may also be biased.Some of the causes of counting errors are: fish arecounted more than once as a result of being sweptover the spillway; fish are not counted through theship navigation lock; and fish are misidentified,especially the smaller salmon with similar appear­ances such as sockeye salmon and chinook salmonjacks. For example, in 1957, only 9,879 chinookjack salmon were counted over Bonneville Dam,but 13,415 were counted over McNary Dam and8,402 into Spring Creek Hatchery, between Bonne­ville and McNary Dams (Junge and Phinney,1963). Thus, more than twice as many jack salmon

20

The quantity L Diqi represents the number ofi=1

mortalities occurring below the dam before count­ing and can be expressed as the difference betweenthe number of untagged recoverable carcasses andthe number of untagged recoverable carcassesdying after counting:

20 20

L Diqi = Y1 - Mara L Ciqi (8)i =1 i-I

(Note that at this point. we are assuming all un­tagged carcasses originating below the dam arerecoverable.) Finally, the total number of mortal­ities after counting can be expressed as follows:

20

Ma L Ciqi = fa(Y 1 + Do)1=1

= f.[Y1 + (1 - ra)faMC~ Ciqi

+ i~ Diqi.)] (9)

To derive an expression for M, (9) is solved forMa, and this result is substituted into (8). Then

of fish dying after counting are mutilated andrendered unrecoverable.

fa = fraction of all deaths occurring aftercounting.

ra = fraction of carcasses of fish dying aftercounting that are recoverable (i.e., notmutilated and rendered unrecover­able).

Do = number of carcasses of fish dying aftercounting that are unrecoverable.

Ma = proportion of those fish that arecounted over the dam that die nearthe dam.

The quantity Do can be expressed as

Do = (1 - ra)faM(?; Ciqi + i~ Diqi) (6)

Because Y 1 is the number of untagged recoverablecarcasses in the river, it is clear that Y1 is the dif­ference between the total number· of untaggedcarcasses and the number of unrecoverable un­tagged carcasses:

y 1 = M(~ Ciqi + i~ Dlqi) - Do

= [i - fa(l - ra)]M(~ Ciqi + i~ Diqi) (7)

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 479

(12)

were tabulated above Bonneville Dam as werecounted in that year at Bonneville.

In developing the model of chinook salmonmortality at Bonneville Dam expressed in equa­tion (10), we assumed that the carcasses of allfish dying below the dam were recoverable. Wealso assumed that all tagged carcasses introducedinto the river below the dam were recoverable.In the event that these assumptions are unjusti­fied, a somewhat more general mortality modelcan be used. This more general model provides forthe possibility that a certain percentage of car­casses released or dying below the dam becomeunrecoverable.

To derive this model, we define the followingquantities:

rb = fraction of carcasses of fish dying be­fore counting or released below thedam that are recoverable.

T = number of tagged carcasses introducedinto the river below the dam.

D'0 = number of carcasses of fish dying be­fore counting that are unrecoverable.

Note that Rx = rbT so that a more generalequation for estimating Y 1 than equation (3) is

y 1 = rS(~2 - 1)An equation for D'0 that is analogous to equation(6) for Do can be written

(

20 20)D'o = (1 - rb)(1 - fa)M i~ Ciqi + i~ Diqi

(13)

Three relationships analogous to those given inequations (7), (8), and (9), respectively, can thenbe written as follows:

(

20 20)Y1 = M L Ciqi + L Diqi - Do - D'o

i ~1 i =1

(

20 20)= (f,.ra - f,.rb + rb)M i~ Ciqi + i~ Diqi

(14)

20 20

fb L Diqi = Y1 - Mara L Ciqi (15)i =1 i =1

20

Ma L Ciqi = fa(Y I + Do + D'o)1=1

480

= fa[ Y1 + (1 - f,.ra + fafb - rb)

MC~ Ciqi + i~ Diq)] (16)

Equations (14), (15), and (16) may be used toderive a general expression for M in the same waythat equations (7), (8), and (9) were used to derivethe expression for M given in equation (10).The result, which allows for the possibility ofcarcasses of fish dying or being introduced intothe river below the dam becoming unrecoverable,is as follows:

M=( 20 ) "20rb I~ Ciqi + Y1 (1 - fa) + faraI~ Ciq"1

(17)

where the value of t 1 from (12) is used. Note thatequations (12) and (17) reduce to equations (3)and (10), respectively, when rb = 1.

In estimating that 16.78 percent of the chinooksalmon run was destroyed near Bonneville Dam,we assumed that all mortalities occurred belowthe dam (Le., that fa = 0). As has already beenpointed out, it is likely that some mortality oc­curs as a result of fish being swept back over thespillway after they have been counted. There­fore, it is of considerable interest to explore theeffect that this mortality of counted fish wouldhave on our estimate of M. Because some of thecarcasses of fish that die by being swept backover the spillway could be so severely mutilatedas to be rendered unrecoverable (Le., have nochance of floating and being recovered), thisfactor must also be considered.

To explore the various possibilities, fa (fractionof deaths occurring after counting) and ra (frac­tion of carcasses of fish dying after counting thatare recoverable) were assumed to take on variouspairs of values, and 1Qr was calculated for eachassumed pair of values by using equation (10).

For fa the following values were assumed: 0.0,0.125,0.250,0.375,0.500,0.625, 0.750, 0.875, and1.0. The same values were assumed for ra, and ~was calculated for all possible pairwise combina­tions of these values. From these results, we con­structed an isopleth diagram giving values of Mcorresponding to values of fa and ra (fig. 8).

Figure 8 shows that our estimate, :M = 0.1678(based on the assumption that fa = 0) is a mini-

U.S. FISH AND WILDLIFE SERVICE

1.0 ...---,.---.----_..._----~-_..._---~-......

ol&J....~ 0.875ool&JII::<t:c 0.750enll.

II::l&J

t;: 0.625<tII::;:)ooo.... 0.500<tJ:....enJ:

!;i 0.375l&Jo-I-I<t~ 0.250

zoi=o~ 0.125ILII

-"o ~~~-=::::==::===::;::::=:::;:==::::;::::::===d1.0 0.875 0.750 0.625 0.500 0.375 0.250 0.125 0

ro=FRACTION OF CARCASSES OF FISH DYING AFTER COUNTING THAT ARE RECOVERABLE

FIGURE S.-Values of M, the fraction of the chinook salmon run dying near Bonneville Damat the time of the 1955 experiments, corresponding to pairs of values for f. and r•.

mum point estimate of the level of chinook salmonmortality at Bonneville Dam. As fa increases, Malso increases; the rate at which 1\1: increasesdepends strongly on the value of ra' For largevalues of ra, corresponding to an assumption thatonly a small percentage of the carcasses of fishthat die by being swept over the spillway arerendered unrecoverable, values of Mincrease onlyslightly as fa increases. On the other hand, if alarge percentage of these carcasses are renderedunrecoverable, ra is small and 1\1: increases rapidlyas fa increases. (Note in figure 8 that points inthe region above and to the right of the linefor lQr = 1.00 correspond to values of fa and fa

that are incompatible with data collected in our1955 experiment.)

FACTORS ASSOCIATED WITH MORTALITYAND THEIR SIGNIFICANCE

We have established in preceding sections thatmany chinook salmon died near Bonneville Damat the time of our 1955 experiment. The nextlogical step is to examine factors associated withthese deaths to determine the actual cause orcauses.

In addition to the 1955 period. we also examinedavailable information for other periods between1943 and 1956 when high or low mortality wasapparent (table 7). The seven periods of appar­ently high mortality were September 1943, Sep­tember 1950. May 1952, July 1954, September1954, June-July 1955, and June-July 1956. Lowmortality periods were September 1946 andApril-May 1955.

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 481

TABLE 7.-Comparative mortality ofchinook salmon at Bonneville Dam in spring, summer, and fall periods between 1948 and 1956

Average Chinook ChinookAverage count of salmon salmon

dally Water chinook Chinook floaters floaters perspillway temperature salmon Search salmon per hour hour per Mortality

Season aild date flow range at dam time floaters of search 10,000 fish rating

C.m.•. • C. Number Him.. Number Number NumberSpring:

HighMay 1-31,1952________________ S,7OO 11.1-13.9 3,478 --------------- (ll ------0:iiii9-----------0:Oi6--.--April 20 to May 10, 1955_______ 1,050 7.8-11.1 6,481 212.6 2 LowSummer:

July 1-31, 19M. _______________ 7,900 14.4-17.8 1,051 26.5 30 1.132 10.771 HighJune 20 to July 22,1955•• _____ 9,020 13.9-17.2 1,836 603.4 194 0.322 1.751 Do.June 20-29, 1956_______________ 10,500 13.3-15.6 4,317 6.5 19 2.923 6.771 Do.

Fall: .Sept. 1-15, 1943________________ 620 18.3-18.9 12,543 8.0 146 18.250 14.550 Do.Sept. 1-30. 1946________________ 140 16.1-20.6 9,235 116.6 21 0.180 0.195 LowSept. 1-20, 1950________________ 480 18.9-20.6 10,OM --------_. ----- 54 ------------------------------ -- HighSept. 1-16, 1954________________ 1,160 17.8-18.3 5,001 125.5 99 0.789 1.577 Do.

I A commercial fisherman reported .. thousands" of dead chinook salmon below the dam.

The "high" and "low" mortality ratings aresubjective classifications based on the number ofchinook salmon floaters relative to the averagedaily chinook salmon counts at Bonn.eville Dam.Major differences in search techniques and in thelengths of search periods make det~iled compari­son between most of the periods of dubious value.Only for our study in 1955, and for a brief followupstudy in 1956, can we be certain that searchtechniques were comparable. We, therefore, gavethe greatest weight to data for these years inreaching our conclusions. Data for earlier yearswere useful, however, in lending support to theconclusions.

In spring 1955, when spillway flows and watertemperatures' were low and numbers of salmonwere high, only 0.009 chinook salmon floater wasfound per hour of search. Later the same year(midsummer), when spillway flows averaged 9,000c.m.s., water temperatures were higher, and num­bers .of salmon were low, 0.322 chinook salmonfloater was found per hour of search.

A pattern of characteristic circumstances ac­companying high mortality periods was evident:floating dead salmon were likely to be especiallynumerous below the dam when either high spill­way flows or exceptionally large numbers of salmonoccurred. If high flows and large numbers ofsalmon occurred simultaneously, even greaternumbers of floating dead salmon could be pre­dicted downstream from the dam. The data intable 7 generally substantiate this conclusion.

Mter completing our study in 1955, we wishedto test the hypothesis that large numbers ofchinook salmon would die and float near Bonne­ville Dam whenever large numbers of fish andhigh spillway flows coincided. Late June 1956 hadsuch a coincidence.

482

From June 20 to 29, spillway flow at the damaveraged 10,500 c.m.s. and coincided with chinooksalmon counts over the dam averaging over 4,300per day. We predicted that at the prevailingwater temperatures of 13.30 to 15.60 C., floatersfrom a given day's mortality would appear down­river about 7 days later and a large number offloaters would be evident below the dam by aboutJhly 3. On July 3, during 1 hour and 50 minutesof search at the Cape Horn station, five chinooksalmon carc~sses were observed; on July 5, 14carcasses were seen during 4 hours and 40 minutesof observation. In terms of numbers of chinooksalmon floaters recovered per hour these numberswer~ greater than for any previous observationperiod (except September 1943), indicating, aspredicted, that serious mortality had occurred.The cause of the high mortality of September1943 is unknown.

FLOW

We were unable to determine which specificconditions at Bonneville Dam contribute to salm­on deaths, but evidence is strong from our studyand from observations in earlier years that highmortality in spring and summer occurs duringhigh flows. The annual peak flow of the ColumbiaRiver at Bonneville Dam is in Mayor June,depending on the time of maximum snowmelt inthe upper watershed; the peak flow dates from1938 to 1955 fell between May 11 and June 28..Flows of more than 8,500 C.m.s. may occur forup to 3 months.

The maximum combined flow capacity of the10 power turbines at Bonneville Dam is about4,100 c.m.s. when the Columbia River is at floodstage. At low riverflows, maximum turbine capac­ity is about 3,400 C.m.s. This means that flows

U.S. FISH AND WILDLIFE SERVICE

ISPRING SUMMER FALLIn, -FLOW

~ ~--- 7,000 C.M.S._75-PERCENT

pERIOD---------

1147

~ ~ ----=-

I 1141

~rL- ---- ------.

1941

J ~---- ---~

1950

,J~4

7 ~ ------ ---""""'"'-

0 - -4 19~~

7 ------ -----~

01952

4 ~~7 ---- -_..:.-

~

0 -1953

~'4

7 ::;;rv ------- ----~

01954

14 ~

7 ---- ------- ---~ ----v

0;;...:..... -.;.1955

14 ~7 ---- ------- ----~ ..........--..

0 - -FIGURE 9.-Columbia River flows (solid lines) at

Bonneville Dam during periods (horizontal bars)when 75 percent of the spring, summer, and fallchinook salmon runs passed the dam, 1946-55.Dashed line at 7,000 c.m.s. indicates flow abovewhich serious mortality occurs.

10 20 30 10 20 30 10 20 30 10 20 30 10 20 30 10 2030API\, MAY JUNE JULY 1lIJG." SEPT.

7

o

14

o

14

o2

14

14

o

iiiCl

:1'.,::>§l....

above 4,100 C.m.s. during high water and above3,400 c.m.s. during low water are dischargedthrough the spillway (flows through the fishladders are an insignificant portion of the totalflow).

Columbia River chinook salmon migrationsmay be separated into ,spring, summer, and fallruns by their time of appearance at BonnevilleDam. For the purpose of this discussion, thespring ruil is defined as occurring in April andMay, the summer run in June and July, and thefall run in August and Septemper. Only a. fewchinook salmon migrate past Bonneville bambefore April or after September.

The three runs are characteristically eXposed todifferent river discharges at the dam. The sum­mer chinook salmon run, the smallest of the three,usually migrates upstream during the period ofpeak river discharge. Less frequently the springrun passes the dam during annual peak flows, al­though both spring and summer runs are alwayssubjected to relatively high flows. During thefall run, flows are relatively low and little wateris discharged through the spillway.

Figure 9 shows the daily flows at the dam in1946-55 for the periods when the central 75 per­cent of the spring, summer, and fall chinooksalmon runs were counted. The 75-percent figurewas arbitrarily selected to. include the major partof each run; 12.5 percent was subtracted from eachend of the rtm to determine ,the 75-percent period.The dashed line indicates the 7,OOO-c.m.s. flowlevel.

The ,?:,OOO-c.m.s. figure was arbitrarily selectedas the ppint above which heavy mortalities occur,based" primarily on 1955 information-the yearfor which the most extensive data are available(fig. 6). The reasons for establishing 7,000 c.m.s.as the critical level are: In 1955, 75 percent of thespring run ,passed the dam when total river flowwas low-between 4,200 and 5,100 c.m.s.; fewfloating chinook salmon were observed on an in­tensive search of the river during a large springchinook salmon run (172.000); the maximumdaily count past the dam was more than" 13,000fish. Therefore, flows under 5,100 c.m.s. were notassociated with a high mortality. Increasing num­bers of floating chinook salmon became apparentdownriver from the dam by June 12, 1955, aftera period when the chinook salmon count was

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 483

about 1,100 fish per day for several days and flowhad increased to between 7.000 and 9,900 c.m.s.

To understand better the relation of numbers ofchinook salmon counted at the dam and numbersfloating below tl~:e dam, it should be noted thatat a water temperature of 14.4° C., salmon car­casses float in about 7 days after death (appendixtable 1). Because water temperature averaged13.3° C. during the first 10 days of June 1955,8 or 9 days would elapse between death and thetime carcasses first floated. The increase in floatersobserved after mid-June was an indicator thatafter June 3, when total riverflow approximated7,000 c.m.s., the numbers of chinook salmon dyingsuddenly increased. We conclude on the basis ofthese observations that mortality may be ex­pected to be high whenever total riverflow risesabove 7,000 c.m.s., and low whenever total flowis less than 5,100 C.m.s.

In some years (1943, 1946, 1950, and 1954) float­ing chinook salmon were recovered below thedam when total flow was less than 5,100 c.m.s.Although data for these periods are limited, therelatively large numbers of floating chinook salm­on in September 1943 and 1950 probably repre­sented a low rate of mortality because very largenumbers of chinook salmon were migratingthrough the area. The search by A. Suomela in1943 (discussed in an earlier section), which re­sulted in the recovery of a large number of car­casses, is not comparable with later searchesbecause he searched in midstream, on shore, ineddies, and any place that fish could be expectedto accumulate, whereas in our later searches,floaters were observed only from fixed locations.Fall chinook salmon runs in these 4 years of ap­parently high mortality were larg~peak dailycounts at Bonneville Dam were 20,000 to 30,000chinook salmon. Water temperatures were about21.1° C., and riverflows were low. Under theseconditions, floating chinook salmon were moreapparent because they floated in a shorter time,thereby reducing their dispersion downstreamfrom the area of death both before and after theybecame buoyant.

The many floaters below the dam in September1954 may have represented a higher rate of lossof chinook salmon than is usual in the fall. Dailycounts were relatively low; the peak was onlyabout 8,000. During the 75-percent period ofpassage of the 1954 fall run, total riverflow was

484

the highest in the 10-year period 1946-55, rangingfrom about 4,700 to 5,500 c.m.s. This higher-than­normal flow may have resulted in a somewhathigher-than-average fall mortality rate.

Because a total flow of 7,000 c.m.s. or greateris associated with high mortality rates, figure 9may be interpreted as follows: (1) The flow overBonneville Dam has never reached 7,000 c.m.s.during the fall run of chinook salmon. Observa­tions have substantiated that fall runs generallyhave experienced relatively low mortality rates.(2) Most of the summer chinook salmon run inevery year has been subjected to flows exceeding7,000 c.m.s., which are associated with a highlevel of mortality. (3) In 4 of the 10 years (1946,1949, 1951, and 1952) the entire 75-percent periodsof spring runs were subjected to flows greater.than 7,000 c.m.s., and in 2 of the remaining 6years (1947 and 1948) flows exceeded 7,000 C.m.s.for about two-thirds of the 75-percent period.The remaining four spring runs (1950, 1953, 1954,and 1955) passed the dam when flows were lessthan 7,000 C.m.s.

In summary, fall runs of chinook salmon werenever subjected to killing flows; spring runs wereexposed to killing flows in some years; and sum­mer runs always encountered killing flows.

WATER TEMPERATURE

River water temperatures affect the floatingqualities of carcasses and thereby make carcassesmore or less evident to observers. Warm watermakes carcasses more apparent and cold watermakes them less apparent.

Flotation experiments described in the ap­pendix showed that carcasses require more timeto float in cold water than in warm. As a result,in the interval between death and floating, car­casses may be swept farther downstream duringperiods of low water temperatures than duringperiods of high temperatures. Furthermore, dur­ing the longer interval between death and float­ing in cold water, scavengers have more oppor­tunity to consume carcasses.

With this in mind, we examined the relationof water temperatures and numbers of chinooksalmon counted at the dam. In 1946-55 the rangeof temperatures during spring runs (April andMay) was 7.8° to 13.9° C.; during summer runs(June and July), 10.0° to 19.4° C.; and duringfall runs (August and September), 17.8° to 21.1°

U.S. FISH AND WILDLIFE SERVICE

C. (fig. 10). Thus, dead salmon are least likelyto be evident as floaters in the spring and mostlikely to be evident in the fall. Figure 10 showsthat in the spring of 1955 average water tempera­tures were the lowest for the 10 years shown(8.9° C.). Unusua:Ily cold water probably con­tributed to the almost complete absence of floatersduring the spring run in 1955. In 1952, a period ofapparently high mortality (table 7, fig. 10), watertemperatures during the spring chinook salmonmigration were highest of the 10-year period.

TURBIDITY

Turbid water, as measured by Secchi diskvisibility, was investigated as a possible factorcontributing to chinook salmon deaths. Wehypothesized that fish rely partially on sight tolocate and negotiate the fish ladders, and reduc­tion of visibility might handicap them.

High flows in the Columbia River are charac­terized by turbid water, and low flows by rela­tively clear water. During low flows the water maybe turbid for short periods when a tributary floodsfl'om heavy rains or rapid snowmelt. For example,in mid-January 1953 after a flash flood on atributary, a Secchi disk visibility reading of 0.06m. was recorded at Bonneville Dam when flowwas only 2,700 c.m.s.

Secchi disk visibility at the dam has seldombeen less than 0.3 m. during periods of majorsalmon migrations (fig. 11). From 1950 to 1955,visibility was 0.3 m. or less for only short, infre­quent periods from April to September, except in1952, when it remained about 0.3 m. or less fromApril 1 through May 30, throughout the springmigration. The 1952 spring run suffered a heavymortality. but because the high turbidity wasaccompanied by high flows and relatively highwater temperatures, it was impossible to evaluatethe separate effects of flow, temperature, andturbidity.

In 1955, from April 10 to June 30, Secchi diskvisibility varied between 0.3 and 0.8 m.; it was0.6 m. at the peak of the spring chinook salmonrun (May 2) and 0.45 m. at the peak of the sum­mer run (June 23). Few floating chinook salmonwere observed in May, but large numbers wereseen in June and July. Because turbidity differedlittle between the two periods, we concluded thatit was not a major factor in mortality at the damin 1955.

Because high turbidity and high flow usuallycoincide. we could not evaluate the separate effectof turbidity, if any.

COMMERCIAL FISHING

The Columbia River gill net fishery has some­times been blamed for dead salmon in the riverbecause some fish escape from nets after becomingenmeshed. An escaped fish usually has charac­teristic net marks-encircling bands where scaleshave been scraped off and cuts on the anterioredges of fins. Hanson et al. (1950: 24) concludedfrom observations in the fish ladders at Bonne­ville Dam and at hatcheries above and below thedam that" Most injuries to the fish observed aretraceables (sic) to fishermen's gill nets; none ofthe injuries were directly traceable to conditionsat Bonneville Dam. Most net injuries were notfatal to fall-run chinook salmon at hatcheriesabove and below Bonneville Dam in 1946." Theconclusion is questionable because only the fishthat survived after being injured were availablefor observation-those that may have died couldnot, of course, be sampled at the hatclleries.

Most significant in discounting gill~ nets as amajor cause of mortality in our study· is the factthat 85,769 spring chinook salmon passed thedam in May 1955 with little apparent mortality,despite an intensive commercial gill net fisheryfrom April 30 to May 27. The spring chinooksalmon run during this period was the largestsince 1939 (the first year runs were counted atBonneville Dam). The catch was 80 percent of thetotal run; the catch below the dam was the secondlargest since 1939. Four stations below the damand one station near The Dalles were searchedintensively for floaters throughout this period,but few were found and none of these bore charac­teristic net marks. This is strong evidence againstattributing the death of floating chinook salmonto the gill net fishery.

We concluded that gill net injuries are not amajor cause of death of chinook salmon foundfloating near the dam.

FISH PASSAGE FACILITIES AT BONNEVILLE DAM

Great effort has been made by fishery agenciesand especially by the U.S. Army Corps of Engi­neers to discover any structure or operation atthe dam that might delay, injure, or kill migrat­ing adult fish. Mechanical failures or routine

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 485

~

SPRING~

FALLI- 1946 -- ~ -

1947 ~~

I- -I-~ -

1948 --A-I-

~-- TEMPERATUR~ -

I-~_75-PERCENT-- PERIOD --

1949 ~

I-

~-

I- ~ -1950

~

I-

~-

JI- -1951 '------

I-

~ -~I- -

1952

~'--'-,

I- -~

I- -1953~

'--- -

-~. -

1954

~---"V"-\.

- --~ -

1955~

r- ~ -I- ~ -

J. J. .l J . J. I. J. J. . 1. J . . 1 I. .1 . 1 J .20 30 10 20 30 10 20 30 10 20 30 10 20 30 10 20

'APR. MAY ~UNE JULY A\,JG. SEPT.

15.6

10.0

15.6

10.0

10.0

15.6

15.6

10.0

21.1

15.6

10.0

15.6

15.6

10.0

10.0

lIJa::::J 15.6~a: 10.0lIJQ,

2lIJI- 15.6a:~IO.O

~15.6

-U 100g •--

FIGURE 10.-Columbia River water temperatures at Bonneville Dam during periods when 75 percentof the spring, summer, and fall chinook salmon runs passed the dam, 1946-55.

486 U.S. FISH AND WILDLIFE SERVICE

1950

1951

1952

2

5 15 25 5APR.

5 15 25 5 15 25 5 15 25JULY AUG. SEPT.

FIGURE 11.-Secchi disk visibility in the Columbia River at Bonneville Dam, 1950-55.

maintenance operations have sometimes neces·sitated shutting down portions of the fish-passagefacilities, but these have rarely affected passagefor long.

Many improvements have been made in thefish passage facilities at Bonneville Dam over theyears. Among them are alteration of the power­house collection system, addition of auxiliaryattraction flows at the entrance to the ladders,installation of flow baffles below the ladder en­trances, and installation of ban"ier screens belowthe spillway bays adjacent to the ladder en­trances.

In 1946 a submarine viewing chamber made ofa section of large steel pipe with the ends sealedoff and with two watertight windows, was usedby the U.S. Army Corps of Engineers and FWS.A series of observations of migrating fish weremade by an observer inside the chamber, whichwas lowered into a fish ladder. Fish within theladder experienced no difficulty in moving throughthe ladder (Hanson et aI.. 1950). Other observa­tions through the years have shown that oncefish have entered the ladders they generally passthrough with little or no injury or delay.

Because there is no evidence that fish are in­jured in the ladders, we eliminated this possiblesource of mortality from further consideration.

DISEASE

Another possible cause of salmon mortality isdisease. In 1954 and 1955, to determine the causesof death we collected all of the recently killedfish we could find below the dam.

None of the fish collected in 1954 were suffi­ciently fresh to warrant a detailed examination,although gross injuries of external origin wereobvious in some instances (table 8).

In 1955. we recovered three chinook and sixsockeye salmon, four white sturgeon, one shad,and two carp in fresh condition (table 8). Seven ofthe fish (two chinook and five sockeye salmon)were immediately frozen and later autopsied byEdward M. Wood, fish pathologist at the FWSFish Nutrition Laboratory, Carson, Wash. Insummarizing the results of his examinations,Wood said:

"In some cases these fish had severe in­juries which clearly caused death. In theseinstances we have attempted to determine ifthere were any contributing factors whichmight have made the fish more susceptibleto injury such as disease. At other times, how­ever, there was no gross evidence of injuryother than slight internal hemorrhage, con­gest~on of various organs with blood, or

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 487

I Autopsy performed.

TABLE 8.-Qbservations at time of rerovery of S moribundand recently dead fish found within 5 miles of Bonnel.illeDam, 1954-55

slight edema over the brain. In these caseswe do not know the cause of death and littleof the nature of the injury. All of these fishwere sexually immature and death was pre­sumably not." related to the reproductivecycle."

1954:September 9 Chinook salmon .Fresh; Jaw reflexes; right eye

turned backward into socket;gills ble",ling; slight hloUllybruises on pectoral fins, isth mus,right operc.le, and caudal pedun­cle.

September 11. White sturgeon Fresh; fin reflexes; abrasions on. pectoral girdle and side.

1956:May 3 do __ . Alive; deep gash near left ventral

fin.May 12. __ •. do Alive; peduncle severed.00 . Carp . Alive, deep dorsal gnsh.Do do. . Fresh; decapitated.May IlL White sturgeon Severed at midbody.June 7 Chinook salmon Fresh; torn premaxillary, isthmus,

and first gill arch; clotted bloodin heart cavity.

June 10 Chinook salmon , Moribund; abrasions on head anddorsal fin; blood clot betweenfirst and second lelt gill nrches.

June 17 American shad Fresh; head severed; caudal pe-duncle severed.

June 21. White sturgeon Fresh; body severed.July 9 Sockeye salmon Fresh; abrnsion on lelt opercle.July 12 Sockeye salmon ' Fungused gashes on right side.00 do . Fresh; no apparent injltries.00 do 00.July 14 do Fresh; massive wound into body

cavity.July 16 . do . Alive; floating.July 21. Chinook salmon 1 •• Large gash on peduncle.

SUMMARY

1. In most years since the completion of Bonne­ville Dam in 1938, floating dead fish-particularlychinook salmon-have been observed downstreamfrom the dam.

2. In 1954 and 1955 the Oregon Fish Commis­sion. under a contract with FWS, studied salmonmortality at Bonneville Dam.

3. The main aims of the studies were to esti­mate the chinook salmon mortality at BonnevilleDam and to determine the causes of death.

4. From ratios of tagged to untagged floatingcarcasses. we estimated that 4,412 recoverablecarcasses of chinook salmon that had died nearBonneville Dam were in the river at the time ofour 1955 experiment.

5. The chinook salmon mortality at Bonneville

<Erling J. Ordal. 1955. Progress Report No. 13, U.S. Fish and WildlifeService, Contract No. 14-19-008-2418,7 pp. On file Bureau of CommerdalFisheries Biological Laboratory, Seattle, Wash. 98102.

delay of chinook salmon below Bonneville Dam(Schoning and Johnson. 1956) could have beenthe immediate cause of death of at least a portionof the chinook salmon included in our estimate ofmortality.

An epidemic of the bacterial disease columnaris,Chondrococcus colmnna.ris. among salmon (par­ticularly sockeye) was reported in the ColumbiaRiver system in late July and August 1955. 6

Sockeye salmon in the latter part of the run seemedmore heavily infected than those in earlier seg­ments. probably because water temperature in­creased during the summer. We do not believethat columnaris was an important contributingfactor to death of chinook salmon in June andearly July 1955 because our study was completedbefore the outbreak of the disease. Becausecolumnaris epidemics are associated with rela­tively warm water, mortalities reported duringspring cold-temperature periods in previous yearswere probably not caused by this disease.

In summary, although we were able to recoveronly a small number of freshly killed fish im­mediately below Bonneville Dam. there was noindication that disease contributed to death in1954 and 1955. Many of the fish had severe recentexternal injuries. Some of the dead fish withoutmassive injuries probably died as a result of nitro­gen poisoning, a potential cause of death notrecognized by us at the time of OIU' study.

Condition at time 01 recoverySpeciesDate found

Since the conclusion of our study of mortality,other investigators have discovered that super­saturation of nitrogen in Columbia River waterat times of high flow may be a significant cause offish mortalities. Some of the physiological symp­toms described by Wood when he autopsied thefresh dead salmon are characteristic of nitrogenpoisoning (such as internal hemol1'haging and con­gestion of spleen), which could well have been adirect cause of mortality that we did not recognize.

Ebel (1969) found that nitrogen saturationlevels potentially dangerous to fish always oc­curred at Bonneville Dam when water was dis­charged through the spillway. Nitrogen super­saturation may occur when atmospheric nitrogenis entrained and dissolved in the plunging turbu­lent spillway flow. Therefore, highest nitrogensaturation values coincide with peak spillwayflows. Nitrogen levels in tail waters below Bonne­ville Dam were substantially higher than levelsbelow seven other Columbia River dams. Highnitrogen levels, coinciding with high flows and a

488 U.S. FISH AND WILDLIFE SERVICE

Dam was estimated to be 16.8 percent of thetotal run during the period of our 1955 experi­ment (June 21 through July 10).

6. If a substantial proportion of the fish diedafter they were counted (by being swept back overthe spillway) and if a substantial number of thesecarcasses were mutilated and rendered unrecover­able, actual mortality was substantially higherthan our estimate.

7. Our experiment probably included the periodof maximum mortality in 1955.

8. The numbers of floating carcasses are relatedto the amount of flow of the Columbia River.The mortality of salmon is highest when flowsexceed 7,000 c.m.s.

9. Fall chinook salmon runs have never beensubjected to flows above 7,000 C.m.s. (killingflows); spring runs were e~posed to such flows insome years; and summer runs nearly always en­countered such flows.

10. Water temperature, turbidity, disease. andinjuries from gill nets had little influence on thenumber of carcasses in the river.

11. The specific causes of death and the pre­cise areas at Bonneville Dam where death oc­curred were undetermined, but the major sourceof chinook salmon mortality is associated withthe spillway during high flows.

12. We did not consider nitrogen poisoningresulting from high flow turbulence as a possiblecause of mortality at the time of our experiment,but evidence subsequently developed by otherinvestigators indicates that this may be a majorspecific mortality factor.

ACKNOWLEDGMENTS

Fred C. Cleaver, formerly Assistant Directorof the Oregon Fish Commission, contributedgreatly to the planning and analysis. Ivan J.Donaldson, Resident Biologist, V.S. Army Corpsof Engineers, Bonneville Dam, helped releasetagged experimental fish and provided informationon mortalities of previous years.

John A. Dudman, Associate Professor of Mathe­matics at Reed College, and Richard F. Link,Assistant Professor of Statistics and Mathematicsat Oregon State College, provided technical as­sistance in the early stages of the statisticalanalyses.

Lee Motley, operator of Beacon Rock Moorage,supplied information on the use of salmon car-

casses by fishermen for sturgeon bait. Finally,Edward M. Wood, former pathologist of the V.S.Fish and Wildlife Service, performed the histo­logical examinations of freshly killed salmon todetermine causes of death.

LITERATURE CITED

EBEL, WESLEY J.1969. Supersaturation of nitrogen in the Columbia

River and its effect on salmon and steelhead trout.U.S. Fish Wild\. Serv., Fish. Bull. 68: 1-11.

HANSON, HARRY W., PAUL D. ZIMMER, and IVAN J.DONALDSON.

1950. Injured and dead fish in the vicinity of Bonne­ville Dam. [U.S.] Fish Wild\. Serv., Spec. Sci. Rep.Fish. 29, 41 pp.

JACKSON, R. I.1950. Variations in flow patterns at Hell's Gate and

their relationships to the migration of sockeye salmon.Int. Pac. Salmon Fish. Comm., Bull. 3: 81-129.

JUNGE, CHARLES 0., JR., and LLOYD A. PHINNEY.1963. Factors influencing the return of fall chinook

salmon (Oncorhynchus tshawytscha) to Spring CreekHatchery. U.S. Fish Wild\. Serv., Spec. Sci. Rep.Fish. 445, iv + 32 pp.

PAULIK, G. J., and D. S. ROBSON.1969. Statistical calculations for change-in-ratio esti­

mators of population parameters. J. Wild\. Manage.33: 1-27.

PEARSON, E. S., and H. O. HARTLEY (EDITORS).1966. Biometrika tables for statisticians. Vo\. 1.

3d ed. Cambridge Univ. Press, New York, 263 pp.SCHONING, ROBERT W., and DONALD R. JOHNSON.

1956. A measured delay in the migration of adultchinook salmon at Bonneville Dam on the ColumbiaRiver. Oreg. Fish Comm., Contrib. 23, 16 pp.

TALBOT, G. B.1950. A biological study of the effectiveness of the

Hell's Gate fishways. Int. Pac. Salmon Fish.Comm., Bull. 3: 1-80.

THOMPSON, WILLIAM F.1945. Effect of the obstruction at Hell's Gate on the

sockeye salmon of the Fraser River. Int. Pac.Salmon Fish. Comm., Bull. I, 175 pp.

U.S. ARMY, CORPS OF ENGINEERS.1943-56. Annual fish passage report, North Pacific

Division, Bonneville, The Dalles, and McNaryDams, Columbia River, Oregon and Washington,1943-56. Prepared by the U.S. Army EngineerDistricts, Portland and Walla Walla, Corps of En­gineers, Portland, Oreg. [Each year publishedseparately:]

APPENDIX

EXP~RIMENTS ON FLOATING QUALITIES OFCHINOOK SALMON CARCASSES

This appendix describes experiments designedto test two key assumptions: (1) that experi-

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 489

1 These obse.rvations omitted from calculations of meaus and statistical~omparisons of floating times.

ApPENDIX TABLE I.-Days required for fresh and frozenchinook salmon to floa.t in cold water (7.2°-11.1 ° C.)

. ApPENDIX TABLE 2.-Days required for fresh and frozenchinook salmon to float in wann water (1:i1.8°-17.fe° C.)

Average AverageDays days Days days

Water required required Water required requiredtemperature to float to float temperature to float to float

• a. Nlt1llber Number • O. Number Number13.3-13.9_____ 7 1~.S_13.3_____ 1113.3-13.9_____ 6 1~.S_13.3_____ 1113.3-14.4__• __ '<6 12.S_13.3____ • 813.3-14.4 ___ ._ 1 <6 13.3-14.4____ • '<615.6-16.1. ____ 3 4.6 13.3-14.4_____ 715.6-16.1. ____ 4 15.6-16.7_____ 4 6.315.6-16.7_.___ 5 15.6-16.7____ • 516.7-17.~. __ ._ 4 15.6-16.7_____ 516.7-17.2. _. __ 4 15.6-16.7___ •• 516.7-17.2•. _., 4 16. 7-17.~_____ 5-._-------------- _. -. -. ---- 16.7-17.2_____ 4-. --_. ---------------- ----- 16. 7-17.~_____ 4)

Number

Averagedays

requiredto float

14.1

Frozen chinook salmon

DaysWater required

temperature to float

Number • a. Number7.2-10.6_____ 15

11.8 7.~-1O.6_____ 77.~-10.6_____ 197.2-11.1•• ~4

8.3-10.0•• __ • 168.3-10.0.____ 148.3-10.0.____ 15

10.0-11.1..___ ~4

10.6-11.1..___ 1010.6-11.1. __ ._ 710.6-11.1...__ 910.6-11.1..___ 9

Averaged'lYs

requiredto float

Nur:}bfr1~

1212

Fresh chinook salmon

DaysWater required

temperature to float

made concerning the distributions of floatingtimes in the populations from which our samplesare drawn. When fresh and frozen carcasses werecompared under cold-water conditions (appendixtable 1), the null hypothesis that there is no dif­ference in floating times between the two typesof carcasses was accepted at the 90-percent level ofsignificance. A similar result was obtained forwarm-water conditions (appendix table 2). Ap­pendix figure 1 shows the similarity of floatingcharacteristics of fresh and frozen carcasses,especially at relatively warm water temperaturesof over 10° C. The temperature of the ColumbiaRiver was 14.4° C. during the experiment to esti­mate the population of dead salmon.

Next we compared floating times of carcassesin warm and cold water. again using the Wilcoxonr.ank sum test. Because no differences had beenfound between fresh and frozen carcasses, bothtypes of carcasses were included in the comparisonbetween water temperatures. We rejected (at the99-percent significance level) the null hypothesis

• a.7.~-1O.0 _7. ~-IO.O _8.3-10.6. "'__8.3-10.6. "'__

mental tagged carcasses and carcasses of naturalriver-killed fish have similar floating charac­teristics; (2) that floating carcasses pass rapidlythrough the recovery area and are available forrecovery on only i day.

Floating Qualities of Fresh and Experimentai Carcasses

Salmon sink after death, but decompositiongases cause them to float after a period of time.In the mark-and-recovery method that we usedto estimate the population of dead fish, it wasessential to ascertain whether tagged experimentalcarcasses had floating qualities similar to thoseof other salmon that die in the Columbia River.Differences between the two kinds of carcassescould influence results. We made a series of ex­periments at various water temperatures withfresh and frozen eXperimental chinook salmon car­casses to determine whether there are differencesin elapsed time between death and rise of thecarcass to the surface.

Some information was already available fromexperiments by Hanson et al. (1950). They de­termined the elapsed time between death andfloating for 21 fresh salmonids at water tempera­tures of 13.9° to 20.6° C. We performed similarexperiments over a wider range of temperatureswith 24 frozen and 14 fresh chinook salmon. Cold­water experiments were done in a spring-fed pondat the Oregon Fish Commission Laboratory,Clackamas, Oreg.; warm-water experiments weredone in the Columbia River near Bonneville Darn.Frozen fish were thawed in air for 24 hours beforesubmersion for testing; they were from the samelots as those that were later tagged and releasedto estimate the population of dead salmon. Freshfish were purchased from commercial fishermenarid were submerged within a few hours afterdeath. The time required to float was calculatedfrom the time the fish was placed in the wateruntil some part was visible at the water surface.

In the cold-water temperature range (7.2°­11.1° C.) frozen salmon took 2.3 days longer tofloat than fresh salmon (appendix table 1),whereas in the warm range (12.8°-17.2° C.) thedifference was reduced to 1.7 days (appendixtable 2). We tested the significance of these dif­ferences by using Wilcoxon rank sum tests forcomparison of group means. This nohparametrictest does not require that any assumption be

490 U.S. FISH AND WILDLIFE SERVICE

26 r---------------:--.,....-------,

Disappearance of Floaters from Recovery Area

We conducted another experiment to estimatethe rapidity of disappearance of floaters from thesurface of the Columbia River within the re­covery area and to trace their dispersion from thepoint of first appearance. From April 4 to July 22,1955, all floaters in reasonably good conditionwere tagged with Petersen disks and releasedwhere they were found (appendix table 3). Of289 chinook salmon floaters tagged. 11 percentwere recovered; 26 were recovered again afterthe first release. and five after a second release.The single recovery from the 17 chinook salmonreleased above the dam at station 8 (The Dalles)was by a fisherman about 4.8 km. downstreamfrom the release point. Nineteen percent of the 11steelhead floaters were recovered. The greatestdistance traveled by any tagged floater beforerecovery below the dam was 74 km.

All the tagged experimental carcasses used forthe population estimates were intact. Some of thenatural population of killed fish had massivewounds. and part of these could not float and berecovered as floaters. Therefore, in the main bodyof this report we have referred to such fish asbeing unrecoverable, and in our basic model weallow for the possibility that a fraction of thefish that die are rendered unrecoverable.

Pertinent results of the flotation experiment aresummarized as follows:

1. All intact dead chinook salmon, fresh orfrozen, floated.

2. The relation between water temperature andtime required to float was inverse, i.e., the colderthe water the greater the time required for thecarcass to float.

3. No statistically significant or practical differ­ences were found between floating times of freshand frozen fish over the range of temperaturestested.

4. A slit or puncture of the abdominal wallusually did not greatly affect the floating qualitiesof the fish. .

5. Fresh and frozen chinook salmon carcasseshave similar floating properties; therefore, frozencarcasses can be used to simulate fresh carcasses.

18.316.1

o FRESH SALMON

~FROZEN SALMON

9.4

aDo

11.7 13.9TEMPERATURE (OC.1

ApPENDIX FIGURE I.-Days required for fresh and frozenchinook salmon to float at water temperatures of 7.2° to17.2° C.

that floating time is unaffected by water tempera­ture.

To determine whethel' massive tissue damagesuch as ruptured body cavities might affect thetime required to float. we made another experi­ment with 15 frozen carcasses. Slits of 2.5 to 7.5cm. were made in the body wall of eight carcasses,and full-length slits (from vent to isthmus) weremade in seven.

Three fish. two with full length slits and onewith a 7.5-cm. slit. took only 4 or 5 days to float attemperatures of 15.60 to 18.30 C. Of the remaining12 fish, tested at water temperatures of 9.40 to11.10 C., all but two floated-one of these had a2.5-cm. slit and the other had a full-length slit.Except for the two that did not float, the timerequired for slit fish to float was about the sameas for intact fish: in the warm-water range. slitfish required an average of 4.3 days to float com­pared with 4.6 days for intact fish; in the cold­water range, slit fish required an average of 12.8days to float compared with 11.8 days for intactfish.

~ 16

9LL 14ot-

fr112II::5~ 10II:

~ 8t-

20

2

22

24

6

4

iii~ 18c

CHINOOK SALMON MORTALITY IN COLUMBIA RIVER NEAR BONNEVILLE DAM 491

ApPENDIX TABLE 3.-Number of floaters tagged at searchstations near Bonnel'ille Dam and percentage recovered,April 4 to 2i, 1955

Chinook s"llIIon SOCkl~ye- salmon St~~lIw"d tront_._----- ------- -----------Tagging Re- R~- Re-location Taggt'd ~ov~red T"gged covcr\\(l Tugged covered

Number Number Number Number Number N"mberThe Dalles. ____ I. 1 4 0 U 0Bonncville •____ 32 (1 5 (1 2 0Cape Horn. ____ 75 G 21 0 0 0Reed Island __ ._ 92 1a a 1 3 0Ellswonh_. ____ I. 4 IS 1 ., 0Willamettc. ____ 63 4 13 1 4 1St. Helens_.____ 3 a 0 0 0 1

---------------_._-------

Tags from eight of the 272 chinook and threeof the 57 sockeye salmon released downstreamfrom the dam were returned by fishermen; therest were recovered at search stations.

Forty-four chinook and 12 sockeye salmon werereleased at stations 1 (St. Helens) and 2 (Willam­ette) when there was no search effort fartherdownriver. These fish. therefore. had no chancefor recovery at search stations. In both years,carcasses released at stations farther downstreamhad less chance of recovery than those releasedfarther upriver because the number of searchstations was smaller downstream from the dam.

Only five of the 22 chinook salmon floatersrecovered at search stations in 1955 traveledmore than 16 km.. and all but one were found the

I McGowan and Moffett Cret'k stations combined.• Including 26 released twice and fiv~ released a third time.I Including five second rec.ovcries.

same day they were released. Thus. the smallnumber of second recoveries of tagged floaterswas indicative of their rapid disappearance.

Tagging and recovery of floaters also provideduseful information on their rate of travel: duringour experiment in July 1955. when river flow washigh, the rate below Bonneville Dam was from3.4 to 5.1 km.p.h. (average 4.2 km.p.h.).

Thus, during the period of recovery of taggedfloaters for the mortality estimate. a floater re­quired a maximum of only 16 hours to passthrough the entire search area between the damand the mouth of the Willamette River, a distanceof 70 km.

From these facts, the following conclusionsmay be drawn regarding disappearance and rateof travel of floaters:

1. Floaters appearing at the surface nearBonneville Dam would pass through the 70-km.search area between the dam and the mouth ofthe Willamette River in less than 16 hours, pro­vided they remained in the current.

2. Floaters would pass through the entire searcharea during the night between successive days ofsampling, eliminating any possibility of beingobserved.

3. Floaters originating near the dam wouldreach the mouth of the Columbia River in 3 or 4days.

19

11361TOI.aL.______ • 2119 • alPercent

recovered_____________ 11

492 U.S. FISH AND WILDLIFE SERVICE