Sampling Strategy for Enumeratingthe Western Arctic Populationof the Bowhead Whale

problems in mind that this paper pre­sents the NMML research strategy forestimating total abundance of thebowhead whale and discusses methods

20·

40·

80· N

60·

70·

30·

120·

NORTH AMERICA

ARCTIC OCEAN

150· W

which eventually may be used tomonitor relative abundance, i.e., an­nual changes in population size.

Study Area

The study area closely approximatesthe range of the western Arctic popula­tion of bowhead whales (Fig. I). Therange extends from the west-central Be­ring Sea north of approximately lat.60 o N, throughout the Chukchi andeastern East Siberian Seas, and east­ward throughout the U.S. Beaufort Seato Banks Island and Amundsen Gulf,Northwest Territories, Canada.

Seasonal movements of bowhead

180·

-"':OR rH PA('Ir fe OCl AN

SIBERIA

150· E

Introduction

The National Marine MammalLaboratory (NMML), National MarineFisheries Service, NOAA, is conduct­ing research on the population enumer­ation of the western Arctic stock of thebowhead whale, Balaena mysticetus.From a management viewpoint this re­search is critical for evaluating thestatus of this severely depleted stock.Because so little is known about thebowhead, inferences regarding thehealth of the population arise mostlyfrom changes in total population size.The basis for this inference is the as­sumption that changes in populationsize reflect the summation of all lifehistory processes of the bowhead. Tostate it rather coarsely, for this depletedstock a decline in total abundance mayindicate a shift toward extinction; anincrease, a shift toward survival.

At present our only scientific meansof assessing the number of whaleswhich can be safely removed from thebowhead stock is to estimate totalabundance and then rely on theories ofpopulation growth to predict acceptablelevels of removal. But estimating totalabundance is a very costly endeavor fora wide ranging species like thebowhead and, furthermore, there is noguarantee that general theories of popu­lation growth will always apply. Re­gardless of the population level there isthe possibility of decline induced,perhaps, from stress relating to ac­tivities of offshore oil development,subsistence harvest, or natural fluctua­tions of the ecosystem. It is with these

BRUCE D. KROGMAN

10·

Bruce D. Krogman is with the National MarineMammal Laboratory, Northwest and AlaskaFisheries Center, National Marine Fisheries Ser­vice, NOAA, 7600 Sand Point Way N.E., Bldg.32, Seattle, WA 98115.

Figure I.-Bowhead whale study area. Hatched lines indicate distribution of thewestern Arctic and Okhotsk populations of bowhead whales prior to commercialexploitation (Braham et aI., footnote 4).

30 Marine Fisheries Review

whales closely parallel the seasonaldistribution of ice. The northwardspring migration from the Bering Sea istimed with the breakup of the pack ice(Bailey and Hendee, 1926) which oc­curs about April. Upon entering theChukchi Sea the migration cuts north­eastward toward Point Hope and alongthe northwest coast of Alaska, towardPoint Barrow (Braham et aI., 1980).The migration occurs closest to shore atPoint Barrow, with most of the popula­tion passing from late April throughMay. Although the direction of migra­tion toward Banks Island does notchange, east of Point Barrow the coast­line slants southeasterly toward Can­ada. Thus the spring migration routeeast of Point Barrow becomes progres­sively more offshore as the whalesmove easterly through the Beaufort Seatoward Banks Island.

During the spring migrationbowheads confine themselves to openwater areas in the pack ice. In theirsouthern range the ice is thin and easilydisturbed by winds and currents; openwater areas are abundant. As bowheadspenetrate further north, however, theice becomes thicker and flaws in the icecommonly appear as long cracks, orleads. Along the north west coast ofAlaska these leads persist between thelandfast ice and the pack ice forming azone which may exceed 50 km in widthtowards the southern end near CapeLisburne (Burns et al. l

). This zone nar­rows dramatically to only a fewkilometers at Point Barrow (Braham etaI., 1979). East of Point Barrow thezone occurs further offshore (Marko,1975).

From June to September bowheadsfrequent areas south and southwest ofBanks Island; the autumn migrationwestward through the Beaufort Sea be­gins in August and September, withmost sightings recorded in October nearthe 20 m depth contour line from De­marcation Bay to Point Barrow

'Burns, J. J., L. H. Shapiro, and F. H. Fay. 1977.The relationships of marine mammal distribu­tions, densities, and activities to sea ice condi­tions. In Environmental assessment of the Alaskacontinental shelf, annual reports of principal in­vestigators for the year ending March 1977, Vol.I, Receptors-mammals, p. 503-554. U.S. Dep.Cammer., Natl. Oceanic Atmos. Admin., Envi­ron. Res. Lab., Boulder. Colo.

Sepl.-Ocl. 1980

(Ljungblad et al. 2). From Point Barrowthe animals move westward towardWrangel Island (Cook, 1926; Town­send, 1935) then south through theChukchi Sea into their winter range, theBering Sea.

Field Methodology

Ice and Land Camps

Each spring from 15 April through 30May since 1976 we have countedbowhead whales as they migrated pastPoint Barrow in the nearshore lead. Be­cause of constant daylight during latespring and summer in the Arctic, a 24­hour observation schedule was main­tained. One camp was deployed in 1976and 1977. In 1978 and 1979 two campswith seven persons each were deployedon the landfast ice next to the nearshorelead approximately 5 km north of PointBarrow.

The two camps, called South Campand North Camp, were located 600­800 m apart depending on the avail­ability of ice platforms or ridges ofsufficient height to provide visual ob­servation across the lead.

Bowheads moved northeastward inthe nearshore lead from South Camptoward North Camp. South Camp ob­servers maintained the primary countwhile North Camp observers estimatedhow many whales were missed at SouthCamp.

Documentation of the ice campcounting methodology is reported inBraham et al. (1979), Braham, Krog­man, Johnson, Marquette, Rugh,Sonntag, Bray, Brueggeman, Dahl­heim, Nerini, and Savage (1980), andKrogman et a1. 3 .

A land camp was located at CapeLisburne in 1978 to study the onset andtermination of spring migration along

2 Ljungblad, D. K ., M. F. Platter- Rieger, and F. S.Shipp, Jr. 1980. Aerial surveys of bowheadwhales, North Slope, Alaska. Naval Ocean Sys­temsCenterTech. Doc. 314, Final rep., Fall 1979,BLM Project No. 00L80AA851-IAO-I­ELEMENT OGB, 182 p. Bureau of Land Man­agement, Code 733,18 and C Streets N.W., Rm.2657, Washington, DC 20240.3B. D. Krogman, G. W. Priebe, and R. M.Sonntag. 1980. Arctic Whale Task ice camp sur­vey data management format, document 1980version. Unpubl. rep., 30 p. Natl. Mar. MammalLab., Natl. Mar. Fish. Serv., NOAA, 7600 SandPoint Way N.E., Bldg. 32. Seattle, WA 98115.

the northwestern coast of Alaska. Landcamp methodology was similar to thatused in the ice camps except that onlyone camp was used (Rugh and Cub­bage,1980).

Aerial Survey

Aerial survey procedures were de­signed to maximize our ability to de­lineate nearshore and offshore distribu­tion of whales in seas covered with packice. The aircraft was flown over leads atelevations ranging from 70 to 300 mdepending upon cloud cover. Docu­mentation of aerial survey methodol­ogy used for this research is reported inBraham et al ..j and Krogman et al. 5 .

Total Abundance

Results from aerial surveys flown todelineate the spring distribution ofbowhead whales in the Bering, Chuk­chi, and Beaufort Seas, and resul ts fromfeasibility studies conducted at St.Lawrence Island, Cape Prince ofWales, Point Hope, Cape Lisburne, andPoint Barrow, indicate that the most re­liable estimate of total abundance willresult from counts made from the ice asmigrating bowhead whales pass PointBarrow, Alaska, during their springmigration. Census methods which re­lied on aerial survey methodoiogy werequickly disqualified because of the con­founding effects of sea ice on samplingdesign. The other above-mentionedsites were considered for staging land/ice based counts, but each eventuallyproved inferior based on the criteria ofproximity of location to migratoryroutes, ice, and prevai ling weatherconditions which influence visibil ity.

Counts made at Point Barrow do notconstitute a complete census, however.In fact, before one can use the Barrow

'Braham, H., B. Krogman, and G. Carroll. J979.Popu lation biology of the bowhead whale(Ba/aena mysticelus) II: Migration, distribution,and abundance in the Bering, Chukchi, andBeaufort Seas, with notes on the distribution ofwhite whales (Delphinapterus leI/cas). Unpubl.final rep., OCSEAP contract R7120807, 118 p ..Natl. Mar. Mammal Lab., Natl. Mar. Fish. Serv.,NOAA, 7600 Sand Point Way N.E., Bldg. 32,Seattle, WA 98115.5Krogman, B. D., R. M. Sonntag, and H. W.Braham. 1979. Arctic Whale Task aerial surveyformat, 1979 version. Unpubl. rep., 28 p., Natl.Mar. Mammal Lab., Natl Mar. Fish. Serv.,NOAA, 7600 Sand Point Way .E., Bldg. 32.Seattle, WA 98115.

31

where T = total population size,C

i= ith component of the popu­

lation, andn = 5.

counts as the basis for estimating totalabundance, it is necessary to considerall segments (components) of the bow­head population not accounted for bythe Barrow counts.

To proceed along this line of reason­ing, a total estimate (T) for the bowheadpopulation can be expressed mathemat­ically as:

The advantage of this technique isthat each component can be studiedseparately and later combined to make atotal abundance estimate. Researchstrategies for estimating each compo­nent will now be presented.

C I is typically greater than actualcounts because there are intervals dur­ing the 15 April through 30 May periodduring which observers are unable towatch for whales. Unstable ice condi­tions and/or periods of poor visibil itybecause of fog or ice-choked leadssometimes prevent observers fromcounting. It is thus necessary to interpo­late for periods of missed watch so thatC

Iwill equal the total number of

bowheads passing the counting stationsduring the field season.

Braham et at. (1979), particularly whenonly a small percentage of the totalwatch period is missed. The ice campindex was computed as the summationof the products of the average rate ofwhales per hour of watch during eachday multiplied by 24 hours. In 1978, itwas computed as 2,264 with a totalrange of uncertainty equal to 1,082(Braham et aI., 1979). Regretfully, theterm "index" has been a source ofconfusion for some, and thus the termC I is introduced here as the "Barrowestimate. "

C2 and C 3 were evaluated by a fieldstudy conducted at Cape Lisburne in theearly spring of 1978 (Rugh and Cub­bage, 1980). The main objective of thatstudy was to determine the onset andtermination of the spring migrationalong the northwestern coast of Alaska.Results were combined with aerial sur­vey results and ice camp counts to esti­mate the magnitude of CJ and C J .

The value of C2 was "negligible ac­cording to results from Cape Lisburne.Bowheads were seen migrating pastCape Lisburne before Point Barrow. In1978, regular watches at Cape Lisburnecommenced 10 April and the first con­firmed sighting of bowheads occurred 8days later; bowheads were seen dailythereafter. At Point Barrow, regularwatches commenced 15 April, and thefirst bowhead was observed 5 days lat­er. Again, they were seen daily thereaf­ter. These similarities in temporal dis­tribution indicate that few, if any,bowheads migrated past Point Barrowprior to the commencement of the PointBarrow ice camp census.

C3

is estimated to be less than 4 per­cent of C I . This estimate for the numberof bowheads moving by Point Barrowafter the spring sampling period wasderi ved from ice camp data and wasverified by results from aerial surveys,as explained below.

A comparison of trends in dailymovements of bowhead whales migrat­ing by Point Barrow for the years1976-78 (Fig. 2) showed a decline inmovement during the last third of the 15April through 30 May period. An aver­age rate per day beyond 30 May wascomputed as 3.08 (0.94 SD). The end­ing dates of our observations for eachyear were: 2 June 1976,3 June 1977,5June 1978, and 30 May 1979. Using

(2)"+ ~ W,i even

W, = summation of the nurn­berofbowheads countedduring each period ofwatch, and

W, = summation of the num­ber of bowheads esti­mated as moving by thecamp during periods ofno watch.

"~

I even

6Braham, H. W, and B. D. Krogman. 1977.Population biology of the bowhead (Sa/aena mys­licelus) and beluga (De/phinaplerl/S tel/cas)whale in the Bering, Chukchi and Beaufort Seas.Processed rep., 29 p. atl. Mar. Mammal Lab.,Natl. Mar. Fish. Serv., NOAA, 7600 Sand PointWay .E., Bldg. 32, Seattle, WA 98115.

The length of each period Wi will varywith environmental conditions. Thefirst watch period of the season WI lastsuntil observers must abandon watch. Atthe moment WI terminates, W 2 beginsand continues until observers return towatch, demarcating the initiation ofw 3 ,

and so on.Obviously the value of each Wi odd

equals the number of whales countedduring each period of watch. It is nowleft to estimate w, e,en' After consider­able investigation of alternative meth­ods, such as polynomial curve fitting, itwas concluded that for any missedperiod x, W, even can best be approxi­mated by interpolation using the aver­age rate(s) of whale movement basedon counts made during the two adjacentperiods, each also equal in length toperiod x. Thus, if 2 hours are missed,an estimate of that missed period iscalculated based upon the preceding 2hours and following 2 hours of data. If3 days are missed, then 3 days preced­ing and 3 days following are used.Also, as is likely to occur in the latterexample, if some portion of an adjacentperiod is also missed, the rate is com­puted on whatever data are available inthat adjacent period.

The computation of C1

yields a valuesimilar to that reported as the "ice campindex" in Braham and Krogman 6 and

"where ~I odd

To proceed, let C I be estimated bye"defined as:

(I)T=~Ci=l I

Letting CI= number of whales passing

the ice camp during a spe­cific sampling period in thespring (15 April-30 May),

then C 2 number of whales passingby Barrow before thespring sampling period,

C 3 = number of whales passingby Barrow after the springsampling period,

C4

number of whales whichnever pass the ice camp,perhaps remaining in theChukchi and Bering Seasthrough summer, and

C 5 = number of whales passingby Barrow far offshore,beyond the range of sightof the observers at the icecamp.

32 Marine Fisheries Review

250---

1976200- 72

J6S 159 153

72

70

Alaska

Chukchi Sea ;;;2~~')~~ ,D ~)'l~

k ,

.r ~//"

f"'~r-tl==t==I~=r=t=l==t=r-:::r-=l-,---"'T-,-".---,-,--,---,---,-,-,--,--r-,-,--,--,--+H 68

70

68

~ 200--------=------

~ :~: ="'~"~;.~"~ 18 21 ~ V ~ 3 6 9 12 e 18 21 M 27 ~

I----- APR 1L

1977

/---APRfL

~ 250--------------

r---APRIL--+----MAV-----1

Figure 3. -Aerial survey tracklines flown in the eastern Chukchi and western Beau­fort Seas on 1,4. and 5 June 1976. Dots represent presence of bowhead whales:20 whales were counted with a mean group size of 1.8 (SD= 1.1).

~t::I:±t::I:±t±±t±±t±±:l±t:i±t±±±±tl±t:l±t:l±tJl±±:l±±:t::I:±t4-I '73

Figure 2. -Comparison amongyears (1976-78) of estimated totalnumber of bowhead whales migrat­ing northward past Point Barrow,Alaska, 15 April-30 May. For pur­poses of comparison, totals are basedon hourly rates per day times 24hours. Estimated yearly totals are796,715, and 2,264 for 1976, 1977,and 1978, respectively. Differencesin total estimates are ascribable to ob­server effort and weather. The 1978estimate is considered the best avail­able to date (Braham et aI., 1979).

73

165

161

159

]55 149

J53

143

Beaufort Sea

Figure 4, -Aerial survey trackl ines flown in the eastern Chukchi and western Beau­fort Seas 18-20 June 1976. The dOl (indicated by an arrow) represents one bowheadwhale.

search conducted since 1976 indicatethat few if any bowhead whales remainin the Bering and Chukchi Seas south of

the pack ice after the closure of the icecamps (Braham et aI., footnote 4; Dahl­heim et aI., 1980),

71

]43149155161

Chukchi Sea

71

only this average as a basis for ex­trapolating through the month of June,it can be estimated that approximately92 whales pass Point Barrow after theice camp counting station closes.

Aerial survey results suggest, how­ever, that dail y rates decl ine through themonth of June. Figures 3 and 4 illus­trate the difference in the number ofbowhead whales observed during thefirst week versus the third week of June1976. Thus, the estimated 92 whalescan be considered a maximum value.The value 92 is 4 percent of 2,264,which is the current estimate of thenumber of bowheads which pass PointBarrow during the 15 April through 30May period (Braham et aI., 1979).

C4 is apparently insignificant. Re­sults from vessel and aerial survey re-

Sepl.-Ocl.19RO 33

Cs is difficult to me3sure because theperimeter of the sample space, i.e., thefarthest distance that observers can reli­ably count whales, is not discrete. Mea­surement of C s has been attemptedprimarily through aerial survey, but re­sults have been slow in coming becausethe number of bowheads is small, andtechnological limitations have pre­vented an accurate determination of theposition of whales relative to the icecamps.

Based on results from aerial survey,few, if any, bowheads migrate morethan 8 km seaward of the ice camp;most are within 3 km (Braham, K rog­man, Johnson, Marquette, Rugh,Sonntag, Bray, Brueggeman, Dahl­heim, Nerini, and Savage, 1980). Onthe other hand, resul!s from the icecamps indicate that almost all bow­heads migrate within 5 km-mostwithin I km -of the landfast ice edge.This discrepancy is ascribable to differ­ences in methodology that have yet tobe resol ved.

Further complicating the determina­tion of the magnitude ofCs is the proba­bil ity that the number of whales passingbeyond the range of sight of observers isnot const3.nt. Yearly fluctuations in iceconfigurations may contri bute to shiftsin whale distribution relative to the icecamps. And, certainly, location andcondition of ice at the camps affect theobserver's ability to count whales.Changes in distributional patterns ofwhales relative to the ice camps there­fore prevent a reliable determination ofCs at this time. A subjective estimatebased on field experience is that theupper bound of C s does not exceed 30percent of the Barrow estimate and insome years, such as 1978, it is muchlower.

In summary, the above analysessuggest that components C

2, C3 , and C4

will probably account for only a smallnumber of whales in the total estimateof abundance. Too little is known re­garding the magnitude of Cs to predictits effect on future estimates. Further­more, this analysis supports the conclu­sions made by Braham, Krogman,Johnson, Marquette, Rugh, Sonntag,Bray, Brueggeman, Dahlheim, Nerini,and Savage (1980) that the "ice campindex" or "Barrow estimate" undercertain conditions can serve as an ap­proximation of total abundance.

34

Relative Abundance

Until now, out of interest in present­ing an overview, 1 ha ve postponed anydiscussion of accuracy or precision. Inall likel ihood, if the estimate of totalabundance is accurate, it will also beprecise, but not vice versa. A statementof accuracy describes how close to thetr~e value a particular estimate falls,whereas precision refers to the close­ness of each repeated measurement ofthe same quantity. In this section, theresearch strategy for determining accu­racy and precision will be presented.Note that the following discussion dealsalmost exclusively with the evaluationof C 1 , as its magnitude so overshadowsthe other components.

A typical approach to studying accu­racy is to uncover sources of error orbias. The sources of error associatedwith estimating C 1 are closely as­sociated with how well observers areable to count whales. An obvious errorwould be for an observer to miss seeinga whale altogether (e l ). This error wasstudied in 1978 and 1979 using two icecamps (see Field Methodology section)and is estimated at approximately 20percent, i.e., 20 percent of all whaleswhich swim by one camp are never seen(Braham, Krogman, Johnson, Mar­quette, Rugh, Sonntag, Bray, Brueg­geman, Dahlheim, Nerini, and Savage,1980). Other errors occur, but thesemust be defined as they relate to the wayobservers score whales.

Observers are asked to score allsightings made during a period Wi odd as:New sightings, duplicate or repetitivesightings, or conditional sightings,which occur when observers are unsureas to which of the previous twocategories in which a whale sightingbelongs. Here, error terms can be de­fined as: e2 , when an observer incor­rectly scores a whale as new after thewhale has already been counted; or e3 ,

when an observer incorrectly scores awhale as a duplicate before it has beencounted as a new whale. No error per secan be made regarding conditionalwhales.

To determine the accuracy of the es­timate of C I it is necessary to determinethe magnitude of e2 and e3 and to de­termine, if possible, the proportion ofconditional whales which were newrather than duplicates. To determine the

preCISIon of the estimate of C I , it isnecessary to determine the variation inmagnitude of these sources of error­observer variabil ity.

Computer Modeling toImprove Accuracy and Precision

Counting errors e 2 and eJ are beingevaluated using computer modeling.Field data collected on the diving pro­files of bowhead whales were used asthe basis for developing a model whichgenerates the surfacing pattern of apopulation of bowhead whales duringmigration. The model creates a database similar to the one based upon rawfield data, except that for the modeledpopulation, it is known exactly whichwhales that "swim by the camp" arenew and which are duplicates. The nextstep in the modeling procedure involvesthe development of a counting programto independently evaluate the modelpopulation for new and duplicatewhales. As the counting program pro­cesses each "sighting" it evaluates itagainst all previous sightings and,based upon probabilities, the programdecides whether or not the sighting isnew or duplicate.

On the average, the counting pro­gram overestimates the number of newwhales by 8 percent, and un­derestimates the number of duplicatewhales by 2 percent. But these resultsmust be further verified through fieldexperiments. The counting programwill be implemented in the field throughuse of a microcomputer. As observers atthe ice camps record observations, theywill also relay their data by radio to anearby laboratory for evaluation by thecounti ng program. Feedback basedupon computer evaluation will thenbe provided to the observer. Throughthis two-way communication much canbe learned about the nature of errorsmade by observers and by the countingprogram.

The results of this study will beapplicable to evaluation of accuracyand precision of the estimate of C I . As­suming that the counting program isverified, it will be applied to previousyears' field data for evaluation of count­ing errors. Upward or downward ad­justments can be made, resulting in amore accurate estimate of C I' Since ad­justments in counts will also have theeffect of negating much of the observer

Marine Fisheries Review

variability apparent from year to year(see Table I, Braham, Krogman,Johnson, Marquette, Rugh, Sonntag,Bray, Brueggeman, Dahlheim, Neri­ni, and Savage, 1980), the precisionof the estimate of C I will also beimproved.

Measurement of Accuracy andPrecision for Missed Data

Unfortunately, there are more factorsthan just observer variability which in­fluence the accuracy and precision ofthe estimate of C 1 by Equation (2). Asenvironmental conditions worsen, andobservational effort becomes more dis­continuous, the accuracy and precisionof the estimate of C 1 become a concern.Simply stated, an estimate of C I basedupon 95 percent watch effort would beconsidered more reliable than an esti­mate based upon only 35 percent. Forthe 95 percent case, interpolation to de­termine the number of whales whichpassed the ice camp during periods ofmissed watch would be required 5 per­cent of the time; for the latter case, 65percent of the time. Thus, as percentwatch effort decreases, the estimate ofC

1becomes less a measurement and

more an inference.The accuracy of Equation (2) and the

rate at which its precision decreases arebeing studied with the aid of a computermodel. The basis of this model is a database which contains whale counts madeduring a continuous period of watchwhich lasted 18 days from 2 to 20 May1978. The total number of whalescounted during this period equaled1,133 and, for the purposes of this ex­periment, is considered equal to C 1, thetrue number of bowheads which passedthe counting station.

The model simulates the effects ofreduced watch effort by introducingperiods of missed watch which vary inlength and frequency. Theoretically,watch effort can be varied from 0 to 100percent, but for practical reasons, ex­periments have been run from 6 to 98percent by increments of 2 percent. Adetermination of accuracy and preci­sion is made for each increment.

As an example of how accuracy andprecision are determined for a givenpercentage of watch effort, consider aseason where 80 percent of the time wasspent watching. The model would ini­tially remove 20 percent of the data, and

Sept.-Oct. 1980

calculate an estimate of C I' The differ­ence between the estimate and the trueC

1is called a residual. Following this,

the model would reselect 20 percent ofthe data for removal, and recompute C I'

again computing a residual. After per­forming this test many times, an aver­age and standard deviation are com­puted for the residuals.

It is the average of the residualswhich is used to evaluate accuracy.Theoretically, if the experiment is re­peated often enough and if Equation(2), which is used to estimate C 1 , isunbiased, then the average residualshould equal zero; that is, there shouldbe no difference between estimated andtrue values of C 1• A significant upwarddeparture from zero indicates thatEquation (2) is upward biased, i.e.,tends to overestimate the number ofwhales which swim by during periodsof missed watch. A significant down­ward departure similarly indicates thatEquation (2) is downward biased. Pre­liminary results suggest that Equation(2) is unbiased, or accurate.

The standard deviation of the residu­als forms the basis for determining pre­cision. When Equation (2) repeatedlyestimates C 1 very closely, the standarddeviation of the residuals will be small,indicating good precision. When re­peated estimates of C 1 vary widely, thestandard deviation of the residuals willbe large, indicating a lack of precision.Through the use of the computermodel, the rate of growth of the stan­dard deviation is being traced as a func­tion of watch effort.

In summary, this research shouldeventually provide the criteria whichwill allow annual changes in estimatesof C 1 to be evaluated statistically, thusimproving the chances of detecting sig­nificant changes in population size ifand when they occur. Furthermore, theevaluation will take into account varia­tions in percent total watch effort,thereby increasing the usefulness ofmarginal counting years for monitoringpopulation stabil ity.

Discussion and Conclusion

This paper was written primarily toexplain the National Marine MammalLaboratory's research strategy forpopulation enumeration of bowheadwhales. Other avenues of research thathave been seriously investigated have

not been addressed in this report. Forexample, censusing techniques usingactive and passive acoustic deviceshave received considerable attention(Braham, Krogman, Johnson, Mar­quette, Rugh, Sonntag, Bray, Brueg­geman, Dahlheim, Nerini, and Savage,1980; Braham, Krogman, Nerini,Rugh, Marquette, and Johnson, 1980).Aircraft have been used as plat­forms for validating ice camp counts(Braham et aI., 1979). Even remotesensing methods which utilize satelliteshave been explored. These alternativecensusing methods are being investi­gated because it is through independentverification of research results thatmeaningful statements regarding popu­lation abundance can be made.

It should be remembered, too, thatno matter how difficult it may be toestimate total population size, it is en­tirely feasible that we can monitor thestatus of this population by detectingchanges in its size through time. Onesimple approach might be to eliminatethe problem of observer judgments andwhale counting by having observersrecord only the number of sightingsthey make without attempting to trans­late the number of sightings intonumber of whales. For example, as­suming environmental factors are stan­dardized, if the number of sightings in­creases 5-fold over a period of years, aninference could be made that the totalpopulation size is also increasing. Itwould be premature to adopt thismethodology, however, since it has notbeen shown that observers do not countaccurately. It may be that inquiries willshow not only that observers are highlyreliable, but that both total abundanceand relative abundance can be reliablyestimated.

Acknowledgments

I gratefully acknowledge H. Brahamand T. Bray for help and guidance in thepreparation of this manuscript. Of par­ticular assistance during data analysisand computer modeling were R.Sonntag, R. Grotefendt, J. Kuhlmann,and G. Priebe. Thanks is extended tothe many seasonal employees who par­ticipated at the ice camps since 1976,especially G. Carroll, S. Savage, J.Smithhisler, and E. Iten. M. Tillmanand J. Breiwick kindly offered sugges­tions during the preparation of this

35

HOWARD W. BRAHAM, MARK A. FRAKER, and BRUCE D. KROGMAN

Spring Migration of the WesternArctic Population of Bowhead Whales

paper. The 1976-77 research seasonswere funded by the Bureau of LandManagement under the Outer Continen­tal Shelf Environmental AssessmentProgram office, NOAA, Juneau,Alaska; 1978-79 seasons were fundedthrough NOAA's expanded BowheadWhale Research Program.

Literature CitedBailey, A. M., and R. W Hendee.

1926. Notes on the mammals of northwesternAlaska. J. Mammal. 7:9-28.

Braham, H. w., M. A. Fraker, and B. D. Krog-

IntroductionEach spring the western Arctic popu­

lation of bowhead whales, Sa/aenamysticetus, migrates from the BeringSea, through the Chukchi Sea, and intothe Beaufort Sea. For centuries, coastalEskimos of western Alaska and easternSiberia have taken bowheads duringspring as the whales moved past theirvillages soon after openings formed inthe pack ice. Traditionally, Eskimossailed or paddled their boats out intocracks and open water areas in the ice,called leads and polynyas, respectively,from April to June to hunt the whales.The breakup of the pack ice and migra­tion pattern of the whales are so regularthat the whales are reliably accessible towhalers each spring, but only for a fewweeks. AtSt. Lawrence Island, Alaska,

Howard W. Braham and Bruce D. Krogman arewith the National Marine Mammal Laboratory,Northwest and Alaska Fisheries Center, NationalMarine Fisheries Service, NOAA, 7600 SandPoint Way N.E., Bldg. 32, Seattle, WA 98115.Mark A. Fraker is with LGL Limited, Environ­mental Research Associates, 2453 Beacon Av­enue, Sidney, B.C. V8L IX7, Canada.

36

man. 1980. Spring migration of the west­ern Arctic population of bowhead whales. Mar.Fish. Rev. 42(9-10):36-46.

____ , B. Krogman, J. Johnson, W Mar­quette, D. Rugh, R. Sonntag, T. Bray, J.Brueggeman, M. Dahlheim, M. Nerini, and S.Savage. 1980. Population studies of thebowhead whale (Balaena mysticetus): Prelimi­nary results of the 1979 spring research sea­son. Rep. Int. Whaling Comm. 30:391-404.

____ , , S. Leatherwood, W.Marquette, D. Rugh, M. Tillman, J. Johnson,and G. Carroll. 1979. Preliminary reportof the 1978 spring bowhead whale researchprogram results. Rep. [nt. Whaling Comm.29:291-306

____, , M. Nerini, D. Rugh, W.Marquette, and J. Johnson. 1980. Re­search on bowhead whales. June-December

for example, present-day whal ing oc­curs from approximately the first weekin April to about the first week in May;at Barrow, Alaska, the peak of bothmigration and whaling activity occursfrom the last week in April to the lastweek in May (Braham and Krogman!;Marquette2

, 1979; Braham et aI., 1979;Braham et al. 3).

I Braham, H .. and B. Krogman. 1977. Populationbiology of the bowhead (Balaeno mysticetus) andbeluga (Delphinopterus leucos) whale in the Be­ring, Chukchi and Beaufort Seas. Processed rep.,29 p. Natl. Mar. Mammal Lab., NOAA, 7600Sand Point Way N.E., Bldg. 32, Seattle, WA98115.2Marquette, W. M 1977 The 1976 catch ofbowhead whales (Balaenu myslicetus) by Alas­kan Eskimos, with a review of the fishery, 1973­1976, and a biological summary of the species.Processed rep., 80 p. Natl. Mar. Mammal Lab.,NOAA, 7600 Sand Point Way N.E., Bldg. 32,Seattle, WA 98115.3Braham, H., B. Krogman, andG. Carroll. 1979.Population biology of the bowhead whale(Balaena mysticetus) II: Migration, distribution,and abundance in the Bering, Chukchi, andBeaufort Seas, with notes on the distribution ofwhite whales (Delphinuplerus leucas). Unpubl.final rep., OCSEAP Contract No. R7 120807 , 118p. Natl. Mar. Mammal Lab., NOAA, 7600 SandPoint Way N.E., Bldg. 32, Seattle, WA 98115.

1978. Rep Int Whalmg Comm 30"405-413Cook, J. 1926 Pursumg the whale, a quarter

century of whaling in the Arctic. HoughtonMifflin Co., Boston, 344 p.

Dahl heim, M., T. Bray, and H. Braham.1980. Vessel survey for bowhead whales inthe Bering and Chukchi Seas, June-july1978. Mar. Fish. Rev. 42(9-10):51-57.

Marko, J. R. 1975. Satellite observations ofthe Beaufort Sea ice cover. Dep. Environ.,Victoria, B.C., Beaufort Sea Tech. Rep. 34,137 p.

Rugh, D. J., and J. C. Cubbage. 1980. Migra­tion of bowhead whales past Cape Lisburne,Alaska. Mar. Fish. Rev. 42(9-10):46-51.

Townsend, C. H. J935. The distribution ofcertain whales as shown by logbook records ofAmerican whaleships. Zoologica (N.Y.) 19,50 p.

Townsend (1935) plotted the loca­tions of bowheads killed by Yankeewhalers operating in the Bering Sea andArctic Ocean from 1848 to 1919. Hismap suggests that bowheads may haveformerly occurred in the Bering andChukchi Seas, as well as in the BeaufortSea, during the summer months, wellbeyond the present-day whaling sea­son. Data collected since 1974, how­ever, indicate that bowheads do occur inthe eastern Beaufort Sea during thesummer (Fraker and Bockstoce, 1980),but that probably only a few occur in theBering and southern Chukchi Seas dur­ing summer (Braham et aI., footnote 3;Braham, Krogman, Nerini, Rugh,Marquette, and Johnson, 1980; Dahl­heim et aI., 1980).

Although the general timing and pat­tern of movements of bowheads duringthe spring migration are known (Baileyand Hendee, 1926; Tomilin, 1957;Foote 4

; Durham 5; McVay, 1973;

Braham and Krogman; footnote 1), theprecise pathway that they take has notbeen fully described-especially in thelargely frozen Beaufort Sea. I n thispaper we discuss the spring migrationroute, March through June, and de­scribe ice conditions encountered bythe whales. In addition to gaining anunderstanding of an important part of

'Foote, D. C. 1964. Observations of the bowheadwhale at Point Hope, Alaska. Unpubl. manuscr.,73 p. McGill Univ., Montreal, Quebec, Can."Durham, F. E. 1972. Biology of the bowheadwhale (Balaena mysticelus) in the western arctic.Unpubl. manuscr., 93 p. Dep. BioI., Univ. South.Calif., Los Angeles, CA 90007.

Marine Fisheries Review