JOURNAL
OF FIELD
ORNITHOLOGY
Published by
Association of Field Ornithologists
VOL. 70, NO. 2
Spring 1999
PAGES
145--282
J. Field Ornithol., 70(2):145-157
USE
OF RADAR
FOR MONITORING
COLONIAL
BURROW-NESTING
SEABIRDS
D. F. BERTRAM
Wildlife & BehaviouralEcologyResearch
Groups
Departmentof BiologicalSciences
SimonFraserUniversity,8888 UniversityDrive, Burnaby
British
Columbia
V5A 1S6
and
EnvironmentCanada, Canadian Wildlife Service,Pacific WildlifeResearch
Centre
RR1 5421 RobertsonRd, Delta British Columbia, V4K 3N2, Canada
L. COWAN
Wildlife & BehaviouralEcologyResearchGroups
Departmentof BiologicalSciences
SimonFraserUniversity,8888 UniversityDrive, Burnaby
British
Columbia
V5A
1S6
A. E. BURGER
Departmentof Biology
Universityof Victmia
P.O.
Box 3020
Victoria, British Columbia
V8W 3N5, Canada
Abstract.--We used high-frequencysurveillanceradar as a non-intrusivemethod to census
nocturnal burrow-nestingCassin'sAuklet (Ptychoramphus
aleuticus)on the world's largest
colony,Triangle Island, British Columbia. In the largestsubcolony,WestBay,the radar unit
wasoperated 2200-0510 h nightly from 30 Apr.-11 May 1996 during the onsetof hatching.
Radar imageswere storedon video cassettefor subsequentimage analysis.We report total
nightly activitybasedupon cumulativesamplesfrom the tapesat 30-sintervalsto determine
the percent of bird activitywithin a fixed area. Estimatesof the averageimage sizeof an
individualbird wereusedto developconservative
countsof birdsin the samplearea.Activity
at the colony began approximately 1.5 h after sunsetand ended at least 15 rain before
sunrise.Activitylevelsincreasedover the studyperiod and showedconsiderablynightly variation. Nightly activitywas continuousat low levels through the samplingperiod from 30
Apr. to 3 May. Late eveningpeaksof activityaround 2300 h were evidentand tended to
increasein intensityfrom 3 May onward.Nightly activityat the colonywascorrelatedwith
the number of chickshatchedthe followingday.The maximumestimateof individualbirds
detectedwas156,327on 10 May.Radar hasgreat potentialto elucidatepatternsof seasonal
activity.We contrastand evaluatethe use of radar and traditional methodsbasedon burrow
145
146]
D. F. Bertramet al.
J.Field
Ornithol.
Spring 1999
countsand identify severalmajor advantagesfor radar in long-term population monitoring
and seabirdresearchprograms.
USO DE RADARES PARA MONITOREAR AVES MARINAS COLONIALES QUE ANIDAN
EN
HUECOS
Sinopsis.--Utilizamosradaresde seguimientode alta frecuenciacomoun m6todo no intrusivo para censarindividuosde Ptychoramphus
aleuticusanidando en huecosen la colonia
mils grande del mundo, en la Isla del Trifingulo, Columbia BritJnica. En la subcoloniamils
grande,WestBay,la unidad de radar se oper6 de 2200 a 0510 h todaslas nochesdesdeabril
30 hastamayo 11 del 1996 durante el comienzodel empollamiento.Se almacenaronlas
imfigenesde radar en cintasde video para analizarlasluego. Reportamosactividadnocturna
total basfindonosen muestrasacumuladasde las cintasa int•rvalos de 30 s para determinar
el porcentajede actividadde las avesdentro de un hrea fija. Los estimadosdel tamafiode
imagen promedio de un ave individualse usaton para desarrollarconteosconservativos
de
avesen el firea de muestreo.La acfividaden la coloniacomenz6cercade 1.5 h despu•sdel
anochecery terminaba al menos 15 min antes del amanecer.Los nivelesde actividadaumentaron a travis del perlodo de estudioy mostratonuna variaci6nnocturna considerable.
De noche, la actividadfu• continua (a bajosniveles)a travis del periodo de muestreode
abril 30 al 3 de mayo.Los picosde actividadtarde en la noche rueton evidentescerca de
las 2300 h y de mayo 3 en adelante hubo la tendenciade aumentaren intensidad.Se correlacion6 la actividadnocturna con el nfimero de pichoneseclosionadosel dia despu•s.E1
estimadomfiximo de avesindividualesdetectadosfu• de 156,327 en mayo 10. E1 uso de
radarestiene un gran potencialpara elucidar patronesde actividadestacional.Contrastamos
y evaluamosel uso de radaresy m•todos tradicionalesbasadosen conteosde huecose
idenfificamosvariasventajasimportantespara el usode radaresen programasa largo plazo
de monitoreo de poblacionesy de investigaci6nde avesmarinas.
Most modern schemesfor censusingand monitoring numbers of burrow-nestingseabirdsare basedupon countsof "apparentlyoccupiedburrows" (Anker-Nilssen and Rostad 1993, Walsh et al. 1995) or estimates of
occupiedburrows(calculatedfrom the productof countsof burrowsand
the percentageof burrows known to be occupied through physicalexamination [e.g.,Rodwayet al. 1988,Bertram1995]). Countsof apparently
occupied burrows, however,are subjectto significantobserverbias, and
the necessitythat surveysbe conducted prior to the arrival of juvenile
prospectorsand when vegetation is low and occupancyis most visible
(Anker-Nilssen and Rostad 1993). In addition, burrow-based counts un-
derestimatetrue adult population size (Anker-Nilssenand Rostad1993).
Estimates of "true" burrow occupancy are not suitable for long-term
monitoring for some speciesbecause physical examination causesfrequent desertionsand thus lowersthe chancethat thoseburrowswill be
usedin subsequent
years(Gastonet al. 1988,Rodwayet al. 1996).Further
drawbacksof this method are reduced repeatabilitybecauseit is difficult
to define objectivelywhat constitutesa burrow; possiblebiasin estimates
of occupancybecausenest chamberscannot alwaysbe inspected;and
intensive labor requirements because,in addition to counts, a subsetof
the burrowsmust also be excavated(Gastonet al. 1988). In fragile habitats, methodsthat require burrow counting (and in particular thoserequiring excavation)are potentiallydestructiveto coloniesthus compromising conservationefforts.
The purposeof thispaper is to report the useof radar asan alternative,
Vol.7o,No.2
RadarMonitoring
ofSeabirds
[ 147
non4ntrusivemethod for monitoring, censusing,and quantifyingcolonyattendancebehaviorof a colonialburrow nestingseabird.Radar hasbeen
used in studiesof seabird movements(Day and Cooper 1995) and for
estimating numbers of the non-colonial Marbled Murrelet (Brachyram-
phusmarmoratus,
Burger1997) but had not yet been deployedfor census
and monitoring studieson a major colonyof burrow nesters.We contrast
the radar and the burrow count (seeabove)methods,identifyimportant
areasfor further research,and highlight the potential utility of radar for
long-term monitoring and researchprogramsfor seabirdpopulationsin
general.
METHODS
We studied Cassin'sAuklets (Ptychoramphus
aleuticus)at the world's
largest colony on Triangle Island (the Anne Vallae EcologicalReserve,
51ø52'N,129ø05'W),in BritishColumbia,Canada.It is noteworthythat
80% of the world's breeding population of Cassin'sAuklet are found
along the rugged islandsof northern coastalBritish Columbia (Rodway
1991). Triangle Island is located 46 km northwestof Cape Scotton VancouverIsland (seeVermeeret al. [1979] for a descriptionof the breeding
habitat). From 30 Apr.-11 May 1996we deployeda high-frequencymarine
surveillance radar unit (Furuno FR810D, 940MHz, 10KW with a 2.0-m
antenna) in WestBay (Fig. 1) where the largestburrow densitiesof breeding Cassin'sAuklet were reported. Cassin'sAuklet was the only seabird
nesting in West Bay in significant numbers (Rodway et al. 1990). The
radar unit was positionedabove high tide at the base of the edge of a
rockyslopebehind a large log and woodenfence that reduced the interference causedby waveactivity.The radar antennawas75 cm abovethe
ground and 30 cm below the top of the log. The radar antennawastilted
15øup from its normal horizontalpositionand,with an anti-clutterscreen
mounted on the lower edge (Cooper et al. 1991, Burger 1997), scanned
an estimated vertical are of 35 ø above the horizon.
Both the rain and sea-
scattersuppressers
were turned off to maximize the sensitivityand hence
the likelihoodof bird detection.The gain wassetat 75% of maximumto
maximize bird detection and minimize noise. Radial line markers at 240 ø
and 340ø were added to the radar screen.The scanningradiuswas0.5
km. The time taken for the antenna to make a completehorizontalrevolution was3.02 m 0.01 s (SD, Burger 1997).
Video imagesof the radar screenwere obtained by installing a video
converter (Furuno RP-6065B) between the radar and a VHS video cassette
recorder.A time-datestampwasalsoplacedon the recording.The system
wasoperatedwith a seriesof 12V batteries.Each night the radar unit was
startedbetween 2200-2300 h and stoppedat 0300-0510 depending on
the availablebatterypower.
The total sampleregion included the area abovethe ocean in WestBay
and horizontally 100-500 m from the radar unit. The two dimensional
samplingarea from the radar imagewas125,732m2 (Fig. 2a). Mostbirds
flew straightin and out of the colony and circling birds were not seen
148]
D. F. Bertramet al.
J.Field
Ornithol.
Spring 1999
N
West
South
Bay
o
lOOO
lOOO
2000
Meter•
TnangleIsland
Vancouver
Island
Islands
FIGURE1. Triangle Island, British Columbia showingthe location of the radar unit in West
Bay in 1996 (©).
on the video tapes.We assumedthat birds circling over the colonywould
staywithin 100 m of the radar unit and hence not be counted twice.
Seabird activitywas sampledfrom videotapesevery 10 min throughout
the night and every 2 min near peak activityusing an image acquisition
device (Play TechnologiesSnappyVideo Snapshot,with the contrastset
to the maximum value) on a computer (Nantron Systems,pentium, 120
mhz). We did not sampleduring periodsof rain becausebird imageswere
obscured.Birds flying at different heightsor distancesproducedvarying
amountsof white on the video imagesthus creating a greyscale.We used
Vol.70,No.2
RadarMonitoring
ofSeabirds
[ 149
A) Low activity
B) High activity
F[cup•2. Theradarimages
captured
fromvideotape
during
lowactivity
(A:1 May,0020
h) andduring
highactivity
(B:10May,2333h) of Cassin's
Auklets
in WestBay.
The
sampling
areaforactivity
determination
isdelineated
in bothpanels.
Theringsare
separated
by 100m andformtheboundaries
for quadrants
1-4.
150]
D. F. Bertramet al.
J.Field
Ornithol.
Spring 1999
the UTHSCSA Image Tool program to create binary images (black and
white) by convertingall graypixelsto white usingthe thresholdfunction.
For each "captured" image the samplingarea in West Bay was selected
and a histogramshowingthe percent white and blackwascreated.Activity
wasdefined as the percent of white pixels due to bird attendancein the
scanned area. This was calculated by subtractingthe number of white
pixels in the imagesfrom control imagesthat were taken from the beginning or end of the tapeswhen no birds were present.Tide heightsof the
control and sampleimageswere matchedas closelyaspossible,to accommodate changesin shorelineland appearingin the image.
The flight speedsof Cassin'sAukletswere estimatedfrom the distance
coveredbetweensuccessive
imageson the radar screen (measuredwith a
ruler with a precisionequivalent to 10 m) divided by the time taken for
the antenna to make a complete horizontal revolution (3.02 s). To travel
the 500 m from the edge of the radar screen to the colony at the minimum speed (60 km/h: max. -- 132 km/h, mean -- 95.8, SD = 14.8 km/
h, n = 60) would require approximately30 s.Samplingevery30 s counted
the slowestbirds only once and thus produced conservativeestimatesof
nightlyactivity.To obtain estimatesof the number of birdsat 30-sintervals
we interpolated from the imagescaptured at 2-min or 10-min intervals.
Samplingboutswere not alwaysthe sameduration, so to comparetotal
nightly activitywe interpolated at the beginning and/or the end of tapes.
We assumed that the earliest (2200, 2210, or 2225 h) and the latest (0510,
0525, or 0535 h) interpolation times had no bird activitybasedupon our
observationsat thosetimes on other nights.Followinginterpolation,col-
ony activitywasgaugedfor a 7 h 10 min period (861 30-sintervals)each
night by the cumulativetotal of white pixelsfrom all 30-sintervals.
To estimate the number of individual birds in the samplingarea we
calculated "bird area" estimatesbased upon echoes of known single
birds. We testedfor differencesin the size of echoesof singlebirds by
sampling 30 individualsin each of four quadrantsat distancesof 100-200
m, 200-300 m, 300-400 m, and 400-500 m from the radar unit (quadrants 1-4). If differences could not be detected we would use the median
valuefor all quadrants.To calculatea conservativeestimateof the number
of birds in the radar imageswe divided the area of activityby the median
area
of an individual
Cassin's Auklet.
The
cumulative
totals
of birds
ob-
servedat 30-sintervalswere reported for each night.
In the Farallon Islands,Manuwal and Thorsen (1993) reported that
incubating and brooding pairs of Cassin'sAuklet alternate duties every
24 h. Brooding lastsfor up to 5 d after hatching. Followingbrooding,
both parentscan deliver food but singlenightly feedsare most common
(provisioningrateson Triangle Island vary from 0-2 feedsper night; DFB,
unpubl. data). Our estimatesof activity and the numbers of individual
birds in West Bay assumedthat all birds were incubating, brooding, or
one parent from each pair was delivering a single meal then brooding.
Hence each parent is detected once, either while returning to or departing from the colony.
Vol.
70,No.2
RadarMonitoring
ofSeabirds
25000
20000
15000
10000
5000
o
0
2
4
6
8
10
12
Date in May
25000
B
ß
20000
15000
10000
5000
ß
ß
ß
ß
ß
o
0
2
Number
4
6
8
of chicks hatched
FIGURE
3. Increasein cumulativenightlyactivityduringthe studyperiod (A). The relationshipbetweenhatchdate and cumulativenightlyactivitylevelsduring the studyperiod
on Triangle Island (B).
On four occasions
concurrentwith radar monitoringwe capturedGassin'sAukletsastheydepartedfrom the colonybetween0200-0430h using
a "pheasantnet" 15-mlongand 3-mhighin WestBay.We scoredthe iris
color type aswhite, offwhite,light brown,or brown,basedupon minor
modifications
of the techniqueof Manuwal(1978). Eyecoloris reported
to becomelighter asthe birdsage (seealsoEmslieet al. 1990).
Hatchingdatesof nestlingsin WestBaywere estimatedby inspection
of a sampleof burrowsat 3-d intervalsbeginning30 d after laying.
RESULTS
Activitylevelsandpatterns.--Activity
levelsincreasedoverthe studyperiod (Fig. 3a) and showedconsiderable
nightlyvariationin pattern (Fig.
4). Activityat the colonybeganapproximately
1.5 h aftersunsetandend-
152]
D. F. Bertram et al.
J. FieldOrnithol.
Spring 1999
6o
/rMay
j
20 •April
•30
•/1
4O
0
0
0
100
200
,.•
0
0
100
200
300
400
,81
May
112 ,/
40
40
•
400
-[
60
I
•
300
100 200 300 400
6o
0
100
200
300
400
May 213
40 ,•-•
20 • • I I•/
•,•
0
.•,
60
•
40
<
0
0• • ] •
0
100 200 300 400
100
200
300
400
300
400
May
314
40
•
0
•
100
200
300
0
400
.
•
0
I
100
4o
200
May
10111
0
0
100
200
300
400
0
100
200
300
400
Time (min)
FIcu• 4. Patternsof "nightly"activity(2200-0510h) from 30 Apr.-11May 1996in West
Bay,Triangle Island as gaugedby radar.Video imageswere "captured"at 2-10-min
intervalsand the othervalueswereinterpolated.
Activitywasgaugedasthe percentage
of the samplingareathat wasoccupiedby birds.Arrowsdepictthe time of sunrise.
ed at least15 min before sunrise.Nightlyactivitywascontinuousat low
levelsthroughthe samplingperiodfrom 30 Apr.-3 May.Late evening
peaksof activityaround2300 h (whichlastedapproximately
2 h) were
evidentand tendedto increasein intensityfrom 3 Mayonward.Activity
levelsextendedprogressively
into the latenightfrom7 Mayonwards
until
the end of sampling.
Breeding
chronology
and activitylevels.--Thefirstaukletnestlinghatched
Vol.70,No.2
RadarMonitoring
ofSeabirds
[ 153
TAI•LE1. Cumulative"nightly" activity,time, and magnitudeof maximumactivityof Cassin's
Auklets as gauged by radar in W•st Bay of Triangle Island in 1996. Estimatesof the
cumulativenightly number of birds in West Baywere basedupon the median estimates
of the area for an individual
bird.
Cumulative
Date
Time
radar
operated
30 April
1 May
2May
3May
2238-0449
2200-0510a
2250-0330•
2245-0535c
4May
2230-0500
6May
7May
8May
9May
10May
2245-0445
2210-0430d
2230-0440
2230-0440e
2220-0440f
activity Maximum Time of
(white
activity maximum
pixels/100)
( %)
activity
3057.34
2157.57
4410.50
2875.52
7.62
6.81
8.52
30.93
2340
0400
0330
2245
Total area
of birds
(me)
Estimated
number of
birds
3,844,032
2,712,733
5,545,369
3,615,430
21,475
15,155
30,980
20,198
5525.16
62.99
2246
6,946,848
38,809
12,501.89
9715.62
12,955.84
17,112.31
22,255.85
59.26
52.36
51.92
62.22
69.01
2258
2304
2318
2314
2302
15,718,773
12,215,567
16,289,530
21,515,509
27,982,542
87,814
68,243
91,003
120,198
156,327g
Except 0220-0400, 0402-0420 h.
Except0027-0306h.
Except0030-0150 h.
Except0240-0300 h.
Except 0215-0225, 0410-0425 h.
f Except2225-2310, 0150-0210, 0240-0300 h.
Estima•s were 1,974,774,44,716 and 142,810usingthe minimum,maximumandmean
valuesforindividual bird area, respect•ely.
on 2 May in 1996. The median hatch date was 11 May with the latest
hatching on 16 June (n = 56). During the radar sampling period the
number of chickshatchinggenerallyincreased.Nightly activityat the colony waspositivelycorrelatedwith the number of chickshatchedthe following day (r = 0.93, n = 10, P < 0.001, Fig. 3b).
Conservativecountsof birdsin WestBay.--The size of individual bird
images did not differ significantlywith the distance from the radar unit
and therefore samplesfrom all qaudrantswere pooled to estimatenumbersof birds. The median bird image was179 me (mean = 196.5, SE =
10.2, n = 124; range = 14.6-661.3).
As expected,the increasein number of birds in early May mirrored
the activityresults (Table 1). The maximum estimate of individual birds
detectedwas 156,327 on the night of 10-11 May. The minimum was
15,155 on 1-2 May.
At the netting site in West Bay eye color became progressively
darker
at later capture sessions.No brown-eyedbirds (presumed to be immatures) were captured on 5 May (n = 23) and 8 May (n = 31) but they
appearedamong6% of captureson 10 May (n = 28) and 11 May (n =
27). The broad category,light brown,increasedfrom approximately18%
on the first two capturebouts to 40% in the last two bouts.Conversely,
the incidenceof white eyedbirdsdeclined49% to 16% during the same
154]
D. F. Bertram
et al.
J.Field
Ornithol.
Spring 1999
time frame. The number of birds with off-white eye color wassimilar for
all capturesessions
(range 25%-43%).
DISCUSSION
Nightly activitylevelsin West Bay varied seven-fold,with the minimum
occurringon 1 May and the maximum on 10 May. Basedon the assumptions of 24-h shiftsin incubation,and singlefeeds/night coupledwith
brooding of young nestlings, the observed increase in activity during
hatchingwasnot expectedand thuswarrantsconsiderationof the validity
of the assumptions.For example, the magnitude of shift duration for
incubationand brooding may be variableand longer than 24 h. Data on
Cassin'sAuklets from Frederick Island, British Columbia (1995-1997) do
vary (e.g., some48 h and 72 h incubationshifts),but mostnestingpairs
(95% or greater) exchangeincubation duties every 24 h (A. Harefensit,
unpubl. data). Information from another project on Triangle Island in
1998 alsosuggests
48 h incubationin 5 of 16 burrowsexamined.Variation
in shift duration coupledwith more regularvisitsaround hatchingcould
have contributedto the observedincreasein activityin our study.
It is also possiblethat somenestlingsmay be fed by a singleadult and
not brooded, resulting in two detectionsof the same individual when
incoming and outgoing.Someof the oldestnestlingsmay alsohavebeen
fed by both parentsand not broodedin the later part of the studyperiod
in which caseboth parentswould be detected twice. Increasesin colony
activityat hatching have alsobeen reported for Atlantic Puffins (Fratercula
arctica;Harris 1984). Increasedattendanceby young, non-breedingbirds
at the colonyduring hatchinglikely contributedto the observedincrease
in activityduring our study.Birds with light brown and brown irisesare
most likely to be 1-2 yr old and immature, while thosewith white and
off-whiteirisesare most likely to be breeders(Manuwal 1978, Emslieet
al. 1990). A larger proportion of dark-eyedbirds were capturedin the
later netting sessions.
Harris (1984) alsonoted that the number of young,
non-breedingAtlantic Puffins on the colony increasedas the breeding
seasonprogressed(Harris 1984).
In British Columbia, a comparisonof four surveymethods (quadrat,
transect,point-centerquarter and Batchder's) identified the quadrat and
transectmethodsas most accurate(Savardand Smith 1985). That finding
led to the developmentof the principal inventorymethod usedfor most
seabirdcoloniesin the province,including Triangle Island. The census
technique (Rodwayet al. 1988; Bertram 1995) is basedon countsof burrowsin plotsplacedat specificintervalsalong transectsperpendicularto
the shore, averageburrow density,estimatesof colonyarea, and burrow
occupancyestimates.The total estimatednestingpopulationon Triangle
Islandbasedon 1989 transectsamplingwas547,637 _ 25,748 (SE) pairs
(Rodwayet al. 1990). In West Bay the breeding population estimates
ranged from 96,346 pairs (Rodwayet al. 1990) to 130,359 pairs (CWS
unpubl. data) depending upon the burrow densityestimateused in calculations.This translatesto approximately195,000-261,000breedingin-
Vol.70,No.2
RadarMonitoring
ofSeabirds
[ 155
dividualsin WestBayin 1989. In the presentstudy,the maximum number
of birds detected was 156,327 based on the median size of individuals on
the radar screen.This is likely to be a conservativeestimatebecausethe
samplinginterval (30 s) wasbasedon the slowestflight speedsobserved,
and we could not detectbirds that flew into WestBayoverland. A further
potential sourceof error, which could produce underestimates,is overlapping and obscuredimagesof birds flying near each other. The relationshipbetweenthe actualnumber of Cassin'sAuklet and the number
detected would be asymptoticif bird overlap and obscuredimagesincreasedwith activitylevels.Note, however,that the maximum observed
activityneverrepresentedmore than 70% of the samplearea,and that
thosehigh levelswere short-livedthus reducing the potential for underestimationdue to bird overlap.Finally,manydepartingbirdsdivesteeply
from the slopesand often staycloseto the water perhapsreducingtheir
chance of detection on radar and thus contributing to the underestimation of individualbirds. The similar magnitude of the estimatesof numbers basedon the 1989 transectsurveyand the presentstudysuggests
that radar can be a useful tool for quantifyingpopulationchanges.We
evaluateand summarizethe merits of radar versusmonitoring techniques
basedupon burrow countsbelow.
Gastonet al. (1988) suggestthat a monitoring techniqueshould provide an accurateindex of populationtrends,be fast,havelittle observer
bias and causeminimal disturbanceto breeding birds. Our work shows
that the radar method has strongpotentialto providean accurateindex
of populationtrends.Prior to implementation,however,general design
principles (e.g., choice of sampleunit, when to sample,and statistical
power; see Peterman 1990, Thomas 1996) for a radar-basedmonitoring
programwill need to be consideredto developthe mostsensitiveindex
from the leastamount of field effort. It is noteworthythat recent analysis
of Norwegianlong-termdata derivedfrom burrow counts(Anker-Nilssen
et al. 1996) showedthat trendsdocumentedfor someseabirdpopulations
could have been detectedwith much lessfield effort (fewer plots and less
frequent visits).By contrast,in British Columbia,population trendshave
been difficult
to detect because few colonies have been visited in less than
5-yr intervals,and plots on transectsurveyshave coveredonly limited
proportionsof the colonyarea. Moreover,on LangaraIsland,BritishColumbia, populationestimatesbasedon transectsurveysdid not reflect a
decline in Ancient Murrelets despitethe fact that colonyextent had diminishedby half from rat predation (Bertram 1995, Bertram and Nagorsen 1995). The present studywas confined to West Bay, but complete
colonycoverageon Triangle Island could be achievedby sequentialsampling in subcolonieswith a mobile radar unit. Unlike burrow-basedmonitoring and censustechniques,the radar techniquehas the potential to
be conductedquicklyand to be repeateddaily.However,the mostappropriate time frame for samplingneeds to be determined. The large increasein activitylevelsduring our studyalsounderscorethe sensitivityof
estimatesof apparentlyoccupiedburrowsto samplingdate. Radar mon-
156]
D. F. Bertramet al.
J.Field
Ornithol.
Spring 1999
itoring shouldbe readily transferablebetweendifferent observers.This is
a major advantageover burrow-basedschemeswhich often require that
countsare made by the sameperson in everyyear (e.g., Anker-Nilssen
and Rostad1993). Finally,radar is non4ntrusiveand doesnot causethe
disturbanceand possibledestructionto breedinghabitatwhichhavebeen
identified as significantdrawbacksof burrow-basedmethods (see above).
Our resultsstronglysuggestthat radar technologycan be a usefultool
in seabirdpopulationmonitoringand researchprograms.Note that the
radar technique could also be applied to monitoring flight activities
around large coloniesof surface-and cliff-nestingseabirdswhich are too
large for countingindividuals.On Triangle Islandwe will investigatefactorswhich affectcolonyattendancepatternsto developa populationindex based on the radar techniquesoutlined herein. Long-term radar
monitoringwill thus facilitatethe developmentof a time seriesdata set
which will be coupledwith traditional demographicstudiesto evaluate
the causesof population changesand to predict future trends.Suchtime
seriesinformation will be fundamental for evaluatingthe effectsof largescaledecadaland interdecadalchangesin marine ecosystem
production
on seabirdpopulations.
ACKNOWLEDGMENTS
Discussions
with G. Kaiser(CWS) prompted this study.L. Dill loaned equipment,D. Otto
located'Snappy'and B. Robertsonprovidedadvicefor imageprocessing.
Thanksto UTHSCSAfor placing Image Tool on the web. Natural Sciencesand EngineeringResearchCouncil (NSERC) of Canadagrantsto F. Cooke,A. E. Burgerand a NSERCCollaborativeResearch
and Developmentgrant to F. Cooke and T D. Williams supportedthe research.Logistic
support and additional funds were providedby G. Kaiser.A. Harfenist, H. Knechtel and M.
Lemon kindly providedunpublisheddata. Thanks to C. Smith for assistance
with data analysis.
LITERATURE
CITED
ANKER-NILSSEN,
T. ^ND O. W. ROSTAD.1993. Censusand monitoring of puffins Fratercula
arcticaon Rost, N Norway, 1979-1988. Ornis Scandinavica4:1-9.
, ERISTAD,
K. E., ANDS.-H. LORENSTEN.
1996. Aims and effort in seabirdmonitoring:
an assessment
basedon Norwegiandata.Wildlife Biology2:17-26.
BERTRAM,
D. F. 1995.The rolesof introducedratsand commercialfishingin the declineof
Ancient Murrelets on Langara Island, British Columbia. Cons.Biol. 9:865-872.
, ANDD. W. NAGORSEN.
1995. Introduced rats, Rattusspp., on the Queen Charlotte
Islands:implicationsfor seabirdconservation.CanadianField-Naturalist109:6-10.
BURGER,
A. 1997. Behaviorand numbersof Marbled Murrelets measuredwith radar.J. Field
Ornithol.
68:208-223.
COOPER,
B. A., R. H. D^¾,R.J. RITCHIE,ANDC. L. CP,^NOR.
1991.An improvedmarine radar
systemfor studiesof bird migration.J. Field Ornithol. 62:367-377.
D^¾,R. H., •tND B. A. COOPER.1995. Patterns of movement of Dark-rumped Petrels and
Newell's
Shearwaters
on Kauai. Condor
97:1011-1027.
EMSLIE,S. D., R. P. HENDERSON,
ANDD. G. AINLEY.1990. Annual variation of primary molt
with age and sex in Cassin'sAuklet. Auk 107:689-695.
G^s•:o•, A. J., I. L. JO•ES,^ND D. G. NO}•LE.1988. Monitoring Ancient Murrelet breeding
populations.ColonialWaterbirds11:58-66.
H•us, M.P. 1984. The Puffin. T&A D Poyser,Staffordshire,United Kingdom. 224 p.
M•'•uw^L, D. A. 1978. Criteria for aging Cassin'sAuklet. Bird Banding49:157-161.
--,
•tNOA. C. THORESEN.
1993. Cassin'sAuklet (Ptychoramphus
aleuticus).No. 50 in A.
Vol.
70,No.2
RadarMonitoring
ofSeabirds
[ 157
Poole and F. Gill, eds. The birds of North America.Academyof Natural Sciences,Philadelphia and American Ornithologists'Union, Washington,D.C. 18 pp.
PETERMAN,
R. 1990. Statisticalpower analysiscan improve fisheriesresearchand management. Can.J. Fish.Aquat. Sci. 47:2-15.
RODWAY,
M. S. 1991. Statusand conservationof breeding seabirdsin British Columbia.ICBP
Technical
--,
Publication
No.
11:43-102.
M.J.F. LEMONANDG. W. KAISER.1988. British Columbia SeabirdInventory: Report
#1.-EastCoast MoresbyIsland. Technical Report SeriesNo. 50. Canadian Wildlife Service, Pacificand YukonRegion, BritishColumbia.
, --,
ANDK. R. SUMMERS.
1990. BritishColumbiaSeabirdcolonyinventory:Report
#4-Scott
Islands.
Census results from
1982 to 1989 with
reference
to the Nestucca
oil
spill. Technical Report SeriesNo. 86. CanadianWildlife Service,Pacificand Yukon Region, British Columbia.
, W. A. MONTEVECCHI,
ANDj. w. CHARDINE.1996. Effectsof investigatordisturbance
on breedingsuccess
of AtlanticPuffins(Fraterculaarctica).BiologicalConservation76:
311-319.
S^VXRD,J.-P. L., AND G. E. j. SMITH. 1985. Comparison of survey techniques for burrow
nesting seabirds.Canadian Wildlife Service ProgressNotes No. 151.
THOMAS,L. 1996. Monitoring long-term population change:why are there so many analysis
methods?Ecology77:49-58.
VERMEER,K., R. A. VERMEER,K. SUMMERS,AND R. R. BILLINGS. 1979. Numbers and habitat
selectionof Cassin'sAuklet Breedingon Triangle Island, BritishColumbia.Auk 6:143151.
WALSH,P.M., D. J. HALLEY,M.P. HARRIS,A. DELNEVO,I. M. W. SUM,ANDM. L. TASKER.
1995. Seabirdmonitoringhandbookfor Britain and Ireland.JNCC/RSPB/ITE/Seabird
Group, Peterborough,United Kingdom.
Received18 Nov. 1997;accepted11 May 1998.