A MINIATURE
ACTIVITY
PLUNGE-DIVING
RECORDER
FOR
SEABIRDS
DAVID J. ANDERSON,l'• PAUL R. $IEVERT,1
JOHN ANDREWS-LABENSKI,
2 AND ROBERTE. RICKLEFS
•
aDepartment
of Biology,
LeidyLaboratory,
Universityof Pennsylvania,
Philadelphia,
Pennsylvania
19104-6018USA, and
2Department
of Psychology,
Universityof Pennsylvania,
Philadelphia,Pennsylvania19104-6196USA
A•STRACr.--Wepresentdesignspecifications
for a self-containedactivity recorder,developed for use on plunge-diving seabirds.Compressedintegrated circuitry and miniature
componentsallow constructionof recorderssmall enough(28 g) to be mountedon the tails
of boobies and other seabirds. The cost is moderate.
Field tests of the recorder
on Brown
Boobies(Sulaleucogaster)
demonstratedits ability to measureforagingbehaviorat seathat
typically cannotbe observeddirectly. Received17 January1990,accepted
25 September
1990.
M•N¾ seabird speciesforage far from land,
and studiesof their activity away from terrestrial breeding coloniesare difficult. Although
seabirdscan be observed from ships, one usually doesnot know the individual's age,breeding status, and colony location. In addition,
tracking individuals from boats is difficult or
impossible.The ability of remote sensingde-
matically by the use of surface-mountcomponents, which lack most of the mass and volume
of conventional integrated circuits.
We describea 28-g battery-poweredCMOS
memory data recorder,that usessurface-mount
componentsand is designedto study foraging
activity of seabirds.We present initial results
of its use in the field. The basic concept of the
vices to record behavior
of seabirds has been
design was inspired, in part, by Mohus (1987).
used to overcometheseproblems.Radio trans- The recorder can store 8 kilobytes (kb) of data
mitters (Trivelpiece et al. 1986, Anderson and on a SRAM memory chip in 8-bit data bins in
Ricklefs 1987), time-activity recorders(Prince time intervals that range from 8 to 512 s for
and Francis 1984; Wilson et al. 1986; Cairns et
total collection periods of 1-48 days.The type
al. 1987a, b; Birt-Friesen et al. 1989), and inteof data collected depends on the sensor. We
grated circuit boards(Kooymanet al. 1983,Mo- recorded output from a miniature acceleromehus 1987, Croxall et al. 1988) have produced ter designed to detect plunge dives of boobies
information on activity at seathat complements (Sulidae).
studies
of the same individuals
and their
off-
spring at the breeding site.
GENERAL FEATURES OF OPERATION
The quality of informationgeneratedby these
deviceshasbeen limited primarily by their size.
The recorder'spresent configuration(Figs. 1
Large transmitters or recording devices may
causeflying birdsto alter their behavioror sub- and 2) detectsand recordsnumber of events (in
ject them to risk of injury (Perry 1981, Cacca- our application,plunge dives) per 64-sinterval.
mise and Hedin 1985, Wilson et al. 1986). Be-
When the bird enters the water, the bird and
cause smaller
attachedrecorder deceleratefaster than a slug
suspendedon a spring (Fig. 3). If the difference
units have either
fewer
or smaller
componentsor both, they permit less sophisticated applications. This constraint may account
for the underutilization
of available
in-
tegrated circuit technology in ornithology.
However,
the size of such units can be cut dra-
between
these two rates of deceleration
the bin starts a data collection
3Present address:Department of Avian Sciences,
University of California, Davis,California 95616-0690
USA.
257
is suf-
ficient, the suspended slug compressesthe
spring, strikes a contactpoint, and closesthe
recorder'scircuit. The resulting signal causes
the number in a data bin to increaseby 1; and
The number
accumulated
after
interval
at zero.
64 s is written
to a memory location.The slug may bounceand
strike the contact more than once per dive, but
The Auk 108:257-263. April 1991
258
ANVmSONETAL.
NOR:
VDD6 VSS CONNECTZDNS
ADDRESS
[Auk, Vol. 108
COUNTERS
A
8Kx8
SRAM
CLOCK
U3E
470K
T /
A¸
SENSOR
DATA
COUNTER
ACTZVZTY
FZGURE
RECORDER
I
Fig. 1. Schematicof recorder'scircuit. Seetext for details.
the circuitry recordsno additional signalswithin a 2-sinterval after the first. The springsused
in this model require < 1 s to equilibrate after
on5
•EZ•
u
/ol
o•*•
•
lull Io%
•o•OlL
o1-•
Data are uploaded from recordersto MS/PC
DOS-based portablecomputerswith a field-usable, battery-powered, RS-232 interface and
customsoftware.The resulting data setis a column of numbers, each of which is the number
•o1•1o•
•
compression.
Io e
F_• o IOL+
of signalssent by the sensorper 64-speriod, in
chronological order. Part of the recorder'swaterproof coating can be removed with a knife
to accessI/O pins, and the coating (we used
marine epoxy) can be restoredin the field. Recorders
can thus be reused
for the life of their
batteries, which can be replaced in the field
with a portable soldering iron.
Fig. 2. Circuit boardconfiguration,shown actual
size. Circular
batteries
are on the back side of the
board;other componentsare on the front side. Large
rectangular SRAM chip is mounted over some components on the front side.
TECHNICAL CIRCUIT DESCRIPTION
A schematicof the recorder'scircuit is given
in Figure 1; the actual layout chosenfor a par-
April1991]
Seabird
Activity
Recorder
259
ticular application will depend on the shape
desired. A full-sized illustration of the configuration that we used is shown in Figure 2. The
circuit is powered by two 3-V lithium cells BT1
pulse. Other time constantscan be chosen by
appropriateselectionsof R3 and C3, where T =
and BT2 connectedin series.Operatingcurrent
is lessthan 100/•A, and nominal battery life is
longer than 1,000 h at 100 /•A using 2,325 or
Each sensorinput advancesU5, a 12-bit binary counterwith its first 8 outputsconnected
2,032 cells (Panasonic1989).Jumper w1 is used
to power on the circuit and may be a 0.1" x
0.1" shorting jumper, twisted leads, or a magnetic reed switch. Schottky diode D1 prevents
a chargingvoltagefrom being applied to the
lithium
cells when the recorder
is connected
to
remote power (as during uploading of SRAM
contents).CapacitorC1 is for power supply bypassingand is mounted near chip U4 to minimize clocknoise.ResistorR19 and capacitorC8
filter current surgesduring memory writes and
are mounted near the Vccpin of the SRAM.
On power-up, the resetsof addresscounters
U1 and U2B are momentarilyheld high to zero
counters.Reset time is determined by resistor
R4, capacitor C2, and Schmitt inverter U3A.
When the charging voltage acrossC2 reaches
the trip level of U3A, the outputgoeslow, which
0.8RC.
times
The
time
maximum
constant
bounce
should
be at least 3
time.
to the data I/O of the memory. The data value
accumulatedin U5 is read into memory and
momentarily reset once per clock period. Data
input rate should not exceed255 countswithin
a single clockperiod. At count256,datacounter
U5 rolls over, returning the first 8 outputs to
zero. This processof binning data continues
once per clock period until memory is full at
address8,191. (Note: a clock of 64 s/pulse with
8,192 data bins equalsapproximately 145 data
collection hours.) At addresscount 8,192, Q1 of
U2B goeshigh, transferringthe memoryfrom
write to read mode and simultaneouslyturning
off the clock. The circuit is then quiescentwith
the SRAM
held
in data retention
mode.
Data
will remain valid until battery supply drops
below
2-2.5
volts.
COMPONENTS
enables the address counters, ca. 30 ms after
power is applied. Reset places the memory in
write mode by holding the write enable (WE)
All componentsexcept the sensor,batteries,
clockcrystal,and the memorychipare surfacemountdevices.All logicdevicesare CMOS.The
of the SRAM low through resistorR7 from the
Q1 output of addresscounterU2B. Data outputs
are disabledby pull-up resistorR8, which holds
the output enable (OE) high.
The circuit clockis a crystal-controlledPierce
oscillatorthat consistsof crystalX1, resistorsR1
memory chip is a low-power static random accessmemory (SRAM), organized 8K x 8-bits in
a standard28-pin plasticpackageand mounted
over the top of the surfacemount components.
The batteries, provided with solder tabs, are
and R2, and capacitorsC6 and C7. X1 is a stan-
mounted on the solder side of the board. I/O
dard 32768 Hz crystal driving U4 (a 24-stage
frequency divider with decoded2•8 to 224outputs), while resistorsR17 and R18 provide for
low-power operation of U4 (Motorola 1988).
Output is decodedat 22•,soa pulseis generated
every 64 seconds.When Q21 of U4 goeslow,
the falling edge--differentiated by capacitorC4
and resistor R5 and buffered by Schmitt inverters U3F and U3E--drives the chip enable
(CE1) low with a 2-3 /•s pulse to write to the
memory. This pulse is in turn differentiated by
connectionsare through a 16-pin edge connector plated directly on the board. The data
recorder was fabricated on a 0.062" (0.157 cm)
glass epoxy printed circuit board by Precision
Graphics,Raritan, New Jersey,and coatedwith
a 2-mm layer of marine epoxy, which can be
removed with a knife for battery replacement.
The fully assembledand coatedunit weighed
28 g in our application,and 18 g beforecoating.
SENSOR
capacitor C5, resistor R6, and Schmitt inverter
U3D, giving a 1-2/•s pulse to resetdata counter
U5 and advance address counters U1/U2B.
Data input occursby mechanicalcontactclosure of sensor S1. The time constant (T) of re-
sistor R3 and capacitor C3 forms a 2 s/0.5 Hz
debounce/low-passfilter. Schmitt invertersU3C
and U3D speedup and buffer the input closure
We used an accelerometer(Fig. 3) designed
to detect deceleration experienced by a plunge-
diving seabird as it enters the water. The recorder's circuit is closed momentarily when a
87-mgslug suspendedon a spring strikesa contact inside the cylinder of the spring. The contact can be screwedthrough the base of the
260
A•aDm•So•a
•r AL.
[Auk,Vol. 108
UPLOADING INTERFACE UNIT
To retrieve data stored in the recorder, we
built a battery-poweredRS-232interfaceutilizing a CY232, Parallel/Serial Interface device
(CyberneticMicro Systems1984).This interface
and associatedsoftwarewere designedspecif-
ically for usewith MS/PC DOS-basedcomputersasthe destination;designspecifications
can
be obtained by contactingAndrews-Labenski.
The
recorder
is connected
to
the
interface
through the 16-pin PC edge connector.Power
to the recorderis provided by the interfaceunit.
To avoidplacinga 50-#Aburden currenton the
lithium cells,jumperW1 mustbe removedafter
power is supplied by the interface. The interface transferred data at 2,400 baud, 7 data bits,
even parity, and 1 stopbit in ASCII Hex format.
A completeuploadingto a ToshibaT1000computer took approximately 13 minutes.
FIELD RESULTS
We field-tested30 recorderson Midway Island, Northwest Hawaiian Islands, and John-
stonAtoll in Juneand July, 1989.Eachrecorder
wasmountedon a breedingBrownBooby(Sula
leucogaster).
We used heavy thread and cyanoacrylate glue to tie each corner of the unit to
the ventral surface of one of four rectrices, which
Fig. 3. Accelerometerto detect plunge dives.
were positionedascloseaspossibleto the bird's
body.Andersonand Ricklefs(1987)give details
and photographsof the mounting method.All
of the birdswere feedingchicksand soreturned
to the nest site approximately daily, which allowed
retrieval
of recorders.
Our first field trial producedno databecause
the recorderswere coatedwith acrylic conforaccelerometerto narrow or widen the gap be- mal coatingfor electronics(Humisealtype 1B73)
tween it and the suspendedslug,sothat a lesser and salt water penetrated the circuit. Subseor greaterdeceleration,respectively,is required quent use of marine epoxy solvedthe problem,
to causecontact.The accelerometeris aligned but many recorderswere destroyedduring this
with its long axisparallel to the long axisof the initial trial.
bird, sothat the slugwill movetowardthe conFive recorders produced data on Johnston
tactonly during head-firstor tail-firstdives,and Atoll birdsfrom 14to 17July 1989(Fig.4). Three
not during side-to-sidemotion of tail during of the five recordersregistered"events" (acflight or preening.During testswith a wooden celerometertriggers)during spotcheckswhen
model of a diving booby entering a swimming we observedthe birds in the colony. This inpool, an accelerometerset to fire only after a dicated that the accelerometer detected some
drop of 2 m or more fired on 10/10 of drops other activities (possibly jumping and brief
from 2 m and 0/10 of drops from 1 m. We set flights)in addition to plunge dives.Thesenonall accelerometersat this sensitivity to avoid feedingeventswere infrequent,whereasevents
recording low-velocity landing dives.
away from the breedingcolony,which presum-
April 1991]
Seabird
ActivityRecorder
40
261
Brown Booby #2
©ooo
3O
2O
10
0
4O
Brown Booby #12
3O
2O
-+•
10
¸
0
(D
4O
¸
3O
b
2O
_C•
10
E
Z
Brown Booby #17
o
4o
o•
•
ßßOD
0
•
O0
Brown Booby #20
3o
2o
10
0
._1, .Ilid.,.. ß, . .I .., ,,,*• fil•ed
4o
Brown Booby #29
3o
2o
10
0
July 14
July 15
July 16
July 17
Fig.4. Activity
ofbreeding
Brown
Boobies.
Triangles
indicate
whentheactivity
recorder
wasattached
(V)andremoved
(V).Circles
indicate
absence
(O)orpresence
(0)ofthebirdduring
spotchecks
ofitsnest
site.Therecorder
attached
toBird20wasremoved
on20July,
buttheSRAM
chiphadfilledtocapacity
on
17 July.
ably indicatedplunge dives,occurredat rates foragedonceperdayon 6 complete
and8 parbetween10 and 40 divesper hour.
tial bird-dayswhen we collecteddata.
2. Two birds showed clear evidence of nocWemadeseveral
observations
fromthesepreliminary data:
turnalforagingonthenightof 14-15July1989.
1. Birdsforagedat leastonceper day.One Several workers have suspectedthat Brown
bird clearlymadetwo tripson oneday(Bird Boobiesforageat night (Nelson1978:501),and
29,Fig.4), butthe otherfourbirdsapparently bright moonlight(the lunar phasewasfull on
262
AuDisou œt^t,.
[Auk, Vol. 108
BrownBooby#17, JohnstonAtoll, July 15, 1989
u•
5
20
Hour of doy
Fig. 5. Fine-scaleactivitypatternof BrownBooby17, a breedingbird, recordedon 15 July 1989.The bird
was absentfrom its nest site during the 5-h period. Number of eventsare displayedper 64-s interval.
3. On an hourly basis,events during fishing
trips appearedgrouped in periods of 1- to 9-h
havior of free-ranging birds and other animals.
With appropriate sensors,these units could record such actionsas vocalizations,approaches
to electromagneticallymarked locations, and
duration (•' = 3.90 h, SD = 2.17, n = 20). Most
movement
trips appeared to contain only one group of
dives. In two instancesthat might contain two
groups,we cannotbe sure that the bird did not
return briefly to the colony between groups.
ties(light / dark, water / air, horizontal/ vertical).
18 July) may have facilitated round-the-clock
foraging.
4. Three
birds were known
across environmental
discontinui-
ACKNOWLEDGMENTS
to be absent from
Developmentand fabricationof theserecordersand
fieldwork were supportedby a grant from the Na29, 15-16 July; Fig. 4). None of these birds tional GeographicSociety.We are indebted to Fred
showed evidence of foraging, and we suspect Letterio and Walter Suplick for production of the acthat they roostedaway from their nest sitesall celerometer;to BobWeissmanand colleaguesat Prenight. If so, estimatesof foraging effort based cision Graphics,Raritan, New Jersey,for board proon presence or absenceat the nest site must duction; and to Steve Heard for assistanceduring
testing of accelerometerprototypes.Andrew Olson
accountfor this possibility.
wrote the down-loading software. W. Montevecchi
5. The temporal pattern of plunge dives was made helpful commentson the manuscript. Fieldconsistentwith that observedvisually in Masked work wasconductedunder permit from the U.S. Navy
(Sula dactylatra) and Blue-footed (S. nebouxii) (Midway Atoll), U.S. DefenseNuclear Agency (Johnboobies (Anderson and Ricklefs 1987), in which
ston Atoll), and U.S. Fish and Wildlife Service (Midbirds at fishing areasspent much of their time way and Johnston atolls). We thank Betty Anne
sitting on the water and watching for other Schreiber and Gary Schenk for field assistance.
the nest site during at least one night (e.g. Bird
plunge-diving birds. Feeding events were
clumped in setsinterspersedthroughout periodsof relative inactivity. Brown Boobiesexhibited this samepattern (e.g. Fig. 5), although we
do not know whether birdswere flying or floating when they were not diving. These two possibilities
neous
could
data
be differentiated
from
water-immersion
with
simultadetectors
(Cairns et al. 1987a).
This activity recorder combines small size,
ease of use in the field, and reasonable cost
(approximately$80.00 each),and it providesa
new tool to study frequency and timing of be-
ANDERSON,D. J., & R. E. PUCKLInuS.
1987. Radio-track-
ing Masked and Blue-footedboobies(Sulaspp.)
in the Gal•pagosIslands.Natl. Geogr.Res.3: 152163.
BIR?-FRIESEN,V. L., W. A. MONTEVœCCHI,D. K. C.•aRNS,
& S. A. MACleo. 1989. Activity-specificmetabolic rates of free-living Northern Gannetsand
other seabirds.Ecology70: 357-367.
CACCAM•Sœ,
D. F., & R. S. HœDIN. 1985. An aerodynamicbasisfor selectingtransmitterloadsin birds.
Wilson
Bull. 97: 306-318.
April1991]
Seabird
Activity
Recorder
263
MOTOROLA.1988. Pp. 6/297-302 in CMOS logic data,
DL131, rev. 1. Phoenix, Arizona, Motorola, Inc.
MONTEVECCHI.
1987a. Electronicactivityrecordersfor aquaticwildlife. J.Wildl. Manage.51:395- NELSON,J.B. 1978. The Sulidae. Oxford, Oxford Univ.
CAmNS, D. K., K. A. BREDIN, V. L. BIRT, & W. A.
399.
Press.
--,
& W. A. MONTEVECCHI. 1987b. Ac-
PANASONIC.1989. Pp. 33-34 in Lithium batteries
tivity
budgets
and
foraging
ranges
ofbreeding technical handbook,
Common
Murres.
Auk
104: 218-224.
CROXALL,
J.P., R. W. DAVqS,& M. J. O'CONNELL. 1988.
Diving patternsin relation to diet of Gentooand
Macaronipenguinsat SouthGeorgia.Condor 90:
157-167.
CYBERNETIC MICRO
lel/Serial
SYSTEMS.
Interface
1984.
Device
CY232
Manual,
BXEC-S0014. Secausus,New
Jersey,PanasonicIndustrial Co.
Paral-
3F:
CY232MAN.014. San Gregorio, California, Cybernetic Micro Systems,Inc.
PERRY,M. C.
1981. Abnormal behavior of Canvas-
backsequippedwith radio-transmitters.
J.Wildl.
Manage. 45: 786-788.
PRINCE,
P. A., & M.D. FRANCIS.1984. Activity budgetsof foraging Grey-headedAlbatrosses.Condor 86: 297-300.
TRIVELPIECE,
W. Z., J. L. BENGsTON,
S. G. TRIVELPIECE,
& N.J. VOLKMAN. 1986. Foraging behavior of
KOOYMAN,G. L., J. O. BILLUPS,
& W. D. FARWELL.1983.
Gentooand Chinstrappenguinsas determined
Two recently developedrecordersfor monitoring diving activity. Pp. 197-214 in Experimental
biology at sea(A. G. MacDonald and I. G. Priede,
bynewradiotelemetry
techniques.
Auk 103:777-
Eds.). New York, Academic Press.
MOHUS,I. 1987. A storing telemetry-transmitter for
recordingbird activity. Ornis Scandinavica18:
227-230.
781.
WILSON, R. P., W. S. GRANT, & D.C.
DUFFY. 1986.
Recording deviceson free-ranging marine animals: Does measurement affect foraging performance?Ecology 67: 1091-1093.