Demography and reproductive ecology of great frigatebirds

2006·5.7 MB·English

by  Donald C Dearborn

DEMOGRAPHYAND REPRODUCTIVE ECOLOGY OF
GREAT FRIGATEBIRDS
BY
DONALD C. DEARBORN' ANDANGELAD. ANDERS-
ABSTRACT
Frigatebirds {Fregata spp.) differfrom most Pacific seabirds in fundamental
ways, making it difficultto include themin generalizations aboutseabirdmanagement.
We present demographic data on great frigatebirds {F. minor) on Tern Island, French
Frigate Shoals, in 1998-2000. In terms ofmating attempts, males were more likely than
females to tryto obtain a mate butwere much less likely to succeed atpairing, andthe
variation in pairing success was greaterformales than females. Althoughfledging
success was high (63.5%), hatching success was below 30% in all three years ofthis
study. Males endtheirparental care offledglings soonerthan do females, butparental
care by both sexes extends into the pair-formation portion ofthe nextbreeding season.
Plumage data indicate that females do notbreed in years following a successfial breeding
attempt. Formales, the findings are less clear; some males may simultaneously feed 1-
year-old offspring andtendnew nests. In colony-wide counts offrigatebirds, we detected
as many as 1,171 males, 1,053 females, and 691 juveniles on the Island at atime. We
estimatedthat in 1999 there were 2,099 males and 1,615 females thatnested, out ofa
pool ofapproximately 11,195 males thattriedto attract a mate and 1,809 females that
evaluatedpotential mates. Because additional birds did nottry to breed atall, the total
numberofadults in the population is largerthanthis. Using amplified fragment length
polymorphism (AFLP) genetic markers, we found no evidence ofspatial genetic structure
withintheTern Island colony, confirmingprevious work showing genetic variation
between, butnotwithin, breeding colonies.
INTRODUCTION
Biological conservation requires abasic understanding ofthe life history and
demography offocal species (Meffe and Carroll, 1994), and sound managementmay be
— —
especially important andproblematic for species such as seabirds that are clustered
into areas ofhigh density duringbreeding. Frigatebirds {Fregata spp.) are colonially
breeding seabirds forwhich detailedreproductive and demographic data are generally
lacking formostparts oftheirrange (Metz and Schreiber, 2002), andgeneralizing
'DepartmentofBiologyandPrograminAnimalBehavior,BucknellUniversity,MooreAve., Lewisburg,PA
17837USA,E-mail: [email protected]
-DepartmentofBiologyandPrograminEcology,PennsylvaniaStateUniversity,321 MuellerLab,
UniversityPark,PA 16802USA
160
from other seabird species maybe unwise, given that frigatebirds are extreme among
seabirds in their sexual dimorphism (Dearborn et al. 2001a), low wing loading (Metz and
Schreiber, 2002), thermal-dependentflight (Weimerskirch et al., 2003), shortpairbonds
(Nelson, 1975), and long parental effort (Nelson, 1975).
Ourprevious work with great frigatebirds {F. minor) breeding onTern Island
(23° 45'N latitude, 166° 17'Wlongitude), Hawaii, has shown thatthesebirds arevery
long-lived (individuals over40 years ofage; Juola etal., in press). They regularly
move thousands ofkilometers between islands, yet they retain broad-scale genetic
differentiation (Dearborn et al., 2003). Theytypically exhibit a very male-biased sex
ratio atthe breeding colony (Dearborn et al., 2001a), inbreed slightlywhen choosing
mates (Cohen and Dearborn, 2004), and onlyrarely exhibit extra-pairpaternity (Dearborn
et al., 2001a). Finally, reproductive success offrigatebirds onTern Island is tiedto the
body condition ofthe breeding adults (Dearborn, 2001).
Here, we presentnew data from 1998, 1999, and 2000 to address fourbasic
aspects ofreproductive ecology in this Tern Islandpopulation ofgreat frigatebirds. First,
we compare the pairing success ofmales and females. Second, we quantify reproductive
success at monitorednests. Third, we compare the duration and frequency ofbreeding
ofmales andfemales. Andfinally, we explorethe size and structure ofthe breeding
population, basedon direct counts, mark-resightdata, and spatial analysis ofAFLP
genetic profiles ofbreeders.
METHODS
Pairing Success ofMales andFemales
To assess breeding attempts and pairing success ofmales and females, we banded
and wing-taggedunpairedbirds atthe startofthe 2000breeding season bycapturing
them in the breeding colonywhile theywere perched in shrubs atnight. Between 23
January2000 and 30 January2000, we tagged 79 males, 76 ofwhich were inbreeding
plumage, and 54 females, 51 ofwhich were in breedingplumage. Subsequently, we
surveyed the breeding colony three times per day (at 0900, 1330, and 1700) from 23
January through 15 May 2000 to record the reproductivebehaviorofthese marked
individuals. An individual was categorized as attempting to mate ifamale's gular
pouchwas inflated orifa female was performing an inspection ofa displaying male,
eitherwhile making stereotypical low-altitude inspectionflights orwhileperched in
contactwith a displayingmale. Fortaggedbirds that eventually builtnests, contents of
nests were monitored daily. Based directly on these observations oftaggedbirds, we
calculated the proportion ofdisplaying males that obtained a mate andthe proportion of
mate-evaluating females that obtaineda mate.
161
Reproductive Success
We measuredhatching success in 1998, 1999, and 2000 by making daily or twice-
dailychecks ofindividually marked nests from the start ofegg laying (typically early
February) through early summer(July in 1998 and 1999, May in 2000). In 1999, nests
were then followedbi-weekly through Decemberto measure fledging success.
Duration ofBreeding Cycle and Breeding FrequencyAcrossYears by Males and Females
Duringthe egg-layingperiod, fromthe end ofJanuarythrough May orJune, many
adults are still feeding 1-year-old offspring from the previous breeding season. Although
these 1-year-olds can fly, they still rely on colony visits by theirparents for most oftheir
food. We recorded all opportunistic observations ofadults feeding 1-year-olds fromthe
first ofJanuaryto early July in 1998 and late January to early May in 1999 and inferred
differences inthe duration ofparental care from temporal patterns in the proportion of
feedings madebymales versus females. During these feeding observations, we also
recorded plumage coloration ofthe provisioning adult. Comparing theirplumage to
that ofbirds currently incubating eggs orbrooding new chicks allows an assessment
ofwhetherthepool ofadults caring for 1-year-olds is different fromthe pool ofadults
with new nesting attempts. We assessed the plumage ofall males that were incubating
eggs orbrooding young chicks onTern Island on 30 March 1999, and we did the same
forfemales on 2April 1999; we then compared the distribution ofplumage coloration in
these groups withthat ofbirds thatwere feeding 1-year-old offspring in 1999.
BreedingPopulation Size and Spatial Genetic Structure
In 1998 and 1999, we conducted daily counts ofthe frigatebird colony by
m
following aregular survey route that took us within 50 ofall individuals on the Island
(see Dearborn et al., 2001a). Duringthese surveys, we counted individuals in three age
classes (1-2 yearoldjuveniles, 3-6 year old subadults, and adults), two sex classes
(foradults only), andthree location classes (perched, on nest, andflying). The number
ofbirds onnests atany giventime is only aminimum countofthe numberofnesting
attempts andnumberofbirds attempting to nestthatyear, because the majority of
nests fail. Tobetterestimate the numberofnesting attempts and the number ofadults
participating inthese attempts, we combinedestimates offledging success atthose nests
thatwere monitored during the 1999 season, the numberofchicks fledged across the
entire colonythat season (as revealed in acensus made inAugust), and the frequency
ofnesting attempts by markedmales andfemales. We calculatedthe total numberof
nesting adults inthe colony as (#fledged)/(estimate ofreproductive success)x(mean #
nest attempts formarkedmales and females). Last, we combined this calculation with
ourmeasurement ofpairing success formales and females, to yield an estimate ofthe
total numberofindividuals that attempted to attract a mate (inthe case ofmales) or
choose amate (inthe case offemales) in thatbreeding season.
162
Ourprevious research on this population had suggested an absence ofstrong
spatial genetic structure to the breeding colony, as assessedwith multilocus minisatellite
fingerprinting (Cohen and Dearborn, 2004). Here, weuse a separate data set to assess
the robustness ofthis finding. In an analysis ofpopulation genetic structure amongTern
Island, JohnstonAtoll, and Christmas Island (Kiribati), we analyzedAFLPdata from 117
polymorphic loci (Dearborn et al., 2003), finding significant differentiation among the
three islands. Here, we use Spatial Genetic Software v. 1.0c (SGS; Degen et al., 2001)
to test for spatial autocon'elation amongthe Tern Island samples. Foreach bird thatwas
sampled, we mapped its breeding location on a coordinate grid ofthe Island and then
used SGS to generate eight sets ofpairwise combinations ofbirds whose nests fell into a
particularcategory based on physical distance between the two nests. In this manner, we
made sets ofall pairs ofbirds whose nests were within 50 m ofeach other, within 50-100
m m
ofeachother, within 100-150 ofeach other, and so forthup to a 350-400m category.
Within each distance category, SGS computes the mean ofthe genetic dissimilarity
between each possible pairofsampledbirds, using Tanimoto distance for dominant
markers such asAFLPs. Plotting the mean genetic dissimilarity ordered across the eight
distance categories tests whetherthere is spatial genetic structure to the population. A
1000-run Monte Carlo pemiutationtestwas usedto generate confidence intervals forthis
relationship.
RESULTS
Pairing Success ofMales andFemales
Ofthe 76 tagged males thatwere in breedingplumage, 64 (84.2%) attempted
to attractamate at some point during the 2000 breeding season, but only 12(18.75%)
succeededin pairing. Fourofthe 12 had a nestwith a female but no egg, and an egg
was laid atthe remaining eight nests. None ofthe markedmales nestedtwice withinthat
breeding season.
Ofthe 51 tagged females thatwere inbreeding plumage, 28 (54.9%) evaluatedthe
pool ofdisplaying males at some point during the breeding season, and 25 (89.3%)) nested
(5 ofthe 25 had a nest with a male butno egg, and 20 laid an egg). Ofthe 25 females
thatnested, 8 nested more than once in the season (followingnest failure), including 1
that laid an egg on three differentbreeding attempts within one season.
Individual males were thus much more likely to try to mate than were females
(84.2% vs. 54.9%o; X- = 13.132, df= 1, p = 0.0003), but the males that triedto attract
amate were only one-fourth as likelyto succeed as females (18.75% vs. 89.3%); X- =
40.307, df= 1, p < 0.000001). Among those birds that didnest, females were more likely
to nest again afternest failure than were males (Fisherexactp = 0.036). Forthose birds
that triedto acquire amate, there was more within-sex variation in pairing success for
males than forfemales (fornumberofnesting attempts: male CV= 2.098, female CV=
0.601, and95% confidence intervals do not overlap).
163
Reproductive Success
Hatching success was 24.7% (45 of 182 nests) in 1998, 28.4% (74 of261 nests) in
1999, and 23.5% (4 of 17 nests) in 2000. Using those nests with definitively known lay
dates, there was no seasonal change in hatching success in 1998 (logistic regression, n =
118 nests from 25 February to 4 June 1998: Wald X- = 0.889, df= 1, p = 0.346). In 1999,
therewas also no significant seasonal change in hatching success (logistic regression,
n = 231 nests from 7 February to 30 May 1999: Wald X- = 2.672, df= 1, p = 0.102).
Fledging success was 63.5% (47 of74 nestlings that hatched) in 1999; thus, overall
reproductive success was 18.0% (47 fledglings from 261 nests).
Duration ofBreeding Cycle and Breeding FrequencyAcrossYears by Males and Females
We recorded 373 feeding events to 1-year-olds in 1998 and 374 feedings to 1-
year-olds in 1999. In January and February, during the early part ofthe breeding season,
roughly one-third to one-halfoffeedings to 1-year-olds was madeby males (Fig. 1). As
the new breeding season progressed, however, male efforttapered off, such that nearly all
feedings observed inApril, May, and June were made by females. This clearly indicates
a difference in the duration ofparental effort by males andfemales, though it does not
address whethertheircare fora 1-year-old nestling prohibits them from attempting to
startanew nestthat same season.
Figure 1. Feedingsto 1-year-oldgreatfrigatebirdfledglingsmadebymalesandfemalesonTernIslandin
1998and 1999. Totalsamplesizewas373 and374feedingobservationsin 1998and 1999.respectively.
Male plumage varied primarily in breast coloration, ranging from blackto gray
to brown. Based onfive categories ofbreast coloration, we foundthatthe males feeding
1-year-old offspring in 1999 were more likely to be brown than were males on new nests
ormales trying to attract amate in thatyear, and males trying to attract a mate were more
likelyto have substantial amounts ofwhite in thebreastplumage than males who
164
100
1
90
60-
„ 80
^50-
I 70
^40- 72 1 60
o i 50
|30-
o 1 40
S so- |iP 1 % 30
^
20
lo- r
10
I _ 1
nU 71 n
' 1 i
ack rayislack rayis rown rown lack hite
Figure2. Plumageofgreatfrigatebirdsfeeding 1-year-oldfledglingsandbirdsengagedinnewbreeding
attemptsonTernIslandin 1999. a)Breastplumagecategoriesofmalesthatwerefeeding 1-year-olds(open
bars)ordisplayingtoattractamate(hatchedbars)ortendinganeweggorchick(solidbars), b)Head
plumagecategoriesoffemalesthatwerefeeding 1-year-olds(openbars)ortendinganeweggorchick
(solidbars).
«
hadalreadybeen chosen as mates that season (X- = 222.837, df= 8, p 0.00001; Fig.
2). This whiterbreast plumage is likely indicative ofmales that arejustreaching sexual
maturity (Metz and Schreiber, 2002).
Female plumage varied primarily in head coloration, ranging from blackto light
brown. Based on categories ofhead coloration, we foundthat females feeding 1-year-
old offspring in 1999 were almost exclusively black-headed,«whereas no females on new
nests that yearwereblack-headed (X- =488.343, df=4, p 0.00001; Fig. 2).
Breeding Population Size and Spatial Genetic Structure
Daily counts in 1998 and 1999 revealed as many as 1,171 males, 1,053 females,
and 691 juveniles and subadults on the Island at a single time (Fig. 3). In bothyears,
the numberofjuveniles and subadults was fairly constant overtime. In contrast, the
total numberofadults on the Islandfluctuated greatly and generally increased all the
way through the pair-formation and egg-laying portion ofthe breeding season, even
though the rate ofpairing declineddramatically in May andJune. By July, mostnests
had failed orhad reached ages when chicks are unattended exceptwhen being fed, and
most ofthe adults on the Island were not engaged in reproductive activity. We previously
(Dearborn et al., 2001a) described a seasonal shift in the ratio ofmales to females thatare
unpaired andpotentially available formating. This ratiobecomes less biasedbecause of
a gradual increase in the numberoffemales on the Island (Fig. 3). Here, we apply the
plumage criteria described in the previous section to assess whetherthese females are
likelybreeders. Plumage-specific daily counts offemales were conducted in 1999 only.
From January through March of1999, most ofthe females on the Islandhadblack head
plumage, indicative ofcurrent breeders; females with brown ormottled heads were rare,
perhaps because they were on the Island only while feeding 1-year-olds. As the number
165
2000- a A
A^^^
1500- A
-
l*l
1000-
^
500^ %
^^
0-
r
1 1 1 1 1 1 1 1 I I
r3e
>
C
0)
E
<Q. Sro —T3
n
CNj CN CM CN ^ CO CD
400
: e Figure 3. Daily population counts of great
^ frigatebirds on Tern Island in 1998 and 1999. (a)
T
300- Total adults (open triangles), immatures (open
circles), and active nests (solid dots) in 1998. (b)
Males (open diamonds) and females (filled circles)
200- that were not on nests in 1998. (c) Total adults
(open triangles), immatures (open circles), and
active nests (solid dots) in 1999. (d) Males (open
100 diamonds) and females (filled circles) thatwerenot
on nests in 1999. (e) Females in black-headed
breedingplumage(filledsquares)orvariousbrown-
headed non-breeding plumages (open circles) in
Qo(1) 4—c(30 uS_I ^irI_a <iQ1_- ^1>(0, n—c313 1 1999.
CCDM CM rsi r^ 00 CD
166
offemales on the Island increased inApril, a decreasing proportion offemales were
black-headed, meaning that most ofthe "new" females on the Island were non-breeders
(Fig. 3e).
In an Island-wide census in lateAugust 1999 (USFWS, unpublished data),
we found 378 nestlings. Because there was no mortality at individually marked nests
between late July and the fledging ofthe last chick in December, andbecause no
nestlings at markednests fledgedbefore late September, it can be assumed that 378 is
a good estimate ofthe number offrigatebirdnestlings thatfledgedonTern Island in
1999. Applying our estimate ofreproductive success at markednests during thatyear
(18.0%), there were approximately 2,099 nests initiated during the 1999 breeding season.
Among those birds thatwere marked at the start ofthe 2000 season (the only yearfor
which we have such data from early-marked females) andthen initiatednests (i.e., laid
an egg) that year, the mean numberofnests initiatedper female was 1.30; formales,
the mean numberofnest attempts was 1.00. Ifthese numbers are relatively constant
across years, the population offrigatebirds actually nesting onTern Island in 1999 likely
consisted ofapproximately 1,615 females (2,099 nests with eggs / 1.30 eggs per female)
and2,099 males. This estimate can be combinedwith ourmeasure ofpairing success
forfemales (89.3%) andmales (18.75%)) to estimate the numberofadults thatattempted
to breed (i.e., including those that did and did not reachthe stage ofnest building). By
this approach, the pool ofbirds attempting to breed in 1999 consisted ofapproximately
1,809 females (1,615 /0.893) and 11,195 males (2,099 / 0.1875). Including those adults
notbreeding in 1999 (whetherat sea, on other islands, oron Tern Islandbut notcurrently
breeding), the total number ofsexually mature adults in the Tern Island population is
likely even larger, given thatplumage evidence (see above) suggests that individuals
whose nests succeed are likely to skipbreeding forat least the following year.
Spatial analysis ofAFLP datarevealed no significant change inpairwise genetic
dissimilarity between breeders across the categories ofdistance between nest sites, as the
confidence interval forthe Tanimoto dissimilarity index within each distance category
spannedthe overall mean of0.3447. Thus,AFLPdata suggestno spatial genetic structure
within theTern Islandbreeding colony, consistent with the finding ofvery little structure
based on the multilocus minisatellite fingerprinting data (Cohen and Dearborn, 2004).
DISCUSSION
This study sheds light onbasic demographic processes in this population. We
foundthat in agiven yearmales were much more likely than females to try to obtain a
matebut were much less likelyto succeed at pairing. This difference inpairing success
between the two sexes is consistentwithbehavioral descriptions ofmate acquisition
(which seems to consistentirely offemale choice; Nelson, 1975) and with amale-biased
operational sex ratio (Dearborn et al., 2001a). Females, butnotmales, occasionallynest
multiple times in a season ifthe firstnest fails; this difference between sexes is due to the
difficultythatmales have in attracting amate, ratherthan to lack ofinterest in re-nesting
by males. Overall, the coefficient ofvariation inpairing success was much greater for
167
males than females; this provides evidence that sexual selection is strongeron males in
this population and supports the notion that the male-biased operational sex ratio is linked
to sexual selection via variation in pairing success.
Following pairformation and egg laying, nesting attempts had a very low
probability ofsuccess. Although fledging success was high, hatching success was
below 30% in all three years ofthis study. Frigatebird reproductive output is affected
by El Nifio conditions at colonies further south in the Pacific (Schreiber and Schrciber,
1989), but ourthree years with low success included an El Niiio year(1998), a La Niiia
year(1999), and ayearofmoderate Southern Oscillation Index. Overall reproductive
success was towards the low end ofthe range reported from otherpopulations (Metz and
Schreiber, 2002), though hatching success on Tern Island was somewhat higher in 2003
than in previous years (Juolaand Dearborn, in press). Nests failed during incubation
foravariety ofreasons, but three main causes seem to be aggressive interactions with
otherfrigatebirds, severe weatherevents, and prolonged foraging trips by the mate that
is currently offthe nest (Dearborn, 2001). Aggressive interactions could reduce nest
success offrigatebirds in a density-dependent manner(Reville, 1988; 1991), though we
do notyet have such data forTern Island. Prolonged foraging trips by a breeding bird
are importantbecause thebody condition ofthe frigatebird currently incubating becomes
a limiting factor in its ability to stay on the nest and continue fasting (Dearborn, 2001).
This, coupled with analyses ofother seabird populations on Tern Island (Dearborn et
al., 2001b), suggests that resource availability is a key component ofindividual- and
population-level reproductive success forthese birds.
Plumage data indicate that females do notbreed in years following a successful
breeding attempt; the plumage ofthose females feeding 1-year-old offspring was
categorically different from that offemales incubating eggs ornew nestlings. Formales,
the findings are more complex. Males feeding I-year-olds were much more likely to have
brownbreastplumage than males with eggs ornew nestlings, butthere was substantial
overlap between the two groups ofmales. Eitherplumage is a less discriminatory
indicatorofbreeding status in males than in females or some males are simultaneously
feeding 1-year-olds and tending new nests. Forboth sexes, featherwear in the absence
ofmolt is the likely mechanism by which birds feeding 1-year-olds are brownerthan
birds involved in new breeding attempts, as brown is the basal coloration ofblack-tipped
head andbreast feathers in great frigatebirds (Metz and Schreiber, 2002). Overall, these
observations are consistent with the long-standing hypothesis that male frigatebirds try
to breed annually and females biennially (Stonehouse and Stonehouse, 1963; Diamond,
1973; Nelson, 1975; Trivelpiece and Fen-aris, 1987; Carmona et al., 1995), butthe
evidencepresentedhere is indirect. Note also thatthis hypothesis ofunequal breeding
frequencies ofmales and females was drivenby observations ofmagnificent frigatebirds
[Fregatamagnificens), in which males abandon the care ofnestlings afterjust a few
months (Osomo, 1999). We have shown in this study thatmales in ourpopulation do
taperofftheirparental care offledglings soonerthan do females, but male care extends
well into the pair-formation part ofthe next breeding season.
Colony-wide counts detected as many as 1,171 males, 1,053 females, and
691 juveniles on the Island at a single time. However, the frequent turnoverofbirds.
168
revealed by mark-resight data, indicates thatthe true numberofadults using the Island
is many times largerthan this. Individuals vary extensively in theirpattern ofIsland
use (unpublished data), such that some individuals visit onlybriefly, others stay for
weeks, and yet others come and go regularly overthe course ofmany months. Similar
complexities were seen in patterns ofvisits to otherislands (e.g. JohnstonAtoll) by
frigatebirdsthatwere wing-tagged on Tern Island (Dearborn et al., 2003). In lightof
these complexities in space use, thebestway to define apopulation may be based on
breeding individuals. Using a combination ofnest counts and reproductive metrics,
we estimatedthat 1,615 females and2,099 males nested onTernIsland in 1999; this
relatively small difference in numberofbreeders ofthe two sexes reduces the effective
population size only slightly (from 3,714 to 3,651 in a given year; Kimura and Crow,
1963). Manyadditional birds attempted tobreedbutdid notprogress beyondthepair-
formation stage. Based on the pairing success ofmarked individuals, we estimatedthat
the pool ofbirds attempting to breed onTern Island in 1999 was larger, particularly for
males: 1,809 females and 11,195 males. Because plumage evidence suggests thatbirds
are unlikely to breed in successiveyears, at least in years following thefledging ofa
chick, the actual size ofthe breeding population maybe even larger. Our estimate ofthe
numberofbirds breeding on Tern Island is substantially higherthan the previous estimate
forthis population (300- 375 pairs; Metz and Schreiber, 2002), andmayreflectmore
exact information oran increasing population, orboth; population increase in French
Frigate Shoals is suggested by nest counts overthe past40 years (summarized in Cohen
and Dearborn, 2004). Given that the global population ofgreat frigatebirds has been
estimated as only 54,000-68,000 breedingpairs (Metz and Schreiber, 2002), theTern
Islandpopulation maybe a demographically significant one.
UsingAFLP genetic markers, we foundno evidence ofspatial genetic structure
within the Tern Island colony. This is consistent with ourunderstanding ofthe historyof
thispopulation and with ourprevious findings with multilocus minisatellite fingerprinting
(Cohen and Dearborn, 2004). The lack ofsmall-scale spatial structure alleviates the
needto account forthis in localizedmanagement decisions, although ourearlierfinding
ofsubstantial genetic differentiation amongTern Island, JohnstonAtoll, and Christmas
Island (Dearborn et al., 2003) is important.
Acrucial gap in ourknowledge offrigatebird ecology in theNorthwestern
Hawaiian Islands is foraging movements and destinations, particularlygiven the
connection between reproductive success and adultbodycondition (Dearborn, 2001).
Diet samples offrigatebirds in Hawaii show apreponderance offlyingfish (Exocoetidae)
and squid (Ommastrephidae; Hamson et al., 1983), butnot knowingwhere the birds are
foraging makes it difficultto assesspotential threats to food availability. Oneplausible
threat is the commercial fisheiy for large predatoryfish, because a reduction in the
numberofsuchfishcould reduce the frequency withwhich frigatebirds'prey are driven
tothe surface (Safina and Burger, 1985). Stocks oflargepiscivorous fishhave declined
markedly, both globally (Myers andWorm, 2003) and inthe central Pacific (Cox et al,
2002). The ecological interactionbetweenpredatoiy fish and frigatebirds is a critical one
because frigatebirds neitherdive norswim aftertheirownprey. Furthermore, because
frigatebirds soar inthemials extensively when traveling (Weimerskirch etal., 2003), they