SHORT
COMMUNICATIONS
Carbon Isotope Ratio of FeathersRevealsFeedingBehaviorof Cormorants
HIROSHI MIZUTANI, • MICHIO FUKUDA,2 YUKO KABAYA,1 AND EITAROWADA•
•Laboratory
of Biogeochemistry
andSociogeochemistry,
Mitsubishi
KaseiInstituteofLifeSciences,
11 Minamiooya,Machida,Tokyo194,Japan,and
2UenoZoological
Gardens,
Ueno,Daito-Ku,Tokyo110,Japan
A colony (ca. 2,000 birds) of the Great Cormorant
(Phalacrocorax
carbo)is present in Shinobazunoike
Pond, Tokyo, Japan.The cormorantsfeed in rivers
and in Tokyo Bay.The major direction of their daily
flight in winter is to the rivers,and in summerto the
Bay (Fukuda1985).The relative importanceof the
two feeding sitesin terms of dietary contributionis
unknown. Dietary studiesconventionallydependon
stomachcontents,castpellets,and field observations.
Althoughusefulto revealbirds'choiceof preyspecies,
such details cannot determine the relative dependenceof cormorantson the two feedingareas.Stable
carbonisotopeanalysisof bird tissuesis a convenient
way to measuresuch dependence,as it tells if the
organic carbon in the tissuesis terrestrial or marine
in origin. Comparedwith conventionaldietarystudies, isotopeanalysisis lesssusceptibleto day-by-day
variation of prey speciesand is in principle free of
numerical bias for easily identifiable foods.
The two stableisotopesof carbonare x2Cand •3C.
In variousbiogeochemicalreactions,they reactat different rates, and the ratio •3C/•2C of various carbon
reservoirs becomes different (Mizutani and Wada
1985a).For instance,organiccarbonproducedby C3
plants that constitutethe major terrestrialproducer
differsby ca.200/•in this ratio from atmosphericCO2.
In marine ecosystems,primary carbon fractionation
occursduring planktonicphotosynthesis
and the resuitant organiccarbonis ca. 70/•higher than the terrestrial organic carbon. This difference is passed
through the food chain, and tissuesof consumersof
tive dietary carboninputs to the birds. Feathersare
readily availablein the field without sacrificeof birds.
The isotopevaluesof feathersrepresentthe diet during their growth period, which enablesthe recognition of seasonalchanges.Each feather reflectsthe
diet of an individual bird. Finally, the breedingseasonof the cormorantsat ShinobazunoikePondbegins
in Septemberand endsin the following June(Fukuda
1980),and the shed feathersare available all year.
To relate the isotoperatiosof feathersto diet, isotope fractionationduring the assimilationof organic
carbon into the body tissue must be considered
(DeNiro and Epstein1978,Minagawaand Wada •985).
We studiedthe fractionationof carbonisotopesinto
feathersof captive cormorantson a constantdiet of
mackerel(Pneumatophorus
japonicus).
The number of
captivecormorantsvaried, but the averagewasca.30.
In 1975 and 1976,we capturedbirds that were less
than one-month-old. Since then, they have eaten a
constant diet of mackerel.
Mackerelsand newly shedprimary feathersof the
captivebirds were collectedperiodicallyfrom May
1984 to June 1989.The sampleswere treated (Mizutani and Wada 1988) and combustedto form CO2 (Mi-
zutani and Wada 1985b).The CO=gaswasintroduced
for ratiometryto either a Hitachi RMU-6R massspectrometerwith dual inlet and doublecollectorsystems
or a Finnigan MAT-251 massspectrometerwith dual
inlet and triple collectorsystems.The carbonisotope
ratio was expressedin o,• deviation (•aC) from PDB
carbonate standard, which is a Cretaceous belemnite
both terrestrial and marine organic carbonswill have
(Belemnitellaamericana)from the Peedee Formation of
a carbonisotoperatio intermediatebetweenthe two
South Carolina, where
sources. Thus, the measurement of the ratio for tissues
may elucidatequestionsof the origin of carboncom•'C(0/•)03C/•2C)
......- (•3C/•=C)PøB
x 1,000.
(•3C/ •=C
poundsin nature.
There have been a number of successfulapplicaThe carbonisotopedatafrom the Hitachi RMU-6R
tions of isotopeanalysisto determine relative pro- were correctedfor •70. Resultsfrom the triple-colportion of terrestrialand marine organicmattersin lectingFinnigan MAT-25! were comparedwith those
varioussamples(e.g.Sackett1964,Tauber1981,Wada from the Hitachi RMU-6R to modify the correction
et al. 1987).Hobson(1987),the first to apply isotope equationof Craig(!957). Workingstandardsof carbon
analysisto avian dietary studies,measuredthe carbon were calibratedagainstU.S. National Bureauof Stanisotoperatio of bone collagenof gulls (Larusspp.)to dardsisotopereferencematerial No. 20 and No. 21.
estimate the relative contributions
of marine and terThe •3C values for the two working standardswere
restrialprotein in their diets.Moors et al. (1988)used -19.40/• and -12.0%o. Standard deviations of the carthe carbonisotoperatio to identify the desertedcol- bon isotopemeasurementswere lessthan 0.10/•.
onies of the RockhopperPenguin (EudypteschrysoKeratin, the majorcomponentof avian feathers,has
come).
a high sulfurcontent.Because
the high sulfurcontent
We chosefeathersas samplesto estimatethe rela- may interfere with completecombustionof feather
400
The Auk 107:400-437.April 1990
April 1990]
401
ShortCommunications
TABLE1. The •3C values of food items regurgitated by Great Cormorants.
•3C of organic carbon(%0)
No. of
measurements'
Food items
Roach
9
Immature gray mullet
Dragonet
4
3
Flathead
5
Goby
7
Gizzard shad
Flat fish
1
6
ϖ SD
-23.5
-24.0
-17.3
-18.6
-17.6
-15.7
-18.3
ñ
+
ñ
+
ñ
Range
1.4
2.0
0.4
0.7
1.6
-26.6
-27.4
-17.8
-19.3
-20.2
+ 0.5
-19.0
to
to
to
to
to
-to
-21.6
-22.7
-16.9
-17.3
-15.7
-17.6
Eachfish was analyzedonly once.
samplesand produce incorrect •3C values, we performed an incomplete combustionof L-cystine (lot
#274, Amino Acid Kit, Kyowa Hakko KogyoCo. Ltd.,
Tokyo) and feathers of the Nankeen Night Heron
(Nycticoraxcaledonicus).
For both samples,measurementswere within 0.1%oof completecombustionvalues.The high sulfur content in feathersdid not seem
mullet (the lowest in •3Ccontent) was the only food
for a cormorant,•3C of its feather should on average
be -20.0%0. Similarly, the •3C of its feather would
be -11.7%oif a cormorantdependssolely on the gizzard shad, which has the highest •3C values. The
to interfere
on the combination
with
an accurate
measurement
of the iso-
tope ratios.
The •3C for the whole
bodies of the mackerel
was
-20.1 + 1.2%o(19 samples),and for the featherswas
-16.1 _+0.6%0(11 samples).Therefore, the apparent
enrichment
of •3C relative
to diet in cormorant
feath-
erswas 4.0%0.The enrichment factor is only apparent
becausethe food item (i.e. mackerel)is composedof
variousorganiccompounds(including lipids) whose
individual enrichment factorsmay be different.
When
wild
cormorants
returned
to Shinobazu-
noike Pond after feeding, they sometimesregurgitated their stomachcontents.Regurgitateditems included a roach(Carassius
auratus),an immaturegray
mullet (Mugil cephalus),
a dragonet(Repomucenus
richardsonii),a flathead (Suggrundus
meerdervoorti),
a goby
(Acanthogobius
fiavimanus),
a gizzard shad (Konosirus
punctatus),
and a flatfish (Limandayokohamae)
(Fukuda
1985).As mentionedabove,regurgitationsonly partly
reflectthe actualdiet. They indicatethe utilization of
both marine and rivefine resourcesby cormorants.
Therefore, we collectedregurgitatedmaterial during May 1984and April 1989,froze them immediately, and treatedthem in the samemanner asthe mackerels (Table 1).
actual
5•C
values
for cormorant
feathers
should
be
between these two extremes,and the values depend
of these food items.
We collected naturally shed primary feathers of
adult cormorantsin the pond from May 1984 to May
1989. Feathers
were
treated
in the same manner
as
the feathersof cormorantsin captivity.Thesefeathers
representa maximumof 46 individualsout of ca.2,000
cormorants in the pond. Their average fi•3C was
-16.0%o, and it indicated an uneven utilization
of
terrestrial and oceanic resourcesby the cormorants
(Fig. 1). If the roach and the immature gray mullet
represent the terrestrial organic carbon sourceand
the other fish are oceanic,the cormorantsin the pond
get on average ca. six tenths of their organic carbon
from Tokyo Bay and the rest from rivers.
The distribution
of 8•C in the feathers of individual
birds was broad (Fig. 1). The range for feathers(9.1%o)
was wider than that for foods (8.3%0).Becausecarbon
in feathersis likely to representthe diet of individuals
only during feather formation, the distribution suggeststhat seasonalor individual variation in diet is
maximal.
The broad distribution of •C does not appear to
result from seasonaldifferences of the flight direction. Of the 46 feathers (Fig. 1), 18 were collected
when the cormorantswere flying mostly to the Bay
The •3C valuesfor the dragonet,flathead, goby,
gizzardshad,and flatfishindicatetheir marine origin.
The roach and the immature gray mullet are from
freshwater (Schoeningerand DeNiro 1984). The average•3C for oceanicfish (-23.8%0) was 6.3%0lower
when they were feeding in the rivers (from 15 December to 15 April). If adult cormorants normally
shed their feathersapproximatelyone year after formation, feathers collected in summer are likely to
than for the rivefine fishes (- 17.5%o).The difference
reflect summer diet, and those collected in winter re-
is similar to the differencebetweenprimary producersof the marine and terrestrialorganiccarbons,and
the two averagescanbe assumedto representthe ratio
flect winter
for the two food sources.
When cormorants feed on these fish, carbon in their
feathersshould be enriched in the heavy isotopeby
ca. 4.0%0relative to the fish. For instance,if the gray
(from 15 June to 15 October), and 14 were collected
diet. The mean
•3C
value
for the feathers
was -16.3 ñ 2.6%0 in summer, and -17.1
_+ 2.6•o in
winter. The averagefor summer was higher than that
for winter (as expectedfrom the seasonalflight direction); however, the difference of 0.8%0is small and
statisticallyinsignificant. Although the life span of
feathers may vary considerably, the broad distribu-
402
ShortCommunications
[Auk,Vol. 107
12
(513C
(%o)
Gray mullet.
Roach ß
FIsthesd.
FIstfish.
Goby,
Dragonetß
Gizzard
shad.
Fig. 1. Frequencydistribution of •3C in cormorantfeathers (n = 46). The •3C values for the foods are
indicatedon the horizontal axis.The length of horizontal portionsof the arrowsin the lower half represents
the enrichment of 4.0c/•obtained from feathersof cormorantson a constantdiet. The positionsof R and M
are at expected•3C valuesof feathersof cormorantsthat feed only on the rivefine (R) or the marine (M) fish.
The asteriskindicatesthe average•C for the 46 cormorantfeathers.
tion was not a compositeof two seasonallydifferent
distributions.
I suggestthe differencemight resultfrom the particularfeedingbehaviorof the cormorants.Therewere
four featherswith •3C below the averageof the two
rivefine fishes,and eight with •3C greaterthan that
of the five oceanic fishes. If we assume these feathers
are from cormorantsthat preyedat the time of feather
formation upon rivefine fish and marine fish, respectively,a few hundred cormorantsout of the total
of 2,000 would have depended mostly on rivefine
resources,whereasapproximatelyone fifth of the total dependedon marine resources.The rest, approximately three quartersof the total, dependedon both
in various proportions. Many cormorants were observedto move daily between rivers and Tokyo Bay
in their searchfor prey. I believe that the majority of
the cormorants
mustalternatetheir daily feedingsites.
Black-tailed Gulls (Laruscrassirostris)at Kabushima,
Aomori Prefecture, Japan, present a different situation (Mizutani and Wada 1988). The average•3C of
available foodsthere ranged from -22.8c/• (sardine)
to - 17.8ø/•(larvaeof Japansoldierfly) (range:5.0ø/•).
The range of •C values for seven gull featherscollected in 1983 was 0.7c/•. The narrow
distribution
in
gull featherssuggestsa narrow variation in their diet.
An additional 17 gull featherscollectedin 1988 and
1989 increasedthe •3C range of the 24 feathers to
1.5c/•with little changein either the averageor the
standarddeviation(Fig. 2).
When Fig. 2 is comparedwith Fig. 1, the narrow
distributionis apparent.Presumablythe gulls feed
heavilyon a singlefoodstuffor eachindividualselects
its food resourcesin a similar way. An alternative
explanationis that gulls molt more synchronously
than the cormorants.Further study is necessaryto
understand the cause of the narrow distribution,
but
the diet of the cormorantsappearsfar more variable
than that of the gulls.The cormorantsare apparently
more individualistic in feeding behavior than the
gulls.
The application of carbon stable isotopesto ornithology and ecologyis not without some complications (Fry 1988). With caution, however, the technique could be of value for ornithology. My results
yield valid informationconcerningboth diet content
and feeding behavior. Combinationwith other elements such as nitrogen would further enhance the
value of the stableisotopetechnique.
The Ueno ZoologicalGardenskindly allowed us to
April 1990]
ShortCommunications
403
20-
L-
.16
T
Sardine.
Squid.
Anchevy.
Mysids.
Japansoldierfly.
Fig. 2. Frequencydistribution of •3C in gull feathers(n = 24). The •3C of various foodsand the extent
of the •3Cenrichmentis indicatedon the horizontal axis.The length of horizontal portionsof the arrows
representsthe enrichmentof 4.0• obtainedfrom gulls on a constantdiet (Mizutani et al. unpubl. data).
collect samples.Kiichi Narita of the Hachinohe Sci-
MIZUTANI, H., & E. WADA. 1985a. Carbon dioxide
ence Museum for Children collected and donated some
and the biosphere.Viva Origino 13: 23-49.
& --.
1985b. Combustion of organic
samplesby infrared furnace for carbon isotope
analysis.Anal. Blochem.146:90-95.
--,&
. 1988. Nitrogen and carbonisotope
ratios in seabird rookeriesand their ecological
implications.Ecology69: 340-349.
gull feathers. We thank Kyoko Karasawafor assistance with
some of the measurements.
LITERATURE CITED
CRAIG,H. 1957. Isotopic standardsfor carbon and
oxygen isotopes.Geochim. Cosmochim.Acta 3:
53-92.
DENIRo,
M. J.,& S.EPSTEIN.
1978.Influence
•f diet
--,
MOORS, P. J., T. W. $PEIR, & G. L. LYON. 1988. Soil
analysesand •3C/•2Cratios identify sites of deserted RockhopperPenguin colonies.Auk 105:
796-799.
on the distributionof carbonisotopesin animals.
SAOCE•r,W.M. 1964. The depositionalhistory and
Geochim. Cosmochim. Acta 42: 495-506.
isotopicorganic carbon compositionsof marine
FRY,B. 1988. Food web structureon GeorgesBank
sediments.Marine Geology2: 173-185.
from stable C, N, and S isotopiccompositions. SCI-IOENINGER,
M. J., & M. J. DENIRO. 1984. Nitrogen
Limnol. Oceanogr.33: 1182-1190.
and carbonisotopiccompositionof bonecollagen
FU•CUDA,M.
Wild
1980. Life of the Common Cormorants.
Birds 45: 679-684.
ß 1985.
Notes on the Common
marine
Cosmochim.
Cormorants
in
TokyoßAnimals and Nature 15: 11-15.
HOBSON,K. A. 1987. Use of stable-carbonisotope
analysisto estimatemarine and terrestrial protein contentin gull diets. Can. J. Zool. 65: 12101213.
MINAGAWA,
M., & E. WADA. 1985ß Biogeochemical
implication of H, C, N, O stableisotopecompositions in plant and animal. Chikyukagaku 19:
39-52.
from
and terrestrial
animals.
Geochim.
Acta 48: 625-639.
TAVBER,H. 1981. •3C evidence for dietary habits of
prehistoric man in Denmark. Nature 292: 332333.
WADA, E., M. MINAGAWA, H. MIZUTANI, T. TsuJI, R.
IMA•ZV•I, & K. KARASAWA.
1987. Biogeochemical studieson the transport of organic matter
along the Otsuchi River Watershed,Japan. Estuarine
Coastal Shelf Sci. 25: 321-336.
Received31 May 1989,accepted20 September
1989.