Cryptic dietary components reduce dietary overlap among sympatric butterflyfishes (Chaetodontidae)

Page 1

Journal of Fish Biology (2009) 75, 1123–1143

doi:10.1111/j.1095-8649.2009.02303.x, available online at www.interscience.wiley.com

Cryptic dietary components reduce dietary overlap amongsympatric butterflyfishes (Chaetodontidae)

I. Nagelkerken*†, G. van der Velde*‡, S. L. J. Wartenbergh*,M. M. Nugues§ and M. S. Pratchett‖

*Department of Animal Ecology and Ecophysiology, Institute for Water and WetlandResearch, Faculty of Science, Radboud University, Heyendaalseweg 135, P.O. Box 9010,6500 GL Nijmegen, The Netherlands, ‡National Museum of Natural History, Naturalis,P.O. Box 9517, 2300 RA Leiden, The Netherlands, §Leibniz Center for Tropical Marine

Ecology (ZMT), Fahrenheitstr. 6, D-28359 Bremen, Germany and ‖ARC Centre of Excellencefor Coral Reef Studies, James Cook University, Townsville, Queensland 4811, Australia

(Received 7 August 2008, Accepted 25 March 2009)

This study used three different methods to ascertain dietary composition for 21 Chaetodontidaespecies co-occurring on a single fringing reef in Derawan Island, East Kalimantan, Indonesia.The combination of in situ feeding observations, examination of gut contents and stable-isotopeanalyses was expected to identify previously unresolved prey items that contribute to dietary sep-aration. In situ feeding observations identified five species that feed predominantly on anthozoans(Chaetodon baronessa, Chaetodon bennetti, Chaetodon lunulatus, Chaetodon punctatofasciatus andChaetodon speculum). Stable-isotope ratios for these species, as well as for Chaetodon ornatissimus(for which no feeding observations were completed), were very similar and consistent with dietscomprising mostly anthozoans. Feeding observations, however, showed that they mostly fed ondifferent coral species, while the identifiable portion of their gut contents showed clear separationbased on cryptic dietary components. For example, C. baronessa and C. bennetti appeared to ingestannelid worms during the course of coral feeding, whereas gut contents of C. punctatofasciatus andC. speculum were dominated by crustaceans. In situ feeding observations further identified the fol-lowing groups: coral–bottom feeders, bottom feeders, sponge feeders and pelagic feeders, feedingon a wide variety of prey items such as Annelida, Crustacea, Cnidaria, Mollusca and macroalgae.Overall, many chaetodontid species had similar reliance on major prey items (e.g. anthozoans orpolychaetes) but differed greatly in the minor prey items that they utilized. Partitioning of minorprey items may be important in reducing interspecific competition and facilitating coexistence ofchaetodontids on coral reefs. © 2009 The Authors

Journal compilation © 2009 The Fisheries Society of the British Isles

Key words: coral reefs; feeding; fishes; gut contents; resource partitioning; stable-isotope analysis.

INTRODUCTION

The extraordinary diversity of fishes on coral reefs has generated considerable inter-est in mechanisms that facilitate coexistence of ecologically equivalent species (Sale,1974, 1978; Munday, 2004; Pratchett, 2005). Established competition theory suggests

†Author to whom correspondence should be addressed. Tel.: +31 24 3652471; fax: +31 24 3652409;email: [email protected]

1123© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles

Page 2

1124 I . NAG E L K E R K E N E T A L .

that species coexistence is achieved largely through resource partitioning (Colwell& Fuentes, 1975; Diamond, 1978). Currently, there is a major dichotomy betweenthose studies that have found segregation in resource use among sympatric coralreef fishes (Clarke, 1994; Wilson, 2001) v. those studies that find strong and con-sistent overlap in patterns of resource use (Munday, 2004; Gardiner & Jones, 2005;Pratchett, 2005). Complementarity in patterns of resource use is often viewed as evi-dence for niche-based competition (Schoener, 1974), though differential patterns ofresource use may arise due to contrasting preferences or morphological or physiolog-ical adaptations (Fulton et al ., 2001). Coexistence of ecologically equivalent speciesmay also occur in the absence of resource partitioning if ecological processes contin-ually act to interrupt the process of competitive exclusion, or when shared resourcesare unlimited (Connell, 1983). On balance, most studies recognize that competitionamong ecologically similar species is intense and has a major influence on patternsof resource use (Munday, 2004), even if it does not adversely affect diversity ofcoral-reef fish assemblages.

To avoid interspecific competition, coexisting fish species may partition resourcesalong three different axes: food, space and time (Schoener, 1974; Nagelkerken et al .,2006). Studies from a variety of ecosystems have shown that of these three factors,dietary separation contributes most to species coexistence (Ross, 1986). When foodis limiting and species compete for food, there is often clear segregation of dietaryresources (Sale, 1974). One notable exception is the high degree of dietary overlapreported for sympatric butterflyfishes (Chaetodontidae). The apparent lack of dietarypartitioning among sympatric chaetodontids, however, may be partly attributed to thelack of resolution in describing patterns of prey use (Pratchett, 2005). For example,many studies have reported very high dietary overlap among obligate coral-feedingbutterflyfishes (Harmelin-Vivien & Bouchon-Navaro, 1983; Bouchon-Navaro, 1986;Zekeria et al ., 2002) simply because they treat all scleractinian corals as a sin-gle distinct prey category. More recent studies have shown that most coral-feedingspecies specialize on specific (sets of) coral species (Cox, 1994; Berumen et al .,2005; Pratchett, 2005). Nonetheless, most corallivorous butterflyfishes tend to exhibithighly convergent feeding preferences and may exhibit very high (up to 73%) overlapin dietary composition (Pratchett, 2007). High dietary overlap occurs despite intenseinterspecific aggression (Berumen & Pratchett, 2006), as well as data showing thatchaetodontids are immediately and severely affected by declines in coral abundance(Pratchett et al ., 2006), implying that coral is limiting.

Many early studies on dietary composition of chaetodontids (Bouchon-Navaro,1986) tended to rely solely on gut-content analysis to establish the range and pro-portional consumption of different dietary items. Disadvantages of using gut-contentdata are that they are snapshots of the feeding habits and that most food items(especially corals) may be unidentifiable. Moreover, different dietary items may bedigested and assimilated at different rates, such that gut-content analysis can givemisleading data on relative consumption of different prey items (MacDonald et al .,1982; Livingston, 1984; Polis & Strong, 1996; Mariani et al ., 2002). Increased res-olution on dietary composition, especially for coral-feeding chaetodontids, may beobtained using in situ feeding observations (Berumen et al ., 2005; Pratchett, 2005).Simple observations of feeding behaviour, however, do not necessarily reveal exactprey items being consumed (Pratchett, 2005). For example, when a chaetodontid bites

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 3

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1125

on the surface of a coral, it might be biting at the coral tissue, ‘sucking’ coral mucusfrom within polyps or picking at micro-invertebrates living on the coral surface.

Stable-isotope analysis is a very useful tool to establish the probable source ofnutrition from a diverse array of prey types. Stable-isotope analysis has severaladvantages over short-term feeding observations and gut content analysis. Mostimportantly, stable-isotope analysis provides a history of food assimilated in terms ofweeks to months, depending on the consumer’s growth rate (Gearing, 1991; Hessleinet al ., 1993). It is therefore less subject to temporal variations in diet and accounts forall food items assimilated. One of the difficulties, however, is that different food itemssometimes show similar stable-isotope signatures (Bootsma et al ., 1996; Pinnegar &Polunin, 2000). Used in combination, in situ feeding observations, examination of gutcontents, and stable-isotope analyses may reveal cryptic dietary components that arefundamental in reducing competition for limiting food resources. Such studies, thatuse a combination of techniques to identify dietary composition for coral-reef fishes,are very rare (Pinnegar & Polunin, 2000; Davenport & Bax, 2002) and non-existentfor the Chaetodontidae. In this study, a combination of in situ feeding observations,examination of gut contents and stable-isotope analyses was used, in an attemptto further improve resolution of dietary composition for sympatric chaetodontids.The objective of the present study was to examine dietary composition and measuredietary overlap among 21 species of Chaetodontidae coexisting on a small fring-ing reef in Derawan Island, Indonesia. More specifically, the role of cryptic dietaryitems in differentiating diets and reducing potential competition among sympatricchaetodontids was explored.

MATERIALS AND METHODS

This study was conducted in October 2003 along a distance of c. 600 m on the southernand most sheltered reef at Derawan Island (2◦ 17′ 05′′ N; 118◦ 14′ 35′′ E), East Kalimantan,Indonesia. Derawan Island is part of a chain of small islands located in front of the BerauDelta in East Kalimantan (Fig. 1). The island is a sand cay (c. 1100 × 550 m) surrounded byan extensive fringing reef which extends up to c. 4·5 km from the cay. The coral reef extendsto a depth of 20 m, with mixed coral assemblages dominated by Acropora sp., Pocilloporasp. and Porites sp.

The broad range of dietary items consumed by each chaetodontid species found at DerawanIsland, as well as their proportional use, was ascertained from in situ feeding observations,following Reese (1975), Berumen et al . (2005) and Pratchett (2005). Observations were con-ducted for two to 10 individuals of each species, with five to eight observations for themajority of the species (Table I). Most chaetodontids continued to feed despite the presenceof divers, though some fishes (especially planktivores) did not exhibit any obvious feedingbehaviour throughout the course of feeding observations. During each feeding observation,an individual fish was followed for 3 min, while recording the total number of bites takenfrom Anthozoa (hard and soft corals), Hydrozoa, Porifera, carbonate pavement, dead coral,rubble, sand or any other conspicuous sessile flora or fauna.

In order to undertake stable-isotope analyses and to examine gut contents, a small number ofindividuals (two to nine) of each chaetodontid species were collected using spearguns. Allenet al . (1998) and Michael et al . (2004) were used for species identification. Furthermore,the identification of the caught specimens was checked during the stay at Derawan Island(G. R. Allen, pers. obs.). For each fish caught, fork length (LF) was measured and the com-plete gut removed immediately and preserved in 96% ethanol to avoid further digestion.Identifiable portions of the gut contents were divided into: (1) Annelida (Polychaeta and otherAnnelida), (2) Crustacea (Amphipoda, Copepoda, Cumacea, Decapoda, Isopoda, Ostracoda,

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 4

1126 I . NAG E L K E R K E N E T A L .

Berau Delta

Panjang

Samama

Sangalaki Kakaban

Maratua

RaburabuDerawan

Borneo

East Kalimantan

10 km

N

Fig. 1. The location of the study area Derawan Island in East Kalimantan, Indonesia.

and other Crustacea), (3) Bryozoa, (4) Cnidaria (Anthozoa and Hydrozoa), (5) Echinodermata,(6) Mollusca (Bivalvia and Gastropoda), (7) Porifera, (8) fish eggs, (9) macroalgae and (10)sediment. Diet composition was determined by estimating the relative volumetric quantity offood items, using a stereomicroscope (Nielsen & Johnson, 1992). Cnidaria (Hydrozoa andAnthozoa) were identified by the presence of polyps.

For stable-isotope analysis, a small piece of white muscle tissue was immediately removedfrom each fish, dried for 48 h at 70◦ C and ground to a fine powder using a mortar, pestleand liquid nitrogen. Only muscle tissue was used because of its slow turnover rate, resultingin a history of food assimilation over periods of months and excluding short-term variability(Gearing, 1991). Various potential food items of Chaetodontidae [see Fig. 2(b)] were col-lected at the study site and were treated in the same way as the fish muscle tissue. The hardcorals consisted of a mix of corals that were consumed by coral-feeding chaetodontids: Acro-pora cytherea, Acropora palifera, Favia sp., Galaxea fascicularis, Montipora sp., Physogyralichtensteini, Pocillopora damicornis, Porites sp. and Stylopora pistillata. The soft coralsconsisted of a mix of Chinonepthys sp., Dendronephthys spp., Junceella juncea, Muricellasp., Sarcophyton sp., Sinularia sp. and three unidentified species. To focus solely on organiccarbon, food items containing calcium carbonate (i.e. Halimeda algae, Crinoidea and somesoft coral species) were decalcified by adding drops of hydrochloric acid to the samples untilthe formation of CO2 gas bubbles stopped (Nieuwenhuize et al ., 1994); this was not done forthe nitrogen isotope analysis, since nitrogen content is affected by acid washing (Bunn et al .,1995). Carbon and nitrogen stable-isotope compositions were measured with a Carlo ErbaNA 1500 elemental analyser coupled online via a Finnigan Conflo III interface with a Ther-moFinnigan DeltaPlus mass-spectrometer (www.thermo.com). Carbon and nitrogen isotoperatios are expressed in the delta notation (δ13C, δ15N) relative to Vienna PDB and atmo-spheric nitrogen. Average reproducibilities based on replicate measurements of standards forδ13C and δ15N were c. 0·18 and 0·15‰, respectively.

Stable-isotope ratios in animals reflect those of their diet, with an average of c. 1‰ enrich-ment in δ13C (indicative of food source) and 3·5‰ enrichment in δ15N (indicative of trophiclevel) (DeNiro & Epstein, 1978; Rau et al ., 1983; Minagawa & Wada, 1984; Fry, 1988),although the variation in δ15N enrichment can sometimes be more substantial (DeNiro &Epstein, 1981; Minagawa & Wada, 1984; Mill et al ., 2007). Stable carbon and nitrogen isotope

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 5

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1127T

able

I.Pe

rcen

tage

ofbi

tes

(mea

n±

s.e.

)ta

ken

from

reef

subs

trat

afo

r20

spec

ies

ofC

haet

odon

tidae

obse

rved

atD

eraw

anIs

land

(ord

ered

acco

rdin

gto

decr

easi

ngco

ntri

butio

nof

Ant

hozo

ato

diet

).C

haet

odon

orna

tiss

imus

was

not

foun

d,so

feed

ing

obse

rvat

ions

coul

dno

tbe

mad

efo

rth

issp

ecie

s.G

rey

area

sin

dica

tem

ajor

feed

ing

subs

trat

a(p

erce

ntag

eof

bite

s≥2

5%).

The

last

colu

mn

(pla

nkto

nfe

edin

g)in

dica

tes

spec

ies

obse

rved

tofe

edon

zoop

lank

ton

(+).

For

feed

ing

onPo

rife

ra,

itco

uld

not

bede

term

ined

whe

ther

the

fishe

sfe

don

the

spon

getis

sue

itsel

for

onsm

all

epif

auna

livin

gon

the

spon

gesu

rfac

e

Fis

hsp

ecie

sN

umbe

rof

fish

Mea

nnu

mbe

rof

bite

sA

ntho

zoa

Hyd

rozo

aPo

rife

raC

arbo

nate

pave

men

tD

ead

cora

lR

ubbl

eS

and

Ann

elid

aA

lgae

Cya

noba

cter

iaPl

ankt

onfe

edin

g

Cor

alfe

eder

s

Cha

etod

onba

rone

ssa

535

100·0

±0·0

Cha

etod

onlu

nula

tus

620

95·2

±4·8

2·4±

2·42·4

±2·4

Cha

etod

onbe

nnet

ti5

1594

·4±

2·72·5

±2·5

2·2±

2·20·9

±0·9

Cha

etod

onpu

ncta

tofa

scia

tus

529

86·8

±13

·213

·2±

13·2

Cha

etod

onsp

ecul

um6

2066

·7±

21·1

33·3

±21

·1C

oral

–bo

ttom

feed

ers

For

cipi

ger

long

iros

tris

71

33·3

±33

·333

·3±

33·3

33·3

±33

·3C

haet

odon

klei

nii

815

32·0

±8·6

2·9±

2·127

·7±

11·9

7·5±

6·512

·8±

7·110

·3±

7·26·7

±6·7

+C

haet

odon

lunu

la8

125

·0±

25·0

75·0

±25

·0F

orci

pige

rfla

viss

imus

86

19·3

±9·9

1·4±

1·421

·8±

11·6

5·7±

5·738

·1±

15·4

13·7

±5·5

Cha

etod

onul

iete

nsis

78

9·9±

6·157

·6±

15·9

10·0

±10

·00·8

±0·8

4·4±

4·47·3

±7·3

10·0

±10

·0B

otto

mfe

eder

s

Cha

etod

onau

riga

512

3·7±

2·310

·0±

10·0

47·0

±16

·19·3

±6·1

27·3

±18

·62·7

±1·7

Cha

etod

onse

mei

on5

162·7

±2·7

16·1

±7·3

5·0±

5·038

·5±

13·7

20·0

±20

·016

·0±

16·0

1·6±

1·6C

haet

odon

ephi

ppiu

m5

1812

·9±

12·9

23·2

±11

·561

·0±

16·5

2·9±

2·9C

haet

odon

vaga

bund

us7

230·6

±0·6

1·6±

1·641

·1±

9·59·8

±6·2

25·7

±8·3

20·2

±8·0

1·0±

1·0H

enio

chus

chry

sost

omus

20

Hen

ioch

usva

rius

60

Spon

gefe

eder

s

Cha

etod

onad

ierg

asto

s6

310

0·0±

0.0

Cor

adio

nch

ryso

zonu

s5

1143

·6±

23·2

38·8

±19

·65·5

±5·5

5·5±

5·56·7

±6·7

Pela

gic

feed

ers

Hem

itau

rich

thys

poly

lepi

s10

0+

Hen

ioch

usdi

phre

utes

20

+

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 6

1128 I . NAG E L K E R K E N E T A L .

7

8

9

10

11

12

δ15N

(‰

)

Coral feedersCoral–bottom feedersBottom feedersSponge feedersPelagic feedersNo observations

(a)

C. semeion

C. ephippium

H. polylepis

Co. chrysozonus

H. diphreutes

C. kleinii

C. lunulatusC. ornatissimus

C. punctatofasciatusC. speculum

H. varius

C. bennetti C. baronessa

C. auriga

C. adiergastos

F. longirostris H. chrysostomus

F. flavissimus

C. vagabundusC. ulietensis

C. lunula

3

4

5

6

7

8

−21 −20 −19 −18 −17 −16 −15 −14 −13 −12 −11 −10

δ13C (‰)

Crabs(13)

Crinoids(2)

Fish larvae(4)

Fleshy green algae(2)

Hermitcrabs (5)

Hydroids (11)

Shrimps(9)

Soft corals (18)Sponges(16)

Tunicates(8)

Zooplankton (4)

Calcareous green algae(3)

Hard corals(9)

Polychaetes (11)

(b)

Fig. 2. Mean ± s.e. δ15N and δ13C values of (a) the 21 species of Chaetodontidae and (b) their potential fooditems collected from the coral reef of Derawan Island. Co, Coradion. Species in (a) are ascribed to fivedistinct trophic groups based on in situ feeding observations (see Table I). Numbers between parenthesesin (b) indicate sample sizes for collected food items.

ratios were compared separately among fish species. As variances were not homogeneous,even after transformation, a Kruskal–Wallis test was used, followed by a Games–Howellpost hoc test which is specifically designed for situations in which populations, variances andsample sizes differ (Field, 2006).

Diet breadth based on gut contents was calculated for each species using Levin’s stan-

dardized index (Hurlbert, 1978): Bi = (n − 1)−1[(∑

j p2ij

)−1 − 1

], where Bi = Levin’s

standardized index for consumer i, n = the number of food categories and pij = the pro-portion of the diet of consumer i that is made up of food item j . The index ranges from 0 to1, with low values indicating diets dominated by few food items.

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 7

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1129

Table II. Mean percentage bites taken by coral-feeding Chaetodontidae on different hard-coral species or genera. Grey areas indicate main coral species fed on (% bites ≥10%)

Chaetodonbaronessa

Chaetodonbennetti

Chaetodonlunulatus

Chaetodonpunctatofasciatus

Chaetodonspeculum

Acropora bottlebrush form 6 0 12 0 0Acropora corymbose form 0 0 3 5 0Acropora cytherea 3 0 0 0 0Acropora divaricata 6 0 10 0 0Acropora florida 0 0 0 2 0Acropora formosa 8 0 2 18 0Acropora gemmifera 0 0 1 0 0Acropora palifera 15 0 0 4 0Anacropora spp. 0 0 3 0 0Astreopora myriophthalma 0 0 8 0 0Cyphastrea serailia 0 20 0 0 5Echinopora lammelosa 0 0 0 4 0Favia–Favites spp. 0 10 0 0 5Galaxea fascicularis 25 0 0 0 53Goniastrea pectinata 2 2 4 6 3Goniopora spp. 0 4 0 0 0Lobophyllia spp. 0 8 0 0 2Madracis kirkyi 0 10 0 0 0Montipora spp. 24 13 24 17 13Pachyseris speciosa 0 0 16 11 0Physogyra lichtensteini 0 12 0 0 0Platygyra daedala 0 5 0 2 4Pocillopora damicornis 6 0 0 13 3Porites cylindrica 0 0 1 0 0Porites massive 0 0 7 4 0Porites tuberculosa 0 0 0 7 0Seriatopora caliendrum 2 0 0 4 4Seriatopora hystrix 0 0 0 0 4Stylophora pistillata 2 4 6 3 5Other scleractinian corals 0 11 3 0 0

Similarity in diet was calculated for all possible pair-wise comparisons between membersof the 21 chaetodontid species. Bray–Curtis similarity was calculated with the programmeBioDiversity Pro version 2 (McAleece, 1997), with high similarities indicative of high dietoverlap. The similarity indices were grouped in classes of 20% similarity. The analysis wasdone separately for diet data from feeding observations (see Table I) and for data fromgut-content analysis (see Table III, but including food items with <4% occurrence).

RESULTS

Based on the results from the in situ feeding observations (see Table I), the 21chaetodontid species studied could be classified into five different feeding groupswhich are treated below.

C O R A L F E E D E R S

In situ feeding observations identified five species, Chaetodon baronessa Cuvier,Chaetodon bennetti Cuvier, Chaetodon lunulatus Quoy & Gaimard, Chaetodon

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 8

1130 I . NAG E L K E R K E N E T A L .

punctatofasciatus Cuvier and Chaetodon speculum Cuvier, which fed predominantly(on average >66% of bites) on hard and soft corals. Three of these species, C.baronessa, C. bennetti and C. lunulatus, fed almost exclusively from the surface oflive corals, whereas C. punctatofasciatus, and C. speculum also took a substantialproportion of bites on Hydrozoa and carbonate pavement, respectively (Table I). Allfive coral-feeding chaetodontids took a considerable percentage of bites (13–24%)from Montipora sp., but otherwise consumed broadly different coral taxa (Table II).There were further differences among these species in minor prey items consumed,as identified in the gut contents.

Gut contents of putative coral-feeding chaetodontids (C. baronessa, C. bennetti,C. lunulatus, C. punctatofasciatus and C. speculum) were largely unidentifiable(Table III) and dominated by indistinct mucus, which may represent partially digestedcoral tissue or gastric juices produced by the fishes themselves. Consequently, it wasdifficult to reconcile volumes of coral slime against persistent dietary items, such aschaetae from polychaetes. Identifiable portions of their gut contents, however, showedclear separation based on cryptic dietary components. Interestingly, C. baronessa,C. bennetti and C. speculum appeared to ingest annelid worms during the courseof coral feeding (Table III). Gut contents of C. punctatofasciatus and C. speculumwere dominated by crustaceans, while C. lunulatus also consumed macroalgae to aconsiderable degree. Diet breadth, based on gut contents, ranged from 0·03 for C.baronessa, where the gut contents were dominated by coral tissue, up to 0·16 forC. punctatofasciatus, which consumed molluscs and a wide variety of crustaceans(Table III).

Chaetodontids showed significant differences among species for δ13C (Kruskal–Wallis test, P < 0·001) and δ15N (Kruskal–Wallis test, P < 0·001) in muscle tissue.For the coral feeders, no significant differences were found in δ13C and δ15N valuesamong species, except one case for each stable isotope (Table IV). When consideringan average enrichment of c. 1‰ in δ13C and c. 3·5‰ in δ15N between food andconsumer, stable isotope ratios of coral feeders were consistent with diets comprisingmostly corals and polychaetes (Fig. 2). Muscle tissue of C. lunulatus was enrichedin δ13C and depleted in δ15N, consistent with a diet partly based on macroalgae(Fig. 2). Also for Chaetodon ornatissimus Cuvier, which is considered a hard-coralfeeder (Cox, 1994; Berumen et al ., 2005; Pratchett, 2005), δ15N values were depletedand its diet consisted considerably of macroalgae (Fig. 2 and Table III).

C O R A L – B OT T O M F E E D E R S

In situ feeding observations identified five chaetodontid species, Chaetodon kleiniiBloch, Chaetodon lunula (Lacepede), Chaetodon ulietensis Cuvier, Forcipiger lon-girostris (Broussonet) and Forcipiger flavissimus Jordan & McGregor, which fedpartly on corals (10–33% of bites) or hydrozoans (58%), and partly on hard substrata(i.e. carbonate pavement, rubble or dead coral; Table I) or on sponges.

Whilst feeding on hard substratum, these chaetodontids appeared to be target-ing either polychaete worms or crustaceans, as revealed by gut content analysis(Table III); stable-isotope analysis supported a diet based mainly on crustaceans (pos-sibly shrimp and hermit crabs, Fig. 2). Various species showed a significant depletionin δ13C and enrichment in δ15N values compared to coral feeders (Table IV), sug-gesting utilization of additional food sources and less dependence on coral as a food

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 9

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1131

Tab

leII

I.R

elat

ive

gut

cont

ent

(bas

edon

lyon

iden

tifiab

leco

nten

ts)

show

nfo

rm

ajor

diet

com

pone

nts

(>4%

volu

me)

for

the

21sp

ecie

sof

Cha

etod

ontid

ae;

grey

area

sin

dica

teim

port

ant

food

item

s(c

onte

nt≥2

0%).

Num

ber

ofsa

mpl

esfo

rgu

tco

nten

tan

alys

is,

%gu

tco

nten

tth

atco

uld

beid

entifi

edan

ddi

etbr

eadt

h(b

ased

onid

entifi

able

gut

cont

ents

)ar

eal

sosh

own.

Fish

spec

ies

are

orde

red

follo

win

gTa

ble

I

Num

ber

ofgu

tco

nten

tsa

mpl

es

%gu

tco

nten

tun

iden

tifia

ble

Lev

in’s

diet

brea

dth

Ann

elid

aC

rust

acea

Bry

ozC

nida

ria

Ech

ino

Mol

lusc

aPo

rife

raE

ggs

Alg

aeSe

d

Poly

chO

ther

Am

phi

Cop

eC

uma

Dec

aIs

oO

stra

Oth

erA

ntho

zH

ydro

zB

ival

Gas

tro

Cor

alfe

eder

sC

haet

odon

399

0·03

2575

baro

ness

aC

haet

odon

598

0·09

205

520

50lu

nula

tus

Cha

etod

on4

970·1

025

1513

45be

nnet

tiC

haet

odon

298

0·04

1767

17or

nati

ssim

usC

haet

odon

499

0·16

2513

2513

25pu

ncta

tofa

s-ci

atus

Cha

etod

on3

990·1

125

833

33sp

ecul

umC

oral

–bo

ttom

feed

ers

For

cipi

ger

542

0·00

98lo

ngir

ostr

isC

haet

odon

586

0·29

78

2518

79

87

klei

nii

Cha

etod

on4

570·0

188

13lu

nula

For

cipi

ger

558

0·01

9010

flavi

ssim

us

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 10

1132 I . NAG E L K E R K E N E T A L .

Tab

leII

I.C

onti

nued

Num

ber

ofgu

tco

nten

tsa

mpl

es

%gu

tco

nten

tun

iden

tifia

ble

Lev

in’s

diet

brea

dth

Ann

elid

aC

rust

acea

Bry

ozC

nida

ria

Ech

ino

Mol

lusc

aPo

rife

raE

ggs

Alg

aeSe

d

Poly

chO

ther

Am

phi

Cop

eC

uma

Dec

aIs

oO

stra

Oth

erA

ntho

zH

ydro

zB

ival

Gas

tro

Cha

etod

on4

990·0

375

1313

ulie

tens

isB

otto

mfe

eder

sC

haet

odon

385

0·05

5836

auri

gaC

haet

odon

46

0·00

98se

mei

onC

haet

odon

655

0·04

1870

ephi

ppiu

mC

haet

odon

984

0·06

606

22va

gabu

ndus

Hen

ioch

us7

950·0

284

14ch

ryso

stom

usH

enio

chus

597

0·14

2140

79

10va

rius

Spon

gefe

eder

sC

haet

odon

694

0·04

7025

adie

rgas

tos

Cor

adio

n6

910·2

09

2514

2514

6ch

ryso

zonu

sPe

lagi

cfe

eder

sH

emit

au-

986

0·15

3639

6ri

chth

yspo

lyle

pis

Hen

ioch

us5

980·0

735

751

diph

reut

es

Poly

ch,

Poly

chae

ta;

Oth

er,

othe

rA

nnel

ida;

Am

phi,

Am

phip

oda;

Cop

e,C

opep

oda;

Cum

a,C

umac

ea;

Dec

a,D

ecap

oda

(mai

nly

shri

mps

);Is

o,Is

opod

a;O

stra

,O

stra

coda

;O

ther

,ot

her

Cru

stac

ea(u

nide

ntifi

able

);B

ryoz

,B

ryoz

oa;

Ant

hoz,

Ant

hozo

a;H

ydro

z,H

ydro

zoa;

Ech

ino,

Ech

inod

erm

ata;

Biv

al,

Biv

alvi

a;G

astr

o,G

astr

opod

a(>

25%

pela

gic

Pter

opod

a:T

heco

som

ata

and

Gym

noso

mat

a);

Alg

ae,

mac

roal

gae;

Sed,

sedi

men

t.

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 11

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1133T

able

IV.

Pos

tho

cco

mpa

riso

n(G

ames

–H

owel

l)of

stab

leca

rbon

(low

er-l

eft

side

)an

dni

trog

en(u

pper

-rig

htsi

de)

isot

ope

ratio

sam

ong

the

21ch

aeto

dont

idsp

ecie

s.Fi

shsp

ecie

sar

eor

dere

dfo

llow

ing

Tabl

eI

Cor

alfe

eder

sC

oral

–bo

ttom

feed

ers

Bot

tom

feed

ers

Spon

gefe

eder

sPe

lagi

cfe

eder

s

Chaetodonbaronessa

Chaetodonlunulatus

Chaetodonbennetti

Chaetodonornatissimus

Chaetodonpunctatofasciatus

Chaetodonspeculum

Forcipigerlongirostris

Chaetodonkleinii

Chaetodonlunula

Forcipigerflavissimus

Chaetodonulietensis

Chaetodonauriga

Chaetodonsemeion

Chaetodonephippium

Chaetodonvagabundus

Heniochuschrysostomus

Heniochusvarius

Chaetodonadiergastos

Coradionchrysozonus

Hemitaurichthyspolylepis

Heniochusdiphreutes

Cor

alfe

eder

sC

haet

odon

baro

ness

a—

∗∗N

SN

SN

SN

S∗∗

NS

NS

∗N

SN

SN

SN

SN

S∗∗

NS

∗∗

NS

NS

Cha

etod

onlu

nula

tus

NS

—N

SN

SN

SN

S∗∗

∗∗∗

NS

∗∗∗

NS

∗∗∗

NS

NS

∗∗∗

∗∗∗

NS

∗∗∗

∗∗∗

∗∗∗

∗∗C

haet

odon

benn

etti

NS

NS

—N

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SC

haet

odon

orna

tiss

imus

NS

∗N

S—

NS

NS

∗N

SN

SN

SN

SN

SN

SN

SN

S∗

NS

NS

∗N

S∗∗

Cha

etod

onpu

ncta

tofa

scia

tus

NS

NS

NS

NS

—N

S∗∗

∗∗

NS

∗∗N

S∗∗

NS

NS

∗∗∗∗

∗N

S∗∗

∗∗∗

∗∗∗

Cha

etod

onsp

ecul

umN

SN

SN

SN

SN

S—

∗∗N

SN

S∗

NS

∗N

SN

S∗

∗∗N

S∗

∗N

S∗

Cor

al–

bott

omfe

eder

sF

orci

pige

rlo

ngir

ostr

is∗

∗∗∗

NS

∗∗∗

∗∗∗

∗∗∗

—∗

NS

NS

NS

NS

∗∗∗

∗∗∗

NS

NS

∗∗∗

NS

NS

NS

NS

Cha

etod

onkl

eini

i∗∗

∗∗∗

∗N

S∗∗

∗∗∗

∗∗∗

∗N

S—

NS

NS

NS

NS

NS

∗N

S∗

NS

NS

NS

NS

NS

Cha

etod

onlu

nula

NS

NS

NS

NS

NS

NS

NS

NS

—N

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SN

SF

orci

pige

rfla

viss

imus

NS

NS

NS

NS

NS

NS

NS

NS

NS

—N

SN

S∗

∗∗N

SN

S∗∗

NS

NS

NS

NS

Cha

etod

onul

iete

nsis

NS

∗∗N

SN

SN

SN

SN

SN

SN

SN

S—

NS

NS

NS

NS

NS

NS

NS

NS

NS

NS

Bot

tom

feed

ers

Cha

etod

onau

riga

NS

NS

NS

NS

NS

NS

NS

NS

NS

NS

NS

—∗

∗∗N

SN

S∗

NS

NS

NS

NS

Cha

etod

onse

mei

on∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

NS

∗∗∗

∗N

S—

NS

∗∗∗∗

∗N

S∗∗

∗∗∗

∗C

haet

odon

ephi

ppiu

m∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗N

S∗∗

∗∗∗

∗N

S—

∗∗∗∗

∗N

S∗∗

∗∗∗

∗∗∗

Cha

etod

onva

gabu

ndus

NS

NS

NS

NS

NS

NS

NS

NS

NS

NS

NS

NS

∗∗∗

∗∗∗

—N

S∗∗

NS

NS

NS

NS

Hen

ioch

usch

ryso

stom

us∗

∗∗∗

NS

NS

∗∗

NS

NS

NS

NS

NS

NS

∗∗∗

∗∗∗

NS

—∗∗

∗N

SN

S∗

NS

Hen

ioch

usva

rius

NS

NS

NS

NS

NS

NS

NS

∗N

SN

SN

SN

S∗∗

∗∗∗

∗N

SN

S—

∗∗∗∗

∗∗

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 12

1134 I . NAG E L K E R K E N E T A L .

Tab

leIV

.C

ontin

ued

Cor

alfe

eder

sC

oral

–bo

ttom

feed

ers

Bot

tom

feed

ers

Spon

gefe

eder

sPe

lagi

cfe

eder

s

Chaetodonbaronessa

Chaetodonlunulatus

Chaetodonbennetti

Chaetodonornatissimus

Chaetodonpunctatofasciatus

Chaetodonspeculum

Forcipigerlongirostris

Chaetodonkleinii

Chaetodonlunula

Forcipigerflavissimus

Chaetodonulietensis

Chaetodonauriga

Chaetodonsemeion

Chaetodonephippium

Chaetodonvagabundus

Heniochuschrysostomus

Heniochusvarius

Chaetodonadiergastos

Coradionchrysozonus

Hemitaurichthyspolylepis

Heniochusdiphreutes

Spon

gefe

eder

sC

haet

odon

adie

rgas

tos

NS

NS

NS

NS

NS

NS

∗∗∗∗

∗N

SN

SN

SN

S∗∗

∗∗∗

∗N

S∗

NS

—N

SN

SN

SC

orad

ion

chry

sozo

nus

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

NS

NS

∗∗N

S∗∗

∗∗∗

∗∗∗∗

∗∗∗

—N

SN

SPe

lagi

cfe

eder

sH

emit

auri

chth

yspo

lyle

pis

∗∗∗

∗∗∗

∗∗∗∗

∗∗∗

∗∗∗

∗∗∗

NS

NS

∗∗N

S∗

∗∗∗

∗∗∗∗

∗∗∗

∗N

S—

NS

Hen

ioch

usdi

phre

utes

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗N

SN

S∗∗

NS

∗∗∗

∗∗∗∗

∗∗∗

∗∗∗

NS

NS

—

NS,

non-

sign

ifica

nt.

∗ P≤

0·05.

∗∗P

≤0·0

1.∗∗

∗ P≤

0·001

.

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 13

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1135

source. The latter is supported by an average enrichment of more than the usual3·5‰ between consumer and prey: coral tissue and coral–bottom feeders differedon average by >5‰ in δ15N.

Whilst feeding on sponges, it could not be discerned whether C. kleinii andF. longirostris fed on the sponge tissue or on macrofauna associated with the spongesurface. Gut content and stable isotope analysis both suggested the latter, as spongetissue was not an important component of their gut contents and the two species weretoo enriched in δ13C to be feeding considerably on sponges. Chaetodon kleinii hadthe widest diet breadth of all species and the most diverse array of dietary items in itsgut contents, but gut contents were dominated by Copepoda (Table III). δ13C valuesin tissue of this species were depleted (although not significantly, Table IV) com-pared to the other four species in this group [Fig. 2(a)]. The depleted values couldpotentially be caused by feeding on pelagic Copepoda and Gastropoda (Table III)which showed depleted δ13C values, as depicted by those of zooplankton, com-pared to food items comprising benthic macrofauna and corals [Fig. 2(b)]. Feedingobservations confirmed pelagic feeding for this species (Table I).

No significant differences in δ13C and δ15N values were found among the otherfour species in this group (Table IV).

B OT TO M F E E D E R S

In situ feeding observations identified four species, Chaetodon auriga Forsskal,Chaetodon ephippium Cuvier, Chaetodon semeion Bleeker and Chaetodon vagabun-dus L., which fed primarily from the hard bottom substratum (Table I). Gut contentanalysis suggested that C. auriga and C. vagabundus mainly preyed on polychaeteworms while feeding on the substratum, but strangely it also showed considerablefeeding on hard corals for C. ephippium and C. semeion (Table III). Stable iso-tope analysis showed yet another pattern (Fig. 2), suggesting C. ephippium and C.semeion to feed on benthic filter feeders (possibly sponges and tunicates) and C.auriga and C. vagabundus to feed on crustaceans. The former two species showedsignificantly depleted values of δ13C and δ15N compared to all other chaetodontidspecies (except C. lunula), while the latter two species showed no significant dif-ferences in stable isotope ratios compared to members of the coral–bottom feeders(Table IV). Chaetodon auriga and C. vagabundus showed significant differences intissue δ15N values with several species of coral feeders.

No in situ feeding observations were obtained for Heniochus chrysostomus Cuvierand Heniochus varius (Cuvier), but gut-content analysis suggested a diet mainlybased on polychaetes and amphipods, respectively (Table III), while stable-isotopeanalysis suggested a diet mainly based on crustaceans and corals, respectively (Fig. 2).Heniochus chrysostomus did not show significant differences in δ13C or δ15N valueswith coral–bottom feeders (Table IV), but did show differences with coral feed-ers (except for C. bennetti ). Heniochus varius showed the almost opposite pattern(except for lack of significant differences with C. lunula and C. ulietensis).

S P O N G E F E E D E R S

In situ feeding observations revealed two species, Chaetodon adiergastos Sealeand Coradion chrysozonus (Cuvier), which fed mostly from the surface of sponges

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 14

1136 I . NAG E L K E R K E N E T A L .

Tab

leV

.M

ean

±s.

d.fo

rkle

ngth

(LF),

num

ber

ofsa

mpl

espr

oces

sed

for

stab

le-i

soto

pean

alys

isan

dan

over

view

ofth

em

ain

diet

and

feed

ing

subs

trat

ade

duce

dfr

omth

eth

ree

used

met

hods

for

the

21sp

ecie

sof

Cha

etod

ontid

ae.

The

orde

rof

spec

ies

follo

ws

that

ofTa

ble

I.T

helo

wer

part

ofth

eta

ble

prov

ides

seve

ral

adva

ntag

esan

ddi

sadv

anta

ges

ofth

eva

riou

sm

etho

ds

Fish

spec

ies

LF

(mm

)N

umbe

rof

isot

ope

sam

ples

Feed

ing

obse

rvat

ions

(FO

);>

25%

cont

ribu

tion

Gut

cont

ent

anal

ysis

(GC

A);

>20

%co

ntri

buti

onSt

able

isot

ope

anal

ysis

(SIA

)

Cor

alfe

eder

sC

haet

odon

baro

ness

a11

0±

36

Cor

als

Cor

als/

anne

lids

Cor

als/

poly

chae

tes

Cha

etod

onlu

nula

tus

117

±3

6C

oral

sC

oral

s/m

acro

alga

eM

acro

alga

e/co

rals

/pol

ycha

etes

Cha

etod

onbe

nnet

ti14

5±

104

Cor

als

Mac

roal

gae/

anne

lids

Cor

als/

poly

chae

tes

Cha

etod

onor

nati

ssim

us14

5±

114

—M

acro

alga

eC

oral

s/po

lych

aete

s/m

acro

alga

eC

haet

odon

punc

tato

fasc

iatu

s92

±3

7C

oral

sC

rust

acea

ns/m

ollu

scs

Cor

als/

poly

chae

tes

Cha

etod

onsp

ecul

um12

8±

15

Cor

als/

hard

bott

omC

rust

acea

ns/a

nnel

ids

Cor

als/

poly

chae

tes

Cor

al–

bott

omfe

eder

sF

orci

pige

rlo

ngir

ostr

is16

3±

106

Cor

als/

spon

ges/

hard

bott

omC

rust

acea

nsC

rust

acea

nsC

haet

odon

klei

nii

100

±6

5C

oral

s/sp

onge

s/ha

rdbo

ttom

Mul

tipl

epr

eyty

pes

Cru

stac

eans

/zoo

plan

kton

Cha

etod

onlu

nula

163

±10

3H

ard

bott

om/c

oral

sPo

lych

aete

sC

rust

acea

nsF

orci

pige

rfla

viss

imus

146

±5

4H

ard

bott

omPo

lych

aete

sC

rust

acea

nsC

haet

odon

ulie

tens

is14

0±

134

Hyd

roid

sPo

lych

aete

sC

rust

acea

nsB

otto

mfe

eder

sC

haet

odon

auri

ga16

4±

22

Har

dbo

ttom

Poly

chae

tes/

mac

roal

gae

Cru

stac

eans

Cha

etod

onse

mei

on18

9±

83

Har

dbo

ttom

Cor

als

Spon

ges

and

tuni

cate

sC

haet

odon

ephi

ppiu

m18

5±

106

Har

dbo

ttom

Cor

als

Spon

ges

and

tuni

cate

sC

haet

odon

vaga

bund

us13

0±

66

Har

dbo

ttom

Poly

chae

tes/

eggs

Cru

stac

eans

Hen

ioch

usch

ryso

stom

us14

2±

65

—Po

lych

aete

sC

rust

acea

nsH

enio

chus

vari

us15

5±

95

—C

rust

acea

ns/p

olyc

haet

esC

oral

s/po

lych

aete

sSp

onge

feed

ers

Cha

etod

onad

ierg

asto

s14

8±

97

Spon

ges

Poly

chae

tes

Cru

stac

eans

Cor

adio

nch

ryso

zonu

s14

2±

166

Spon

ges/

hard

bott

omM

ulti

ple

prey

type

sSp

onge

s/m

acro

inve

rteb

rate

sPe

lagi

cfe

eder

sH

emit

auri

chth

yspo

lyle

pis

131

±5

5Pl

ankt

onZ

oopl

ankt

onZ

oopl

ankt

onH

enio

chus

diph

reut

es15

3±

65

Plan

kton

Zoo

plan

kton

Zoo

plan

kton

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 15

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1137T

able

V.

Con

tinue

d

Fish

spec

ies

LF

(mm

)N

umbe

rof

isot

ope

sam

ples

Feed

ing

obse

rvat

ions

(FO

);>

25%

cont

ribu

tion

Gut

cont

ent

anal

ysis

(GC

A);

>20

%co

ntri

buti

onSt

able

isot

ope

anal

ysis

(SIA

)

Adv

anta

ges

–id

enti

fied

sign

ifica

ntco

ral

feed

ing

for

som

esp

ecie

s–

iden

tifi

edcr

ypti

cco

mpo

nent

sin

diet

ofpr

edom

inan

tly

cora

lfe

eder

s(e

.g.

anne

lids

,cr

usta

cean

san

dal

gae)

–id

enti

fied

for

cora

lfe

eder

sth

atco

rals

are

prob

ably

also

the

mai

nfo

odso

urce

over

long

erti

me

span

s

–id

enti

fied

H.d

iphr

eute

san

dH

.pol

ylep

isas

pela

gic

feed

ers

vs.

C.c

hrys

ozon

usas

asp

onge

feed

er(n

otfo

und

bySI

A)

–id

enti

fied

cora

las

cryp

tic

diet

com

pone

ntfo

rpr

edom

inan

tly

non-

cora

lfe

eder

sC

.sem

eion

,C

.ep

hipp

ium

and

C.

chry

sozo

nus

–id

enti

fied

mac

roal

gal

feed

ing

byC

.lu

nula

tus

and

C.o

rnat

issi

mus

over

long

erti

me

span

s

–id

enti

fied

spec

ific

cora

lsp

ecie

sse

lect

edas

food

sour

ce

–ta

xon-

spec

ific

iden

tific

atio

nof

food

sour

ces

–id

enti

fied

lack

oflo

ng-t

erm

mac

roal

gal

feed

ing

for

C.b

enne

tti

assu

gges

ted

byG

CA

–id

enti

fied

mic

roha

bita

tsus

edfo

rfe

edin

g–

iden

tifie

dhi

gher

depe

nden

ceon

crus

tace

ans

than

onpo

lych

aete

sas

food

sour

cefo

rC

.aur

iga

and

C.

vaga

bund

usth

ansu

gges

ted

byG

CA

–id

enti

fied

C.

sem

eion

and

C.

ephi

ppiu

mas

mai

nly

spon

gefe

eder

s–

supp

orte

dsp

onge

feed

ing

(by

FO)

for

Cor

adio

nch

ryso

zonu

s(n

otfo

und

byG

CA

)–

iden

tifie

dlo

wde

pend

ence

onsp

onge

asfo

odso

urce

for

C.a

dier

gast

os(i

nco

ntra

stto

FO)

Dis

adva

ntag

es–

coul

dno

tid

enti

fysp

ecifi

cpr

eyta

xaon

subs

trat

aot

her

than

cora

ls

–co

uld

not

dete

ctco

ral

asfo

odso

urce

for

cora

lfe

eder

s–

low

taxo

nom

icre

solu

tion

ofpr

eyit

ems

cons

umed

–sn

apsh

otof

feed

ing

beha

viou

r–

larg

epr

opor

tion

ofgu

tun

iden

tifia

ble

–sn

apsh

otof

diet

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 16

1138 I . NAG E L K E R K E N E T A L .

(Table I). Gut contents for these species, however, rarely contained sponge spicules,but comprised a diverse assemblage of invertebrates (Table III), suggesting that theydid not feed on the sponges themselves, but on the sponge-associated fauna. Thiswas particularly apparent for C. chrysozonus which showed the second widest dietbreadth (0·20; Table III) of any chaetodontid. The two species differed significantlyin their δ13C values (Table IV). Stable isotope analysis supported the notion of thediet consisting of macrofauna (possibly crabs) for C. adiergastos, but suggested amixed diet on sponge tissue and benthic invertebrates for C. chrysozonus (Fig. 2).

P E L AG I C F E E D E R S

In situ feeding observations for Hemitaurichthys polylepis (Bleeker) andHeniochus diphreutes Jordan revealed that these chaetodontids never fed from ben-thic substrata, but spent considerable time in the water column (Table I), up to 10 mfrom the reef matrix. Though it was not possible to obtain quantitative data on feedingrates, these observations suggest that both these species are obligate pelagic feed-ers, ostensibly feeding on planktonic organisms. Support for planktonic feeding wasobtained during other scuba dives (20+ dives, 45–60 min. each) at Derawan Islandand surrounding islands (see Fig. 1), where these two species were commonly seenfeeding from the water column (I. Nagelkerken & G. van derVelder., pers. obs.). Gutcontent analysis confirmed that the predominant dietary items consumed by these twofishes are that of Copepoda (many of which are known to be pelagic) and Gastropoda(>25% in the guts consisting of pelagic species) (Table III). Moreover, stable iso-tope ratios in the tissues of both H. polylepis and H. diphreutes were consistent withfeeding on zooplankton (Fig. 2) and were significantly depleted compared to almostall species of coral, coral–bottom and bottom feeders (Table IV).

S I M I L A R I T Y I N D I E T

The Chaetodontidae had a specialized diet, shown by the overall low values(<0·30) of diet breadth (Table III), which can reach a maximum value of 1.0 whenspecies are generalist feeders and target a wide range of food items. These specializeddiets did not overlap much among species, as shown by each of the three differentmethods used. Overall, the 21 species showed relatively little similarity in diet basedon gut contents: 67% of all pair-wise comparisons between species showed <20%similarity in diet (Fig. 3). For diet based on feeding observations, which is generallya less taxon-specific method of diet determination (except for cases such as coralpolyp feeding), 67% of all pair-wise comparisons between species showed <40%similarity in diet. Of a total of 210 pair-wise comparisons (see Table IV) for δ13Cvalues between species, 94 were significantly different. Of the remaining ones, anadditional 25 comparisons that did not show significant differences for δ13C showeddifferences for δ15N, resulting in 57% of all possible species comparisons beingsignificantly different in stable-isotope values.

DISCUSSION

This is the first-ever study to utilize stable isotopes to explore dietary habits ofChaetodontidae, providing novel insights into significant sources of nutrition for these

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 17

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1139

0

10

20

30

40

50

60

70

0·0 –20·0

20·1–40·0

40·1–60·0

60·1–80·0

89·1–100·0

Bray–Curtis similarity (%)

Freq

uenc

y (%

)

Fig. 3. Frequency histograms for degree of similarity in diet among 21 species of Chaetodontidae, based onfeeding observations ( ; see Table I) and on gut-content analysis ( ; see Table III). Similarity in dietwas calculated for all possible pairwise comparisons between species and subsequently subdivided intoclasses of 20% similarity.

fishes. Previous studies based on gut-content analyses or in situ feeding observations(Harmelin-Vivien & Bouchon-Navaro, 1981, 1983; Harmelin-Vivien, 1989; Wrathallet al ., 1992; Cox, 1994; Berumen et al ., 2005; Graham, 2007) have emphasizedscleractinian corals as the most critical component in the diet of most chaetodontids.The δ13C values recorded for the Chaetodontidae, however, ranged from −20·1 to−11·3‰, which is as large as that of the food organisms found on the coral reef.This shows that chaetodontids target a very wide range of different food items, pro-viding considerable opportunity for differences in the main dietary items consumedby sympatric chaetodontids. Moreover, diet breadth was very low indicating a highdegree of food specialization by chaetodontids. These specialized diets did not over-lap much among species as shown by feeding observations, and gut-content andstable-isotope analyses. The chaetodontid species differed in their diet by specializ-ing on sponges, zooplankton, polychaetes, corals or decapods, or were generalistsfeeding on a wide variety of macrofauna, sometimes including macroalgae (Table V).While previous studies have suggested that there is often very limited dietary par-titioning among chaetodontids (Bouchan-Navaro, 1986; Pratchett, 2005), increasedresolution of dietary composition provided in this study suggests that species withostensibly similar diets (e.g. specialized coral feeders) may consume very differentcryptic dietary items, as shown by gut content analysis, or feed on a different suiteof coral species. Differences in the diets of sympatric species may be fundamentalin minimizing interspecific competition and facilitating coexistence among coral reeffishes (Sale, 1974).

Stable isotope ratios tend to distinguish two major groupings (in terms of num-ber of species) and several minor groupings among the 21 chaetodontid species atDerawan Island, East Kalimantan. The most prominent group included all chaetodon-tids categorized as coral and bottom feeders (C. kleinii, C. lunula, C. ulietensis, F.flavissimus and F. longirostris), two (out of four) bottom feeders (C. auriga andC. vagabundus), a sponge feeder (C. adiergastos) and H. chrysostomus. Assumingnormal levels of dietary enrichment, these species are presumed to feed mainly on

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 18

1140 I . NAG E L K E R K E N E T A L .

crustaceans, possibly with a moderate component of their diet comprising hard andsoft corals, polychaetes or other prey items. These findings are largely consistent withother studies where most of these species (C. auriga, C. kleinii, C. lunula, C. ulieten-sis, C. vagabundus and H. chrysostomus) are regarded as facultative coral feeders,which feed to a greater or lesser extent on corals (Harmelin-Vivien & Bouchon-Navaro, 1983; Harmelin-Vivien, 1989; Sano, 1989; Pratchett, 2005; Graham, 2007),though the importance of corals in the diets of these species is unclear. For example,C. auriga feeds extensively on hard and soft corals in some geographic locations(e.g. French Polynesia, Harmelin-Vivien & Bouchon-Navaro, 1981; Indian Ocean,Graham, 2007), but not in others (e.g. Great Barrier Reef in Australia, Pratchett,2005; Indonesia, this study). If coral is a critical component in the diet of thesespecies, they are likely to be susceptible to widespread coral loss caused by climate-induced coral bleaching (Pratchett et al ., 2006; Graham, 2007) and other majordisturbances (Wilson et al ., 2006). Geographic variation in dietary composition offacultative corallivores, however, suggest that these species opportunistically usecorals for food and can survive on alternate prey items (e.g. polychaetes) whereexcessive competition or limited availability precludes feeding on corals.

The other major grouping comprised chaetodontid species observed to feed osten-sibly on corals (C. baronessa, C. bennetti, C. lunulatus, C. ornatissimus, C. punc-tatofasciatus and C. speculum) as well as H. varius, which was not observed feedingin this study, but is reported to feed on mainly on scleractinian corals (Sano, 1989).Elsewhere in the Pacific, C. baronessa, C. bennetti, C. lunulatus, C. ornatissimus,C. punctatofasciatus and C. speculum are all considered to be obligate hard-coralfeeders (Cox, 1994; Berumen et al ., 2005; Pratchett, 2005). The stable isotope resultsdo support a diet based predominantly on hard corals, but feeding on different speciesof corals and on minor prey items (e.g. algae, polychaetes and crustaceans), whichmay be overlooked during feeding observations (see Table V) or by using stableisotope analysis, is potentially very important in providing sufficient dietary sepa-ration to facilitate competitive coexistence among these species. The relative useof non-coral prey may also be critical in providing greater resilience for chaetodon-tids during increasing disturbances (e.g. climate-induced coral bleaching) that reducecoral availability (Pratchett et al ., 2006; Wilson et al ., 2006).

The combination of methods used provided more detail on the specific diets ofChaetodontidae and was useful in evaluating the data in cases where one methodprovided a very different outcome than the other methods. Table V lists severaladvantages and disadvantages of the three methods and shows how use of one methodmay reveal dietary items that are not discovered by other techniques. Each of thethree methods was effective in distinguishing such dietary items. Most importantly,gut-content analysis frequently identified cryptic food items that did not form themain component of the diet (as shown by stable isotope analysis). Feeding obser-vations provided ecological data on the microhabitat from which the food itemswere selected, but also specifically which coral species were targeted by coral feed-ers. Identification of coral as the main food source for coral feeders, however, wasnot successful through gut-content analysis, probably due to the soft tissue of coralpolyps that quickly disintegrated or were rapidly digested in the gut. Stable iso-tope analysis showed an apparently good match in δ13C and δ15N values betweenchaetodontids and their expected prey, reflecting commonly observed degrees ofenrichment between prey and predator. This match indicated the major prey items

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 19

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1141

were being assessed or caught by prey sampling procedures. Stable carbon-isotopeanalysis often provided data on whether food items consumed formed part of theregular diet (i.e. consistent over longer time periods) or were consumed opportunis-tically; in addition, it separated species that directly consumed sponge tissue fromthose consuming macrofauna on sponges. Stable nitrogen-isotope analysis was usefulin identifying the degree to which crustaceans, corals, or macroalgae contributed tothe diet of species that were considered coral feeders on basis of field observations.Although sample numbers were relatively low for the techniques used, this did notaffect the conclusions of the study. The division of species into groups based onfeeding observations was unambiguous considering the individual variation amongspecies as shown in Table I. The sample size was low for gut-content analysis, butthis did not affect the conclusion that cryptic components contributed to the diet.Finally, stable-isotope value mostly showed small s.e. leading to many significantdifferences among species of the different groups.

The present study reveals extreme diversity in the dietary habitats of chaetodontidswhereby Chaetodontidae assemblages comprise planktivores, corallivores, zooben-thivores, generalist feeders and partial herbivores (omnivores). The combination ofin situ feeding observation, gut content analysis and stable-isotope analysis has fur-ther improved resolution of dietary habitats for coral-reef chaetodontids (Table V),thereby suggesting that dietary overlap among sympatric chaetodontids may havebeen overestimated in the past (Bouchan-Navaro, 1986; Pratchett, 2005). Accord-ingly, coexistence of chaetodontids may be facilitated by resource partitioning(Schoener, 1974), contributing to the high local diversity of chaetodontids on coralreefs (Pratchett, 2005; this study). These data, combined with increasing understand-ing of competitive hierarchies among sympatric chaetodontids (Wrathall et al ., 1992;Berumen & Pratchett, 2006), suggest that despite stochastic recruitment and impor-tant priority effects (Sale, 1974) interspecific competition may still play an importantrole in structuring of reef-fish assemblages.

This study was funded by the Netherlands Foundation for the Advancement of TropicalResearch (WOTRO), EKP-Pilot Project WT87-301. We are indebted to B. W. Hoeksema ofthe National Natural History Museum Naturalis, Leiden, for inviting us to join his researchteam and for organizing the logistics of the expedition. We thank Derawan Dive Resort for alltheir support on the island, M. G. Versteeg for logistic support in the field, M. van Couwelaarfor collecting the zooplankton samples, M. E. Orbons for help with the gut-content analysis,J. Eygensteyn for analysing the stable isotopes, G. R. Allen for checking the identification ofthe fish species and M. L. Berumen for his comments on an earlier version of the manuscript.I. N. was supported by a Vidi grant from the Netherlands Organisation for Scientific Research(NWO). This is a Centre for Wetland Ecology publication nr 410.

References

Allen, G. R., Steene, R. & Allen, M. (1998). A Guide to Angelfishes and Butterflyfishes . SanDiego, CA: Odyssey Publishing.

Berumen, M. L. & Pratchett, M. S. (2006). Effects of resource availability on the competitivebehaviour of butterflyfishes (Chaetodontidae). Proceedings of the 10th InternationalCoral Reef Symposium 1, 644–650.

Berumen, M. L., Pratchett, M. S. & McCormick, M. I. (2005). Within-reef differences in dietand body condition of coral-feeding butterflyfishes (Chaetodontidae). Marine EcologyProgress Series 287, 217–227.

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 20

1142 I . NAG E L K E R K E N E T A L .

Bootsma, H. A., Hecky, R. E., Hesslein, R. H. & Turner, G. F. (1996). Food partitioningamong Lake Malawi nearshore fishes as revealed by stable isotope analysis. Ecology77, 1286–1290.

Bouchon-Navaro, Y. (1986). Partitioning of food and space resources by chaetodontid fisheson coral reefs. Journal of Experimental Marine Biology and Ecology 103, 21–40.

Bunn, S. E., Loneragan, N. R. & Kempster, M. A. (1995). Effects of acid washing on stableisotope ratios of C and N in penaeid shrimps and seagrass: implications for food-webstudies using multiple stable isotopes. Limnology and Oceanography 40, 622–625.

Clarke, R. D. (1994). Habitat partitioning by chaenopsid blennies in Belize and the VirginIslands. Copeia 1994, 398–405.

Colwell, R. K. & Fuentes, E. R. (1975). Experimental studies of the niches. Annual Reviewof Ecology and Systematics 6, 281–310.

Connell, J. H. (1983). On the prevalence and relative importance of interspecific competi-tion – evidence from field experiments. American Naturalist 122, 661–696.

Cox, E. F. (1994). Resource use by corallivorous butterflyfishes (family Chaetodontidae) inHawaii. Bulletin of Marine Science 54, 535–545.

Davenport, S. R. & Bax, N. J. (2002). A trophic study of a marine ecosystem off southeasternAustralia using stable isotopes of carbon and nitrogen. Canadian Journal of Fisheriesand Aquatic Sciences 59, 514–530.

DeNiro, M. J. & Epstein, S. (1978). Influence of diet on the distribution of carbon isotopesin animals. Geochimica et Cosmochimica Acta 42, 495–506.

DeNiro, M. J. & Epstein, S. (1981). Influence of diet on the distribution of nitrogen isotopesin animals. Geochimica et Cosmochimica Acta 45, 341–351.

Diamond, J. M. (1978). Niche shifts and the rediscovery of interspecific competition. Amer-ican Scientist 66, 322–331.

Field, A. (2006). Discovering Statistics Using SPSS , 2nd edn. London: Sage Publications.Fry, B. (1988). Food web structure on Georges Bank from stable C, N, and S isotopic com-

positions. Limnology and Oceanography 33, 1182–1190.Fulton, C. J., Bellwood, D. R. & Wainwright, P. C. (2001). The relationship between swim-

ming ability and habitat use in wrasses (Labridae). Marine Biology 139, 25–33.Gardiner, N. M. & Jones, G. P. (2005). Habitat specialisation and overlap in a guild of coral

reef cardinalfishes (Apogonidae). Marine Ecology Progress Series 305, 163–175.Gearing, J. N. (1991). The study of diet and trophic relationships through natural abundance

13C. In Carbon Isotope Techniques (Coleman, D. C. & Fry, B., eds), pp. 201–218.San Diego, CA: Academic Press.

Graham, N. A. J. (2007). Ecological versatility and the decline of coral feeding fishes fol-lowing climate driven coral mortality. Marine Biology 153, 119–127.

Harmelin-Vivien, M. L. (1989). Implications of feeding specialization on the recruitment pro-cesses and community structure of butterflyfishes. Environmental Biology of Fishes 25,101–110.

Harmelin-Vivien, M. L. & Bouchon-Navaro, Y. (1981). Trophic relationships amongchaetodontid fishes in the Gulf of Aqaba (Red Sea). Proceedings of the 4th InternationalCoral Reef Symposium 2, 537–544.

Harmelin-Vivien, M. L. & Bouchon-Navaro, Y. (1983). Feeding diets and significance ofcoral feeding among chaetodontid fishes in Moorea (French Polynesia). Coral Reefs 2,119–127.

Hesslein, R. H., Hallard, K. A. & Ramlal, P. (1993). Replacement of sulfur, carbon, andnitrogen in tissue of growing broad whitefish (Coregonus nasus) in response to achange in diet traced by δ34S, δ13C, and δ15N. Canadian Journal of Fisheries andAquatic Sciences 50, 2071–2076.

Hurlbert, S. H. (1978). The measurement of niche overlap and some relatives. Ecology 59,67–77.

Livingston, R. J. (1984). Trophic response of fishes to habitat variability in coastal seagrasssystems. Ecology 65, 1258–1275.

MacDonald, J. S., Waiwood, K. G. & Green, R. H. (1982). Rates of different digestion of dif-ferent prey in Atlantic cod (Gadus morhua), ocean pout (Macrozoarces americanus),winter flounder (Pseudopleuronectes americanus), and American plaice (Hippoglos-soides platessoides). Canadian Journal of Fisheries and Aquatic Sciences 39, 651–659.

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143

Page 21

D I E TA RY OV E R L A P A M O N G C H A E T O D O N T I DA E 1143

Mariani, S., Maccaroni, A., Massa, F., Rampacci, M. & Tancioni, L. (2002). Lack of consis-tency between trophic interrelationships of five sparid species in two adjacent centralMediterranean coastal lagoons. Journal of Fish Biology 61 (Suppl. A), 138–147.

McAleece, N. (1997). Biodiversity Pro. London: The Natural History Museum.Michael, S. W., Bavendam, F., Hoover, J. P. & Kuiter, R. H. (2004). Angelfishes and Butter-

flyfishes . New Jersey, NJ: T.F.H. Publications.Mill, A. C., Pinnegar, J. K. & Polunin, N. V. C. (2007). Explaining isotope trophic-step frac-

tionation: why herbivorous fish are different. Functional Ecology 21, 1137–1145.Minagawa, M. & Wada, E. (1984). Stepwise enrichment of 15N along food chains: further

evidence and the relation between δ15N and animal age. Geochimica et CosmochimicaActa 48, 1135–1140.

Munday, P. L. (2004). Competitive coexistence of coral-dwelling fishes: the lottery hypothesisrevisited. Ecology 85, 623–628.

Nagelkerken, I., van der Velde, G., Verberk, W. C. E. P. & Dorenbosch, M. (2006). Segre-gation along multiple resource axes in a tropical seagrass fish community. MarineEcology Progress Series 308, 79–89.

Nielsen, L. A. & Johnson, D. L. (1992). Fisheries Techniques . Bethesda, MD: American Fish-eries Society.

Nieuwenhuize, J., Maas, Y. E. M. & Middelburg, J. J. (1994). Rapid analysis of organic car-bon and nitrogen in particulate materials. Marine Chemistry 45, 217–224.

Pinnegar, J. K. & Polunin, N. V. C. (2000). Contributions of stable-isotope data to elucidatingfood webs of Mediterranean rocky littoral fishes. Oecologia 122, 399–409.

Polis, G. & Strong, D. R. (1996). Food web complexity and community dynamics. AmericanNaturalist 147, 813–846.

Pratchett, M. S. (2005). Dietary overlap among coral-feeding butterflyfishes (Chaetodontidae)at Lizard Island, northern Great Barrier Reef. Marine Biology 148, 373–382.

Pratchett, M. S. (2007). Dietary selection by coral-feeding butterflyfishes (Chaetodontidae) onthe Great Barrier Reef, Australia. Raffles Bulletin of Zoology Supplement 14, 171–176.

Pratchett, M. S., Wilson, S. K. & Baird, A. H. (2006). Long-term monitoring of the GreatBarrier Reef. Journal of Fish Biology 69, 1269–1280.

Rau, G. H., Mearns, A. J., Young, D. R., Olson, R. J., Schafer, H. A. & Kaplan, I. R. (1983).Animal 13C/12C correlates with trophic level in pelagic food webs. Ecology 64,1314–1318.

Reese, E. S. (1975). A comparative field study of the social behaviour and related ecologyof reef fishes of the family Chaetodontidae. Zeitschrift fur Tierpsychologie 37, 37–61.

Ross, S. T. (1986). Resource partitioning in fish assemblages: a review of field studies. Copeia1986, 352–388.

Sale, P. F. (1974). Overlap in resource use, and interspecific competition. Oecologia 17,245–256.

Sale, P. F. (1978). Chance patterns of demographic change in populations of territorial fishin coral rubble patches at Heron Reef. Journal of Experimental Marine Biology andEcology 34, 233–243.

Sano, M. (1989). Feeding habits of Japanese butterflyfishes (Chaetodontidae). EnvironmentalBiology of Fishes 25, 195–203.

Schoener, T. W. (1974). Resource partitioning in ecological communities. Science185, 27–39.

Wilson, S. K. (2001). Multiscale habitat associations of detrivorous blennies (Blenniidae:Salariini). Coral Reefs 20, 245–251.

Wilson, S. K., Graham, N. A. J., Pratchett, M. S., Jones, G. P. & Polunin, N. V. C. (2006).Multiple disturbances and the global degradation of coral reefs: are reef fishes at riskor resilient? Global Change Biology 12, 2220–2234.

Wrathall, T. J., Roberts, C. M. & Ormond, R. F. G. (1992). Territoriality in the butterflyfishChaetodon austriacus. Environmental Biology of Fishes 34, 305–308.

Zekeria, Z. A., Dawit, Y., Ghebremedhin, S., Naser, M. & Videler, J. J. (2002). Resourcepartitioning among four butterflyfish species in the Red Sea. Marine and FreshwaterResearch 53, 163–168.

© 2009 The AuthorsJournal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 75, 1123–1143