FULL PAPER
Trophic niches of thirteen damselfishes (Pomacentridae)
at the Grand Re
´
cif of Toliara, Madagascar
Bruno Fre
´
de
´
rich Æ Gre
´
gory Fabri Æ Gilles Lepoint Æ
Pierre Vandewalle Æ Eric Parmentier
Received: 19 November 2007 / Revised: 15 March 2008 / Accepted: 18 March 2008 / Published online: 16 April 2008
Ó The Ichthyological Society of Japan 2008
Abstract The damselfishes, with more than 340 species,
constitute one of the most important families that live in the
coral reef environment. Most of our knowledge of reef-fish
ecology is based on this family, but their trophic ecology is
poorly understood. The aim of the present study was to
determine the trophic niches of 13 sympatric species of
damselfishes by combining stable isotope (d
15
N and d
13
C)
and stomach content analyses. Isotopic signatures reveal
three main groups according to their foraging strategies:
pelagic feeders (Abudefduf sexfasciatus, A. sparoides,
A. vaigiensis, Chromis ternatensis, C. dimidiata, Dascyllus
trimaculatus and Pomacentrus caeruleus), benthic feeders
(Chrysiptera unimaculata, Plectroglyphidodon lacrymatus
and Amphiprion akallopisos) and an intermediate group
(D. aruanus, P. baenschi and P. trilineatus). Stomach
contents reveal that planktonic copepods and filamentous
algae mainly represent the diets of pelagic feeders and
benthic feeders, respectively. The intermediate position of
the third group resulted from a partitioning of small
planktonic prey, small vagile invertebrates and filamentous
algae. In this last feeding group, the presence of a wide
range of d
13
C values in P. trilineatus suggests a larger
trophic niche width, related to diet-switching over time.
Some general considerations about the feeding habits of
damselfishes reveal that their choice of habitat on the reef
and their behavior appear to be good predictors of diet
in this group. Benthic (algae and/or small invertebrates)
feeders appear to be solitary and defend a small territory on
the bottom; zooplankton feeders remain in groups just
above the reef, in the water column.
Keywords Feeding habits Pomacentrids
Stable isotopes Stomach contents
Introduction
Coral reefs have an extraordinary diversity of fishes, and a
central goal of reef community ecology is to understand the
mechanisms allowing the coexistence of so many species.
The partitioning of resources (e.g., food and habitat) may be
viewed as one of the key factors in the diversifying process,
which promotes the coexistence of closely related and
ecologically equivalent species (Colwell and Fuentes 1975).
The trophic niche, defined as the place of an organism in the
environment in relation to its food (Silvertown 2004), is
probably one of the main axes of diversification in verte-
brates (Streelman and Danley 2003). The trophic ecology of
reef fishes has been broadly studied. For instance, extensive
evidence of trophic adaptations and wide diversity in
diet were highlighted in Labridae (Wainwright 1988;
Wainwright et al. 2004) and Chaetodontidae (Motta 1988;
Ferry-Graham et al. 2001; Pratchett 2005).
The damselfishes (Pomacentridae) include more than
340 species that live mainly in coral reef environments
(Allen 1991; Nelson 2006). Pomacentrids are perhaps the
most conspicuous inhabitants of coral reefs (Allen 1991).
Their abundance has led to a legitimate focus on damself-
ishes in reef ecological studies; most of our knowledge
of reef-fish ecology is based on this family. This is partic-
ularly true for habitat use (Waldner and Robertson 1980),
larvae (Wellington and Victor 1989), settlement (Leis and
B. Fre
´
de
´
rich (&) G. Fabri P. Vandewalle E. Parmentier
Laboratoire de Morphologie fonctionnelle et e
´
volutive, Institut
de Chimie (B6c), Universite
´
de Lie
`
ge, 4000 Lie
`
ge, Belgium
e-mail: bruno.frederich@ulg.ac.be
G. Lepoint
MARE, Laboratoire d’Oce
´
anologie, Institut de Chimie (B6c),
Universite
´
de Lie
`
ge, 4000 Lie
`
ge, Belgium
123
Ichthyol Res (2009) 56:10–17
DOI 10.1007/s10228-008-0053-2
Carson-Ewart 2002) and recruitment (Lecchini and Galzin
2005). Diverse topics on coral reef ecology related to the
diet of some damselfishes have been investigated. For
instance, the effects of herbivorous damselfishes on benthic
communities are increasingly being studied (Hata and Kato
2002; Gobler et al. 2006; Ceccarelli 2007). Damselfishes
have also served as a model fish for the study of planktivory.
For example, the impacts of ecological factors like rank
within social groups (Coates 1980), zooplankton dynamics
(Noda et al. 1992) and current speeds (Mann and Sancho
2007) on the feeding behavior of planktivorous species
were illustrated in damselfishes. However, very few works
focus on all of the various trophic niches present in the
family Pomacentridae. Currently, only two studies are
entirely devoted to comparative feeding habits in several
species of damselfishes: one at Alligator Reef, Florida Keys
(Emery 1973) and another in Taiwan (Kuo and Shao 1991).
In general, three main feeding guilds are considered (Allen
1991): herbivorous, planktivorous and omnivorous.
In trophic niche analysis, two main approaches are
usually used: stomach content analysis and the use of stable
isotope ratios. The combination of these approaches in the
same study has the advantage of compensating for the
inaccuracy of each method. Stomach content data are a
snapshot that reflect the most recent meal and may not be
representative of the overall diet. The use of stable isotope
ratios of carbon (
13
C/
12
C) and nitrogen (
15
N/
14
N) provides
complementary data in the analysis of gut contents (e.g.,
Cocheret de la Morinie
`
re et al. 2003; Parmentier and Das
2004). Stable isotope analysis has emerged as a powerful
tool for tracing dietary sources, as the isotope ratios of a
consumer are related to those of its food (Peterson and Fry
1987). This method provides an integrated measure of the
dietary components over a much longer period of time than
gut content analysis. Although stable isotope analysis does
not provide a detailed picture of dietary preferences, it
gives an average estimate of an organism’s preferred diet
that is much less subject to temporal bias (Pinnegar and
Polunin 1999). Moreover, stable isotope analysis was
recently revealed to be a powerful tool for assessing the
trophic niche widths of species (Bearhop et al. 2004).
The purpose of the present study is to determine the
trophic niches of 13 sympatric species of damselfishes
(Table 1) through an approach combining stable isotopes
(d
15
N and d
13
C) and stomach contents. This framework
will permit an informed discussion about the diversity of
their feeding habits.
Materials and methods
The 13 species were collected on the ‘‘Grand Re
´
cif’’ of
Toliara (SW Madagascar, Mozambique Channel) (23.36°S,
43.66°E) in November 2005. All specimens (Table 1) were
collected on the inner reef at depths of between 2 and 15 m
after being anesthetized by rotenon or by a solution of
quinaldin. As all damselfishes exhibit diurnal feeding
activity (Emery 1973), all fishes were captured in the
morning (between 9 and 11 a.m.). After their capture, the
fishes were brought to the surface and killed as quickly as
possible by overdose immersion in MS-222. They were
then placed on ice.
Each fish was weighed and its standard length (SL) was
measured to the nearest millimeter. Samples (±2cm
3
)of
lateral muscle tissue of each fish were used for stable
isotope analysis. The entire digestive tract was removed
and conserved in 70% alcohol for stomach content analy-
sis. All fish samples showed little variation in size except in
Amphiprion akallopisos, Abudefduf vaigensis, Dascyllus
trimaculatus and Plectroglyphidodon lacrymatus.
Different primary food sources (plankton, benthic
invertebrates and algae) were taken from the fish collection
site. Suspended organic matter was obtained by filtering 5 l
per sample of seawater from the collection site on GF/C
Whatman filters, after pre-filtering through a 250 lm sieve.
Meso-zooplankton was trapped using a net with a mesh of
250 lm, towed on the reef at a depth of 2 m. Erect mac-
roalgae (three species of chlorophytes, one rhodophyte and
three phaeophytes) colonizing the reef and turf algae from
dead coral areas, where some studied damselfishes lived,
were taken. Benthic small invertebrates (such as isopods,
amphipods, harpacticoid copepods, annelids) were col-
lected from the reef during the night using light traps
attached to the substratum above the bottom (50 cm).
These traps consisted of plastic bottles (1.5 l) containing
glowsticks, which were alight for 8–12 h.
Table 1 Species studied in this work
Species Abbreviations nn
0
SL (mm)
Abudefduf sexfasciatus A. sex 11 2 83–99
Abudefduf sparoides A. spa 13 1 80–89
Abudefduf vaigiensis A. vai 9 3 93–127
Amphiprion akallopisos Am. aka 10 2 34–80
Chromis dimidiata C. dim 12 1 33–53
Chromis ternatensis C. ter 5 0 63–82
Chrysiptera unimaculata Ch.uni 15 0 53–75
Dascyllus aruanus D. aru 11 0 28–50
Dascyllus trimaculatus D. tri 13 0 37–79
Plectroglyphidodon lacrymatus Pl. lac 10 0 44–76
Pomacentrus baenschi P. bae 4 0 53–77
Pomacentrus caeruleus P. cae 11 0 44–67
Pomacentrus trilineatus P. tri 10 1 50–66
n Number of specimens, n
0
number of specimens in which stomach is
empty, SL standard length
Diet of damselfishes 11
123
Stomach content analysis. Stomachs were opened and
all dietary constituents were identified using a Wild M10
binocular microscope or a polarizing microscope (DM
1000, Leica, Solms, Germany). In an ecomorphological
context, the preys were separated into six categories
reflecting the imposed functional constraints (Barel 1983):
phytoplankton, benthic algae, sessile invertebrates, vagile
invertebrates, zooplankton and detritus. The stomach con-
tents were expressed by numerical methods (Hyslop 1980):
the number of items in each food category was recorded for
each stomach, and the total was expressed as a percentage.
The variation in diet with size was also investigated in
Amphiprion akallopisos, Abudefduf vaigensis, Dascyllus
trimaculatus and Plectroglyphidodon lacrymatus.
Stable isotope analysis. Samples of lateral muscle tissue
and potential food sources were dehydrated for 24 h at
50°C before being ground into a homogeneous powder.
After grinding, samples containing carbonates (macroal-
gae, zoobenthos) were placed for 24 h under a glass bell
with fuming HCl (37%) (Merck, Darmstadt, Germany, for
analysis quality) in order to eliminate calcareous material.
Lipids were not removed from all samples. Seawater
samples were filtered through precombusted glass filters
(Whatman GF/F) (at 450°C, for 4 h) for the isotopic
analysis of particulate organic matter (POM). Carbon and
nitrogen gas were analyzed on a V.G. Optima (Micromass,
Beverly, MA, USA) IR-MS coupled to an N–C–S ele-
mental analyzer (Carlo Erba, Milan, Italy). Routine
measurements were precise to within 0.3% for both d
13
C
and d
15
N. Stable isotope ratios were expressed in d nota-
tion according to the following:
dX ¼ R
sample
R
standard
1
1; 000
where X is
13
Cor
15
N and R is the corresponding ratio
13
C/
12
Cor
15
N/
14
N for samples or standards. Carbon and
nitrogen ratios are expressed relative to the vPDB (Vienna
Peedee Belemnite) standard and to the atmospheric nitro-
gen standard, respectively. Reference materials were
IAEA-N1 (d
15
N = +0.4 ± 0.2%) and IAEA CH-6
(sucrose) (d
13
C =-10.4 ± 0.2%).
Statistical analyses. The Kolmogorov–Smirnov test was
used to test the normality of the data. One-way ANOVA
were used to compare isotope ratios between the different
species. These analyses were computed using STATISTI-
CA, Version 7.1 (Statsoft 2005). The computer package
PRIMER (Plymouth Routines in Multivariate Ecological
Research Ltd., UK, Version 5.2.9) was used to generate
hierarchical clustering based on applying Euclidian dis-
tances to isotope ratios. Stomach content data were also
analyzed in PRIMER using the following techniques.
Hierarchical clustering was performed based on applying a
Bray–Curtis similarity coefficient to percentage composi-
tion data. The nonparametric ANOSIM test (analysis of
similarities) was then used to statistically test the inter-
specific differences in stomach contents. Finally, in order
to characterize the trophic niche of each species, PRIMER
was used to perform a hierarchical clustering, based on
applying the Bray–Curtis similarity coefficient to all data
from the two approaches.
Results
Stomach content analysis. The stomachs of 10 individuals
from 134 examined specimens of the 13 species were
found to be empty (Table 1). One specimen of Abudefduf
vaigiensis and one of Chromis ternatensis had stomachs
filled with a great number of demersal fish eggs. This
finding is informative about feeding opportunism, but these
two specimens were not included for calculating the mean
diet composition. Relatively low variations are observed in
the stomach contents of each species (standard deviations
vary from 0 to 15% of the total content according to the
categories of prey).
There were significant differences in the gut contents of
the 13 species, as shown in Fig. 1 (ANOSIM global test
statistic R = 0.565, P \ 0.001). However, pairwise com-
parisons revealed that some species show nonsignificant
differences in diet (pairwise R \ 0.5, Table 2). According
to these results, the 13 species can be grouped into three
clusters (Fig. 1, Table 2). The three species of Abudefduf
and C. ternatensis form a first cluster where zooplankton
Fig. 1 Summarized dietary compositions of the 13 pomacentrid
species. For abbreviations of the species, see Table 1. The six prey
categories (phytoplankton, benthic algae, sessile invertebrates, vagile
invertebrates, zooplankton and detritus) are represented by different
color patterns
12 B. Fre
´
de
´
rich et al.
123
accounts for [95% of items in the stomach contents. In
these species, the prey were essentially pelagic copepods
but, to a lesser extent, chaetognaths were also found in
almost all of the stomachs of A. sexfasciatus and A.
sparoides (7 and 10 specimens, respectively). The second
group of species is composed of Pomacentrus caeruleus,
Chromis dimidiata and both Dascyllus species. Their
stomach contents reveal that they feed mainly on zoo-
plankton, mainly consisting of planktonic copepods (58%
in Dascyllus aruanus to 83% in P. caeruleus), but that they
also consume benthic algae and vagile benthic inverte-
brates in variable proportions according to each species
(Fig. 1). For example, vagile invertebrates (essentially
harpacticoid copepods) represent 38% of the items found in
D. aruanus stomachs, and benthic filamentous algae con-
stitute 28% of C. dimidiata stomach contents. Finally, the
cluster III groups species which consume mainly benthic
algae ([50% of the items in their stomach contents):
Plectroglyphidodon lacrymatus, Amphiprion akallopisos,
Pomacentrus baenschi, Pomacentrus trilineatus and
Chrysiptera unimaculata. This last species even showed
more than 90% of benthic algae in its stomach contents.
Within cluster III, P. baenschi was the species that had the
most varied stomach contents: zooplankton (22%), benthic
algae (53%), sessile invertebrates (19%) and vagile inver-
tebrates (4%).
Stomach contents revealed no size-related diet varia-
tions in Am. akallopisos, A. vaigiensis, D. trimaculatus and
Pl. lacrymatus (in these four species, linear regressions of
each prey category versus fish size were not significant;
P [ 0.05).
Stable isotope analysis. All of the food sources had
lower d
15
N values than the fish species (Fig. 2). Isotopic
values in algae (including turf algae) ranged from -17.5 to
-8.9% for d
13
C and from 2.0 to 6.1% for d
15
N. Zoo-
plankton had lower d
13
C and higher d
15
N than both algae
and zoobenthos.
There were highly significant differences between the
delta values of the 13 pomacentrid species (one-way
ANOVA: d
13
C, F
(12,121)
= 146.6, P \ 0.0001; d
15
N,
F
(12,121)
= 53.65, P \ 0.0001). The three species clusters
could be distinguished according to their mean d
13
C values
(Figs. 2, 3). The first group (A), which included Ch. uni-
maculata, Pl. lacrymatus and Am. akallopisos, showed the
least negative d
13
C values. Each Abudefduf species, the two
Chromis species, D. trimaculatus and P. caeruleus had the
most negative d
13
C values and comprised a second
assemblage (C). Pomacentrus trilineatus, P. baenschi and
D. aruanus formed an intermediate cluster (B) along the
d
13
C axis between the first two groups. Among all species,
P. trilineatus, Am. akallopisos and D. trimaculatus had the
greatest variation in d
13
C values. This d
13
C variation is
significantly correlated with the SL only in Am. akallopisos
(r
2
= 0.67; P \ 0.01), where the small specimens showed
lower d
13
C values than the largest. Dascyllus trimaculatus
and Ch. unimaculata, respectively, showed the most posi-
tive and the most negative d
15
N values. However, even
when significant differences existed between the mean
values of d
15
N of some species, none of the clusters could
be easily distinguished along this axis using this parameter.
Discussion
Stable isotopic analyses show that the 13 species can be
divided into three main feeding guilds according their
Table 2 Pairwise comparisons of stomach contents (ANOSIM)
A. sex A. spa A. vai Am. aka C. dim C. ter Ch.uni D. aru D. tri Pl. lac P. bae P. cae
A. spa –
A. vai ––
Am. aka ** ** **
C. dim –*–*
C. ter –––** –
Ch.uni ** ** ** – ** **
D. aru * ** * ** * * **
D. tri – – – * – – ** –
Pl. lac ** ** ** – * * – * *
P. bae ** ** ** – * ** * ** * –
P. cae – – – * – – ** – – * **
P. tri ** ** ** – * ** – ** ** – – **
For abbreviations of the species, see Table 1
** Highly significant differences between species (R [ 0.75), * overlapping but clearly different (R C 0.50)
– No differences (R \ 0.50)
Diet of damselfishes 13
123
foraging strategies: benthic feeders/foragers (group A),
pelagic feeders (group C) and an intermediate group (group
B), feeding on both pelagic and benthic prey (Figs. 2, 3).
This separation can be made along the d
13
C axis, which
represents a continuum of food sources from plankton (the
most negative values) to zoobenthos and algae (the least
negative values) (Bootsma et al. 1996). Pelagic feeders
(Abudefduf species, Chromis species, D. trimaculatus and
P. caeruleus) and benthic feeders (Ch. unimaculata, Pl.
lacrymatus and Am. akallopisos) had d
13
C values that
matched the isotopic signatures of, respectively, plankton
and benthic prey. Pomacentrus trilineatus, P. baenschi and
D. aruanus (Fig. 3) form an intermediate group in which
prey could come from benthic and pelagic food webs.
Generally speaking, enrichment in
15
N relative to assimi-
lated food is expected between two trophic levels (DeNiro
and Epstein 1981). In this study, the d
15
N values of the
fishes are higher than their potential food sources. Never-
theless, d
15
N does not allow us to discriminate between
different trophic niches. Dascyllus trimaculatus displays
the highest d
15
N values but also has a large size, and so
probably occupies a higher trophic position than other
pomacentrids.
Stomach content analysis provides more detail (e.g., the
kind of prey generally selected) about the diet, revealing
feeding preferences and some niche partitioning not shown
by stable isotopes. For example, stomach contents revealed
that planktonic copepods were the dominant prey of pela-
gic feeders (group C + D. aruanus), as they are for
the majority of diurnal planktivore fishes of coral reefs
(Hobson and Chess 1978). In this group, Abudefduf
sparoides, A. sexfasciatus, A. vaigiensis and C. ternatensis
can be considered exclusively zooplanktivorous species
(Fig. 1). Dascyllus aruanus is a carnivorous species that
eats both zooplankton and vagile benthic invertebrates,
explaining the intermediate position given by the stable
isotope analysis. Pomacentrus caeruleus, C. dimidiata and
D. trimaculatus feed mainly on zooplankton (group C), but
Fig. 2 Average (±SD) d
15
N
and d
13
C values of the 13
pomacentrid species and of food
items collected from the coral
reef. Isotopic signatures of each
benthic algae species are
summarized in one mean and a
SD that reflects their disparity.
For abbreviations of the species,
see Table 1. The groups A, B
and C of fishes are respectively
the benthic feeders/foragers, the
intermediate group and the
pelagic feeders (see the text and
Fig. 3). Pomacentrid species
belonging to the groups A, B or
C and the potential primary food
sources (zooplankton,
zoobenthos and algae) are
represented by different icons.
POM is the particulate organic
matter
14 B. Fre
´
de
´
rich et al.
123
