Journal of Fish Biology (2003) 62, 1137́–1158

doi:10.1046/j.1095-8649.2003.00108.x, available online at http://www.blackwell-synergy.com

Dietarý–morphological relationships in a fish

assemblage of the Bolivian Amazonian floodplain
M. P O U I L L Y *†‡, F. L I N O †, J.-G. B R E T E N O U X * AND
C. R O S A L E S †
*Institut de Recherche pour le De´veloppement (IRD), Universite´ Lyon 1, Laboratoire
d’Ecologie des Hydrosyste`mes Fluviaux, 43, Bd du 11 Novembre 1918, F-69622
Villeurbanne Cedex, France and †Instituto de Ecologı´a, Universidad Mayor
de San Andre´s, Bolivia, CP 10077, La Paz, Bolivia

(Received 24 January 2002, Accepted 7 April 2003)

Morphological correlates of diet were examined in 48 species of freshwater fishes from

floodplain lakes in the central part of the Mamoré River (Bolivian Amazon). The species
were classified, according to the percentage occurrence of seven food items, into eight broad
trophic categories: mud feeders, algivores, herbivores, terrestrial invertivores and omnivores,
carnivores, zooplanktivores, aquatic invertivores and piscivores. There were significant rela-
tionships between the diet and morphology of the fishes even when the effect of taxonomical
relatedness between species was eliminated. Relative gut length was the main morphological
variable used to order species on a carnivore to mud feeder gradient. Standard length and
head and mouth size were the morphological variables most closely associated with prey size.
Mud feeder, algivore and piscivore species appeared as the most dietary and morphologically
specialized. These results support both the hypotheses that species morphology influences the
diet and that morphological similarity is conserved even in comparison with taxonomically
unrelated species. # 2003 The Fisheries Society of the British Isles

Key words: Bolivia; diet; ecomorphology; river floodplain; tropical freshwater fishes.

INTRODUCTION
A correlation between morphological attributes and diet has been sought in
testing ecomorphological hypotheses in fish assemblages. The results of studies
on diet and morphological relationships in temperate and tropical river fish
assemblages, however, have excluded the idea of a general pattern. Some studies
have revealed strong relationships between diet and morphology (Gatz, 1979;
Wikramanayake, 1990; Winemiller et al., 1995; Piet, 1998; Hugueny & Pouilly,
1999; Xie et al., 2001) while others found relatively weak and indistinct relation-
ships (Grossman, 1986; Douglas & Matthews, 1992; Motta et al., 1995;
Winemiller & Adite, 1997). Some relationships between morphological attributes
and diet have been encountered repeatedly, e.g. gut length is frequently positively

‡Author to whom correspondence should be addressed. Tel.: þ33 4 72446299; fax: þ33 4 72431141;
email: pouilly@univ-lyon1.fr

1137
# 2003 The Fisheries Society of the British Isles
1138 M. POUILLY ET AL.

correlated with herbivory, gill raker length is often associated with the con-
sumption of plankton and fish size (and related head and mouth size) is com-
monly associated with prey size. Relationships also exist between morphological
attributes and feeding behaviour, e.g. the orientation of the mouth and the
position of the eyes are linked with the position of the fish relative to its food
(Gatz, 1979).
At present, even though many studies have reported strong relationships
between species ecology and morphology, no powerful predictive model exists
for fish assemblages. This discrepancy may be partially due to the importance of
fish behaviour in determining the type of prey that can be used (Grossman,
1986), the relatively high level of opportunism in freshwater fish species
(Hugueny & Pouilly, 1999), the influence of phylogeny on limiting morpho-
logical adaptation (Douglas & Matthews, 1992; Motta et al., 1995) and the
selection of morphological and dietary variables.
Ecomorphological hypotheses include two main considerations: (1) species
morphology is likely to be similar within an ecological group and to differ
between ecological groups depending on the nature of the resource used and
the strategy developed by the species to use it and (2) the morphological
variations correspond to a response to selective pressure and result in the
phenomenon of convergence: ‘morphological similarity of phylogenetically
unrelated species’ (Winemiller, 1991).
The present study investigated the trophic ecology and morphology of
48 dominant fish species inhabiting lakes of the floodplain of the Mamoré
River, one of the principal drainages of the Bolivian Amazon.

MATERIALS AND METHODS

STUDY AREA
The Bolivian Amazonian plain is situated in the upper watershed of the Madera River,
one of the primary tributaries of the Amazon River (Fig. 1). The majority of this territory
(c. 550 000 km2) comprises the ‘Llanos de los Moxos’ (province of Beni), a landscape
dominated by savannah with some patches of forest confined to the higher part of the
plain, and forest galleries (bands of vegetation) along the rivers. The Mamoré River
drains the south Bolivian Andes and >85% of the Moxos savannah area, and corres-
ponds to a large floodplain system with a potential flood extension of c. 150 000 km2
(Denevan, 1980). Local climatic conditions are marked by the alternation of a wet season
(October to March) and a dry season (April to September). A substantial annual flood
(maximum flow in the main channel in February 1987 was 8010 m3 s
1) generally occurs
at the end of the wet season (December to April) and can last as long as 3 or 4 months
(Loubens et al., 1992).
The study area is situated in the central part of the Mamoré River (between 14 300 and
14 520 S and 64 510 and 65 010 W) near the city of Trinidad (capital of Beni province).
Loubens et al. (1992) described the aquatic habitats of the Mamoré River floodplain. The
study site included eight lakes in the central Mamoré River floodplain that corresponded
to four different ecological habitats: oxbow lakes at three locations (near the river, in the
middle of the forest ‘band’ and at the limit between forest and savannah) and savannah
lakes. Pouilly et al. (1999) presented a preliminary description of limnological para-
meters, plankton and fish communities.

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1139

82° 78° 74° 70° 66° 62° 58° 54° 50°

12°

8°

4°

0° Amazon Basin

4°

8° Rio Madera

12° Beni
Study area
Rio Mamoré

16°

0 500 1000 km

Projection Lambert’s Azimuthal Bolivia

20°

FIG. 1. Map of the Amazon watershed showing the Bolivian part of the upper Madera basin (Beni plain
and Andes) and the study area in the central Mamoré River.

FI S H S A M P L I N G
Fishes were sampled using 13 gillnets (25 m long and 2 m depth) with a wide range of
mesh sizes (10, 15, 20, 25, 30, 35, 40, 50, 60, 70, 80, 90 and 110 mm). Sampling was
conducted during eight periods between March 1998 and March 2000. For each sampling
(lake and period), gillnets were left in place for 2 h in the evening (1700́–1900 hours) and
for 2 h in the morning (0500́–0700 hours). They were placed near the shore and their
locations were approximately the same throughout the study.
Captured fishes were preserved in buffered formaldehyde (4%), transported to the
laboratory and then placed in buffered alcohol (75%). They were identified at the species
level (or at the genus level if systematic knowledge was inadequate). Identification was
based on voucher specimens left by a previous systematic research programme (Lauzanne
& Loubens, 1985; Lauzanne et al., 1991) at the Trinidad fish collection (CIRA-UTB), the
Museo Nacional de Historia Natural, La Paz and at the Musée National d’Histoire
Naturelle, Paris.
The 48 species analysed represented 89% of the total number of individuals captured
during the study (M. Pouilly, unpubl. data). Species were selected in order to present
a broad range of systematic groups. The 48 species studied belonged to five main
orders (Characiformes, Siluriformes, Gymnotiformes, Perciformes and Clupeiformes)
and represented 18 families out of the 26 registered during the study (six families of

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1140 M. POUILLY ET AL.

Characiformes out of eight collected; seven families of Siluriformes out of nine; two
families of Gymnotiformes out of five; one family of Perciformes out of two; two families
of Clupeiformes out of two collected).
Changes in fish diet and morphology depend on the stage of development and the size
of the fish (Mérigoux & Ponton, 1998). Consequently, for each species, a range of sizes
that were assumed to correspond to the adult stage was studied. All the fishes included
in the morphological analysis were also used for the analysis of diet. In order to optimize
the estimation of diet in a given species, however, extra fishes captured were included in
the diet analysis.

FISH DIET ANALYSIS

Estimation of diet was based on the analysis of stomach contents. After identification
of the fish, the stomach was extracted by dissection. Empty stomachs or stomachs with
almost fully digested contents were eliminated. The contents of the remaining stomachs
were examined under a microscope and items were separated into seven categories: soft
substratum (MUD); algae or periphyton (ALG); aquatic or terrestrial vegetation, fruits
or seeds (VEG); zooplankton (cladocerans, rotifers or copepods, ZOO); aquatic inverte-
brates (AIN); terrestrial invertebrates (TIN); fishes (FISH).
The invertebrate categories (terrestrial and aquatic) mainly corresponded to insect
prey. The soft substratum category did not correspond to a biological feeding resource.
It is likely that fishes that ingest soft substratum are in fact consuming periphyton and
algae aggregated to the substratum and other associated microorganisms. Soft substra-
tum, however, was conserved as an indicator for a particular feeding habit. Following
Sibbing & Nagelkerke’s (2001) classification, implicit prey size categories were used for
the interpretation: ALG ¼ MUD < ZOO < AIN ¼ TIN < VEG < FISH.
In order to obtain a general qualitative diet for a given species, results were expressed
by the occurrence method. The relative frequency of an item in the diet was estimated by
the number of stomachs that contained that item divided by the total number of non-
empty stomachs analysed in that particular species.

MORPHOLOGICAL MEASUREMENTS
Following the results of other authors who have examined fish ecomorphology (Gatz,
1979; Watson & Balon, 1984; Winemiller, 1991) and especially diet ecomorphology
(Motta et al., 1995; Piet, 1998; Hugueny & Pouilly, 1999), 10 morphological variables
associated with prey capture and feeding strategy were selected and were measured on
26́–30 individuals of each species. Quantitative morphological variables were entered as a
ratio of standard length (LS) in the analysis in order to reduce the allometric (size-
dependency) effect and to standardize data for body size differences (Winemiller, 1991).
The 10 variables were: LS, the distance from the tip of the snout to the last vertebra;
relative head length (HEAD), the ratio between head length (distance from the tip of the
snout to the posterior edge of the opercle) and LS; relative mouth height (MOHE), the
ratio between mouth height at maximum extension and LS. These three variables were
assumed to be directly linked with prey size: fish with a larger head and mouth gap are
able to handle larger prey (Gatz, 1979). These variables were expected to be positively
correlated with larger items (VEG, AIN, TIN and FISH) and negatively correlated with
smaller items (ALG, ZOO and MUD); mouth orientation (MORI), coded following
Watson & Balon (1984) and Hugueny & Pouilly (1999) (1, dorsal; 2, terminal; 3, ventrally
oblique; 4, ventral). The position of the mouth was assumed to indicate the position of
food relative to the fish (Gatz, 1979; Hugueny & Pouilly, 1999). Fishes with a dorsal or a
terminal mouth were assumed to obtain their food from the pelagic area (ZOO, TIN and
FISH), while fishes with a ventral mouth were thought to feed on benthic resources
(ALG, MUD and AIN); relative eye diameter (EYDI), the ratio of the horizontal
diameter of the eye and LS. Eye size was assumed to be positively correlated with the
importance of vision for feeding (Gatz, 1979). It may also indicate vertical position in the
water column, because species that inhabit deeper water tend to have smaller eyes

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1141

(Wikramanayake, 1990); eye vertical position (EYPO), the ratio between the vertical
distance from the middle of the eye to the lower limit of the head and the total head depth
measured at eye level. EYPO was expected to indicate the foraging position of the species
in the water column. Nektonic fishes were assumed to have lateral eyes, while benthic
fishes were expected to have more dorsally located eyes (Gatz, 1979; Hugueny & Pouilly,
1999); teeth (TEETH) are considered to be a good index of diet (Gatz, 1979), but no
simple measure exists and different coding systems based on number and shape have been
proposed (Gatz, 1979; Motta et al., 1995). The number and shape of the premaxillar and
mandibular teeth were combined into one code to describe the mode of intake of the diet
[0, lack of teeth (associated with suction or snapping up of passive prey); 1, few teeth
(<50) of conical, insiciform or multicuspidal shape (associated with biting or tearing);
2, numerous teeth (>50), filiforms (associated with scraping); 3, band of numerous teeth
(associated with suction or snapping up of active prey)]; number of gill rakers on the first
gill arch (GRNU) and relative maximum length of gill rakers on the first gill arch
(GRLE), the ratio of the length of the longest gill raker of the first gill arch and LS.
GRNU and GRLE were assumed to be positively linked with filtration activity and
indirectly indicated planktivorous species (Motta et al., 1995); relative length of the
digestive tract (GUTL), the ratio between the total length of the digestive tract and LS.
GUTL was assumed to be higher in herbivores and detritivores, and lower in carnivores
(Gatz, 1979; Hugueny & Pouilly, 1999).

ST A TI S T ICA L A NA LY S IS
Fish diets were compared and grouped by cluster analysis of the trophic distance
matrix (UPGMA algorithm; Legendre & Legendre, 1998). Trophic distances between
the 48 species were calculated from species scores along the three first axes of a corres-
pondence analysis (explaining 76% of the total variability) of the species-food items table
(Table I). Correspondence analysis is an ordination method, based on w2, recom-
mended for analysis of occurrence data (Legendre & Legendre, 1998). The first few
axes (generally the first three or four) model the majority of variation of the matrix
(Winemiller, 1991) and extract the main components of ordination of the matrix.
In order to show the relative level of diet specialization of the species, diet breadth of a
given species was estimated by Levin’s standardized index Bi ¼ [(j Pij2 )
1
1](n
1)
1
where Pij is the proportion of prey j in the diet of predator i and n is the number of prey
categories (Hurlbert, 1978). The index ranges from 0 (diet specialized on few prey items) to 1
(generalist diet).
Relationships between fish diet and morphology were studied by canonical correspond-
ence analysis, a non-linear ordination method specially adapted to explore unimodal
relationships (CCA, Ter Braak, 1986). CCA is a correspondence analysis in which the
axes are a linear combination of explanatory variables (morphological variables). It
allows measurement of the amount of variation in the trophic data that can be explained
by the linear combination of the morphological variables. The statistical significance of
the diet́–morphology correlations extracted by the CCA was estimated by a permutation
test (1000 simulations).
Diet́–morphology relationships could be an artifact derived from the phylogenetic
distance between the species (Winemiller, 1991; Douglas & Matthews, 1992; Hugueny
& Pouilly, 1999). Phylogenetic distances were not available for all Amazonian freshwater
fish groups. Following Hugueny & Pouilly (1999), the phylogenetic link between two
species was approximated by means of an ordinal taxonomic distance. A value of 1 was
set for congeneric species, 1́5 for consubfamilial species, 2 for confamilial species, 3 for
species from the same order and 4 for species from different orders. Statistical compari-
sons of data matrices were performed in order to quantify the agreement between the
diet, morphology and taxonomy of the fish. The non-parametric Mantel test was used to
evaluate the null hypothesis which assumes that the distance among fishes in one matrix
is not correlated with the corresponding distance in another matrix (Legendre &
Legendre, 1998). By extension, a similar procedure could be performed to evaluate the
partial correlation between two matrices while controlling for the effect of a third one

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1142 M. POUILLY ET AL.

TABLE I. Diet composition and breadth (Levin’s index, B) of 48 fish species in the central
Mamoré River expressed by % occurrence of seven food categories. MUD, soft sub-
stratum; ALG, algae and periphyton; AIN, aquatic invertebrates; TIN, terrestrial inver-
tebrates; ZOO, zooplankton; FISH, fishes; VEG, aquatic and terrestrial vegetation.
Species are ordered according to the results of cluster analysis (see Fig. 2). See Table IV
for species abbreviations

Number of
Species code stomachs MUD ALG VEG INT INA ZOO FISH B

Zooplanktivores
HYAMA 17 0́00 23́53 5́88 0́00 11́76 58́82 11́76 0́315
ENTBE 91 0́00 0́00 7́69 31́87 58́24 43́96 0́00 0́357
APYAN 95 0́00 30́53 2́11 9́47 97́89 89́47 0́00 0́305
AGEBR 39 0́00 0́00 0́00 0́00 23́08 97́44 0́00 0́075
Mud feeders
POTAL 38 55́26 52́63 2́63 0́00 0́00 5́26 0́00 0́215
CUEAI 165 70́91 68́48 3́03 0́00 0́00 0́00 0́00 0́181
PSHLA 20 70́00 100́00 0́00 0́00 0́00 0́00 0́00 0́157
PSECU 60 58́33 80́00 11́67 0́00 5́00 0́00 0́00 0́235
POTLA 211 40́76 77́73 0́00 0́00 0́00 5́21 0́00 0́163
Algivores
RHYMI 26 0́00 88́46 11́54 7́69 3́85 0́00 0́00 0́091
PSEAM 51 0́00 100́00 0́00 0́00 0́00 0́00 0́00 0́000
STUNI 52 9́62 100́00 0́00 0́00 0́00 0́00 0́00 0́032
HYTJO 70 5́71 92́86 0́00 1́43 0́00 0́00 0́00 0́026
ENGSP 68 5́88 95́59 0́00 0́00 7́35 0́00 0́00 0́047
LOIMA 25 28́00 88́00 12́00 0́00 16́00 0́00 0́00 0́220
ANOME 144 25́69 90́28 0́00 0́00 6́94 4́17 0́00 0́137
Herbivores
LEOFR 10 0́00 50́00 80́00 20́00 10́00 0́00 10́00 0́340
DORSP 58 18́97 79́31 82́76 5́17 34́48 12́07 1́72 0́450
SCHFA 43 2́33 30́23 74́42 0́00 4́65 0́00 0́00 0́154
Terrestrial invertivores and omnivores
STTCR 68 0́00 10́29 32́35 64́71 32́35 0́00 1́47 0́353
POPCO 52 1́92 0́00 28́85 69́23 25́00 0́00 7́69 0́298
TRPAL 94 0́00 8́51 31́91 74́47 26́60 3́19 2́13 0́321
PAUST 18 0́00 5́56 11́11 55́56 33́33 0́00 27́78 0́412
PIUMA 140 0́71 25́71 23́57 37́14 53́57 5́00 37́14 0́644
TRPAN 277 0́36 19́86 20́94 53́79 55́96 9́75 1́44 0́463
MOEDI 241 0́41 8́30 5́81 46́06 60́17 26́56 0́41 0́389
Aquatic invertivores
OPSSP 27 11́11 40́74 48́15 0́00 77́78 11́11 0́00 0́412
EIAVI 57 0́00 52́63 42́11 14́04 91́23 5́26 3́51 0́388
MYLDU 18 5́56 11́11 44́44 5́56 38́89 5́56 0́00 0́389
CTEPI 16 0́00 6́25 50́00 6́25 62́50 6́25 6́25 0́313
TRYPA 25 12́00 4́00 24́00 4́00 96́00 16́00 0́00 0́230
CENSP 32 0́00 0́00 9́38 28́13 87́50 0́00 0́00 0́138
BROSL 7 0́00 14́29 14́29 14́29 85́71 0́00 0́00 0́179
AUCNU 26 0́00 3́85 0́00 7́69 96́15 3́85 0́00 0́055

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1143

PAECY 26 3́85 30́77 3́85 0́00 76́92 0́00 3́85 0́176

ADOSA 27 0́00 18́52 3́70 3́70 85́19 7́41 0́00 0́138
Carnivores
PYGNA 87 0́00 2́30 32́18 12́64 5́75 1́15 62́07 0́275
TYMSP 55 1́82 1́82 9́09 18́18 16́36 5́45 67́27 0́291
SEREI 32 0́00 6́25 21́88 9́38 34́38 0́00 59́38 0́374
ROBMY 33 0́00 0́00 0́00 18́18 75́76 6́06 42́42 0́261
ROBAF 154 0́00 1́30 0́65 2́60 55́19 9́74 61́69 0́246
PELFL 120 0́00 0́00 2́50 6́67 52́50 3́33 66́67 0́231
Piscivores
HOPMA 28 21́43 10́71 0́00 7́14 3́57 0́00 71́43 0́213
SERRH 76 0́00 1́32 5́26 7́89 10́53 0́00 94́74 0́094
PLGSQ 80 0́00 0́00 5́00 2́50 15́00 0́00 90́00 0́086
SERHO 90 0́00 2́22 11́11 2́22 6́67 1́11 90́00 0́092
CAOMA 28 0́00 3́57 14́29 3́57 7́14 0́00 85́71 0́119
ACEAL 51 1́96 1́96 1́96 3́92 0́00 0́00 96́08 0́035

(Smouse et al., 1986). In the present case, to factor out the taxonomic effect, a partial
Mantel statistic was calculated between the matrices of residuals of morphology and
taxonomy on the one hand and between diet and taxonomy on the other (Douglas &
Matthews, 1992; Hugueny & Pouilly, 1999). Statistical significance was estimated by a
permutation test (1000 simulations). Because of the multiple comparisons, the Bonferroni
technique was used to assign the level of significance (a ¼ 0́05, divided by 3 ¼ 0́017)
(Douglas & Matthews, 1992). The trophic distance matrix was the same as in cluster
analysis. Similarly, the morphological distances were generated using species scores on
the first three axes (explaining 73́8% of total variability) of a principal component
analysis on the correlation matrix of morphological data (Winemiller, 1991; Douglas &
Matthews, 1992).

RESULTS

D I E T C O M P O S I T I O N A N D C LA S SI F I C A T I O N
Diet varied among the species, and cluster analysis allowed eight groups
of species showing similar diet profiles to be identified (Table I and Fig. 2);
zooplanktivores: Hypophthalmus marginatus Valenciennes, Entomocorus benja-
mini Eigenmann, Aphyocharax anisitsi Eigenmann & Kennedy and Ageneiosus
brevis Steindachner fed primarily (>40%) on zooplankton. Ageneiosus brevis
appears as the only species that is highly specialized in the consumption of
zooplankton, whereas the other species completed their diet with aquatic inver-
tebrates (E. benjamini and A. anisitsi) or algae (H. marginatus and A. anisitsi).
Diet breadth (Levin’s index, B) reflected the difference in specialization. It was
equal to 0́075 for A. brevis and ranged between 0́305 and 0́357 for the other
three species; mud feeders: Potamorhina altamazonica (Cope), Potamorhina
latior (Spix & Agassiz), Curimatella alburna (Müller & Troschel), Pseudo-
hemiodon laticeps (Regan) and Psectrogaster curviventris Eigenmann & Kennedy
showed a similarly specialized diet based on algae and soft substratum asso-
ciated with a low diet breadth (B between 0́16 and 0́24); algivores: Rhytiodus

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1144 M. POUILLY ET AL.

HYAMA
ENTBE Zooplanktivores
APYAN
AGEBR
RHYMI
PSEAM
STUNI
HYTJO Algivores
ENGSP
LOIMA
ANOME
POTAL
CUEAI Mud feeders
PSHLA
PSECU
POTLA
LEOFR
DORSP Herbivores
SCHFA
STTCR
POPCO
TRPAL
PAUST Terrestrial invertivores
PIUMA
TRPAN
MOEDI
OPSSP
EIAVI
MYLDU
CTEPI
TRYPA Aquatic invertivores
CENSP
BROSL
AUCNU
PAECY
ADOSA
PYGNA
TYMSP
SEREI Carnivores
ROBMY
ROBAF
PELFL
HOPMA
SERRH
PLGSQ
SERHO Piscivores
CAOMA
ACEAL

FIG. 2. Cluster analysis dendrogram (UPGMA algorithm) of trophic distances between 48 fish species in
the central Mamoré River floodplain. (See Table IV for species abbreviations.)

microlepis Kner, Psectrogaster amazonica Eigenmann & Eigenmann, Sturisoma

nigrirostrum Fowler, Hypoptopoma joberti (Vaillant), Engraulidae, Loricari-
ichthys maculatus (Bloch) and Anodus melanopogon Agassiz fed exclusively on
algae associated with a low diet breadth (B between 0́00 and 0́2); herbivores:
Leporinus friderici friderici (Bloch), Doras sp. and Schizodon fasciatus Spix &
Agassiz fed on vegetation and algae. Leporinus friderici friderici and Doras sp.
also consumed invertebrates (AIN and TIN) but only as a small proportion of
their diet. These food items were probably complementary or resulted from
feeding on aquatic macrophytes. Diet breadth reflected the difference in special-
ization between species of this group (B ¼ 0́15 for S. fasciatus and B > 0́34 for
the two other species); terrestrial invertivores and omnivores: Stethaprion
crenatum Eigenmann, Poptella compressa (Günther), Triportheus albus Cope,
Triportheus angulatus (Spix & Agassiz), Parauchenipterus striatulus (Steindachner),
Pimelodus gr. maculatus-blochii Valenciennes and Moenkhausia dichroura (Kner)

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1145

mainly fed on invertebrates and especially on terrestrial invertebrates. They also

consumed other items such as fishes (P. striatulus and P. maculatus-blochii),
zooplankton (M. dichroura) and vegetation (S. crenatum, P. compressa, T. albus,
T. angulatus and P. maculatus-blochii). Although these species shared a high rate
of consumption of TIN, they also presented noticeable variations in their
complementary diet that resulted in this group being classified as omnivores.
The diet breadth (B between 0́3 and 0́64) confirmed that no species presented a
high level of specialization in the consumption of terrestrial invertebrates;
aquatic invertivores: Opsodoras sp., Eigenmannia virescens (Valenciennes),
Mylossoma duriventre (Cuvier), Ctenobrycon spilurus (Valenciennes), Trachy-
doras paraguayensis (Eigenmann & Ward), Centromochlus sp., Brochis splendens
(Castelnau), Auchenipterus nuchalis (Spix & Agassiz), Parecbasis cyclolepis
Eigenmann and Adontosternarchus sachsi (Peters) fed predominantly on aquatic
invertebrates (>60% of occurrence except in the case of M. duriventre). It was
possible, however, to distinguish two trends. Trachydoras paraguayensis, Cen-
tromochlus sp., B. splendens, A. nuchalis, P. cyclolepis and A. sachsi had a highly
specialized diet (>75% of occurrence of AIN and secondary resources <31%).
Conversely, Opsodoras sp., E. virescens, M. duriventre and C. spilurus consumed
vegetation as a complementary resource (>40% of occurrence for the four
species) as well as algae (>40% of occurrence for Opsodoras sp. and E. virescens).
Diet breadth confirmed the distinction between a group of five more
specialized species (B between 0́06 and 0́18) and four species with a more
omnivorous diet (B between 0́32 and 0́41). Trachydoras paraguayensis was in
an intermediate position between the two trends (B ¼ 0́23); carnivores: Pygo-
centrus nattereri Kner, Tympanopleura sp., Serrasalmus eigenmanni (Norman),
Roeboides myersii Gill, Roeboides affinis (Günther) and Pellona flavipinnis
(Valenciennes) consumed fishes as their primary resource (42́–68% of occur-
rence) and invertebrates (S. eigenmanni, R. myersii, R. affinis and P. flavipinnis)
or vegetation (P. nattereri and S. eigenmanni) as their main complementary
resource. Diet breadth reflected the distinction between the two groups. Three
more specialized species (R. myersii, R. affinis and P. flavipinnis) presented a
lower diet breath (B between 0́23 and 0́26) with two co-dominant resources
(FISH and AIN). The three other species (P. nattereri, S. eigenmanni and
Tympanopleura sp.), which were less specialized, presented a higher diet breath
(B between 0́28 and 0́37); piscivores: Serrasalmus rhombeus (L.), Serrasalmus
hollandi Eigenmann, Plagioscion squamosissimus (Heckel), Calophysus macro-
pterus (Lichtenstein) and Acestrorhynchus cf. altus Menezes comprised a special-
ized group (B between 0́1 and 0́12) with a high percentage of fish consumption
(>85%). Hoplias malabaricus (Bloch) was also associated with this group but
presented secondary consumption of soft substratum and algae, and thus a
higher diet breadth (B ¼ 0́21).

D I E T́ – M O R P H O L O G Y R E L A T I O N SH I P S
The CCA revealed a significant relationship between diet and morphology
(P < 0́01, tested with 1000 permutations). Interpretation was conducted on the
three first axes, accounting for 90́5% of the total diet variation explained by the
morphological variables (Table II). The fraction of variation of each food item

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1146 M. POUILLY ET AL.

TABLE II. Results of canonical correspondence analysis (CCA) between diet and morphology
of 48 fish species of the Mamoré River

Axis 1 Axis 2 Axis 3

Variance of dietary data (%) 23́9 14́6 7́6

Cumulated variance of dietary data 38́5 46́1
Variance of diet́–morphology relationship (%) 46́8 28́7 15́0
Cumulated variance of diet́–morphology relationship 75́5 90́5
Canonical Eigenvalue CCA 0́46 0́28 0́15
Diet́–morphology relationship (r2) 0́71 0́58 0́52

explained by morphology was 32́4% for terrestrial invertebrates, 38́4% for

zooplankton, 44́8% for aquatic invertebrates, 45́8% for vegetation, 57́6% for
algae, 58́4% for fishes and 68́2% for soft substratum. These results showed
that terrestrial invertebrate and zooplankton consumption had the lowest
associations with morphology. Conversely, soft substratum, algae and fishes
appeared to be consumed by species that had more specific morphology.
The axes derived from the CCA provided a geometric space that highlighted
the distribution pattern of species according to the relation between their
morphology and diet. Mud feeders and algivores were distributed in the right
part of the first CCA graph (scatter plots of axes 1 and 2) and were associated
with a large relative gut length (GUTL), which thus appeared as the most
important morphological variable to distinguish microphagous species (mud
feeders and algivores) from the others (Fig. 3). Carnivores and piscivores were
mainly distributed in the upper left part of the first CCA graph (Fig. 3). These
species were characterized by large size (LS), mouth height (MOHE) and head
length (HEAD). Zooplanktivores were positioned on the upper left part of the
second CCA graph (scatter plots of axes 2 and 3, Fig. 4). They were mainly
characterized by a higher number of gill rakers of larger size. Herbivores that
consumed macrophytic vegetation tended to be positioned on the lower left part
of the second CCA graph, and were characterized by a larger eye diameter.
Invertivores were mostly positioned on the lower left part of the CCA graphs
but did not show clear differentiation that could be linked to a particular
morphological feature. Teeth characteristics (TEETH) appeared to be of
secondary importance and showed a negative correlation with consumption of
mud and algae. This result indicated that the codification used enabled species
without teeth (consuming mud or algae) to be approximately distinguished from
the others. Eye position (EYPO) and mouth orientation (MORI) also appeared
as secondary morphological features to explain fish diet, and presented no clear
pattern associated with consumption of a particular diet item.
In the species ordination diagram, species were positioned at the centroid of
their morphological attributes (Figs 3 and 4). Their positions showed overall
similarity with the diet classification resulting from the cluster diagram (Fig. 2).
Some species, however, appeared positioned with a morphology that did not
correspond to their observed diet. For example, P. nattereri was located at the
extremity of the two first axes and thus appeared to be more morphologically

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1147

FISH Axis 2
(a)

MOHE
LS

MUD
HEAD EYPO

GUTL
MORI
TEETH ALG Axis 1

EYDI GRLE GRNU

INT
VEG

INA

ZOO

PYGNA Axis 2
PLGSQ (b)

HOPMA

PELFL
ACEAL
SERRH

POTAL
SERHO SEREI POTLA PSECU
CAOMA
LOIMA
PAUST
ENGSP
PIUMA
ROBMY STUNI
ANOME
STTCR TRYPA Axis 1
PSEAM PSHLA
SCHFA RHYMI CUEAI
TRPAN
AGEBR CENSP TRPAL LEOFR OPSSP HYAMA
POPCO MYLDU
APYAN EIAVI DORSP ADOSA
TYMSP PAECY HYTJO
BROSL
ROBAF
ENTBE MOEDI

AUCNU

CTEPI

FIG. 3. Graphical output of the axes 1 and 2 of canonical correspondence analysis (CCA) linking diet
(seven food items) and morphology (10 attributes) of 48 fish species of the central Mamoré River.
Ordination plots for morphological attributes and food items (a) and species (b) are presented
separately to avoid cluttering. Species are presented by diet categories: , zooplanktivores; *, mud
feeders; , algivores; ~, herbivores; &, terrestrial invertivores; &, aquatic invertivores; , carnivores;
, piscivores. (See Tables I and IV for abbreviations.)

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1148 M. POUILLY ET AL.

ZOO Axis 3
(a)

MOHE

GUTL
GRNU GRLE
TEETH

ALG FISH
EYPO Axis 2
INT
INA MUD
MORI
HEAD

VEG

LS

EYDI

HYAMA Axis 3
AGEBR (b)

HYTJO

ENGSP
STUNI

ENTBE
ANOME
HOPMA
TYMSP
ROBMY
PYGNA
ROBAF PAUST
APYAN STTCR
AUCNU PIUMA
CENSP SERHO
MOEDI
PSHLA ACEAL Axis 2
CAOMA
PAECY ADOSA PELFL
CUEAI SERRH
TRPAN PSECU
POPCO TRPAL PSEAM LOIMA
DORSP POTAL
CTEPI SEREI
RHYMI POTLA
EIAVI
OPSSP PLGSQ
BROSL MYLDU
TRYPA

LEOFR

SCHFA

FIG. 4. Graphical output of the axes 2 and 3 of canonical correspondence analysis (CCA) linking diet
(seven food items) and morphology (10 attributes) of 48 fish species of the central Mamoré River.
Ordination plots for morphological attributes and food items (a) and species (b) are presented
separately to avoid cluttering. Species are presented by diet categories (see Fig. 3). (See Tables I and
IV for abbreviations.)

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1149

TABLE III. Correlation coefficient and probabilities for Mantel test and partial Mantel
test on diet, morphology and taxonomy distances matrix between 48 fish species of the
Mamoré River

Covariable r (Mantel) P*

Diet́–morphology 0́263 0́006

Diet́–taxonomy 0́102 0́025
Morphologý–taxonomy 0́406 0́001
Diet́–morphology Taxonomy 0́244 0́011
*
Bonferroni-adjusted probability ¼ 0́017 (0́05 divided by 3).

specialized for piscivory than other species that eat a higher proportion of
fishes. Sturisoma nigrirostrum was located at the extremity of the second axis,
and could thus be considered as morphologically specialized for detritivory
despite the fact its diet is based on algae. Two carnivorous species Tympano-
pleura sp. and R. affinis were located in a pool of invertivores and zooplankti-
vores distributed in the lower left part, and thus appeared to not present a high
correlation with any of the morphological variables.
The result of Mantel and partial Mantel tests showed the correlation coeffi-
cient between the trophic, morphologic and taxonomic distances matrix to be
significant, except for the diet and taxonomy comparison (P ¼ 0́025, evaluation
by a test of 1000 permutations, Table III). Trophic distances between species
were positively correlated with morphological distances. Morphological dis-
tances were also positively correlated with taxonomic distances, suggesting
that the observed diet́–morphology relationships may be the result of a taxo-
nomic link between species. The partial Mantel test using taxonomic distance as
a covariate showed that the diet́–morphology relationship was still significant
(P ¼ 0́011) when controlling for the effect of taxonomy.

DISCUSSION

D I E T C O M P O S I T I O N A N D C LA S SI F I C A T I O N
Forty-eight fish species of the Mamoré River floodplain were classified into
eight trophic categories: mud feeders, algivores, herbivores, terrestrial inverti-
vores and omnivores, aquatic invertivores, zooplanktivores, carnivores and
piscivores. The trophic categories defined in this study correspond to a large
range of categories presented in other Amazonian fish trophic studies (Marlier,
1968; Knöppel, 1970; Soares, 1979; Goulding, 1980; Soares et al., 1986). The
high level of taxonomic diversity of the Mamoré River floodplain (c. 140
species, 26 families) corresponds to a high level of trophic diversity even when
only the most abundant species are taken into consideration. ‘Forest feeders’ is
the only main category absent from the present data that is generally repre-
sented in the Amazon by herbivorous species such as Colossoma macropomum

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1150 M. POUILLY ET AL.

(Cuvier). Although this species is present in the central Mamoré region (Lauzanne
et al., 1991), it was not abundant in the present sampling.
From the point of view of species-specific diet, it is difficult to compare
studies, not only because of differences in methodology but also because the
patterns of resource availability are not presented. Species that present a broad
distribution and a rather specialized diet, however, are generally consistently
classified. Species belonging to the Curimatidae are widely recognized as micro-
phytophages and classified as detritivores (Soares et al., 1986) or mud feeders
(Goulding, 1980) or separated, as in the present results, into mud feeders and
phytophages (Marlier, 1968). Species of the genus Hypophthalmus have been
shown to be zooplanktivores (Marlier, 1968; Carvalho, 1980). Species of the
genus Leporinus and S. fasciatus are described as herbivorous species (Soares
et al., 1986) or omnivores (Goulding, 1980). Rhytiodus microlepis is classified as
algivorous by Soares et al. (1986). A majority of piscivorous species have also
been described as specialized and as having an unvarying diet dominated by fish
prey: A. cf. altus (Menezes, 1969), H. malabaricus (Soares et al., 1986) and
P. squamosissimus (Goulding, 1980). Species that are highly omnivorous are also
classified consistently, such as species of the genera Triportheus and Pimelodus
(Goulding, 1980; Soares et al., 1986).
A majority of the fishes analysed in the Mamoré River floodplain were shown
to have a diet (at the general level of classification used here) similar to other
South American systems including white, clear and black waters. Food resource
availability is likely to be different in each kind of system, so it may be assumed
that fish diet not only depends on the availability of different food items but
also on the capacity of fishes to capture and make use of this food. Apparently,
specialized species do not modify their diet from one system to another, but the
global trophic structure of the assemblage seems to change with the system. For
example, the fish assemblage of the Machado River, a clear-water tributary
of the Madeira River, is highly supported by forest resources and shows a
broad partitioning between microphages, carnivorous and herbivorous species
(Goulding, 1980). In the present results, as in the white waters of the central
Amazon (Soares et al., 1986), the herbivores tended to be under-represented,
although microphages (algivores and mud feeders) and carnivores (invertivores
and piscivores) included the majority of the species.

D I E T́– M O R P H O L O G Y R EL A T I O N S H I P S
The 48 fish species analysed had taxonomical links ranging from intra-genus
to inter-order. Unfortunately, no reliable phylogenies exist for South American
fishes and thus the analysis had to be based on the less reliable data of
taxonomic relatedness. Weathers & Siegel (1995), however, indicated that ‘if a
data set is reasonably diverse phylogenetically (i.e. has a high ratio of number of
clades to number of species) and exhibits a reasonably strong correlation,
phylogenetic analysis is unlikely to alter conclusions’. Thus, as the analysis
showed high taxonomic diversity (five orders and 18 families for 48 species), it
is likely that integration of taxonomic relatedness rather than phylogenies
would give reliable conclusions with respect to the effect of evolutionary
relationships.

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1151

Three main associations between morphology and fish diet emerged from the
results. Among the 10 morphological variables analysed, relative gut length
appeared to be the main feature broadly associated with fish diet. Relative gut
length was positively and strongly associated with herbivory and iliophagy and
negatively with carnivory, as has been reported in various fish assemblages
(Paugy, 1994; Winemiller & Adite, 1997; Piet, 1998; Hugueny & Pouilly, 1999)
and among species belonging to the same family (Fryer & Ilies, 1972). Besides
gut length, the second main trend corresponded to a positive link between
carnivory and prey size and the morphological variables LS, HEAD and
MOHE. This association has also been reported in various fish assemblages
(Gatz, 1979; Wainwright & Richard, 1995; Piet, 1998; Hugueny & Pouilly,
1999). Finally, the third main association was the positive link between gill
raker length and number and zooplankton consumption (Motta et al., 1995).
Other patterns presented in the literature were not at all or only weakly
validated in the results. In the CCA results, mouth orientation, eye diameter
and eye position appeared as a weak descriptor of fish diet. Benthic fishes
generally tend to possess dorsally positioned eyes and a ventrally orientated
mouth, while pelagic fishes have laterally positioned eyes and an upward
orientated mouth (Gatz, 1979; Watson & Balon, 1984; Winemiller et al., 1995;
Hugueny & Pouilly, 1999). The Curimatidae, which includes a majority of
benthic feeding species characterized by a terminal mouth and lateral eyes, is
an exception to this generalization. A majority of Curimatidae species, however,
have evolved to a benthic detritivorous diet (Géry, 1977) and display other
adaptations, like a lack of teeth and a long intestine adapted to detritivory,
that indirectly indicate benthic feeding. Hugueny & Pouilly (1999) noted a
similar case of Mormyridae benthic feeding species that did not have dorsally
positioned eyes.
Gatz (1979) linked eye size to the importance of vision in feeding and it is
likely that sight is important in foraging and in the capture of animal food and
especially in piscivory. Among the species analysed, eye size appears to be
greatly influenced by taxonomy, as 80% of the species with an EYDI above
the mean were Characiformes (Table IV) which consequently limits the ecological
interpretation of this character in the data.
The results call attention to the fact that the selection of morphological
variables is one of the major difficulties in ecomorphological analysis. In
particular, other morphological variables than those examined could play
important roles in food selection and ingestion and consequently a failed
correlation does not necessarily indicate that there is no morphological correl-
ation with diet. Morphological variables, categories of diet items and statistical
methods used are generally different and specific to individual studies. Even
though the situations analysed were different, the three main associations
observed in the present study remain generally valid, and thus point to a strong
trend towards the matching of an organism’s morphology to its theoretical or
potential diet. Increase in gut length, in the size and number of gill rakers, and
in the size of the mouth and head appear as constant adaptations resulting from
a specialized herbivore, plantkivore and piscivore diet. Consequently, it is
recommended that these morphological features should not be avoided in diet
ecomorphological studies.

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 1152

TABLE IV. Classification, standard length (range in parentheses) and mean values (S.D. in parentheses where appropriate) of nine
morphological parameters for 48 fish species of the Mamoré River. HEAD, relative head length; MOHE, relative mouth height; MORI,
mouth orientation; EYPO, eye vertical position; TEETH, type of dentition; EYDI, relative eye diameter; GRNU, number of gill rakers on
the first gill arch; GRLE, relative maximum length of gill rakers on the first gill arch; GUTL, relative length of the digestive tract

#
HEAD MOHE EYPO EYDI GRLE
LS 10 102 10 102 102 GUTL
Order, Family Species Code n (range) (S.D.) (S.D.) MORI (S.D.) TEETH (S.D.) GRNU (S.D.) (S.D.)

Characiformes
Anostomidae Leporinus friderici LEOFR 30 130́3 2́76 5́53 2 5́97 1 7́71 25 1́21 1́37
friderici (Bloch) (54́8́–195) (0́18) (1) (1́16) (0́91) (0́34) (0́24)
Rhytiodus microlepis RHYMI 26 126́8 2́72 6́64 2 5́79 1 7́1 32 1́89 1́85
Kner (77́8́–197́7) (0́17) (1́3) (0́92) (1́3) (1́2) (0́17)
Schizodon fasciatus SCHFA 26 172́6 2́46 5́55 2 5́18 1 6́98 17 1́24 1́65
Spix & Agassiz (80́5́–277́7) (0́36) (1́23) (0́67) (1́54) (0́31) (0́43)
Characidae Aphyocharax anisitsi APYAN 30 54́7 2́83 10́53 2 5́64 2 9́99 20 3́03 1́21
M. POUILLY

Eigenmann & Kennedy (50́–58́6) (0́15) (0́56) (0́33) (0́29) (0́35) (0́1)
Triportheus albus Cope TRPAL 30 105́2 2́36 8́86 2 5́9 1 8́06 44 1́18 1́09
(91́2́–159́2) (0́1) (1́33) (0́63) (0́45) (0́25) (0́13)
Triportheus angulatus TRPAN 30 100 2́66 8́84 1́5 6́23 1 8́61 43 2́53 1́31
ET AL.

(Spix & Agassiz) (54́–148) (0́21) (0́8) (0́29) (1́04) (0́34) (0́19)
Acestrorhynchus cf. ACEAL 30 157́5 2́94 13́75 2 6́45 1 6́24 27 0́88 1́03
altus Menezes (92́–241́5) (0́26) (1́94) (0́54) (0́63) (0́13) (0́17)
Roeboides affinis ROBAF 30 63́1 2́64 11́19 1́5 5́32 1 9 16 3́68 0́69
(Günther) (51́1́–116́4) (0́27) (1́74) (0́53) (0́63) (0́63) (0́17)
Roeboides myersii Gill ROBMY 30 88́8 2́88 13́43 2 6́4 1 9́74 26 6́02 0́93
(63́5́–134́2) (0́14) (1́71) (0́48) (0́66) (0́56) (0́14)
Parecbasis cyclolepis PAECY 30 54́7 2́71 10́1 2 6́34 1 9́96 26 3́29 0́8
Eigenmann (51́7́–65́4) (0́22) (1́15) (0́77) (0́88) (0́65) (0́14)
Poptella compressa POPCO 30 64́2 2́77 11́51 2 5́35 1 10́96 20 2́75 0́92
(Günther) (36́5́–83́2) (1́65) (2́49) (0́64) (2́15) (0́61) (0́2)

2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 #
Stethaprion crenatum STTCR 30 45́7 3́06 15́6 1́5 5́49 1 12́44 23 3́04 1́08
Eigenmann (41́3́–49́8) (0́15) (1́21) (0́54) (0́51) (0́31) (0́27)
Ctenobrycon spilurus CTEPI 30 42́9 2́45 8́61 2 4́76 1 10́59 19 2́04 0́96
(Valenciennes) (38́6́–49́5) (0́44) (1́08) (0́43) (0́44) (0́45) (0́16)
Moenkhausia dichroura MOEDI 30 55́8 2́44 10́84 2 5́36 1 9́78 25 3́39 0́82
(Kner) (47́–63́9) (0́29) (0́77) (0́43) (0́53) (0́36) (0́12)
Curimatidae Curimatella alburna CUEAI 30 96́2 2́73 7́75 2 5́12 0 10́14 26 0́73 10́06
(Müller & Troschel) (87́6́–118́5) (0́13) (0́56) (1́04) (0́97) (0́15) (2́09)
Potamorhina POTAL 30 119́9 3́49 10́42 2 5́4 0 8́75 0 — 9́03
altamazonica (Cope) (81́76́–174) (0́5) (1́75) (0́72) (0́91) (2́26)
Potamorhina latior POTLA 30 126́7 3́38 9́46 1́5 5́78 0 8́07 0 — 8́04
(Spix & Agassiz) (71́8́–154́6) (0́27) (1́92) (0́51) (0́78) (2́21)
Psectrogaster amazonica PSEAM 30 71 3́49 12́1 2 5́74 0 10́97 96 0́82 1́58
Eigenmann & Eigenmann (60́2́–102́5) (0́31) (1́07) (0́41) (0́51) (0́17) (0́17)
Psectrogaster curviventris PSECU 30 109́3 3́23 10́07 2 5́76 0 9́26 0 — 10́27
Eigenmann & Kennedy (63́–158́4) (0́19) (1́32) (0́66) (1́07) (2́8)
Erythrinidae Hoplias malabaricus HOPMA 30 150́9 3́24 16́82 2 6́34 1 6 24 2́19 1́07
(Bloch) (63́4́–223́3) (0́73) (2́98) (0́66) (0́84) (0́57) (0́2)
Hemiodontidae Anodus melanopogon ANOME 30 135́6 3́38 9́77 2 5́57 0 6́65 105 6́96 1́77
Agassiz (78́7́–175́5) (0́26) (1́93) (0́49) (0́77) (0́61) (0́3)
Serrasalmidae Mylossoma duriventre MYLDU 30 110 2́91 7́88 2 5́1 1 9́23 21 2́9 1́96
(Cuvier) (64́4́–150́2) (0́23) (0́97) (0́42) (0́84) (0́38) (0́43)
Serrasalmus eigenmanni SEREI 30 118́9 3́24 12́94 2 5́95 1 9́04 18 1́28 2́00
(Norman) (66́9́–154́8) (0́2) (2́01) (0́83) (0́63) (0́21) (0́88)
Pygocentrus nattereri PYGNA 30 133́7 3́84 20́03 1 6́37 1 8́3 18 2́12 1́64
DIET AND MORPHOLOGY OF AMAZONIAN FISHES

Kner (105́3́–202́5) (0́18) (1́3) (1́9) (0́9) (0́59) (0́31)

Serrasalmus hollandi SERHO 23 84́5 3́85 14́46 1 5́66 1 9́68 32 1́27 1́23
Eigenmann (39́5́–146́2) (1́32) (1́84) (2́21) (2́54) (0́4) (0́18)

2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
Serrasalmus rhombeus (L.) SERRH 30 98́5 3́73 14́94 1 6́27 1 9́3 11 1́33 1́22
(37́9́–142́5) (0́2) (2́62) (0́39) (1́05) (0́39) (0́21)
1153
 1154

TABLE IV. Continued overleaf

#
HEAD MOHE EYPO EYDI GRLE
LS 10 102 10 102 102 GUTL
Order, Family Species Code n (range) (S.D.) (S.D.) MORI (S.D.) TEETH (S.D.) GRNU (S.D.) (S.D.)

Clupeiformes
Clupeidae Pellona flavipinnis PELFL 30 153́2 2́78 15́21 1 7́61 1 8́53 27 3́96 0́81
(Valenciennes) (138́3́–183́2) (0́12) (1́3) (0́56) (0́61) (0́39) (0́13)
Engraulidae Engraulidae ENGSP 30 141́4 3́03 13́13 2 4́74 0 5́19 87 6́29 2́61
(124́–177) (0́12) (1́51) (0́99) (0́41) (0́55) (0́43)
Gymnotiformes
Apteronotidae Adontosternarchus ADOSA 30 152́4 1́3 3́76 2 7́58 0 1́26 0 — 0́5
sachsi (Peters) (105́3́–189́3) (0́14) (0́41) (0́47) (0́17) (0́12)
Sternopygidae Eigenmannia virescens EIAVI 30 221́5 1 2́37 2 6́2 3 1́79 14 0́71 0́36
M. POUILLY

(Valenciennes) (123́4́–608́7) (0́21) (0́67) (0́58) (0́45) (0́26) (0́07)

Perciformes
Sciaenidae Plagioscion PLGSQ 30 249́8 3́38 13́03 2 6́87 2 5́94 15 4́47 0́93
squamosissimus (Heckel) (185́2́–347) (0́16) (1́9) (0́45) (0́46) (0́4) (0́16)
ET AL.

Siluriformes
Ageneiosidae Ageneiosus brevis AGEBR 30 59́8 2́85 11́71 3 3́75 3 4́04 26 2́89 0́96
Steindachner (56́4́–65́7) (0́15) (0́65) (0́75) (0́42) (0́22) (0́13)
Tympanopleura sp. TYMSP 30 84́9 2́91 9́05 4 3́88 3 5́66 19 1́87 1́06
(68́5́–108́1) (0́4) (0́82) (0́73) (0́84) (0́25) (0́14)
Auchenipteridae Auchenipterus nuchalis AUCNU 30 128́4 2́09 7́38 2 3́14 1 4́91 33 2́27 1́13
(Spix & Agassiz) (87́5́–148́4) (0́22) (0́95) (0́51) (0́71) (0́45) (0́16)
Centromochlus sp. CENSP 30 66́1 2́8 9́16 2 5́32 3 7́6 8 0́23 1́33
(54́7́–76́5) (0́23) (1́07) (0́5) (1́02) (0́12) (0́26)

2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 #
Entomocorus benjamini ENTBE 27 61 2́82 9́71 2 4́04 3 7́48 18 2́22 0́99
Eigenmann (54́1́–70́2) (0́46) (1́27) (0́68) (0́64) (0́29) (0́12)
Parauchenipterus PAUST 30 140́5 2́9 9́56 2 5́21 3 3́97 8 0́48 1́18
striatulus (Steindachner) (103́–165́6) (0́25) (1́11) (0́53) (0́38) (0́12) (0́21)
Callichthyidae Brochis splendens BROSL 25 46́7 3́48 6́23 4 5́96 1 8́68 15 1́11 1́48
(Castelnau) (37́3́–65́8) (0́43) (0́91) (0́51) (1́01) (0́19) (0́21)
Doradidae Doras sp. DORSP 30 56́3 3́11 8́18 3 6́11 0 7́98 14 1́44 2́09
(46́4́–110́1) (0́15) (1́16) (0́55) (0́61) (0́44) (0́45)
Opsodoras sp. OPSSP 29 64́2 3́23 6́69 4 7́59 0 6́79 11 1́55 1́12
(43́3́–84́6) (0́3) (0́69) (0́5) (0́6) (0́23) (0́17)
Trachydoras paraguayensis TRYPA 27 96́6 2́93 6́79 4 7́69 0 8́1 14 0́64 1́74
(Eigenmann & Ward) (84́3́–116́1) (0́3) (0́89) (0́52) (0́96) (0́16) (0́24)
Hypophthalmidae Hypophthalmus HYAMA 30 195́6 3́04 10́54 2 2́21 0 2́87 246 6́4 1́46
marginatus Valenciennes (107́–272) (0́28) (2́96) (0́38) (0́51) (0́99) (0́36)
Loricariidae Hypoptopoma joberti HYTJO 30 70́5 3́14 8́32 4 3́09 2 5́63 84 1́73 8́33
(Vaillant) (61́4́–79́6) (0́18) (1́68) (0́53) (0́6) (0́21) (1́56)
Loricariichthys maculatus LOIMA 30 168́4 2́31 5́34 4 8́01 1 3́07 21 0́47 1́79
(Bloch) (84́3́–230́9) (1́49) (0́56) (0́48) (0́58) (0́11) (0́4)
Pseudohemiodon laticeps PSHLA 30 137́5 1́74 4́7 4 8́18 1 2́94 12 0́48 1́45
(Regan) (84́3́–199́8) (0́15) (0́72) (0́64) (0́34) (0́13) (0́23)
Sturisoma nigrirostrum STUNI 30 121́7 2́31 5́16 4 7́46 2 2́61 68 0́57 7́09
Fowler (88́7́–188́7) (0́16) (0́38) (0́58) (0́23) (0́13) (0́62)
Pimelodidae Calophysus macropterus CAOMA 26 169́5 2́43 6́54 3 7́92 1 2́86 25 2́03 1́04
DIET AND MORPHOLOGY OF AMAZONIAN FISHES

(Lichtenstein) (118́6́–241́7) (0́11) (1́35) (0́47) (0́52) (0́21) (0́2)

Pimelodus gr. maculatus- PIUMA 30 73 4́14 8́38 3 6́54 3 7́35 24 3́1 1́05
blochii Valenciennes (51́2́–112́7) (5́32) (1́22) (0́68) (0́71) (0́81) (0́22)

2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1155
1156 M. POUILLY ET AL.

Several authors have noted that piscivorous and detritivorous species have
the most highly specialized morphology (Piet, 1998; Hugueny & Pouilly, 1999;
Sibbing & Nagelkerke, 2001; Xie et al., 2001). The other diets (invertivore and
omnivore) seem to require a lower level of morphological specialization at least
for the features analysed. At the level of assemblage, the trophic composition
may thus play a role in determining the degree of correlation between diet and
morphology. Studies that concluded the existence of a strong relationship gen-
erally include specialized piscivores, detritivores and planktivores (Piet, 1998;
Hugueny & Pouilly, 1999; Sibbing & Nagelkerke, 2001; Xie et al., 2001). In
contrast, studies that include a majority of invertivore or omnivore and general-
ist species concluded the existence of a relatively weak relationship between diet
and morphology (Douglas & Matthews, 1992; Motta et al., 1995; Winemiller &
Adite, 1997).
The results suggest significant relationships between diet and morphology in the
most abundant fish species of the Mamoré River, even when the taxonomicaĺ–
morphological relationships between species were eliminated. This conclusion
supports the ecomorphological hypothesis and its two main arguments, first
that species morphology is associated with their diet (Wainwright & Richard,
1995) and second that morphological similarity is conserved even in the compari-
son of taxonomically unrelated species (Winemiller, 1991). Ecomorphological
diversification in an assemblage is a function of the richness of species
(Winemiller, 1991). Accordingly, the results obtained in the Mamoré River flood-
plain could be related to the relatively high trophic and taxonomic diversity and
especially to the presence of species with specialized diet requiring particular
morphological features.

This work was part of the BIOBAB project (aquatic biodiversity in the Bolivian
Amazon basin) developed by IRD, the Instituto de Ecologı́a from La Paz University
(Universidad Mayor de San Andrés) and the Centro de Investigación de los Recursos
Acuáticos’ of Trinidad University (Universidad Técnica del Beni). We thank these
institutions for their support. J.G. Bretenoux was a student at the French Polytechnic
School. T. Yunoki, A. Parada from Trinidad University and J.L. Menou from IRD
helped with logistics, fieldwork and identification of specimens. We would also like
to thank B. Hugueny (IRD) for reviewing the manuscript and L. Torres (Trinidad
University) for his helpful participation in the BIOBAB project.

References
Carvalho, F. M. (1980). Alimentação do maparã (Hypophthalmus edentatus Spix, 1829)
do lago Castanho, Amazonas (Siluriformes, Hypophthalmidae). Acta Amazonica
10, 545́–555.
Denevan, W. M. (1980). La Geografı´a Cultural Aborigen de Los Llanos de Moxos. La Paz,
Bolivia: Juventud.
Douglas, M. E. & Matthews, W. J. (1992). Does morphology predict ecology? Hypothesis
testing within a freshwater stream fish assemblage. Oikos 65, 213́–224.
Fryer, G. & Ilies, T. D. (1972). The Cichlid Fishes of the Great Lakes of Africa. Edinburgh:
Oliver & Boyd.
Gatz, A. J. (1979). Ecological morphology of freshwater stream fishes. Tulane Studies in
Zoology and Botany 21, 91́–124.
Géry, J. (1977). Characoids of the World. Neptune City, NJ: T.F.H. Publications.

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
 DIET AND MORPHOLOGY OF AMAZONIAN FISHES 1157

Goulding, M. (1980). The Fishes and the Forest. Exploration in Amazon Natural History.
Berkeley, CA: University of California Press.
Grossman, G. D. (1986). Food resource partitioning in a rocky intertidal fish assemblage.
Journal of Zoology 1, 317́–355.
Hugueny, B. & Pouilly, M. (1999). Morphological correlates of diet in an assemblage of
West African freshwater fishes. Journal of Fish Biology 54, 1310́–1325. doi:
10.1006/jfbi.1999.0962.
Hurlbert, S. H. (1978). The measurement of niche overlap and some relatives. Ecology 59, 67́–77.
Knöppel, H. A. (1970). Food of central Amazonian fishes, contribution to the nutrient-
ecology of Amazonian rain-forest streams. Amazoniana 2, 257́–352.
Lauzanne, L. & Loubens, G. (1985). Peces del rio Mamore´. Collection Travaux et Docu-
ments no. 192. Paris: ORSTOM.
Lauzanne, L., Loubens, G. & Le Guennec, B. (1991). Liste commentée des poissons de
l’Amazonie bolivienne. Revue d’Hydrobiologie Tropicale 24, 61́–76.
Legendre, P. & Legendre, L. (1998). Numerical Ecology. Amsterdam: Elsevier.
Loubens, G., Lauzanne, L. & Le Guennec, B. (1992). Les milieux aquatiques de la région
de Trinidad (Béni, Amazonie bolivienne). Revue d’Hydrobiologie Tropicale 25,
3́–21.
Marlier, G. (1968). Etudes sur les lacs de l’Amazonie centrale. III Les poissons du lac
Redondo et leur régime alimentaire; les chaines trophiques du lac Redondo; les
poissons du rio Prêto da Eva. Cadernos Amazonia (INPA, Manaus) 11, 21́–57.
Menezes, N. A. (1969). The food of Brycon and three closely related genera of the tribe
Acestrorhynchini. Pape´is Avulsos Zoologia Sao Paulo 22, 217́–223.
Mérigoux, S. & Ponton, D. (1998). Body shape, diet and ontogenetic diet shifts in young
fish of the Sinnamary river, French Guiana, South America. Journal of Fish
Biology 52, 556́–569.
Motta, P. J., Clifton, K. B., Hernandez, P. & Eggold, B. T. (1995). Ecomorphological
correlates in ten species of subtropical seagrass fishes: diets and microhabitat
utilization. Environmental Biology of Fishes 44, 37́–60.
Paugy, D. (1994). Ecologie des poissons tropicaux d’un cours d’eau temporaire (Baoulé,
haut bassin du Sénégal au Mali): adaptation au milieu et plasticité du régime
alimentaire. Revue d’Hydrobiologie Tropicale 27, 157́–169.
Piet, G. J. (1998). Ecomorphology of a size-structured tropical freshwater fish commu-
nity. Environmental Biology of Fishes 51, 67́–86.
Pouilly, M., Ibañez, C., Guttierez, M. & Yunoki, T. (1999). Funcionamiento ecológico de
las lagunas de la zona de inundación del rı́o Mamoré (Beni ´ – Bolivia). Revista
Boliviana de Ecologı´a 6, 41́–54.
Sibbing, F. A. & Nagelkerke, L. A. J. (2001). Resource partitioning by Lake Tana
barbs predicted from fish morphometrics and prey characteristics. Reviews in
Fish Biology and Fisheries 10, 393́–437.
Smouse, P. E., Long, J. C. & Sokal, R. (1986). Multiple regression and correlation
extensions of the Mantel test of matrice correspondence. Systematic Zoology 35,
627́–632.
Soares, M. G. M. (1979). Aspectos ecológicos (alimentação e reprodução) dos peixes do
Igarapé do Porto, Aripuana, MT. Acta Amazonica 9, 325́–352.
Soares, M. G. M., Almeida, R. G. & Junk, W. J. (1986). The trophic status of the fish
fauna in Lago Camaleão, a macrophyte dominated floodplain lake in the middle
Amazon. Amazoniana 9, 511́–526.
Ter Braak, C. F. J. (1986). Canonical correspondence analysis: a new eigenvector
technique for multivariate direct gradient analysis. Ecology 67, 1167́–1179.
Wainwright, P. C. & Richard, B. A. (1995). Predicting patterns of prey use from
morphology of fishes. Environmental Biology of Fishes 44, 97́–113.
Watson, D. J. & Balon, E. K. (1984). Ecomorphological analysis of fish taxocenes in
rainforest streams of northern Borneo. Journal of Fish Biology 25, 371́–384.
Weathers, W. W. & Siegel, R. B. (1995). Body size establishes the scaling of avian
postnatal metabolic rate: an interspecific analysis using phylogenetically independent
contrasts. Ibis 137, 532́–542.

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158
1158 M. POUILLY ET AL.

Wikramanayake, E. D. (1990). Ecomorphology and biogeography of a tropical stream

fish assemblage: evolution of assemblage structure. Ecology 71, 1756́–1764.
Winemiller, K. O. (1991). Ecomorphological diversification in lowland freshwater fish
assemblages from five biotic regions. Ecological Monographs 61, 343́–365.
Winemiller, K. O. & Adite, A. (1997). Convergent evolution of weakly electric fishes from
floodplain habitats in Africa and South America. Environmental Biology of Fishes
49, 175́–186.
Winemiller, K. O., Kelso-Winemiller, L. C. & Brenkert, A. L. (1995). Ecomorphological
diversification and convergence in fluvial cichlid fishes. Environmental Biology of
Fishes 44, 235́–261.
Xie, S., Cui, Y. & Li, Z. (2001). Dietarý–morphological relationships of fishes in Liangzi
Lake, China. Journal of Fish Biology 58, 1714́–1729. doi: 10.1006/jfbi.2001.1580.

# 2003 The Fisheries Society of the British Isles, Journal of Fish Biology 2003, 62, 1137́–1158