Trophic ecology of piranhas (Characidae : serrasalminae) from savanna and forest regions in the Orinoco River basin of V...


Material Information

Trophic ecology of piranhas (Characidae : serrasalminae) from savanna and forest regions in the Orinoco River basin of Venezuela
Physical Description:
xiii, 209 leaves : ill., photos ; 29 cm.
Nico, Leo G., 1954-
Publication Date:


Subjects / Keywords:
Piranhas   ( lcsh )
Fishes -- Orinoco River Region (Venezuela and Colombia)   ( lcsh )
Zoology thesis Ph. D
Dissertations, Academic -- Zoology -- UF
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1991.
Includes bibliographical references (leaves 198-208).
Additional Physical Form:
Also available online.
General Note:
General Note:
Statement of Responsibility:
by Leo G. Nico.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 25622507
System ID:

Table of Contents
    Title Page
        Page i
        Page ii
        Page iii
        Page iv
    Table of Contents
        Page v
        Page vi
    List of Tables
        Page vii
        Page viii
    List of Figures
        Page ix
        Page x
        Page xi
        Page xii
        Page xiii
    Chapter 1. Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
    Chapter 2. Environmental setting
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
    Chapter 3. Methods
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
    Chapter 4. Composition of piranha assemblages
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
    Chapter 5. Trophic ecology of savanna piranhas: The apure drainage
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
    Chapter 6. Trophic ecology of piranhas from the upper Orinoco
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
        Page 146
        Page 147
    Chapter 7. Nutrient values of piranha prey
        Page 148
        Page 149
        Page 150
        Page 151
        Page 152
        Page 153
        Page 154
    Chapter 8. Discussion and conclusions
        Page 155
        Page 156
        Page 157
        Page 158
        Page 159
        Page 160
        Page 161
        Page 162
        Page 163
        Page 164
        Page 165
        Page 166
        Page 167
        Page 168
        Page 169
        Page 170
        Page 171
        Page 172
        Page 173
        Page 174
        Page 175
        Page 176
        Page 177
        Page 178
        Page 179
        Page 180
    Appendix A. Material examined
        Page 181
        Page 182
        Page 183
        Page 184
        Page 185
        Page 186
        Page 187
        Page 188
        Page 189
        Page 190
    Appendix B. Data analysis
        Page 191
        Page 192
        Page 193
        Page 194
        Page 195
        Page 196
        Page 197
    References cited
        Page 198
        Page 199
        Page 200
        Page 201
        Page 202
        Page 203
        Page 204
        Page 205
        Page 206
        Page 207
        Page 208
    Biographical sketch
        Page 209
        Page 210
Full Text







I owe a debt of gratitude to many friends and colleagues for their help during the field work, data analysis, and manuscript preparation. During my development as a tropical biologist, I greatly benefited from the guidance of several people. Dr. Jamie Thomerson launched me into the study of neotropical fishes. Dr. Donald Taphorn provided me with the opportunity to do long-term research in the Orinoco River basin. In 1980, he took me to the Llanos where we netted my first piranha while seining the muddy waters of Calo Maraca. The botanist Dr. Basil Stergios invited me to accompany his 1985 expedition to the upper Rio Negro and the Brazo Casiquiare where I fell in love with the rivers and humid forests of southern Venezuela. My graduate committee chairman, Dr. Horst Schwassmann, opened my eyes to the Amazon River; together we crossed part of its estuary, from Bel6m to the Uha de Maraj6, a journey I shall always remember.

My investigations have prospered from discussions with various members of my committee as well as from the exchange of ideas with colleagues and friends, in particular Drs. Donald Taphorn, Steve Walsh, and Kirk Winemiller. Dr. Horst Schwassmann, and other members of my committee -- Drs. Carter Gilbert, Martha Crump, Frank Nordlie, and Nigel Smith -- were always helpful. I thank them for their advice and for reviewing the tedious draft manuscripts. I am also grateful to Drs. Stephen Walsh, Jamie Thomerson, Thomas Keevin, and Carmine Lanciani, and to Frank Jordon for their comments on various sections and drafts of the manuscript.

I thank the following persons for their assistance with collecting fishes in the Venezuelan Llanos: Anielo Barbarino, Linda Delashmidt, Terry Dye, Guillermo Feo, Carter Gilbert, Oscar Leon, Craig Olds, Stewart Reid, Eric Sutton, Donald Taphorn, and


Kirk Winemiller. I am particularly grateful to Donald C. Taphom for providing both a home and work area in Guanare, and for sharing his knowledge of the Venezuelan fish fauna. He has always been unselfish in the sharing of fish specimens and field data from his work in the Orinoco Llanos.

Most of my work in the upper Orinoco resulted from participation in expeditions as part of a natural resource inventory of southern Venezuela, organized and sponsored by Corporaci6n Venezolana Guayana Tecnica Minera (CVG-TECMIN). I thank all those associated with CVG-TECMIN whose assistance made my work possible, for administrative and logistic support these include Dr. Fernando Susach, Victor Fernandez, and Pedro Mata. For occasional help in the field, I express my sincere appreciation to several members of the TECMIN technical staff, Angel Fernandez, Luz:Delgado, Andres Garcia, Guillermo Fuenmayor, Sandra Giner, and Jesus Santiago. I relied heavily on the fishing skills of my field assistants and camp workers hired by CVG-TECMIN: Fabian Morillo, Yovani Aragua, Juan Carlos Castillo, Ambrosio Guayamore, Antonio Gaveln, and Hilario Bolivar. Assistance in the upper Orinoco was also given by Dr. Basil Stergios, Carlos Rivas, Carlos Templa, and Mike Dawson; many of the Yanomamo Indians including Chirino, Pablo and Ram6n, Cesar Timanawd, Ram6n Pokorai; the Curipaco Indians Carlos Antonio Guaruya, Eliasa Guaruya, and Eruerto L6pez; and many individuals whose names I did not record. Together with my field assistants, their intimate knowledge of the fishes, the rivers, and the forests of the upper Orinoco was invaluable to me.

I thank Drs. William Fink and Antonio Machado-Allison for identification of

voucher specimens of many of the serrasalmine fishes, and for providing me with their unpublished key to Venezuelan piranhas. I am also obliged to Dr. Donald Taphorn for identification of fish specimens, and to Dr. Basil Stergios, Gerardo Aymard, and Nidia de Cuellar for identifying plant material. I thank Darryl Harrison for preparing base maps, Angelina Licata for the fish drawings used in many of my figures, Kevin Schuck for help


with computer drawing of dendrograms, and Craig Lilyestrom for help with a few of the graphs and introducing me to the MacIntosh computer.

Parts of my research were funded by National Geographic Society grants no. 381188 and 4183-89. The Department of Zoology, University of Florida, provided additional funds and equipment. I thank all those in the Department of Zoology and Florida Museum of Natural History who gave of their time. I thank George Burgess for use of his library. The Universidad Nacional de los Llanos Occidentales and Museo de Zoologfa (UNELLEZMCNG) gave me lab space and administrative assistance in Venezuela. I am indebted to many of the UNELLEZ administrators and staff for their help during the last ten years. In addition, I thank the Venezuelan people, campesinos, Llaneros, and indigenas, for allowing me do research in their beautiful country.

While conducting this study I came to know many very special people. During my stay in Venezuela, I enjoyed the warm hospitality and friendship of Luis and Sioux Strebin, Jose and Sharon Sutera, and Basil and Anita Stergios; along with Terry Dye, Donald Taphorn, Lucas van Balen and his wife, and Dr. H. Kasem. they brought me through my bout withfalciparum malaria. I express appreciation to Dr. Horst Schwassmann and his wife Satiko for frequently inviting me into their home during my passages through Gainesville. I also thank Dr. Carter Gilbert and all my fellow graduate students, especially Dennis Haney and Dr. Steve Walsh, for their many hospitalities. Finally, I thank my parents and family for their constant love and support, both at home and during the many long separations, and I thank Linda for always being there.



ACKNOWLEDGEMENTS ........................................................................ ii

LIST O F TA BLES .................................................................................. vii

LIST OF FIGURES ................................................................................. ix

A B ST R A CT .... .................................................................................... x ii


I INTRODUCTION ...................................................................... 1

Introduction to Piranhas ................................................................ 1
Purpose of Present Study .............................................................. 3
Literature R eview ......................................................................... 7
Organization of Dissertation ......................... ................................. 16

2 ENVIRONMENTAL SETTING ..................................................... 18

The Orinoco River Basin .............................................................. 18
History of the Basin .................................................................... 22
The Low Llanos ........................................................................ 24
The Upper Orinoco ...................................................................... 28
Savannas and Forests During the Quaternary ....................................... 30
Present Climate and Seasonality ..................................................... 32

3 M E T H O D S .............................................................................. 35

Field Sam pling .......................................................................... 35
Field Site Locations and Sampling Periods ......................................... 37
Evaluation of Habitats ................................................................. 42
Identification and Nomenclature of Piranhas ........................................ 43
Faunal Comparisons ................................................................... 47
A nalysis of D iets ....................................................................... 49
Nutrient Content Analysis of Food Items ........................................... 53
Intestine L ength ......................................................................... 54
Field and Aquarium Observations ..................................................... 55
Statistical A nalyses ...................................................................... 55

4 COMPOSITION OF PIRA A ASSEMBLAGES ............................... 57


DRAINAGE ...........................................................71

Study Area .............................................................. 71
Species Accounts.........................................................75
Comparison of Diets.....................................................91

ORINOCO............................................................ 97

Species Accounts ....................................................... 100
Summary of Upper Orinoco Trophic Patterns............................. 123
Comparison with Savanna Populatons................................... 131
Ecomorphological Correlates: Intestine Length and Diet ..................143

7 NUTRIENT VALUES OF PIRANHA PREY............................148

Animal Matter .........................................................148
Plant Matter............................................................. 153

8 DISCUSSION AND CONCLUSIONS.................................. 155

Composition of Piranha Assemblages ................................... 155
Trophic Patterns ............................ .......................... 163
Ecomorphological. Correlates ............................................ 175
A Phylogenetic Perspective ............................................... 177


A MATERIAL EXAMINED ................................................ 181

B DATA ANALYSIS....................................................... 191

LITERATURE CITED ............................................................ 198

BIOGRAPHICAL SKETCH ...................................................... 209



1-1 Summary of principal field studies investigating piranha feeding and diets in
order of publication date .............................................................. 9
3-1 List of abbreviations used in figures and tables for piranhas and other
serrasalm ine fishes ...................................................................... 46
4-1 Occurrence of piranha species in samples from nine drainages in the
Orinoco River basin, Venezuela ................................................. 59

4-2 Distance between drainages and Coefficient of Biogeographic Resemblance
among nine sampled drainages in the Orinoco River basin ..................... 62

4-3 Number of shared piranha species and Coefficient of Biogeographic
Resemblance among nine sampled drainages in the Orinoco River basin ........ 64 4-4 Summary of piranha species occurrence by habitat and water type, based on
samples in Orinoco River basin, Venezuela ..................................... 70
5-1 Food items of Pygocentrus caribe from the Apure River drainage (Calo
Caicara area) by size class ........................................................... 79

5-2 List of vertebrate prey taxa identified from the stomach contents of the four
most common piranhas from the Cafio Caicara area, Apure River drainage,
in the low Llanos of Venezuela ........................................................ 80
5-3 Food items of Serrasalmus irritans from the Apure River drainage (Caflo
Caicara area) by size class ........................................................81

5-4 Food items of Serrasalmus medinai from the Apure River drainage (Calo
Caicara area) by size class ........................................................ 83
5-5 Food items of Serrasalmus rhombeus from the Apure River drainage (Caflo
Caicara area) by size class ........................................................ 84
5-6 Food items of Serrasalmus elongatus from the Apure River drainage (Cafro
Caicara area) by size class ........................................................ 86
5-7 Food items of Serrasalmus altuvei from the Apure River drainage (Caio
Caicara area) by size class .................................... 88



5-8 Food items of Pristobrycon striolatus from the Apure River drainage (Caiio
Caicara area) by size class ........................................................ 89

5-9 Food items of Catoprion mento from the Apure River drainage (Calo
Caicara area) by size class ........................................................ 90

5-10 Food items of small juvenile piranhas (Size Class I, 10-19 mm SL),
tentatively identified as Pygocentrus caribe, from the Apure River drainage
(Caflo Caicara area) .................................................................. 92

5-11 Matrix of diet overlaps among different size classes of piranhas from Apure
River drainage, low Llanos ........................................................ 95

6-1 List of taxa identified from the stomach contents of the three most common
piranha species from upper Orinoco River drainages, Venezuela ............... 103

6-2 Food items of Serrasalmus rhombeus from the upper Orinoco River basin
by size class ............................................................................ 104

6-3 Food items of Serrasalmus manueli from the upper Orinoco River basin by
size class ............................................................................... 107

6-4 Food items of Serrasalmus cf. eigenmanni from the upper Orinoco River
basin by size class ................................................................... 111

6-5 Food items of Serrasalmus altuvei from the upper Orinoco River basin by
size class .............................................................................. 113

6-6 Food items of Pristobrycon striolatus from the upper Orinoco River basin
by size class .......................................................................... 115

6-7 Food items of Pygopristis denticulatus from the upper Orinoco River basin
by size class .......................................................................... 118

6-8 Matrix of diet overlaps among different size classes of serrasalrnine fishes
from upper Orinoco River drainages .............................................. 124

6-9 Incidence of herbivory in piranhas ( 80 mm SL) as associated with the low
Llanos and upper Orinoco ......................................................... 136

6-10 One-tailed Mann-Whitney U-test results testing the prediction that piranha
species (> 80 mm SL) from upper Orinoco River drainages fed more on
plant material than those from the low Llanos ..................................... 140
7-1 Estimates of the lipid, protein, ash, carbohydrate, and caloric contents of
fins, scales, and whole fish for two taxa of typical prey fish .................... 149

8-1 Orinoco River basin piranhas categorized by adult diet ........................... 158



Figure PAZe

1-1 Hypothesized phylogenetic relationships among genera of the subfamily
Serrasalminae as proposed by Machado-Allison (redrawn from MachadoA llison 1985) ............................................................................. 2

2-1 Map of northern South America showing Orinoco River basin ................... 19

2-2 Major savanna and forest ecosystems in the Orinoco River basin ................. 25

2-3 Approximate distribution of open savanna (low Llanos) in Venezuelan state
of Apure, Orinoco River basin ....................................................... 26

2-4 Approximate distribution of lowland forests in Federal Amazon Territory of
V enezuela ................................................................................ 29

2-5 Monthly rainfall recorded at representative sites in Ofinoco River basin,
Venezuela. (1) Upper Orinoco: monthly average 1971-1977 from Santa
Maria de Los Guaicas, mouth of Ocamo River, Federal Territory of
Amazonas, and (2) Low Llanos: for year 1972 from Cafto Caicara area,
Apure River drainage, Apure State .................................................... 34

3-1 Map of Venezuela showing nine selected drainages in Orinoco River basin
sampled during present study ........................................................ 38

3-2 Map of Apure State, Venezuela, showing principal low Llanos sampling
locations .................................................................................. 39

3-3 Map of 1988-1991 sampling sites in upper Orinoco, Federal Territory of
A m azonas, Venezuela .................................................................. 41

4-1 Schematic representation showing relative distances among the nine
sampled drainages and their approximate location along Orinoco River main
channel ................................................................................... 58

4-2 Scatter diagram showing relationship between number of piranha species
per drainage and log of the drainage area size (km2) for nine sampled
drainages in Orinoco River basin, Venezuela ........................................ 63

4-3 Scatter diagram showing similarity in piranha species composition between
adjacent sampled drainages versus distance between adjacent drainages ......... 65


Fiz=r Plg=
4-4 Scatter diagram showing similarities in piranha species composition
between all possible paired drainages and the distance between paired
drainages .............................................................................. 66
4-5 Dendrogram from cluster analysis depicting similarities among nine sampled
drainages of Orinoco River basin based on piranha species composition ........ 67

5-1 Location of Carlo Caicara study area, Apure River drainage, in low Llanos
of Apure State, Venezuela .......................................................... 72

5-2 General body form and major fin markings of seven piranha species (55-70
mm SL) and Catoprion mento from Calo Caicara study area, Apure River
drainage, in the low Llanos of Venezuela ............................................ 73
5-3 Diets by size class of seven piranha species and Catoprion mento from Calo
Caicara study area, Apure River drainage, in the low Llanos of Venezuela ...... 76
5-4 Diet breadths estimated for seven piranha species and Catoprion mento, by
size class, from the Caflo Caicara area of the low Llanos, Apure River
drainage, Orinoco River basin, Venezuela ....................................... 94

5-5 Dendrograrn from cluster analysis depicting similarities among diets of
piranha species (by size class) from the Calo Caicara area of the low
Llanos, Apure River drainage, Orinoco River basin, Venezuela .................. 96

6-1 Eight piranha species from the upper Orinoco River basin of Venezuela ......... 98

6-2 Diets by size class (>40 mm SL) of eight piranha species from the upper
Orinoco River basin, Venezuela ..................................................... 101

6-3 Diets of selected small juvenile piranhas (10-19 mm SL) from upper
Orinoco River drainages and the low Llanos, Venezuela ......................... 108

6-4 Diets by size class ( 80 mm SL) of four common serrasalmine species from
the upper Orinoco River basin: Catoprion mento, Myleus asterias, M.
schomburgkii, and M. torquatus .................................................... 120

6-5 Dendrogram from cluster analysis depicting similarities among diets of
piranha and related species (by size class) from the upper Orinoco River
basin, V enezuela ...................................................................... 127
6-6 Diet breadths estimated for eight piranha species, by size class, from the
upper Orinoco River basin, Venezuela ............................................. 130

6-7 Diets by size class (_ 40 mm SL) of three piranha species from the Cinaruco
and Capanaparo river drainages of the low Llanos, Orinoco River basin,
V enezuela .............................................................................. 133



6-8 Comparison of the proportion of plant material in the diets of carnivorous
and herbivorous piranha species ( 80 mm SL) from the low Llanos to those from upper Orinoco River drainages in terms of percent adjusted
volume (%Va), percent dominance (%D), and percent frequency of
occurrence (% O ) ..................................................................... 138

6-9 Diets by size class ( t40 mm SL) of Serrasalmus manueli comparing low
Llanos (Cinaruco drainage) and upper Orinoco populations ..................... 141

6-10 Scatter diagram showing relationship between mean intestine length/standard length and percent volume of plant material in diet for
selected serrasalmine fish ( >80 mm SL) ........................................... 145

6-11 Scatter diagram showing relationship between intestine length and standard length for the three most common piranha species in samples from the upper
Orinoco River basin of Venezuela ................................................... 146

6-12 Scatter diagram showing relationship between intestine length and standard length for Serrasalmus manueli comparing upper Orinoco and Llanos
p opulations ............................................................................. 146

7-1 Bar graph comparing dry mass composition of small whole fish, scales and
fin s ...................................................................................... 150

8-1 Proposed hypothetical model showing the dietary responses of carnivorous
versus herbivorous piranha species to changes in various habitat parameters
and food resources ........................................ 171

8-2 Diets and intestine length mapped onto the proposed phylogeny of
Machado-Alhison (1985) for genera of the subfamily Serrasalminae ........... 179


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy


Leo G. Nico

December, 1991

Chairman: Dr. Horst 0. Schwassmann
Major Department: Zoology

This study emphasizes interspecific and intraspecific comparisons of the trophic

ecology of piranhas (Characidae: Serrasalminae) from two vegetatively distinct regions in the Orinoco River basin of Venezuela: (1) the upper Orinoco in the Federal Amazon Territory, an area dominated by dense tropical forests, and (2) the low Llanos of Apure State, a region characterized by immense open savannas. Eleven piranha species were taken from the two regions studied. Since 1979, ten species have been collected from the low Llanos of Apure State, whereas my 1989-1991 samples indicate that at least eight species occur in the upper Orinoco. Seven species were widely distributed and inhabited both regions. The high similarity in species composition between the two regions suggests that there are really no distinct savanna versus forest piranha assemblages.

All piranhas exhibit changes in diet with age. The smallest juveniles, < 20 mm standard length (SL), prey mainly on microcrustaceans and small aquatic insects. Older juveniles of most species, between 20 and 80 mm SL, commonly eat fins. Fin eating by xii

young piranhas is common in both the low Llanos and the upper Orinoco. Chemical analysis indicated that fins are similar to scales in energy and protein content, but less than that of small whole fish. Orinoco piranhas 80 mm SL fall into two general dietary categories: species that are primarily carnivorous and others that are largely herbivorous. The seven carnivorous piranhas (Pygocentrus caribe, Serrasalmus altuvei, S. manueli, S. rhombeus, S. elongatus, S. irritans, and S. medinai) feed heavily on fish or fish fins. Adults of the four herbivorous species (Serrasalmus cf. eigenmanni, Pygopristis denticulatus, and two Pristobrycon spp.) are seed predators, biting seeds and hard fruits into small fragments before ingesting them. Serrasalmus manueli was one of the few carnivorous species that occasionally took seeds or fruit in mass. Overall, relative intestine length differed among serrasalmine species and was closely correlated with diet. Seedeating piranhas tended to have long intestines, similar to those of highly herbivorous Myleus, Mylossoma, and Metynnis species, but I found no significant difference in relative intestine lengths between forest and savanna populations of S. manueli even though diets differed.

Both carnivorous and herbivorous species are somewhat flexible in their use of

food resources. I found that upper Orinoco piranhas, at both the group and species levels, exploit plant material more than piranhas from the low Llanos. Many piranhas, especially those that are largely herbivorous, shift their diets in response to spatial as well as temporal variation in the availability of fish prey versus plant food. Even though savanna and forest regions have a similar piranha species makeup, herbivorous species are much more common in forested regions. Nevertheless, even though seeds and fruits are a locally abundant resource, assemblages of forest piranhas were typically dominated by one or two large, highly carnivorous species (S. rhombeus, S. manueli, or both).



Introduction to Piranhas

Piranhas (Characidae: Serrasalminae) are neotropical freshwater fishes found in the major Atlantic drainages of South America, from 100 North latitude in the Orinoco River basin of Venezuela and Colombia, through the Amazon basin, and south in the La PlataParaguay-Parana basin to about 350 South latitude. Called caribes in Venezuela, they comprise a monophyletic group of at least 30 species in the genera Serrasalmus, Pygocentrus, Pristobrycon and Pygopristis (Fig. 1-1) (Machado-Allison 1983, 1985, Fink 1988). Piranhas possess many unique morphological traits that distinguish them from other serrasalmines (Machado-Allison 1985), but they are most easily recognized by their unusual and impressive dentition. They have a single row of sharp, somewhat triangular teeth in both jaws; the teeth are tightly packed, partially interlocking, and are typically used for rapid puncture and shearing.

Examples of other serrasalmine fishes are species of Mylossoma, Myleus, and

Metynnis (silver dollars, pdmpanos, palometas), the scale-eating characin Catoprion mento, and the nearly one-meter long Piaractus brachypomus (morocoto) and Colossoma macropomwnum (cachama or tambaquO. In contrast to the piranhas, the dental morphologies of other serrasalmine fishes are variable. Most species have combinations of relatively broad and strong molariform and incisive teeth, often with more than one row of teeth in the upper jaw. Serrasalmine fishes are laterally compressed and most are fairly deepbodied; however, maximum adult size, body shape, and color patterns vary among species. Piranhas range from medium-sized fishes such as Serrasalmus irritans, some 20 cm from



Colossoma Piaractus Mytossoma Myleus

Mylesinus A Utiarichthys


Metynnis B Catoprion

P ygop rist is Pygocentrus PRNA

Pristobrycon Serrasalmus.

Fig. 1-1. Hypothesized phylogenetic relationships among genera of the subfamily
Serrasalmninae as proposed by Machado-Allison (redrawn from MachadoAllison 1985).


tip of snout to base of tail (Standard Length, SL) to large species such as Serrasalmus rhombeus, which grow to well over 35 cm SL. The general body form ranges from sharpsnouted and elongate, Serrasalmus elongatus, to blunt-headed and robust species such as the red-bellied piranhas Pygocentrus caribe of the Orinoco and P. nattereri of the Amazon.

Of the approximately thirty species of piranhas, at least twelve occur in the Orinoco basin where, as in many other areas of tropical South America, they are exploited for subsistence fisheries and also commercially. Piranhas occur in rivers and floodplains of both open savannas and densely forested regions and often make up a high percentage of fish biomass and numbers of individuals (Mago-Leccia 1970, Taphom and Lilyestrom 1984). Thus, they play a major role in the food web of many lowland aquatic communities. Contrary to common belief that they are strictly carnivorous, piranhas and their close relatives exploit a broad spectrum of food resources including arthropods, fishes and other vertebrates, fish fins and scales, carrion, seeds, fruits, leaves, and flowers (Goulding 1980, Sazima 1983, Sazima and Guimardes 19S7, Nico and Taphorn 1988, Winemiller 1989a, Sazima and Machado 1990). Serrasalmine fishes living in tropical savannas encounter substantially different habitat conditions in comparison to those populations occupying forested regions. Thus, comparative studies of piranha diets and trophic ecology can provide insights into their responses and adaptations to the selective forces of different environments

Purpose of Present Study

Tropical South America possesses the most diverse freshwater fish fauna of any continent (Bt5hlke et al. 1978). Considering both described and undescribed taxa, Lundberg et al. (1987) estimated that the Orinoco River basin alone probably contains almost 1,000 fish species, more species than are found in all the freshwater systems of temperate North America. Unfortunately, the extremely diverse faunas and floras of many


tropical environments are seriously threatened. Because of expanding human populations and increased development, natural habitats are being destroyed or degraded at a rapid pace. Forests are being cleared for agriculture, ranching, timber, and minerals; rivers and their floodplains are being modified by dams, levees, and pollution. Aquatic and terrestrial systems in the lowland tropics are intricately interwoven, the changes to each inevitably affecting the other. The profound and likely irreversible changes now occurring throughout tropical America make it increasingly important to understand more about the interrelationships between tropical environments and their diverse fauna and flora. Such an understanding is necessary for successful conservation of habitats, species, and genetic stocks.

The science of ecology attempts to understand the complex interrelationships

between an organism and its physical and biotic environment. Studies in trophic: ecology are a way to investigate such interrelationships, and often perrnit analysis at several levels or scales, Obviously, what an animal eats is limited by the kinds and availability of food resources in its local surroundings. Valid interpretations of feeding patterns, however, also require consideration of past envirom-nental conditions, inasmuch as the probability of an animal taking a particular type of prey depends on its array of interrelated morphological and behavioral characters that result from its particular phylogenetic history. Trophic studies provide insights into the biology of organisms, because food, unlike many other resources or niche parameters, can be measured, analyzed, and compared in a large number of ways using a variety of techniques. Evidence for the importance and practicality of research on trophic ecology is provided by the tremendous volume of literature on the subject (for review of fishes see Ross 1986). Accordingly, the study of diets and food resources has contributed to the development of many ecological concepts, including resource partitioning, niche theory, optimality theory, energetics, food webs, and predatorprey relationships.


The aim of this study is to increase our knowledge of the ecology and natural

history of piranhas and a few of their close relatives. I have attempted to document and explain similarities and differences in trophic ecology within and among closely related species by comparing populations from two floristically distinct regions of the Orinoco River Basin of Venezuela: (1) the upper Orinoco, an area dominated by dense forests, and

(2) the low Lianos of Apure State, a region characterized by immense open savannas.

Piranhas are widely distributed geographically and very abundant in many lowland tropical habitats, but we know relatively little about their ecology, behavior, and natural history. Goulding (1980) found many fishes in the Amiazonian forest ecosystems, including several piranhas, to be herbivorous or omnivorous and to depend heavily on flooded forests for food resources. Influenced by Goulding's findings., Donald Taphorn and I began, in late 1983, to study the trophic ecology and natural history of piranhas and other fishes in savanna ecosystems of the Venezuelan Llanos (Nico and Taphorn 1988, Nico 1990). Although my initial research on piranhas focused on savanna regions in the Orinoco Basin, I also made visits to southern Venezuela and the Rio Casiquiare in 1985, and to MaraJ6 Island in the lower Amazon of Brazil in 1986. Subsequently, between May 1988 and March 1991, I participated in five expeditions to the upper Orinoco basin and its forests to broaden my investigation of the feeding ecology of Orinoco piranhas as related to their habitat.

Waters of the low Llanos are typically high in fish and invertebrate biomass, and many sites support large numbers of carnivores (Mago-Leccia 1970, Taphorn and Lilyestrom 1984, Machado-Allison 1987, Saunders and Lewis 1988, Rodriguez and Lewis 1990). However, large woody plants are uncommon, and in conjunction with water stress caused by the yearly dry periods, production of seeds and fruits is highly seasonal (Sarmiento 1984). As a result, availability of plant materials, especially fruits and large seeds, as food for savanna fishes is low, markedly seasonal, and very localized (personal observations). On the other hand, the few phenological studies of wet forests in lowland


areas of tropical South America (Terborgh 1983, Goulding et at. 1988, Janson and Ernmons 1990), or information gleaned from botanical exploration of the upper Orinoco (B. Stergios, personal communication), suggest that such habitats make a wide variety and abundance of fruits and seeds available to fishes during most of the year (Goulding 1980, Gouilding et al. 1988). Even though species diversity is often high in the upper Orinoco, our collections made with a variety of fishing gear indicate that fish biomass in tropical forest regions is generally lower than in the low Lianos.

If present ecological conditions are important, environment should strongly influence both species composition and trophic ecology of piranha assemblages in particular regions. Accordingly, comparisons of piranhas from the more heavily forested upper Orinoco basin with those from the open savannas or low Llanos were made to address the following questions: (1) Do the same species of piranhas occur in both savanna and forested regions? (2) Do diets differ, and, if so, are fish from populations in forested habitats significantly more herbivorous than their saivanna counterparts? (3) Do ontogenetic shifts in diet occur and, if so, how do they compare among species and do they correlate with regional environments? (4) Because gut length is often an indicator of natural diet in vertebrates (i.e., longer in herbivores), is intestinal length of piranhas and other serrasalmine fishes correlated with diet? (5) What are the nutritional benefits associated with different diets or food preferences (e.g., scale- or fin-eating versus flesheating)? (6) How do diets of piranhas compare with other serrasalmine species? (7) What is the relationship between piranha ecology and serrasalmidne phylogeny?

The comparative approach is useful in helping us understand aspects of the

evolution and adaptations of closely related species (Clutton-Brock and Harvey 1984, Huey and Bennett 1986). Recent comparative studies in tropical vertebrate ecology have focused on birds (Schluter 1988), mammals (Terborgh 1983), and reptiles and amphibians (Duellman 1990). An objective in many comparative studies is discovery of evolutionary trends in morphological and behavioral adaptations to different environments, thereby


making it possible to recognize how particular characters arise under the influence of contrasting selective forces (Clutton-Brock and Harvey 1984). Furthermore, comparisons across a broad biogeographical scale offer what Wiens et al. (1986) suggested as perhaps the best opportunity to understand a species' limits or responses to different environmental conditions, revealing patterns that suggest something of its adaptive abilities or evolutionary potential.

Literature Review

The following is a brief review of the literature relating to piranha feeding behavior and trophic ecology. A summary of field studies investigating piranha feeding and diets is given in Table 1 -1. Unless otherwise stated, scientific names are as used in the original literature. All references to Serrasalmus nattereri should be Pygocentrus nattereri.

Early naturalists and explorers were the first to report on piranhas, and most accounts, such as those by Alexander von Humboldt, Henry Bates, and Theodore Roosevelt, were primarily devoted to describing attacks by piranhas on animals and humans, or relating the destruction done by piranhas to fishing gear (Bates [1863] 1975:202, Roosevelt 1914). Hum-boldt, who traveled extensively in South America between 1799 and 1804, wrote the following passage while on the Orinoco River:

On the morning of the 3rd of April our Indians caught with a hook the fish
known in the country by the name of caribe* [*Caribe in the Spanish
language signifies cannibal, or caribito, because no other fish has such a
thirst for blood. It attacks bathers and swimmers, from whom it often bites
away considerable pieces of flesh. The Indians dread extremely these
caribes; and several of them showed us the scars of deep wounds in the calf
of the leg and in the thigh, made by these little animals. They swim at the
bottom of rivers; but if a few drops of blood be shed on the water, they rise
by thousands to the surface, so that if a person be only slightly bitten, it is difficult for him to get out of the water without receiving a severer wound.
Humboldt ([1816-1831] 1971:166-167)


Eigenmann (1915), Eigenmann and Allen (1942), and Myers (1972) have reviewed many of the earlier stories about piranhas. Although published anecdotal information brought notoriety to this group of South American fishes, early accounts did not significantly contribute to our knowledge of piranha biology and natural history, and many stories simply aided to perpetuate myths about piranha ferocity.

There have been a moderate number of studies that have addressed the natural diets of piranhas, (Table 1-1). Most of these have focused on only one or a few of the most common and widespread species, from a single locality or region. The diets of several of the more uncommon species have never been described. Most studies of piranha diets have been of fishes inhabiting artificially altered environments. Because a few of the more common piranha species seem to flourish in man-made lakes, several studies focused on populations inhabiting reservoirs or impoundments. Work on piranhas has also been done in watersheds modified to some extent by cattle ranching, agriculture, and by construction of roadways. Few investigations have been carried out in Pristine habitats.

Several of the earliest efforts to characterize piranha diets were carried out in the

Jaguaribe River basin, a coastal drainage located in northeastern Brazil. The first was that of Menezes and Menezes (1946). Along with information on several other species, they presented a simple listing of food items, primarily fish remains and insects, found in the stomachs of 138 specimens of a single piranha species taken in 1944 from the Lima Campos Reservoir in Ceard State. First identified by Menezes and Menezes as Serrasalmo immaculatus, Braga (1954) later indicated the species was Serrasalmus rhombeus. Braga also studied Lima Campos piranhas. He recorded the frequency of occurrence and volume of stomach contents of a large sample of S. rhombeus (n = 2,222; 100-290 mmn total length) during a 12-month study, 1952-1953 (Braga 1954). Of the 1,713 individuals containing food, predominant food items were freshwater shrimp (Palaemonidae) and fish, 65 and 28 percent of the total food volume, respectively. Braga later published results from other studies on piranhas in a single volume (Braga 1975), providing diet information

s S


0 ;g


cn N gi 6 xA
00 Cf)
cd N rcz d) i .- 11 11 11 11 cq
z 0


t46) r-o

Con 1,
-j C0,0.1 "i C'n cn
Cd C14 M
-4 S

Cc,' 0', r. O'\

tn .- 0 C14
C3% 00 E c
W) c ON
w C4 (ON r- C;,\ cn


4-4 co






9.5 0=0


to cn
00 CN
W) eq



zs :i M

co cn t
cn 'Tr W) 1-4 cq M

-0 00
5 Cc, "o cn

w 4 C

P 00 i I a 2 u 2 .6 00 cc
C6 00 ON
E- -tt

48, i
> 0 11 A .1 -0 R




to to 0>,

a o co

000 Cc,
C 1-4 ,Ct
to 00 cq
A 8 1 -V ". C
!ZI C"n lc l C) 10
x C-- cq N 00
r-- cq 0 c oll M 00 0 10
W) N W) .t 1 0 W) rto

Lt zi
tz Ft

0, cn Cn Ln Cn Ln
Cn tn Cn M

(ON 00
r- 00
ON 79 Cl\
00 00 r.
0 0 00 0

cq A
all C



ZA o

4t. Itz

0 PQ 5


on S. nattereri from three other reservoirs of the Jaguaribe basin. In that work, Braga concluded that piranhas were carnivorous, preferring to feed on other fishes, although they also prey on other vertebrates (i.e., birds and amphibians), some invertebrates, and only occasionally on plant material, such as seeds, parts of leaves, and algae.

Bonetto et al. (1967) reported on the feeding habits of two piranhas, Serrasalmus spilopleura (n = 104) and S. nattereri (n = 21), taken from the middle Parand River and its floodplain in Argentina. Both species fed predominantly on fish, but prey were not identified to lower taxa. Stomachs of many of the individuals examined contained small amounts of plant material, for example, roots and leaves of the aquatic plant Salvinia sp. and roots of Eichhornia sp. (water hyacinth); however, these items were thought to have been taken accidentally during capture of their principle prey.

Lowe-McConnell (1964) studied the ecology of fishes in the Rupununi savanna of the Essequibo River basin, British Guiana (present day Guyana). Although not focusing on fish diets, she briefly mentioned that the stomachs of fishes trapped in drying pools, including the piranhas Serrasalmus nattereri, S. rhombeus, and Pygopristis denticulatus, contained only mud and bottom debris; however, these same fishes had large amounts of body fat. From her work in the Essequibo River drainage, Lowe-McConnell concluded that the dry season was a period when fishes exhibited reduced feeding and high diet overlap.
Roberts (1970) was the first to report on the more unusual feeding practices of piranhas when he found large amounts of fins and some scales in the stomachs of seven Serrasalmus elongatus from several sites in the Amazon River basin. He also discussed scale-eating by the serrasalmine Catoprion mento. Saul (1975) collected two piranha species during his study of habitat and food preferences of fishes at a site along the Aguarico River, a tributary of the Napo River in the western Amazon Basin, Ecuador. He found the stomachs of Serrasalmus nattereri (n = 6) to contain fish remains and insects;


those of S. marginatu (n = 4) held fish and fish remains. Both species were found primarily in floodplain lakes or side channels, as opposed to the main river.

The relationship between piranha feeding patterns and environment had not been

investigated in any detail, however, until Goulding (1980) studied the diets of medium and large-sized adult fishes, including seven piranha species, from in and around the Machado River, a clearwater tributary of the turbid-water Madeira River in Amazonia Brazil, a region of seasonally inundated forests. He found that large Serrasalmus rhombeus (n = 254), the most common piranha taken, were primarily pisciv ores that often bit chunks of flesh out of large fish. Of added interest, fruits and seeds made up an estimated 10 percent of the total volume of food eaten by S. rhombeus. Adult S. elongatus (n = 85) collected by Goulding were similar to the juveniles examined by Roberts in that they had fed heavily on fins and scales. Goulding reported that five of the seven species fed heavily on seeds, which they masticated before swallowing. These five piranhas, which were relatively uncommon in samples, included Serrasalmus serrulatus (n = 36), Serrasal.'Mus cf. striolatus (n = 9), and three unidentified Serrasalmus species (n = 5?). Goulding concluded that Amazonian fishes had evolved a close interrelationship with lowland forests, and that many fishes depended on flooded forests for food in the form of fruit, seeds, and other allochthonous materials. In addition to being seed predators, Goulding (1983) later indicated that piranhas sometimes took seeds whole, and suggested that these fish also acted in dispersal of seeds.

After Goulding completed his work in the Madeira River basin, he was joined by Carvaiho and Ferreira (Goulding et al. 1988) in a study on the natural history of Amazonian fishes in habitats associated with the main channel of the lower Rio Negro. Of approximately 450 species taken in the Rio Negro, six were piranhas. They did not discuss piranha diets in much detail, but reported that fin and scale eating by adult piranhas seemed to be common. Their collections of juvenile piranhas from the Rio Negro are yet to


be analyzed. In the published proceedings of a symposium Ledo et al. (1989) reported on the morphology, reproduction, and diet of Serrasalmus altuvei from the Rio Negro.

Sazima and his colleagues have explored the behavior of piranhas and related species, particularly in reference to fin and scale-eating, through laboratory work with captive fishes and underwater observations of free-living fishes in the Pantanal region, western Brazil (Sazima 1983, 1986, 1988; Sazima and Zamprogno 1985, Sazima and Pombal 1988, Sazima and Machado 1990). Sazima and Guimardes (1987) recognized Pygocentrus nattereri and Serrasalmus spilopleura as scavengers, feeding on dead and sometimes decaying fishes, birds, and mammals. They suggested that deaths of humans attributed to piranhas might often be the result of scavenging on victims already dead from drowning or other causes. Sazima and Machado (1990) presented data from their underwater observations on the behavior of three species, S. marginatus, S. spilopleura, and P. nattereri, with emphasis on their predatory tactics, feeding behavior, and social interactions while foraging.

Ferreira (1984b) studied the food and feeding habits of the principal fish species inhabiting the Curudi-Una River Reservoir, in the Amazon Basin near Santar6m, Brazil. Ferreira found that the diet of Serrasalmus rhombeus, taken from five localities in the reservoir, showed slight differences among stations, feeding mainly on fish at four sites, and on aquatic insects (i.e., ephemeroptera) at a fifth locality. In total, 250 specimens were examined (ranging from 50 to 320 mm SL) of which 196 piranhas contained food.

Northcote et al. (1986, 1987) reported fin feeding by Serrasalmus spilopleura from the Americana Reservoir, an impoundment on the Atibaia River in southeastern Brazil. Analysis of prey fishes with damaged fins showed that S. spilopleura usually nip the lower lobe of fins, indicating that attacks are made from below and behind prey fishes. They also suggested that S. splilopleura was the main, if not the only, fin-feeder in the reservoir fish community. Mota et al. (1982) studied the seasonal variation in the intensity of feeding, as measured by stomach fullness, of S. spilopleura from the Bariri Reservoir in Sdo Paulo


State, Brazil. They reported that heaviest feeding took place during the summer months with a corresponding heavy build up of body fat by autumn.

Several recent works dealing with piranhas in the Venezuelan Llanos have

documented diet shifts during ontogeny (Machado-Allison and Garcia 1986, Nico and Taphorn 1988, and Winemiller 1989a). All found that early juveniles ate principally zooplankton, while older individuals were predominantly piscivores. In addition to discussing ontogenetic changes in dental morphologies, Machado-Allison and Garcia (1986) examined 134 specimens, mostly juveniles, of three species (Pygocentrus caribe Pristobrycon striolatus, and Serrasalmus rhombeus) taken from the Camaguin lagoon, a large wetland between the Portuguesa and Apure rivers, Guarico State. Predominant food items, in terms of frequency of occurrence, of young piranhas were microcrustaceans, aquatic insects, fins, and scales. In addition, they found small seeds (i.e., Cyperaceae) in the stomachs of 33 of 134 specimens, but did not provide information on volume and numbers of seeds eaten. Nico and Taphorn (1988) examined the diets of seven piranha species from streams and flooded savannas in the western Llanos of the Orinoco River Basin. Their work suggested that savanna piranhas do not depend on fruits and seeds as had been reported by Goulding (1980) for piranhas in flooded tropical rain forests. Nico and Taphom found that juveniles and sub-adults of six of seven piranhas from a single locality fed heavily on fins. Winemiller (1989a) studied the fish community of Cario Maraca, a seasonal stream in the Apure River drainage that meanders through a region of wooded savannas used for cattle ranching and cropland in Portuguesa State. His findings on four piranha species (P. caribe, S. irritans, S. rhombeus, and S. medinai), mostly juveniles and sub-adults, supported the earlier work of Nico and Taphorn.

For the most part, piranhas have been found to be daytime feeders, usually most active during twilight periods (Lowe-McConnell 1964, Goulding 1980, Barthem 1987, Nico 1990, personal observations). Barthem (1987) reported on activity patterns of several vdrzea lake fishes from near Manaus, Brazil. His results, based onanalysis of gilinet


catches, suggest an early morning peak of activity in Pygocentrus nattereri, Serrasalmus elongatus, and S. rhombeus, and a late afternoon peak for the latter two species. Nico (1990) found that young Pygocentrus notatus ( = P. caribe) had a peak feeding activity in the morning, whereas adults seemed to be more active near dusk.

Intestine length of a few piranha species has been measured by Braga (1954),

Luengo (1965), Bonetto et al. (1967), and Jdgu and dos Santos (1988). Braga reported that the digestive tract of Serrasalmus rhombeus averaged 1.4 times the standard length. Luengo (1965), relating intestine length to natural diet, reported that the highly carnivorous Serrasalmus nattereri ( = Pygocentrus caribe) taken in the Venezuelan Llanos had a short intestine compared with that of Prochilodus reticulatus ( = Prochilodus mariae?), a prochilodid fish that feeds primarily on organic mud and detritus. Martfnez (1976:67) studied the anatomy of what was identified as Serrasalmus nattereri ( = Pygocentrus caribe) and reported 10 to 12 intestinal caeca originating near and slightly past the pylorus. In a taxonomic treatment, J6gu and dos Santos (1988) provided information on intestine length and number of pyloric caeca for several piranha species from the Rio Tocantins of Brazil, but they did not suggest dietary correlations.

Organization of Dissertation

The content and sequence of the chapters reflect the major components of my study. Chapter 2 provides an overview of the Orinoco River basin and describes the environment of the two study regions. Chapter 3 identifies the study sites and details field and laboratory methods used. Chapter 4 compares the species composition of piranha assemblages between the two regions and among the different drainages. Chapter 5 presents results from long-term studies on the trophic ecology and natural history of savanna populations of piranhas from a site in the Apure River drainage. Chapter 6 focuses on the trophic ecology of populations of serrasalmine fishes from six drainages in


the upper Orinoco River Basin, a region characterized by extensive forests that undergo seasonal flooding. In addition, my findings from the upper Orinoco are compared to results from my studies of savanna populations, including data on a few savanna species not reported in previous studies. Chapter 7 then gives results of an analysis of the nutritive values of various food items commonly taken by piranhas. Finally, Chapter 8 summarizes the major findings of the study, and attempts to interpret and synthesize the results in terms of relevant mechanistic, ecological, and evolutionary issues.


In this chapter I give an overview of the Orinoco River basin and describe the two regions dealt with in the present study: (1) the open savannas or low Llanos of Apure State, and (2) the upper Orinoco in the Venezuelan Federal Territory of Amazonas, an area dominated by forests. Little is known about the geological and ecological history of tropical lowlands in South America (Whitmore and Prance 1987). Because historical events undoubtedly influenced present-day ecological patterns, I briefly review several hypotheses concerning the basin's and the river's formation, as well as climatic and vegetation changes that purportedly took place in northern South America during the recent past.

The Orinoco River Basin

The Orinoco River, although dwarfed by the Amazon, is a major tropical river. The basin is located in northern South America between 2' and 10' N latitude, and empties into the Atlantic Ocean (Fig. 2-1). The Orinoco's mean annual discharge is about 35,800 m3/sec from a drainage area of approximately 1, 100,000 km2, located in Venezuela (70%) and Colombia (30%) (Lewis 1988). Compared to the Mississippi River. the Orinoco has more than twice the annual discharge, yet drains an area of only one-third the size. The basin is geologically and ecologically diverse, having large mountain ranges and broad low-lying plains, with habitats ranging from inu-nense open grasslands to lush tropical forests.






0 V)


0. (n cli cc
w co w




The headwaters of the Orinoco are traditionally considered to be in southern

Venezuela where it receives water from both the western slopes of the Guyana Shield and from extensive lowland areas dominated by deciduous or evergreen tropical forest. In contrast, most tributaries in the northern and western parts of the basin originate in the Andes and adjoining coastal mountains and then flow through the Lianos, a vast alluvial plain supporting tropical savanna vegetation. The main channel of the Orinoco forms an unusual U-shaped pattern, the river winding around the western borders of the Guyana Shield. From its source the Orinoco flows several hundred kilometers west-northwest until it is met by the Atabapo and Guaviare rivers; here the river makes an abrupt turn to the north. Along its northward segment the Orinoco forms the border between Colombia and Venezuela.

At about 5'30'N latitude the river courses through several series of treacherous cataracts near Puerto Ayacucho, the most prominent being the Atures Rapids where the river drops 17.5 m over a distance of 9.3 km (CODESUR 1979). The Attires Rapids and the other cataracts are spread along a reach of nearly 100 km, marking the boundary between the upper and lower sections of the Orinoco. Farther downstream the Orinoco is joined from the west by the Meta, Cinaruco, Capanaparo, and Apure rivers as it gradually swings to the east. Finally, below the confluences with the Caura and Caroni rivers the Orinoco arrives at its large delta. Here the huge main river is subdivided into a multitude of smaller channels that empty into the Atlantic Ocean near the island of Trinidad. The main causeway of the Orinoco, from its origin in the Sierra Parima in the Guyana Shield to its mouth, is estimted to be 2,140 km long (DISCOLAR 1983). High-order tributaries draining highland areas of the Andes and the Guyana Shield are characterized by large numbers of waterfalls, rapids, and fast-flowing waters. However, the larger streams and rivers common to lowland areas typically have slight gradients. The upper Orinoco, at the


community of La Esmeralda, nearly 1,800 km by river from the sea, is only 135 m above sea level, an average drop of less than 1 m for every 10 km of river.

In summary, large rivers of lowland forests that feed the upper Orinoco from the south include the Mavaca and Atabapo rivers. Big-river tributaries originating in the Guyana Shield and feeding into the upper Orinoco include the Ocamo, Padamo, Cunucunuma, Ventuari, and Sipapo; those flowing into the lower Orinoco include the Caura and Caroni rivers. Large tributaries of the Orinoco that originate on or near the Andean slopes are the Inirida, Guaviare, Vichada, Tomo, Meta, Cinaruco, Capanaparo, Arauca, and Apure rivers. About 31 percent of the upper Orinoco volume is continually diverted into the Amazon basin by way of the Brazo Casiquiare, a natural canal (CODESUR 1979).

In addition to a diverse flora and topography, the Orinoco River basin is a mosaic of water types. Alfred Russel Wallace ([1853] 1972) was one of the first to mention the striking differences in water colors among the various South American waters using the terms whitewater, blackwater, and clearwater rivers. In a later and more detailed analysis, Sioli (1975, 1984) summarized differences among these three water types based on differences in water color, amount of suspended matter, and chemical properties. Whitewater streams are turbid because of inorganic sediments, of milk-coffee color (cafe con leche), relatively rich in nutrients, and of near neutral pH. Black-water streams are transparent but reddish brown, the color of tea due to humic acids (the color appears black where water runs deep), very poor in nutrients and very acidic. Clearwater streams are also transparent, often greenish in color, typically poor in nutrients, and have a pH that can range from slightly acidic to slightly alkaline.

In the Orinoco River basin, whitewater and blackwater habitats predominate. Clear waters are more typical of small carlos and the upland reaches of certain rivers. However, the classification of rivers by water color is not always easy to apply and thus the usefulness Sioli's classification scheme is somewhat limited. Many rivers are a mixture of


two or more of the basic water types; some change seasonally as flows from a diverse array of smaller tributaries rise and fall. For example, during high water the Guayapo River of the Sipapo drainage changes from a clearwater river in its upper and middle reaches to a blackwater river in its lower segments, due to increased input of tannin-stained waters from adjacent flooded forests. To a large extent, the hydrochemistry of the Orinoco tributaries is controlled by the geomnorphological and floristic character of their watersheds. In general, rivers of the Andes and Lianos are high in suspended sediments, whereas streams originating in the weathered Guayana highlands and the forests of the upper Orinoco transport low sediment loads. Nevertheless, exceptions to the above-mentioned hydrochemical pattern can be found in all regions as a result of local conditions.

History of the Basin

The details and precise timing of geological events 1 evolvedd in the formation of the Orinoco River basin are not well understood. The oldest land form in the basin is the Guyana Shield, ancient mountains composed of Precambrian material that underwent additional uplift and exposure during the Tertiary (Whitmore and Prance 1987). Because they have been exposed for many millions of years, these highlands have been heavily worn and dramatically shaped by erosion. The mountains of the Andes were created later, uplifted as a result of plate tectonic events during the Miocene-Pliocene, 2-8 million years ago (Whitmore and Prance 1987:7). The Llanos are thought to be the most recent major topographical feature in the basin. During Quaternary periods of raised sea-level (interglacials) the Llanos plains were the bed of a large interior sea. Thus, similar to many lowland areas in South America, these central plains are primarily composed of Quaternary sediments that niantle Tertiary and older sediments, with recent additions of alluvial river deposits carried down from the surrounding mountains during the Pleistocene (Beek and Bramdo 1968:86, Walter 1973: 72-73, Cole 1986).


An ancient Orinoco River is thought to have flowed to the northwest, emptying into the Caribbean before the Miocene uplift of the Andes near or through the present day Maracaibo Lake Basin. In addition to geologic evidence, this idea of a northward flow has received added support from comparisons of fossil and living fishes found in northern Colombia and Venezuela (Lundberg et al. 1986, 1988). According to the proposed scenario, the river slowly shifted to the east some time between the Late Cretaceous and early Oligocene, finally assuming its present course by the Oligocene epoch, about 30 million years ago (Lundberg et al. 1988). However, it is still uncertain how the river's movements affected or coincided with the fon-nation and eventual draining of the sea that covered the interior lowlands at about the same time periods (Taphom 1990).

The Orinoco and Amazon rivers are connected today by the Casiquiare canal, a natural waterway flowing southward from the upper Orinoco into the upper Rio Negro, permitting interchange of fishes between the two basins. However, the Orinoco and Amazon basins may have had an even closer association in the past, partly based on studies in western Brazil. The Amazon once flowed westward to the Pacific before the major rise of the Andes in the Miocene (Goulding 1980, Putzer 1984). After the Andes uplift blocked its westward flow, a huge lake, or more likely, a series of large lakes and wetlands, were formed covering much of the lowland areas in the basin (Frailey et al. 1988). Eastward flow was prevented at the time by higher elevations joining the Brazilian and Guyana shields. Before the Amazon had completely carved its way east between the two shields, the proposed Pleistocene/Holocene lake or lakes had theil prime outlet north into the Orinoco. Indeed, tectonic events may have resulted in shifting the primary outlet several times between the north and east (Frailey et al. 1988).


The Low Llanos

The Lianos (Fig. 2-2) are a vast plain of up to 400 kmn width along the north bank of the Orinoco, stretching from near the delta mouth westward for over 1,000 km through Venezuela, far into Colombia (Walter 1973). It occupies about 500,000 km2. is essentially unbroken by major forests, and is the largest continuous savanna in the neotropics north of the equator (Sarmiento 1984). A combination of factors, for instance soil type and rainfall patterns, prevent the establishment of forest in large parts of the Llanos. Dry season fires, of mostly human origin, are also important in limiting the growth of woody vegetation.

Several systems have been used to divide the Llanos into different categories using combinations of physical and biological features. In most cases, a distinction is made between the low Llanos and high or upper Lianos, based on elevation and extent of flooding during the rainy season (Walter 1973: 73, Cole 1986). The low Llanos are characterized by open grass savannas with fairly level terrain that undergo frequent or long term flooding during the rainy season. According to Cole (1986:92), savanna grasslands cover some 150,000 km2 of the low Llanos in the Orinoco basin. Apure State, in western Venezuela, encompasses a large portion of the low Llanos. Here the terrain is relatively flat, with elevations of 100 mn or less above sea level. In Apure, the landscape is dominated by herbaceous cover, mainly bunch grasses and sedges, with occasional shrubs and small trees (Sarrniento 1984, Cole 1986). Most trees, including palms, exist in small isolated groups (matas) or in gallery forest situations fringing the rivers and larger streams (Sarmiento 1984, Cole 1986). These gallery forests range from narrow bands, only a few trees wide, to fairly extensive stands, in some places many hundreds of meters wide, as seen existing along reaches of the lower Cinaruco and Capanaparo rivers.

The low Llanos of Apure State (Fig. 2-3) comprise a vast open floodplain

containing a diversity of aquatic habitats (Mago-Leccia 1970). All undergo pronounced seasonal changes. Because of its low elevation and flat physical relief, most of the


;D0 0

V, X0

.. . . .. . . .-- .
i~~~......... ......... !, i

.... ......
...... .... ........:::::::@

tD~ .... ...
*0z C.

. . . . .

.......... ....

..... I t

o-'- /




1, 4


14 4,-q


O.L 0i


landscape floods during the rainy season, whereas surface water during the dry season is usually limited to a few large permanent lagoons, large streams, and rivers. As the dry season progresses, the savanna vegetation turns yellow and the earth becomes hard and is fissured with cracks from the intense heat of the sun. Rivers and streams in the low Llanos are of low gradient, with many wide meanders and anastomoses. The Apure River and many of its tributaries are whitewater in character, with a relatively high sediment load and high conductance (Saunders and Lewis 1988). In contrast, the Capanaparo and Cinaruco are essentially blackwater rivers, with moderate sediment loads. In addition to stream channels, the landscape of the low Llanos is dotted with shallow marshes (esteros) that hold water after floods or rain. Lagoons (lagunas) form in deeper depressions, some created from abandoned stream meanders, and are more permanent aquatic habitats. Human population density is low and most of the low Llanos is used for cattle ranching. Since the early 1970's extensive networks of low earthen dikes, called modules (mdulos), have been constructed, surrounding large areas of the floodplain savanna. These modules lessen the extreme effects of seasonal changes within their boundaries by controlling flooding and by holding water in lower areas, available to cattle year round. The hundreds of borrow pits (prestamos) excavated during the construction of these dikes, as well as from building of roads, collect water and serve as permanent and semi-permanent artificial lagoons. Typically less than 25 m wide, some prestamos stretch unbroken for several kilometers.
During high water, adjacent rivers and streams in the Llanos are often connected by sheet flooding of the savanna or by filling of small interconnecting channels. Thus, fish are generally able to move across the entire floodplain. However, where dikes or roadways are present, run-off patterns have changed and local fish movements are somewhat restricted (Taphorn and Lilyestrom 1984, Nico and Taphorn 1988).

Many of the wetlands in the Llanos contain large amounts of organic material and become highly productive during the rainy season (Mago-Leccia 1970, Saunders and


Lewis 1988). During high water, the flooded savanna habitats are important to fish as breeding and feeding areas. Aquatic vegetation flourishes and emergent and floating plants may cover large portions of the surface of both permanent and temporary water bodies. Large mats of water hyacinth (Eichhornia species) are especially abundant in many of the lagoons, prestamos, and some of the streams. Partially inundated grasses and sedges commonly grow in the shallows of streams, prestamos, and other standing waters. Other common aquatic and semi-aquatic plants of the low Llanos include Ludwigia, Benjarnania, and Pontederia. Some common aquatic plants with floating leaves are Salvinia, Lemna, and Pistia.

The Upper Orinoco

The upper Orinoco drainage is situated in the Federal Territory of Amazonas

(Fig. 2-4) in southern Venezuela, a vast area covering 178,095 km2 (CODESUR 1979). Main features of the heavily forested landscape are flat alluvial plains, large mountain valleys, and massive granitic hills (lajas) (CODESUR 1979, Huber and Wurdack 1984). About 70 percent of the territory is drained by the Orinoco, with the remainder flowing into the Rio Negro of the Amazon basin (CODESUR 1979). Together with the adjacent state of Bolivar, this southern section of the country is often referred to collectively as Venezuelan Guyana because the Precambrian Shield dominates the topography. Little developed and with few inhabitants, the great majority of intact forests in Venezuela occur in this southern region. Steyermark (1982) estimated that forests covered 82 percent of the region, or about 146,000 km2. Due to forest clearing, the total coverage is probably slightly less today. Because forests of the upper Orinoco are continuous with those of the northern Amazon Basin, the region is part of the largest area of humid forests within the neotropics (Whitmore and Prance 1987).




0 0

670 660 650 640

Fig. 2-4. Approximate distribution of lowland forests (shaded area) in Federal Amazon
Territory of Venezuela.


The flat lowlands or peneplains of southern Venezuela are domidnated by wet tropical forests (Figs. 2-2 and 2-4). The altitude of these lowland regions in the upper Orinoco range from 100 to 500 m, with most less than 200 m (Huber and Wurdack 1984). Similar to the situation described for Amazonian lowlands (Prance 1979, Kubitski 1989), many lowland forests in the upper Orinoco are inundated by the annual high water, with some sites being permanently waterlogged. Aerial photos of the lowlands reveal a history of streams meandering across their adjacent floodplains. Large and small streams, cutoffs, and oxbows abound in many of the drainages (e.g., Mavaca River) and, similar to the low Llanos, interconnections with adjacent drainages within the upper Orinoco are common during high water. However, aquatic vegetation in the region is much more sparse and localized in comparison to the low Llanos. Although a few backwaters are choked with submergent and emergent vegetation, most floodplain lakes in the upper Orinoco have few or no aquatic plants. The reasons for such scarcity are likelIy related to high shade, unconsolidated bottoms, and few available nutrients in the water or substrate. In flowing waters there are often small bands of flooded grasses along bank edges of inner bends. The fish fauna of the upper Orinoco is the least known in all of Venezuela, largely because of the inaccessibility of the region. With very few roads, modes of transportation within the area are limited almost exclusively to small and medium-sized boats and small planes.

Savannas and Forests during the Quaternary

Both tropical forest and tropical non-forest vegetation originated in the Cretaceous and Tertiary, about 100 million years ago, periods when world climates were apparently rather stable and tropical conditions much more widespread (Whitmore and Prance 1987). However, during the last two million years there have been periodic and dramatic fluctuations in the distribution of savannas versus forest environments in northern South


America that are associated with climatic oscillations (Prance 1982, Whitmore and Prance 1987). Although lowland regions remained essentially tropical, during major glaciations the equatorial lowland regions experienced dryer and hotter climates (but see Colinvaux 1976). As a result, savanna and open-land environments expanded, whereas humid forests contracted. During these dry periods, forests were often reduced to combinations of variably-sized isolated refuges and fingers of gallery forests along major rivers. Indeed, new data on Pleistocene mammals indicate that savanna habitat may have expanded to such an extent as to form a continuous band connecting the Orinoco Llanos with the pampas of Argentina by way of the western Amazon basin (Rancy 1991). The last major glacial event causing a cold, dry Pleistocene climatic phase occurred 13,000-18,000 years ago.

Much controversy exists about the extent and precise location of forests during the Pleistocene dry periods. Haffer (1987:12) presented a map outlining possible distribution of both relatively humid areas and tropical forests, forest "refugia", surviving in northern South America during that period of savanna expansion. His map delineates three major regions of hurnid forests in the Orinoco basin: several sections in the Andes; a block near the southern tip of Venezuela; and a long broad belt (100-200 krn wide) south of the main stem Orinoco, extending from near the mouth of the Meta River eastward through the Guianas.

In contrast to glacial periods, interglacial or pluvial periods were wetter and relatively cooler, contributing to the expansion of humid forests and contraction of savannas. Isolated savannas curTently found in the upper Orinoco are thought to be savanna refugia that survived the cooler and wetter phases. The present climatic condition is wet and warm (i.e., as opposed to dry and hot), and humid forests are at or near their maximum extent, whereas the extent of natural savannas and dry forests is restricted (Whitmore and Prance 1987:63).


Present Climate and Seasonality

Regional topography and the fact that the entire Orinoco Basin lies immediately

north of the equator are important factors controlling climate and the timing of precipitation (Walter et al. 1975). There is a trend toward increased rainfall from north to south. Nevertheless, rainfall patterns can be quite variable from year to year in a single locality, as well as among different localities within any particular region.

The Lianos have consistently high temperatures and fairly distinct wet and dry

periods. The rainy season is associated with the inner-tropical convergence zone (Walter et al. 1975), and lasts from seven to nine months, usually from late April to October or November (Fig. 2-5). Little or no rain falls during the two or three driest months, when the inner-tropical convergence zone is in the southern hemisphere. Total rainfall in the Llanos ranges from 1200 to 1800 mnm per year (Taphorn 1990), with savanna grasslands common to the low Llanos typically receiving 1300-1400 mm (Cole 1986:97). The recording station at San Fernando de Apure, located in the center of the low Lianos, reported a mean annual precipitation of 1416 mm, based on 2 6 years of records (Walter et al. 1975); mean monthly temperatures for the locality varied between 26.6 and 29.2 TC (Cole 1986:97). However, daily temperatures can fluctuate widely.

Although the two regions have a fairly similar range of air temperatures, the upper Orinoco is much more humid than the Llanos. Average annual rainfall is 3,250 mm/yr, with very humid areas receiving well over 4,000 mm (CODESUR 1979). Although the amount of rainfall varies seasonally, the dry season in the upper Orinoco is not as pronounced as in the Llanos (Fig. 2-5). Few areas of record receive less than 75 mm in any given month during the dry season (CODESUR 1979). Nevertheless, the alternation of low and high water seasons is distinct. Highest precipitation occurs between May and September and the lowest between December and March (CODESUR 1979). Monthly mean temperatures range from 26 to 28.5 TC (CODESUR 1979). Both the low Llanos of


Apure State and the lowland forests of the upper Orinoco undergo extensive flooding during the wetter part of the year by a combination of rising river water and local rains.


-'-- Upper Orinoco
Low Llanos 300


S 200


Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month

Fig. 2-5. Monthly rainfall recorded at representative sites in Orinoco River basin, Venezuela. (1) Upper Orinoco: monthly average 1971-1977 from Santa Maria
de Los Guaicas, mouth of Ocamo River, Federal Territory of Amazonas, and (2)
Low Llanos: for year 1972 from Caflo Caicara area, Apure River drainage,
Apure State. Sources: CODESUR (1979) and Taphom and Lilyestrom (1984),



Fl Sapng

In the field I tried to collect as many individuals, sizes, and species of serrasalm-ine fishes as possible from each region and drainage, but I did not attempt to measure fish biomass. Sampling of fishes was done in all major habitats using appropriate combinations of collecting gear and fishing tactics so as to obtain the most information in a short period of time. Although piranhas were taken in greatest numbers during the day, especially near twilight, night collections were also made in the two major regions studied. For sampling piranhas, I depended mainly on various combinations of seines, gill nets, cast nets, and hook and line. Harpoons, spear gun, midnnow traps, trotlines, and commercial (5% rotenone) and natural (barbasco) fish toxicants were used only rarely; these capture methods contributed little to the total sample of serrasalmine fishes taken. In the field, most fishes collected were preserved immediately in 10-20% formialin; in the laboratory, preserved specimens (those less than about 200 mm SL) were transfered to 70% ethanol or 35% isopropanol. A few young piranhas were kept alive and maintained in aquaria.

Each type of sampling gear targeted a certain size range of fishes. For collecting

young piranhas and other small fishes, I used small nylon seines (6x2 and 5x2 m, 6.4 mm. mesh), nylon and monofilament cast nets ranging from 1.8 to 3 mn diameter (2 to 2.5 cm mesh), and hand nets (3 mm mesh); these were normally limited to use in shallow areas (<1.5 m deep). Small seines and hand nets were the most effective equipment for sampling in areas with dense aquatic vegetation. In instances where vegetation made it difficult to



effectively puff the seine forward, a seine was set stationary, and one or more "kickers" drove fishes from the vegetation into the net.

Nylon bag seines (6x2 m, with 2x2 rn bag, 6.4 mm mesh) or large beach seines (10x2 m, 6.4 mm mesh) were used in more open shallow water habitats. These nets sampled for a broad range of fish sizes, but because few areas had clean substrates, large seines were usually limited to open beaches along rivers or streams, and lagoons, where there was little submerged debris.

Nylon and monofilament gill nets with float lines were used to sample selectively for medium and large-sized fishes, including adult and subadult piranhas. Gin net sizes were 20x2 rn (20 cm stretch mesh), lOx5 m. (5 cm mesh), 50x2 rn (with two 25-m long panels of 5 and 7 cm mesh, respectively), and lOOx2 m. (6 cm mesh). For taking piranhas, gill nets were set in areas with little or no current, used in conjunction with seines, and in deeper water (>1.5 m) where use of seines was impractical. Set gill nets were monitored and all fish removed every 0.5 to 6 hours, depending on tliine of day and local fish abundance. Because piranhas can damage gill nets, nets set in areas with high piranha densities were checked more frequently.

All piranha species occurring in the Orinoco River basin take baited hooks (personal observation). Hook-and-line fishing was done mostly with natural bait, using hand lines and occasionally rod and reel. Different sizes and types of commercial hooks were used to sample for fishes of different sizes. Type of bait employed was noted so as to not be mistakenly recorded later as part of their natural diet. In general, sampling by hook and line yielded the same piranha species, and proportions of individuals, as those caught by gill net.

Overall, piranha species that were common in any one area were taken by two or more types of sampling gear. No one capture technique seemed to be selective for a particular piranha species, or group of species. Seines, gill nets, cast nets, and hook and


line were all occasionally effective in capturing species considered uncommon in a particular drainage or region.

To reduce the risk of missing uncommon or rare species in upper Orinoco

drainages, where most sampling was limited to relatively short periods, I supplemented my own samples with collections of others. In the field, I recorded the catch of hired fishermen and Indians, including the daily hook-and-line collections of camp workers.

Field Site Locations and Sampling Periods

Nine drainages of the Orinoco River basin in Venezuela were sampled during the present study, three in the low Llanos and six in the upper Orinoco (Fig. 3-1).

Low Llanos or Savanna Sites

Three major drainages were studied in the low Llanos of Apure State: the Apure, Cinaruco, and Capanaparo rivers (Fig. 3-2). Principal habitats sampled were rivers, streams, natural and artificial lagoons and pools, and flooded savanna. The Apure drainage was the only site where long-term research on piranhas was carried out.

Apure Drainage. Almost all fieldwork was done in the Caio Caicara watershed of the Apure drainage in and around the Fernando Corrales research station and ranch of the Universidad de Los Llanos (UNELLEZ) (07'25'50"N, 69'35'30"W). This area is bordered by Carlo Caicara, its tributary Calo Maporal, and a smaller stream, Caflo Guaritico. I made periodic samples in the Caio Caicara study area during December 1983; January, March, May, July, September, and November 1984; and March, April, and May 1985; these were followed by samples taken during August 1988, March, April, and December 1989, and January, March, and May 1990. I also examined material taken


0 0 0
C\j 0 LO

o (D

C\j C\j
m Q

(D 11-1 (0 (D 00

cr 0 ca
0 Cd
0 cr z on 04
(D Lli 0
(D E =
CL ca

U) 04 LO
0 0 0
co co
W z (D

V) 0 >
0 r
o LLJ C\j 2
0 -1 0
N (\j (D


0 CC) C"i





cn cn

0 CA



cn cn
0 0

0- 0 ..

0 01



between 1979 and 1984, which included monthly samples of a one-year biomass study of the area in 1982-1983 (see Taphorn and Lilyestrom 1984). A total of about two hundred fish collections were made in the Apure River drainage in carlos Caicara and Maporal and their adjacent floodplains from 1979 to 1990.

Cinaruco and Capanaparo drainages. I examined collections made by others from the lower Cinaruco and Capanaparo rivers and their floodplains. Most sites sampled were within 50 km of the north-to-south road joining San Fernando de Apure with Puerto Paez, a route roughly following 67030' W longitude. Samples were taken in January 1982, December 1986, January 1987, and March and April 1989. The 1989 collections were the result of a fish inventory of the two drainages made by D. C. Taphom and A. Barbarino using seines, cast nets, and hook and line. Overall, sites collected were the main channels and near-shore areas of the two rivers, as well as Laguna Larga of the Cinaruco River, and Caflo La Pica and Laguna Brava of the Capanaparo drainage.

Upper Orinoco or Forest Sites

Six major drainages in the upper Orinoco, Amazon Territory of Venezuela, were sampled for analysis of piranha populations from forested regions (Fig. 3-3). These included the Sipapo, Atabapo, Ventuari, Matacuni, Ocamo, and Mavaca drainages, as well as the main channel of the Orinoco River. I collected fishes and evaluated habitats during five expeditions, between early 1989 and 1991, as part of a resource inventory of southern Venezuela in association with the mineral consortium CVG-TECMIN (Corporaci6n Venezolana Guyana Tecnica Minera). Data on serrasalmine fishes from the region are based on intensive sampling of many sites over relatively short periods of time. Each expedition lasted between 25 and 40 field days and covered a different drainage or set of


60- + +/




4*- +


0 100

670 6 6" 650 640

Fig. 3-3. Map of 1988-1991 sampling sites in upper Orinoco, Federal Territory of
Amazonas, Venezuela Solid circles represent localities sampled by the
author; solid squares represent areas collected by others (specimens from all sites deposited at MCNG and examined by the author). Circles and squares
may represent more than one sampling site.


drainages. Work in any one drainage was usually limited to one part of the year, either during the high or low water period.

Lowland habitats sampled in the upper Orinoco were channel borders and beaches of rivers, streams, side channels, old meanders and other back waters, including isolated floodplain lakes, as well as inundated forest. Elevations ranged from 90 to about 240 m above sea level. Collection sites and sampling dates were as follows: (1) Guayapo and Sipapo rivers, from 12 May to 7 June, 1989; (2) lower Ventuari River, including its tributaries the Yureba, Pari, Asisa, Guapuchi ( = Picure), Marueta rivers and the carlos CucuriW, Moriche and Tabi-Tabi, from 17 September to 16 October, 1989; (3) Atabapo River, including its tributaries the Atacavi and Temi rivers and the caflos Patacame, Cuchakrn, Bocachico, and Chimita, from 23 October to 17 November, 1989; (4) Ocamo, Putaco, Padamo, and Matacuni rivers, and the carlos Jayuwapuei, Jenita, Maveti, and Yaraca-bawei, including the upper Orinoco River main channel above its confluence with the Ventuari, from 16 January to 16 February, 1990; and (5) Mavaca River, its tributary Cario Mavaquita ( = Caflo Hauyapiwei), and the upper Orinoco main channel, from 18 January to 20 February, 1991. A total of 200 fish collections representing 180 individual sites was made in this region, resulting in over 18,000 specimens, of which 500 were serrasalmine fishes.

Evaluation of Habitats

A variety of habitat parameters was recorded whenever possible for most collection sites. Qualitative and quantitative estimates were as follows: (1) season (wet, dry, or transitional); (2) water type (clear, black, white, or combination); (3) water level (isolated pools, continuous in bank, bank full, or flood condition); (4) watershed topography (level, moderately rolling, rolling, mountains); (5) stream gradient (flat, slight, moderate, or >30 degrees); (6) land use (natural, cattle, agricultural, or developed); (7) major terrestrial


habitat (grassland, savanna, gallery forest, or forest); (8) dominant riparian vegetation (none, grasses/forbes, shrubs/trees, narrow forest border, or forest); (9) water depths at site and capture depths (very shallow [0-5 cm], shallow [5-20 cm], moderate [20-50 cm], deep [50-100 cm], very deep [ 100-150 cm], extremely deep [>150 cm]); (10) water velocities, as measured with a mechanical flowmeter (General Oceanics) or using a floating stick or cork, were classified into six categories (zero [<0.02 m/sec], very slow [0.02-0.10 m/sec], slow [0.10-0.25 m/sec], moderate [0.25-0.75 m/sec], fast [0.75-1.25 m/sec], and torrential [>1.25 m/sec]); (11) water temperature; (12) water pH measured with a Hach pH Pocket Pal meter or a Hach colorimetric (color disc) wide range pH test kit (Model 17-N);

(13) water color and turbidity (very-clear, clear-tannin, clear-green, slightly-turbid, moderately turbid, or very turbid); (14) water clarity as measured with a 20-cm diameter Secchi disk; (15) estimated maximum and average water width; (16) degree of shading (full sun, partial shade, temporary full shade, permanent full shade).

Substrate types and types of aquatic or instream cover for each collecting site were recorded by estimating their proportion of the total surface area: (0)=none/few; (1)<10%;

(2)10-50%; and (3)>50%. Substrate types recognized were: silt/mud; sand; gravel; rubble; boulder; bedrock/claypan; vegetation; litter; and detritus/decomposed organics. Fish cover categories recognized were: floating aquatic vegetation; emergent vegetation; submergent vegetation; litter on bottom; overhanging vegetation; undercuts; bank cover; brush/roots; rocks; logs; and artificial structures (e.g., culverts, bridge piers).

Identification and Nomenclature of Piranhas

At least 12 species of piranha occur in the Orinoco River basin (personal

observation). As with most groups of South American freshwater fishes, there has been a great deal of confusion surrounding the classification and nomenclature of the subfamily Serrasalminae, due to a combination of factors. Some species exhibit marked changes in


body shape with growth. Individuals are also fairly variable in terms of body pigmentation and color patterns; such differences can be associated with age, size, or sexual condition, and may also be environmentally induced. To further complicate piranha taxonomy, the complete geographic distribution of most species is unknown. Moreover, because little if any comparative material was available to previous taxonomists, most earlier descriptions were incomplete, and thus inadequate for making accurate identifications. Fink (1988) reported that fewer than half of the about 60 nominal species of piranhas were valid. Nevertheless, sampling in previously uncollected regions has resulted in the discovery of a few new species (personal observation). Except for difficulties in identifying some of the smallest specimens (less than 25 mm. SL), I believe I have successfully sorted out the different species encountered in this study. Preserved specimens examined are deposited in the Museo de Ciencias Naturales (MCNG), UNELLEZ, Guanare, Venezuela. Voucher material has also been donated by MCNG under an official cooperative agreement to the Florida Museum of Natural History (UF), Gainesville, Florida. Institutional acronyms are from Leviton et al. (1985).

Because of the taxonomic confusion, piranhas have often been misidentified in the literature. Drs. William Fink (UMMZ) and Antonio Machado-Allison (MBUCV) are currently revising the piranhas of Venezuela; I have relied on both for identifications of certain voucher specimens. They have also made available their unpublished key to Venezuelan piranhas. Taphomn (1990) prepared keys to the characiform fishes of the Apure drainage, and also listed probable generic and specific synonyms of serrasalmrine fishes. His nomenclature for piranhas follows that of Fink and Machado-Allison. Abbreviations used for piranhas and other serrasalmine fishes in illustrations and tables are given in Table 3-1. In the following section species names used in the present report are listed together with names used by me and colleagues for the same species in recent publications:


(1) Serrasalmus altuvei Ramirez 1965: Serrasalmus altuvei in Nico and Taphorn (1988); color illustrations in Nico and Taphorn (1986:33), photos reproduced in Schulte (1988:119).

(2) Serrasalmus cf. eigenmanni Norman 1928: Serrasalmus cf eigenmanni may involve a species complex (W. Fink, personal communication): however, I distinguished only one species in my samples. This species, to my knowledge, has not been previously referenced in publications on Venezuelan piranhas. The piranha Serrasalmus nalseni described and figured by Fernmndez-Y6pez (1969) is a possible synonym, but I have not seen the type series. All previous references to Serrasalmus eigenmanni from the Orinoco probably refer to other species (see Serrasalmus irritans below).

(3) Serrasalmus elongatus Kner 1860: Serrasalmus cf. elongatus in Taphorn and Lilyestrom (1984), Nico and Taphomrn (1988); color illustrations in Nico and Taphomrn (1986), photos reproduced in Schulte (1988:1, 119); photos as S. pingke in Roman (1983:56).

(4) Serrasalmus irritans Peters 1877: S. eigenmanni in Taphomrn and Lilyestrom (1983); S. irritans in Nico and Taphomrn (1988) and Winemiller (1989a, 1989b); color illustrations as Serrasalmus eigenmanni in Nico and Taphorn (1986:31, 33), same photos reproduced in Schulte (1988:120), photos as S. eigenmanni in Rom,.n (1983:57).

(5) Serrasalmus manueli Femndez-Y6pez and Ramnfrez 1967: This species has rarely been referenced in the literature since its original description. I am unaware of misidentifications. There is no published information on its biology or natural history.

(6) Serrasalmus medinai Ramfrez 1965: Serrasalmus caribe in Nico and Taphomrn (1988); incorrectly spelled as S. medini in Winemiller (1989a, 1989b); color illustrations as Pristobrycon sp. in Nico and Taphorn (1986:31, 33, 41), photos reproduced as Serrasalmus (Pristobrycon) sp. in Schulte (1988:119), and as Serrasalmus (Pristobrycon) "iridescent" (incorrect locality given) in Schulte (1988:128); photos as S. rhombeus in Romdn (1983:31,60).


Table 3-1. List of abbreviations used in figures and tables for piranhas and other serrasah-nine fishes.

Piranhas: Other Serrasahnine Fishes:

PC Pygocenftw caribe Cm Catoprion mento
Psp Pristobrycon sp. Mag Metynnis argenteus
Pst Pristobrycon sololatus Mya Myleus asterias
Pyp Pygoprislis denticulatus Mys Myleus schomburgkii
Sa Serrasalmus altuvei Myt Myleus torquatus
Seg Serrasalmus cf eigenmanni Md Mylossoma duriventfis
Sel Serrasalmus elongatus
Sir Serrasalmus irfitans
Sm Serrasalmus manueli
Smd Serrasalmus medinai
Sr Serrasalmus rhombus


(7) Serrasalmus rhombeus (Linnaeus) 1766: Serrasalmus rhombeus in Nico and

Taphom (1988) and Winemiller (1989a, 1989b); and color illustration in Nico and Taphorn (1986:40-41).
(8) Pygocentrus caribe Valenciennes 1849: Pygocentrus notatus in Taphorn and Lilyestrom (1984), Nico and Taphorn (1988), Winemiller (1989a, 1989b), Nico (1990); color illustrations are given as P. notatus in Nico and Taphomrn (1986:31, 40, 41), photos reproduced as Serrasalmus (Pygocentrus) notatus in Schulte (1988:109, 110), as Serrasalmus niger in Schulte (1988: 111), and as Serrasalmus nattereri (incorrect locality given) in Schulte (1988:12-13).
(9) Pristobrycon striolatus (Steindachner) 1908: Pristobrycon striolatus in Nico and Taphom (1988); color illustrations as Pristobrycon striatus in Nico and Taphorn (1986:31, 41), and as Pygocentrus striolatus in Nico and Taphorn (1986:33), photo reproduced in Schulte (1988:120). Machado-Allison et al. (1989) concluded that Pygopristis antoni Ferndndez-Y6pez, 1965 is a junior synonym of P. striolatus.

(10) Pristobrycon sp.: This species is in the process of being described as new by W. Fink and A. Machado-Allison.

(11) Pygopristis denticulatus Muller and Troschel 1844: photograph as Serrasalmus (Pygopristis) antoni in Roman (1983:51).

Faunal Comparisons

The Coefficient of Biogeographic Resemblance (CBR) proposed by Duellman (1990) was used to compare the composition of piranha assemblages between the two regions studied and among the different drainages. This qualitative index of similarity has been reinvented a number of times, consequently it is known by several names including the Faunal Resemblance Factor, the Sorenson Index, and the Dice Coefficient (Boesch 1977, Romesburg 1984, Magurran 1988, Duellman 1990). This index is one of the more


commonly used measures for comparing species presence-absence data (Krebs 1989), and is calculated as:

CBR = 2C/ (Ni + N2),

where C is the number of species common to two areas, NI is the number of species in the first area, and N2 is the number of species in the second area. The resulting values range from 0, if there are no species in common, to 1, if both areas share the same number and kinds of species. This measure is simple, is not affected by relative numbers of individuals, and thus is best used when there are no reliable quantitative data on abundance of species (Magurran 1988, Krebs 1989). Binary similarity measures are not totally independent of sample size (Wolda 198 1, Krebs 1989). However, comparisons when maximum number of possible species is low give results less affected by sample size differences than those dealing with many species (see Krebs 1989). CIBR is very similar to the Jaccard Coefficient (Boesch 1977, Romesburg 1984, Duellman 1990); it doubly weights shared positive attributes (i.e., joint presences), and is preferred over the Jaccard Coefficient if the entities vary widely in their number of positive attributes (Boesch 1977:22).

To make presence-absence data more reliable, I supplemented my data with

information from collections deposited at MCNG and UF. I also reviewed the literature for records of species occurrence. A Spearman rank correlation test was used to test for an association between species composition and drainage area size, and between faunal similarities and distances among drainages. The similarity or resemblance matrix generated from the CBR equation was used as a basis for clustering drainages, employing the unweighted pair-group method using mathematical averages, or UPGMA.


In order to analyze diets and document ontogenetic shifts in feeding, sampled fishes were divided into five size classes: Group 1) 10-19 mmn SL small juveniles; Group 11) 2039 mm SL; Group HII) 40-79 mmn SL; Group IV) 80-159 num SL; and Group V) 160 mm SL. Several species were not represented by all size classes.

During the study, more than 2,000 serrasalmine fishes were captured; total number of specimens examined for stomach contents was 1,538 from the low Lianos and 393 from the upper Orinoco. Standard length (SL) of each specimen was measured to the nearest mm. Although a few of the largest fish were dissected and examined in the field, most specimens were preserved after capture and dissected later in the laboratory to determine stomach contents. I examined stomach contents and recorded frequency of occurrence

(0), number (N), and dominance by bulk (D) for each food item. I also recorded volume

(V) of food items found in stomachs of fishes taken in the field after 1986. Small food organisms were counted using a Ward counting wheel under 25X magnification. Volume was determined by water displacement using appropriately- sized graduated cylinders. Volumes of smaller items were calculated by measuring dimensions with Vernier calipers or by examidnation under a dissecting microscope and comparison with known volumes. An estimate of relative stomach fullness was made using scores ranging from 0 (empty) to

3 (if full, or almost full) (Thomerson and Wooldridge 1970, Nico and Taphomn 1988).

For specimens collected in the Apure River drainage before 1986, I estimated

volume using a point system, referred to as adjusted volume (Va), and derived from D and stomach fullness (Nico and Taphorn 1988). That is, only the dominant food categories of a given stomach were included and each was assigned the fullness points (1-3) for that stomach. If two food items in a single stomach were of about equal volume, each received half the fullness value. Thus, %Va = total fullness points for each food category as a percentage of all fullness points, and can be shown as:


%10Va Y, Fj /YTi,) x 100,
n n
where Fi is the fullness points for stomachs whose dominant food item is F; Ti is total fullness points for all stomachs; and n is the number of stomachs examined of a species or size class.

I attempted to identify food items to the lowest taxonomnic level, whenever possible. For comparing piranha diets, I recognized eleven general food categories: (1) masticated seeds crushed or fragmented seeds together with the hard parts (mainly endocarp)that surround the seed; (2) other plant material mostly leaves, root fragments, fleshy fruits, and flowers; (3) decapods crabs and shrimp; (4) microcrustaceans copepods, cladocerans, ostracods and conchostracans; (5) aquatic insects various families, including larvae and adults; (6) other invertebrates mostly arthropods, including terrestrial forms, but also aquatic nematodes and snails; (7) small whole fish at least two-thirds of total body; (8) fish flesh pieces of flesh bitten from fish; (9) fish fins fins not taken with or attached to pieces of flesh; (10) fish scales scales not taken with or attached to flesh; and

(11) other detritus, sand, nonfish vertebrates, and unidentifiable items.

I compared and contrasted diets within and among the different species and size groups by calculating estimates of diet breadth and overlap based on percent food volume or adjusted volume. Because I did not quantitatively measure the relative availability of food resources in the field, I chose two widely used indices of breadth and overlap that do not take into account the relative abundance of the various food items in the natural environment.

Diet breadth, or width, measures the diversity and evenness of resource use. Diet breadth was computed for each species' size class using Levins' index of niche width (Levins 1968, Krebs 1989):


Diet breadth (B) = 1/ jp2

where Pi is the proportion of food category i in the diet. B ranges from 1, when one resource state is used exclusively, to n (the total number of food categories recognized), when all resource states are used in equal proportion. This measure is the reciprocal of Simpson's index of diversity (Krebs 1989).

Ii used the Schoener Overlap Index (Schoener 1970) to estimate the similarity in diets between and within species. This particular index is also known as the Renkonen Index, the Percent or Proportional Similarity Index, Percentage Overlap, and Czekanowski's Index (Krebs 1989); the Bray-Curtis Coefficient gives the same results as the above index if its scores are standardized (Boesch 1977:26, Bloom. 198 1). The Schoener Overlap Index is calculated as:

Overlap = 1 0.5( 1 x pi I (minimum pxi, pyi),

where pxi is the proportion of food category i in the diet of species x, pyi is the proportion of food category i in the diet of species y, and n is the number of food categories. The Schoener Index gives values from 0 to 1 (or 0 to 100%) indicating no overlap to complete similarity in the proportion of food categories used by the two groups compared.

There are many indices available for estimating niche or diet width and overlap, but as yet there is no general consensus as to which of the many indices is best (Huribert 1978, Abrams 1980, Wallace 1981). Recent reviews of frequently used indices include Ludwig and Reynolds (1988) and Krebs (1989). All measures of breadth and overlap are biased in some way, the bias usually being reduced with increasing sample size (Schluter 1988, Krebs 1989). It has also been shown that measures of breadth and overlap are sensitive to the number of resource categories used. Although lumping resource states can inflate niche


overlap values (Greene and Jaksic 1983), bias increases as the number of resource categories used increases or is large (Ricklefs and Lau 1980, Smith and Zaret 1982, Krebs 1989). Overlap and breadth indices also give somewhat different results depending on the methods used to tabulate dietary components, for example, percentages based on total food volume, weight, frequency of occurrence, or food item number (Wallace 198 1). An additional problem implicit in all overlap and breadth measurements is that the investigator may recognize resource categories that animals do not. For instance, Goulding et al. (1988:52) suggested that Arnazonian frugivorous fishes "select their preferred fruits and seeds at what taxonomically are referred to as genera" rather than at the species level. Conversely, organisms may distinguish resources lumped together by human observers (Ludwig and Reynolds 1988, Krebs 1989). Finally, most overlap and width indices are based on the relative use of different resources, without accounting for their relative availabilities. However, resource availability is difficult to assess and therefore rarely measured. Those indices that do incorporate some estimate of food availability into their equations are generally less simple to interpret and more variable between sites (Schluter 1988). Abrams (1980) and Wallace (198 1) recommended the Schoener Index as the best of the measures of niche overlap when available resources to the organism have not been quantified. Krebs (1989:304-306), referencing Wolda (1981), regarded this measure as one of the better quantitative similarity coefficients available, showing that it is relatively little affected by sample size and number of species (i.e., resources states).

Because percentage of occurrence and percentage of total number of food items tend to over-emphasize the importance of small food items, I used proportions based on total food volume for calculating diet widths and overlaps. Using food volume introduces the possible bias that a large prey item in a single large piranha might overshadow all other items when pooling stomach contents data. However, this problem is overcome to a large extent by dividing fish into different size classes and treating each size class as a separate "trophic unit" (sensu Stoner and Livingston 1984).


In my analysis of diets, the similarity or resemblance matrix generated from the Schoener Index was used as a basis to cluster species using the unweighted pair-group method using mathematical averages, or UPGMA. I used 2 x 2 contingency table X2 tests (one-tailed) and Mann-Whitney U-tests to test the prediction that forest populations fed more on plant material than their savanna counterparts.

Nutrient Content Analysis of Food Items

In order to compare nutrient values, analyses were carried out to determine ash, crude protein, total fat, carbohydrate, and energy contents of different food items. Three types of food items were examined: small whole fish, fish fins, and fish scales. I also searched the literature for information on the nutrient value of other types of general food items, for example insects and various types of plant material. Prey fishes used for analysis were taken from the Apure River drainage, Venezuela, using cast nets and seines during the dry season in December 1989 and March 1990. Tissues from members of two unrelated families of fishes were tested: (1) Curimatidae, represented by Curimatella immaculata, Cyphocharax spilurus, and Steindachnerina argentea; and (2) Cichlidae, Aequidens pulcher and Cichlasoma orinocense. These species are common in the Orinoco River basin and are frequently preyed upon by juvenile and adult piranhas.

In total, 135 small cichlids (SL range 39-77 mm) and 96 curimatids (SL range 4578 mm) were used for nutrient content determinations. Fish were placed on ice immediately after capture, transported to the laboratory at the Universidad de Los Llanos Occidentales (UNELLEZ) in Guanare, Venezuela, and refrigerated until analysis was conducted. Body scales were removed from partially frozen specimens with a small fish scaler, and no attempt was made to clean off the body's outer mucous layer. Only caudal and dorsal fins were used for fin samples; these were removed by clipping with scissors above the fin base to avoid scales. Two or three entire cichlid and curimatid specimens


were used for tests as small whole fish. All subsequent nutrient analyses were carried out at UNELLEZ under the direction of the biochemist Dr. Margioly de Morales. Samples were oven-dried to constant weight at 60 T to prevent lipids in the body tissues from volatilizing. After drying, samples were ground in a Willey mill and stored in desiccators. Unless otherwise stated, methods of analysis used follow AOAC (1980) guidelines. Results are expressed as percentages of dry weight, rather than as percentages of ash-free dry weight (i.e., on organic matter basis), for ease of comparison to previously published data. Ali measurements were duplicated; the differences between replicate subsamples of a particular food item were generally less than 1%. Ash contents were determined by combustion of samples of dried body tissue at 550 T for 3 hours (Pierce et al. 1980). Total fat content was determined by extracting lipids for four hours with diethyl ether. Crude protein was determined by the Kjeldahl method for nitrogen, where crude protein is total nitrogen concentration multiplied by 6.25. Carbohydrates were deten-nined as the sum of crude fiber and nitrogen-free extract; total carbohydrates are the sum of both values. Energy content was measured using a Parr bomb calorimeter and expressed as 0 per g; this can be converted to kilocalories per g by dividing by 4.184.

Intestine Length

Length of the intestinal tract (IL) was determined by removing the entire

gastrointestinal tract, placing the extended intestine (without stretching) on a flat surface, and measuring the distance from the pylorus to the anus (Ribble and Smith 1983). Fecal material was left in the intestine during measurement. Relative intestine length for each fish was calculated by dividing IL by standard length (SQ. An analysis of covariance (ANCOVA) was carried out in order to compare the regression lines of IL versus SL among species. A Spearman rank correlation test was used to test for an association between relative intestine length and the percent volume of plant material among the


different species. My data indicated allometric changes in intestine length. In order to reduce variation and make comparisons more valid, only specimens > 80 mm SL were used in correlation analysis.

Field and Aquarium Observations

I made qualitative and quantitative observations on the feeding behaviors of wildcaught juvenile and subadult piranhas maintained live in aquaria or large outdoor concrete tanks (L. Nico, unpublished data). Species represented from low Llanos populations were Pygocentrus caribe, Serrasalmus elongatus, S. irritans, S. medinai, S. rhombeus, and Pristobrycon striolatus; those from upper Orinoco sites were Pygopristis denticulatus and Serrasalmus manueli. In the field, most observations of piranha feeding behavior were done from shore or from boats while fishing, these qualitative observations totaled over one hundred hours. I also made brief underwater observations on the feeding behavior of several species totaling about five hours. Underwater observations were made using mask and snorkel in clearwater and blackwater streams in the states of Bolivar, Guarico, and Monagas. Species observed were juveniles or subadults of Serrasalmus irritans, S. medinai, S. rhombeus, and Pygocentrus caribe.

Statistical Analyses

All nonparametric tests were run using the computer software program Basic

Statistical Subroutines by DYNACOMP (1983) following Siegel (1956) and Siegel and Castellan (1988). Statistical tests of ANCOVA were done using the SuperANOVA program on a Macintosh SE/30 computer. For cluster analysis, I used the computer program NTSYS-pc (Numerical Taxonomy and Multivariate Analysis System) version 1.5 by Rohlf (1988). Cluster analysis has been shown to be a useful and reliable multivariate


procedure for identifying feeding patterns in assemblages of fishes (Stoner and Livingston1984, Henderson and Walker 1986) and other vertebrates (Jaksfc and Medel 1990). The method has also been used extensively to describe fish distribution patterns (Hocutt and Wiley 1986). Basically, the UPGMA clustering technique generates a hierarchical tree, or dendrogram, grouping together those samples or objects that are most similar. UPGMA is widely applied because it can be used with a variety of resemblance coefficients and also because it forms clusters in a more conservative manner than many other clustering methods (Gauch 1982:199, Romesburg 1984, James and McCulloch 1990:147). As a measure of secondary validity, an additional clustering technique, complete linkage (farthest neighbor), was used to assess whether resulting clusters were more-or-less real rather than mathematical artifacts (Gauch 1982, Romesburg 1984).

COMPOSITION OF PIRANHA ASSEMBLAGES In this chapter I compare the species composition of piranha assemblages from the low Llanos to those from the upper Orinoco. I also examine and compare species composition and richness among the nine drainages sampled, and present possible habitat correlates. Little has been published on the distribution of Orinoco River fishes, with the upper Orinoco being that part of the basin least explored ichthyologically. Because sampling effort was not equal among all the drainages, I supplementedmy data with information from museum collections and published records. Although additional collecting will likely add a few species to faunal lists for several of the drainages, the data presented represent a first attempt at understanding piranha distribution patterns and habitat preferences.

Figure 4-1 is a schematic representation showing the relative distances among drainages in relation to their confluence with the main channel of the Orinoco River. Table 4-1 is based on my samples supplemented by museum material, and shows the species composition by region and drainage. A total of eleven piranha species were taken from the two regions studied. Since 1979, ten species have been collected from the low Llanos of Apure State, whereas my 1989-1991 samples indicate that at least eight species occur in the upper Orinoco of the Amazon Territory of Venezuela. Seven species were widely distributed and inhabited both regions. The Coefficient of Biogeographic Resemblance (CBR) value comparing piranha species composition between the low Llanos and the upper Orinoco was 0.78.

Piranha assemblages in five of the six upper Orinoco drainages were numerically dominated by either S. rhombus or S. manueli, or both. The exception was the Mavaca



Ocamo Mavaca
Matacuni 1000 km

Ventuari AtabapoQ Sipapo -500 km

Cinaruco Capanaparo Apure

Fig. 4-1. Schematic representation showing relative distances among the nine sampled
drainages and their approximate location along Orinoco River main channel.
Arrow indicates division between upper and lower Orinoco.




cn cn



00 &OD
a4 04 CIA

a4 u u 04

a4 a., CLO a4 0-( a,

U U a4 U 0.4 CL4

iz "I "ts ;t

4 lo vi

to qj
co C11 C-n


River, where Serrasalmus cf. eigenmanni was the most common piranha collected. Pygocentrus caribe is the predominant piranha throughout most of the Llanos, and was abundant in samples from the Apure River drainage although absent or uncommon in the Cinaruco and Capanaparo. Pygocentrus caribe was rare in the upper Orinoco. Serrasalmus irritans and S. medinai were locally common in the low Llanos, but neither was taken in the upper Orinoco. Serrasalmus manueli, although absent from much of the Llanos, was commonly taken along with S. rhombeus in the Cinaruco River. Pristobrycon striolatus, although widespread, was never common in any one drainage.

Species occurred in one to seven of the nine drainages sampled (Table 4-1); most were found in several (mean = 4.1) drainages, with Serrasalmus rhombeus and Pristobrycon striolatus (taken in seven drainages) being the most widespread. Several species were absent from two middle drainages, the Sipapo and Atabapo rivers. Two species were limited to single drainages. In the Orinoco River basin, Pristobrycon sp. is known only from the Atabapo drainage. Serrasalmus medinai was represented only in Apure River samples (although it is found in other Orinoco tributaries that fringe the Llanos, drainages not included in this study).

The number of piranha species per drainage ranged from 1 to 8 (mean = 5.0) (Table 4-1). Sipapo drainage samples yielded only one piranha species (Serrasalmus manueh), whereas all other drainages were inhabited by three or more species. The Apure had the highest number of species (eight), but it was also the largest and best sampled drainage. There was a significant correlation between numbers of piranha species and size of drainage area (Spearman rank correlation, rs = 0.67, one-tailed P < .05, n = 9; Table 42, Fig. 4-2).

Species composition of piranha assemblages across drainages was highly variable; CBR values ranged from 0 (i.e., no species in common) to 0.83 (Table 4-2). Adjacent drainages often shared many of the same species (Table 4-3), but waterways that had many species in common (i.e., high CBR) were not always those drainages that were


geographically close (Fig. 4-3, Table 4-2). For instance, the Mavaca and Cinaruco were somewhat similar in terms of species composition (CBR = 0.73), yet the two drainages are far apart geographically. Similarly, the Apure drainage had four species in common with the Ocamo, even though these two rivers are separated by roughly 1000 km of waterway (Fig. 4-1). As expected, there was a significant negative correlation between species composition resemblance (i.e., CBR) and the distance between adjacent drainages (rs =

- 0.66, one-tailed P <.05, n = 8; Table 4-2, Fig. 4-3). However, the relationship was not significant at the 0.05 level when plotting all possible combinations of drainage-pair distances against their CBR values (rs = 0.23, one-tailed, .10 > P > .05, n = 36; Table 42, Fig. 4-4).

Figure 4-5 summarizes results of a UPGMA cluster analysis showing affinities

among the nine drainages based on CBR similarity values given in Table 4-3. In addition to similarity in species composition, most resulting groups of drainages were related by either geographic proximity, characteristics of the drainage (e.g., water type), or some combination of these two factors. The first major cluster consisted of the three savanna drainages (Capanaparo, Cinaruco, and Apure) and one upper Orinoco tributary (Ventuari). The Capanaparo and Cinaruco are adjacent drainages and both were classified as blackwater, but each receives high input from clearwater streams. The Apure River is a whitewater river with many small blackwater and clearwater tributaries. The Ventuari also is mainly a whitewater river although it has several large blackwater and clearwater tributaries. The second major cluster was formed by three drainages in the upper Orinoco. This cluster consisted of the Matacuni, Ocamo, and Mavaca; all are whitewater rivers and geographically close. The remaining two drainages, the Sipapo and Atabapo, did not cluster with any of the other drainages. They were blackwater rivers with relatively few piranha species. Clustering based on the farthest neighbor strategy resulted in drainage groupings identical to that of the UPGMA method.


Table 4-2. Distance between drainages and Coefficient of Biogeographic Resemblance (in parenthesis) among nine sampled drainages in the Orinoco River basin. Distances measured in kilometers between drainage mouths by way of Orinoco River main channel. CBR values range from 0 (no species in common) to 1 (complete similarity in species composition)

River Drainages

Low Llanos Upper Orinoco

Apu Cap Cin Sip Ata Ven Mat Oca Mav

Drainage area
(km2 x 103) 167 35 25 13 9.4 40.5 10.1 9.8 5.3

Number of 8 7 6 1 4 6 3 5 5
piranha species

Apure 110 160 400 530 610 945 965 1000
(0.80) (0.57) (0) (0.33) (0.57) (0.36) (0.62) (0.62) Capanaparo 50 290 420 500 835 855 890
(0.77) (0.25) (0.55) (0.77) (0.20) (0.50) (0.50) Cinaruco 240 370 450 785 805 840
(0.29) (0.60) (0.83) (0.44) (0.55) (0.73) Sipapo 130 210 545 565 600
(0.40) (0.29) (0) (0) (0) Atabapo 80 415 435 470
(0.60) (0) (0.22) (0.44)

Ventuari 335 355 590
(0.44) (0.73) (0.73)

Matacuni 20 55
(0.75) (0.75)

Ocamo 35




g Cap

, 6 Cin oVen
S0 ooca
o May
. 4 OAta
2- = Low Llanos
0o O = Upper Orinoco
0 .......i.. I I ...... I 1 1
3.0 3.5 4.0 4.5 5.0 5.5 6.0

Log drainage area (km 2)

Fig. 4-2. Scatter diagram showing relationship between number of piranha species per
drainage and log of the drainage area size (km2) for nine sampled drainages in
Orinoco River basin, Venezuela. Drainage abbreviations are: Apu = Apure, Ata = Atabapo, Cap = Capanaparo, Cin = Cinaruco, Mat = Matacuni, May =
Mavaca, Oca = Ocamo, and Sip = Sipapo.


Table 4-3. Number of shared piranha species and Coefficient of Biogeographic Resemblance (CBR) (in parenthesis) among nine sampled drainages in the Orinoco River basin. CBR values range from 0 (no species in common) to 1 (complete similarity in species composition).

River Drainages

Low Llanos Upper Orinoco

Apu Cap Cin Sip Ata Ven Mat Oca Mav Number of
piranha species 8 7 6 1 4 6 3 5 5

Apure 6 4 0 1 4 2 4 4
(0.80) (0.57) (0) (0.33) (0.57) (0.36) (0.62) (0.62) Capanaparo 5 1 2 5 1 3 3
(0.77) (0.25) (0.55) (0.77) (0.20) (0.50) (0.50) Cinaruco 1 2 5 2 3 4
(0.29) (0.60) (0.83) (0.44) (0.55) (0.73) Sipapo 1 1 0 0 0
(0.40) (0.29) (0) (0) (0) Atabapo 2 0 0 1
(0.60) (0) (0.22) (0.44)

Ventuari 2 4 4
(0.44) (0.73) (0.73)

Matacuni 3 3
(0.75) (0.75)

Ocarno 4




0.8 00 0
E 0.6 0

4- 0.4 00
C- 0

0.0 ,
0 100 200 300 400

Distance between adjacent drainages (kin)

Fig. 4-3. Scatter diagram showing similarity in piranha species composition between
adjacent sampled drainages versus distance between adjacent drainages.
Similarity measure was Duellman's (1990) Coefficient of Biogeographic



0 0.8E *

U 0.60
( ~0.62 0 .0
0 .

vausaditnesaefo 0.0l 4- 2 .


0 0.5 1.0
I Water Types Dominant








0 0.5 1.0
Coefficient of Biogeographic Resemblance

Fig. 4-5. Dendrogram from cluster analysis depicting similarities among nine sampled
drainages of Orinoco River basin based on piranha species composition. Major and minor (in parenthesis) water types, W = whitewater, B = blackwater, and C = clearwater; Dominant regional flora, S = savanna (low Llanos), and F = forest
(upper Orinoco). Similarity values are from Table 4-3. Clustering technique was UPGMA, and similarity measure was Duellman's (1990) Coefficient of
Biogeographic Resemblance.


A summary of species occurrence by habitat type is given in Table 4-4. Most species were found in a variety of environmental conditions, but relative abundances differed among sites. Serrasalmus rhombeus and S. manueli were frequently taken in riverine habitats; other piranhas were more common in floodplain habitats (e.g., lagoons or old meanders) and in caflos with slow currents. In those cases where piranhas were found in large streams or rivers, they commonly congregated in adjoining slackwaters near the interface between fast-flowing and still waters, for example near the mouths of sloughs or deep carios.

Most piranhas were found in two or all of the three major water types (Table 4-4), but greatest numbers occurred in whitewater habitats. Three species (i.e., Serrasalmus manueli, Pygopristis denticulatus, and Pristobrycon sp.) were usually associated with lowpH, blackwater habitats. Nevertheless, most species seem to tolerate a wide range of temperatures and pH, as shown by wild-caught juveniles of both S. manueli and P. denticulatus that have survived nine months (to date) in aquarium water with pH as high as

8.5. Water temperatures at sample sites ranged from 24.5 to 38 'C, and pH 4.5 to 7.2; water transparency ranged from very turbid (10 cm as measured with a Secchi disk), to very clear (>3 m).

Where piranhas occur, two or more species are commonly taken together. In the low Llanos, juveniles and subadults of as many as five species (i.e., Pygocentrus caribe, Serrasalmus altuvei, S. irritans, S. medinai, and S. elongatus) were occasionally taken in single seine hauls in floodplain pools of the Apure drainage. In the upper Orinoco, adults of as many as four species of piranha (e.g., Serrasalmus rhombeus, S. cf. eigenmanni, Pygocentrus caribe, and Pristobrycon striolatus) were caught together in gill nets set during evening twilight. Sites of highest species richness were characterized by a large amount of instream cover or other protection, including mats of aquatic vegetation, adjacent flooded forest, or deep water. Small juvenile piranhas (<40 mm SL) were only found in


microhabitats where there was high cover provided by herbaceous plants. In the Llanos, small piranhas relied on water hyacinth or flooded grasses for shelter and as a foraging place. In the upper Orinoco, young juveniles were limited to a few sites where there were mats of flooded grasses (e.g., stream edges) or shallow waters with dense growths of submergent plants, such as beds of the aquatic plant Elodea.


cq C14 C14


tA C14 C14 tn
cq C14

COS tn
cq kn
tm. A


x x





LZ tu C-0




This chapter focuses on the diets of piranhas from the area around the Fernando Corrales ranch and research station of the University of the Western Llanos (UNELLEZ), located in the Caflo Caicara region, Apure River drainage, in the low Llanos (Fig. 5-1). I give this drainage separate treatment because it is the only site where long-term sampling (1979-1990) has been carried out (Taphorn and Lilyestrom 1984, Nico and Taphorn 1984, 1985, 1988, Nico 1990, Taphorn 1990). The existing data base on piranhas from the Calo Caicara region is the most complete of any single locality or drainage in the Orinoco basin. The seven species examined were Pygocentrus caribe, Serrasalmus altuvei, S. medinai, S. elongatus, S. irritans, S. rhombeus, and Pristobrycon striolatus (Fig. 5-2). I also report on the diet of the serrasalmine Catoprion mento, a close relative of the piranhas.

Study Area

The study area is located in the low Llanos of Apure State (07025'50"N,

69035'30"W) about 80 km W of the town of Mantecal (Fig. 5-1). Several permanent streams border the 12,600-ha ranch, catlos Caicara, Maporal, and Guaritico. These lowgradient streams have mostly silty bottoms, and are bordered by narrow gallery forests. The ranch is maintained primarily for grazing of cattle and encompasses open flat savanna with a very gradual sloping terrain west to east. Most of the savanna floodplain of the ranch is surrounded by a 2.5 m high earthen dike constructed to control flooding during the rainy season and hold water during the drier part of the year. During the rains most of the



w GO






> w 4.
0 0


Fig. 5-2. General body form and major fin markings of seven piranha species (55-70 mmn
SL) and Catoprion mento (Cm) from Cafto Caicara study area, Apure Drainage,
in the low Lianos of Venezuela: Pc = Pygocentrus caribe; Pst =Pristobrycon
striolatus; Smd = Serrasalmus medinai; Sr = Serrasalmus rhombeus,
Sa =Serrasalmus altuvei, Sir =Serrasalmus irritans, and Sel = Serrasalmus



PC 0

Smd Sr

Sa 311

Sel cm


savanna becomes flooded, with low areas covered by as much as 2 m of water. Lower areas and deeper borrow pits contain water throughout the year, but most permanent water sites are reduced to less than 1 m depth during the dry season. The fish species composition of borrow pits resembles that of natural lagoon habitats. Large free-floating mats of vegetation, primarily Eichhornia crassipes and E. azurea, are common in streams and savanna pools. During high water, fishes can move between the streams and the interior of the diked areas through culverts or low sections of the dike, but this interchange has been greatly reduced as dike stability has been achieved. During the late dry season (January to early April), the streams have little or no flow and are often reduced to isolated pools. Rainy season flow is continuous with depths to 5 m or more. Width ranges from 3 to 12 m during low water to over 30 m during peak flooding. Although the Apure River itself is white water in character, streams of the study area are essentially blackwater habitats with low transparency. The flooded savanna pools are white water habitats, usually very turbid (< 10 cm Secchi disk), but the larger poolIs often become quite clear (> 1 m Secchi disk) late in the rainy season.

Species Accounts

In this section, I describe the general morphology, local distribution and habitat, and the diets of the piranha species found in the Cafto Caicara study area. Information on the smallest size class of piranhas, juveniles less than 20 mm SL, were pooled; all were tentatively identified as Pygocentrus caribe. Figure 5-3 is a summary of the diets by size class (Group HI and larger) of the seven piranhas and Catoprion mento from the Caflo Caicara area.

> 4 > 4.4




Ln > t C4. cn 0




z e

en e
CL ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ e .........Q

.........c .. .. .
...... ...... ......
.. ....
mu ......... .....
.. . .. .. ... .
.. . . . .. .....

< If





CD... .


N Ci

g0 co ID

C c

j A4


Pygocentrus caribe (Valenciennes 1849)

Pygocenrus caribe was the most abundant piranha in the Carlo Caicara area,

occurring in both flooded savanna and flowing waters. This blunt-headed, heavy-bodied species (Fig. 5-2) reaches approximately 300 mm SL. Stomach contents of 516 P. caribe in the four larger size classes are given in Table 5-1 and Figure 5-3. Unlike most other piranha species, fish fins and scales were of minor importance in stomachs of juveniles. Group II fish fed most heavily on aquatic insects, followed by microcrustaceans, but for Group 1II fish flesh was the most important food item, followed by plant material (primarily vascular plant debris) and aquatic insects. Larger P. caribe of groups IV and V took mostly fish flesh and whole small fish.
Thirty-two species of prey fishes (mostly characins) were identified from stomachs of Groups IV and V P. caribe (Table 5-2). There were also several instances of cannibalism by adults on small juveniles. In addition to fish, larger juveniles and adults sometimes ate other vertebrates, perhaps as carrion. Different stomachs contained a chunk of flesh from a small caiman (Caiman crocodilus), skin of a lizard (probably Ameiva), and a small adult leptodactylid frog. Three individuals taken together had stomachs full of feathers, flesh, and bone fragments of a white-faced whistling duck (Dendrocygna viduata). Mammals were not found in P. caribe from the study area, but I have found flesh and fur of small rodents in specimens from nearby localities.

Serrasalmus irritans Peters 1877

Serrasalmus irritans was very common, usually second only to P. caribe in

numbers of individuals at most sampling sites. My sample consisted of 271 specimens (Table 5-3 and Fig. 5-3). It has a moderately pointed snout (Fig. 5-2) and is seldom longer than 160 mm SL. My study shows that juvenile S. irritans (Groups II and III) specialized on fins of other fishes. Stomachs of the young of this piranha were typically packed exclusively with fins, mostly from fish smaller or similar in size to the predator.


4 oc
C oc
u 06

C-9 oc Al r,

CN -kn m


oc 00 >

> C-i C-i

C,4 cq 00 CN C06 C,4 r,

6 C'i ON \6

6 C'i
"o C-i


oc 00 en


In z z


Table 5-2. List of vertebrate prey taxa identified from the stomach contents of the four most common piranhas from the Cafto Caicara area, Apure River drainage, in the low Llanos of Venezuela.

Species Vertebrate Pre y
(fiNuency of occurrence)

Pygocentrus cailbe Fish (whole or flesh): Characidae Aphyocharax erythrurus (6), Astyanax
bimaculatus (1), Astyanax sp. (3), Charax sp. (2), Cheirodon pulcher (1), Ctenobrycon spiturus (22), Gymnocorymbus thayeri
(15), Hemigrammus ntarginatus (5), Hemigrammus sp. (5), Moenkhausia dichroura (2), Odontostilbe pulcher (7), Poptella orbicularis (6), Pygocentrus caribe (6), and Roeboides affinis (1), Roeboides dayi (8), Roeboides sp. (3), Serrasalmus eigenmanni (1), Serrasalmus sp. (1), Triportheus sp. (1); Curimatidae Curintata Pxtae (10), curitnatids (14); Lebiasinidae Pyrrhulina cf. lugubris (2); characoid (8); Cichlidae Microgeophagus ramirezi (2), Gymnotidae Gymnow carapo 1); Stemopygidae Sternopygus macrurus (1), and Eigentnannia spp. (10); Hypopomidae Hypopornus sp. (1); Auchenipteridae Entoffwcorus garneroi (1) and Parauchenipterus galealus (1); Pimelodidae Pimelodella sp.(I); Caffichthyidae Hoplosternum littorale (1); Rivuhdae (1). Fins of Hypophthalmidae Hypophthalms cf. edentatus (1); Characidae Triportheus sp. (1). Reptiles: Flesh of Caiman crocodiles (1); flesh of lizard, probably Ameiva (1).
Amphibians: Small adult leptodactylid frog. Birds: Feathers and flesh of white-faced whistling duck (Dendrocygna viduata) (3).

Serrasalmus irdtans Fish (whole or flesh): Characidae Ctenobrycon spilurus (1),
Odontostilbe pulcher (2), Hemigrammus sp.(l), small juvenile Pygocentrus caribe (1); Curimatidae Curinwta n7etae (1), curimatid
(1); characoid (4); gymnotoid (1); Cichlidae Microgeophagus randrezi (1), cichlid (1). Fins of adult Pygocentrus caribe (2).

Serrasalmus metgnai Fish (whole or flesh): Characidae Aphyocharax erythrurus (1),
Hemigrammus nwrginatus (2), Hemigrammus sp. (1), and Odontostilbepulcher (2), characid (2); Curimatidae Curinwta ffwtae
(2), curitnatid (1); Cichlidae Microgeophagus rapdrezi (1); gymnotoid (1); catfish (1); Fins from Loricariidae (1).

Serrasalnws rhombus Fish (whole or flesh): Characidae Astyanax bitnaculatus and Charax
sp.(l), characid (1); characoid (1); Doradidae Agamxis? (3); unidentified catfish (1); Fins of Characidae Pygocentrus caribe (1); Erythrinidae? (1).
Amphibians: Small adult frog (1).




"C 00 00 C-4
1-11 C14

c oc C,6
en cn

cn N IC


7N C' 00 N in.




g CN

4.4 0


cz Col


It was usually not possible to identify the fins eaten, but the thick red anal fins of Pygocentrus caribe were found in two individuals. Large S. irritans preyed mostly on small whole fish, fish flesh, fins, and occasionally scales. Fishes eaten whole or in pieces by S. irritans are given in Table 5-2. A small juvenile P. caribe was found in the stomach of one individual. No nonfish vertebrate remains were found.

Serrasalmus medinai Valenciennes 1849

Serrasalmus medinai was the third most common piranha in the study area; 124 specimens were analyzed for diet (Table 5-4 and Fig. 5-3). It is a medium-sized piranha, usually less than 160 mm SL with a head shape that is intermediate between the robust P. caribe and species with pointed snouts (Fig. 5-2). Unlike Pygocentrus, but similar to other Serrasalmus, it has a series of ectopterygoid teeth that is easily seen even in small individuals (from about 25 mm SL). It is primarily a fin-eater when small (20-79 mm SL). The largest individuals (Group IV) fed chiefly on fish flesh, fins and scales, and small whole fish. Prey fish identified from the stomachs of S. medinai are given in Table 5-2. The only fin identified was from an armored catfish. Shrimp and crabs were also eaten.

Serrasalmus rhombeus (Linnaeus) 1766

Serrasalmus rhombeus was uncommon in flooded savanna of the Caflo Caicara

study area, although on several occasions adults were taken in large numbers by hook and line at night in Carlo Maporal. Juvenile Serrasalmus rhombeus have a somewhat elongate body and sharp snout (Fig. 5-2). They resemble S. irritans in body form, but are slightly more robust. Adults have heavy rhomboidal bodies. The body and fins of large individuals, some well over 300 mm SL, are black or blue-black, and the iris is often deep red. I examined the stomach contents of 51 S. rhombeus from the Caio Caicara area (Table 5-5 and Fig. 5-3).



00 oc

C14 I c

oc C14 00 I r

C-i W) C-4



U cu
N C6
z z


C14 rn Al C'i C,;
rn C14 M

Lei Cfi cn "Zr W)

cl C14 in

OC C'i C'i

4 06 cli


Q 09 r


00 en v 00
ZZ C14 C4 r,:

C14 It cn 0 0 >,

tA u u



Juvenile S. rhombeus in size classes iH-IV specialized on fish fins. Aquatic insects, mostly plecopterans, were packed in the guts of several specimens from Carlo Guaritico. The ten largest S. rhombeus (Group V) had taken almost equal volumes of whole small fish, chunks of fish flesh, and fish fins. Fishes identified from the stomachs of larger S. rhombeus are given in Table 5- 2. Three large S. rhombeus collected at night in Calo Maporal contained entire or partial individuals of small 20-30 mm SL doradid catfishes. One individual had eaten a small adult frog.

Serrasalmus elongatus Kner 1860

Serrasalmus elongatus was rare to uncommon in my samples in the study area. It is a pikelike piranha (Fig. 5-2), having the most elongate body, relative to body depth, of all piranhas. I examined 42 individuals, including a few from the small size class (Group II) from other locations in the Apure River drainage (Table 5-6 and Fig. 5-3). Stomachs of three small juveniles (20-39 mm SL) were packed with fins of other small fishes, and one also contained a nematode (possibly a parasite). Stomachs of large specimens contained fish fins and many scales. In addition, small whole fish and chunks of fish flesh were taken by the largest S. elongatus.

Serrasalmus altuvei Ramirez 1965

Serrasalmus altuvei was rare in the Caflo Caicara area, and was found in both streams and floodplain pools. It is a medium-sized piranha; maximum length in my samples was less than 200 mm SL, with a deep narrow body and a slender pointed snout (Fig. 5-2). It has a silver body and a broad, black terminal band on its unpaired fins, none of my specimens from the Apure drainage had any red on the body. I examined the stomachs of ten of 17 individuals collected in Caflo Maporal and the flooded savanna



Z 09


N C14



in C14

tn C'i 6 r-:

"t all cn N 00



cn C4 To


(Table 5-7 and Fig. 5-3). Fish fins, scales, flesh and small whole fish were found in the two size classes (il and IV) represented in our samples.

Pristobvcon striolatus (Steindachner) 1908
Pristobrycon striolatus was rare in the Cario Caicara study area, and was collected only on several occasions from Cailo Maporal. It is a medium-sized species (Fig. 5-2), less than 200 mm SL; the largest specimen taken from the Calo Caicara region was 156 mm. Adults sometimes had a red-orange chest and occasionally a pattern of orange vertical bars along the sides. Pristobrycon striolatus has a rather blunt snout, but smaller and weaker jaws and jaw musculature than Pygocentrus caribe. The body is deep and strongly compressed. Small juveniles are similar to young Serrasalmus rhombeus but are slightly stouter, lack ectopterygoid teeth, and have little if any side spotting. The largest individual collected in the study area was 159 mm SL. I examined the stomach contents of 16 of the 38 specimens collected. Food items by size class are given in Table 5-8 and Figure 5-3. Specimens in Group II fed mostly on fish fins as did the one Group III specimen examined. Three adults from Group IV contained fish flesh, a few scales, shrimp fragments, an adult aquatic beetle, an aquatic hemipteran and unidentified insect fragments.

Catoprion mento (Cuvier) 1819

Catoprion mento was common in many of the streams and savanna pools. It is a small, deep-bodied species (Fig. 5-2) that is closely related to piranhas. Although maximum size is around 120 mm SL, in the Apure drainage most specimens sampled were under 80 mm SL, and none was greater than 100 nm. The teeth of C. mento are somewhat everted and its long lower jaw swings open to more than 180 degrees, making it well suited for scraping scales off other fish. The stomachs of all 104 specimens examined contained fish scales (ranging up to 23 mm in diameter) (Table 5-9 and Fig. 5-3). Plant material, primarily comprising vascular debris and root wads, and occasionally filamentous