Dung beetles, monkeys, and seed dispersal in the Brazilian Amazon


Material Information

Dung beetles, monkeys, and seed dispersal in the Brazilian Amazon
Physical Description:
viii, 116 leaves : ill. ; 29 cm.
Vulinec, Kevina
Publication Date:


Subjects / Keywords:
Seeds -- Dispersal   ( lcsh )
Dung beetles -- Amazon River Region   ( lcsh )
Monkeys -- Amazon River Region   ( lcsh )
Forest regeneration   ( lcsh )
Animal ecology   ( lcsh )
Entomology and Nematology thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Entomology and Nematology -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1999.
Includes bibliographical references (leaves 102-115).
Statement of Responsibility:
by Kevina Vulinec.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 021581372
oclc - 43707593
System ID:

This item is only available as the following downloads:

Full Text







To the memory of my brothers, blood and spirit:

David Clover

John E. Fischer

Zack (The King)


I wish to thank the members of my committee, Carl Barfield, Colin Chapman,

Jim Lloyd, and Mike Thomas, for their support, advice, and encouragement. I

especially thank my committee chair, John Sivinski, for his enthusiasm, support, and

intellectually intriguing discussions.

I thank the following organizations for their financial support of my research:

The Entomology and Nematology Department at the University of Florida, the

Fulbright Commission for International Studies, the Gahan Foundation at the

University of Florida, the Charles A. and Anne Morrow Lindbergh Foundation, the

Florida Center for Systematic Entomology, the Dickinson Award for Tropical

Agriculture, the Women in Agriculture Club, and the J. D. Turner Foundation.

In Brazil, I am grateful to 0 Conselho Nacional de Desenvolvimento

Cientifico e Tecnol6gico (CNPq), Instituto National de Pesquisas da Amaz6nia

(INPA), and Museu Paraense Emflio Goeldi (MPEG). I am especially thankful to

Claudio R. V. da Fonseca of INPA for his friendship and immeasurable help. I also

wish to thank Pedro L. B. Lisboa for his assistance.

Special thanks are due Phoebe Wilson for the maps, David Almquist and

Freida Ansoanuur for sorting insects, Bruce Gill and Dave Edmonds for help with

identifications. I am very grateful to Ken Portier, Galin Jones, and Victor Hu for help

with the statistics. I thank Gina Posey and Sid Mayer for their computer and

programming help.

I also thank my sons, Coleman and Corey Kane for their help in Brazil and the

States, and their never-ending confidence that I would eventually finish and get a real

job. I thank my father, Tony Vulinec, with love, for a lifetime of support. Thanks are


also due Pette Blanchard, Julie Candela, Alfonso Diaz, and Joanna Lambert for

continual friendship and encouragement (we will survive). Finally, without the help

and support of my field assistant, packer, monkey-spotter, and entertainment

committee, my husband, Dave Mellow, this project would have died a lonely death in



ACKNOW LEDGMENTS....................................................iii

A B ST R A C T ....................................................................................... .................. vii

INTRODUCTION ................................. ....................................

Evolution of Dispersal Characteristics of Seeds..................................................3

Overview of Seed D ispersal ...............................................................5

Community Ecology and Seed Dispersal ........................................................8

Ecology of Dung Beetles ........................ ...............................

D ung Beetles as Seed D ispersers .................................................................................... 13

Background on Monkeys as Seed Dispersers......................... ...................... 16

Dung Beetles and Deforestation........................................................... 25

Forest Regeneration and Seed Dispersal.......................... ....... ......... 27

O objectives ........................................................ ............. ............. 28

M ETHODS .............................................................. 31

Study Sites ........................... ...... ....... ......................... 31

S urveys................................................................................................... .............. ......... 34

Seed Burial Experiments................................................... 38

RESULTS ........................................................ ............ ............ 40

B eetle Surveys ................................................................ 40

Monkey Censuses ............... ......................... .................... 41

M onkey and Beetle Comparisons.................................................. 42

Beetles and Seed Burial ........................................ ........................42



Community Analysis.................................................47

D ISCU SSIO N ......................................................... ............ 88

APPENDIX 1. BEETLE SPECIES COLLECTED....................... ...................... 98

APPENDIX 2. MONKEY OBSERVATIONS............................................ 99

LITERATURE CITED ............................................. 102

BIOGRAPHICAL SKETCH .......................................................................... ............. 116

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



Kevina Vulinec

December 1999

Chairman: Dr. John Sivinski
Major Department: Entomology and Nematology

Tropical rainforest trees have evolved strategies to move seeds away from the

parent plant. These strategies are thought to reduce transmission of predators and

fungal diseases from parent to offspring. One of the most common strategies to move

seeds is through fruit carried in vertebrate guts and defecated later. Seeds embedded in

dung and deposited on the forest floor are vulnerable to seed predation by rodents

unless buried by dung beetles. In this project, I explore the dynamics of dung beetle-

monkey-seed interactions and how this process may contribute to forest regeneration,

especially in secondary growth and disturbed habitat. Work was done at three sites on

the Amazon River in Brazil: Estalio Ferreira Penna in Para and Reserva Adolfo Ducke

in Amazonas, and private land near Cacaulandia, Rond6nia. I determined that different

species of dung beetles have varying efficiencies at seed burial, and beetle species were

ranked into seed burial guilds based on scores from a principal component analysis.

The potential for seeds to be buried is dependent on the dung beetle and monkey

communities at each site. Communities among habitats within the same site also

differ: clear-cut areas harbor a very low biomass of dung beetles, and the species are

different than either primary forest or secondary growth. Secondary growth contains a

subset of the same species found in primary forest, but lower in species richness and

abundance. Correspondence analysis of habitats and seed burial guilds suggests that

effective seed dispersal and burial in secondary growth depends on the amount of

disturbance of that habitat. Some management recommendations are made that may

encourage monkey and beetle seed dispersers to colonize disturbed areas and contribute

to forest regeneration.



Rainforests are defined and structured by their trees. These forests are also

dynamic; gaps are created by tree falls, allowing light tolerant species to grow, and gaps

are filled in successionally, becoming primary forest again (Martinez-Ramos &

Alvarez-Buylla 1995; Brokaw 1986, 1985). Regeneration in tropical forests is

dependent on seed dispersal, mostly accomplished by vertebrates. Although dispersal

can be by wind, water, or other factors, in the Neotropics, 45-90% of canopy trees and

almost 100% of shrubs and subcanopy trees have fruit adapted for animal dispersal

(Terborgh 1983; Howe & Smallwood 1982; McKey 1975). In Neotropical wet

forests, only 10% of trees are dispersed by wind; in contrast, 21% of trees in temperate

forests have wind dispersed seeds (Dirzo & Dominguez 1986). In the Neotropical

rainforests, monkeys are one of the primary dispersers of tree seeds, swallowing seeds

whole at the fruiting tree, and defecating them intact a distance away from the parent

tree (Emmons 1990; Estrada & Fleming 1986; Janson 1983). Primates are estimated

to be between 25-40% of frugivore biomass in tropical forests (Chapman 1995;

Terborgh 1983; Eisenberg & Thorington 1973). Nevertheless, more than 90% of

seeds dispersed by monkeys may be destroyed by rodents (Estrada & Coates-Estrada

1991). Secondary dispersal and burial by dung beetles may provide refugia for seeds

from rodent predation and be an important force in rainforest regeneration. Previous

work has emphasized the potential role of these insects in dispersal and seed burial


(Andresen 1999, 1994; Feer 1999; Estrada & Coates-Estrada 1991; Howard & Zanoni

1989) and in germination (Shepherd & Chapman 1998).

Dung beetles play integral roles in tropical ecosystems as decomposers, nutrient

recyclers, and seed dispersers. Perturbations of the environment change dung beetle

community composition and, often, decrease diversity (Vulinec, in press; Halffter et

al. 1992; Klein 1989; Mor6n 1987; Howden & Nealis 1975). As cattle ranching,

logging, and concomitant deforestation increase in the tropics (an annual mean

deforestation rate of 0.8% between 1980 and 1990; World Resources Institute 1994),

many species of dung beetles that are adapted to rain forest conditions may vanish.

Not only would extinction of this beetle fauna affect biodiversity, but the impact of its

loss on soil structure, nutrients, predatory phoretic mites and nematodes, and forest

regeneration is unknown.

Recently, much attention has focused on tropical forest regeneration and

reclamation (Laurance et al. 1997; Wunderle 1997; Laurance 1991), particularly in

places that have been deforested for centuries, such as areas in the Northeast of Brazil

(Uhl et al. 1997). In some cases, abandoned pastures and plantations may regenerate

back to the original forest, if enough intact rainforest is left in the surrounding area, and

if use of the degraded area was light or moderate, e.g. not bulldozed (Buschbacher et

al. 1992). However, not only must there be intact forest left, but the fauna involved in

ecosystem functions also must be preserved, including the pollinators and dispersers

(Pannell 1989).

My research has two sets of objectives: (1) delineate the characters that make

particular species of dung beetles good seed dispersal agents (dispersal defined as either

horizontal movement or burial), establish an index of seed dispersal abilities, and rank

each species in terms of relative quality as a seed disperser, and (2) examine whether

beetles high on the dispersal index list are found in primary forest, secondary growth,

and clear-cut land, and what factors might influence their distributions in these three

habitats. The total index of a dung beetle community over time, in addition to other

factors, such as soil quality, amount of disturbance, or other seed dispersers, could be

used as a score for a particular location or habitat, and an indication of the partial

potential of this area for regrowth and continued regeneration. Thus, this work will

contribute towards determining some of the factors that affect primary forest

regeneration from areas of secondary growth.

The three study sites were located in upland moist tropical forest within the

Amazon Basin, in the states of Rondonia, Amazonas, and Para (Figure 1).

Evolution of Dispersal Characteristics of Seeds

While seeds in dung may only exert weak selection on dung beetles, possibly

by increasing handling time of dung, or reduction of dung in a brood ball (Andresen

1994; Estrada & Coates-Estrada 1991), beetle behavior may impose stronger selection

on plants. Measuring the force of selection exerted by beetles on whole suites of

morphological features of tree seeds may be possible. This goal could be

accomplished by measuring a performance gradient: a score in an ecologically relevant

activity. In this case, seeds facilitate through their size and shape their own

transportation and burial by dung beetles. The performance gradient is the partial

regression of performance on the stated character, holding all other characters constant.

The fitness gradient is the partial regression of relative fitness on performance, holding

all other performance variables constant (Arnold 1983). Measuring how a seed's shape

and size contributes to fitness directly may be impossible, but one can measure how

shape and size affect a seed's performance in dispersal.

Seeds might be subjected to opposing selection by monkeys and dung beetles.

For example, large seeds pass more slowly through a monkey's gut, and therefore may

have an increased chance of being deposited away from the parent's seed shadow

(Janzen 1986). Small seeds, on the other hand, are less likely to be removed from

dung by beetles (Andresen 1994; Estrada & Coates-Estrada 1991). If trade-offs occur,

the covariance between performance measures should be negative; the magnitude

indicates the strength of the conflict (Arnold 1983).

The study presented here addresses in the abstract sense the "performance" of

seeds under a range of possible selection pressures (the difference in each dung beetle

species' mean contribution to the fitness of particular tree seeds). Only one component

of an entire performance gradient of a tree species is being measured. However, other

measures of performance, such as the effects of monkeys, could be measured and

fitness gradients, at least for juvenile survival, could be constructed with laboratory

studies of germination success under different conditions. By looking at seed size and

shape, one could construct a selection surface and determine the force of both

stabilizing and directional selection on seed morphology (Lande & Arnold 1983).

While there is much debate about the extent of coevolution between trees and

their seed dispersers (Levey & Benkman 1999; Herrera 1998; Lambert & Garber

1998), 45-90% of tropical fruit trees appear to be adapted for vertebrate seed dispersal.

Large seeds can better survive environmental stress such as drought, shade,

competition, and greater depth of burial (Westoby et al. 1996; Gulmon 1992).

Nevertheless, stabilizing selection from dispersers may occur in an upper limit to seed

size, or in the number of seeds in a given fruit. Monkeys swallow seeds only up to a

certain size, usually no more than 3 cm in the largest dimension (Lambert 1999; Castro

1991; Garber 1986). If dung beetles also exert directional selection, large seeds would

probably be selected against, while it is doubtful that there would be a lower limit on

seed size, unless larvae eat seeds in their brood balls. Larval feeding may be

encountered less frequently than adult feeding behavior, and adults do not have

adequate mouthparts for fiber mastication (Halffter & Edmonds 1982). The selection

pressures on seeds exerted by dung beetles have not been studied, and their importance

relative to primary dispersers is largely unknown (Herrera 1998). Nevertheless, given

the large number of dung beetles in tropical forests, possibly more than 2000 beetles

per ha (Peck & Forsyth 1982), their effect on seed survival may be important, and

influence forest regeneration and tree distribution patterns, if not the evolution of

character states in seeds.

Endozoochorous seeds will have a probability of being eaten, a probability of

being dispersed away from the parent plant, a probability of being buried by dung

beetles, and a probability of surviving in the environment they subsequently encounter.

The effect of dung beetles on seed survival may be negligent or substantial. A seed,

once defecated, may be left on the surface of the ground (in dung or not), buried

shallowly, buried deeply, buried in a food ball, buried in a brood ball, buried with dung,

buried without dung, moved from the dung source, buried under the dung source,

buried with many other seeds, buried with few other seeds. All these possibilities may

impact a seed's survival, even before the seedling has to deal with the environmental

factors at the site. Dung beetles may play a significant role in seed survival during this

phase of a seed's life.

Overview of Seed Dispersal

Seed dispersal and the possible evolutionary tracking between dispersers and

fruiting plants has been increasingly questioned in recent literature (Levey and

Benkman 1999; Herrera 1998; 1986). In one 12-yr. study, Herrera (1998) found that

although there was large supra-annual variation in fruit supply and fruit-eating bird

abundance, there was essentially no correlation between the two; bird abundance was

best explained by autumn weather patterns. These results imply that mutual selection

pressures between plants and their seed dispersers are constrained (Levey & Benkman

1999). On the other hand, another study, using a multivariate approach, suggests that

there was significant selection by brown capuchin monkeys on some plant traits

influencing the energy, carbohydrate, and protein content of fruit, and spatial aspects of

fruit displays (Janson et. al. 1986). Furthermore, some seeds must pass through

vertebrate guts to germinate (Lambert 1999; Rowell & Mitchell 1991; Chapman 1989;

Idani 1986), and plant adaptations to attract dispersers, such as nutritionally rich fleshy

fruit, suggest past coevolution between plants and vertebrate dispersers (Dirzo &

Dominguez 1986)

The selection pressures exerted by extant dispersers may be variable or weak,

nevertheless, the ecological impact of dispersers on forests and plant communities can

be important (Holl 1999; Chapman & Chapman 1996, 1995; Hubbell & Foster 1990;

Janzen 1986, 1970). Within limits, seedling survival increases with distance from the

parent plant (Howe et al. 1985; DeSteven & Putz 1984; Augspurger 1983). Specific

mortality factors and light gaps complicate many studies, so the results are not always

clear-cut (Howe et al. 1985; Clark & Clark 1984), however, most studies support the

escape hypothesis (Dirzo & Dominguez 1986). In this hypothesis, it is proposed that

escape from predators and pathogens of the adult tree is the driving force behind plant

adaptations to move seeds to other sites (Howe & Smallwood 1982; Connell 1971;

Janzen 1970). Additionally, dispersal reduces competition for resources, such as

nutrients, light, and water, among relatives (Stiles 1989). Even in Paleotropical forests,

35-45% of tree species and 70-80% of understory plants had animal dispersed seeds

(Howe & Smallwood 1982). One study estimates that if frugivorous seed dispersers

at Kibale National Park in Uganda were removed, 60% of tree species would be lost

(Chapman & Chapman 1995).

Soil, water availability, cover, microclimate, and seed predators all affect

seedling establishment. Seed predators, however, may have the greatest impact on

early survivorship. Janzen (1986) found seed removal from experimentally placed

cow and horse dung to depend on the type of dung, the number of seeds, the size of the

seeds, and the placement of dung. Mice removed a majority of seeds from dung, and

all the seeds from dung with high seed density (Janzen 1986). Moreover, mice were

attracted directly to the dung. This experiment was carried out in Santa Rosa National

Park in Costa Rica during the dry season; at this time there were no active dung beetles

to bury seeds and influence the results of rodent seed removal.

Other studies show how detrimental predation on the forest floor is to seed

survival. In Manu National Park, Perfi, Andresen (1999) found that while dung odor

attracted rodents, 76-80 % of seeds buried experimentally in dung at 3 cm or deeper

escaped rodent predation, whereas only 33% of seeds at 1 cm escaped. In Kibale,

Uganda, 95 98% of seeds experimentally buried in dung escaped removal even at 1

cm depth (Shepherd & Chapman 1998).

Vertebrate dispersal results in non-random patterns of tree growth. Forget &

Sabatier (1997) identified the arboreal pathway of a group of spider monkeys (Ateles

paniscus) in French Guiana and it showed a good relationship between the 1984

seedling shadow of Virola and current trees with diameters > 30 cm. Similar seed

shadows predicted from monkey activity patterns were found in tamarins (Garber

1986), and howlers (Julliot 1997).

Community Ecology and Seed Dispersal

Communities of trees are often used to give a definition to a particular forest

type (e.g. Caatinga, liana forest, chavascal, buritizal; Pires & Prance 1985).

Communities of animals are harder to delineate, as they are mobile, shorter lived, and

often difficult to measure. Nevertheless, for animals with distinct home ranges or

specific habitat requirements, it is sometimes possible to define a community.

Animals that depend on ephemeral resources often experience intense

competition, and the populations in these groups is limited. The habitat will only

support a certain number of animals, and, especially during times of food shortages,

many animals may die (Wright et al. 1999).

Monkey community structure in a particular location will often be determined

by the local productivity (Kay et al. 1997), annual flooding of forests (Peres 1997),

and hunting and other disturbance (Chapman & Balcomb 1998; Terborgh 1985).

Species composition and abundance of monkey communities may in turn influence the

fruiting tree communities, as evidenced by seed shadows predicted by monkey activity

patterns (Forget & Sabatier 1997; Julliot 1996; Garber 1986). While monkeys are not

the only seed dispersers in Amazon forests, they are the ones that provide the most

seeds for dung beetle burial, given the size of the dung pat, place of deposition, and the

preference for primate dung in the majority of dung beetles (Estrada et al. 1993;

Halffter & Matthews 1966).

In the same way, dung beetle communities will have an influence on the seed

shadow and possibly the adult tree community in a locality. For example, recent

research (Vulinec, in press; Feer 1998) shows vast differences in the performance of

specific dung beetles in their seed burying abilities. Dung beetles vary in size, activity

schedule, and dung processing behavior. Beetles are broadly grouped into three guilds

by behavior: endocoprids, which live and nest within the dung pat, paracoprids, the

burrowers, which bury dung directly under the dung pat, and telocoprids, the rollers,

which form balls and roll the ball away from the deposition site to be buried later

(Halffter & Edmonds 1982; Halffter & Matthews 1966). A community made up

largely of small rollers will not be as effective in seed burial as communities that

contain species of large burrowers. Furthermore, the community of dung beetles at a

site may have a role in the pattern of seed burial, and consequently, the seed shadow in

the forest. Small diurnal rollers will bury seeds along monkey foraging routes, or at

foraging trees. These seeds are predicted to be buried shallowly, and dispersed widely

along the route. On the other hand, large nocturnal burrowers will bury seeds in a large

bolus of dung; the predicted pattern of soil seed bank would be clumped and deep.

Additionally, the mostly smaller diurnal beetles do not bury large seeds well, so these

seeds would be left vulnerable to rodent predation unless buried at night.

Ecology of Dung Beetles

Dung beetles (Family Scarabaeidae; subfamily Scarabaeinae) are important

components of ecosystems, especially in the tropics, where the highest diversity of

dung beetles in the world occur (Halffter & Matthews 1966). As mentioned, this

group of beetles is split into three main categories or guilds, separated by dung

processing behavior: rollers, burrowers, and dwellers; these categories can be separated

further by more detailed descriptions of nesting behavior (Halffter & Edmonds 1982).

All dung beetles considered in this study were rollers (those that make balls of dung at

the source, and roll it away to be buried later) or burrowers (those that bury dung

directly under a dung pat). Both of these groups are considered as secondary

dispersers, whether the dispersal is vertical or horizontal. Dwellers, such as the genera

Aphodius and Ataenius, do not bury dung and are generally small (< 8 mm;

Woodruff 1973), characteristics that limit their effect on seed dispersal. Of the

burrower and roller guilds, several tribes are important dung handlers that do influence

seed dispersal. Adult dung beetles bury dung for their own consumption and to make

nests for their larvae. These beetles have low reproductive rates for insects, in fact

some may have only three to eight young per female (Halffter & Edmonds 1982).

Concomitant with fewer young, these beetles exhibit a high degree of parental care, and

often biparental cooperation in nest building (Halffter & Edmonds 1982; Halffter &

Matthews 1966). Burrows made for feeding are usually similar in construction to nest

burrows, but less complex and more shallow, and significantly more abundant

(Otronen 1988; Halffter & Matthews 1966).

In addition to the decomposition and nutrient recycling roles played by dung

beetles, they also help reduce pest insects in dung (Fincher et al. 1981; Bomemissza

1970). While dung beetles themselves do not eat dung breeding pests, they often carry

phoretic predatory mites and nematodes that do (Klein 1989; Halffter & Matthews

1966). They also bury a considerable amount of feces, which limits the amount

available to flies or other dung breeding insects. For these reasons, they are considered

beneficial to humans, and there is a thriving introduction program in Australia (Doube

1986; Kirk & Ridsdill-Smith 1986; Bomemissza 1976; 1979), and a discontinued one

in the United States (Fincher 1981, 1986). Brazil has started to import African dung

beetles adapted to pastureland as biological control agents (Fincher pers. com.;

EMBRAPA pers. com.).

Preferences for dung exist, although most beetles, especially in Neotropical

rainforests, are generalists and opportunistic (Gill 1991; Halffter & Matthews 1966).

Some beetles are attracted to both dung and carrion, and a large group, of many genera,

are strictly carrion feeders (Gill 1991; Estrada et al. 1993; Peck & Forsyth 1982).

The following are descriptions of some of the more important dung beetle taxa

at my study sites.

The Onthophagini (genus Onthophagus) are primarily small (< 10 mm), and

make simple or compound burrows directly under the dung pat for nesting (Halffter &

Edmonds 1982) and eating (Vulinec, pers. obs.). Males of many species have horns

or projections on the head or pronotum, and bisexual cooperation in nesting is common

(Halffter & Edmonds 1982; Halffter & Matthews 1966). These beetles are common

on mammal dung, but are sometimes found on reptile dung, bird dung, carrion, and

fruit (Gill 1991; Young 1981). One large African species, Onthophagus gazella,

introduced in 1974 into North America for control of dung breeding pests (Fincher

1981; Vulinec & Eudy 1993) has now spread from the release site in Georgia to

Chiapas, Mexico (Montes de Oca & Halffter 1997), and is currently being released in

Brazil (Fincher pers. com.; EMBRAPA pers. com.).

Phanaeini (primarily the genera Phanaeus, Coprophanaeus, Oxysternon, and

Sulcophaeus) are large (10 mm 40 mm), and abundant in South America (Gill 1991).

Phanaeus spp. in South America are diurnal dung feeders, often flying for long

periods during the day to forage (Gill 1991). Those species occurring in Amazonian

forests are normally dull brown, with some metallic reflections on the pronotum and

the venter (Vulinec 1997; Edmonds 1994). Dung is pulled or pushed into holes dug at

the edge of a dung pat, although sometimes the dung will be pushed a distance away

from the pat; bisexual cooperation in nest building is common in this genus (Edmonds

1994; Halffter & Matthews 1966). The genus Coprophanaeus contains some of the

largest dung beetles in the new world; Coprophanaeus lancifer may be as large as 50

mm in length. In this species, both sexes are homed and may jointly defend the

nesting burrow or food resources (Otronen 1988). They are crepuscular and iridescent

violet colored (Vulinec 1997; Edmonds 1994). C. lancifer is reported to be

necrophagous (Edmonds 1972), but is often caught at dung and rotting fruit (Vulinec,

in press). Other members of the genus appear to be strictly necrophagus (Otronen

1988; Edmonds 1972). Oxysternon is large, brightly colored, iridescent, diurnal, and a

very strong flyer. Like Phanaeus they also spend considerable time during the day

on foraging flights (Gill 1991). They appear to be solely coprophagus (Peck &

Forsyth 1982; Edmonds 1972). All beetles in this tribe make large burrows and bury

brood balls deeply (Halffter & Edmonds 1982), and in South America are restricted to

forested areas (Gill 1991).

The Dichotomiini tribe encompasses almost half of the dung beetle species in

the Neotropics, including the common genera Dichotonmius, Ateuchus, and

Canthidium. Dichotomius is generally a dung feeder, but occasionally feeds on carrion

or fruit (Gill 1991). These beetles are nocturnal, large (10 mm 30 mm), and most

abundant during the rainy season (Janzen 1983). There are some reports of this genus

pushing food away from the source (Luederwaldt 1914). Canthidium are small,

between 2 mm 13 mm, mostly diurnal, and often perch in wait for food odors (Gill

1991; Diaz & Vulinec in prep.). They may follow and perch near foraging mammals,

even before dungfall (Gill 1991). Some species will roll pellets of rodent dung (Gill

1991). The smaller species bury dung in shallow burrows (Jones 1998).

The Eurystemini all exhibit complex ball forming and pair-bonding behavior,

but rarely bury dung more than a centimeter below ground (Halffter & Matthews


The Canthonini are a large group of ball-rollers that are very common in Central

and South America (Halffter & Matthews 1966). Deltochilum, with 76 New

World species, are large beetles, between 8 mm 30 mm (Gill 1991). These beetles

come to both carrion and dung (Halffter & Matthews 1966). Canthon, the largest

supragenus (including Glaphyrocanthon, Trichocanthon, Scybalocanthon, and others)

in tropical America, with 148 species, are normally small, diurnal ball-rollers (Gill

1991; Halffter & Matthews 1966). One notable exception is the relatively large

nocturnal Canthon aequinoctialis (Howden & Young 1981). Beetles of the Canthon

group may roll balls up to 5 m away from the site of deposition (Halffter & Matthews

1966). Many are adapted to savanna conditions, and are the dominant genus found in

clear-cut areas in Amazonia (Halffter 1991; Klein 1989).

Dung Beetles as Seed Dispersers

Dung beetles of the roller and burrower guilds dig nests under dung pats or roll

the dung away from the site of deposition to be buried later. Most species make

feeding balls that are buried, consumed, or may be abandoned uneaten. Additionally,

beetles use dung to supply the larvae with sufficient food for development. Nests are

provisioned with dung, usually formed into balls, and eggs laid in the balls. The larvae

develop and pupate within these brood balls, and emerge as adults. Brood balls and

feeding balls may differ in the amount of contaminants, such as seeds, that beetles

remove, as some beetle species construct elaborate and immaculate brood balls

(Halffter & Edmonds 1982).

Recent research indicates that dung beetles are an important link in forest

regeneration through their activities as secondary seed dispersers (Andresen 1999; Feer

1999; Shepherd & Chapman 1998; Estrada & Coates-Estrada 1991; Wicklow et al.

1984). Because monkeys are one of the primary tree seed dispersers in tropical forests

(Estrada & Fleming 1986), and the majority of seeds on the ground in these forests are

destroyed by rodents (Andresen 1999; Shepherd & Chapman 1998; Hulme 1993),

dung beetles that bury monkey dung may play an important role in seed survival. At

Los Tuxtlas, Mexico, more than 90% of seeds in monkey dung were lost to rodent

predation unless relocated, primarily by dung beetles, which may bury up to 60% of

these seeds, contributing to enhanced germination (Estrada & Coates-Estrada 1991).

In the study at Los Tuxtlas, rodents were able to locate 90-100% of seeds on the soil

surface, but only 56% when seeds were 2.5 cm under ground. In another study, at

Manu, Peru, only 5% of seeds buried experimentally in monkey dung at a depth of 5

cm were discovered by seed predators (Andresen 1994; 1999). Work in Africa also

suggests increased seed survival in dung beetle buried seeds (Shepherd & Chapman


Dung beetle species will differ in their quality as seed dispersers. In

experiments using PVC cylinders, Estrada & Coates-Estrada (1991) found that 41% of

total experimental seeds were buried by dung beetles (30% by burrowers and 11% by

ball rollers). Eighty-three percent were buried at a depth greater than 2.5 cm and 54%

were buried at depths greater than 5 cm. Burrowers buried seeds of all 20 species of

trees examined (range 1-18 mm in length), but ball-rollers buried the seeds of only 11

of the 20 possible plant species (Estrada & Coates-Estrada 1991). Burrowers should

be better than rollers at seed protection, but rollers tend to relocate seeds, and this

dispersal is thought to contribute to greater escape from the parent tree and enhance the

survival of seedlings (Chapman & Chapman 1995). Seed size is negatively correlated

with the percentage of seeds buried by beetles (Andresen 1999; Estrada & Coates-

Estrada 1991). Small seeds may be more readily buried; however, small seeds such as

Ficus may not germinate at great depths. In the study by Estrada and Coates-Estrada

(1991), the majority of seeds were buried at depths between 2-5 cm, which may be an

optimal depth for both germination and rodent escape, at least for larger seeds.

Nevertheless, some large species of beetles, which would be expected to bury a large

quantity of dung and seeds, may bury brood balls at depths of 40 cm (Peck & Forsyth

1982), depths which may exclude them from being high-quality dispersers. Dung

preferences also exist; species of beetles attracted to herbivore dung would be better at

dispersing dung and seeds (Estrada et al. 1993; Halffter & Matthews 1966). Diurnal

beetles may follow monkey troops on foraging trips (Gill 1991), and nocturnal beetles

may forage at monkey sleeping trees (Feer 1999).

In French Guiana, Feer (1999) found that the proportion of seeds buried by

dung beetles was inversely correlated with seed size. A high percentage of seeds < 5

mm are buried by dung beetles, but seeds of 30 mm in the largest dimension were only

buried 13% of the time. Feer also found, as in other studies (Estrada & Coates-Estrada

1991; Vulinec, in press), that rollers buried fewer and smaller seeds than burrowers.

Also, in this study, burrowers buried seeds an average of more than 12 cm, a depth that

would reduce germination success substantially (Shepherd & Chapman 1998). Feer

(1999) suggests that rodent scatterhoarding and caching of seeds may result in greater

secondary dispersal and survival of large seeds than the activities of dung beetles.

Background on Monkeys as Seed Dispersers

Along with birds and bats, monkeys are one of the most important seed

dispersers in the Neotropics (Castro 1991; Estrada et al. 1991; Chapman 1989; Howe

1989; Garber 1986; Rowell & Mitchell 1991). Some authors consider monkeys to be

poor seed dispersers compared to birds, because they defecate seeds in clumps rather

than scattering them, contributing to density dependent mortality in the seeds (Howe

1980, 1989). Other authors suggest that monkeys are high quality dispersers. Of fruit

species consumed by three monkey species in Costa Rica, 53% had seeds that were

found intact in primate dung (Chapman 1989). Chapman (1989) states that of seeds

dispersed by primates and deposited on the forest floor, the vast majority are either

killed by predators or moved by secondary dispersers. Post-dispersal burial by dung

beetles may effect seed survival, and consequently raise the quality of monkeys as seed

dispersers over birds, because bird dropping are much less attractive to dung beetles

(especially large burrowers) than monkey droppings (Halffter & Matthews 1966). On

the other hand, some beetles may bury seeds too deeply for germination, negatively

affecting seed survival (Feer 1999).

Zhang and Wang (1995) state that a dispersal agent defecating few seeds in

many small dung piles is better than one defecating many seeds in a few large dung

piles. When these dung piles are dropped may be important also. For example, large

piles at night may be buried more completely by the large nocturnal burrowers than

piles dropped during the day, when small rollers are more active. Monkey behaviors,

amount of fecal material, and number of intact seeds defecated by a monkey

community in a given area will greatly influence the dung beetle communities'


I concentrate my discussion on the most abundant seed dispersing monkeys in

my three study locations and on what is known of their seed dispersal and defecation

patterns: saddleback tamarin (Saguinus., i../l -. i. midas tamarin (Saguinus midas),

Brazilian bare-face tamarin (Saguinus bicolor), brown capuchin (Cebus apella), red

howler (Alouatta seniculus), and red-handed howler (Alouatta belzebul). I give brief

background information on the others.

Silvery Marmoset (Callithrix argentata argentata)

These elf-like monkeys are found in the lowland forests of the eastern Amazon.

Group sizes range from 5 14 (Rylands & de Faria 1993). At Caxiuani, my study site

in Pari, 8 individuals were recorded in 1995, with a home range size of 8 ha, typical of

more gummivorous Callitrichids (Veracini 1996). Veracini reported greater use of

exudates in this species than for other Callithrix. Although sympatric with Saguinus

midas, there appears to be segregation in habitat use, as S. midas can use primary

forest, while C. argentata appears to need at least some disturbed forest (Veracini

1996). With its small home range, use of secondary forest, and large consumption of

plant exudates, this primate not a major factor in primary forest seed dispersal

dynamics, although they may prove to be very important in dispersal in capoiera

(natural secondary regrowth), where capuchins and howlers are less frequent.

Saddleback Tamarin (Saguinusfusicollis)

One of the most commonly seen species in my study site in Rondania was the

saddleback tamarin, Saguinusfuscicollis. The saddleback tamarin (with 14

subspecies: Hershkovitz 1977) ranges between 308 535 grams, smaller than some

squirrels (Garber 1986). Territories are between 30 120 hectares in size (Terborgh

1985). Group sizes are generally small, between 1-17 individuals, with a mean of

almost six individuals.

In general, tamarins tend to concentrate their feeding on only a few species of

plant at one time. In Cocha Cashu, Peru, one troop used only six major resource

species over one year (Terborgh 1985). Saguinusfusicollis appear to be aware of the

location and fruiting schedules of many tree and liana species in their home ranges

(Garber 1986).

At the Rio Blanco field station in Peru, Garber (1986) collected data on the seed

dispersal abilities of a mixed group of tamarins, Saguinus nmystax, the mustached

tamarin, and Saguinusfusicollis. These monkeys were found to consume and excrete

intact seeds as large as 25 mm. The mean seed size passed by S. fusicollis in this

study was 14.7 mm. Nevertheless, these large seeds were often voided without much

attendant fecal material. Smaller seeds, such as those of Ficus, were more likely to be

passed in a fecal mass. According to Garber (1986), these monkeys are high-quality

dispersal agents; dispersing seeds between distances of 34 to 513 m, the distances

positively correlated with the time of gut passage. Seeds with a higher specific gravity

were passed more quickly, and thus, would be deposited closer to the parent plant.

Some 87% of seeds consumed by these monkeys were moved 100 m, 51% were

moved more than 200 m, and 33% were moved more than 300 m. The present

distribution of adult trees in his study was consistent with tamarin-dispersed seed

shadows. Additionally, 70% of seeds that pass through tamarin guts germinated.

Castro (1991) found S. fusicollis in Peru consumed seeds as large as 2.4 cm, and he

states that tamarins may serve as dispersers for 90% of the seeds they consume.

The tamarins at Rio Blanco tended to consume fruit early in the morning, and

as seeds passed through their digestive tract in one to five hours (usually less than three

hours), generally their peak of seed defecation was on or before noon (Garber 1986). In

another study, S. fusicollis experience peaks in defecation at 1000 hrs and at 1300 hrs,

and gut passage of seeds is between 3 4 hrs, but may be longer for some species of

fruit (Castro 1991).

Midas Tamarin (Saguinus midas)

The midas or golden-handed tamarin is one of the largest tamarins, with a mean

body weight of 590 g (Pack et al. 1999). They form groups of 2 6 (Mittermeier et.

al. 1977). I saw a solitary midas tamarin in Reserva Ducke, my site in the central

Amazon, an area where they may be competitively displacing the Brazilian bare-faced

tamarin (Ayers et al. 1980; Van Roosemalen, pers. comm.) and a group of 3 in

Caxiuan5. They forage primarily in open high forest, unlike other members of the

genus, and at height of 5 25 m. They also often travel in the canopy, and may also be

found in dense viney habitats such as streamsides (Mittermeier et al. 1977). They are

also one of the more wide-spread tamarins, occurring in much of northeast Brazil to the

Rio Negro, Guyana, Surinam, and in French Guiana (Emmons 1997). Their annual

diet consists of 47.1% fruit (32.2% of this total being seeds), 50.2% invertebrates, and

fiber and vegetable matter as the remainder. S. midas consumes more seeds during the

dry season, and may be important in seed dispersal during the peak fruiting season, but

be more granivorous during the dry season (Pack et al. 1999).

Brazilian Bare-Face Tamarin (Saguinus bicolor bicolor)

The bare-faced tamarin has one of the most restricted ranges of any tamarin. It

thrives in secondary forests, but its range is only a small strip along the Rio Negro.

Due to destruction of its habitat it is listed as CITES Appendix I, US-ESA endangered,

and IUCN Red List Endangered (Emmons 1997). Tamarins are not widely hunted,

and the species listed as critically endangered got that designation due to habitat

fragmentation and destruction (Emmons 1997).

S. bicolor lives in groups of 2 9 individuals, and usually forage in dense viney

vegetation in secondary forests and along edges. One study records annual

consumption of plant parts as fruit 96.1%, 3% exudates, and 0.9% flowers. Of all

attempts at prey foraging, arthropods were only caught in 5.4% of the records, although

the group spent more time searching for arthropods than feeding on plant material

(Egler 1992). These tamarins were observed to discard the skin and seeds of pulp,

except where the seeds were small and numerous (e.g. Rollinia). The group in this

study repeatedly used the same travel routes (Egler 1992).

Night Monkey (Aotus sp.)

Night monkeys (weighing 780 1249 g; Emmons 1997) eat fruit and insects,

and appear to eat the same species of fruit as larger diurnal monkeys. At densities of

40 individuals per km2 in Cocha Cashu it could be a large contributor to seed dispersal;

their nocturnal habits make them difficult to study (Wright cited in Terborgh 1985).

Dusky Titi (Callicebus moloch)

I found these small monogamous monkeys only at the Rond6nia site, and only

one family group was seen. Like howlers, these monkeys are primarily folivores, and

spend 60% of the day at rest (Terborgh 1985). They prefer forest edges, and this

habitat was where I saw them. The group in RondBnia spent a few days eating

Cecropia fruit.

Common Squirrel Monkey (Saimiri sciureus)

Squirrel monkeys are medium sized (480 1400 g), form large groups of 25 -

100, and use home ranges of 100 500 ha (Costello et al. 1993). They often associate

with capuchins, which appear to be better at finding fruit, and appear more alert to

danger than squirrel monkeys (Terborgh 1985).

Brown Capuchin (Cebus apella)

The brown capuchin has the largest range of any primate in the Amazon basin,

being found in all of Brazil and north into Venezuela, Peru, and Ecuador (Terborgh

1983). Capuchins have omnivorous diets, eating any prey from insects to lizards and

small birds, to nuts, pith, flowers, nectar, fungi, and sap; although fruit is the main

element in the diet. They are able to open tough palm nuts that are not available to

smaller monkeys, and can use these in the early dry season when fruit is scare. During

the day, they are almost constantly moving, feeding continuously as fruit trees are

found (Terborgh 1985). Weight ranges from 1.7 4.5 kg, and their group sizes vary

from 5 20, foraging is usually in the middle to lower levels of the forest, but they will

utilize the upper canopy (Janson 1985).

Capuchin feces consists mostly of seeds held together by a small amount of

organic matter. These loosely structured feces are dropped from heights of 5 30 m,

and seeds are often scattered as they fall (Rowell and Mitchell 1991). Another study

found capuchin feces in small clumps with a small quantity of seeds (mean of 3 seeds

per clump; Zhang & Wang 1995). Ingested seeds may range in size from 0.5 18

mm in the maximum dimension (Rowell and Mitchell 1991). Seeds of 35 of 39

species found intact in capuchin dung were planted, and 31 germinated (median

percentage germinated 60.5, with a range of 0 100%; Rowell and Mitchell 1991).

Capuchins move at a rate of 200 m / hr. Their gut passage rate is 1 5 hours, allowing

seeds to be dispersed 200 1000 m from the parent tree (Mitchell 1989; Rowell and

Mitchell 1991).

In one study comparing capuchin and spider monkey seed dispersal patterns

(Zhang & Wang 1995), capuchins were found to swallow seeds less frequently than

spider monkeys (61.2% of fruits ingested whole, as opposed to 88.3% for spider

monkeys), but because capuchins spent less time in the tree, fewer seeds were

defecated under the feeding tree (4.5% vs. 26.2%). Zhang and Wang (1995) found that

capuchin defecations had fewer Ziziphus seeds than spider monkey droppings (mean

2.8 vs. 30.6), and that spider monkeys dispersed 51.1 times more seeds than

capuchins. However, because spider monkey dung was scattered widely by striking

understory plants, seeds in capuchin dung persisted longer on the forest floor, and had

a higher survival rate (9.8 % for capuchins, and 3.8% for spider monkeys). They

conclude that capuchin seed dispersal resulted in a higher seed survival rate than spider

monkey dispersal, due to fewer seeds per dung pile, and more seeds left on the forest

floor. Nevertheless, they did not consider that seeds may have been removed by dung

beetles rather than predators, and this removal would increase survival of seeds. Both

these species are considered good dispersal agents (Chapman 1989; Mittermeier &

Van Roosmalen 1981).

Guianan Saki (Pithecia pithecia), Monk Saki (Pithecia monachus), Brown
Bearded Saki (Chiropotes satanus)

All the sakis are considered seed predators, although they probably disperse

some seeds (Norconk 1998). The proportion of ripe and unripe seeds in their diets

ranged between 18 46% (Peres 1993; Van Roosemalen et al. 1988). In one study,

Chiropotes satanas masticated seeds in 86.4% of feeding samples (Norconk &

Kinzey 1994). Nevertheless, the relative amount of biomass of seed predators to seed

dispersers will influence the total amount of primate dispersed seed rain.

Red Howler (Alouatta seniculus) and Red-Handed Howler (Alouatta belzebul)

Red howler monkeys average 6.5 kg in weight (range 3.6 11.1; Peres 1997;

Emmons 1997). Their group sizes vary from 5 8 (Crockett & Eisenberg 1987).

Red-handed howlers are slightly smaller (4.8 8 kg; Emmons 1997). Group sizes for

this species range from 2 8 (Branch 1983). Vocalizations are similar; the red-handed

howlers have a higher pitched voice (Emmons 1997). The density of these species

varies widely with location. In Cocha Cashu, Peru, density was 30 individuals per

km2 (Terborgh 1985), while in Venezuelan Ilanos it was 150 per km2 (Rudran 1979).

At several transects along the Rio Jurud in Brazil densities ranged from 2.6 9.2 per

km2 (Perez 1997). Howlers in general have widely variable densities from under 10 to

over 100 per km2 (Crockett 1998; Chapman & Balcomb 1998).

Alouatta belzebul is an Amazonian forest species, while A. seniculus is found

in a wider range of habitats, often along rivers (Crockett 1998; Peres 1997). Because

of their small home ranges, typically < 25 ha (Crockett 1998), howlers can exist in

forest fragments, disturbed habitats, and near human habitation, although many may

survive on sub-optimal diets (Crockett 1998; Neves & Rylands 1991). Chapman &

Balcomb (1998) found howler populations to be primarily influenced by the local

history of disturbance, hunting, food crop failure and disease, rather than productivity

of an area. Howlers are easily hunted, due to their vocalizations and sedentary habits,

and have been exterminated from many forest fragments with high hunting pressure

(Chiarello & Galetti 1994). While in Reserva Ducke, a protected reserve, I saw

howlers that had been killed by hunters; especially unfortunate because the standard of

living among the local people was far better than many areas along the Amazon and its

tributaries (Vulinec & Mellow, pers. obs.).

Howlers are more sedentary than other monkeys; allowing them to take

advantage of longer-to-digest foliage, and they have the ability to detoxify some plant

defensive compounds (Milton 1981). They often eat only 3 times a day, spending the

remainder of the time resting and digesting (Terborgh 1985). Although considered

folivores, fruit makes up 13 73% of their annual diet (Julliot 1994), and, in one

instance, only 15.1% of dung samples from howlers contained no seeds (Chapman

1989). Gut transit time for A. seniculus is 35 hours (Lambert 1998; Crissey et al.

1990). Feces are compact (Van Roosmalen 1985), and may be excreted up to 5 times

per day (Gilbert 1997). For A. seniculus, one study in French Guiana estimated a

troop of 6 8 individuals defecated 1.5 kg fresh weight of feces per day, which

extrapolates to 55 g per ha per day (Juliott 1996b). The estimate for A. palliata from

Barro Colorado Island in Panama was 54 g (Gill 1991), and in Los Tuxtlas, Mexico, it

was 11 g (Estrada et al. 1991). In French Guiana, A. seniculus defecated seeds up to

40 mm; 49% of these were in the range of 10 15 mm (Feer 1999). Julliot (1996b)

found 60% of defecations occurred under sleeping sites, the rhythm of defecations

being bimodal. Seeds can be dispersed more than 550 m from the parent tree (mean

260 m; Julliot 1996b). Gilbert (1997) observed that A. seniculus near Manaus always

moved to a new location before defecating, and always after resting. The monkeys

would move from the resting site to a large horizontal branch with a mean height of

22.6 m above the ground, and look down before defecating. Additionally, defecations

occurred in areas relatively free of understory leaf surfaces. Gilbert suggests that this

behavior reduces the chance of encountering the ova or larvae of endoparasites in the

vertical range of the forest between 25 28 m, which is used by the monkeys for

foraging and traveling. A large amount of dung in one place may have implications for

secondary seed dispersal by dung beetles. The smaller diurnal roller beetles may not

bury all the dung, and much may be left over for the large crepuscular and nocturnal


Howler monkeys may not be as good dispersers as some other primates.

While 87.4% of seeds that pass through these monkeys germinated, compared to only

50.2% of control seeds, they often tend to pass most of the seeds underneath the parent

tree (de Figueiredo 1993). Spider and cebus monkeys eat more fruit than howlers,

travel more extensively, and may be better as dispersers (Chapman 1989; Howe 1980,

1986). Julliot (1997) found an increase in seedling abundance of preferred plant

species at sleeping sites; a pattern she suggests may increase seedling mortality through

density dependent factors, or alternatively, decrease it by overwhelming the effect of

predators and pathogens.

White-Bellied Spider Monkey (Ateles belzebuth)

Spider monkeys, weighing from 5.9 -10.4 kg, form large social groups of 20 -

40, but split up to forage in smaller size groups of 1 5. Home range size may be 150

- 250 ha (Symington-McFarland 1988). Spider monkeys forage in tall canopy trees

and are highly mobile. They move from tree to tree on medium sized branches and

rarely enter the lower part of the canopy or the understory (Van Roosmalen 1985;

Fleagle & Mittermeier 1980). Chapman (1989) found that spider monkeys spent

77.9% of their time eating fruit. Dispersed seeds fall into mature forest or maturing old

gaps rather than new gaps, which are often avoided by this species. Spider monkey

feces is loosely textured and widely spread (Zhang & Wang 1995; Van Roosmalen

1985; Forget & Sabatier 1997). Gut transit time averages 5.25 hrs (Lambert 1998;

Milton 1984), and seeds would typically be defecated as the monkeys forage through

the forest. In one area in French Guiana, the current distribution of adult Virola sp.

trees corresponds well with the seedling shadow along spider monkey arboreal transit

paths (Forget & Sabatier 1997).

Dung Beetles and Deforestation

Deforestation contributes to a loss of biological diversity. Several studies

throughout the New World tropics have shown significant decreases in dung beetle

diversity as rain forest is cut for crops or pasture (Estrada et al. 1998; Halffter et al.

1992; Klein 1989; Mor6n 1987). Howden & Nealis (1975) collected dung beetles in

both forested areas and clearings in Amazonas, Colombia. Only 6 species were

collected in clearings (4 of which were taken only in that habitat). On the other hand,

50 species were collected in the forest. The number of individuals trapped was also

nearly 10 times less in the clearings than in the forest. In Chiapas, Mexico, Mor6n

(1987) found four times the diversity of necrophilous Scarabaeinae in undisturbed rain

forest compared to that in the coffee-cacao plantations. However, the beetle fauna in

the plantations was closer to the original species assemblages than those found in cattle

pastures or row crops. Collection from matched pairs of traps in disturbed habitats

(pasture, roadsides, plantations) versus primary rain forest in Costa Rica also showed

significantly fewer dung beetles frequenting the disturbed habitats (Vulinec, unpub.).

Fragmentation of forested areas also affects dung beetle diversity. Klein (1989)

showed progressively fewer species in contiguous, 10 ha tract, I ha tract, and clear-cut

rain forest near Manaus, Brazil. Similar results were obtained on four sites in Mexican

rainforests with nearby derived pastures and secondary growth forest edges. Both

species richness and diversity indices declined with the intensity of deforestation.

Additionally, beetles were smaller in mean size (not attributable to lower food

availability), diumal species increased, and the guild structure changed; in clear-cut

areas, the dominance of burrowers as seen in intact forest decreased (Halffter et al.

1992). In reference to this last point, Halffter and Matthews (1966) hypothesized that

rolling originated in open areas as a response to the massive excrement of large

herbivorous Pleistocene mammals; on the other hand Chin and Gill (1996) suggest that

burrowing beetles date to the late Cretaceous. Additionally, in Los Tuxtlas, Mexico,

forest fragments held more species than disturbed habitats, followed by mixed and

cacao plantations, forest edges, live fences and coffee, citrus and allspice plantations.

Pastures held the fewest species. Dung beetle species richness and abundance was also

associated with the number of non-volant mammals recorded at the study sites

(Estrada et al. 1998).

Forest Regeneration and Seed Dispersal

In the last 20 years, 6 million ha of Amazon forests have been clear-cut and

turned into cattle pastures (Whitmore 1997). Much of this land, if subjected to heavy

use, particularly bulldozing, may never become reforested in the foreseeable future, but

instead remain shrubland. Even less degraded forests could take 500 years to

regenerate as opposed to pastures subjected to light use which may only require 100

years to regenerate (Buschbacher et al. 1992). Forest recovery in the Pard region of

Brazil is dependent on many factors other than soil nutrients. The most important

factors include: loss of on-site regeneration, slow seed dispersal, seed and seedling

predation, unfavorable microclimates, and soil compaction that discourage tree

establishment and survival (Buschbacher et al. 1992). According to several workers,

the most important factor in regeneration is getting seeds to deforested sites (Holl 1999;

Nepstad et al. 1991; Pannell 1989; Young et al. 1987).

Dispersers tend not to leave forests to enter cleared areas. For example, only 5

of the 150 bird species in Paragominas, Brazil, both consume fruits and move regularly

between forests and large clearings (Buschbacher et al. 1992). Additionally, rodents

are more prevalent in pasture than primary forest, thus seed predation increases in

disturbed areas (Nepstad et al. 1991; Schupp 1988), and regeneration is slowed.

Forest dung beetles do not forage in clear-cuts, even at distances as little as 10 meters

from the forest edge (Klein 1989; Vulinec, in press).

The potential for reforestation in secondary growth is much greater than is for

degraded pastures. Many seed dispersers readily enter secondary growth (Chapman

1987). Dung beetles are more abundant in secondary growth than pastures (Halffter et

al. 1992; Vulinec, in press). Even in land that was once pasture, trees that emerge

above the grass and shrubs increase the probability of other trees invading. Initially,

they serve as roosts for dispersers; then, as they shade out grass, there will be fewer

rodents and a better microclimate for other trees' establishment (Nepstad et al. 1991).

Dung beetles may prove important in secondary growth regeneration. In this

study, I examine which species enter these regrowth areas and what their value is as

secondary seed dispersers.


My study has two main objectives: (1) establish an index of dispersal abilities;

that is, what size of seeds a particular species of dung beetle will bury and how deep

each beetle species buries seeds, and (2) determine whether beetles high on the

disperser index are found in both primary, secondary growth, and clear-cuts, and what

factors might influence their distributions in these three habitats. This study provides a

survey of the dung beetle and monkey fauna in three localities of the Amazon Basin,

Brazil. Given the community of monkeys, community of dung beetles, and the type of

disturbance of the habitat, some predictions can be made about the potential soil seed

bank in a specific area. The study also contributes new knowledge about the natural

history of several species of dung beetles. Most importantly, this work provides some

recommendations for management of primary and secondary forests in tropical


Seed Dispersing Abilities

A particular dung beetle species has characteristics that make it an effective seed

disperser (i. e. burying a large amount and variety of seeds and burying them at optimal

germination depths), just as some monkey species are better than others in this respect

(Garber 1986; Chapman 1989). A combination of many factors are probably

important in determining whether a species is a good disperser. For example, one

would expect a large dung beetle that makes large brood balls to be a good disperser,

because many more seeds would be contained within the ball than in smaller balls from

smaller species. However, many of the largest dung beetles are also the least abundant;

furthermore, large beetles also tend to bury brood balls deep underground (sometimes

over 1 m), where germination of most seeds would be impossible (Halffter &

Edmonds 1982). Some factors that may be important in determining dispersal ability

include: abundance, size, depth of dung burial, male-female cooperation in nest

building (which can increase the amount of dung buried; Cook 1991), distance from

the source (seeds removed from the parent tree have higher survival; Janzen 1970),

nest structure (simple single burrows, or compound branched burrows; Halffter &

Edmonds 1982), the manner in which the dung is processed (do beetles remove

seeds?), whether beetles are nocturnal or diurnal (most Neotropical monkeys are

diurnal), whether the beetles follow monkey troops (beetles are suspected of following

mammals (Gill 1991)), if the beetles make feeding burrows in addition to nesting

burrows. To determine beetle burial abilities, I conducted a series of observations,

collections, and experiments to determine which beetles are most important for seed

dispersal, and from these data, I constructed an index of dispersal abilities, and

developed a guild score for each species.


Species Composition Between Primary, Secondary Forests, and Clear-cuts.

Collections were made in Rondonia, Amazonas, and Pard in primary forest,

secondary growth forest, and clear cuts. I surveyed all three habitats at least 15 times at

all three localities. From these collections, species richness and abundance measures

were used to differentiate sites and habitats within sites. Correspondence analysis was

used to delineate communities and the species that best describe them. Further, I

compared the communities of species at each site and each habitat within sites and the

communities in niche space (guild score) at each site and each habitat within sites.

Comparing community structure and correlations between environmental factors may

yield some recommendations for management practices. For example, is a more

efficient suite of dung beetles, in terms of their disperser index, attracted to secondary

growth areas of lower soil density than secondary growth with high soil density, or to

natural regrowth or banana plantations?


Study Sites

Overview of the Amazon Basin

Modem Amazonia falls between 5N and 10'S, and occupies an area of 6

million km2, more than half within Brazil (Pires & Prance 1985). Day length varies

little during the year, and solar energy at the soil level in the central Amazon varies

between 300 and 600 cal. cm2 per day (Salati 1985). Of the possible habitats in the

Amazon basin (including vlrzea white water flooded forest, igap6 black water

flooded forest, and dry forest), I examined only areas of terra firma (mata densa),

classified as Tropical Moist Forest under the Holdridge life-zone system (Croat 1978),

and ecosystems derived from that habitat. The majority of dung beetles and the highest

species diversity of seed dispersing mammals in the Neotropics occur in this habitat

(Peres 1997; Gill 1991; Halffter & Matthews 1966). The three localities of my study

each were separated by around 800 km linearly (Figure 1), and are referred to

throughout the text as: Rond6nia, Reserva Ducke (the site in Amazonas), and CaxiuanA

(the site in Para). I designated abbreviations for habitats as: R=primary rainforest,

S=secondary growth or disturbed forest, P=clear-cut. Trap stations within each habitat

site are given numbers along with the letter of the habitat, e.g. RI, R2, etc. Qualitative

site characteristics were compiled in the same manner as Peck and Forsyth (1982), and

are presented in Table 1.


The Rondonia site is private land located near Cacaulandia, and encompasses a

cattle ranch, primary forest, overgrown cacau and banana plantation, and natural

secondary growth called capoeira or, more specifically, babaquzal (the dominant tree

being the babacu palm Orbignya barbosiana). The total holding comprises over 2000

ha. The majority of collecting was done in a 250 ha remnant terra firma forest

contiguous with the secondary areas mentioned above (Figure 2). Mean annual

rainfall is 2289 mm (range 1768 2832), and mean annual temperature is 27.50C

(range 23.57 30.93). Relative humidity averages 73.5% (Emmel & Austin 1990).

Rondonia was undeveloped until the late 1960's, when the construction of roads and

government incentives opened the area. It now has one of the highest rates of

deforestation in the world (Stone et al. 1991). By 1991, over 37, 000 km2 were

cleared, more than 16% of the total state (Stone et al. 1991; Fearnside 1993). Despite

the rapid deforestation, biological surveys have estimated this area to have one of the

most diverse butterfly faunas in the world (possibly as many as 1800 species) (Emmel

& Austin 1990). In addition to monkeys, collared peccaries were common, and deer,

large tortoises, tayra, jagarundi, agouti, and some large birds, such as curassows were

often seen.

The primary forest at this site is relatively intact, although some selective

logging was done in the past. During my stay, surrounding land-holdings were being

cut. The secondary growth is overgrown cacao-banana at S1 (see map), natural

secondary regrowth from pasture dominated by babaqu palm at S2, and similar but

more open habitat at S3.

The clear-cut areas are cattle pasture, mostly planted in an exotic African grass,

with isolated trees along fence rows.

Reserva Ducke

The city of Manaus is located in the state of Amazonas at the confluence of the

Rio Negro and Rio Solimoes (Amazon). The biological station Reserva Adolfo Ducke

is 25 km northeast of the city (Figure 3), and is surrounded by land that is rapidly being

developed. Reserva Ducke has 10,000 ha with extensive trails and a 45 m rainforest

tower. Poaching is common, and most of the large mammals have been extirpated.

While I was there, a worker (matero) reported seeing a jaguar. Agoutis were common,

but no tapir or peccaries have been seen for some time (personal communication with

guards and materos). Monkeys are the most common large mammal, but are also

hunted regularly within the reserve by people in the surrounding communities. As

these communities often have a higher standard of living than many parts of the

Amazon, this hunting pressure is particularly unfortunate. Dead howlers that had been

shot and escaped to die later were found during my time at this site.

Rainfall in the area averages 2200 annually, with a dry season from July to

September (Anon. 1978). The secondary growth was cleared and replanted in timber

trees (not all native) 20 years ago; some has grown back to native secondary growth

(see Table 1).

The secondary growth is land that had been cleared and replanted in timber

species, not all native. No extraction currently occurs. This habitat harbors a high

canopy with little understory, but thick ground cover. S2 is very much like primary

forest, and Sl is used as a short-cut by local residents. The clear-cut areas are: S3

is a 10 ha soccer field near the station's buildings, S2 is a 10 ha area cleared for the

meteorological station (slightly upland), and S1 bordered the highway outside the

station and was near domestic animals, including cows.


The recently created (established 1990) Ferreira Penna Scientific Research

Station is located near the community of Caxiuana, Para, off the Rio Para on the

southwest side of the Ilha do Maraj6. This station includes 33,000 ha, 80% of which is

terra firma forest, and 20% is floodplain forest. Trails are cut in a 50 meter square grid

pattern (Figure 4). Surrounding communities are sparsely populated, and hunting

pressure is low (Lisboa 1997). Monkeys, especially howlers and the silvery

marmosets, have been acclimated to humans by previous research. Pacas and agoutis

are common; I regularly saw coati and deer: occasionally tayra and grey-winged


I used the 5 ha patch of cleared land which housed the station's meteorological

equipment as my clear-cut site. No other clear-cuts were available, so only one

collection point was possible. Additionally, there was little secondary growth, therefore

disturbed habitat was used as secondary growth: S l=a treefall gap, S2=the edge of the

primary forest, and S3=the edge of the road leading to the river.



I censused beetle species composition and abundance in three habitat types at

each research site: primary rain forest, secondary growth (edges of primary forest, old

plantations, and previously cleared areas), and clear-cut areas. I used baited pitfall traps

set along the same route as the monkey transect. Figures 2, 3, and 4 show the trap

stations (a set of three traps set 20 m apart, and nearly 1 km from other trapping

stations), monkeys transect paths, and general habitat maps of the three study sites.

Beetle traps were I liter plastic cups buried to the rim in soil, and topped with a

3 cm aperture funnel. A 50 ml cup for bait was suspended by wire above the trap. The

census traps, and all others except where noted, were baited with 25 ml human dung.

Human dung has been shown to be as attractive to beetles as monkey dung (Howden

& Nealis 1975), although the majority of dung beetles are generalists (Gill 1991). The

volume of dung in the traps attracted even the largest beetles in the area (see Peck &

Howden 1984). I set a total of 27 traps each census, which were left up for 24 hrs,

then collected. Three traps were set at 20 m intervals at 9 different trapping stations (3

sites in primary forest, 3 in secondary growth, and 3 in clear-cut). This trapping

regime yielded a total of 1242 trap days (sensu Klein 1989) for the three locations.

Each trapping station was as distant from the others as possible, typically 1 km, to

maximize independence (Hurlbert 1984).

Data collected during these surveys included: habitat, beetle species, abundance,

sex (in some species), air temperature, rainfall at all three locations, and relative soil

density at two of the locations.

Beetles were identified by me, Bruce Gill of the Entomology Unit, Canadian

Food Inspection Agency, Dave Edmonds of California State Polytechnic University,

and Fernando Vaz de Mello of Viqosa Universidade. Beetles identified by me were

compared to collections at INPA in Manaus, Brazil, and the Florida State Collection of

Arthropods, much of whose material was from Bert Klein's collection (1989), and

identified by him and Robert Woodruff of the Florida State Collection of Arthropods,

or through the keys of Edmonds (1994), Jessop (1985), Halffter & Martinez (1962).

Beetle biomass was measured as dry weight; the beetles were dried at room

temperature for 1 week prior to weighing. Selected species, those most common or

largest, were weighed on an Ohaus balance at the USDA-ARS laboratory in

Gainesville, Florida. Mean biomass of each species was calculated as the mean of 10

individuals. In a few cases, only 3-6 individuals were weighed. For beetle species that

were not weighed, biomass was assumed to be similar to beetles of the same size, and

the weights of those species were used. This extrapolation was necessary only with

small, less common beetles.

Beetle abundance at all three sites was plotted against time, and the abundance

differences between primary and secondary growth were tested with a Student's paired

sample t-test. The differences among mean beetle abundance in various types of

secondary growth were compared with a one-way ANOVA.

Species richness among all three habitats (primary forest, secondary growth,

and clear-cuts) within sites was examined for significance using ANOVA, and

significant differences among the habitats were isolated using a Waller-Duncan K-ratio

T test.

I performed a correspondence analysis to cluster similar beetle communities

among sites and habitats, and to determine the species that influence the clustering

(Legendre & Legendre 1998; Terborgh & Andresen 1999; Jongman et al. 1995;

Manly 1995; Klein 1989). I also examined the abundance of certain common species

in secondary and primary forest using a t-test on abundance.


In order to determine if beetle abundance and diversity correlated with monkey

abundance, I censused monkey populations at the three sites: 36 times in Rondonia and

Ducke, and 27 times in Caxiuani. Monkey transects were stratified by the proportion

of secondary and primary growth at each site. These transects were walked by 2

people, and transect pace was about I km/hr. The following data were collected:

monkey species, number of individuals (seen or heard), time, location, observer

distance, height at first sighting, sighting angle. This last measurement allows

calculation of the perpendicular distance of the first sighting from the transect. Transect

summaries are presented in Table 2.

I used methods described in Chapman et al. (in press) to calculate monkey

density and relative abundance. Density was calculated using a 50% cut-off rule to

select the sighting distance. If Xi is the number of sightings in distance class i for a

given species, the last distance used in the analysis is at the end of the first class such

that Xi+1/Xi and Xi+2/Xi were both equal to 0.50 or less. This method compensates

for intraobserver differences, as sightings at the farthest distances are excluded from the

census if they occur infrequently (if 2 distance classes are less that 50% of the next

closer distance class). Observer to animal distance, as opposed to perpendicular

distance was used, as perpendicular distance underestimates transect width for forest

primates (Chapman et al. 1988). Densities of groups were then calculated as the

number of groups sighted within the truncated sighting distance divided by the area

sampled, i.e. the length of the transect times the truncated distance (Chapman et al., in

press). Additionally, relative abundance was calculated as a second density measure to

determine agreement between estimates, as interobserver variation may occur

(Chapman et al., in press). Relative abundance was calculated as the number of

groups seen per kilometer walked, and relative densities were calculated for groups per

square kilometer and number of individuals per square kilometer.

The proportion of monkeys that are seed predators as opposed to non-predators

was evaluated using a Chi-square 2x2 contingency table test for differences among the

three sites.

Seed Burial Experiments

Because the number of dung beetle species was high at each site (> 50), I

limited the number of species included in the experiments. Some species were

eliminated immediately (e. g., all endocoprids, as they would not contribute to seed

burial and are usually small). Other species were chosen by their size, abundance, and

known or expected behavioral characteristics.

These experiments determined how deep particular species of dung beetles bury

seeds artificially imbedded in dung. For most of the trials, I used five different sizes

and shapes of seeds: (1) small round; 3 mm diameter, (2) small oval; 7 x 2 mm, (3)

large flat oval (squash seeds), (4) large round; 10 mm diameter, (5) large round; 15

mm diameter. The "seeds" were beads and some natural seeds. I generally used 10

each of #1 and #2 type seeds, 5 of #3, 3 of #4, and 2 of #5. Only a few trials with #3

were conducted because the seeds disintegrated quickly. In the final analysis, seeds

were divided into small (# 1 and # 2) and large (#4 and # 5).

In general, experiments were conducted as follows. I placed beetles in buckets

(40 cm diameter x 36 cm depth) filled with 150 mm sandy soil. Fifty cc of fresh cow

dung with embedded seeds was placed in the center of each bucket. I covered the

buckets with screen mesh covers, and left them up for 72 hrs. The number of

individuals of each species varied depending on the size of the beetles. Beetle numbers

were varied to keep biomass of beetles burying 50 cc of dung approximately equal.

Sample sizes varied depending on the beetles captured alive.

My procedure varied for smaller species (e.g. Syballocanthon sp.). These

beetles were placed in 1-liter containers with 100 mm sandy soil, and were presented

with 10 cc fresh cow dung containing only small seeds.

After 72 hrs., I excavated the burrows, 10 mm at a time. I recorded depth of

each seed, and whether it was in a brood ball or a food ball. I also recorded depth of

beetles. In some cases, when beetles attempted to disperse before the 72 hours were

over, I recorded the time of day they flew, to determine foraging activity patterns.

To construct an index of seed dispersal abilities, a principal components

analysis was performed on the beetle species and their following characteristics:

average biomass, standard deviation of biomass, percentage small seeds buried,

percentage large seeds buried, average depth of burial of small seeds, average depth of

burial of large seeds. Principal component analyses allow a large dataset of highly

correlated variables to be reduced to a smaller number of uncorrelated variables, called

the principal components, which are then arranged by those that explain the majority of

the variance (Legendre & Legendre 1998; Birch 1997; Bookstein 1991). Based on

their PCA scores, beetle species were classified into guild categories that reflect both

their abilities to bury seeds at optimal depths for germination, that is, less than 5 cm for

small seeds, and less than 10 cm for large seeds (Shepherd & Chapman 1998), and

their biomass mean and variance. This index was substituted for beetle species in a

correspondence analysis (Legendre & Legendre 1998) and compared to the plot of

habitats and communities to illustrate differences between community species

partitioning and guild structure partitioning in the different habitats.

To determine the effect of disturbance on beetle guilds, habitats were divided

into 5 levels of disturbance, from least to most, I and 2 in primary forest, and 3, 4, 5 in

secondary growth (see Table 1).


Beetle Surveys

Habitat characteristics are presented in Table 1. From all three sites, a total of

14,505 beetles from 88 species were collected (Appendix 1). Both beetle species

numbers and overall abundance differed significantly among the three sites, comparing

total number of species and total numbers at each site (ANOVA; species:

F(2,43)=79.26, P<0.001; abundance: F(2,43)=41.94, P<0.001). Abundance also

varied significantly among habitats within each of the three sites, using paired trap days

as the comparison (Rondonia: F(2,44)=23.18, P<0.001, Figure 5; Ducke:

F(2,40)=6.06, P=0.005, Figure 6; Caxiuana: F(2,42)=27.14, P<0.001, Figure 7).

Secondary growth, which yielded a subset of the same species found in primary

growth, had lower abundance of beetles at all three sites compared to primary growth

(Rond6nia: t(15)=5.68, P<0.001; Ducke: t(14)=3.82, P=0.0009; Caxiuana: t(14)=3.79,

P=0.0009). At Reserva Ducke, the only site that had clear-cuts of varying ages,

numbers of species in clear-cuts correlated with amount of time since an area was

cleared (r=.962, df=2, p<0.025). There were also significant differences in beetle

abundance between secondary growth habitats of differing amounts of disturbance at

two of the sites (Rond6nia: F(2,42)=1.36, P=0.26 ns; Ducke: F(2,42)=4.15, P=0.023;

Caxiuanl: F(2,42)=5.42, P-0.008). At Ducke and Caxiuana, the more disturbed the

habitat, the fewer beetles were caught. Some species occurred in both secondary

growth and primary forest in similar abundances, but others appeared more restricted

(Table 3).

Species richness was significantly different among all habitats within sites

(Table 4). However, habitats among sites were different only between Rondonia's

primary and secondary forest and those habitats at the other sites (Table 5). Species

richness was not significantly different between Ducke and Caxiuana, and only

Caxiuana differed significantly from the other two sites in the species in clear-cuts; not

unexpected as only two beeltes were caught in clear-cut at Caxiuana.

Biomass of individual beetles is presented in Table 6. Biomass varied widely

over time and among sites (Figures 8, 9, 10). Biomass did not differ significantly

between primary forest and secondary growth in Rond6nia (t(15)=0.22, P=0.41), or at

Reserva Ducke (t(14)=0.18, P=0.56). Biomass was significantly greater in primary

growth than secondary growth only at Caxiuana (t(14)=0.35, P=0.05).

Monkey Censuses

Monkey densities varied significantly among the three sites (F(2, 5)=4.5,

P<0.01; Table 7). Total densities for all species at each site were 28.42 individuals per

square kilometer for Rond6nia, 8.94 for Reserva Ducke, and 38.55 for Caxiuana.

Caxiuani had higher density of monkeys than Rond6nia, although fewer number of

species. Reserva Ducke had low densities of all species, and the fewest species, which

may be due to its proximity to a major metropolitan area.

Monkeys differ in their abilities to disperse seeds, and as with a dung beetle

community, the proportion of good seed dispersers may be quite different from site to

site. From my estimates of monkey densities, I calculated the proportion of monkeys

that are seed predators as opposed to seed dispersers at each site (Figure 11), and found

the difference between sites to be significant (Chi-squares: Caxiuana vs. Ducke =44.7,

p<0.001; Caxiuana vs. Rond6nia =10.4, p<0.005; Ducke vs. Rond6nia =101.0,

p<0.001). The monkey fauna at Reserva Ducke was composed of almost 50% seed


Monkey and Beetle Comparisons

Beetle abundance varied widely among the three sites, but did not appear to be

related to monkey density. Beetle abundance was much higher at the Rond6nia than

the other two sites, but monkey densities were essentially the same between Caxiuana

and Rondonia.

Beetles and Seed Burial

Many factors may influence a beetle species' ability to bury seeds. Some taxa,

such as Aphodius spp., do not bury seeds at all. Individual species may have

characteristics that determine their quality as seed dispersers, for example, size,

abundance, interring a large amount of dung under a dung pile, or rolling small balls of

dung away from a dung pat. Over 20 species of beetles were tested for their ability to

disperse seeds (Table 8). These beetles were the most abundant at all three locations,

and were typically among the larger species, as these beetles would be expected to bury

more dung, and therefore more seeds. One of the critical questions I addressed was

whether beetles buried seeds too deeply for germination. Most seeds will not

germinate at depths greater than 10 cm, and for smaller seeds the germination threshold

may be less than 5 cm (Shepherd & Chapman 1998). Table 8 shows average depth of

burial for small and large seeds of beetles tested. Figures 12 and 13 show proportions

of small (:< 5 mm) and large seeds (> 10 mm) left on the surface by different beetle

species for which three or more trials were performed. Beetles differed in the

proportion of seeds that were taken underground (F(2, 19)=3.6, P<0.05). Waller-

Duncan multiple comparisons showed Coprophanaeus lancifer, Oxysternon

conspicillatum, and severalDichotomius species to be most efficient at getting seeds


Guild Analysis

In order to analyze beetle community structure in relation to seed burial, beetle

species were differentiated into guilds that define their efficiency at seed burial. Guild

partitioning was accomplished by principal component analysis using the following

variables: biomass, variance in biomass, percent of small seeds buried, percent of large

seeds buried, average depth of burial of small seeds, and average depth of burial of

large seeds. One species, Coprophanaeus lancifer, was an extreme outlier, due to its

very large size (Table 6), and was excluded from the PCA analysis (Haas & Tolley

1998), although was later given a guild score (6) based on its large size and efficient

seed burial behavior. Eigenvalues (percent of variance explained) of the correlation

matrix and the scoring coefficients for each character are presented in Table 9. The first

two factors explained 94% of the variance. The loadings on the first principal

component (factor 1) are all positive and relatively equal; 74% of the variance is

explained by this factor, which can be interpreted to characterize absolute size. The

second principal component, which contains more information about seed burial

characters, correlated most highly with biomass, and negatively with all burial

characters, especially the effect on small seeds. From the principal component analysis

scatterplot of the first two factors (Figure 14), I grouped species into 7 categories, and

assigned a score to each, for a total of 6 guilds (Table 10). One beetle, Dichotomius

boreus, although it buried almost all seeds, was placed in guild 3 due to the extreme

depth of burial of both seed sizes (averages of 9 cm for small seeds, and 8 cm for large

seeds) which limits its effectiveness as an efficient seed burier. In general, members of

the same genus and relative size had similar PCA scores, therefore, beetles that were

not tested in burial experiments were given the same guild scores as their congeners.

Guilds were constructed to mirror effectiveness at seed burial, given all variables used

in the PCA. Guilds were ranks of effectiveness from I to 6, with guild 1 being

inefficient or poor seed buriers, to guild 6 which contained beetles that buried a high

proportion of seeds at depths that theoretically would allow a high germination rate.

Natural History Observations

The following accounts illustrate some of the differences in behavior and

natural history of the more abundant dung beetles used in the guild and community


Coprophanaeus lancifer is one of the most conspicuous and ubiquitous

species. It is also one of the largest dung beetles in the Amazon Basin (in my

collections, exceeded only by Sulcophanaeusfaunus), and is an iridescent blue-violet

color. This beetle flies consistently at dawn and dusk (Vulinec, in press). I have found

it in Reserva Ducke at dusk flying on trails, and rarely at lights at the field station.

While the literature states that it is primarily or wholly necrophagous (Edmonds 1972;

Halffter & Matthews 1966; Gill 1991), I collected it in considerable numbers at human

dung, at carrion (chicken) traps, and have found it in the forest on horse dung, rotting

bananas, and a dying two-toed sloth. It is a very active and hardy beetle, living for over

a month in captivity on cow dung. C. lancifer buries seeds of all sizes tested, but

sometimes buries them deeper than 10 cm (Table 8). Additionally, while it buries

some seeds too deeply, it appears to remove others on the way down its burrow, thus

depositing seeds in a horizon between 1 15 cm. Because this beetle grabs a large

chunk of dung (as much as 50 cc), and pulls the entire piece into the burrow, it is

potentially a very important seed burier. As it also forages in the evening and early

morning, it would be active during the times monkeys are in sleeping trees, and

considerable dung is being deposited in one place.

Coprophanaeus jasius, which occasionally came to dung baited traps, was

more common in carrion baited traps. This species would not eat cow dung while in

captivity, nor did it bury seeds well (Table 8).

One of the most conspicuous beetles in Rond6nia is the metallic green and

black Oxysternon conspicillatum. This beetle was also collected in Reserva Ducke,

but was less abundant. It was not collected in CaxiuanI, although the congener 0.

selenum was caught. They are very active and strong fliers. Peck and Forsyth (1982)

recorded one specimen as flying 50 m in under two minutes, and another I km after

two days. This species would often fly to dung traps as they were being set during the

day. They often flew down trails, but then would circle into undergrowth on the edge

of the trail and perch for several minutes before approaching the ground. This behavior

may be a defense against predation from whatever animal deposited the dung; large day

flying beetles such as 0. conspicillatum may be particularly vulnerable to monkey

predation. They bury dung very quickly, often working in male-female pairs. A male

and female were put together in a plastic container with 50 cc cow dung. The male

pursued the female for several minutes, and then mated; mating lasting approximately

I minute. Immediately after, both began to dig holes below the dung pat and lowered

the entire pat into the hole. The female then sectioned the dung and formed balls

(approximately 2 cm in diameter), one of which she pushed out of the hole onto the

surface the soil. She then dug a deeper hole to the bottom of the container (10 cm).

Fifteen minutes later, she emerged and pushed the surface ball into this burrow and

buried it. 0. conspicillatum is one of the most effective seed buriers found.

Eurysternus is a very common genus, especially in Rond6nia. While species

of this genus have particularly complex nesting behavior, including nuptial feasts, and

brood ball abandonment (Halffter and Edmonds 1982), they are not effective seed

buriers. The most common species, E. caribaeus, was one of the dominant

components of the dung beetle biomass in Rond6nia, and contributed a significant

proportion of the biomass in Amazonas. I tested this genus three times in the seed

experiments, and out of a total of 50 seeds presented, it only buried 2.

Dichotomius is one of the most common beetles, and has the greatest number

of species of all genera in the Amazon. It also contains some of the larger beetles in the

New World. The taxonomy of this group requires revision (Howden & Young 1981).

Species are mostly nocturnal (Peck & Forsyth 1982; Halffter & Matthews 1966), and

differ in their seed burial abilities; one species consistently buried seeds too deeply for

germination (Table 8). They typically dig deep single burrows for both feeding and

nesting (Halffter & Edmonds 1982). Although generally dung feeders, I have collected

some species of this genus in rotting fruit. They were also common in horse feces on

the forest trails at the Rondonia site.

While most Canthon species are small, diurnal rollers, C. aequinoctialis are

large (Table 6), and nocturnal. Howden and Young (1981) found them to be most

active between 1700 to 2100 hrs. While they would seem to be able to bury a large ball

of dung, my experiments showed they buried very few seeds, even fewer than other

smaller species of Canthon. In fact, all species of this genus and related genera tested

were poor seed dispersers, even though they may transport dung up to 5 m away from

the deposition site. Two individuals of the diurnal Scybalocanthon pygidalis were

observed removing and eating dung off seeds before leaving them on the soil surface

while rolling their ball away. This species was one of the first to arrive at dung in the

primary forest during the day.

Of the four Phanaeus species caught, only two, P. chalcomelas and P.

cambiforti were caught in large enough numbers to test for seed burial. These two

differed little in their behavior as seed buriers (Table 8), and had similar activity

patterns. In appearance, only close examination reveals that they are separate species,

and they often cannot be distinguished in some of the minor males. P. alvarengai is a

rare beetle, discovered in 1984, and described from two specimens, thought to be

endemic to Rond6nia and nearby Bolivia (Edmonds 1994; Amaud 1984). I caught 15

at the Rondonia site, and one female several hundred kilometers away in Caxiuana.

One live female in Rondinia was tested with seeds and showed similar burial patterns

to its congeners, although its activity period lasted for only 10 minutes around 0630

hrs, as opposed to the mid-day, extended activity period shown by P. cambiforti and

P. chalcomelas.

The small-sized Canthidium do not bury seeds very deeply (1 2 cm; Jones

1994; Gill pers. comm.).

Some large rollers, such as Telhybomia orbiculare and Deltochilum

pseudicarus tested as very poor seed dispersers. Seeds were often left on the surface,

possibly being removed during ball-making, although this behavior was not observed.

Community Analysis

I performed a correspondence analysis with species and abundance at each site,

subdividing the sites by habitats where the trapping stations were located. Clear-cut

trapping stations were excluded because they had a completely different complement of

species, and caused the analysis of other sites to clump in one cluster, making

differentiation impossible. Each site had six trapping stations, three in primary forest

and three in secondary growth. Figures 15, 16, 17, 18, 19, and 20 show the

correspondence analysis of species abundance and habitats at the three sites for groups

of species with distinct behavioral characteristics. These separate graphs are necessary

because one graph of all species at a site becomes too complicated to view, and

divisions into burrowers versus rollers (Figures 15, 17, 19), and nocturnal versus

diurnal beetles (Figures 16, 18, 20) make the data more comprehensible. However,

since dividing beetle guilds into only burrowers versus rollers obscures some real

differences in their seed dispersal activities; a more complex subdivision of guilds is

done in the second correspondence analysis (see below).

Distances between points are calculated as Chi-square distances based on the

contingency table of abundances of each species at each station, the points themselves

are the centroids of the reciprocal averages of species among stations and stations

among species (Legendre & Legendre 1998). These distances represent the abundance

of a species at a given station (proximity of the species centroid to the station), the

defining species of a given station (those species that cluster with the station), and how

different each station is in terms of its species and their abundances (the distance

between stations). Axes dimension I and dimension 2 are those that explain the

greatest amount of variance in the data (87.7% in Rond6nia; 94.8% in Ducke; 95.3% in

Caxiuana). Further dimensions are not displayed.

In Rondonia, more nocturnal beetles than diurnal beetles are caught in

secondary growth (Figure 16). There were some exceptions, as rare species clustered

at the far end of the primary forest on the graph. Generally, diurnal beetles tended to

cluster around the three primary growth trapping stations. Rare species were found on

the periphery of the graph, for example, Canthon sordidus. Burrowers were more

evenly distributed between primary and secondary sites than rollers (Figure 16). The

majority of rollers were found in primary growth, although some species seemed to

prefer secondary growth, for example, Deltochilum septemstriatum was most

abundant at site Sl, the most disturbed station, but also the secondary-growth station

with the greatest monkey sightings.

A different situation occurred at Ducke in Amazonas. One primary forest

station (RI) was clearly separate from the others. All the secondary sites and R2 and

R3 clustered near each other. RI was in the occasionally flooded lowlands, and

contained unusual numbers of two species of Deltochilum, relatively large nocturnal

rollers. Diurnal beetles were less common in all the secondary stations than in primary

forest, while nocturnal beetles were more or less equally abundant in all the stations

(Figure 18). Again, rollers clustered more towards the left side (primary upland

habitats), and the center (i.e. they were less abundant in the more extreme secondary

sites) than burrowers (Figure 17).

CaxiuanA also had most of the primary stations and secondary stations

clustering together. The most divergent station was S3; it was also the most disturbed,

being located along a road built to move passengers and boats from the river. There

was a wide spread among the abundances of many nocturnal beetles, indicating a large

amount of overlap in their abundances among the habitats, while the majority of diurnal

beetles clustered at R2, the station with the most monkey sightings (Figure 20).

Rollers were somewhat more restricted to primary growth than burrowers (Figure 19).

The second part of the analysis begins first with a categorization of beetle

species into various "seed dispersal guilds", followed by a correspondence analysis

with beetle seed dispersal guild score and abundance at the various stations within sites.

Guild score, discussed earlier, from the principal components analysis (Table 8), was

substituted for species and multiplied by the abundance of all members of that guild

that occurred at a station. I plotted the recalculated guild scores (weighted by

abundance) and trapping stations at each site (Figures 21, 22, 23). Again, only the first

two axes were used. In all cases, guild 5 clustered closest to a disturbed site. In

Rond6nia, guild 4 was at the periphery of the plot, indicating that the whole area was

poor in this guild, but when it did occur, it was in the river floodplain habitat of the

primary forest. Guild 4 was not abundant at any site.

Guild distances describe the importance of that guild in seed dispersal times its

abundance at a station. The average guild distances from each trapping station were

plotted against the stations ranked as a continuum of habitat disturbance (Figure 24).

The greater the guild distance on the y-axis, the less abundant that guild is in each

habitat rank. Each guild was correlated with habitat disturbance using Spearman's rank

correlation coefficients (Table 11). Guild I distance was significant with increasing

habitat disturbance. As correspondence analysis distance increases from a trapping

station, the abundance decreases, therefore a positive correlation in this test indicates a

decrease in abundance of that guild with increasing habitat disturbance. Guild 2

approached significance. Guild 3 and guild 4 were not significant. Guild 5 was

negatively correlated with habitat disturbance (that is, more abundant in more disturbed

habitats). Guild 6 distance was significantly correlated to disturbance. Guilds 3 and 5

also showed significant quadratic regressions, guild 5 displaying a pattern of greatest

abundance in moderately disturbed habitats. Habitats 1 and 2 (the least disturbed), both

in primary forest, had a high abundance of guilds 1, 2, 3, and 6, but low abundances of

guilds 4 and 5. More disturbed habitats had relatively large numbers of guilds 1, 2, 3,

and 6, but also show an increase in guild 5. The most disturbed stations at all three


sites (disturbance level 5) had the lowest abundance of all guilds, but were especially

low in guild 6 species, the most efficient seed buriers.

Table 1. Habitat characteristics at sites and trapping stations.

Trap station RI R2 R3 S1 S2 S3

Elevation upland upland upland upland upland upland
Canopy dense dense dense sparse medium sparse
Understory dense dense sparse dense dense sparse
Ground cover sparse dense sparse dense dense dense
Vegetation type md md md cac/ban capoiera capoiera
Disturbance slight low low high slight moderate
Soil clay clay clay/grv clay/grv clay/grv clay/grv

Trap station RI R2 R3 S1 S2 S3

Elevation lowland upland upland upland upland upland
Canopy dense dense dense medium medium medium
Understory dense dense dense dense dense sparse
Ground cover dense medium sparse dense dense sparse
Vegetation type md md md plant plant plant
Disturbance low low slight moderate slight high
Soil sand clay clay clay clay clay

Trap station RI R2 R3 Sl S2 S3

Elevation upland upland upland upland upland lowland
Canopy dense med dense sparse medium sparse
Understory dense dense dense sparse dense dense
Ground cover medium sparse dense dense dense dense
Vegetation type md md md gap edge edge
Disturbance low slight slight sight moderate high
Soil clay clay clay clay clay/grv clay/grv

Upland=above flood level; Lowland=below flood level; Dense=cover almost entire;
Medium-cover less than entire but not sparse; Sparse=light coverage. Vegetation types:
md=mata densa (terra firma; unflooded upland relatively undisturbed climax rainforest);
cac/ban=overgrown cacao/banana plantation; capoiera=secondary natural growth (in
Rondonia, dominated by Babaqu palm); Planta=old timber plantation; Gap=natural tree
fall gap (0.5 ha); Edge=edges of primary forest bordered by clear-cuts. Disturbance
(primarily anthropogenic): low < slight < moderate < high. Soil: grv=gravel.

Table 2. Characteristics of monkey transects.

Site Transect Length Number of Transects Kilometers Censused
Rond6nia 6 36 216
Reserva Ducke 7 36 252
Caxiuand 3 27 81


Table 3. Abundance of some beetle species in primary (R) and secondary (S) growth.

Species mean R sd R mean S var S p-value

Canthon aequinoctialis 194.6 202.45 52.6 57.72 0.01
Dichotomiusbatesi 212.3 218.3 114.3 125.66 0.08
Dichotomiuslucasi 51.33 53.9 33.66 35 0.13
Dichotomiuspodalirius 15.6 15.8 16.6 16.85 0.11
Canthon oprl 47.3 51.05 10 10.1 0.06
Dichotomiuscarinatus 21.6 21.7 16.6 17.19 0.08
Deltochilum septemstriatum 8 8.29 7 8.23 0.42
Eurysternuscaribaeus 374 379.18 246.6 269.67 0.18
Onthophagus bidentatus 190 192.25 110.3 115.18 0.05
Hansreia affinis 55.6 61.54 15.3 17.38 0.05
Phanaeus chalcomelas 33.3 34.02 18 18.87 0.11
Oxysternon conspicillatum 13 13.53 4.6 4.83 0.03


Table 4. Species richness among habitats (Primary, secondary, clear-cut) at all three
sites. Means sharing a letter are not significantly different.

Habitat Mean # of Species Standard Error
Primary Forest a 35.31 1.62
Secondary Growth b 27.44 2.03
Clear-cuts c4.81 0.93

Habitat Mean # of Species Standard Error
Primary Forest a 13.60 1.23
Secondary Growth b 9.87 0.85
Clear-cuts c 3.67 0.64

Habitat Mean # of Species Standard Error
Primary Forest a 11.47 1.03
Secondary Growth b 8.46 0.87
Clear-cuts c 0.07 0.07


Table 5. Species richness among sites (Rondonia, Ducke, and Caxiuana) in all three
habitats. Means sharing a letter are not significantly different.

Site Mean # of Species Standard Error
Rondonia a 35.31 1.62
Ducke b 13.60 1.23
Caxiuana b 11.47 1.03

Site Mean # of Species Standard Error
Rondonia a 27.44 2.03
Ducke b 9.87 0.85
Caxiuana b 8.47 0.88

Site Mean # of Species Standard Error
Rondonia a 4.81 0.93
Ducke a 3.67 0.63
Caxiuana b 0.07 0.07

Table 6. Beetle species average biomass (bio mean), variance in biomass (bio var).
Species abbreviations are found in Appendix 1.

Species bio mean bio var
A3sb 0.009 2.96E-05
a3sg 0.005 5.06E-06
a4br 0.009 2.96E-05
aconn 0.019 5.21E-05
aset 0.028 9.39E-05
c3opr 0.001 0.00000008
c3sb 0.001 8.66E-08
ca 0.129 0.00216445
can2spots 0.014 1.4216E-05
can7mb 0.014 1.4216E-05
canacut 0.009 1.7897E-05
canbrun 0.014 1.4216E-05
canlit 0.009 1.7897E-05
canmb 0.009 1.7897E-05
canquad 0.019 0
cansbr 0.013 4.6806E-05
cansep 0.037 5.7584E-05
cansord 0.027 3.18E-05
cantri 0.037 5.7584E-05
canvir 0.014 1.4216E-05
cdey 0.012 1.65E-05
cdor 0.025 0.00010
cger 0.025 0.00010
cj 0.618 0.03328842
cl 3.260 1.04222977
cp 0.618 0.03328842
cpin 0.034 0.00025418
csb 0.017 3.5096E-05
csplen 0.019 0.00011
d2knobs 0.363 0.01120278
dbat 0.073 4.61E-05
dbor 0.451 0.02630228
dear 0.437 0.01070775
dcarb 0.286 0.01019375
delamaz 0.172 0.00018769
delcar 0.181 0.00569541
delguy 0.086 0.00086438
delpse 0.676 0.00815155
delsep 0.059 0.00029308
dimit 0.567 0.0423451
diuc 0.099 0.00042
dmam 0.417 0.01865058
dmel 0.241 0.00447345
dpod 0.451 0.02630228
drob 0.133 0.00136649

dwor 0.412 0
ec 0.102 0.00081101
ecay 0.034 0.00012687
ef 0.156 0.0010041
eh 0.008 8.9525E-06
eham 0.156 0.0010041
ei 0.034 0.00012687
ep 0.034 0.00012
ev 0.15635 0.001
gs 0.011 1.22E-07
ha 0.037 5.75E-05
o2tub 0.797 0.059
oc 0.797 0.059
om 0.146 0.0018
on2knobs 0.016 3.85E-05
on4mbr 0.003 2.5E-07
onbi 0.009 1.78E-05
onhir 0.006 1.32E-06
onscp 0.020 0.0009
ontlOb 0.023 0.0002
os 0.797 0.059
pal 0.135 0.0018
pbi 0.135 0.0018
pcam 0.135 0.0018
pcanxan 0.009 1.7897E-05
pch 0.146 0.0018
scycanoprl 0.027 3.1812E-05
sp 0.027 3.18E-05
stri 0.011 1.22E-07
to 0.455 0.0014
uroxpyg 0.009 2.96E-05

Table 7. Mean group densities, mean group sizes, and mean number of individuals per
kilometer for each primate species in the three locales.



Group Density Average group
(groups/km2) size

# individuals/km2

Ateles paniscus 0.43 2.00 0.86
Cebus apella 3.15 3.91 12.32
Callicebus moloch 0.62 2.75 1.71
Pithecia monachus 0.58 3.00 1.74
Saguinus fusicollis 1.23 5.75 7.07
Saimiri sciureus 0.62 9.00 5.58

Reserva Ducke

Species Group Density Average group # individuals/km2



Alouatta seniculus 0.40 5.00 2.00
Cebus apella 0.10 8.00 0.80
Chiropotes satanus 0.20 10.00 2.00
Pithecia pithecia 0.20 3.00 0.60
Saguinus bicolor 0.69 5.14 3.54


Species Group Density Average group # individuals/km2
(groups/km2) size

Alouatta belzebul 4.94 4.38 21.61
Cebus apella 0.93 6.34 5.90
Chiropotes satanus 0.62 10.00 6.20
Saguinus midas 0.62 3.00 4.84

Table 8. Beetle species tested for seed burial abilities.

No. of Beetles % % Ave. Depth Ave. Depth
Trials per Small Large Small (SD) Large (SD)
Trial Seeds Seeds
Buried Buried

Canthon "small matte black" I
Canthon aequinoctialis 2
Canthon triangularis 2
Coprophanaeus lancifer 9
Coprophaneus jasius 3
Deltochilum pseudicarus 2
Deltochilum septemstriatum 3
Dichotomius batesi 6
Dichotomius boreus 4
Dichotomius lucasi 3
Dichotomius podalirius 7
Eurysternus caribaeus 3
Glaphyrocanthon subhyalinus 4
Hansreia affinis 1
Oxysternon conspicillatum 17
Oxysternon selenus 1
Phanaeus cambiforti 4
Phanaeus chalcomelas 9
Scybalocanthon pygidalis 4
Telhyboma orbiculare 1

6 50 0.6 (0.28)
6 8 0 1 0
6 19 0 2.02 (1.6) 0
2 66 72 5.26(4.03) 3.95(2.94)
2 3 0 1 0
2 0 0 0 0
4 17 0 0.57 (0.22) 0
4 74 14 3.68 (2.85) 2(1.22)
2 98 94 9.23(3.76) 8.28(4.06)
4 30 0 3.79 (2.5) 0
2 59 18 4.21 (2.87) 3 (2.45)
10 5 0 1 0
10 2 0 1.55 (0.9) 0
6 0 0 0 0
2 81 64 5.65(3.75) 2.57(2.52)
1 55 40 6.7 (1.49) 7
4 61 12 3.39(2.04) 1
4 64 21 4.32(2.74) 2.86(2.41)
6 6 0 1.07(1.02) 0
2 15 0 1 0



Table 9. The standardized scoring coefficients for the principal component analysis of
beetle seed burial guilds. sm=small seeds, lg=large seeds.

Variable PC 1 PC 2 PC 3 PC 4
Mean biomass 0.151 0.647 0.301 1.088
Variance biomass 0.182 0.485 0.260 -1.602
% sm buried 0.187 -0.393 1.625 -0.174
%l Ig buried 0.214 -0.124 -0.256 1.911
Average depth sm 0.214 -0.248 -0.377 -0.677
Average depth Ig 0.211 -0.155 -1.238 -0.489
explained (%) 75.3 18.8 3.7 2.1

Table 10. Guild scores for beetle species.

Ateuchus bronze
Ateuchus connexus
Ateuchus histrio
Ateuchus murrayi
Ateuchus setulosus
Canthidium deyrollei
Canthidium dorhrini
Canthidium gerstaeckeri
Canthidium pinotoides
Canthidium ruficolle
Canthidium shiny black
Canthidium splendidumn
Canthon 7mm matte black
Canthon acutus
Canthon aequinoctialis
Canthon brunneus
Canthon literatus
Canthon podagrius
Canton septemmaculatus
Canthon sordidus
Canthon triangularis
Canthon virens
Coprophanaeus lancifer
Deltochilum amazonicum
Deltochilum guyanensis
Deltochilum pseudicarus
Deltochilum septemstriatum
Dichotomius batesi
Dichotomius boreus
Dichotomius carbonarius
Dichotomius carinatus
Dichotomius imitator
Dichotomius lucasi
Dichotomius mamillatus
Dichotomius melzeri
Dichotomius ohaust
Dichotomius worontzowi
Eurysternus caribaeus
Eurystermus cayennensis
Eurysternus confusus
Eurysterus cymescens
Eurystermus hamaticollis
Eurystermus hirtellus
Eurysternus inflexus
Euysternus velutinus
Glaphyrocanton subhyalinus
Hansreia affinis

2 Neocanthidium lentum 2
2 Neocanthidium miscellum 2
2 Ontherus 10 mm black 3
2 Ontherus carnfrons 3
2 Ontherus lamijnfer 3
2 Onthophagus bidentatus 3
3 Onthophagus hircutus 3
3 Onthophagus rubresens 3
2 Onthophagus scoop clyp 3
3 Ontophagus 2 knobs 3
2 Oxysternon conspicillatum 6
2 Oxysternon macleyi 6
3 Oxysternon selenus 6
2 Phanaeus alvarengai 4
2 Phanaeus bispinus 4
2 Phanaeus cambiforti 4
2 Phanaeus chalcomelas 4
2 Pseudocanthon xanthurus 2
3 Scybalocanlhon org pro 3
2 Scybalocanthon pygidalis 2
3 Scybalocanthon trimaculatus 2
2 Telhyboma orbiculare 1


Table 11. Correlations and P-values for amount of disturbance versus beetle guild

Guild Spearman's P-value
Score Rank
Guild 1 0.284 0.05>p>0.025
Guild 2 0.259 0.10>p>0.05
Guild 3 0.098 ns
Guild 4 0.001 ns
Guild 5 -0.325 0.05>p>0.025
Guild 6 0.334 0.05>p>0.025





.I. B'


- ^







+ Research Sites

Figure 1. Map of study sites.

Primary Forest
Secondary Growth
Clear Cul
Trapping Stalion
Transect Route

Figure 2. Map of study site in Rondonia


H Primary Forest
* Secondary Growth
* Clear Cut
Trapping Station
B Transect Route
H River

Figure 3. Map of study site at Reserva Adolfo Ducke

4A ffi idy Forest lo
I Pmary Forest Trapping Station
Secondary Growth
*CeIr ut TransectRoute
Figure 4. Map of study site at CaxiuanaCut
Figure 4. Map of study site at Caxiuana



Figure 5. Abundance of individual beetles over time at the Rondonia site.





AA^ "
^V<7 \ A-V^^
U^-^^A A*<-flr^A~~^^,^JM,----^.^^A^A






Figure 6. Abundance of individual beetles over time at the Ducke site.





Foi 7 Ad -o iPt
6 0 -0--Voom---

M 40
z 20

0' i A A A 4' 4' 4'- A 4' 4' 4' 4' 4' 4'


Figure 7. Abundance of individual beetles over time at the Caxiuana site.




wI ---Pniuyn
I- ClSeconduy




Figure 8. Average beetle biomass over time at the Rondonia site. Error bars are + or one standard deviation.



Figure 9. Average beetle biomass over time at the Ducke site. Error bars are + or one standard deviation.

-- Primary


U 4 I




0.25 T




Ob* 0 0 0 0 0 0 0 0 0 0 '


Figure 10. Average beetle biomass over time at the Caxiuana site. Error bars are + or one standard deviation.



--Pnnwry t





SNSeed predator
I *NW49Wedat om
6 1 so




Figure 11. The proportion of monkeys that are seed predators versus non-seed predators at all three sites.











Ec Ct Sp DOat DIuc Dbor Dwc Oc Peam Pch CI
Beetle Species

Figure. 12. The proportion of small seeds (< 5mm) left on soil surface by some beetle species.



0.4 -


Ec Ct Sp Dbat DIuc Dbor Dear
Beetle Species

Pcam Pch CI

Figure 13. The proportion of large seeds (>10 mm) left on soil surface by some beetle species.

GGuiud I


Guild 2 Guild 5

--- / 0E --- 0------------, ____________
.5 1) -0 I 0.5 1 1.5 2 2.5

Guild 3 _\ Guild 4

Guild -"

Factor 1

Figure 14. Plot of factor 1 and factor 2 of the principal component scores of beetle seed dispersal guilds.

* dbat
* duc

a oc


(a) Burrowers


1 .

---^----- .-----*---------------

1 -0.5 s 1 1
a --

Dim 1

(b) Rollers

---------------------------- -------------- ^ ------


ux O
------------------------e4 ----------------------

I 8 46 -4A 02 as 0


Dim I







Figure 15. Correspondence analysis of trap stations and beetle species in Rondonia divided by
(a) burrowers, (b) rollers. Dimn=dimension.


(a) Diurnal

-- ----------0



~-- -------------,4



E705 1

Dim 1

(b) Nocturnal

as !





0.2 06 (

Figure 16. Correspondence analysis of trap stations and beetle species in Rondonia
divided by (a) diurnal, (b) nocturnal. Dimn=dimension.


x 1


(a) Burrowers

Dim I

(b) Rollers




Es ^

Dim 1

Figure 17. Correspondence analysis of trap stations and beetle species in Ducke
divided by (a) burrowers, (b) rollers. Dim=dimension.

-a- t-

- --0.5


I 05 1

15* 2 25







(a) Diurnal

--------------------- --- --------------------

ll5 0 i 1 1 5 2 A


Dim 1

Figure 18 Correspondence analysis of trap stations and beetle species in Ducke
dividedby (a) diurnal, (b) nocturnal. Dim-dimension.


(a) Burrowers


15 -1 -0.5 05 1 1.5


(b) Rollers


U __-_ __x- _________

-I J5 D 0.5 1 1.5 2 2.5 3 3
*+I. g

Dm 1

Figure 19. Correspondence analysis of trap stations and beetle species in Caxiuana
divided by (a) burrowers, (b) rollers. Dim=dimension.





(a) Diurnal


S5 0.5 1 1.5 2 2.5 3

1 A


Dim 1

(b) Nocturnal



S-1.5 -1 -0.5 0.5 1 1.5

I~~~~ ~~ ~ ~ ~~~~~ ----------9 ----------------

Dim I

Figure 20. Correspondence analysis of trap stations and beetle species in Caxiuana
divided by (a) diurnal, (b) nocturnal. Dim=dimension.









______ -- __--- n--

0 1

5 -0.2 -0.15 -0.1 0.05 0.05 0.1 015 0.2 OP.25 C

Figure 21. Correspondence analysis of trap stations and seed burial guilds
in Rondonia. Dim=dimension.



*Guild 4



- Ce


04 0.6

0.8 1 1.2 14


Figure 22. Correspondence analysis of trap stations and
seed burial guilds in Ducke. Dim-dimension.

0 _


01 *

06 0.2 0.2 04 0.6 0.8


Figure 23. Correspondence analysis of trap stations and seed burial guilds
in Caxiuana. Dim=dimension.

10 X

x x

Sa z A *Guildl
0 0 Guild 2
2 A AGuild3
0 XGuHd4
3 X xGuild5
SA 0 Guild 6


0 1 2 3 4 5 6
Amount of Disturbance

Figure 24. The average guild distance (Euclidian) from correspondence analyses
versus a ranking of habitats by amount of disturbance.


N% A Guild3
S- Guild 6

15 Guild


1 2 3 4 5
Habitat Disturbance Level

Figure 25. The hypothetical allowable disturbance of
a habitat in relation to the abundance of important seed burial guilds.
A=transition primary to secondary forest; B=beetles become too scarce to influence seed survival.



This study addresses the problem of rainforest regeneration from the

community perspective, the beetle community being one step in the process of seed

dispersal that also involves the tree community, and primary disperser community.

The actual amount of secondary dispersal accomplished by dung beetles, how many

seeds are dispersed, what sizes of seeds, distribution patterns underground,

germination, and survival through the seedling stage to become reproductively active

adult trees, is far from determined. Nevertheless, I have illuminated some important

factors that may influence seed dispersal dynamics: dung beetle species vary in their

efficiency as secondary seed dispersers; beetles can be classified into guilds on the

basis of these variations; guild structure differs at different sites, and the abundance of

the guilds are related to levels of habitat disturbance.

Establishing the guild index required testing beetle species individually to

determine their seed burial abilities. Several authors (Andresen 1999; Feer 1999;

Estrada et al. 1991) have presented beetles with seeds imbedded in dung to determine

burying abilities. In these previous experiments, beetles were allowed to bury dung in

PVC cylinders measuring 20 cm in diameter. Results from some of these experiments

should be reexamined, as restricting beetles to a small cylinder may force them to bury

dung and seeds deeper than they normally would. Most beetles that burrow deeply

make tunnels at an angle from the dung pat (Halffter & Edmonds 1982). A beetle

confined to a small diameter vertical-sided bucket may be forced to dig down rather

than at an angle. My experiments were conducted in buckets with a top diameter of 40

cm, and this procedure yielded much more shallow burial of seeds (even with the

largest beetles) than that found by Feer (1999). Andresen (1999) and Estrada et al.

(1991) reported shallower seed burials, but the beetles were smaller in both these

studies than in Feer's (Andresen allowed beetles (not identified) to colonize the

cylinders and Estrada et al. worked in Mexico with a community of smaller sized

dung beetles, mostly Canthon).

Beetle seed burial guilds were constructed by combining the variables of

biomass, variance in biomass, proportion of small (< 5 mm) and large (> 10 mm)

seeds buried, and average depth of burial for large and small seeds in a principal

component analysis (PCA). This analysis allows multiple, highly correlated variables

to be reduced to fewer axes. These axes are the principal components that describe the

greatest amount of variation in the data and by preserving Euclidean distances between

the original variables can be used as descriptors of the objects. In this case, species

cluster by their PCA scores such that similar species will be close to each other, and the

axes define their position in the two dimensional space that explains the majority of the

variance. Guild scores were assigned based on the clumping in the plot and knowledge

of the biology of the beetles. For example, Telhyboma orbiculare, Deltochilum

pseudicarus, and the unrelated Coprophanaeus jasius were all placed in guild 1, due to

their large size and inefficient seed burial. These beetles were given the lowest rank

because they dispose of a large amount of dung but do not bury seeds. Guild 2 was

mostly composed of small beetles (usually < 6-7 mm) that roll or bury a small amount

of dung. Guild 3 contain medium sized beetles, both rollers and burrowers (7-13

mm) that bury small seeds effectively. Guild 4 contained medium beetles that bury

seeds at good germination depths, under 2 cm but less than 10 cm; these beetles were

rare at all sites (Figure 24). Guild 5 was composed of large effective buriers, mostly of

the nocturnal genus Dichotomius. The most effective buriers and the largest beetles,

primarily Oxysternon and Coprophanaeus lancifer, were placed in guild 6. The one

outlier in the graph (lower, right) was Dichotomius boreus, which often, but not

always, buried seed too deeply for germination; this species was placed in guild 3.

Communities are categorized by the species richness and abundance in a habitat

or site. Communities may vary widely in their species composition and abundance,

even over dstances a few meters, and the dung beetle communities of the Amazon

basin are no exception. Numbers of species and numbers of individuals were

statistically different in primary forest, secondary forest, and clear-cuts. More species

and individuals were collected in primary forest than either other habitat. Species in

secondary growths were subsets of primary forest species, less abundant and lower in

species richness. There was a strikingly total lack of overlap in species composition

between primary forest and pasture.

Habitat disturbance clearly affects beetle community composition in three sites

in the Amazon basin, and increased disturbance resulted in fewer beetle species and

abundance, even though the amount of dung may be higher in disturbed habitats.

Examination of the effect of increased disturbance on the populations of high quality

seed buriers shows that some of the best dispersers, such as Oxysternon, and

Phanaeus were the first to disappear when disturbance was high. On the other hand, a

few species of good buriers, for example, Dichotomius batesi, are eurytopic, and can

withstand moderate disturbance.

Dung beetle community composition may have an important effect on seed

dispersal, and the present research suggests that division of beetles into only roller and

burrower guilds, as has historically been the case, will not yield a complete picture of

the potential for seed burial in an area. I suggest that guild composition or ecological

niche (quantified by the guild index) is a more accurate description of community

structure than species composition and abundance when examining seed dispersal

potential. While the Rond8nia site had a greater abundance of beetles than the Reserva

Ducke site, a large part of the biomass was in species that were poor seed dispersers,

such as Canthon aequinoctialis and Eurysternus caribaeus. This pattern is

demonstrated by the relatively greater distance of guild 2 (low efficiency seed

dispersers) to guild 3 (more efficient seed dispersers) at more stations in Ducke and

Caxiuana than Rondonia in the correspondence analysis plots.

Each site showed different patterns of dung beetle community structure in the

comparisons of guilds. In Caxiuana and Ducke, guild 6, the most efficient seed

buriers, were nearest the stations that had the most monkey sightings. Additionally, at

Caxiuana, guild 3 was also nearest the area of greatest monkey abundance. At Ducke

guild 1 was in higher abundance at stations with the most monkey sightings, although

guild 3 clustered with 4 stations, two of which were in primary forest with the highest

monkey abundances. At Rond6nia, only guild 4, which tended to be rare, showed an

association with numbers of monkeys sighted. Monkey sighting and beetle abundance

were correlated significantly at Ducke and Caxiuana, but not Rond6nia. At all three

sites, guild 5 clustered with a disturbed habitat; this pattern is clear in the correlations

among guilds and habitats.

When average guild distance at all three sites was compared with the amount of

disturbance, several patterns emerge (Figure 24). Guilds 1, 2, 3, and 6 are in highest

abundance in the primary forest (disturbance number I and 2); these guilds are also

abundant in secondary growth of mild to moderate disturbance, and only become rare

in the most disturbed habitats studied. Guild 3 seems the least affected by habitat

disturbance, while guild 6, the large diurnal and crepuscular phanaeines, has the most

extreme reaction to the level of disturbance. Guild 4 is rare in all habitats. The one

negative correlation, guild 5, is rarest in primary forest, and one of the most abundant

guilds in the mild and moderately disturbed habitats. These beetles, primarily

Dichotomius species, are highly efficient at seed burial, and may seek out more

disturbed habitats, such as edges or treefall gaps. These are also good foraging areas

for many monkeys. Most guild 5 species are nocturnal, and might congregate at

monkey sleeping trees. On the other hand, these beetles may be exploiting a niche that

other beetles find intolerable in terms of microclimate. Being nocturnal might allow

greater latitude in foraging habitat; the damp and cooler nights may equalize

microclimates between disturbed and primary forest. Nevertheless, even this guild

appears to have a threshold of disturbance beyond which it becomes rare. All the

disturbed habitats sampled were forested to a large extent, and the characteristics that

reduce the presence of a guild in a particular area need to be investigated.

The most disturbed habitat outside of cleared land (disturbance level 5) in all three sites,

human activity was highest and included some clearing, maintenance of roads or trails,

and foot (and sometimes horse) traffic.

Figure 25 shows a distribution of 3 guilds, based on the correlations discussed

above. This graph represents the optimum beetle guilds for effective seed burial, and

possibly forest regeneration. Guilds I and 2 are low efficiency buriers (guild I may

even be detrimental to seeds in that these beetles process a lot of dung, but do not bury

seeds), and are excluded from the graph. Guild 4 is always rare, and while these

beetles are efficient buriers, they were also excluded from the graph. Guild 3 is

important; these beetles are small, and do not bury a lot of dung, but they bury small

seeds at optimal germination depths (2-3 cm; Shepherd & Chapman 1998). They are

mainly diurnal, therefore could bury seeds at monkey foraging and travel routes.

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EFQ6GT4GQ_67V7R8 INGEST_TIME 2013-02-07T19:20:58Z PACKAGE AA00012955_00001