PLANT-ANIMAL INTERACTIONS - ORIGINAL PAPER

Look before leaping: foraging selectivity of capuchin monkeyson acacia trees in Costa Rica

Hilary Young Æ Linda M. Fedigan Æ John F. Addicott

Received: 31 August 2006 / Accepted: 3 October 2007 / Published online: 27 October 2007

� Springer-Verlag 2007

Abstract Acacia trees in Costa Rica have an obligate

mutualism with three species of Pseudomyrmex ants, which

vigorously defend their host tree from insect and mam-

malian herbivores. Depending on the size and species of

ant colony, individual acacia trees may be differentially

protected. For animals able to discern between weakly and

highly aggressive ant colonies, costs of ant stings from less

active colonies might be offset by nutritional value

acquired from feeding on acacia fruit or ant larvae in

swollen thorns. We examined foraging selectivity of

capuchin monkeys on acacia trees in Santa Rosa National

Park, Costa Rica. We measured four characteristics of the

acacia trees from which capuchins fed and of acacias

immediately adjacent to those in which the monkeys fed:

diameter at breast height (DBH), accessibility, species of

closest tree and ant species present. We found that capu-

chins prefer to forage in acacias that are large and

accessible. We also made two measurements of ant colony

activity on each tree, one before and one after disturbing

the ant colony. We found that the three species of mutu-

alistic ants differ in baseline activity levels and that

mutualistic ants are more active than non-mutualistic ant

species found in acacia trees. We also found that capuchins

foraged more frequently in trees colonized by non-mutu-

alistic ants, but the explanatory value (r2) of this model was

low. Furthermore, monkeys did not discriminate between

acacias on the basis of baseline ant activity or the ant

colony’s response to disturbance. We conclude that these

monkeys select acacia trees in which to forage based on

characteristics of the trees rather than the ants. In addition,

our study suggests that white-faced capuchins act as pre-

dators on the acacia ants but they probably benefit the

dispersal and reproductive success of acacia trees. Capu-

chins may in fact function as an additional mutualistic

partner for acacia trees via seed dispersal, but they must

overcome the ants’ defense of the trees to do so.

Keywords Acacia collinsii � Ants � Cebus capucinus �Mutualism � Pseudomyrmex

Introduction

The obligate mutualism of swollen-thorn acacias (Acacia

collinsii) with three species of Pseudomyrmex ants in Costa

Rica has long been a topic of study for ecologists (e.g.,

Janzen 1966, 1967, 1969; Young et al. 1990; Cronin 1998).

The swollen thorns of A. collinsii provide nesting sites for

ant colonies, and the leaves provide food for ants in the

form of Beltian bodies and extra-floral nectar. In turn, the

ants guard the acacias from predation by most herbivores.

Recently, primatologists have become interested in this

symbiotic mutualism because white-faced capuchin mon-

keys (Cebus capucinus) can overcome an ant colony’s

stinging workers to feed on both the acacia fruit and the

larvae-filled acacia thorns (e.g., O’Malley and Fedigan

2006). Monkeys of the genus Cebus employ a variety of

extractive foraging tactics over a range of intensities, from

forceful ripping to less destructive peeling and picking

(e.g., Janson 1985; O’Malley and Fedigan 2005). These

Communicated by Craig Osenberg.

H. Young � J. F. Addicott

Department of Biological Sciences, University of Calgary,

T2N IN4 Calgary, AB, Canada

L. M. Fedigan (&)

Department of Anthropology, University of Calgary,

T2N 1N4 Calgary, AB, Canada

e-mail: Fedigan@ucalgary.ca

123

Oecologia (2008) 155:85–92

DOI 10.1007/s00442-007-0883-z

monkeys are adept at circumventing the pain-inducing

defense systems of their intended prey. Thus it is perhaps

not surprising that the white-faced capuchin is one of the

few vertebrate species that frequently forages on ant-

defended acacia trees in the neotropics (Janzen 1969;

Panger et al. 2002; Fragaszy et al. 2004). C. capucinus

occur sympatrically with ant-acacia trees throughout most

of the tropical dry forests of Mesoamerica, with the

exception of Mexico. This study investigates whether

capuchins use selective foraging to successfully feed on

both acacia fruit and ant larvae inside swollen-thorns,

despite the obligate mutualism between A. collinsii trees

and acacia ants.

Acacias, ants, and the ant–acacia mutualism

Acacia collinsii (Fabaceae: Mimosoideae) trees range

along the west coast of Mexico through Central America to

Colombia and reach heights of up to 10 m (Seigler and

Ebinger 1995; Ebinger et al. 2000). Acacia seeds are

embedded in a yellow aril and each legume pod contains an

average of eight seeds (range = 1–12, n = 102; Valenta

2007). Throughout their range, A. collinsii trees have an

obligatory mutualistic relationship with three species of

ants, Pseudomyrmex spinicola, Pseudomyrmex nigrocinc-

tus, and Pseudomyrmex flavicornis (Ward 1993; Cronin

1998). Extra-floral nectaries at the base of the petiole

provide ants with sugar, and swellings on the leaflet tips

(Beltian bodies) provide ant colonies with protein, lipids,

and vitamins (Janzen 1966). Each acacia tree generally

houses only one species of ant, which nest in the trees’

hollow thorns.

In return for food and shelter, mutualistic ants protect the

tree in which they nest. Without this protection acacias show

lower growth and poor survival (Janzen 1966, 1967, 1975).

Acacia ants defend their host trees from insect and mam-

malian herbivores by swarming, stinging, and biting the

invader (Janzen 1966, 1969). Previous research on these

three mutualistic Costa Rican acacia ants has investigated

whether the ants differ in anti-herbivore effectiveness in

terms of patrolling activity, response time, attack rate, and

post-disturbance activity (Janzen 1966; Young et al. 1990;

Cronin 1998). Both Janzen (1966) and Young et al. (1990)

concluded that P. spinicola is the most aggressive of the

three. However, Cronin found that differences between the

activity levels of Pseudomyrmex species only emerge when

time of day is considered. In addition to the three mutualistic

Pseudomyrmex species, A. collinsii trees can host non-

mutualistic ants, such as Pseudomyrmex nigropilosa (Janzen

1975). Although in the same genus as mutualists, this species

and some other species of ant apparently do not protect the

tree from herbivores. These studies suggest that, depending

on temporal factors and the species of ant colony, individual

acacia trees may be differentially protected. Varying levels

of protection might provide an opportunity for the mutualism

to be exploited by predators like capuchins.

When foraging on acacias, capuchins typically extract

the aril and seeds from the bean pods, and spit out the

latter. However, intact acacia seeds have been found in the

monkeys’ feces (Wehnke et al. 2004; Valenta 2007), sug-

gesting that consumed seeds are neither chewed to the

point of destruction nor digested during gut passage. The

aril of acacias is brightly colored, rich in lipids and prob-

ably evolved to attract seed dispersers (Janzen 1969;

Handel and Beattie 1990). McCabe and Fedigan (2007)

found that the aril of one A. collinsii pod contains 0.024 g

of fat, 0.082 g of protein, and 0.182 g of carbohydrates,

which is highly nutritious compared to other capuchin food

items. They also found that ant larvae from the swollen

thorns of the acacia are very high in lipids (McCabe and

Fedigan 2007).

Many mutually beneficial relationships have apparently

evolved from exploitative interactions (e.g., Beattie 1985;

Connor 1995). And there are many cases of non-mutualists

taking advantage of a mutualistic system without providing

reciprocal benefits; that is, of being parasitic on a mutu-

alistic system (e.g., Janzen 1975; Letourneau 1990;

Bronstein 1991; Stanton et al. 1999; Hibbett 2002; Cheney

and Cote 2005). However, it is much rarer for a third

organism to exploit both parties in a mutualism. Examining

the ways in which capuchins interrupt the obligatory

mutualism of acacia trees and acacia-ants should provide

insight into the variable success of Pseudomyrmex ants in

protecting the trees and the success of the acacias in

defending themselves against arboreal herbivores. Miller

(1996) found that acacia seed dispersal by large herbivores

was advantageous to future seedling recruitment on the

African savannah. Given that capuchins are attracted to the

aril produced by the tree and that they swallow and pos-

sibly disperse acacia seeds, we also consider whether these

monkeys are purely exploitative of the mutualism or may

play a more complex role in the reproductive ecology of

A. collinsii.

In order to assess whether capuchins forage indiscrimi-

nately on any ant acacia tree, or instead select those trees

with less effective defenders, we address the following

questions:

1. Do capuchins forage preferentially in acacia trees with

given physical characteristics, such as greater size,

accessibility, and non-acacia neighbors?

2. Do ant species show differences in activity and

response levels and size of trees occupied?

3. Do capuchins forage preferentially in acacia trees that

house one of the Pseudomyrmex ant species and/or in

86 Oecologia (2008) 155:85–92

123

acacia trees with lower ant colony activity and

response levels?

Materials and methods

Study site and study subjects

We conducted this research at Santa Rosa National Park

[SRNP; the original sector of the Area de Conservacion

Guanacaste (ACG)]. SRNP encompasses 108 km2 of dry

tropical forest in northwestern Costa Rica. The subjects of

this study consisted of 44 wild adult and juvenile white-

faced capuchin monkeys; all belonged to one of two

habituated social groups: CP and LV. Group members were

individually identified by natural markings like facial scars,

patterns in fur coloring and size and shape of the black cap.

These two groups have been the focus of studies on the

behavioral ecology and life history of C. capucinus since

1983 (e.g., Fedigan and Jack 2005).

Data collection

We collected behavioral information on the foraging of

capuchins on acacias in two ways. Our main method was to

conduct 10-min focal follows on individuals, during which

we recorded all observed incidents of foraging on acacias

in detail. We also recorded data on acacia foraging ad

libitum between focal follows. We collected 192 h of focal

data in June and July 2003, and from January up to and

including May 2004. Observations took place between

0700 and 1800 hours.

Whenever a capuchin came into contact with an acacia

tree, we recorded monkey identity and number and type of

food items eaten (e.g., aril or the ant larvae inside thorns).

Capuchins only enter acacia trees for the purpose of feeding

on acacia ant larvae or acacia aril, and they spend very little

time in acacia trees, presumably because of the presence of

the stinging acacia ants. Physical measurements were taken

for every acacia tree in which we observed capuchins for-

aging. Once a capuchin’s foraging bout ended, we tagged the

tree, recorded its location and returned at the same time of

day within the next 2 days to measure four variables:

1. Acacia tree diameter at breast height (DBH).

2. A yes/no measure of the tree’s accessibility to a

capuchin entering from another tree. Capuchins tend to

travel from tree to tree at an average height of 10 m

and rarely travel on the ground (Melin et al. 2007). For

this reason, an acacia tree was categorized as acces-

sible to an arboreal monkey if its crown was B5 m

from the crown of a different species of tree;

otherwise, the tree was deemed inaccessible.

3. A yes/no measure of whether the closest adjacent tree

was an acacia.

4. The ant species present, determined as being one of the

three species of mutualistic Pseudomyrmex, or

‘‘other.’’ Many of the ‘‘other’’ ants were P. nigropil-

osa, the non-mutualistic ant described by Janzen

(1975). Voucher specimens of acacia ant were iden-

tified by J. Longino, an expert in Costa Rican ant

taxonomy.

We also took two measurements on ant colony activity for

each tree. We measured baseline ant activity as the bi-

directional flow of ants across a pre-marked 2-cm section

of the acacia tree trunk during a 1-min span. This

measurement was considered indicative of activity level

over the whole tree because previous studies have shown

that colony activity is generally constant over the entire

tree (i.e., Cronin 1998). For a measure of the response of

the ant colony to disturbance, we reassessed ant activity

after we tapped the tree 5 times with a stick.

We could count the bi-directional flow of ants accurately

up to *35 ants in the 1-min span. However, when ant

activity was greater than this, we employed a categorical

scale, with levels 2, 3, and 4 corresponding to 36–70, 71–

105, and 106–140 ants min@1, respectively. For analyses,

we used medians of these categories (i.e., 53, 88, and 123).

To compute a measure of ant response to disturbance,

we took the log10 of the ratio of ant activity after tapping

the tree to the baseline ant activity. Because ant activity

could be 0, we added 1 to both counts prior to taking their

ratio. Therefore, our measure of response to disturbance

was log 10Ant activity following disturbanceþ1

Baseline ant activityþ1

� �: An

increase in ant response is assumed to reflect an increase in

the encounter rate of a potential predator with the stinging

ants. This may affect whether a predator risks entering an

acacia tree to forage (Young et al. 1990).

Those acacias in which the monkeys foraged are refer-

red to as ‘‘trees used’’ (n = 303). In addition, we took the

same set of measurements on trees and on ant activity (see

above) for acacias over 1.5 m in height immediately

adjacent (nearest neighbors) to those in which the monkeys

fed. When a tree used had two equidistant nearest neigh-

bors, measurements of both trees were included in the

sample. The adjacent trees are referred to as ‘‘trees not

used’’ (n = 333). However, ant activity was only measured

on 80% of the trees not used. If we subsequently observed a

capuchin foraging in a tree that we had already measured as

a tree not used, we changed its classification to ‘‘tree used,’’

removed it from the ‘‘tree not used’’ category and re-

measured it. The two categories remained mutually

Oecologia (2008) 155:85–92 87

123

exclusive, allowing us to maintain the integrity of a com-

parison between the characteristics of trees in which

capuchins were observed foraging and those in which they

were not observed foraging.

Data analyses

All data analyses were conducted using the generalized

linear model (GLM) and the linear model (LM) procedures

in R, version 2.3.1 (R Development Core Team 2006). The

GLM procedure allows analyses of regression, ANOVA or

analysis of covariance problems using data with a variety

of distributions of errors (e.g., normal, binomial, Poisson,

and quasi-Poisson), and relationships between means and

variances (Crawley 2005).

One set of analyses considered the use of acacia trees by

capuchins as a function of: (1) whether DBH was[3 cm, (2)

accessibility, (3) whether the acacia tree’s nearest neighbor

was another acacia, and (4) the species of ant occupying the

acacia. For these analyses we treated tree use (0 or 1) as the

dependent variable, and applied the GLM procedure using

binomial errors. The analysis of dichotomous dependent

variables (e.g., tree use) against categorical independent

variables (DBH, accessibility, nearest neighbors, and ant

species) are essentially logistic ANOVAs and give results

that are identical to two-way v2 analyses.

As with most analyses with a large number of obser-

vations, statistical significance does not necessarily imply

importance of the independent variables for explaining

variation in the dependent variable. To provide a way of

addressing these concerns, statisticians have developed

various measures of goodness-of-fit, such as r2 for standard

regression problems. For regressions/ANOVAs involving

categorical dependent variables such as those used here,

several pseudo-R2 measures have been developed (Veall

and Zimmermann 1996). However, there is no consensus as

to which measures are most appropriate or how to interpret

the absolute value of these measures (Veall and Zimmer-

mann 1996). We choose to present Maddala’s pseudo-R2-

value (Maddala 1983; Eq. 2.50) in order to facilitate

comparison among our results and to indicate which results

are more likely to represent important and not just statis-

tically significant results.

For analysis of baseline ant activity as a function of ant

species we used log(ant activity + 1) and applied the GLM

procedure using the quasi-Poisson error distribution family.

For the analysis of the response to disturbance and tree

DBH as a function of ant species, we used the LM pro-

cedure, which assumes normal errors.

For those analyses using ant species as the independent

variable we used planned comparisons to test three mutu-

ally exclusive hypotheses: (1) P. spinicola differs from

P. flavicornis, (2) P. spinicola and P. flavicornis differ

from P. nigrocinctus, and (3) the three mutualistic ants

differ from the other ants. The results of these comparisons

are presented as t-tests with 1 df.

Results

Capuchins spent 23.6% of their overall foraging time on

acacia arils and ant larvae, which was a greater proportion

of time than was spent on the fruits of, or insects in, any

other tree species. Capuchins targeted the aril from A.

collinsii during 81.4% of acacia foraging time, 4 times as

often as the larvae inside acacia thorns (18.6%).

Physical characteristics of acacia trees used

in comparison to trees not used

To investigate whether capuchins foraged preferentially in

trees with certain qualities, we compared three measures of

tree phenotype between the two categories of acacia trees.

Capuchin tree use was associated non-randomly with

whether a tree’s DBH was [3 cm (v2 = 111.6, P \ 0.001,

n = 636, and pseudo-r2 = 0.214) (Fig. 1a). The sampled

trees whose DBH was [3 cm were more likely to be used

by capuchins (0.646, n = 275) than were the acacias whose

DBH were B3 cm (0.237, n = 361) (Fig. 1a).

Tree use by capuchins was also associated non-randomly

and positively with accessibility (v2 = 63.1, df = 1,

P \ 0.001, n = 634, and pseudo-r2 = 0.126) (Fig. 1b).

Sampled trees judged to be accessible to capuchins were

much more likely to be used by them. Finally, capuchin tree

mc 3 > si HBD

seYoN

Pro

po

rtio

n o

f sa

mp

led

tre

esu

sed

by

cap

uch

ins

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

0.1

)572(

)163(

elbissecca si eerT

seYoN

)825(

)601(

aicaca si NN

seYoN

)894(

)731(

(c)(b)(a)

Fig. 1 Proportion of sampled acacia (Acacia collinsii) trees used by

capuchin monkeys (Cebus capucinus) as a function of a whether trees

were [3 cm diameter at breast height (DBH), b whether trees were

judged to be accessible to capuchins, and c whether a tree’s nearest

neighbor (NN) was another acacia. The horizontal dotted line in each

graph represents the overall proportion of sampled trees that was used

by capuchins. The number above each bar represents the number of

trees in each category

88 Oecologia (2008) 155:85–92

123

use was associated non-randomly with whether a tree’s

nearest neighbor was or was not another acacia (v2 = 11.6,

df = 1, P \ 0.001, n = 635, and pseudo-r2 = 0.024). The

trees whose nearest neighbors were other acacias were less

likely to be used by capuchins (0.441, n = 498) than were the

acacias whose nearest neighbors were not acacias (0.605,

n = 137) (Fig. 1c). However, relative to DBH and accessi-

bility, whether a sampled tree’s neighbor was another acacia

was a weak predictor of tree use by capuchins.

Differences among the ants: comparison of ant activity

and size of trees occupied

We examined whether ant species that occur on A. collinsii

trees exhibit different levels of baseline activity, of

response to disturbance, and whether they occur in differ-

ent sizes of trees.

The undisturbed bi-directional flow of ants across a 2-cm

line during a 1-min span differed among the ant species

(F3,400 = 14.6, P \ 0.001, and r2= 0.122) (Fig. 2a). Baseline

activities of P. flavicornis and P. spinicola differed from

each other (t = 2.72, df = 1, and P = 0.006). The baseline

activity level of P. nigrocinctus was significantly greater

than that of the other two mutualist species (t = 6.32, df = 1,

and P \ 0.001), and the baseline activity of the three

mutualist species combined was greater than for the ‘‘other’’

category of ants (t = 2.85, df = 1, and P = 0.004). The

response of ants to disturbance, measured as

log 10Ant activity following disturbanceþ1

Baseline ant activityþ1

� �also varied

among the types of ants (F3,400 = 6.339, P \ 0.001, and

r2 = 0.053) (Fig. 2b). However, of the three planned com-

parisons, only the contrast between P. spinicola and

P. flavicornis was significant (t = 4.05, df = 1, and P \ 0.001).

The DBH of trees occupied differed among ant species

(F3,464 = 5.23, P = 0.001, and r2 = 0.092) (Fig. 2c). P. ni-

grocinctus occupied smaller trees in our sample than the

combination of the other two mutualistic ants (P. flavicornis

and P. spinicola) (t = 3.05, df = 1, and P = 0.002) and the

‘‘other’’ ants occupied larger trees than the mutualistic ants

(t = 3.199 and P = 0.001). However, even though the results

are statistically significant, the explanatory value of the

model (i.e., r2) is very low. The number of trees occupied by

P. flavicornis, P. spinicola, P. nigrocinctus, and other ants

for which DBH was measured were 79, 307, 52, and 37,

respectively.

Ant characteristics of acacia trees used in comparison

to trees not used

To investigate the possibility that capuchins foraged pref-

erentially in trees with certain ant species or ant activity

characteristics, we analyzed tree use by capuchins as a

function of ant species, baseline ant activity and ant

response ratio.

Capuchins showed differential use of trees occupied by

different species of ants (v2 = 9.12, df = 3, P = 0.027,

n = 475, and pseudo-r2 = 0.025) (Fig. 3), with the proba-

bility of use being highest for the trees with the ‘‘other’’ ants

and decreasing in order from P. spinicola and P. flavicornis

to P. nigrocinctus. Capuchins were significantly more likely

to use the few trees in our sample occupied by the other non-

mutualistic ants than by the three mutualistic species

(z = @2.71 and P = 0.006). In addition, although not a large

enough difference to reach significance (z = 1.41 and

P = 0.157), the use of trees occupied by the highly active

Species of ant occupying tree

P. flavicornis

P. spinicola

P. nigrocinctusOthers

)mc(

HB

D

0

2

4

6

8

10

stna f

o esn

opse

Rec

nabr

utsid

ot

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

tna e

nilesaB

nim .

on( ytivitca

1-)

0

20

40

60

80

100

120

140(a)

(b)

(c)

Fig. 2 a Baseline ant activity, b the response of ants to disturbance

measured as log 10Ant activity following disturbanceþ1

Baseline ant activityþ1

� �; and c the

DBH of trees occupied by ants as a function of the species of ant

occupying a tree. In each box and whisker plot the shaded boxrepresents the 25th and 75th percentiles; the line represents the

median, the whiskers represent 10th and 90th percentiles, and dotsrepresent more extreme observations (sample sizes for trees occupied

by Pseudomyrmex flavicornis, Pseudomyrmex spinicola, Pseudomyr-mex nigrocinctus, and other ants were 65, 276, 48, and 15,

respectively for a and b, and, for c, n = 79, 307, 52, and 37,

respectively)

Oecologia (2008) 155:85–92 89

123

P. nigrocinctus ants appeared to be less than the use of trees

with the other two mutualistic species. Note also that

although there was a significant effect of ant species, the

extent to which ant species explain tree use by capuchins is

quite low relative to tree DBH and tree accessibility.

To investigate the possibility that capuchins forage pref-

erentially in acacias with low levels of ant activity,

regardless of the ant species, we assessed the regression of

tree use by capuchins as a function of baseline ant activity.

Although the relationship was significant (P = 0.025 and

pseudo-r2 = 0.015), the explanatory power of baseline ant

activity was low, and the significance of the relationship was

marginal for such a high sample size. Therefore, we do not

consider this to be an important relationship. Furthermore,

capuchins did not forage preferentially in sampled trees with

lower response levels of ants to disturbance. The response

level of ants to disturbance (regardless of the species of ant)

was not a good predictor of whether or not a tree was used by

capuchins (P = 0.129 and pseudo-r2 = 0.007).

Discussion

Acacia tree characteristics and capuchin foraging

decisions

Capuchins preferred to forage in larger diameter acacias.

This variable explained more variance in tree use than any

other characteristic of trees or the ants that occupied them.

Preferential use of large-diameter trees is likely a natural

consequence of the obvious correlation of DBH and tree

height, and of our observations that capuchins preferred to

travel through all trees at heights[5 m. This is in keeping

with Melin et al. (2007) who found that capuchins usually

forage and feed at heights averaging 10 m above the forest

floor.

Another key to the use of acacia trees by capuchins is

whether they can access acacias from a neighboring tree

without actually having to be in the acacia. We found that

capuchins almost always entered acacias by swinging into

them from large neighboring trees not harboring acacia

ants and that the trees used by capuchins were not sur-

rounded by other acacias as often as were the trees not

used. Capuchins almost never enter an acacia from the

ground and indeed, the distance of the closest branch of a

tree without acacia ants may be the most telling factor in a

capuchin’s decision to forage in an acacia.

Ant characteristics and capuchin foraging decisions

Of the four types of ants that occupy A. collinsii trees at

this site, P. nigrocinctus was most active when undisturbed

and more likely to be found in smaller trees. Non-mutu-

alistic ants were the least active when undisturbed and

more likely to be found in larger trees. P. spinicola

exhibited the greatest response to tree disturbance, sup-

porting Janzen’s (1966) and Young et al.’s (1990) findings

that this species is the most aggressive and protective of the

mutualistic ants.

Capuchin monkeys were more likely to forage in acacias

occupied by non-mutualist ants and more likely to use trees

when ants exhibited low activity levels prior to disturbance.

However, these were both weaker predictors of capuchin

foraging decisions than were tree size and accessibility.

Additionally, since non-mutualist ants were more likely to

be found in larger acacias also preferred by monkeys, we

conclude that tree characteristics are more important to

capuchin foraging decisions than ant characteristics. Use of

acacias by capuchins showed no relationship with response

levels (i.e., aggressiveness) of the ants, which also indi-

cates that the monkeys are not making decisions based on

the ants’ differential potential to defend the trees.

What role do capuchins play in the ant–acacia

mutualism?

Willmer and Stone (1997) argued that the interests of a

plant and its defender may sometimes conflict, and this

may be the case for A. collinsii trees and mutualistic

Pseudomyrmex ants. While ants defend their host tree

against herbivores in order to ensure the tree’s (and con-

sequently their own) survival, it has been hypothesized that

eert gniypucco tna fo seicepS

P. flavicornis

P. spinicola

P. nigrocinctusOthers

Pro

po

rtio

n o

f sa

mp

led

tre

esu

sed

by

cap

uch

ins

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

)97(

)25(

)703(

)73(

Fig. 3 Proportion of sampled trees used by capuchins as a function of

the mutualistic (black bars) or non-mutualistic (gray bars) ant species

occupying the acacia. The horizontal dotted line represents the overall

proportion of sampled trees that was used by capuchins, and the

number above each bar represents the number of trees occupied by

each species of ant

90 Oecologia (2008) 155:85–92

123

acacia trees may depend on the attraction of birds (Janzen

1969) and/or ants (Handel and Beattie 1990) to the nutri-

tious and colorful aril for seed dispersal. Birds are the most

likely candidates as primary dispersers of acacia seeds

(Janzen 1969), although we rarely witnessed birds feeding

on the seed pods of acacia trees. Ants are unlikely to be

primary dispersers of these seeds because of the protective

Pseudomyrmex ants. Is it possible that capuchins play some

role in acacia seed dispersal?

Capuchins certainly prey on ant larvae found in swollen-

thorns of acacias and they consume arils produced by these

trees. We might conclude, therefore, that these monkeys

are simply a parasite on the system (sensu Letourneau

1990), an exploitative non-mutualist that takes advantage

of both parties in the system without providing reciprocal

benefits. However, we found that the vast majority of

acacia foraging by monkeys is directed at arils rather than

ant larvae and that the arils rather than the seeds are the

preferred target of consumption.

A. collinsii seeds do not germinate unless the seed

coat is damaged (Janzen 1969; Or and Ward 2003) and it

has been argued that passage through a monkey’s

digestive system may enhance the ability of the seed to

germinate (Wehnke et al. 2004). White-faced capuchins

are very mobile and travel about 4 km day@1, moving

seeds considerable distances from the parent plant and

defecating them *2 h after ingestion (Wehnke et al.

2003). Valenta (2007) found that capuchins at our site

defecate acacia seeds an average of 278 m from the

parent tree and that 41% of 282 planted seeds germinate

after passage through a capuchin’s gut as compared to

31% of 600 seeds planted without benefit of monkey

passage. However, it may be premature to conclude that

capuchins are effective acacia seed dispersers without

knowing what proportion of ingested seeds the monkeys

pass whole.

Capuchins are pre-adapted by their foraging patterns to

overcome the defense of protective ants, by their dietary

patterns to prefer arils to seeds and by their ranging pat-

terns to carry seeds over long distances before depositing

them. In addition, capuchins in this study tended to pluck

and rip off only a few acacia pods from the terminal ends

of an acacia tree’s branches before moving into another

tree. Individual acacia trees did not sustain heavy structural

damage, and of the damage incurred, little occurred on the

vital main stem. Therefore, we suggest that C. capucinus

may act as predators (albeit weak ones) on the acacia ants

but they also benefit the dispersal and reproductive success

of A. collinsii trees. Our evidence suggests that capuchins

may function as an additional mutualistic partner for

A. collinsii via seed dispersal, but they must overcome the

ants’ defense of the trees to do so.

Acknowledgements We thank the staff of ACG, especially R.

Blanco Segura for help with this study and permission to do research

in the park. We are grateful to Dr J. Longino for help with ant

identification and to Dr R. Longair for advice on methodology. This

project was funded by a Sigma XI Grant (H. Y.), by the Canada

Research Chairs Program (L. M. F.) and by NSERC Grants (L. M. F.

and J. F. A.).

References

Beattie AJ (1985) The evolutionary ecology of ant–plant mutualisms.

Cambridge University Press, Cambridge

Bronstein JL (1991) The non-pollinating wasp fauna of Ficus pertusa:

exploitation of a mutualism? Oikos 61:175–186

Cheney KL, Cote IM (2005) Frequency dependent success of

aggressive mimics in a cleaning symbiosis. Proc R Soc Lond

B 272:2635–2639

Connor RC (1995) The benefits of mutualism: a conceptual frame-

work. Biol Rev Camb Philos Soc 70:427–457

Crawley MJ (2005) Statistics. An introduction using R. Wiley,

Chichester

Cronin G (1998) Between-species and temporal variation in acacia-

ant herbivore interactions. Biotropica 30:135–139

Ebinger JE, Seigler DS, Clarke HD (2000) Taxonomic revision of

South American species of the genus Acacia subgenus Acacia(Fabaceae: Mimosoideae). Syst Bot 25:588–617

Fedigan LM, Jack K (2005) The demographic and reproductive

context of male replacements in Cebus capucinus. Behaviour

141:755–775

Fragaszy D, Visalberghi E, Fedigan LM (2004) The complete

capuchin. Cambridge University Press, Cambridge

Handel SN, Beattie AJ (1990) Seed dispersal by ants. Sci Am 263:76–

83

Hibbett DS (2002) Plant–fungal interactions: when good relationships

go bad. Nature 419:345–346

Janson CH (1985) Aggressive competition and individual food

consumption in wild brown capuchin monkeys (Cebus apella).

Behav Ecol Sociobiol 18:125–138

Janzen DH (1966) Co-evolution of mutualism between ants and

acacias in Central America. Evolution 120:249–275

Janzen DH (1967) Fire, vegetation structure and the ant X acacia

interaction in Central America. Ecology 48:26–35

Janzen DH (1969) Birds and the ant 9 acacia interaction in Central

America, with notes on birds and other myrmecophytes. Condor

71:240–256

Janzen DH (1975) Pseudomyrmex nigropilosa: a parasite of a

mutualism. Science 188:936–937

Janzen DH (1983) Costa Rican natural history. University of Chicago

Press, Chicago, Ill.

Letourneau DK (1990) Code of ant-plant mutualism broken by

parasite. Science 248:215–217

Maddala GS (1983) Limited-dependent and qualitative variables in

econometrics. Cambridge University Press, Cambridge

McCabe G, Fedigan LM (2007) Effects of reproductive status on

energy intake, ingestion rates and dietary composition of female

white-faced capuchins (Cebus capucinus) at Santa Rosa, Costa

Rica. Int J Primatol 28:837–851

Melin A, Fedigan LM, Hiramatsu C, Sendall C, Kawamura S (2007)

Effect of colour vision phenotype on insect capture by a free-

ranging population of white-faced capuchins (Cebus capucinus).

Anim Behav 73:205–214

Miller MF (1996) Dispersal of Acacia seeds by ungulates and

ostriches in an African savanna. J Trop Ecol 12:345–356

Oecologia (2008) 155:85–92 91

123

O’Malley R, Fedigan LM (2005) Variability in food-processing

behavior among white-faced capuchins (Cebus capucinus) in

Santa Rosa National Park, Costa Rica. Am J Phys Anthropol

128:63–73

O’Malley R, Fedigan LM (2006) Age/sex class differences in

foraging on Acacia-dwelling ants by white-faced capuchins

(Cebus capucinus). Am J Phys Anthropol [Suppl] 42:141

Or K, Ward D (2003) Three-way interactions between Acacia, large

mammalian herbivores and bruchid beetles—a review. Afr

J Ecol 41:257–265

Panger M, Perry S, Rose L, Gros-Louis J, Vogel E, MacKinnon K,

Baker M (2002) Food processing technique differences across

three capuchin (Cebus capucinus) populations in Costa Rica. Am

J Phys Anthropol 119:52–66

R Development Core Team (2006) R: a language and environment for

statistical computing. R Foundation for Statistical Computing,

Vienna. http://www.R-project.org. Cited 6 Mar 2003

Seigler DS, Ebinger JE (1995) Taxonomic revision of the ant-acacias

(Fabaceae, Mimosoideae, Acacia, series Gummiferae) of the

new world. Ann Mo Bot Gard 82:117–138

Stanton ML, Palmer TM, Young TP, Evans A, Turner ML (1999)

Sterilization and canopy modification of a swollen thorn acacia

tree by a plant-ant. Nature 401:578–581

Valenta K (2007) Quantity, quality and spatial patterns of seed

dispersal in Cebus capucinus. MA thesis, University of Calgary

Veall MR, Zimmermann KF (1996) Pseudo-R2 measures for some

common limited dependent variable models. J Econ Surv

10(3):241–259

Ward PS (1993) Systematic studies on Pseudomyrmex acacia-ants

(Hymenoptera: Formicidae: Pseudomyrmecinae). J Hymenopt

Res 2:117–168

Wehnke EV, Hubbell SP, Foster RB, Dalling JW (2003) Seed

dispersal patterns produced by white-faced monkeys: implica-

tions for the dispersal limitation of neotropical tree species. J

Ecol 91:677–685

Wehnke EV, Valdez CN, Dominguez CA (2004) Seed dispersal and

defecation patterns of Cebus capucinus and Alouatta palliata:

consequences for seed dispersal effectiveness. J Trop Ecol

20:535–543

Willmer PG, Stone GN (1997) How aggressive ant-guards assist seed-

set in Acacia flowers. Nature 388:165–167

Young BE, Kaspari M, Martin TE (1990) Species-specific nest site

selection by birds in ant-acacia trees. Biotropica 22:310–315

92 Oecologia (2008) 155:85–92

123