The effect of an ant-hemipteran mutualism on the coffee berry borer (Hypothenemus hampei) in southern Mexico

Agriculture, Ecosystems and Environment 117 (2006) 218–221
www.elsevier.com/locate/agee
Short communication
The effect of an ant-hemipteran mutualism on the coffee berry
borer (Hypothenemus hampei) in southern Mexico
Ivette Perfecto a,*, John Vandermeer a,b
a School of Natural Resources and the Environment, 440 Church, Dana Building, University of Michigan, Ann Arbor, MI 48109, USA
b Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI 48109, USA
Received 5 August 2005; received in revised form 6 April 2006; accepted 11 April 2006
Available online 5 June 2006
Abstract
The indirect effect of an ant-hemipteran mutualism was investigated in the coffee agroecosystem of Southern Mexico. The ant, Azteca
instabilis, forms a mutualistic relationship with the coccid, Coccus viridis, on coffee plants. Through ?eld surveys and experimental studies,
the indirect effect of this mutualism on the main coffee pest in the region, Hypothenemus hampei, the coffee berry borer (CBB), was
investigated. Results indicate a signi?cant negative relationship between the number of coccids on a plant and the proportion of berries with
damage by the CBB. Results also indicate that the effect of the ants is signi?cant on per plant basis but not on per branch basis. Finally, a
signi?cant negative linear relationship was found between ant activity and the time it took ants to remove arti?cially placed borers on coffee
berries. This study indicates that the mutualistic relationship between Azteca ants and the coccids has a positive indirect effect on the plant by
reducing the numbers of the main insect pest of coffee.
# 2006 Elsevier B.V. All rights reserved.
Keywords: Integrated pest management; Coffee; Natural enemies; Pest control; Shade trees; Mutualism; Ants
1. Introduction
and the green scale Coccus viridis (Green), in which the ant
acts as a predator on the CBB, signi?cantly reducing its
The main pest of coffee in Latin America is the coffee
impact as a pest.
berry
borer
(CBB),
Hypothenemus
hampei
(Ferrari)
It has previously been demonstrated that the tending of
(Coleoptera: Scolytidae) a small beetle that bores into the
various hemipterans (mainly aphids, coccids, mealybugs
coffee beans as they are maturing (Baker, 1999; Damon,
and treehoppers) by ants frequently has a net bene?t on the
2000). Effects of the borer are variable but can reach
host plant (Buckley, 1987; Dansa and Rocha, 1992;
epidemic proportions and cause signi?cant economic
Oliveira and Del-Claro, 2005). In the case of A. instabilis
damage (Baker, 1999). Biological control potential has
and its mutualist C. viridis, the potential for an indirect
been reported for parasitic wasps (Barrera et al., 1990;
positive effect using arti?cially placed larvae of Pieris
Damon and Valle, 2002), entomopathogens (Samuels et al.,
rapae (L.) has been reported earlier (Vandermeer et al.,
2002; Mendez-Lopez et al., 2003; Neves and Hirose, 2005)
2002). While the ants tend coccids in large clumps
and ants (Armbrecht, 2003; Armbrecht and Perfecto, 2003;
concentrated on individual branches, they also patrol the
Ve´lez et al., 2003; Gallego and Armbrecht, in press). This
rest of the plant and, in the case of an herbivore that
study examines a complicated interaction involving a
requires some time to in?ict damage, the ants are potential
mutualism between the ant Azteca instabilis (F. Smith)
predators of that pest.
This study uses ?eld surveys and experimental manip-
ulations, to investigate whether the ant A. instabilis also acts
* Corresponding author. Tel.: +1 734 764 1433; fax: +1 734 936 2195.
E-mail address: perfecto@umich.edu (I. Perfecto).
as a predator on the coffee berry borer.
0167-8809/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.agee.2006.04.007

---

I. Perfecto, J. Vandermeer / Agriculture, Ecosystems and Environment 117 (2006) 218–221
219
2. Methods
coccids were counted on each branch to determine which
class the branch falls into: 0–6 individuals = low; 7–30
2.1. Field survey
individuals = medium; 30–70 individuals = high; 70 and
above = super high category. There was no need to actually
All A. instabilis nests were located in a 45 ha plot on
count coccids unless the actual number was near the border
Finca Irlanda, an organic shaded coffee farm located in
between the categories. Normally a minute or so of counting
southern Chiapas, Mexico (158100North and 928200West).
made it evident to which class the branch belonged. To make
The plot contains 10,647 shade trees (>10 cm dbh) in about
an estimate of the ?nal number estimate, all low branches
100 species, most of which seem perfectly suitable for
were multiplied by 0; medium branches by 15; high
nesting for this species of ant (nests have been found in 35 of
branches by 46; and super extra branches by 150. Scaling
the tree species, about what would be expected with a
numbers were determined from the initial study of 21 plants.
random allocation). Located near each of ?ve nests that were
The full count category included those plants that had no
isolated from other nests, six coffee trees were assessed for
berries and fewer than 25 branches. All leaves and stems
total number of coccids (C. viridis) (see below). The ?ve
were examined and all coccids greater than 2 mm were
nests were taken from widely separated areas within the
counted. Estimating scales with this procedure and
45 ha plot. Independent observers then assessed the same
comparing the estimate to the known numbers from the
coffee trees, for total number of coffee berries and CBB
preliminary study of 21 trees provides 93% ef?ciency in
damage. The proportion of damaged fruits per coffee plant
estimation (r2 = 0.926).
was estimated as the number of coffee berries that had CBB
(as evidenced from the entrance hole), over the total number
2.3. Experimental methods
of berries. The procedure was repeated in three areas that
were isolated from any A. instabilis nests, again sampling
A coffee plant with actively foraging A. instabilis
areas widely separated from one another in the 45 ha plot. In
individuals was located near to each of the ?ve ant nests
the end, a total of 34 coffee plants were surveyed for total
mentioned above. A total of nine branches were chosen so as
number of C. viridis, coffee berries and coffee berries
to represent a range of ant activity from high to low. On each
damaged by the CBB.
of the branches, ant activity was assessed by watching a
An additional survey was conducted on six coffee plants
single point for a 5 min period and counting the number of
(haphazardly chosen to represent a range of densities for C.
ants passing that point, or, if ant activity was especially high,
viridis) in which the total number of coccids on each branch
for a single minute. Then, one by one, individual borers were
was directly counted, and subsequently the rate of borer
placed on a berry on each of the branches and the time to
attack determined for the same branches.
removal by an ant was recorded. Since it takes a borer a little
more than an hour to completely burrow into a coffee berry
2.2. Coccid sampling and estimations
(personal observations), the beetles were watched for a
maximum of 70 min, if they persisted that long.
The following procedures were adapted, based on a
preliminary study of 21 coffee plants for which the actual
number of coccids were ?rst determined in absolute terms
3. Results
and then compared with relative estimates obtained in
several ways.
The relationship between number of coccids and
On a given coffee plant a quick survey was undertaken to
percentage of coffee berries attacked by the CBB was
determine whether there were any obvious concentrations of
assessed with a simple regression, which was signi?cant
coccids or ant activities. Based on this quick survey the plant
( p = 0.014, r2 = 0.18) for a linear regression of proportion of
was placed into one of three categories – a 5-min category, a
fruits with borers versus log of number of C. viridis (Fig. 1).
four-class category, or a full count category. For those plants
Thus the total number of coccids on a coffee plant was
in the 5-min category the entire plant was scanned
inversely related to the proportion of berries attacked by the
systematically from top to bottom during a 5 min period.
CBB. Assessment on a branch-by-branch basis revealed a far
All coccids (larger than 2 mm in length) on leaves, stems,
weaker relationship, which, while suggestive, was not
and berries, were counted concentrating on berries and
signi?cant statistically. It appears to be protection of the
terminal shoots where the coccids tended to concentrate.
coffee plant as a whole that is affected by the presence of the
The count of total number of coccids greater than 2 mm
ant, and not strictly localized to particular branches with a
encountered in a 5 min period is the base number. The
high concentration of coccids, although some degree of such
estimated actual number was computed as, estimated
localization is suggested.
number = 2.55 Â base number. If, during the course of
The results of the experimental manipulations indicated
counting, more than six individual coccids were encountered
that there was a negative relationship between ant activity
on any given branch, the plant was transferred to the four-
and the time of removal of CBB by ants (Fig. 2), with a
class category. For those plants in the four-class category,
signi?cant regression of the log of time to removal versus the

---

220
I. Perfecto, J. Vandermeer / Agriculture, Ecosystems and Environment 117 (2006) 218–221
log of ant activity ( p < 0.01; r2 = 0.68). If 1 h is taken as
approximately the critical time necessary for a borer to
become invulnerable to A. instabilis, the simple linear
regression of Fig. 2 suggests that the critical ant activity is
about 45 ants per minute crossing a given point, which is
signi?cantly lower than the maximum ant activity observed
in this experiment (114/min). As in previous experiments
with P. rapae, even occasional patrolling by ants can thus
effect control, although the results with the coffee berry
borer are not as evident as with P. rapae (Vandermeer et al.,
2002).
Furthermore, it is worth noting that in 44% of the cases,
the beetle was taken away in the mandibles of an ant,
suggesting predation. Unlike the case of P. rapae in which
most of the time the caterpillar simply fell off of the branch
and escaped predation, the ants frequently carried the borers
away, likely using them as a food source (Vandermeer et al.,
2002).
4. Discussion
The results suggest that the mutualistic relationship
between A. instabilis and C. viridis affords indirect bene?ts
Fig. 1. Relationship between proportion of coffee berries with coffee berry
to the coffee plant through a reduction of the coffee berry
borers and the log of the number of scale insects per coffee plant. Regression
borer. Thus, in practical terms, the damage done by the
equation is above box and explains about 18% of the variance.
mutualistic herbivore (the coccid) must be judged relative to
the protection against the coffee berry borer. This type of
indirect interaction has been noted before for other
associations in natural systems (Buckley, 1987; Dansa
and Rocha, 1992; Fagundes et al., 2005; Oliveira and Del-
Claro, 2005) but is not well studied in agroecosystems where
the net outcome of the interaction can have important
economic implications. The only other documented case for
an agroecosystem is that of the ant Dolichoderus thoracicus
(Smith), its associated mealybugs and several economically
important herbivores of cocoa in Malaysia (Khoo and
Chung, 1989). In the case of D. thoracicus, the bene?t of the
ant as predator of cocoa pests is so much higher than the
negative effect of the mealybugs that the farmers actively
introduce ants with mealybugs to their cocoa plantations as a
way of pest control (Perfecto and Castin˜eiras, 1998). The
costs and bene?ts of the A. instabilis complex in coffee have
not been studied suf?ciently to determine whether the net
outcome is likely to be positive for coffee production. Multi-
species mutualistic systems can have a wide range of
outcomes (Bronstein and Barbosa, 2002), however, the
negative impact on CBB of the mutualism suggest that the
ant-hemipteran mutualism can have an overall positive
effect on coffee production.
In the case of the Azteca-coccid–CBB complex, the ant
colonies – and therefore also the CBB control – are patchily
distributed (Vandermeer and Perfecto, 2006). These patches
Fig. 2. Relationship between log of time to removal of coffee berry borer by
constitute a very small proportion of the plantation (about
ants and the log of the ant activity. Dashed lines indicate approximate
4% of the shade trees in the case of Finca Irlanda). However,
position of critical ant activity that will attack an ant before the critical 1 h
period (see text).
since the ant acts as a predator, that 4% represents a sink for

---

I. Perfecto, J. Vandermeer / Agriculture, Ecosystems and Environment 117 (2006) 218–221
221
the CBB population. Removal of 4% of the CBB population
Buckley, R.C., 1987. Interactions involving plants, Homoptera, and ants.
may have only a small effect on the overall population
Ann. Rev. Ecol. Syst. 18, 111–135.
Damon, A., 2000. A review of the biology and control of the coffee berry
dynamics, but it does represent one more tool in the toolbox
borer. Hypothenemus hampei (Coleoptera: Scolytidae). Bull. Entomol.
of potential biological control of the most important coffee
Res. 90, 453–465.
pest in Latin America, the coffee berry borer.
Damon, A., Valle, J., 2002. Comparison of two release techniques for the
This local level dynamics leads to further speculation
use of Cephalonomia stephanoderis (Hymenoptera: Bethylidae), to
about what the dynamics might be on a more regional level.
control the coffee berry borer Hypothenemus hampei (Coleoptera:
Scolytidae) in Soconusco, southeastern Mexico. Biol. Control 24,
In a landscape dominated by shade coffee rather than
117–127.
unshaded coffee, the additional 4% control could have a
Dansa, C.V.A., Rocha, C.F.D., 1992. An ant-membracid plant interaction in
major effect, depending on how close the CBB population is
a cerrado area of Brazil. J. Trop. Ecol. 8, 339–348.
to being a stable one. In the current context, it appears that
Fagundes, M., Neves, F.S., Fernandez, G.W., 2005. Direct and indirect
isolated shade plantations in an area of predominantly
interactions involving ants, insect herbivores, parasitoids and the host
plant Baccharis dracunculifolia (Asteracea). Ecol. Entomol. 30, 28–35.
unshaded coffee may provide sinks that help control the
Gallego, M.C., Armbrecht, I. Depredacio´n por hormigas sobre la broca del
borer over a broad region.
cafe´ en cafetales cultivados bajo dos niveles de sombra en Colombia.
Revista Manejo Integrado de Plagas, Costa Rica, in press.
Khoo, K.C., Chung, G.F., 1989. Use of the black cocoa ant to control mirid
damage in cocoa. Plant. Kuala Lampur 65, 370–383.
Acknowledgments
Mendez-Lopez, I., Basurto-Rios, R., Ibarra, J.E., 2003. Bacillus thurin-
giensis serovar israelensis is highly toxic to the coffee berry borer.
We wish to thank the Peters family, the owners of Finca
Hypothenemus hampei Ferr. (Coleoptera: Scolytidae). Fems Microbiol.
Irlanda, for allowing us to conduct this study on their farm.
Lett. 226, 73–77.
G. Lopez-Batista and B. Estevan Chilel, helped with data
Neves, P.M.O.J., Hirose, E., 2005. Beauveria bassiana strains selection for
biological control of the coffee berry borer. Hypothenemus hampei
collection and the establishment of the 45 ha plot. Guillermo
(Ferrari) (Coleoptera: Scolytidae). Neotrop. Entomol. 34, 77–82.
Ibarra and Alvaro Ballinas provided logistical support. The
Oliveira, P.S., Del-Claro, K., 2005. Multitrophic interactions in a neotro-
study was supported by NSF grant DEB-0349388.
pical savanna: ant-hemipteran systems, associated insect herbivores and
a host plant. In: Burslem, D., Pinard, M., Hartley, S. (Eds.), Biotic
Interactions in the Tropics: Their Role in the Maintenance of Species
Diversity. Cambridge University Press, Cambridge, pp. 414–438.
References
Perfecto, I., Castin˜eiras, A., 1998. Deployment of the predaceous ants and
their conservation in agroecosystems. In: Barbosa, P. (Ed.), Perspectives
Armbrecht, I., 2003. Leaf litter ant diversity and function in coffee under
on the Conservation of Natural Enemies of Pest Species. Academic
different management systems. Ph.D. Dissertation. University of Michi-
Press, San Diego, pp. 269–289.
gan, Ann Arbor, MI, USA.
Samuels, R.I., Pereira, R.C., Gava, C.A.T., 2002. Infection of the coffee
Armbrecht, I., Perfecto, I., 2003. Litter-twig dwelling ant species richness
berry borer Hypothenemus hampei (Coleoptera: Scolytidae) by Brazi-
and predation potential within a forest fragment and neighboring coffee
lian isolates of the entomopathogenic fungi Beauveria bassiana and
plantations of contrasting habitat quality in Mexico. Agric. Ecosyst.
Metarhizium anisopliae (Deuteromycotina: Hyphomycetes). Biocontrol
Environ. 97, 107–115.
Sci. Techn. 12, 631–635.
Baker, P.S., 1999. The coffee berry borer in Colombia. Final Report of the
Vandermeer, J., Perfecto, I., 2006. A keystone mutualism drives pattern in a
DFID-Cenicafe´-CABI Bioscience IPM for coffee project. 143 pp.
power function. Science 311, 1000–1002.
Barrera, J.F., Baker, P.S., Valenzuela, J.E., Schwarz, A., 1990. Introduction
Vandermeer, J., Perfecto, I., Ibarra Nun˜ez, G., Philpott, S., Garcia Ballinas,
of two African parasitoid species to Mexico for biological control of the
A., 2002. Ants (Azteca sp.) as potential biological control agents in
coffee borer Hypothenemus hampei (Ferrari) (Coleoptera: Scolytidae).
shade coffee production in Chiapas, Mexico. Agroforest. Syst. 56, 271–
Folia Entomol. Mex. 79, 245–247.
276.
Bronstein, J.L., Barbosa, P., 2002. Multitrophic/multispecies mutualistic
Ve´lez, M., Bustillo, A.E., Posada, F., 2003. Depredacio´n de Hypothenemus
interactions: the role of non-mutualists in shaping and mediating
hampei por Solenopsis geminata y Gnamptogenys sp. (Hymenoptera:
mutualisms. In: Tscharntke, T., Hawkins, B.A. (Eds.), Multitrophic
Formicidae). In: Sociedad Colombiana de Entomolog?´a (eds) Libro de
Level Interactions. Cambridge University Press, Cambridge, pp. 44–66.
Resumenes XXX Congreso. Cali, Colombia, p. 26.

---

Document Outline

• The effect of an ant-hemipteran mutualism on the coffee berry borer (Hypothenemus hampei) in southern Mexico
  • Introduction
  • Methods
    • Field survey
    • Coccid sampling and estimations
    • Experimental methods
  • Results
  • Discussion
  • Acknowledgments
  • References

---