Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Received: 11 November 2017 | Accepted: 28 August 2018

DOI: 10.1111/1365-2664.13283

RESEARCH ARTICLE

Landscape structure regulates pest control provided by ants in


sun coffee farms

Natalia Aristizábal | Jean Paul Metzger

Department of Ecology, University of São


Paulo, São Paulo, Brazil Abstract
1. Ants play a fundamental role in coffee pest control. Despite this, there is a lack of
Correspondence
Natalia Aristizábal understanding about how landscape configuration and composition regulate the
Email: nati.aristizabal1@gmail.com ecosystem service provided by ants in sun coffee farms within highly fragmented

Present Address
landscapes.
Natalia Aristizábal, , Gund Institute for 2. We measured whether landscape structure influences ants’ ability to regulate cof-
Environment, University of Vermont,
fee berry borer (CBB) in sun coffee farms in Southeastern Brazil. Considering the
Burlington, Vermont
ecological interactions between ants and CBB at three different stages of pest
Funding information control (reduction of CBB presence, CBB infestation, and CBB bean damage), we
The Brazilian National Council for Scientific
measured pest control among 10 landscapes that represented a gradual differ-
and Technological Development (CNPQ);
Coordination for the Improvement of Higher ence in forest and coffee cover. We manipulated ants through exclusion experi-
Education Personnel (CAPES); São Paulo
ments and tested whether interactions between ants and different landscape
Research Foundation (FAPESP), Grant/
Award Number: 2013/23457-6; Brazilian structure metrics (distance to forest fragments, 2-km-level forest cover, and
Ministry of Education
300-m-level forest and coffee cover) influenced pest control.
Handling Editor: Owen Lewis 3. The presence of ants reduced CBB presence and CBB damage by up to 40%. We show
how pest control service indicators change depending on the landscape level. The prob-
ability of CBB presence increased with expanding coffee and forest cover at the 300-m-
level but decreased at the 2-km-level. CBB infestation reduced further after 25 m from
forest edges, suggesting ants that provide these ecosystem services are adapted to ma-
trix conditions in sun coffee farms. Ants reduced CBB presence, CBB infestation, and CBB
damage in landscapes with at least 40% 2-km-level forest cover. Beyond this threshold,
there is a turning point for ecological processes involved in pest control.
4. Synthesis and applications. This is the first long-term branch-level exclusion experi-
ment to present strong evidence of ants as efficient providers of pest control in sun
coffee farms. We show how landscape structure modulates key ecological pro-
cesses involved in three different ant-CBB interactions that regulate CBB popula-
tions. Forest cover measured at different landscape levels yielded different results
for CBB presence, emphasizing the importance of multi-scale studies for landscape
management. Considering surrounding forest cover and crop proximity to forest
fragments in planning the spatial arrangement of coffee farms can thus both im-
prove pest control as well as contribute to biodiversity conservation.

KEYWORDS
ants, Atlantic Forest, coffee berry borer, coffee borer beetle, coffee farms, ecosystem
services, land-use change, pest control

J Appl Ecol. 2019;56:21–30. wileyonlinelibrary.com/journal/jpe


© 2018 The Authors. Journal of Applied Ecology | 21
© 2018 British Ecological Society
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
22 | Journal of Applied Ecology ARISTIZÁBAL and METZGER

1 | I NTRO D U C TI O N landscape simplification, has sometimes been associated with an in-


crease in CBB removal by ants (De la Mora, García-­Ballinas, & Philpott,
Humans have altered more than 75% of the Earth’s ice-­free land 2015). A better understanding of the effects of landscape change on
(Verburg, Erb, Mertz, & Espindola, 2013), including at least 40% the ecological processes involved in ant-­mediated regulation of CBB
of natural habitats transformed into agricultural landscapes (Foley in monoculture sun coffee farms could reconcile higher yield coffee
et al., 2011). Agricultural intensification alters landscape structure, production and forest conservation (Morris, Jiménez-­Soto, Philpott, &
ecological processes, and species diversity, as well as the provi- Perfecto, 2018; Vandermeer & Perfecto, 2007).
sion of ecosystem services (Cardinale et al., 2012; Folgarait, 1998; Therefore, the fundamental goal of this study was to examine ant-­
Mitchell et al., 2015; Tscharntke et al., 2008). In agroecosystems, mediated pest control in sun coffee farms across different landscape
changes in landscape composition or configuration may modify conditions. Specifically, we tested whether proximity to a forest frag-
ecological processes that regulate pest control, acting on the per- ment and the amount of forest and coffee cover at different landscape
sistence and movement of predators at the landscape level (Bianchi, levels modulates the regulation of CBB by ants. We did this in sun-­
Booij, & Tscharntke, 2006; Boesing, Nichols, & Metzger, 2017). Land exposed coffee farms by experimentally excluding ants from selected
conversion also increases isolation from native habitats and typically branches during an entire coffee production season in Southeastern
decreases abundance and activity of crop pests’ natural enemies Brazil. We measured three indicators of pest control in which ants
(Karungi et al., 2014). Given the increasing tendencies of habitat loss could possibly interact with and regulate CBB: (a) CBB presence: we
and the possible changes in community composition with land con- counted presence or absence of CBB per branch to test the ability of
version, it is fundamental to test the effects of extreme land-­use ants to prevent CBB from colonizing coffee berries; (b) CBB infestation:
change for monoculture agriculture in the provision of ecosystem we measured the percentage of CBB in branches to test the capacity of
services; specifically, how pest control by mobile organisms is mod- ants to impede CBB population growth once at least one coffee berry
ulated by landscape structure changes. had been colonized; and (c) CBB damage: we calculated the percentage
Coffee agroecosystems benefit from several ecosystem services of spoiled coffee beans to test the ability of ants to prevent further
(Classen et al., 2014; Jha et al., 2014). One of these services is the reg- fruit damage by predating adult and immature CBB inside bored ber-
ulation of the world’s major coffee pest, Hypothenemus hampei Ferrari ries. We expected our response variables (CBB presence, CBB infes-
(Coleoptera: Curculionidae) or “coffee berry borer” (CBB) (Vega, tation, and CBB damage) would be higher on ant-­excluded branches.
Infante, Castillo, & Jaramillo, 2009). CBB is a coffee plant specialist, Furthermore, we predicted CBB presence, CBB infestation, and CBB
where females bore through coffee berries to reproduce. The brood damage to decrease as proximity to forest fragments and forest cover
then feed off coffee beans which reduces coffee’s quality, weight, increased, and as coffee cover decreased. Hence, we presumed ants
and yield (Vega et al., 2009). Although bored coffee berries are still would provide more pest control of CBB in landscapes with higher per-
marketable, in Brazil alone, the world’s leading coffee producer, CBB centages of forest and in coffee crops near forest fragments.
activity represents $215–358 million dollars loss per year (Oliveira,
Auad, Mendes, & Frizzas, 2013). Considering that, over 10 million
2 | M ATE R I A L S A N D M E TH O DS
hectares of native habitat in the world have been transformed for cof-
fee crops (FAO, 2016), and that more than 100 million livelihoods de-
2.1 | Study area
pend on coffee production worldwide (Bunn, Läderach, Ovalle Rivera,
& Kirschke, 2015; Fair Trade Foundation, 2012), the search for natural Our study took place in Southeastern Brazil, in a coffee-­producing
control methods for CBB is fundamental for economic and conserva- region called Mogiana and Sul de Minas in the border between the
tion reasons (Jaramillo, Borgemeister, & Baker, 2006). States of Minas Gerais and São Paulo (Figure 1, see Supporting
Among species known as natural enemies of CBB, ants stand out Information Figures S1 and S2a,b). Brazil produces around 35%
because they are extremely abundant in tropical systems (Fernández, of the commercialized coffee in the world (USDA, 2015) and the
2003; Hölldobler & Wilson, 1990), where coffee is best produced. In Mogiana and Sul de Minas region produces two-­thirds of it for two
the lab, ant colonies can efficiently remove adult CBB (Armbrecht & main reasons. First, the region is a culturally traditional area for cof-
Gallego, 2007; Larsen & Philpott, 2010) and predate on immature CBB fee production since the expansion of the railroad system at the end
directly inside bored berries (Morris & Perfecto, 2016). Studies in tra- of the 19th century (EMBRAPA, 2018). Second, the region has ideal
ditionally grown shaded coffee agroecosystems show several general- conditions for coffee production with a subtropical highland climate,
ist ant species defend coffee plants from CBB colonization (Gonthier, annual mean temperatures between 19°C and 22°C, and altitudes
Ennis, Philpott, Vandermeer, & Perfecto, 2013), and predate on CBB between 800 and 1,300 m (EMBRAPA, 2018).
(Bustillo, Cárdenas, & Posada, 2002; Philpott & Armbrecht, 2006). Nearly 90% of the coffee in Brazil is produced as sun coffee
Reduction of forest habitat and landscape heterogeneity, as well as a (Ricci, Araújo, & Franch, 2002) in an area expanding more than 2
decline in habitat quality, can lead to a decrease in richness of twig-­ million hectares (EMBRAPA, 2016). Within this region, we selected
nesting and leaf litter ants that provide pest control services (De la 10 independent landscapes following the procedures presented
Mora, Murnen, & Philpott, 2013; Perfecto & Vandermeer, 2002). On in previous studies from the same region (see Boesing, Nichols, &
the contrary, expansion of coffee crop area, which can be related to Metzger, 2017; Librán-­Embid, De Coster, & Metzger, 2017; Saturni,
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ARISTIZÁBAL and METZGER Journal of Applied Ecology | 23

F I G U R E 1 Study region and


geographical location of the 10 landscapes
(2-­km-­radius)

Jaffe, & Metzger, 2016). These landscapes, or circular sample areas only have two sites for logistical restrictions (one of the farms was
of 2-­km-­radii, were composed of an agricultural matrix and Atlantic abandoned and sold). Therefore, we had a total of 29 experimental sites
Forest remnants. The Atlantic Forest is an endangered biome con- involving 17 different farms. We only used Coffea arabica varieties of
sidered a biodiversity hotspot (Myers, Mittermeier, Mittermeier, Catuaí and Catucaí, which are physiologically similar because Catucaí
Da Fonseca, & Kent, 2000) and an important supplier of ecosystem is a cross from Catuaí. We controlled for type of soil and management
services regionally and globally (Ferraz et al., 2014; Joly, Metzger, & practices. Farms in the study applied similar herbicides, fertilizers, and
Tabarelli, 2014). Unfortunately, only 12%–16% of its known distribu- pesticides and harvested semi-­manually (except for one property—
tion remains (Ribeiro, Metzger, Martensen, Ponzoni, & Hirota, 2009). which included three experimental sites—that harvested mechanically).
The 10 2-­km-­level landscapes varied gradually in their amount
of forest cover from 10% to 55% (Figure 1). The matrix in our land-
2.2 | Landscape metrics and site selection
scapes was a heterogeneous mosaic dominated by coffee and pasture,
with some sugar cane and Eucalyptus production (see Supporting We used high-­resolution satellite images from ArcGIS 10.3 basemap
Information Figure S1). In each landscape, we established three sites imagery (2009–2011) with a reference scale of 1:5,000 and 1 m res-
with ant-­exclusion experiments except for one landscape that could olution to map the landscapes, classify and quantify land-­uses (see
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
24 | Journal of Applied Ecology ARISTIZÁBAL and METZGER

Supporting Information Table S1). Land-­use barely changed from any effect of Tanglefoot on CBB (see Supporting Information Figure
years of used satellite images (2009–2011) to the years in which we S2d). We collected a few individuals of the frequent visitor ant
executed our experiments (2014–2015). Nonetheless, during our morphospecies seen in our experimental branches for genus-­level
fieldwork, we actively revised and corrected land-­use classifica- identification.
tions when needed. To prevent spatial overlap, we kept landscape
centroids at a minimum distance of 6 km. We analysed landscape
2.4 | Pest control indicators: CBB presence/absence,
composition and configuration at two different levels: a 2-­km-­radius
infestation, and bean damage
level for each selected landscape and a 300-­m-­radius level around
each experimental site (details below). Land-­use in selected land- We considered three response variables, or indicators of CBB regula-
scapes did not vary greatly in 1, 2, or 3 km-­radius. We chose a tion, to test different ecological processes where ants and landscape
2-­km-­radius as our landscape level because some ant species com- structure could be decreasing the probability of CBB presence and
monly disperse within this distance after nuptial flights (Hölldobler & the percentages of CBB infestation and CBB damage. Bored berries
Wilson, 1990), and a 300-­m-­radius based on ant-­foraging distances are left with a 2 mm diameter hole drilled by adult females that lay
(Carrol & Janzen, 1973). their eggs inside the coffee beans (Damon, 2000). The presence of
Using FRAGSTATS 4.2 (McGarigal, Cushman, & Ene, 2012), we CBB, referred from here on as CBB presence, was considered when
measured the following landscape metrics as explanatory variables: there was at least one bored coffee berry on a branch. Absence was
distance from experimental sites to the nearest forest fragment, considered when none of the coffee berries on a branch had CBB.
2-­km-­level forest cover, and 300-­m-­level forest and coffee cover. The second indicator was the percentage of bored coffee berries
To select a representative range of these variables for our exper- out of the total number produced per branch, or “CBB infestation.”
imental sites, we first analysed all pixels inside radii of 300 m for Finally, the third pest control indicator was the percentage rotten or
each landscape to better understand the relationship between dis- damaged of each bored coffee bean, referred from here on as “CBB
tance to forest fragments and amount of native forest cover around damage.” On our last visit in June 2015, right before farmers’ harvest
coffee crops in this region (see Supporting Information Figure S3). time, we collected all coffee berries in our experimental branches.
Controlling for coffee variety restricted the selection of experi- We then separated bored from non-­bored berries and opened all
mental sites, but ultimately our study design included a range of bored berries. To calculate bean damage percentage or “CBB dam-
distances to forest fragments (5–110 m), 2-­km-­level forest cover age,” we measured the length of the galleries made by CBB and di-
(10%–55%), 300-­m-­level forest cover (2%–60%), and 300-­m-­level vided it by the length of the coffee beans inside each bored berry
coffee cover (5%–80%). (Jiménez-­Soto, Cruz-­Rodríguez, Vandermeer, & Perfecto, 2013).

2.3 | Ant-­exclusion experiments 2.5 | Data analyses


To test the role of ants in CBB control, we experimentally manipulated We removed from the dataset two experimental sites that did
the presence of ants in selected branches of coffee plants during not show evidence of CBB presence on any of the branches.
the 2014–2015 production season. We set up 10 ant-­exclusion Consequently, we analysed a subsample that included all 10 land-
experiments (Gonthier et al., 2013) per experimental site, right after scapes, but only 27 (out of 29) experimental sites. To test the ef-
the main flowering in October 2014 (before berries developed to ficiency of ant-­exclusion experiments, we performed a paired t-­test
ensure absence of CBB). In each coffee plant, we randomly selected of ant-­activity between all control and all treatment branches. We
two neighbouring branches with similar number of flowers, height, tested the collinearity among the explanatory variables [distance to
and position, and arbitrarily assigned one as the treatment branch forest fragments, 2-­km-­level forest cover, and 300-­m-­level forest
(ant-­excluded) and the other as the control branch. We then manually and coffee cover] and found no significant correlation. Therefore,
removed ants (if there were any) and isolated the branches by tying all four explanatory variables were used in further analyses. To test
away any other branch that could potentially come into contact. Next, the effects of forest and coffee cover change in exclusion of ants
we spread Tree Tanglefoot Insect Barrier (The Scotts Company, LLC) on each of the three response variables (CBB presence, CBB infesta-
along 10 cm of the ant-­excluded branch after the first seven nodes, tion, and CBB damage), we performed forward model selection using
counting from the tip to the base (see Supporting Information Figure Akaike Criterion Information (AICc). We used GLMM with a bino-
S2c). This physical barrier excluded any terrestrial invertebrates, mial distribution, the “glmer” function, and a logit-­link function in the
including ants. We visited each experimental site once a month to “lme4” package (Bates, Mächler, Bolker, & Walker, 2015; Zuur, 2009)
maintain exclusion experiments until June 2015. We used monthly in r version 3.2.3 (Development Core Team, 2018). We performed
visits to count on each control and ant-­excluded branch (a total of model selection using the “dredge” function from the MuMIn pack-
580 branches) the number of ants seen during five minutes of active age in r (Bartoń, 2018), starting with a full model that included sepa-
observations and the number of coffee berries produced and bored. rate fixed effects from the interaction of ant-­exclusion (presence or
We marked but did not manipulate control branches, except for a absence of ants in branches), all landscape metrics, and “ant-­activity”
dot of Tanglefoot that would not exclude ants but would control for (number of ants seen in five minutes at every visit). We selected
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ARISTIZÁBAL and METZGER Journal of Applied Ecology | 25

equally plausible models with ΔAICc < 2 (see Supporting Information cover. The probability to detect CBB presence was higher in ant-­
Table S2). To represent our experimental design, we included three excluded branches and increased generally as 300-­m-­level coffee
random effects in all models: the landscape identity, the experimen- cover increased (p < 0.001) and 300-­m-­level forest cover increased
tal site identity nested within the landscape, and the plant identity (p < 0.001), and it decreased as 2-­km-­level forest cover increased
nested within the experimental site. Percentages were calculated (p < 0.001, Figure 2a–c).
internally in models with the “cbind” function.

3.3 | Percentage of CBB infestation per branch


3 | R E S U LT S
Out of the 275 models created with the “dredge” function for CBB
infestation data, there were three models selected as equally in-
3.1 | Provision of CBB control by ants
formative (ΔAICc < 2). The three models included as fixed effects
We monitored 20,308 coffee berries from which 2,217 (c. 10%) were the interaction of ant-­exclusion with 2-­km-­level forest cover and
bored (1,112 in ant-­excluded branches and 1,105 in control branches). distance to the nearest forest fragment (see Supporting Information
Across all 10 landscapes and 17 farms, 47% of the branches had CBB Appendix S2, Figure S7, Tables S2, S5). The percentage of CBB infes-
presence (at least one bored coffee berry) from which 51% were in tation was higher in ant-­excluded branches. Ants provided more CBB
ant-­excluded branches and 43% in control branches. There were 892 infestation control in landscapes with less than 40% 2-­km-­level for-
ants seen in control branches and 57 ants in ant-­excluded branches est cover and after 25 m from forest edges (p < 0.001, Figure 3a,b).
during a total of 30 minutes observations (5 minutes × 6 visits).
The average “ant-­activity” (ants seen) in control branches and ant-­
3.4 | Percentage of CBB damage per coffee bean
excluded branches were 0.55 and 0.04, respectively (p < 0.0001,
see Supporting Information Figure S4). Frequent ant morphospecies Model selection analysis from 275 possible models for CBB damage
visitors in our experiments were identified to the genus level (see directly in coffee beans predicted five models that included as fixed
Supporting Information Table S3). Ants reduced the probability of effects the interaction of ant-­exclusion with 2-­km-­level forest cover
CBB presence and the percentage of CBB damage up to 40%. Out of and ant-­activity (see Supporting Information Appendix S3, Figure
the 2,078 bored berries we opened, coffee beans in ant-­excluded S8, Tables S2, S6). Reduction in CBB damage by ants was more
branches had an average of 51% damage and control branches had intense in landscapes with less than 40% 2-­km-­level forest cover and
41% damage (see Supporting Information Figure S5). decreased as forest cover increased (Figure 4a,b).

3.2 | Presence or absence of CBB per branch


4 | D I S CU S S I O N
For the binary data indicating CBB presence, the “dredge” func-
tion ran 275 possible models and there were three indistinguish- This study presents new evidence that ants regulate CBB popula-
able models selected (ΔAICc < 2) (see Supporting Information tions in sun coffee farms and that landscape structure regulates
Appendix S1, Figure S6, Tables S2, S4). These three models included key processes involved in the provision of this ecosystem service.
as fixed effects the interaction of ant-­exclusion with 2-­km-­level As forest cover increases in 2-­km-­level landscapes, the probability
forest cover, 300-­m-­level forest cover, and 300-­m-­level coffee to detect CBB presence and the percentage of CBB infestation and

F I G U R E 2 CBB presence final model representation (lowest AICc). CBB presence probability per branch in response to the interaction of
ant-­exclusion and (a) 2-­km-­level forest cover, (b) 300-­m-­level forest cover, and (c) 300-­m-­level coffee cover. Shaded areas between curves
show the interaction of exclusion experiment (ants presence/absence) and intensity of pest control provision. Shaded areas in yellow
represent where ants had positive effects and in black negative effects
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
26 | Journal of Applied Ecology ARISTIZÁBAL and METZGER

F I G U R E 3 CBB infestation final model representation (lowest AICc). CBB infestation percentage per branch in response to the interaction
of ant-­exclusion and (a) 2-­km-­level forest cover, (b) distance to the nearest forest fragment, and (c) ant-­activity. Shaded areas between
curves show the interaction of exclusion experiment (ants presence/absence) and intensity of pest control provision. Shaded areas in yellow
represent where ants had positive effects and in black negative effects

F I G U R E 4 CBB damage final model representation (lowest AICc). CBB damage percentage per bored coffee bean in response to the
interaction of ant-­exclusion and (a) 2-­km-­level forest cover and (b) total number of ants seen in 5-­minute visits. Shaded areas between
curves show the interaction of exclusion experiment (ants presence/absence) and intensity of pest control provision. Shaded area in yellow
represents where ants had a positive effect and in black a negative effect

CBB damage decreases. However, ants provide stronger CBB pres- rates were tested using sentinel caterpillar experiments in sun cof-
ence, CBB infestation, and CBB damage regulation in landscapes with fee farms in Kenya and contrary to our findings, the probability
lower forest cover and at c. 40% forest cover in 2-­km-­level, this ant-­ of pest removal decreased as distance to the forest fragment in-
mediated CBB reduction ceases. Ants’ ability to reduce CBB infesta- creased, specifically after 25 m (Milligan, Johnson, Garfinkel, Smith,
tion intensifies beyond 25 m from forest edges. As coffee cover and & Njoroge, 2016). This could suggest there are different ecological
forest cover increases in 300-­m-­level landscapes, ants represent up processes involved in the reduction of CBB infestation in Brazilian
to 40% less CBB presence detectability. sun coffee farms other than predation or prey removal by ants such
as non-­consumptive attacks or indirect chemical deterrence (Morris
et al., 2018). This could also indicate that ants receive higher pre-
4.1 | Proximity to forest fragments and pest control
dation pressures themselves near forest fragments (Librán-­Embid
by ants
et al., 2017) and hence are less able to reduce CBB infestation.
Proximity to forest fragments typically enhances pest control ser-
vices, mainly due to spillover edge effects from forests adjacent ag-
4.2 | Coffee cover and pest control by ants
ricultural areas (Boesing et al., 2017; Rand, Tylianakis, & Tscharntke,
2006). However, in Southeastern Brazilian sun coffee farms, ant-­ CBB presence was partially explained by the amount of coffee cover
mediated CBB infestation control seems to increase as the distance around a 300-­m-­level, as observed in other studies (Avelino, Romero-­
from forest fragments increases. These results indicate that inside Gurdián, Cruz-­Cuellar, & Declerck, 2012; De la Mora et al., 2015).
the coffee matrix ants are considerably important to prevent fur- The probability to detect CBB presence increased as 300-­m-­level
ther CBB infestation once they have established. Ant prey removal coffee cover increased, where ants reduced up to 40% CBB presence
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ARISTIZÁBAL and METZGER Journal of Applied Ecology | 27

when there is more than 60% coffee cover. Accordingly, in a study landscapes with >30%–35% forest cover (Banks-­Leite et al., 2014).
in Mexican coffee farms, ants removed more CBB as the number Interactions among different functional guilds, or natural enemies
of coffee plants increased (De la Mora et al., 2015). Avelino et al. with common prey, may have negative impacts on pest control in
(2012) showed that increasing coffee cover raised the abundance of landscapes with increasing complexity, or >25% semi-­natural habi-
CBB and other coffee pests in low-­shade coffee farms in Costa Rica. tat (Martin, Reineking, Seo, & Steffan-­D ewenter, 2013). Thus, we
Coffee expansion may offer more area for CBB colonization, but it is suggest that in our study ants’ regulation of CBB at the 2-­k m-­level
also providing more opportunities for CBB removal by ants that nest was an indirect response to insectivorous bird abundances and
and forage in the coffee matrix. predating pressures on ants.
Farmers indicated in interviews that the major source of CBB col- Forest cover in 300-­m-­level was selected as an important pre-
onization is fallen bored berries not harvested in the previous year. dictor of CBB presence control by ants. Contrary to what we found
Most Brazilian farms in our study did not sweep between coffee for forest cover at 2-­km-­level, the probability to detect CBB pres-
plants after harvesting to prevent fallen bored berries to propagate ence in the absence of ants increased as forest cover in 300-­m-­level
new CBB populations. Perhaps, the bigger the area transformed into landscapes increased. This coincides with previous findings in
coffee crops, the more bored berries that fall to the ground and the 200-­m-­level studies (De la Mora et al., 2013) along with our expec-
more chances of CBB dispersing and colonizing. Thus, in 300-­m-­level tations because forest fragments generally offer habitat and other
landscapes, ground-­foraging ant species that help prey on bored resources to ants and natural enemies such as insectivorous birds
fallen berries year-­round (Armbrecht & Gallego, 2007; Morris et al., (Boesing et al., 2017). Throughout our entire field season, even be-
2018) may be important to regulate CBB presence in sun coffee farms. fore coffee berries were growing, we observed ants foraging and
Two of the five models better explaining CBB damage also in- nesting in the coffee matrix. The most commonly observed ant gen-
cluded 300-­m-­level coffee cover as a predictive variable. There was era in our study have been associated with human-­disturbed hab-
some evidence that ants reduce the percentage of CBB damage re- itats and are omnivorous generalists (Fernández, 2003). However,
gardless of the amount of coffee cover and this trend strengthened ants involved in our experiments, and hence in pest control, may not
as the coffee crop area grew. Altogether, in our study, coffee expan- necessarily be forest-­dependent and that may be why 300-­m-­level
sion hinted at two effects of pest control services by ants: less CBB forest cover did not predict an association for CBB infestation or CBB
presence and CBB damage. This could mean that some ant species damage. On the other hand, 2-­km-­level forest cover predicted asso-
involved in preventing CBB colonization are also small enough to for- ciations for every response variable in our study (CBB presence, CBB
age in CBB galleries. It seems that these ant species that adapt well infestation, and CBB damage). Forest cover in larger landscape-­levels
to the sun coffee matrix are dominating the local ant community. may change the abundance and behaviour of other CBB predators,
such as birds spilling over from forest fragments into coffee farms
(Boesing et al., 2017; Kellermann, Johnson, Stercho, & Hackett,
4.3 | Forest cover and pest control by ants
2008). Hence, affecting ant communities and their provision of eco-
The relationships among ants, pest control, and forest cover are system services. We found conflicting CBB presence results associ-
more complex and far more interesting than we had predicted. The ated with forest cover at the 300-­m and 2-­km levels. This highlights
2-­k m-­level landscapes seem to have a threshold in pest control the importance of considering multi-­scale studies to plan agricul-
provision around 40% forest cover. The probability of CBB presence tural landscapes. There are different ecological processes involved
decreases with more forest area, but there is more CBB presence at each level and different conservation decisions that could be de-
detection in ant-­excluded branches only until this threshold. Also, rived from them.
ant-­excluded branches have less CBB infestation in landscapes
with lower forest cover. Likewise, ants and 2-­k m-­level forest cover
4.4 | Ants contribution to pest control
are likely predictors of CBB damage prevention. Independently of
the processes behind the arrival of CBB, once a coffee berry is The most commonly observed genera in our experimental sites
colonized by CBB, and bigger natural enemies (such as birds) can were Linepithema, Pheidole, Solenopsis, Crematogaster, Camponotus,
no longer predate it, ants become an even more important source Neoponera, and Brachymyrmex, which were also found to be linked
of pest control. Coffee berries had less CBB damage in landscapes with coffee agroecosystems in previous studies (Armbrecht &
with lower forest area, but ants reduced bean damage only until Gallego, 2007; Armbrecht & Perfecto, 2003; De la Mora & Philpott,
around 40% forest cover in 2-­k m-­level landscapes. To explain this 2010; Perfecto & Vandermeer, 2002). Ant species involved in the
unexpected result, we suggest that above 40% forest cover the suppression of CBB presence may not be the same as the species
community of species in the landscape shifts and predatory pres- that regulate CBB infestation or CBB damage. Presumably, the most
sures on ants increase. A study in sun coffee farms in Costa Rica territorial species out of these genera would prevent more CBB
showed bird abundances were higher with more forest cover (Karp presence and CBB infestation and other smaller and arboreal or
et al., 2013), potentially increasing predation on ants. Particularly ground-­foraging species would reduce more CBB damage. Out of
for this study, in the Atlantic Forest, bird communities that could the ant genera commonly observed in our experimental sites, spe-
be influencing ant populations, apparently recover and stabilize in cies from Linepithema, Pheidole, and Solenopsis are small enough to
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
28 | Journal of Applied Ecology ARISTIZÁBAL and METZGER

fit through CBB galleries and could potentially predate CBB inside AC K N OW L E D G E M E N T S
coffee berries. Individuals from the genus Solenopsis were found
We thank farm owners and workers for their fundamental
on several occasions inside bored coffee berries (authors’ obser-
cooperation. We would like to acknowledge Inara R. Leal and
vations). Evaluating CBB damage from field collected samples is
Sebastián F. Sendoya for their input on the experimental design
not typically utilized to quantify pest control. Using CBB damage
and contributions throughout this project. Also, we thank Carla
as an indicator of pest control was fundamental to better under-
R. Ribas and Paulo S. Oliveira for valuable discussions over the
stand this system. Pest control provided by arboreal and ground-­
first version of this manuscript, Gustavo Bravo for his support,
foraging ants is not restricted to preventing CBB colonization or
Danielle Rock for English revisions, and Sheina Koffler for
decreasing infestation. It also minimizes CBB damage directly in
help with statistical analyses. This study also benefitted from
coffee beans, which is ultimately what will profit farmers.
teamwork from all members of the Interface Project, particularly
from Guilherme Prata Gonçalves with ant identifications (SISBIO/
4.5 | Conclusion and management implications ICMBio permit 48059), and from Adrian Gonzalez, Elizabeth
Nichols, Francisco d’Albertas, Greet De Coster, Larissa Boesing,
Sun coffee is usually considered “low-­quality habitat” and sometimes
and Leandro R. Tambossi with assistance in spatial and statistical
even disregarded for ecosystem services. We show here that this is
analyses. Finally, we would like to thank assistance in the field,
not the case in Southeastern Brazilian sun coffee farms. This study
especially from Emilien Rottier. We are grateful to the São Paulo
highlights the significance of ants in pest control by reducing CBB
Research Foundation, which supported this study embedded
presence, infestation, and damage through their interaction with the
in funding for Interface Project (FAPESP, 2013/23457-­6). The
surrounding landscape. This is the first long-­term branch-­level ant-­
Brazilian Ministry of Education supported N.A. (CAPES-­DS;
exclusion study in sun coffee farms and it conveys new information
2014–2016). The Brazilian National Council for Scientific and
that allows us to propose useful landscape management recommen-
Technological Development funded J.P.M.
dations that could reduce CBB and increase coffee yield in one of
the world’s most economically important regions for coffee produc-
tion. These recommendations reinforce ecological CBB control while AU T H O R S ’ C O N T R I B U T I O N S
contributing to the protection of the endangered Atlantic Forest.
N.A. and J.P.M. conceived the project and wrote the manuscript.
Increasing forest cover by promoting forest restoration within pas-
J.P.M. obtained funding. N.A. collected and analysed the data.
ture areas and allocating forest near to coffee crops could lower the
probability of CBB presence. Guaranteeing 35%–40% of forest cover
in a 2-­km-­radius around coffee crops will maintain coffee branches DATA AC C E S S I B I L I T Y
with less CBB presence, CBB infestation, and CBB damage, while also
maintaining forest cover above extinction thresholds for several ver- Data available via the Dryad Digital Repository https://doi.
tebrates and plant groups from the Atlantic Forest (Lima & Mariano-­ org/10.5061/dryad.3rh2043 (Aristizábal & Metzger, 2018).
Neto, 2014; Martensen, Ribeiro, Banks-Leite, Prado, & Metzger,
2012; Pardini, Bueno, Gardner, Prado, & Metzger, 2010; Rigueira, da ORCID
Rocha, & Mariano-Neto, 2013).
We recognize CBB colonization is also affected by other fac- Natalia Aristizábal http://orcid.org/0000-0002-1001-8824
tors not tested in this study, such as farm management, humidity,
and local temperature. We advise incorporating these variables
REFERENCES
as local factors in future studies, as well as more variables that
measure connectivity processes and matrix quality (Boesing et al., Aristizábal, N., & Metzger, J. P. (2018). Data from: Landscape structure
2017). Also, we observed in our study different landscape and regulates pest control provided by ants in sun coffee farms. Dryad
Digital Repository, https://doi.org/10.5061/dryad.3rh2043
ecological processes acting on several organisms involved in the
Armbrecht, I., & Gallego, M. C. (2007). Testing ant predation on the cof-
provision of pest control. We recommend the inclusion of ecolog- fee berry borer in shaded and sun coffee plantations in Colombia.
ical networks data because the provision of ecosystem services The Netherlands Entomological Society, 124, 261–267.
is mostly the result of an interaction of organisms. We argue that Armbrecht, I., & Perfecto, I. (2003). Litter-­t wig dwelling ant species rich-
landscapes composed of sun coffee farms, surrounded by c. 35% ness and predation potential within a forest fragment and neigh-
boring coffee plantations of contrasting habitat quality in Mexico.
forest cover, can sustain ecological processes capable of provid-
Agriculture, Ecosystems and Environment, 97, 107–115. https://doi.
ing CBB control and other ecosystem services, such as pollination org/10.1016/S0167-8809(03)00128-2
(Saturni et al., 2016) and “coffee leaf miner” (Leucoptera coffeella) Avelino, J., Romero-Gurdián, A., Cruz-Cuellar, H. F., & Declerck, F. J.
control (Librán-­Embid et al., 2017). Thus, we emphasize the impor- (2012). Landscape context and scale differentially impact coffee leaf
rust, coffee berry borer, and coffee root-­knot nematodes. Ecological
tance of landscape planning in sun coffee agricultural landscapes
Applications, 22, 584–596. https://doi.org/10.1890/11-0869.1
to help bridge the gap between biodiversity conservation and cof- Banks-Leite, C., Pardini, R., Tambosi, L. R., Pearse, W. D., Bueno, A. A.,
fee production. Bruscagin, R. T., … Metzger, J. P. (2014). Using ecological thresholds to
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ARISTIZÁBAL and METZGER Journal of Applied Ecology | 29

evaluate the costs and benefits of set-­asides in a biodiversity hotspot. Ferraz, S. F. B., Ferraz, K. M. P. M. B., Cassiano, C. C., Brancalion, P.
Science, 345, 1041–1045. https://doi.org/10.1126/science.1255768 H. S., da Luz, D. T. A., Azevedo, T. N., … Metzger, J. P. (2014). How
Bartoń, K. (2018). MuMIn: Multi-model inference. R package version 1.42.1. good are tropical forest patches for ecosystem services provision-
Retrieved from https://CRAN.R-project.org/package=MuMIn ing? Landscape Ecology, 29, 187–200. https://doi.org/10.1007/
Bates, D., Mächler, M., Bolker, B. M., & Walker, S. C. (2015). Fitting linear s10980-014-9988-z
mixed-­effects models using lme4. Journal of Statistical Software, 67, 1. Foley, J. A., Ramankutty, N., Brauman, K. A., Cassidy, E. S., Gerber,
https://doi.org/10.18637/jss.v067.i01 J. S., Johnston, M., … Zaks, D. P. M. (2011). Solutions for a culti-
Bianchi, F. J., Booij, C. J., & Tscharntke, T. (2006). Sustainable pest reg- vated planet. Nature, 478, 337–342. https://doi.org/10.1038/
ulation in agricultural landscapes: A review on landscape composi- nature10452
tion, biodiversity and natural pest control. Proceedings of the Royal Folgarait, P. J. (1998). Ant biodiversity and its relationship to ecosystem
Society: Biological Sciences, 273, 1715–1727. https://doi.org/10.1098/ functioning: A review. Biodiversity and Conservation, 7, 1221–1244.
rspb.2006.3530 https://doi.org/10.1023/A:1008891901953
Boesing, A. L., Nichols, E., & Metzger, J. P. (2017). Effects of landscape Food and Agriculture Organization of the United Nations, Rome. (2016).
structure on avian-­mediated pest control services. Landscape http://www.fao.org/statistics/en/
Ecology, 32, 931–944. https://doi.org/10.1007/s10980-017-0503-1 Gonthier, D. J., Ennis, K. K., Philpott, S. M., Vandermeer, J., & Perfecto, I.
Boesing, A. L., Nichols, E., & Metzger, J. P. (2017). Land use type, for- (2013). Ants defend coffee from berry borer colonization. BioControl,
est cover, and forest edges modulate avian cross-­habitat spill- 58, 815–820. https://doi.org/10.1007/s10526-013-9541-z
over. Journal of Applied Ecology, 55, 1252–1264. https://doi. Hölldobler, B., & Wilson, E. O. (1990). The ants. Chapter 1. The importance
org/10.1111/1365-2664.13032 of ants. Cambridge, MS: Harvard University Press.
Brazilian Agricultural Research Corporation (EMBRAPA). (2016). Ministry Hölldobler, B., & Wilson, E. O. (1990). The ants. Chapter 3. The colony life
of Agriculture, Livestock, and Food Supply. Retrieved from http://www. cycle. Cambridge, MS: Harvard University Press.
embrapa.br Jaramillo, J., Borgemeister, C., & Baker, P. (2006). Coffee berry borer
Brazilian Agricultural Research Corporation (EMBRAPA). (2018). Ministry Hypothenemus hampei (Coleptera: Curculionidae): Searching for
of Agriculture, Livestock, and Food Supply. Retrieved from http://www. sustainable control strategies. Bulletin of Entomological Research, 96,
embrapa.br 223–233. https://doi.org/10.1079/BER2006434
Bunn, C., Läderach, P., Ovalle Rivera, O., & Kirschke, D. (2015). A bit- Jha, S., Bacon, C. M., Philpott, S. M., Méndez, V. E., Läderach, P., & Rice,
ter cup: Climate change profile of global production of Arabica R. A. (2014). Shade coffee: Update on a disappearing refuge for bio-
and Robusta coffee. Climatic Change, 129, 89–101. https://doi. diversity. BioScience, 64, 416–428. https://doi.org/10.1093/biosci/
org/10.1007/s10584-014-1306-x biu038
Bustillo, A. E., Cárdenas, R., & Posada, F. J. (2002). Natural enemies Jiménez-Soto, E., Cruz-Rodríguez, J. A., Vandermeer, J., & Perfecto,
and competitors of Hypothenemus hampei (Ferrari) (Coleoptera: I. (2013). Hypothenemus hampei (Coleoptera: Curculionidae) and
Scolytidae) in Colombia. Neotropical Entomology, 31, 635–639. its interactions with Azteca instabilis and Pheidole synanthropica
https://doi.org/10.1590/S1519-566X2002000400018 (Hymenoptera: Formicidae) in a Shade Coffee Agroecosystem.
Cardinale, B. J., Duffy, E., Gonzalez, A., Hooper, D. U., Perrings, C., Venail, Environmental Entomology, 42, 915–924. https://doi.org/10.1603/
P., … Naeem, S. (2012). Biodiversity loss and its impact on humanity. EN12202
Nature, 486, 59–67. https://doi.org/10.1038/nature11148 Joly, C. A., Metzger, J. P., & Tabarelli, M. (2014). Experiences from the
Carrol, C. R., & Janzen, D. H. (1973). Ecology of foraging by ants. Brazilian Atlantic Forest: Ecological findings and conservation ini-
Annual Review of Ecology and Systematics, 4, 231–257. https://doi. tiatives. New Phytologist, 204, 459–473. https://doi.org/10.1111/
org/10.1146/annurev.es.04.110173.001311 nph.12989
Classen, A., Peters, M. K., Ferger, S. W., Helbig-Bonitz, M., Schmack, J. Karp, D. S., Mendenhall, C. D., Figueroa Sandí, R., Chaumont, N., Hadly,
M., Maassen, G., … Steffan-Dewenter, S. (2014). Complementary E. A., & Daily, G. C. (2013). Forest bolsters bird abundance, pest
ecosystem services provided by pest predators and pollinators in- control and coffee yield. Ecology Letters, 16, 1339–1347. https://doi.
crease quantity and quality of coffee yields. Proceedings of the org/10.1111/ele.12173
Royal Society B: Biological Sciences, 281, 20133148. https://doi. Karungi, J., Nambi, N., Ijala, A. R., Jonsson, M., Kyamanywa, S., & Ekbom,
org/10.1098/rspb.2013.3148 B. (2014). Relating shading levels and distance from natural vege-
Damon, A. (2000). A review of the biology and control of the coffee tation with hemipteran pests and predators occurrence on coffee.
berry borer, Hypothenemus hampei (Coleoptera: Scolytidae). Bulletin Journal of Applied Entomology, 139, 669–678.
of Entomological Research, 90, 453–465. Kellermann, J. L., Johnson, M. D., Stercho, A. M., & Hackett, S. C. (2008).
De la Mora, A., García-Ballinas, J. A., & Philpott, S. (2015). Local, landscape, Ecological and economic services provided by birds on Jamaican
and diversity drivers of predation services provided by ants in a coffee Blue Mountain coffee farms. Conservation Biology, 22(5), 1177–1185.
landscape in Chiapas, Mexico. Agriculture, Ecosystems & Environment, https://doi.org/10.1111/j.1523-1739.2008.00968.x
201, 83–91. https://doi.org/10.1016/j.agee.2014.11.006 Larsen, A., & Philpott, S. M. (2010). Twig-­nesting ants: The hidden pred-
De la Mora, A., Murnen, C. J., & Philpott, S. M. (2013). Local and land- ators of the coffee berry borer in Chiapas, Mexico. Biotropica, 42,
scape drivers of biodiversity of four groups of ants in coffee land- 342–347. https://doi.org/10.1111/j.1744-7429.2009.00603.x
scapes. Biodiversity and Conservation, 22, 871–888. https://doi. Librán-Embid, F., De Coster, G., & Metzger, J. P. (2017). Effects of bird
org/10.1007/s10531-013-0454-z and bat exclusion on coffee pest control at multiple spatial scales.
De la Mora, A., & Philpott, S. (2010). Wood-­nesting ants and their Landscape Ecology, 32(9), 1907–1920. https://doi.org/10.1007/
parasites in forests and coffee agroecosystems. Environmental s10980-017-0555-2
Entomology, 39, 1473–1481. https://doi.org/10.1603/EN09295 Lima, M. M., & Mariano-Neto, E. (2014). Extinction thresholds for
Fair Trade Foundation. (2012). Fair trade and coffee: Commodity briefing. Sapotaceae due to forest cover in Atlantic Forest landscapes. Forest
Retrieved from http://www.fairtrade.net/fileadmin/user_upload/ Ecology and Management, 312, 260–270. https://doi.org/10.1016/j.
content/2009/resources/2012_Fairtrade_and_coffee_Briefing.pdf. foreco.2013.09.003
Fernández, F. (2003). Introducción a las Hormigas de la Región Neotropical. Martensen, A. C., Ribeiro, M. C., Banks-Leite, C., Prado, P. I., & Metzger,
Bogotá, Colombia: Instituto de Investigación de Recursos Biológicos J. P. (2012). Associations of forest cover, fragment area, and con-
Alexander von Humboldt. nectivity with neotropical understory bird species richness and
13652664, 2019, 1, Downloaded from https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.13283 by CAPES, Wiley Online Library on [09/06/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
30 | Journal of Applied Ecology ARISTIZÁBAL and METZGER

abundance. Conservation Biology, 26(6), 1100–1111. https://doi. Ribeiro, M. C., Metzger, J. P., Martensen, A. C., Ponzoni, F. J., & Hirota, M.
org/10.1111/j.1523-1739.2012.01940.x M. (2009). The Brazilian Atlantic Forest: How much is left, and how
Martin, E. A., Reineking, B., Seo, B., & Steffan-Dewenter, I. (2013). is the remaining forest distributed? Implications for conservation.
Natural enemy interactions constrain pest control in complex agri- Biological Conservation, 142, 1141–1153. https://doi.org/10.1016/j.
cultural landscapes. Proceedings of the National Academy of Sciences, biocon.2009.02.021
110, 5534–5539. https://doi.org/10.1073/pnas.1215725110 Ricci, M. S. F., Araújo, M. C. F., & Franch, C. M. C. (2002). Cultivo orgânico
McGarigal, K., Cushman, S. A., & Ene, E. (2012). FRAGSTATS v4.2: do café: recomendações técnicas. Brasilia: Embrapa Informação
Spatial pattern analysis program for categorical and continuous Tecnológica (72–79), Rio de Janeiro.
maps. Computer software program produced by the authors at the Rigueira, D. M. G., da Rocha, P. L. B., & Mariano-Neto, E. (2013).
University of Massachusetts, Amherst. Forest cover, extinction thresholds and time lags in woody plants
Milligan, M. C., Johnson, M. D., Garfinkel, M., Smith, C. J., & Njoroge, (Myrtaceae) in the Brazilian Atlantic Forest: Resources for conser-
P. (2016). Quantifying pest control services by birds and ants in vation. Biodiversity and Conservation, 22(13–14), 3141–3163. https://
Kenyan coffee farms. Biological Conservation, 194, 58–65. https:// doi.org/10.1007/s10531-013-0575-4
doi.org/10.1016/j.biocon.2015.11.028 Saturni, F., Jaffe, R., & Metzger, J. P. (2016). Landscape structure in-
Mitchell, M. G. E., Suarez-Castro, A. F., Martinez-Harms, M., Maron, fluences bee community and coffee pollination at different spatial
M., McAlpine, C., Gaston, K. J., … Rhodes, J. R. (2015). Reframing scales. Agriculture, Ecosystems & Environment, 235, 1–12. https://doi.
landscape fragmentation’s effects on ecosystem services. Trends org/10.1016/j.agee.2016.10.008
in Ecology & Evolution, 30, 190–198. https://doi.org/10.1016/j. Tscharntke, T., Sekercioglu, C. H., Dietsch, T. V., Sodhi, N. S., Hoehn, P.,
tree.2015.01.011 & Tylianakis, J. M. (2008). Landscape constraints on functional di-
Morris, J. R., Jiménez-Soto, E., Philpott, S. M., & Perfecto, I. (2018). Ant-­ versity of birds and insects in tropical agroecosystems. Ecology, 89,
mediated (Hymenoptera: Formicidae) biological control of the coffee 944–951. https://doi.org/10.1890/07-0455.1
berry borer: Diversity, ecological complexity, and conservation bio- USDA. (2015). PSD online – Home. In production, supply and distribution
control. Myrmecological News, 26, 1–17. online. Retrieved from http://www.fas.usda.gov/psdonline/
Morris, J. R., & Perfecto, I. (2016). Testing the potential for ant predation Vandermeer, J., & Perfecto, I. (2007). The agricultural matrix and a fu-
of immature coffee berry borer (Hypothenemus hampei) life stages. ture paradigm for conservation. Conservation Biology, 21, 274–277.
Agriculture, Ecosystems and Environment, 233, 224–228. https://doi. https://doi.org/10.1111/j.1523-1739.2006.00582.x
org/10.1016/j.agee.2016.09.018 Vega, F. E., Infante, F., Castillo, A., & Jaramillo, J. (2009). The coffee berry
Myers, N., Mittermeier, R. A., Mittermeier, C. G., Da Fonseca, G. A., & borer, Hypothenemus hampei (Ferrari) (Coleoptera: Curculionidae):
Kent, J. (2000). Biodiversity hotspots for conservation priorities. A short review, with recent findings and future research directions.
Nature, 403, 853–858. https://doi.org/10.1038/35002501 Terrestrial Arthropod Reviews, 2, 129–147.
Oliveira, C. M., Auad, A. M., Mendes, S. M., & Frizzas, M. R. (2013). Verburg, P. H., Erb, K., Mertz, O., & Espindola, G. (2013). Land system sci-
Economic impact of exotic insect pests in Brazilian agriculture. ence: Between global challenges and local realities. Current Opinion
Journal of Applied Entomology, 137, 1–15. https://doi.org/10.1111/ in Environmental Sustainability, 5, 433–437. https://doi.org/10.1016/j.
jen.12018 cosust.2013.08.001
Pardini, R., Bueno, A. D. A., Gardner, T. A., Prado, P. I., & Metzger, J. P. Zuur, A. F. (2009). Mixed effects models and extensions in ecology
(2010). Beyond the fragmentation threshold hypothesis: Regime with R. New York, NY/London, UK: Springer, 574 pp. https://doi.
shifts in biodiversity across fragmented landscapes. PLoS ONE, 5(10), org/10.1007/978-0-387-87458-6
e13666. https://doi.org/10.1371/journal.pone.0013666
Perfecto, I., & Vandermeer, J. (2002). Quality of agroecological ma-
trix in a tropical montane landscape: Ants in coffee plantations in
S U P P O R T I N G I N FO R M AT I O N
Southern Mexico. Conservation Biology, 16, 174–182. https://doi.
org/10.1046/j.1523-1739.2002.99536.x Additional supporting information may be found online in the
Philpott, S. M., & Armbrecht, I. (2006). Biodiversity in tropical agro-
Supporting Information section at the end of the article.
forests and the ecological role of ants and ant diversity in pred-
atory function. Ecological Entomology, 31, 369–377. https://doi.
org/10.1111/j.1365-2311.2006.00793.x
R Core Team. (2018). R: A language and environment for statistical com- How to cite this article: Aristizábal N, Metzger JP. Landscape
puting. Vienna, Austria: R Foundation for Statistical Computing. structure regulates pest control provided by ants in sun coffee
Retrieved from https://www.R-project.org/
farms. J Appl Ecol. 2019;56:21–30. https://doi.
Rand, T., Tylianakis, J. M., & Tscharntke, T. (2006). Spillover edge effects:
The dispersal of agriculturally subsidized insect natural enemies into org/10.1111/1365-2664.13283
adjacent natural habitats. Ecology Letters, 9, 603–614. https://doi.
org/10.1111/j.1461-0248.2006.00911.x

You might also like