See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236191524

Effect of Two Agroecological Management Strategies on Ant

(Hymenoptera: Formicidae) Diversity on Coffee Plantations in
Southwestern Colombia

Article in Environmental Entomology · April 2013

DOI: 10.1603/EN11084 · Source: PubMed

CITATIONS READS

37 329

2 authors:

María Ximena Urrutia Escobar Inge Armbrecht

Independent Universidad del Valle (Colombia)
2 PUBLICATIONS 37 CITATIONS 150 PUBLICATIONS 5,219 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Inge Armbrecht on 23 May 2014.

The user has requested enhancement of the downloaded file.

 Effect of Two Agroecological Management Strategies on Ant
(Hymenoptera: Formicidae) Diversity on Coffee Plantations in
Southwestern Colombia
Author(s): M. X. Urrutia-Escobar and I. Armbrecht
Source: Environmental Entomology, 42(2):194-203. 2013.
Published By: Entomological Society of America
URL: http://www.bioone.org/doi/full/10.1603/EN11084

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the

biological, ecological, and environmental sciences. BioOne provides a sustainable online
platform for over 170 journals and books published by nonprofit societies, associations,
museums, institutions, and presses.
Your use of this PDF, the BioOne Web site, and all posted and associated content indicates
your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.
Usage of BioOne content is strictly limited to personal, educational, and non-commercial
use. Commercial inquiries or rights and permissions requests should be directed to the
individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers,
academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research.
 COMMUNITY AND ECOSYSTEM ECOLOGY

Effect of Two Agroecological Management Strategies on Ant

(Hymenoptera: Formicidae) Diversity on Coffee Plantations
in Southwestern Colombia
M. X. URRUTIA-ESCOBAR1,2 AND I. ARMBRECHT3

Environ. Entomol. 42(2): 194Ð203 (2013); DOI: http://dx.doi.org/10.1603/EN11084

ABSTRACT SimpliÞcation of agroecosystems because of industrialization of agriculture may cause
the loss of associated animal biodiversity of both vertebrates and invertebrates. To measure how the
agricultural intensiÞcation on coffee plantations affects ant biodiversity, we intensively sampled ants
in Caldono (Cauca, Colombia). We surveyed 15 sites classiÞed into three management types: sun
coffee plantations, shaded coffee plantations, and forest patches. Fifteen 50-m linear transects, each
one consisting of 5 pitfall traps and 5 tuna baits, were set at each sampling location between December
of 2009 and February of 2010. We collected 18,186 ants that represent 82 ant species, 34 genera, and
9 subfamilies of Formicidae (Hymenoptera: Formicidae). The management intensiÞcation index
showed an increasing intensiÞcation gradient along the 15 sampling locations from forest patches to
shaded coffee to sun coffee plantations. Shaded coffee plantations harbored the highest number of
species (60), followed by forest (56) and sun coffee (33). Ant species composition and plant structure
on shaded coffee plantations resembled the forest patches more than the sun coffee plantations. Forest
and shaded coffee plantations had a more equitable distribution of ant species, whereas in sun coffee
plantations, Linepithema neotropicum (Emery) and Ectatomma ruidum (Roger) typically outnum-
bered all other ant species. Evidence from functional groups indicated that speciÞc habitat and feeding
requirements exist among the species that are found together. Our results conÞrmed that intensiÞ-
cation of agriculture negatively affects ant diversity, despite the fact that farms were located in a
heterogeneous landscape, suggesting that agroecological management is a strong determinant in the
conservation of wild fauna.

KEY WORDS agroecological strategy, ant community, coffee plantation, management intensiÞca-
tion index

Coffee is traditionally grown under a canopy of shade
trees (Perfecto et al. 1996). However, the National
Federation of Coffee Growers in Colombia (FNC
2010a) reported on how the planted area has changed
in the last 2 yr. An increase of modern or sun coffee
plantations (20,000 ha), contrasts with an evident de-
crease of the traditional or shaded coffee plantations
(10,000 ha). This conversion process, from shaded to
sun coffee farms, happened very fast. Colombia has
the largest cover of nonshaded coffee plantations in
northern Latin America (Perfecto et al. 1996, Arm-
brecht et al. 2005).
According to Philpott et al. (2006), the increase in
intensive agricultural practices drives the loss of bio-
diversity. This increase can have drastic consequences
for ecosystem services (e.g., predation, pollination,
1 Institute for Biodiversity and Ecosystems Dynamics, University of
Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands.
2 Corresponding author: Marṍa Ximena Urrutia-Escobar, Departa-
mento de Biologṍa, Universidad del Valle- Meléndez, Calle 13 No
100-00. Cali, Colombia (e-mail: mxurrutia@gmail.com).
3 Departamento de Biologṍa, Universidad del Valle, Apartado Aéreo
25360, Cali, Colombia.
seed dispersal), especially in tropical areas that have
been hard hit by deforestation. The use of alternative
management strategies in the agricultural matrix plays
an important role in the conservation of biodiversity.
In recent years, several authors have tested the idea
that shaded coffee plantations (acting as diverse agro-
ecosystems) are likely to be a critical refuge, providing
a high-quality matrix within which biodiversity is con-
served, somehow resembling a “good habitat” for for-
est species moving between forest fragments (Per-
fecto et al. 1996, Armbrecht and Perfecto 2003,
Gallego-Ropero 2005, Philpott et al. 2006).
Shaded coffee plantations represent complex agri-
cultural lands in terms of vegetation, some of which
incorporate high diversity, density, and different lev-
els of arboreal strata that can maintain levels of bio-
diversity similar to nearby native forests (Moguel and
Toledo 1999, Armbrecht et al. 2005, Philpott et al.
2006). The canopyÕs structural complexity affects the
microclimate of the coffee understory, and offers liv-
ing and nesting sites for a variety of organisms (Per-
fecto et al. 1996). They also offer sustainable economic
beneÞts to farmers (Moguel and Toledo 1999, Van-

0046-225X/13/0194Ð0203$04.00/0 䉷 2013 Entomological Society of America

April 2013 URRUTIA-ESCOBAR AND ARMBRECHT: COLOMBIAN COFFEE ANTS 195

dermeer and Carvajal 2001). In addition, the diversity

and complexity of habitat and feeding resources of-
fered by shaded coffee plantations may provide some
degree of niche diversiÞcation (Armbrecht et al. 2004)
in which generalist and specialist ants may coexist.
Ants (Hymenoptera: Formicidae) are known to be
highly sensitive to habitat change (Armbrecht et al.
2005). Nowadays, they are particularly useful in agro-
ecological research, because ant assemblages include
tramp species, forest and predatory specialists that are
sensitive to habitat change, showing strong negative
responses to agroecosystem intensiÞcation (Philpott
and Armbrecht 2006). Ants are a major focus for test-
ing hypotheses regarding species richness, community
dynamics, mutualisms, and invasions (Philpott and
Armbrecht 2006). IntensiÞcation processes mainly af-
fect the associated shade component. However, the
disappearance of other habitat variables such as leaf
litter, twigs, or plant cavities may increase nest-site
limitations of many ant species (Wielgoss et al. 2010).
In this sense, multistrata agroforestry systems such
as shaded coffee plantations can be used as templates
for sustainable and environmental friendly agricul-
tural production with a high potential to preserve
associated biodiversity (Armbrecht et al. 2005). Thus,
it is important to examine the conservation value of
shaded coffee plantations in other contexts in Colom-
bia as well as the variables that most inßuence the ant
distribution patterns. Because functional ant groups
(guilds) respond differently to the intensiÞcation pro-
cess on coffee plantations, it is also appropriate to use
them to measure the impact of agricultural intensiÞ- Fig. 1. Distribution of 15 sampling locations in Caldono,
cation on associated fauna instead of using only spe- Colombia.
cies identity (Rivera and Armbrecht 2005). In turn,
this will also contribute to more understanding of how
the transformation from shade to sun coffee planta- producer in Colombia with 74,000 ha (FNC 2010b). By
tions affects ant diversity and ecosystem services. 2009, shaded coffee plantations dominated the area
To understand how agricultural intensiÞcation pro- covering 72% of the planted area, the remaining area
cesses may affect the associated ant fauna, we mea- (20,500 ha) was cover by sun coffee plantations (FNC
sured different agroecological management strategies 2010b).
on southwestern Colombian coffee plantations. We We visited the villages of Pescador, El Descanso,
investigated whether ant diversity, species composi- and El Rosal. The farmers were asked about manage-
tion and functional assemblages were affected by the ment strategies and main characteristics of their coffee
intensiÞcation gradient on coffee plantations. We also plantations. Based on their information and the loca-
determined important habitat characteristics associ- tion of the sites, a classiÞcation of farms and forest
ated with differences in species distributions. patches was divided into three management catego-
ries: 5 shaded coffee plantations (Sh), 5 sun coffee
plantations (S), and 5 forest patches (F) (Table 1). To
Materials and Methods
avoid a “site effect,” we tried to maximize the spatial
Study Area. Intensive samplings were carried out mixture of the three categories in the study area,
between December of 2009 and March of 2010 in the which was ⬇15 km2. The studied coffee plantations
municipality of Caldono (2⬚ 49⬘ N, 076⬚ 32⬘ W) in the ranged between 3 and 6 yr of age.
Department of Cauca (Fig. 1). The average annual The forests were isolated patches of secondary nat-
temperature is 21.5⬚C. Annual precipitation is 2191 ural vegetation, not managed for any type of agricul-
mm, distributed in a bimodal pattern with two high ture, with closed canopies ⬇10 m above the ground.
rain peaks in AprilÐMay and OctoberÐNovember No agrochemical inputs were used at any of the
(Pardo-Lorcano et al. 2003). Altitude ranges from shaded coffee farms. In addition, they contained sev-
1,000 to 2,000 m at sea level (asl). The area is pre- eral tree species, particularly fruit trees such as guamo
dominantly agricultural, Caldono being the leading (Inga edulis Martius), tangerine (Citrus reticulate
producer of sisal, although coffee and cassava produc- Blanco), lime [Citrus aurantifolia (Christmas) Swin-
tion has increased in recent years (Pardo-Lorcano et gle], avocado (Persea americana Miller), and mango
al. 2005). The Department of Cauca is the sixth coffee (Mangifera indica L.) among others. All sun coffee
196 ENVIRONMENTAL ENTOMOLOGY Vol. 42, no. 2

Table 1. Description of the 15 studied locations in Caldono, Colombia

No. of tree species (100 m2) Canopy Area Elevation

Code Owner⬘s name Geographic location
or dominating species cover (%) (ha) (m asl)
S1 Blanca Pantoja 02⬚ 48⬘43.5⬙ N, 076⬚ 32⬘32.4⬙ W Unshaded 0 1.5 1,545
S2 Carlos Mosquera 02⬚ 48⬘45.3⬙ N, 076⬚ 32⬘47.8⬙ W Unshaded 0 4 1,541
S3 Eduardo Gómez 02⬚ 49⬘18.5⬙ N, 076⬚ 31⬘57.3⬙ W Unshaded 0 1 1,532
S4 Elciario Hernández 02⬚ 49⬘29.9⬙ N, 076⬚ 32⬘31.2⬙ W Unshaded 0 4 1,547
S5 Segundo Erazo 02⬚ 48⬘49.0⬙ N, 076⬚ 32⬘05.7⬙ W Unshaded 14 6 1,535
Sh1 Blanca Pantoja 02⬚ 48⬘47.2⬙ N, 076⬚ 32⬘35.2⬙ W Inga edulis 66 1 1,484
Sh2 Celimo Argote 02⬚ 49⬘13.2⬙ N, 076⬚ 32⬘20.2⬙ W Inga edulis 91 0.5 1,549
Sh3 Elciario Hernández 02⬚ 49⬘27.5⬙ N, 076⬚ 32⬘34.4⬙ W Inga edulis 80 4 1,553
Sh4 José Daza 02⬚ 49⬘29.4⬙ N, 076⬚ 32⬘24.4⬙ W Inga edulis 86 1.5 1,555
Sh5 Maximiliano Ortiz 02⬚ 51⬘44.6⬙ N, 076⬚ 33⬘52.7⬙ W Inga edulis 89 1.6 1,398
F1 Arturo Prieto 02⬚ 51⬘31.0⬙ N, 076⬚ 33⬘44.0⬙ W Lauraceae sp. 2 83 2 1,391
F2 Dolly Zúñiga 02⬚ 51⬘27.7⬙ N, 076⬚ 33⬘07.6⬙ W Piper sp. 2 100 2 1,443
F3 Elciario Hernández 02⬚ 49⬘36.5⬙ N, 076⬚ 32⬘34.9⬙ W Lauraceae sp. 1 100 3 1,540
F4 Reinel Paz 02⬚ 51⬘26.0⬙ N, 076⬚ 33⬘04.7⬙ W Melastomataceae sp. 4 91 0.5 1,335
F5 Segundo Erazo 02⬚ 48⬘50.2⬙ N, 076⬚ 32⬘07.8⬙ W Lauraceae sp. 1 97 6 1,511

plantations were unshaded, with the exception of S1,
which contained four guamo trees very distant of each
other. Plantains (Musa x paradisiaca L.) occasionally
were found among coffee bushes or forming barriers
on some of the sun coffee plantations. In this study,
Musa x paradisiaca L. is not considered a shade tree
because it has no woody stem and a life cycle of about
1 or 2 yr (Armbrecht et al. 2005). However, this crop
may provide some shade and resources for ants, and
canopy cover measures may increase when plantains
are close.
Ant Sampling in Leaf Litter and Coffee Bushes. We
visited each study site four times to sample ants and
one additional time to measure habitat variables. Five
stations were established at each location following a
50-m linear transect. The stations were located 12.5 m
from one another. Two traps (pitfall and tuna bait)
were installed at each station. Pitfall traps were es-
tablished on the soil surface (burrowed) and left for
48 h for collecting ground foraging ants. These pitfall
traps consist of a 150-ml plastic cup partially Þlled
(50%) with a solution of water plus detergent in a very
low concentration. Tuna baits consisted of a 10- by
15-cm sheet of paper with 3Ð 4 g of tuna-in-oil placed
on top. Tuna baits also were attached to trees or on top
of coffee bushes 1 m from the soil, and left for 5 hr. For
the pitfall method, we collected samples of ant work-
ers, majors, and queens, in 60-ml vials with 96% eth-
anol. Tuna bait contents and samples were collected
in 11- by 13-cm Ziploc bags (S.C. Johnson & Sons Inc.,
Racine, WI).
Habitat Characterization. In total, 11 habitat vari-
ables were measured and grouped in three vertical
strata: 1) arboreal stratum, 2) coffee bush stratum, and
3) soil stratum. The arboreal stratum included per-
centage of canopy cover, tree species richness, tree
density, tree height, percentage of tree dominance,
diameter at breast height (dbh), and number of plant
strata per station. The coffee bush stratum included
coffee bush height and foliage width of coffee bushes.
Finally, the soil stratum considered number of twigs
and leaf-litter volume.
At each location, the 11 habitat variables were mea-
sured in 1-m2 plots established at each station over the
50-m linear transect. All trees located inside or with
their canopy touching the 1-m2 sampling plot were
identiÞed to species or morphospecies. In addition,
their height and the elevation where they were touch-
ing the station were estimated using the vertical in-
tercept line technique described in Mills et al. (1991).
This was done to determine how many stratum layers
were present at each station. The dbh of live trees, the
height of coffee bushes, and foliage width also were
measured. Canopy cover per plot was measured with
a GRS densitometer (Forestry Suppliers Inc., Jackson,
MS). The percentage of shade cover then was esti-
mated based on the proportion of covered and un-
covered points (7) around the coffee bush in the
station.
The estimation of the tree density and percentage
of dominance of one shade tree variables were esti-
mated in a 100-m2 area over the already established
50 m linear transect. Thereafter, all extant trees 1 m
away from both sides of the transect were counted.
For the soil stratum, we recorded the number of twigs
lying on the soil per management strategy. The leaf-
litter layer was measured with a 30-cm ruler (KUM,
Erlangen, Germany) placed vertically by dividing the
1-m2 plot in nine subsquares of 33.3 by 33.3 cm and
averaging the nine points.
Laboratory Phase. The samples were cleaned and
separated at the laboratory, and then placed in vials
with 85% ethanol. Taxonomic identiÞcation was car-
ried out at the Entomology Section of the Biology
Department at the Universidad del Valle (Cali, Co-
lombia) by using keys from Bolton (2000), Fernández
(2008), Longino (2008), Galvis and Fernández (2009),
and Mackay and Mackay (2008). In addition, the sam-
ples were compared with those in the extant ant col-
lection of the Museum of Entomology (MEUV), Uni-
versidad del Valle. Finally, voucher specimens were
deposited at the MEUV.
The Intensification Index. In total, 11 habitat vari-
ables were used to calculate the Management Inten-
siÞcation Index (MI). The Management IntensiÞca-
tion Index was calculated as is described in Mas and
Dietsch (2003), taking into account the adaptations
made by Armbrecht (2003), thus making the MI ap-
April 2013 URRUTIA-ESCOBAR AND ARMBRECHT: COLOMBIAN COFFEE ANTS 197

plicable to the study of ground and leaf-litter organ-

0.02
0.58
0.48
0.52
0.14

0.52
0.42

0.40
0.34
3.42
2.59
F5
isms such as ants. Four habitat variables not consid-

0
0
ered previously (average leaf-litter height, average

0.007
tree height, average foliage width, and number of

0.08

0.32
0.49

0.56

0.46
0.27
2.68
2.92
F4

0.5
plant strata) were included in this study. This index

0
0
weighs each habitat variable equally along a scale from
0.0 (less managed or more natural condition) to 1.0

0.33

0.63
0.20

0.79
0.26

0.35
0.24

2.92
F3

2.8
0

0
0
(most intensive condition).
The 11 variables were standardized differently fol-

0.08
0.48
0.60

0.77

0.55
0.34
3.52
3.12
lowing the protocol of Mas and Dietsch (2003). Vari-

F2

0.7
0

0
0
ables 1, 2, 3, 8, 11, and 13 (Table 2) were calculated
obtaining the proportion of the average value for each

0.17

0.61
0.52
0.10

0.64
0.64

0.43
0.38
3.99
3.19
F1

0.5
location divided by the highest value found for that

0
0
variable among the 15 locations. The proportion then
was subtracted from 1.0 so that a higher value would

0.11
0.75
0.82
0.88
0.28

0.60
0.86
0.41
0.49
0.19
0.46
5.85
2.99
Sh5
reßect a higher intensiÞcation grade. For example,
tree species richness variable tends to decrease with
agricultural intensiÞcation. Therefore, in this study,

0.14
0.41
0.86
0.85
0.81

0.77
0.82
0.53
0.72
0.88
0.37
7.16
2.21
Sh4
for each location, the average value for that variable
was divided by the most natural condition (richest

0.20
0.75
0.88
0.72
0.14

0.70
0.82
0.59
0.65
0.64
0.47
6.56
2.67
Sh3
location), in this case 12 (Table 2) and then subtracted
from 1.0. Variables 5, 6, 7, 9, and 10 (Table 2) were
calculated using the proportion of the relative differ-

0.08

0.71
0.56
0.39

0.50
0.78
0.63
0.54
0.70
0.32
5.71
2.68
Sh2

0.5
ence (Armbrecht 2003). Thus, the dominance of one
Standarized values for 11 habitat variables included in the MI plus Shannon Index values for each location

shade tree (DST) variable tends to increase with ag-

ricultural intensiÞcation, on the assumption that more
0.34
0.91
0.98
0.44

0.72
0.94
0.41
0.51
0.58
0.34
7.17
2.96
Sh1

intensive management practices include a lower tree

1
diversity. The most natural condition for DST (0.0
value) was based on the location with the overall
0.85

0.65
0.77
0.96
0.68
9.91
2.42
S5

lowest dominant tree value. Thus, the standardized

1
1
1
1

1
1
DST value for each location was calculated as follows:

0.92
0.90
0.70
0.91
10.43
2.64
1Ð [(DST location Ð lowest DST)/(highest DST -
S4
1
1
1
1
1

1
1
lowest DST)]. The standardized values for each vari-
able were added in such a way that the number of
variables included in the index constitute the maxi-
0.96
0.82
0.79
0.82
0.86
10.25
2.55
S3
1
1
1
1
1

mum value possible reached by the MI, 11 in this case.

Data Analyses. The total sampling effort was tested
by means of three nonparametric estimators: inci-
0.66
0.60
0.95
0.57
9.78
2.71
S2
1
1
1
1
1

1
1

dence-based coverage estimator (ICE), Chao 2, and

Jacknife 1, which were chosen for being the most
0.73
0.79
0.89
0.69

2.61

appropriate for the analysis of hyperdiverse groups

S1

10.1
1
1
1
1
1

1
1

that exhibit grouped distributions (Gallego-Ropero

2005, Rivera and Armbrecht 2005). The sampling ef-
Avg no. of plant strata (0.0 ⫽ 10, 1.0 ⫽ 0)

Avg foliage width (0.0 ⫽ 5.66, 1.0 ⫽ 0.51)

fort also was measured by categories (forest, shaded

Avg dbh live trees (0.0 ⫽ 154, 1.0 ⫽ 9.5)

Avg leaf-litter ht (0.0 ⫽ 8.33, 1.0 ⫽ 0)

coffee, and sun coffee plantations). To estimate the

Avg coffee ht (0.0 ⫽ 3.65, 1.0 ⫽ 0.72)
% canopy cover (0.0 ⫽ 100, 1.0 ⫽ 0)
Tree sp. richness (0.0 ⫽ 12, 1.0 ⫽ 0)

Avg no. of logs (0.0 ⫽ 69, 1.0 ⫽ 0)

number of ant species at the 15 sampling locations, the

Avg tree ht (0.0 ⫽ 1.2, 1.0 ⫽ 14.9)
Tree density (0.0 ⫽ 52, 1.0 ⫽ 0)

% dominance of one shade tree

nonparametric ICE was obtained. These calculations

Habitat variables

plus Shannon diversity values (H⬘) were carried out

(0.0 ⫽ 22.2, 1.0 ⫽ 100)

using the EstimateS Program version 8.2 (Colwell

2006). The ICE values then were compared by means
of one-way analysis of variance (ANOVA) (ANOVA;
3 management types, 5 sampling locations each) when
normality assumptions and homogeneity of variances
were met. This was done using the PAST Program for
Windows (Hämmer et al. 2001). A posthoc Tukey
analysis was carried out to further determine which of
the treatments was statistically different.
We tested whether the MI results were consistent
Shannon Index

with the qualitative classiÞcation done at each sam-

Table 2.

No.

pling location by means of a one-way ANOVA and

(H⬘)

Tukey test. The assumptions of normality and homo-

Total

geneity of variances were met. A correlation analysis

10
11
13
1
2
3
5
6

7
8
9
198 ENVIRONMENTAL ENTOMOLOGY
(Pearson correlation coefÞcient) was made between
MI values and Shannon diversity index (H⬘) at each
sampling location. Both analyses were performed us-
ing the R program for Windows (R Development Core
Team 2008).
Differences in species composition between the
three studied categories were detected by means of
the ordination methodology nonmetric multidimen-
sional scaling (NMDS) using the BrayÐCurtis distance
measure. Subsequently, we tested these differences by
means of a single factor ANOSIM (analyses of simi-
larities), which allowed us to test the H0 of no differ-
ence between the three habitat types (Quinn and
Keough 2002). For performing NMDS and ANOSIM,
we used the PAST Program for Windows (Hämmer et
al. 2001). The SIMPER test (similarity percentages)
was used to calculate the percentage of contribution
of each species to the dissimilarities between groups
and similarities within groups (Quinn and Keough
2002). The SIMPER test was performed using the
Community Analysis Package (CAP version 3.01)
(Seaby et al. 2004). For determining which habitat
variables inßuenced the ant distribution patterns
most, a canonical correspondence analysis (CCA) was
carried out using the CANOCO Program for Windows
version 4.5 (ter Braak and Šmilauer 1998). The signif-
icance of all canonical axes was measured by means of
a 999 permutations Monte Carlo Test. A set of eight
habitat variables was used (the most representative for
ants), but two of them presented autocorrelation and
were withdrawn from the analysis (average tree spe-
cies richness and average coffee height).
The relationship between habitat management
strategy and functional attributes or traits of species
was assessed using the fourth-corner approach (Roy
and de Blois 2006, Dray and Dufour 2007). Only 73
species were considered in this analysis. The remain-
ing nine species were not included in the analysis
because of a lack of information about their habitat
preferences and feeding habits. The traits classiÞca-
tion was based on the ant guilds reported by Ramṍrez
et al. (2010), Lozano-Zambrano et al. (2009), Jiménez
et al. (2008), and unpublished data of E. Jiménez (E.
Jiménez, personal communications), which are the
result of several years of ecology research of southwest
Colombian ant fauna. The trait classiÞcation includes
the habitat and feeding requirements of the ant spe-
cies, for offering relevant information of how ant
groups with similar ecological requirements respond
to agricultural intensiÞcation. Abundance data (num-
ber of times one ant species was recorded for one trap)
and a functional characterization were used to com-
pute the fourth-corner statistic. To perform the anal-
ysis, the ade4 package (Dray and Dufour 2007) for R
was used. These associations were tested by 999 per-
mutations and, in this case, abundance values were
permuted in such a way as to simulate the absence of
habitat control over species assemblages (permuta-
tion model 2 sensu Legendre et al. 1997).
Abundance data were used for all analysis with
the exception of the species accumulation curves,
ANOSIM and SIMPER that used presence and ab-
Vol. 42, no. 2
sence data. In this study, “occurrence” represents an
event in which one ant species was recorded, regard-
less of the number of individual ant species captured
by a sampling method. In this sense, abundance data
are deÞned as the number of occurrence events for a
particular ant species in a particular trap during the
whole sampling period.
Results
During sampling time, a total of 18,186 ants were
collected. They represented 82 ant species grouped in
34 genera and 9 subfamilies of Formicidae. The sub-
family Myrmicinae showed the highest number of
species and represented 46% of the total number of ant
species. This subfamily dominated both in number of
genera (15), morphospecies (38), and capture events
(741). Formicinae was the second largest subfamily
representing 18% of all ant species, followed by the
Poneromorph subfamilies (mostly solitary hunters)
(15%), Pseudomyrmecinae (11%), and Dolichoderi-
nae and Ecitoninae both with 5%. Pheidole, Pseu-
domyrmex, and Camponotus were the most species-
rich genera and were represented by 14, 9, and 7
species, respectively.
The three nonparametric estimators (ICE [87%],
Chao 2 [86%], and Jacknife 1 [83%]), showed that the
collected ants represented an average of 85% of the
total estimated number of ant species in the study area.
The sampling effort also was measured for each man-
agement strategy by using the nonparametric estima-
tor Chao 2, it being the most rigorous and less biased
for small samples (IAvH 2006). According to Chao 2,
the best sampled management strategy was forest
(82%), followed by sun coffee (79%) and shaded cof-
fee (70%).
Species Richness per Management Strategy. Ant
species richness per management strategy was highest
in shaded coffee with 60 ant species, followed by forest
(56) and sun coffee (33). The richest locations were
Sh5 and F1 both with 32 ant species each; the less
diverse location was S5 with only 14 ant species. The
three management strategies showed signiÞcant dif-
ferences in terms of the incidence-based coverage
estimator (ICE) mean values (Colwell 2006) (F ⫽
3.89; df ⫽ 2,12; P ⫽ 0.039). Posterior analyses revealed
that the mean number of ant species in the shaded
coffee plantations was higher than that on sun coffee
plantations (P ⫽ 0.035) but not different from the
forest patches. Shaded coffee plantations harbored 15
ant species not found elsewhere (25% of its ant fauna)
whereas 13 (23%) and six (18%) ant species were
found exclusively in forest and sun coffee plantations,
respectively. Capture methods also varied in efÞ-
ciency. Pitfall traps recorded 80% of occurrence
events; the remaining 20% were registered by tuna
baits.
As a general tendency, species diversity (H⬘) by
location tends to decrease as the MI values increase.
This is shown by a negative and signiÞcant correlation
coefÞcient (r ⫽ ⫺0.5714, df ⫽ 13, P ⫽ 0.026) between
species diversity and the correspondent MI values.
April 2013 URRUTIA-ESCOBAR AND ARMBRECHT: COLOMBIAN COFFEE ANTS
The management index values (Table 2) discrimi-
nated three principal groups, corresponding to the
categories deÞned at the beginning of this study. The
three groups of values do not overlap (sun: 9.78 Ð10.43,
shaded: 5.85Ð7.17, forest: 2.68 Ð3.99). MI results per
location followed a gradient in terms of agricultural
intensiÞcation (Table 2). The gradient reßected the
most (S4) and less (F4) intensively managed locations
at the two ends of the intensiÞcation gradient for the
studied area in the municipality of Caldono.
The one-way ANOVA conÞrmed that the MI mean
values per management strategy were statistically dif-
ferent (F ⫽ 206.43; df ⫽ 2,12; P ⬍ 0.001), and a posthoc
Tukey analyses revealed that the MI means in the two
strategies and forest were all different (P ⫽ 0.0001).
Species Composition per Management Strategy. A
separation of the three strategies was evident from the
NMDS diagram (stress value: 0.18), which may be a
result of the variation in shade cover. The ANOSIM
analysis detected highly signiÞcant differences be-
tween all management strategies (ANOSIM R
global ⫽ 0.29; P ⬍ 0.0001), rejecting the H0 of no
difference between the three habitat types. The
SIMPER analysis detected that forest and sun coffee
plantations showed the highest dissimilarity (92%),
followed by forest and shaded coffee plantations
(90.8%) and sun and shaded coffee plantations
(82.7%). Shaded coffee plantations and forest patches
shared a total of 20 ant species, all of them highly
associated with the arboreal component. In contrast,
sun coffee plantations and forest patches only shared
three ant species.
Some ant species were considered typical based on
their occurrence for each management strategy. Het-
eroponera inca (Brown), Pachycondyla aenescens
(Mayr), Camponotus novogranadensis (Mayr), Crema-
togaster nigripilosa Mayr, and Pseudomyrmex pallens
(Mayr) were very frequently present in the forests.
Camponotus sp. 11 and Atta cephalotes (L.) were also
very common but were found in forest and shaded
coffee plantations. Linepithema neotropicum (Emery),
Ectatomma ruidum (Roger), Pheidole sp. 4, Cypho-
myrmex rimosus (Spinola), and Solenopsis geminata
(F.) were common in sun coffee plantations. Partic-
ularly, E. ruidum (8.4%), L. neotropicum (7.6%), and
Pheidole sp. 4 (9.1%) were the ant species that most
contributed to establishing the differences between
categories. For instance, L. neotropicum represented
12.1% of total occurrence events during the whole
sampling period. However, 55% of L. neotropicumÕs
occurrence events were reported for sun coffee plan-
tations. A similar situation occurred for Pachycondyla
obscuricornis (Emery), Ectatomma ruidum, and Phei-
dole sp. 4.
Habitat Variables Affecting Ant Distribution Pat-
terns. The canonical correspondence analysis deter-
mined that 38% of the variation in the data were
explained by this analysis. Here, 20 and 10% of the
variation in the species data were explained by axis 1
(eigenvalue: 0.474) and axis 2 (eigenvalue: 0.247),
respectively. Based on the canonical correlation co-
efÞcients, it was inferred that the Þrst CCA axis rep-
199
Table 3. Canonical and correlation coefficients of six habitat
variables analyzed by CCA
Canonical Correlation
Habitat variables coefÞcients coefÞcients
Axis 1 Axis 2 Axis 1 Axis 2
Avg tree ht 0.3508 ⫺0.5455 0.8844 ⫺0.0148
Avg no. plant strata ⫺0.1694 0.4027 0.839 ⫺0.1126
Avg leaf-litter ht 0.5973 ⫺0.86 0.8138 0.2888
Coffee foliage width ⫺0.2327 0.8977 ⫺0.4321 0.7069
Tree cover (%) ⫺0.278 1.8243 0.8673 0.372
Avg no. of twigs 0.5534 ⫺0.4221 0.829 0.1474
Numbers in bold show the highest coefÞcients.
resented a gradient in tree height and the second CCA
axis represented a gradient in coffee foliage width
(Table 3). The canonical coefÞcients of leaf-litter and
tree cover were high for axis 1 and axis 2, respectively,
indicating that these two variables determined the ant
distribution patterns in the study area (Table 3).
These results are also shown by a CCA plot (Fig. 2)
where longer arrows represent the habitat variables
that are more strongly correlated with the ordination
axes and, therefore, more closely related to the pat-
terns of variation in species composition (Jongman et
al.1995). The Monte Carlo Test was highly signiÞcant
(P ⫽ 0.001).
Functional Ant Groups. The fourth-corner analysis
determined that the relationship between categories
and functional attributes (feeding and habitat) were
globally signiÞcant. On the one hand, feeding attri-
butes were highly signiÞcant (G ⫽ 77.14, P ⫽ 0.001)
for exudates and omnivorous. However, habitat attri-
butes were signiÞcant (G ⫽ 90.68, P ⫽ 0.001) for
arboreal, rotten sticks, and soil (Table 4).
Discussion
Species Richness by Management Strategy. The re-
sults of this study clearly demonstrated that differ-
ences in management strategies on Colombian coffee
plantations inßuence the number and identity of ant
species present at each management strategy. The
shaded coffee strategy showed the highest ant species
richness and the highest number of exclusive species,
even above the forest management strategy, which
was expected to harbor the highest ant diversity and
species exclusivity. Several explanations are possible:
1) Shaded plantations vary in degree of shade, some-
times with big gaps where high sunlight exposure
may reach high values, whereas other areas may be
100% covered by tree canopy. This situation allows
shaded-tolerant ants to move from forested areas at
lower elevations where their habitat had almost
disappeared to shaded coffee plantations in search
of additional food or habitat resources. For in-
stance, Pachycondyla impressa (Roger) is usually
registered on forest ßoor but, in this study, it was
also collected at the Sh1 shaded coffee plantation,
close to a riparian forest. It is possible that this
species crossed forest boundaries in search of nest-
ing or extra feeding resources.
200 ENVIRONMENTAL ENTOMOLOGY Vol. 42, no. 2

Fig. 2. CCA ordination triplot of ant data with six habitat variables measured at 15 sampling locations in Caldono,
Colombia.

2) A wide range of feeding and habitat resources for ties for the associated fauna (Armbrecht et al. 2005,
ants may be found, depending of the agricultural Mera et al. 2010).
intensiÞcation gradient. Some ant species may be 3) The sampled forest patches were small and sec-
favored by the Þrst stages of agricultural intensi- ondary, usually embedded in pasture matrixes. For
Þcation. In addition, low-intensity agriculture may this reason, they may have already lost part of their
promote habitat diversiÞcation. This may provide ant species richness, especially the very sensitive
more resources such as plant biomass and fruits, and rare ones (usually present at low densities).
not only for ants but also for other animals Although forests did not register the highest num-
(Tscharntke et al. 2005). The structural complexity ber of species, they harbored 13 species not found
offered by shaded coffee plantations may increase elsewhere and that probably have special habitat
the number of associations (mutualism and pre- requirements. These exclusive species mainly are
dation) between ants and other insects (Blattaria, associated with different strata in the forest and
Orthoptera, Hemiptera, Coleoptera, and Hyme- their presence is possible because of the structural
noptera), thus providing more nesting opportuni- complexity of the vegetation that forests exhibit
(Azteca cf coeruleipennis Emery, Myrmelachista sp.,
Table 4. Fourth-corner diagram: relationship among catego-
Acanthognathus teledecus [Brown and Kempf], and
ries and functional attributes Heteroponera inca). Particularly, H. inca is endemic
from Colombia (Jiménez et al. 2008) and A. tele-
Functional Categories decus was only collected twice at F4, thus a very
attributes Forest Shaded Sun rare species in this study.
Feeding
Exudates ⫹ ⫺ The results of this study are highly consistent with
Fungivorous studies in other parts of Colombia. Armbrecht et al.
Generalist (2005) found that ant fauna in forests were less rich
Omnivorous ⫺ ⫹
Opportunistic than on organic shaded coffee plantations in Apṍa,
Predator Risaralda, and Tamesis, Antioquia (northwestern Co-
Habitat lombia). Statistical differences with ICE estimators
Arboreal ⫹ ⫺ also were detected between shaded and sun coffee
Rotten sticks ⫹
Generalist plantations in that study, although the methodology
Leaf-litter was visual (manual) capture and winkler extraction
Opportunistic from the leaf litter (Armbrecht et al. 2005). Highly
Soil ⫹ similar results were reported by Gallego-Ropero
Subterranean
(2005) in Apṍa using tuna baits as sampling units.
Positive (⫹) and negative (⫺) relationships between categories Gallego-Ropero et al. (2009) and Mera et al. (2010)
and functional attributes are shown. found similar results in the Department of Cauca
April 2013 URRUTIA-ESCOBAR AND ARMBRECHT: COLOMBIAN COFFEE ANTS
when shaded and sun coffee plantations were studied,
although no forests were sampled in the region. The
consistency among studies developed on Colombian
coffee plantations demonstrates that effectively
shaded coffee plantations represent a high quality
matrix for ants where greater species richness and
more equitable ant communities can be found. The
current study goes far beyond because, here, we fur-
ther analyze the composition of ant fauna by using
other statistical tools. Implications for conservation
are strong. As demonstrated by Perfecto et al. (2009),
migration among forest fragments in heterogeneous
tropical landscapes is fundamental for maintaining
metacommunities in the long-term and, therefore, a
requirement for the conservation of biodiversity.
Our results show a negative effect of agricultural
homogenization on species richness, species identi-
ties, and the possibility of equitable distribution.
Therefore, the higher the MI value (or intensiÞcation
condition), the lower species diversity. Overall, our
previous qualitative characterization matched the
quantitative analyses derived from 11 measured hab-
itat variables, evidencing that there is an agricultural
intensiÞcation gradient in the coffee plantations of the
region.
Changes in Species Composition Because of Differ-
ences in Management Strategies. The results of this
study determine that differences in management strat-
egies on coffee plantations also inßuence the way in
which these ant species are distributed over the ag-
ricultural intensiÞcation gradient. As expected,
shaded coffee plantations resembled forest patches
more than sun coffee plantations in terms of ant spe-
cies composition. This may be a result of the similarity
of the arboreal component between these two man-
agement strategies (Table 4). In this sense, the great-
est dissimilarity between forests versus sun coffee
plantations is explained as the lack of shade trees and
other derived components (e.g., twigs, leaf litter) on
sun coffee plantations. This situation limited the avail-
ability of certain kinds of habitat and feeding resources
needed by specialist ants.
The differences in species composition among the
three management strategies are basically determined
by the ant species that were present in the three
sampled strategies, but differed in terms of the number
of occurrence events reported for each management
strategy. Linepithema neotropicum, Pachycondyla ob-
scuricornis, Ectatomma ruidum, Cyphomyrmex rimo-
sus, and Pheidole sp. 4 were present in all categories
but were particularly dominant in sun coffee planta-
tions. This trend is in agreement with Abadṍa et al.
(2010) where the species found in pastures are a
subset of the species found in the remaining categories
(riparian corridor, secondary, and mature forest) and
the difference in species composition is mainly estab-
lished by species abundance recorded in pastures.
Habitat Variables that Influence Ant Distribution
Patterns. Variation in species composition was driven
by the average tree height gradient, corroborating
the assumption made from the NMDS plot where the
three management strategies were separated by the
201
canopy cover gradient. Clearly, the species present at
highly intensive management plantations (e.g., Phei-
dole sp. 4, C. rimosus, E. ruidum) were separated from
species present at more natural locations (P. impressa,
C. novogranadensis, C. nigripilosa).
The importance of leaf litter and twigs demon-
strated by this study was consistent with other studies.
Armbrecht et al. (2005) found that ants on the litter
stratum (where twigs are found) are inßuenced neg-
atively by coffee production intensiÞcation and may
even disappear in the Þrst stages of the agricultural
modernization process. Not only do the number of
twigs increases from sun coffee plantations to forest,
but also their length and thickness, thus offering ad-
ditional habitat resources and quality (tree diversity)
that may contribute to the diversity of ant species
(Armbrecht et al. 2004). Canopy cover not only reg-
ulates the temperature inside sampling locations, but
also, directly affects the establishment of different ant
populations and other arthropods (Gallego-Ropero
2005), offering more nesting and feeding resources.
Ant Functional Groups. Our results demonstrated
that ant species assemblages were effectively depen-
dent upon the habitat characteristics of the three mea-
sured management strategies. Ant guilds ecologically
responded to the agricultural intensiÞcation gradient
on coffee plantations. Some ant guilds were favored by
the habitat characteristics of less complex agricultural
systems (e.g., omnivorous), whereas others were quite
sensitive and were only present at more natural or less
managed locations (e.g., exudates). This implies that
strong biotic ties (speciÞc habitat and feeding require-
ments) exist among the species that are found to-
gether.
The species possessing functional attributes favored
by agricultural intensiÞcation were always dominant
and in great abundance in sun coffee plantations. In
this study, 1,450 “occurrence” events in total were
recorded. Shaded coffee plantations and forests only
registered 29 and 28% of the total occurrence events,
respectively. In contrast, sun coffee plantations ob-
tained the highest number of occurrence events
(44%), albeit with the lowest species richness among
the three measured management strategies.
The relative homogeneity of sun coffee plantations
seemed to favor soil dominant and omnivorous ant
species such as S. geminata, E. ruidum, and Pheidole sp.
4. E. ruidum particularly, may reach great abundance
in open pasturelands in the southwestern region of
Colombia (Santamaria et al. 2009). E. ruidum can also
be a dominant species in economically important
crops (Lachaud 1990). In this study, it represented
12% of the total occurrence events for sun coffee
plantations after L. neotropicum (15%). According to
the SIMPER results, these dominant species deter-
mined the differences between management strate-
gies and forest patches.
Regarding forests, the relationships between envi-
ronment and functional traits mainly were associated
with the arboreal component. The detected positive
relationships were a result of the presence of exclusive
ants with strict habitat requirements (Cephalotes spp.:
202 ENVIRONMENTAL ENTOMOLOGY
arboreal and exudates attributes, A. teledecus: rotten
sticks). Finally, no signiÞcant relationship was found
for shaded coffee plantations. This may be related to
the possibility that some ant species may even disap-
pear during the Þrst stages of the agricultural mod-
ernization process (Armbrecht et al. 2005). Neverthe-
less, more information is needed. It has been estimated
that 30% of ant species on the shaded coffee planta-
tions were not sampled.
In synthesis, our results determined the following:
1) Ants are highly sensitive to increases in manage-
ment intensity in coffee agriculture. We evidenced a
decrease of ant species diversity as management in-
tensiÞcation increases. 2) Shaded coffee plantations
presented the highest species richness and were more
similar to ant forest fauna than the one in sun coffee
plantations with respect to forest. The explanation is
that shaded coffee plantations are ecologically more
complex than other agricultural systems such as sun
coffee plantations. 3) This study also found that the
agricultural intensiÞcation process beneÞted a few ant
species that were present in high quantities in inten-
sive management systems, demonstrating that species
assemblages are dependent upon the physical char-
acteristics of the locations where they were found.
Our results also highlight the importance of main-
taining the associated arboreal component to maintain
landscape heterogeneity, thereby preserving the as-
sociated biodiversity, their equitability and their eco-
logical interactions. However, forests harbored ex-
clusive ant species which were never found in
productive systems such as coffee plantations (e.g.,
A. teledecus, H. inca). This indicates that no other
land uses can replace the conservation value that
forest patches possess.
Acknowledgments
Joost Duivenvoorden, University of Amsterdam, provided
invaluable orientation to this study. Elizabeth Jimenez con-
tributed classifying several ant species into feeding and hab-
itat guilds. Elena Paz for Þeld assistance. Members of the
Entomology Section, Universidad del Valle for their valuable
support and orientation. We are also grateful to the reviewers
for invaluable comments on earlier versions of this manu-
script. This Project was partially Þnanced by the Netherlands
Fellowship Programme (NFP), Stunt and AUV grants.
References Cited
Abadı́a, J. C., C. Bermúdez, F. H. Lozano-Zambrano, and P.
Ulloa-Chacón. 2010. Hormigas cazadoras en un paisaje
subandino de Colombia: riqueza, composición y especies
indicadoras. Rev. Colomb. Entomol. 36: 127Ð134.
Armbrecht, I. 2003. Habitat changes in Colombian coffee
farms under increasing management intensiÞcation. En-
dangered Species Update 20: 163Ð178.
Armbrecht, I., and I. Perfecto. 2003. Litter-twig dwelling
ant species richness and predation potential within a
forest fragment and neighboring coffee plantations of
contrasting habitat quality in Mexico. Agric. Ecosyst. En-
viron. 97: 107Ð115.
Vol. 42, no. 2
Armbrecht, I., I. Perfecto, and J. Vandermeer. 2004. Enig-
matic biodiversity correlations: ant diversity responds to
diverse resources. Science 304: 284 Ð286.
Armbrecht, I., L. Rivera, and I. Perfecto. 2005. Reduced
diversity and complexity in the leaf-litter ant assemblage
of Colombian coffee plantations. Conserv. Biol. 19: 897Ð
907.
Bolton, B. 2000. The ant tribe Dacetini. With a revision of
the Strumigenys species of the Malgasy Region by Brian
L Fisher, and a revision of the Austral epopostrumiform
genera by Steven O Shattuck. Mem. Am. Entomol. Inst.
65: 1Ð1028.
Colwell, R. K. 2006. Estimate S: statistical estimation of spe-
cies richness and shared species for samples, version 8.
(http://www.viceroy.eeb.uconn.edu/estimates).
Dray, S., and A. B. Dufour. 2007. The ade4 package: imple-
menting the duality diagram for ecologists. J Stat. Softw.
22: 1Ð20.
Fernández, F. 2008. Subfamilia Ponerinae. In F. Lozano-
Zambrano, F. Fernández, E. Jiménez, and T. Arias. (eds.),
Sistemática, biogeografṍa y conservación de las hormigas
cazadoras de Colombia. Instituto de Investigaciones de
Recursos Biológicos Alexander von Humboldt. Bogotá
D. C., Colombia.
[FNC] Federación Nacional de Cafeteros. 2010a. Área cul-
tivada. (www.federaciondecafeteros.org/particulares/es/
quienes_somos/119_estadisticas_historicas/).
[FNC] Federación Nacional de Cafeteros. 2010b. Área cul-
tivada según exposición solar por departamento. (www.
federaciondecafeteros.org/static/Þles/Cauca09.pdf).
Gallego-Ropero, M. C. 2005. Intensidad de manejo del agro-
ecosistema de café (Coffea arabiga L.) (monocultivo y
policultivo) y riqueza de especies de hormigas general-
istas. Bol. Mus. Entomol. Univ. Valle Colomb. 6: 16 Ð29.
Gallego-Ropero, M. C., J. Montoya-Lerma, and I. Armbrecht.
2009. ÀEs la sombra benéÞca para la diversidad de hor-
migas y el peso del café? Una experiencia en Pescador,
Cauca, Colombia. Bol. Cient. Mus. Hist. Nat. 13: 106 Ð116.
Galvis, J. P., and F. Fernández. 2009. Ants of Colombia X.
Acanthognathus with the description of a new species
(Hymenoptera: Formicidae). Rev. Colomb. Entomol. 35:
245Ð249.
Hämmer, Ø, D. Harper, and P. Ryan. 2001. Past: paleonto-
logical statistics software package for education and data
analysis. Palaeontol. Electronica 4: 1Ð9.
[IavH] Instituto de Investigación de Recursos Biológicos
Alexander von Humboldt. 2006. Métodos para el análisis
de datos: una aplicación para resultados provenientes de
caracterizaciones de biodiversidad. Programa de inventarios
de biodiversidad. (http://www.bio-nica.info/biblioteca/
HumboldtAnalisisDatos.pdf).
Jiménez, E, F. H. Lozano-Zambrano, J. Rodrı́guez, and D. P.
Ramı́rez. 2008. Conservación de hormigas cazadoras:
rareza y endemismo. In E. Jiménez, F. Fernández, T. M.
Arias, and F. H. Lozano-Zambrano (eds.), 2008. Sis-
temática, biogeografṍa y conservación de las hormigas
cazadoras de Colombia. Instituto de Investigación de Re-
cursos Biológicos Alexander von Humboldt. Bogotá D.C.,
Colombia.
Jongman, R.H.G., C.J.F. Ter Braak, and O.F.R van Tongeren.
1995. Data analysis in community and landscape ecology.
Cambridge University Press, United Kingdom.
Lachaud, J. P. 1990. Foraging activity and diet in some neo-
tropical ponerine ants. I. Ectatomma ruidum Roger (Hy-
menoptera, Formicidae). Folia Entomol. Mex. 78: 241Ð
256.
April 2013 URRUTIA-ESCOBAR AND ARMBRECHT: COLOMBIAN COFFEE ANTS
Legendre, P., R. Galzin, and M. L. Harmelin-Vivien. 1997.
Relating behavior to habitat: solutions to the fourth-cor-
ner problem. Ecology 78: 547Ð562.
Longino,J. 2008. HormigasdeCostaRica.(http://academic.
evergreen.edu/projects/ants/Genera).
Lozano-Zambrano, F. H., P. Ulloa-Chacón, and I. Armbecht.
2009. Hormigas: relaciones especies-área en fragmentos
de bosque seco tropical. Neotrop. Entomol. 38: 44 Ð54.
Mackay, W., and E. Mackay. 2008. Subfamilies of living ants in
the new world. (http://www.utep.edu/LEB/antgenera.
htm).
Mas, A. H., and T. V. Dietsch. 2003. An index of manage-
ment intensity for coffee agroecosystems to evaluate but-
terßy species richness. Ecol. Appl. 13: 1491Ð1501.
Mera, Y. A., M. C. Gallego, and I. Armbrecht. 2010. Inter-
acciones entre hormigas e insectos en follaje de cafetales
de sol y sombra, Cauca, Colombia. Rev. Colomb. Ento-
mol. 36: 116 Ð126.
Mills, G. S., J. B. Dunning, and J. M. Bates. 1991. The rela-
tionship between breeding bird and vegetation volume.
Wilson Bull. 103: 468 Ð 479.
Moguel, P., and V. M. Toledo. 1999. Biodiversity conserva-
tion in traditional coffee systems of Mexico. Conserv.
Biol. 13: 11Ð21.
Pardo-Lorcano, L. C., J. Montoya, A. C. Bellotti, and A. van
Schoonhoven. 2003. Abundancia de chizas rizofagas
(Coleoptera: Melolonthidae) en agroecosistemas de Cal-
dono y Buenos Aires, Cauca, Colombia. Rev. Colomb.
Entomol. 29: 177Ð184.
Pardo-Lorcano, L. C., J. Montoya, A. C. Bellotti, and A. van
Schoonhoven. 2005. Structure and composition of the
white grub complex (Coleoptera: Scarabaeidae) in agro-
ecological systems of northern Cauca, Colombia. Fla.
Entomol. 88: 355Ð362.
Perfecto, I., R. A. Rice, R. Greenberg, and M. E. van der
Voort. 1996. Shade coffee: a disappearing refuge for bio-
diversity. Bioscience 46: 598 Ð 608.
Perfecto, I., J. H. Vandermeer, and A. Wright. 2009. Na-
ture⬘s matrix, linking agriculture, conservation and food
sovereignity. Earthscan. London, Sterling, VA.
Philpott, S. M., and I. Armbrecht. 2006. Biodiversity in trop-
ical agroforests and the ecological role of ants and ant
diversity in predatory function. Ecol. Entomol. 31: 369 Ð
377.
Philpott, S. M., I. Perfecto, and J. Vandermeer. 2006. Effects
of management intensity and season on arboreal ant di-
View publication stats
203
versity and abundance in coffee agroecosystems. Biodiv-
ers. Conserv. 15: 139 Ð155.
Quinn, G. P., and M. J. Keough. 2002. Experimental design
and data analysis for biologists. Cambridge University
Press, New York.
Ramı́rez, M., J. Herrera, and I. Armbrecht. 2010. ÀBajan de
los árboles las hormigas que depredan en potreros y
cafetales colombianos? Rev. Colomb. Entomol. 36: 106 Ð
115.
R Development Core Team. 2008. R: a language and envi-
ronment for statistical computing. R Foundation for Sta-
tistical Computing, Vienna, Austria. ISBN 3-900051-07-0
(http://www.R-project.org).
Rivera, L., and I. Armbrecht. 2005. Diversidad de tres
gremios de hormigas en cafetales de sombra, sol y bosques
de Risaralda. Rev. Colomb. Entomol. 31: 89 Ð96.
Roy, V., and S. de Blois. 2006. Using functional traits to assess
the role of hedgerow corridors as environmental Þlters
for forest herbs. Biol. Conserv. 30: 592Ð 603.
Santamaria, C., Y. Dominguez-Haydar, and I. Armbrecht.
2009. Cambios en la distribución de nidos y abundancia
de la hormiga Ectatomma ruidum (Roger 1861) en dos
zonas de Colombia. Bol. Mus. Entomol. Univ. Valle Co-
lomb. 10: 10 Ð18.
Seaby, R.M.H., P. A. Henderson, and J. Prendergast. 2004.
Community analysis package, version 3.01. Pisces Con-
servation Ltd., Hampshire, United Kingdom.
Ter Braak, C.J.F., and P. Šmilauer. 1998. CANOCO refer-
ence manual and user⬘s guide of Canoco for Windows:
software for Canonical Community Ordination, version 4.
Microcomputer Power, Ithaca, NY.
Tscharntke, T., A. Klein, A. Kruess, I. Steffan-Dewenter, and
C. Thies. 2005. Landscapes perspectives on agricultural
intensiÞcation and biodiversity Ð ecosystem service man-
agement. Ecol. Lett. 8: 857Ð 874.
Vandermeer, J., and R. Carvajal. 2001. Metapopulation dy-
namics and the quality of the matrix. Am. Nat. 158: 211Ð
220.
Wielgoss, A., T. Tscharntke, D. Buchori, B. Fiala, and Y. Clough.
2010. Temperature and a dominant dolichoderine ant spe-
cies affect ant diversity in Indonesian cacao plantations.
Agric. Ecosyst. Environ. 135: 253Ð259.
Received 23 March 2011; accepted 24 January 2013.