526 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

Basic ecology of ant gardens in a dry-premontane transitional forest

Ángela Marcela Barrera-Bello1,2; http://orcid.org/0000-0003-2819-0259

Alba Marina Torres-González1,2*; http://orcid.org/0000-0002-3010-2505

1. Universidad del Valle, Departamento de Biología, Calle 13 No. 100-00, Cali, Colombia; angela.barrera@correouni-

valle.edu.co, alba.torres@correounivalle.edu.co (Correspondence*)

2. Grupo de Investigación Ecología y Diversidad Vegetal, Universidad del Valle, Calle 13 No. 100-00, Cali, Colombia.

Received 23-III-2022. Corrected 22-VII-2022. Accepted 26-VII-2022.

ABSTRACT

Introduction: “Ant gardens” are ant nests located at different heights on trees on which vascular epiphytic

plants that have been transported and sown by ants have germinated. Although this mutualistic relationship has

been studied in humid tropical ecosystems, information on other tropical and Colombian ecosystems is scarce.

Objective: To characterize the distribution, building, diversity, demography and phenology of ant gardens in

dry tropical forest.

Methods: In January and February, 2018, we identified 170 ant gardens on a 100x5 m transect on the banks

of the Quesada River, Colombia, and in adjacent secondary dry premontane transitional forest; we monitored

changes, for gardens and plants, in 28 of them, every two weeks (March 2018-February 2019).

Results: The gardens, built by Azteca ulei, were aggregated near water bodies; had 10 species of epiphytes and

were on 13 tree species. Larger gardens had more epiphyte species. Some epiphytes had a bimodal phenologi-

cal pattern. Less seedlings become established in the dry season, and less adults remain in the gardens. Garden

characteristics benefit both epiphytes and ants.

Conclusions: The gardens built by A. ulei have ecological characteristics that favor the germination, establish-

ment, and reproduction of diverse epiphytes in this dry tropical ecosystem, including aggregation near water

flows.

Key words: Azteca ulei; dry forest; epiphytes phenology; plant-animal interaction; premontane forest.

https://doi.org/10.15517/rev.biol.trop.2022.41703

TERRESTRIAL ECOLOGY

Ant gardens (AGs) are spherical carton

nests built by ants on trees; in them, the ants

deposit seeds of vascular epiphytic plants,

which subsequently germinate (Davidson,

1988). The ants provide these epiphytes with

protection against herbivory through patrol-

ling and a suitable substrate for establishment

(i.e. growth of epiphytes up to the reproduc-

tive phase) through the transport of organic

material (Kleinfeldt, 1978). In turn, the plants

offer structural support and a variety of food

resources to the ants (Madison, 1979).

AGs have been reported in humid tropi-

cal ecosystems and near water bodies (Yu,

1994). The composition of epiphytes in AGs is

influenced by ant species and AG size, among

other factors (Leal et al., 2017). AGs have high

ecological value because the mutualistic rela-

tionship allows them to hold water and nutri-

ents, harbor an associated fauna, and facilitate

527

Revista de Biología Tropical, ISSN: 2215-2075, Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

vascular epiphytes (Céréghino et al., 2010).

However, there are still questions regarding

the natural history of AGs (Leal et al., 2017).

In addition, studies on ecosystems other than

humid tropical forests and Colombian ecosys-

tems, in general, are scarce (Bader, 1999).

Therefore, the objective of this study was

to characterize the spatial distribution, forma-

tion process, diversity, demography, and phe-

nology of AGs in a secondary dry premontane

transitional forest in Colombia.

MATERIALS AND METHODS

Study site: We conduct the study in Colom-

bia, in the rural area of the municipality of Cali,

Pance jurisdiction, El Peón district, in Loma

Larga Reserve, near Farallones National Park.

The geographic coordinates are (3°19’0.7’’ N

- 76°34’39’’ W), with an altitude ranging from

1 100 to 1 250 m.a.s.l. The mean temperature

is 24.8 °C, as recorded at Universidad del Valle

meteorological station (IDEAM, 2018). The

mean annual rainfall is 1 997 mm, as recorded

at La Pajarera gauging station, located in Loma

Larga Reserve (L. G. Naranjo, personal com-

munication, 2019). According to the system

of Espinal (1964), the life zone is a transition

between dry forest and premontane forest. The

habitat corresponds to secondary forest patches

surrounded by ravines, scrub, and livestock

pastures (Sardi et al., 2018).

Species identification and formation

process of Ant gardens: Ants and insects

associated with Ant gardens (AGs) were col-

lected and identified at the entomology sec-

tion of Universidad del Valle using the key by

Longino (2007) and confirmed by J. Longino

(personal communication, 2018). AG epiphyte

and tree samples were collected, processed, and

identified at the CUVC Herbarium of the Uni-

versidad del Valle (Cali, Colombia).

We conducted searches for AGs during

January and February of 2018, on a transect

of 100 m long and 5 m wide, on the banks of

the Quesada River and in the forest. In this

transect, we recorded a total of 170 AGs to

calculate the density (Appendix 1). In addi-

tion, we georeferenced each AG with a Garmin

GPSMAP 60CSx GPS receiver.

We selected and marked 28 AGs of dif-

ferent sizes and heights for monitoring every

two weeks between March 2018 and February

2019. In the biweekly follow-up, we recorded

the use (i.e., presence of ants and epiphytes)

and the size of the AGs, the phenology, and the

demography of their epiphytes. Additionally,

we performed day and night observations of

AGs for 40 hours on different days to describe

ant behavior during AG formation. We priori-

tized the observation of AGs with epiphytes in

the reproductive stage. Finally, we dissected a

large AG and described its structure.

We made AG size and ant behavior obser-

vations for demography and took photographs

and videos with a Canon Powershot Sx530

Hs 50x camera. For heights greater than two

meters, we held a ruler next to the AG attached

to the tip of a rod. We used the photographs to

calculate the size of the AGs.

Spatial distribution and size of Ant gar-

dens: We established six lines 100 m long at

each side and at different distances from the

Quesada River: 0 (bank), 10, 20, 30, 40, and

50 m, for a total of 12 lines, to ensure cover-

age of all distances from the river. Within each

line, we selected at random eight points using a

table of random numbers in Excel. Surrounding

each point, we established a plot of 5 × 5 m, to

complete 16 plots for each distance from the

river and 96 plots in total (2 400 m2). In each

plot, we registered only trees with the presence

of AGs. We recorded the location, height (m),

estimated size (cm), and proximity to the river

(m) of the AGs found, as well as the epiphyte

species present in them. We used the total AGs

registered to calculate the density.

We measured the height of the AGs in

the trees using rods of known length, from the

ground to the point where the AG was located.

The size of the AGs was visually estimated and

categorized following Davidson (1988). We

established three size categories according to

diameter: small (0-19.9), medium (20-39.9),

528 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

and large (≥ 40 cm). We used metric tape to

measure the distance to the river. We performed

observations of AGs located more than 3 m

high with Nikon 8 × 35 binoculars.

Diversity, abundance, phenology, and

demography of epiphytes in Ant gardens:

We recorded the diversity, richness, or number

of species of epiphytes in the 28 AG selected

to follow up. For abundance, we recorded

the number of individuals of each species in

the AGs. For phenology, we established four

phenophases according to Fournier (1974):

Flower buds (Fb), flowers (Fl), immature fruits

(If), and mature fruits (Mf). In each AG, we

recorded the presence of each phenophase

for each epiphyte. For epiphyte demography,

we established three growth stages: seedling,

juvenile and adult. We recorded the number of

individuals of each epiphyte for each growth

stage in the visible half of each AG.

Data analysis: We calculated the den-

sity of AGs on the river bank using the river’s

length measured in ArcGIS® (ESRI, 2011)

and the number of recorded AGs. We evalu-

ated the distribution pattern of the AGs using

a distance-based hypothesis test, with the G

function, within the context of spatial statistics,

the methodology of spatial statistics area data

(point patterns). The G function measures the

distribution of the distances from an arbitrary

point xi to its nearest neighbor, expressed

as follows:

G(r)=(N°{ri ≤ r}) ⁄ n

Where n is the total number of AGs found;

ri is the minimum distance between an AG and

its neighbor; and r is the random variable of the

distance to the nearest AG (Bivand et al., 2013).

We tested the relationship between the

number of AGs and the distance to the water

body with a Pearson (t) correlation analysis

(Freedman et al., 2007). We used a simple

linear model (lm) to establish the relation-

ship between AG size and proximity to the

water body (lm (size ~ proximity)). Likewise,

we used a fitted linear model, using the gls

function (nlme package, weights: VarPower;

Dobson & Barnett, 2008), to establish the

relationship between AG height and proximity

to the water body (gls (height ~ proximity)).

The fitted linear model was necessary due to

the absence of normality in the height variable.

We used a simple linear model (lm) to

evaluate the relationship between AG size and

epiphyte richness. In addition, we performed

a post-ANOVA Tukey’s Test (Tukey, 1977) to

assess which garden size presented greater epi-

phyte richness. Finally, we performed analyses

of richness based on the number of adults and

juvenile individuals because these stages are

more stable in the AGs.

We performed Pearson (t) correlation anal-

ysis between AG growth and size and between

AG growth and rainfall. The growth data were

obtained by standardizing the AG size data dur-

ing the months of measurement (March-July

2018) and fitting them using functions in the

collection of the tidyverse package (Wickham

et. Al, 2019). Adjustment through tidyverse

packages allowed us to evaluate the AG growth

regardless of size. We evaluated assumptions

about the error in the adjusted models through

the Shapiro-Wilk test (Shapiro & Wilk, 1965)

for normality and the Levene’s Test (Levene,

1960) for homoscedasticity. These tests were

performed for all datasets, using a significance

level of < 0.05. For the correlation analysis

among variables, we used the cor.test function

with the Spearman (ρ) and Pearson (t) tests,

according to the normality of each pair of vari-

ables. We performed all the statistical models

and correlation analyses presented above in the

statistical software R® (R Core Team, 2013).

We used the photographs to measure

AG size and perform the demographic count

were processed in ImageJ® software (Version

1.51j8.; Rasband, 2018). We completed the

demographic count of seedlings using the cell

counter tool of this software.

We performed the phenological analysis

using the percentage of intensity of the phe-

nophase in each species. We calculated the

intensity from the number of AGs that had a

529

Revista de Biología Tropical, ISSN: 2215-2075, Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

certain phenophase over the total AGs with

the presence of the epiphyte in the adult stage.

We compared phenological and demographic

data with rainfall using correlation analysis.

We used Spearman correlation for phenologi-

cal analyses except for Epiphyllum phyllanthus

(Pearson’s test). We performed demographic

analysis using Spearman’s correlation test. We

used the classification reported by Newstrom

et al. (1994) to describe the frequency of the

phenophases of the epiphyte species.

RESULTS

Species, density, and formation process

of Ant gardens: We recorded ten epiphyte

species (Table 1, Fig. 1) and 13 tree spe-

cies associated with the Ant gardens (AGs)

(Table 2). We identified the AG-forming ant as

Azteca ulei (Fig. 2). The hemipterans associ-

ated with ants in the AGs belonged to the fam-

ily Coccidae and were tended by the ants inside

channel-shaped cavities.

A summary of the natural history obser-

vations can be found in the supplementary

information (Appendix 2) and discussed in

the following section. Patrolling and foraging

were the main activities performed by A. ulei.

This ant performed patrolling on some of the

AG epiphytes (Appendix 2). During patrol-

ling, A. ulei attacked the majority of insects or

vertebrates that visited the AGs. The objects

foraged by A. ulei included various materials,

particularly seeds and pieces of fruits from

certain epiphytes in the AGs (Appendix 2). In

general, the observed AGs had active ants dur-

ing the monitored year, except for some AGs

that were abandoned.

The dissected AG was constructed from

plant material, was light brown in its inte-

rior, and contained multiple connected cavities.

Internally we observed multiple roots, pieces of

leaves, and small hollow branches that bounded

the cavities, all held together by a matrix of

apparently processed plant material. The A.

ulei from the dissected AG were polymorphic

TABLE 1

Diversity and abundance of epiphytic plants from the ant gardens in the Loma Larga Reserve, Valle del Cauca, Colombia

Ant garden size Small

(0-19.9 cm)

Medium

(20-39.9 cm)

Large

(≥ 40 cm)

Family Species Total

Araceae Anthurium gracile (Rudge) Schott 1,2,5 2 6 3 11

Araceae Anthurium obtusum (Engl.) Grayum 1,2,5 0 0 3 3

Araceae Philodendron barrosoanum G.S. Bunting5,6 0 0 1 1

Bromeliaceae Aechmea angustifolia Poepp. & Endl. 1,2,5 7 13 4 24

Cactaceae Epiphyllum phyllanthus (L.) Haw. 1,2,4 3 13 5 21

Cactaceae Rhipsalis baccifera (Sol.) Stearn.60 1 0 1

Gesneriaceae Drymonia serrulata (Jacq.) Mart. 6,7 - - - -

Moraceae Ficus paraensis (Miq.) Miq. 1,2,4 5 11 4 20

Orchidaceae Epidendrum flexuosum G. Mey. 1,2,3 3 6 3 12

Piperaceae Peperomia rotundifolia (L.) Kunth 5,6 1 0 0 1

Reproductive status (%) 28.6 66.7 83.3

Average number of species 3.0 3.3 3.8

1. Epiphytes where Azteca ulei were observed carrying their seeds to the ant garden.

2. Epiphytes in which Azteca ulei patrolled seedlings and reproductive adults, epiphytes.

3. Found only in ant gardens.

4. Found rarely outside of ant gardens (usually near them).

5. Found inside and outside of ant gardens.

6. Recorded only once in ant gardens.

7. D. serrulata was not found in monitored ant gardens, but it was found in the initial census.

530 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

Fig. 1. Epiphytic plants of the Azteca ulei ant gardens in the Loma Larga Reserve, Valle del Cauca, Colombia. a) Anthurium

gracile, b) Anthurium obtusum, c) Drymonia serrulata, d) Epidendrum flexuosum, e) Epiphyllum phyllanthus, f) Ficus

paraensis, g) Philodendron barrosoanum (Photo: J.M. Ruiz), h) Peperomia rotundifolia, i) Rhipsalis baccifera, j) Aechmea

angustifolia. Photos: Á. Barrera.

531

Revista de Biología Tropical, ISSN: 2215-2075, Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

Fig. 2. Individuals of Azteca ulei collected in ant gardens. a.) Pupa, b.) General view of worker, c.) Winged male, d.) Frontal

view of worker’s head. Photos: V. Sarria.

TABLE 2

Diversity, abundance and proportion of trees from the ant gardens in the Loma Larga Reserve, Valle del Cauca, Colombia

Family Species Abundance (n) Proportion (%)

Araliaceae Dendropanax cf. arboreus (L.) Decne. & Planch. 1 3.6

Erythroxilaceae Erythroxylum citrifolium A. St.-Hil. 1 3.6

Fabaceae Inga cf. ornata Kunth 2 7.1

Fabaceae Calliandra pittieri Standl. 5 17.9

Lauraceae Cinnamomum triplinerve (Ruiz & Pav.) Kosterm. 3 10.7

Melastomataceae Henriettea seemannii (Naudin) L.O. Williams 3 10.7

Melastomataceae Miconia matthaei Naudin 3 10.7

Meliaceae Guarea guidonia (L.) Sleumer 1 3.6

Meliaceae Trichilia pallida Sw. 2 7.1

Moraceae Ficus paraensis (Miq.) Miq. 2 7.1

Myrtaceae Eugenia cf. egensis DC. 3 10.7

Myrtaceae Myrcia cf. popayanensis Hieron. 1 3.6

Myrtaceae Syzygium jambos (L.) Alston 1 3.6

532 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

workers, winged males, and pupae (Fig. 2). We

did not found queens.

Spatial distribution and size of Ant gar-

dens: The AG density was 65/ha (1/154 m2) in

the plots and 1/10 m along the river. Within the

study site, the distribution of AGs presented an

aggregate pattern, such that 32 % of AGs (n

= 170) were recorded in groups of two to five

AGs in the same tree. Regarding vertical distri-

bution, AGs were found at ground level (fallen)

up to 15 m, with a mean height of 5 m.

The number of AGs increased signifi-

cantly near the river (t = -3.9907, P = 0.0162)

and was almost zero 20 m from the riverbank

(Fig. 3). However, we did not find a relation-

ship between AG height and proximity to the

river (gls, P = 0.5791) or between AG size and

proximity to the river (lm, P = 0.5018).

The AGs had a mean diameter of 30.1

cm, ranging from 4-61.6 cm. We did not find

a trend for AG growth, nor was there varia-

tion in the growth of AGs of different sizes (t

= -0.5182, P = 0.5549). Moreover, AG growth

was not dependent on rainfall.

Diversity, abundance, phenology, and

demography of epiphytes in Ant gardens:

Epiphyte richness in AGs consisted of ten

species and was higher in large AGs (lm, P =

0.0110). The abundance of these epiphytes and

their restriction to the AGs are shown in Table

1. Differences were found between the richness

of large and small AGs (Tukey, P = 0.037) and

between small and medium AGs (Tukey, P =

0.030). On average, we recorded three species

of epiphytes in each AG, with variations in AG

size (Table 1). The richness of trees was 13

species; their abundances are shown in Table 2.

Among the observed AGs, 60 % had

epiphytes with some reproductive phenophase

during the year of study and had diameters

between 21.5 and 62 cm. In addition, we regis-

tered reproductive phenophases in 83 % of the

large AGs, 87 % of the medium AGs, and none

of the small AGs during the studied year.

Aechmea angustifolia and Epidendrum

flexuosum had a sub-annual (bimodal) pheno-

logical pattern for most of its phenophases (Fig.

4A, Fig. 4B). For both species, we observed

an increase in the intensity of the flower bud

and flower phenophases in the dry season. The

flower bud and flower phenophases showed a

significant and negative relationship with the

cumulative rainfall for two, three, and four

weeks before the event for both species (Spear-

man correlation). The mature fruit phenophase

increased in the rainy season in both species,

with a significant positive relationship between

this phenophase and cumulative rainfall four

Fig. 3. Number of Azteca ulei ant gardens found at different distances from the Quesada River, in the Loma Larga Reserve,

Valle del Cauca, Colombia. Vertical lines = ± S.D.

533

Revista de Biología Tropical, ISSN: 2215-2075, Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

months before the event only for E. flexuosum.

For Anthurium gracile (Fig. 4C), the pattern of

phenophases was the same as A. angustifolia

and E. flexuosum, with no correlation between

the phenophases and rainfall. Anthurium obtu-

sum did not show a trend in its phenophases

(Fig. 4D), and Epiphyllum phyllanthus showed

a continuous pattern in its phenophases (Fig.

4E), without any correlation between the phe-

nophases and rainfall.

The proportion of the epiphyte growth

stages varied according to AG size, and there

was also a variation in the number of individuals

at each stage during the count months. Only the

seedlings had a positive correlation with rain-

fall (ρ = 0,6914; P = 0,0064), where the number

of individuals was reduced in the dry season.

The population size of the epiphytes of

AGs also varied during the count months (Fig.

5). A. obtusum, E. flexuosum, and A. gracile

had smaller populations (< 100 individuals),

while E. phyllanthus, A. angustifolia, and Ficus

paraensis had larger population sizes (160-753

individuals). F. paraensis was the only species

with a positive relationship with rainfall (ρ =

0.5859, P = 0.0226).

Fig. 4. Phenology of epiphytic plants from ant gardens growing in the Loma Larga Reserve, Valle del Cauca, Colombia. A.

Aechmea angustifolia (n = 16). B. Epidendrum flexuosum (n = 9). C. Anthurium gracile (n = 5). D. Anthurium obtusum

(n=3). E. Epiphyllum phyllanthus (n = 18). Blue bars ( ) represent precipitation (mm); dotted line ( ) represents flower

buds; dashed line ( ) represents flowers; open squares line ( ) represents immature fruits; closed squares line ( )

represents mature fruits.

534 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

DISCUSSION

Species and formation process of Ant

gardens: The morphological characteristics

of the ants in the Ant gardens (AGs) resemble

those of Azteca gnava from Costa Rica and A.

ulei from Brazil. We chose A. ulei because it is

the oldest name (Longino, personal communi-

cation, 2018). Furthermore, there are no known

differences between A. gnava and A. ulei, sug-

gesting that they are closely related, are the

same species, or perhaps are a species complex

(Longino, 2007; Ward, 2019).

The Coccidae found in the Loma Larga

AGs are phloem-sucking insects associated

with AGs (Youngsteadt et al., 2008) and with

ants of the genus Azteca (Catling, 1997) and

serve as a food source for the ants (Way, 1963).

In this regard, Weissflog et al., (2017) suggest

that the feeding of Coccidae on the host plant

phloem is a mechanism that has a nutritional

function and provides moisture to the AG.

The care of hemipterans on E. flexuosum

by A. ulei could be related to the restricted

occurrence of this epiphyte in AGs and could

compromise the condition of this epiphyte

(Kaufmann, 2002). However, we did not record

plants of E. flexuosum in poor condition.

A. gracile, E. flexuosum, E. phyllanthus, P.

rotundifolia and F. paraensis were reported as

epiphytes of AGs in previous studies (David-

son, 1988; Morales-Linares et al., 2017; Orivel

& Leroy, 2010). However, the identification of

A. angustifolia and A. obtusum as epiphytes of

neotropical AGs is novel.

As in other ant species that build AGs

(Kleinfeldt, 1978; Vantaux et al., 2007), patrol-

ling and foraging are performed throughout the

day by A. ulei. The foraging performed by AG

ants such as A. ulei includes the lodging of par-

ticles in the walls of the internal cavities of the

AGs, which enriches the substrate (Weissflog et

al., 2017). The patrolling carried out by A. ulei

safeguards the food and structural resources

offered by the plants. Among the frequent

epiphytes in AGs, A. angustifolia, A. gracile,

A. obtusum, E. flexuosum, E. phyllanthus,

and F. paraensis provided structural and food

rewards for A. ulei, which corroborates their

contribution to the mutualistic relationship and

have been reported previously (Catling, 1995;

Davidson, 1988; Davidson & Epstein, 1989).

Fig. 5. Population size changes of epiphytes found in Azteca ulei´s ant gardens, during March 2018 to February 2019, in

the Loma Larga Reserve, Valle del Cauca, Colombia. Closed squares line ( ) represents Aechmea angustifolia; dotted

line ( ) represents Anthurium gracile, black rhombuses line ( ) represents Anthurium obtusum; black triangles line

() represents Epidendrum flexuosum; double continuous line ( ) represents Epiphyllum phyllanthus; dashed line

() represents Ficus paraensis. Blue bars ( ) represent precipitation (mm).

535

Revista de Biología Tropical, ISSN: 2215-2075, Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

We explain the presence of the orchid

E. flexuosum in AGs by the active transport

of their tiny seeds (5-9.5 mm), which they

carry, held on their mandibles, one by one, as

observed in this study and by Morales-Linares

et al., (2018). These findings are evidence

against other hypotheses for the presence of

orchids in AGs by mechanisms such as wind

dispersal (Catling, 1995) or by the opportu-

nistic establishment of AGs around orchids

(Kaufmann et al., 2001).

The observation of the growth stages of F.

paraensis in the AGs shows that this hemiepi-

phyte, in addition to providing food via its

fruits (Yu, 1994), offers two types of structural

support for AGs. First, in the juvenile stage, F.

paraensis generates roots that support the AG;

then, in the adult stage, it develops adventi-

tious roots in the fork of its branches that

create spaces conducive to the accumulation

of detritus by the ants, which allows the forma-

tion of AGs (Bain et al., 2014). Consequently,

the F. paraensis individuals sown by the ants

increase the space for the A. ulei colony to

build new AGs.

The abandonment by the ants of certain

AGs with big plants showed the AG allows the

maintenance of the substrate and prevents the

plants from dehydrating. Schmit-Neuerburg

& Blüthgen (2007) observed that ants con-

stantly repair the substrate and incorporate

new materials.

Spatial distribution and size of Ant gar-

dens: The AGs of the studied dry premontane

transitional forest showed an aggregated distri-

bution pattern, which coincides with the poly-

domy of A. gnava colonies (Longino, 2007).

This aggregated distribution pattern could also

be explained by the abundance of AGs in the

riparian habitat. The abundance of AGs in these

open habitats could be due to a greater supply

of resources and moisture (Yu, 1994). However,

the vicinity to the river of AGs did not mean an

increase in the size of the gardens. The size of

AGs in this study was similar to that reported

for AGs of A. gnava in other ecosystems (> 60

cm) (Longino, 2007; Morales-Linares et al.,

2017). We suggest that the ants’ behavior deter-

mines the size of AGs and not the microhabitat.

Diversity, abundance, phenology, and

demography of epiphytes in Ant gardens:

The diversity of epiphytes in the dry forest is

lower than in other ecosystems due to the water

stress experienced by these plants (Gentry &

Dodson, 1987; Werner & Gradstein, 2009).

This pattern is consistent with the richness

of ten epiphytes found in the AGs compared

with that involving up to 26 epiphytes in

AGs of A. cf. gnava in humid tropical forests

(Morales-Linares et al., 2016; Morales-Linares

et al., 2017).

Nevertheless, the mean diversity of three

epiphytes per AG was similar to that found

by Morales-Linares et al., (2016), along with

a pattern of greater richness in larger AGs

(Catling, 1997; Morales-Linares et al., 2017).

This positive relationship between richness and

size can be understood through island bioge-

ography theory (MacArthur & Wilson, 1967),

where a larger area provides greater possibili-

ties of colonization by epiphytes.

A. angustifolia and E. phyllanthus were

the most frequent epiphytes in AGs, possibly

due to their adaptations to water stress (Benz-

ing, 1990; Kaufman, 2002). Despite not having

evident anatomical or metabolic adaptations

against desiccation, F. paraensis was also very

common in AGs, which shows that the mutual-

istic interaction with A. ulei could be protecting

it against seedlings desiccation (Schmit-Neuer-

burg & Blüthgen, 2007), in addition to the high

number of its seeds that were observed germi-

nating in the AGs.

Regarding the trees, arboreal ants are more

abundant in trees with a capacity to host hemip-

terans or those that have extrafloral nectaries

(Blüthgen et al., 2000; Morales-Linares et al.,

2016). However, Calliandra pittieri was the

most frequent tree species for AGs despite not

having these characteristics. Therefore, the fre-

quency of AGs in C. pittieri may be because it

is one of the most abundant trees in this forest

(Sardi et al., 2018) and not a possible prefer-

ence by A. ulei.

536 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

The AGs allowed the establishment of

adult epiphytes in reproductive status, where

the bimodal sub-annual phenological pattern

was the most common (i.e., A. angustifolia,

A. gracile, E. flexuosum). The flowering peaks

in the dry season and fruiting peaks during

the rainy season registered for these species

is a frequent pattern in tropical forest plants

(Sahagun-Godinez, 1996). This pattern can be

explained by proximate causes, which correlate

with environmental phenomena (Van Schaik

et al., 1993; Williams-Linera & Meave, 2002).

The flowering of A. angustifolia, A. grac-

ile, and E. flexuosum in the dry season could

be due to endogenous factors such as the water

state of the plant, which in turn responds to

the seasonal availability of water in the envi-

ronment. In this case, these epiphytes have

organs or mechanisms to accumulate moisture;

therefore, they can trigger flowering in the dry

season by collecting water during the rainy sea-

son (Borchert, 1994). In turn, fruit production

in the rainy season is common in tropical dry

forests (Levey, 1988). Therefore, it could have

as a proximate cause the high water demand

required by fleshy fruits (e.g., A. angustifolia,

A. gracile) for their development and turgor

(Lieberman, 1982).

E. phyllanthus was the only epiphyte with

a continuous phenological pattern. The succu-

lence of its leaves likely allows it to retain water

at a level where its reproductive phenophases

are independent of rainfall.

Aridity is the most significant physical

threat to seedlings because they dehydrate

more quickly than at more advanced growth

stages due to their higher surface/volume ratio

and smaller size (Benzing, 1990). The number

of seedlings established during the dry sea-

son decreases, and it marks the limit of the

number of adult individuals that remain in the

AGs of A. ulei.

Similar to seedling survival, the variation

in the population sizes of the epiphytes could

be related to the presence of adaptive, mor-

phological, and metabolic strategies against

desiccation, which includes their occurrence

in AGs. For this reason, all epiphytes, except

F. paraensis, showed a demographic pattern

independent of rainfall because they have these

strategies. However, the high number of F.

paraensis individuals that germinate during the

rainy season could compensate for the intense

pressure imposed by the dry season on the

growth of these seedlings and, again, explain

why F. paraensis is one of the most frequent

epiphytes in AGs.

In conclusion, the mutualism between A.

ulei and the epiphyte species growing in their

AGs is partially influenced by the seasonal-

ity of an ecosystem with dry forest elements.

Nevertheless, the information on the spatial

distribution, the formation process of AGs,

diversity, phenological and demographic pat-

terns indicate that AGs built by A. ulei are

favorable microhabitats for epiphytes germina-

tion, establishment, and reproduction in this

dry premontane transitional forest.

Ethical statement: the authors declare

that they all agree with this publication and

made significant contributions; that there is no

conflict of interest of any kind; and that we fol-

lowed all pertinent ethical and legal procedures

and requirements. All financial sources are

fully and clearly stated in the acknowledge-

ments section. A signed document has been

filed in the journal archives.

ACKNOWLEDGMENTS

We thank the Ecological Committee of

Loma Larga Reserve for allowing us to conduct

the study at this site. We also thank D. Ramírez,

E. López, M. Zuluaga, G. Vargas, V. Sarria, J.

Giraldo, W. Wallens, A. Ramírez, J. Sarmiento,

C. Castaño, M. Erazo, G. Torres, N. and E.

Wheelwright for participating in the fieldwork.

Special thanks to G. Corredor, J. M. Ruiz, D.

Ruiz, J. Montaño, A. Betancourt, P. Silverstone,

N. Wheelwright, A. Zuluaga, and G. Kattan

for data analysis support and the continuous

feedback; to L.G. Naranjo for supplying the

rainfall data; to A. Betancourt and J. Longino

for the identification of the entomological

material. We thank Universidad del Valle for

537

Revista de Biología Tropical, ISSN: 2215-2075, Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

the translation support through native English-

speaking editors at AJE.

RESUMEN

Ecología básica de jardines de hormigas en un bosque

transicional seco-premontano

Introducción: Los “jardines de hormigas” son nidos de

hormigas en árboles, que se localizan a diferentes alturas,

sobre los que germinan plantas epífitas vasculares que han

sido transportadas y sembradas previamente por hormigas.

Aunque esta relación mutualista ha sido estudiada en

ecosistemas húmedos tropicales, la información en otros

ecosistemas tropicales y en Colombia, es escasa.

Objetivo: Caracterizar la distribución, construcción, diver-

sidad, demografía y fenología de los hormigueros en bos-

que tropical seco.

Métodos: En enero y febrero de 2018, identificamos 170

hormigueros en un transecto de 100x5 m a orillas del río

Quesada, Colombia, y en el bosque secundario premontano

seco de transición adyacente; monitoreamos cambios, para

jardines y plantas, en 28 de ellos, cada dos semanas (marzo

2018-febrero 2019).

Resultados: Los jardines, construidos por Azteca ulei, se

agregaron cerca de cuerpos de agua; tenían 10 especies de

epífitas y estaban sobre 13 especies arbóreas. Los jardines

más grandes tenían más especies epífitas. Algunas epífitas

tuvieron un patrón fenológico bimodal. Se establecen

menos plántulas en la estación seca y quedan menos adul-

tos en los jardines. Las características del jardín benefician

tanto a las epífitas como a las hormigas.

Conclusiones: Los jardines construidos por A. ulei tienen

características ecológicas que favorecen la germinación,

establecimiento y reproducción de diversas epífitas en este

ecosistema tropical seco, incluyendo la agregación cerca

de cursos de agua.

Palabras clave: Azteca ulei; bosque seco; fenología de

epífitas; interacción planta-animal; bosque premontano.

APPENDIX 1

Study area in the Loma Larga Reserve, Valle del Cauca, Colombia. The blue line corresponds to the

La Quesada River, the red dots to the ant gardens and the blue box to the sampling plot.

538 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

REFERENCES

Bader, M. (1999). The ecology of ant gardens in the Colom-

bian Amazon. Wageningen University, Netherlands.

Bain, A., Harrison, R., & Schatz, B. (2014). How to be an

ant on figs. Acta Oecologica, 57, 97–108. https://doi.

org/10.1016/j.actao.2013.05.006

Benzing, D. (1990). Vascular epiphytes. General biology

and related biota. Journal of Tropical Ecology, 8(1),

55–56. https://doi.org/10.1017/S0266467400006088

Bivand, R., Pebesma, E., & Gómez-Rubio, V. (2013).

Applied Spatial Data Analysis with R. Springer.

https://doi.org/10.1007/978-1-4614-7618-4

Blüthgen, N., Verhaagh, M., Goitía, W., Jaffé, K., Morawetz,

W., & Barthlott, W. (2000). How plants shape the ant

community in the Amazonian rainforest canopy:

the key role of extrafloral nectaries and homopteran

honeydew. Oecologia, 125(2), 229–240. https://doi.

org/10.1007/s004420000449

Borchert, R. (1994). Soil and stem water storage deter-

mine phenology and distribution of tropical dry

forest trees. Ecology, 75(5), 1437–1449. https://doi.

org/10.2307/1937467

Catling, P. (1995). Evidence for partitioning of Belezean

Ant Nest substrate by a characteristic flora. Biotropi-

ca, 27(4), 535–537. https://doi.org/10.2307/2388970

APPENDIX 2

Behavior and main activities of Azteca ulei recorded during observation periods in the ant gardens

(AG) growing in the Loma Larga Reserve, Valle del Cauca, Colombia

Observed feature Description

Tending of coccids Generally, individuals of A. ulei tended the coccids inside cavities built on branches of

shrubs near the ant garden or the tree of the ant garden. In cleaning the coccids, the ants

constantly moved their antennae over them. We frequently observed that those ants that

emerged from the cavities had more bulging and transparent abdomens.

Accumulation of debris on

Epidendrum flexuosum

We observed accumulation of debris by individuals of A. ulei. These ants formed similar

structures to coccidian cavities, but in this case, on stems and sutures of Epidendrum flexuo-

sum fruits. However, we did not observe coccidians inside. The height of AG did not allow

a detailed observation. The AG that was more easily observable, the accumulation of debris

did not yet form these cavities.

Patrolling Individuals of A. ulei performed continuous movements, although they also remained

motionless, on leaf blades, flower buds, flowers, and fruits of epiphytes. The density of ant

individuals or their greater patrolling did not seem to be determined by the epiphyte species

but by the growth stage to which it belonged. We observed greater patrolling in seedlings

and adults with flowers or fruits.

Patrolling Insect attack (e.g. Salticidae spiders, Camponotus ants).

Patrolling Individuals of Azteca ulei climbed on the paws and beaks of birds (e.g. Catharus ustulatus,

Manacus vitellinus, and Tolmomyias sulphurescens) when they were attacking ant gardens.

Patrolling In contrast to the continuous attack, we recorded other species of ants (Pseudomyrmex

or Crematogaster) on immature fruits of E. flexusosum. These ants were not attacked by

Azteca ulei or by spiders that wove their webs a few inches from the ant garden.

Foraging We observed the ants carried dry plant material outside and fresh material inside the AG.

The dry plant material was fragments of stems, leaves, and the fresh plant material was flow-

ers, fruits, and seeds of epiphytes. Also, the ants moved exoskeletons (e.g. exuvia, wings,

legs, whole bodies of insects) and feathers, to the interior of the AG.

Debris accumulation and

abandonment of ant gardens

In some small ant gardens, the accumulation of debris on the surface was noticeable.

However, these smaller gardens also lost substrate after heavy rain or were abandoned by

the ants. At the end of the observations, the abandoned ant gardens showed a decrease in

seedlings and substrate.

Abandonment of ant gardens A large ant garden remained ant-free during the monitoring year but had adult epiphytes

of E. phyllanthus, on which the bird Forpus conspicillatus foraged on the immature fruits

539

Revista de Biología Tropical, ISSN: 2215-2075, Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

Catling, P. (1997). Influence of aerial Azteca nests on

the epiphyte community of some belizean orange

orchards. Biotropica, 29(2), 237–242. https://doi.

org/10.1111/j.1744-7429.1997.tb00031.x

Céréghino, R., Leroy, C., Dejean, A., & Corbara, B. (2010).

Ants mediate the structure of phytotelm communities

in an ant-garden bromeliad. Ecology, 91(5), 1549–

1556. https://doi.org/10.1890/09-1534.1

Davidson, D. (1988). Ecological studies of neotropical

ant gardens. Ecological Society of America, 69(4),

1138–1152. https://doi.org/10.2307/1941268

Davidson, D., & Epstein, W. (1989). Epiphytic associations

with ants. In U. Lüttge (Ed.), Vascular plants as epi-

phytes (Vol. 76, pp. 200–233.). Springer. https://doi.

org/10.1007/978-3-642-74465-5_8

Dobson, A. J., & Barnett, A. G. (2008). An introduction to

generalized linear models. Chapman & Hall/CRC.

Espinal, L. (1964). Formaciones vegetales del departa-

mento de Antioquia. Revista Facultad Nacional de

Agronomía, 24(60), 1–81.

ESRI (2011). ArcGIS Desktop (Version 10.3, Computer

software). Environmental System Research Institute,

Redlands, CA, USA. http://desktop.arcgis.com/en/

arcmap

Fournier, L. (1974). Un método cuantitativo para la

medición de características fenológicas en árboles.

Turrialba, 24(4), 422–423.

Freedman, D., Pisani, R., & Purves, R. (2007). Statistics

(4th Ed). W. W. Norton & Company.

Gentry, A., & Dodson, C. (1987). Contribution of

Nontrees to Species Richness of a Tropical Rain

Forest. Biotropica, 19(2), 149–156. https://doi.

org/10.2307/2388737

IDEAM. (2018). Datos meteorológicos en la estación de

la Universidad del Valle de Cali. www.ideam.gov.co

Kaufmann, E. (2002). Southeast Asian Ant-gardens:

Diversity, Ecology, Ecosystematic Significance, and

Evolution of Mutualistic Anti-epiphyte Associations

(Doctoral dissertation). Universität Frankfurt am

Main, Germany. https://core.ac.uk/reader/14505280

Kaufmann, E., Weissflog, A., Hashim, R., & Maschwitz,

U. (2001). Ant-gardens on the giant bamboo Giganto-

chloa scortechinii (Poaceae) in West-Malaysia. Insec-

tes-Sociaux, 48(2), 125–133. https://doi.org/10.1007/

PL00001754

Kleinfeldt, S. (1978). Ant-gardens: the interaction of Codo-

nanthe crassifolia (Gesneriaceae) and Crematogaster

longispina (Formicidae). Ecology, 59(3), 449–456.

https://doi.org/10.2307/1936574

Leal, L., Jacovak, C., Bobrowiec, P., Camargo, J., &

Peixoto, P. (2017). The role of parabiotic ants and

environment on epiphyte composition and protection

in ant gardens. Sociobiology, 64(3), 276–283. https://

doi.org/10.13102/sociobiology.v64i3.1219

Levene, H. (1960). Robust tests for equality of variances.

In I. Olkin & H. Hotelling (Eds.), Contributions to

Probability and Statistics: Essays in Honor of Harold

Hotelling (pp. 278–292). Stanford University Press.

Levey, D. (1988). Spatial and temporal variation in Costa

Rican fruit and fruit-eating bird abundance. Eco-

logical Monographs, 58(4), 251–269. https://doi.

org/10.2307/1942539

Lieberman, D. (1982). Seasonality and phenology in a dry

tropical forest in Ghana. British Ecological Society,

70(3), 791–806. https://doi.org/10.2307/2260105

Longino, J. (2007). A taxonomic review of the genus Azte-

ca (Hymenoptera: Formicidae) in Costa Rica and a

global revision of the aurita group. Zootaxa, 1491(1),

1–63. https://doi.org/10.11646/zootaxa.1491.1.1

MacArthur, R., & Wilson, E. (1967). The Theory of Island

Biogeography. Princeton University Press.

Madison, M. (1979). Additional observations on ant-gar-

dens in Amazonas. Selbyana, 2(1), 133–148.

Morales-Linares, J., García-Franco, J., Flores-Palacios, A.,

& Valenzuela-González, J. (2017). Spatial structure

of ant-gardens: vertical distribution on host trees

and succession/segregation of associated vascular

epiphytes. Journal of Vegetation Sciences, 28(10),

1036–1046. https://doi.org/10.1111/jvs.12559

Morales-Linares, J., García-Franco, J., Flores-Palacios,

A., Valenzuela-González, J., Mata-Rosas, M., &

Díaz-Castelazo, C. (2016). Vascular epiphytes and

host trees of ant-gardens in an anthropic landscape in

southeastern Mexico. Naturwissenschaften, 103(11-

12), 96. https://doi.org/10.1007/s00114-016-1421-9

Morales-Linares, J., García-Franco, J., Flores-Palacios, A.,

Valenzuela-González, J., Mata-Rosas, M., & Díaz-

Castelazo, C. (2018). Orchid seed removal by ants

in Neotropical ant-gardens. Plant Biology, 20(3),

525–530. https://doi.org/10.1111/plb.12715

Newstrom, L., Frankie, G., & Baker, H. (1994). A new clas-

sification for plant phenology based on flowering pat-

terns in lowland tropical rain forest trees at La Selva,

Costa Rica. Biotropica, 26(2), 141–159. https://doi.

org/10.2307/2388804

Orivel, J., & Leroy, C. (2010). The diversity and ecology

of ant gardens (Hymenoptera: Formicidae; Sperma-

tophyta: Angiospermae). Myrmecological News, 14,

73–85.

R Core Team (2013). R: A language and environment for

statistical computing (Version 3.5.2, R Core Team,

540 Revista de Biología Tropical, ISSN: 2215-2075 Vol. 70: 526-540, Enero-Diciembre 2022 (Publicado Ago. 03, 2022)

Computer software). Vienna, Austria. http://www.r-

project.org

Rasband, W. (2018). ImageJ (Computer software). Bethes-

da: National Institutes of Health. https://imagej.nih.

gov/ij

Sahagun-Godinez, E. (1996). Trends in the Phenolo-

gy of Flowering in the Orchidaceae of Western

Mexico. Biotropica, 28(1), 130–136. https://doi.

org/10.2307/2388778

Sardi, A., Torres, A., & Corredor, G. (2018). Diversidad flo-

rística en un paisaje rural del piedemonte de los Fara-

llones de Cali, Colombia. Colombia Forestal, 21(2),

142–160. https://doi.org/10.14483/2256201X.10866

Schmit-Neuerburg, V., & Blüthgen, N. (2007). Ant-garden

epiphytes are protected against drought in a Venezue-

lan lowland rain forest. Ecotropica, 13(2), 93–100.

Shapiro, S. S., & Wilk, M. B. (1965). An analysis of

variance test for normality (complete samples). Bio-

metrika, 52(3–4), 591–611. https://doi.org/10.1093/

biomet/52.3-4.591

Tukey, J. W. (1977). Exploratory data analysis (Vol. 2).

Reading Mass.

Van Schaik, C., Terborgh, J., & Wright, J. (1993). The

Phenology of Tropical Forests: Adaptive Significance

and Consequences for Primary Consumers. Annual

Review of Ecology and Systematics, 21, 353–377.

https://doi.org/10.1146/annurev.es.24.110193.002033

Vantaux, A., Dejean, A., Dor, A., & Orivel, J. (2007). Para-

sitism versus mutualism in the ant-garden parabiosis

between Camponotus femoratus and Crematogaster

levior. Insectes-Sociaux, 54(1), 95–99. https://doi.

org/10.1007/s00040-007-0914-0

Ward, P. (2019). Species: Azteca gnava - AntWeb. www.

antweb.org/description.do?name=gnava&genus=azte

ca&rank=species&project=worldants

Way, M. (1963). Mutualism between ants and honeydew-

producing homoptera. Annual Review of Entomolo-

gy, 8(1), 307–344. https://doi.org/10.1146/annurev.

en.08.010163.001515

Weissflog, A., Kaufmann, E., & Maschwitz, U. (2017). Ant

gardens of Camponotus (Myrmotarsus) irritabilis

(Hymenoptera : Formicidae : Formicinae) and Hoya

elliptica (Apocynaceae) in Southeast Asia. Asian

Mirmercology, 9, e009001. https://doi.org/10.20362/

am.009001

Werner, F., & Gradstein, S. (2009). Diversity of dry

forest epiphytes along a gradient of human dis-

turbance in the Tropical Andes. Journal of

Vegetation Science, 20(1), 59–68. https://doi.

org/10.1111/j.1654-1103.2009.05286.x

Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan,

D. A. L., François, R., Grolemund, G., Hayes, A.,

Henry, L., Hester, J., Kuhn, M., Lin Pedersen, T.,

Miller, E., Milton Bache, S., Müller, K., Ooms, J.,

Robinson, D., Seidel, D. P., Spinu, V., … Yutani, H.

(2019). Welcome to the Tidyverse. (Version 1.3.0, R

package). Journal of Open Source Software, 4(43),

1686. 10.21105/joss.01686

Williams-Linera, G., & Meave, J. A. (2002). Patrones

fenológicos. In M. Guariguata, G. Kattan, & M. Gua-

riguata (Eds.), Ecología y Conservación de Bosques

Neotropicales (pp. 408–431). Ediciones LUR.

Youngsteadt, E., Nojima, S., Häberlein, C., Schulz, S.,

& Schal, C. (2008). Seed odor mediates an obliga-

te ant-plant mutualism in Amazonian rainforests.

Proceedings of the National Academy of Scien-

ces, 105(12), 4571–4575. https://doi.org/10.1073/

pnas.0708643105

Yu, D. W. (1994). The structural role of epiphytes in

ant gardens. Biotropica, 26(2), 222–226. https://doi.

org/10.2307/2388813