Behavioral Ecology and Sociobiology

, Volume 70, Issue 12, pp 2195–2201

The frog Lithodytes lineatus (Anura: Leptodactylidae) uses chemical recognition to live in colonies of leaf-cutting ants of the genus Atta (Hymenoptera: Formicidae)

  • André de Lima Barros
  • Jorge Luis López-Lozano
  • Albertina Pimentel Lima
Original Article

DOI: 10.1007/s00265-016-2223-y

Cite this article as:
de Lima Barros, A., López-Lozano, J.L. & Lima, A.P. Behav Ecol Sociobiol (2016) 70: 2195. doi:10.1007/s00265-016-2223-y

Abstract

Chemical-based mimicry and camouflage are known to be employed by invertebrate parasites of social insect colonies, but the use of this strategy by vertebrates to avoid being detected by social insects has received less attention. In this paper, we examine the hypothesis that frog Lithodytes lineatus has skin chemicals that imitate chemical recognition used by leaf-cutting ants of genus Atta. We show that individuals of Lithodytes lineatus were never attacked by the leaf-cutting ants of genus Atta, while 100 % of four other anuran species were. In addition, none of the ten individuals of frog Rhinella major coated with skin extracts of frog L. lineatus were attacked, whereas controls (coated with ultrapure water) were attacked on each occasion. Our results demonstrate that the skin of frog Lithodytes lineatus has chemicals that prevent the attack of both species of leaf-cutting ants, Atta laevigata and Atta sexdens.

Significance statement

In order to reduce the risk of predation, some frog species engage in commensal or mutualistic relationships with invertebrates, but associations between frogs and ants are rarely reported in literature. We show that frog Lithodytes lineatus are not attacked by ants Atta laevigata and A. sexdens; however, other frog species are aggressively attacked. Our results suggest that the biomolecules present in the frog skin are capable of inhibiting the attack of ants, allowing coexistence. This is the first study reporting the possible mechanism for association between frog L. lineatus and ants of genus Atta.

Keywords

Chemical recognition Ant Atta laevigata Atta sexdens Lithodytes lineatus 

Introduction

Communication between colony members of social insects or parasites of these colonies often has a chemical basis. In the case of parasites, chemicals serve to mimic or mask their presence within the colony (Lenoir et al. 2001; Hojo et al. 2009; Tsuneoka and Akino 2009, 2012; Menzel et al. 2014). Frogs have many chemical defenses in their skins, generally used to protect against fungi, bacteria, and predators (Mebs et al. 2010; Jeckel et al. 2015). However, the role of skin-derived chemicals in interactions between frogs and invertebrates has been rarely recorded, except for cases of mutualism and/or commensal relationships involving some species of microhylid frogs and theraphosid spiders (Cocroft and Hambler 1989; Dundee 1999; Siliwal and Ravichandran 2008; Karunarathna and Amarasinghe 2009; Dundee et al. 2012). Similarly, associations between frogs and ants are poorly reported in literature, being known only for anuran species Kassina fusca, K. senegalensis, and Phrynomantis microps living in ant nests in the African savanna (Déjean and Amiet 1992; Rödel and Braun 1999).

In the Americas, the only known association between ants and frogs occurs between frog Lithodytes lineatus and ants of genus Atta (Schlüter 1980; Schlüter and Regös 1981; Lamar and Wild 1995). L. lineatus not only uses the nests as shelter, but also uses the colony as a breeding site, building “nests within the nests” (Schlüter et al. 2009). In Africa, peptides present in the skin secretions of frog Phrynomantis microps are responsible for inhibiting the attack of ant Paltothyreus tarsatus (Rödel et al. 2013). In the Amazon, the mechanisms used by L. lineatus to live with Atta ants have not yet been studied. However, it is known that colonies of Atta use chemical substances for recognition and communication (Whitehouse and Jaffe 1995; Hernández et al. 2002, 2006). The pheromones produced and stored in glands located in different parts of the body of Atta ants is the main route of communication between the members of the colony and trigger specific behaviors, making it possible to classify them in the pheromones of alarm, territorial, marking, and nestmate recognition (Della Lucia 2011). Communication and recognition of conspecifics are made by means of chemical signals and other ant species that do not have chemical odors similar to those of colony members are attacked (Hernández et al. 2002, 2006). Conspecifics of Atta laevigata isolated from the colony, for a certain period, diminish the ability of recognition (Whitehouse and Jaffe 1995).

We hypothesized that frog L. lineatus produces chemicals upon skin that are recognized by Atta ants as a cue for preventing attacks. In order to test this hypothesis, we conducted two field bioassays. In the first treatment, individuals of L. lineatus and two genetically (Adenomera spp.) and phenotypically similar species of anurans (Ameerega picta and Allobates femoralis) were exposed to Atta ants. In the second treatment, Rhinella frogs coated with the substance extracted from the skin of L. lineatus and coated with ultrapure water (controls) were presented to Atta ants to evaluate whether the chemicals that prevent attacks are present in the skin of L. lineatus.

Natural history of organisms used in the present study

Lithodytes lineatus is a frog species widely distributed throughout the Amazon and Orinoco river basins of northern South America. It is known to be associated with leaf-cutting Atta ants. Males of L. lineatus call from ant nests and represent the only species of frog known to breed in the nests of ants (Schlüter 1980; Schlüter and Regös 1981; Lamar and Wild 1995; Schlüter et al. 2009). Like many leptodactylids, L. lineatus lays its eggs outside of the water in foam nests and tadpoles develop until undergoing metamorphosis in the anthill (Schlüter 1980; Schlüter and Regös 1981; Lamar and Wild 1995; Schlüter et al. 2009).

Lithodytes lineatus with orange or red femoral patches contrasting against a dark dorsal coloration has been considered by several authors to mimic poisonous and aposematic frogs of species Ameerega picta and the Allobates femoralis species complex (Schlüter and Regös 1981; Lamar and Wild 1995). Predation risk is reduced in Batesian and Müllerian mimicry because their body coloration resembles aposematic animals. Prates et al. (2012), investigated the morphology, distribution of poison glands, secreted skin compounds, and behavior between L. lineatus and A. picta. Their results suggest that L. lineatus is as toxic or unpalatable to potential predators as A. picta and apparently represents a case of Müllerian mimicry. Since the A. femoralis complex is considered non-toxic (Daly et al. 1987; Grant et al. 2006), Batesian mimicry is a plausible explanation, but no study has been conducted to substantiate such speculations.

Atta spp.

Leaf-cutting Atta ants are restricted to the Neotropics and feed mostly on a fungal symbiont cultivated within the nests, and therefore do not require external prey (Quinlan and Cherrelt, 1979; Della Lucia 2011). Among the eusocial insects, leaf-cutting Atta ants represent one of the most complex social organizations, mainly because of the great difference in morphology between individuals of each caste, which play distinct functions in the anthill (Della Lucia 2011). The A. laevigata ants, for example, recruit larger soldiers for defending the nest against vertebrates. On the other hand, for defense against conspecifics, a large number of smaller are recruited, demonstrating selectivity in the use of soldiers for defense (Whitehouse and Jaffe 1996).

Materials and methods

Animals

Field work was carried out in three areas: 1- Reserva Adolpho Ducke (Ducke/2° 58′ 9.883″ S, 59° 58′ 13.454″ W), Manaus City, Amazonas State; 2- The left bank of the Jaci Paraná River (Jaci Paraná/ 9° 24′ 41.602″ S, 64° 26′ 38.749″ W), Rondônia State; and 3- The left bank of the Madeira River (Madeira/9° 8′ 47.912″ S, 64° 30′ 34.293″ W), Rondônia State. Ten individuals of L. lineatus were collected: two individuals were collected in Ducke (size 33 and 42 mm), two in Jaci Paraná (size 31 and 39 mm) and six individuals in Madeira (size ranging from 21 to 26 mm). Five individuals were captured in nests of Atta laevigata, three individuals in nests of Atta sexdens and two individuals collected in pitfall traps in Ducke. For controls, Ameerega picta individuals (N = 6, size ranging from 21 to 23 mm), Allobates femoralis (N = 4; size ranging from 28 to 33 mm) and two species of genus Adenomera (N = 10; size ranging from 20 to 21 mm) were collected in the same areas as the L. lineatus.

Chemical recognition test using L. lineatus and four other anuran species

The frequency of attacks by ants upon L. lineatus was used as an index of the presence of aggression-inhibiting skin chemicals. Two species of Adenomera, a sister clade to that containing genus Lithodytes (De Sá et al. 2014), and A. femoralis and A. picta, which are phenotypically similar to L. lineatus, were used as controls. Species of Atta feed on a fungus grown inside the nests (Quinlan and Cherrelt 1979) and attack other organisms only to defend the colony. For this reason, potential predators are only attacked when trying to enter the colony, which led us to carry out the experiments at the entrances of the ant nests. Experiments were conducted in Atta sexdens colonies, located in Jaci Paraná (N = 2) and in Ducke (N = 2), and in three colonies of Atta laevigata located in Madeira. All experiments were conducted between 09:00 and 11:00 am for all species tested. The frogs were placed near the entrances of the nests of ants and placed under open-top clear glass containers to facilitate observation. The clear glass containers have 14 cm in height × 11 cm in diameter × 18 cm in width. The open-top clear glass container was positioned approximately 3 cm above the soil to allow the passage and contact of ants with the tested frogs. In order to avoid frogs from climbing or escaping from the clear glass containers during the experiments, all were tied with a cotton thread in the inguinal region. The experiments consisted of three treatments: (1) individuals of Adenomera spp. positioned near the entrances of the nests, (2) individuals of A. femoralis or A. picta positioned near the entrances of the nests, and (3) individuals of L. lineatus positioned near the entrances of the nests. In each experiment, the sequence of the anuran species was randomized to demonstrate that the order in which they were presented did not confound the results. In nests in which more than one experiment was carried out, a different entrance was used for each experiment. Each frog was used only once, and the experiment had a maximum duration of 10 min. After such time, the frogs were removed and the number of ants attached to their skin was counted, except in cases where frog’s bodies were covered in too many ants and they had to be immobilized, in which case the experiment was terminated to prevent those frogs from dying. The response time (attack) of ants was defined as the time of the beginning of the experiment until an ant bit and remained attached to a frog (see Electronic Supplementary material 1).

Obtaining L. lineatus skin extracts

Extracts were obtained from the skin of the frogs. Frogs were killed by overdose with an anesthetic (ointment with 2 % benzocaine) applied only within the mouth, so that the skin was not contaminated before it was removed. Skin extracts of males and females were obtained separately from immersed skins in 1.5 ml of methanol at room temperature. After 72 h, the methanol was evaporated using a vacuum concentration system at 30 °C. Subsequently, 10 mg of frog-skin extract was dissolved in 1.5 ml of ultrapure water. We used the gross composite without identifying the molecules present in the extract.

Extraction of the extract was made in the molecular toxinology laboratory of the Centro de Ofidismo “Prof. Dr. Paulo Bührnheim” of Fundação de Medicina Tropical – HVD, Manaus, Amazonas, Brazil.

Chemical recognition test using L. lineatus skin extracts

We used a methodology adapted from Rödel et al. (2013) to test if the chemical compounds present in the skin of L. lineatus were responsible for avoiding ant attacks. We used juvenile (size ranging from 20 to 23 mm) individuals of frog Rhinella major as experimental subjects. Rhinella major is an abundant frog in the region and its dry integument quickly absorbs liquids. All juveniles were immersed in ultrapure water or L. lineatus skin extracts.

The experiments had two treatments: (1) ten individuals of R. major were impregnated with 1.5 ml of ultrapure water and (2) ten individuals of R. major impregnated with 1.5 ml of L. lineatus skin extracts. The quantities of solutions used were sufficient to immerse the entire body of the animals. Five individuals were impregnated with skin extracts of males and five individuals were impregnated with skin extracts of females to test whether there was a gender effect upon ant attacks.

We used the skin extracts of individuals collected in the colonies of both Atta species (A. laevigata and A. sexdens). In each treatment, R. major individuals were placed near the entrances of Atta nests. Frogs were tethered by a cotton thread tied around their waists. Each frog was used only once. After 5 min, we counted the number of ants that attacked the frog’s skin. The response time (attack) of ants was defined from the time of the beginning of the experiment until an ant bit and remained attached to a frog (see Electronic Supplementary material 2).

In order to minimize observer bias, blinded methods were used during the recording (movies) and analysis of the behavioral data.

Statistical analyses

For analyzing differences in the response of ants (attack/non attack) in both experiments, we use a Kruskal-Wallis Test (nonparametric test). The analyses were performed using R ver. 3.3.0 statistical software (R Core Team 2016).

Results

Chemical recognition test using L. lineatus and four other anuran species

None of the ten individuals of L. lineatus placed in the Atta nest were attacked, whereas 10 individuals of genus Adenomera underwent on average 20.1 attacks (maximum = 50 and minimum = 9). Individuals of phenotypically similar species [Allobates femoralis (N = 4) and Ameerega picta (N = 6)] underwent on average 20.7 attacks (maximum = 52 and minimum = 12) (Fig. 1). There was a significantly lower attack rate (Kruskal-Wallis Test: H = 20.011, df = 2, P < 0.001) on L. lineatus than on the four other species tested.
Fig. 1

Response (attack) of the ants Atta sp., when exposed to different species of anurans. A corresponds to Lithodytes lineatus (N = 10), B corresponds to Ameerega picta (N = 6) and Allobates femoralis (N = 4), and C corresponds to species of the genus Adenomera (N = 10). The horizontal dark lines in the middle of the rectangles represent the median, the box quartiles, the vertical line range (maximum and minimum), and empty point outliers

The time between the ants’ arrival from the colony and the number of attacks varied among the frog species. There was a significant difference (Kruskal-Wallis Test: H = 13.028, df = 2, P = 0.001) between the latency it took for the ant to attack L. lineatus compared to the other four species tested. Individuals of A. picta were attacked on average in 0.36 min, after the first contact (maximum = 0.71, minimum = 0.13 min, N = 6), followed by the Adenomera species (mean = 1.19 min, maximum = 4.37, minimum = 0.05 min, N = 10) and Allobates femoralis (mean = 3.78 min, maximum = 8.34, minimum = 0.78 min, N = 4) (Fig. 2).
Fig. 2

Response time of the ants Atta sp., when exposed to different species of anurans. A corresponds to Lithodytes lineatus (N = 10), B corresponds to Ameerega picta (N = 6) and Allobates femoralis (N = 4), and C corresponds to species of the genus Adenomera (N = 10). The horizontal dark lines in the middle of the rectangles represent the median, the box quartiles, the vertical line range (maximum and minimum), and empty point outliers

Chemical recognition test using L. lineatus skin extracts

No Rhinella major individuals impregnated with L. lineatus skin extracts were attacked by ants (Fig. 3), but all individuals impregnated with ultrapure water only were attacked. On average, 14.3 ants (maximum = 46 and minimum = 5) bit the frogs treated with ultrapure water (Fig. 3). There was a significant difference (Kruskal-Wallis Test: H = 16.365, df = 1, P < 0.001) in ant response between treatments. The ant attacks occurred on average 0.35 min. After the first contact with R. major individuals impregnated in ultrapure water (maximum = 0.56 and minimum = 0.05 min) (Fig. 4), there was a significant difference (Kruskal-Wallis Test: H = 16.323, df = 1, P < 0.001) in the ants’ time of attack among frog species. Some individuals impregnated with L. lineatus skin extracts were bitten only when they jumped on the ants, but they were released shortly after the bite. The forebody of one R. major individual was impregnated with L. lineatus skin extracts and its hind legs were coated with ultrapure water. The ants walked over the region impregnated with L. lineatus skin extracts, but did not bite. However, the hind legs were bitten by ants at first contact. The ants did not attack R. major individuals that were impregnated with from either male or female L. lineatus skin extracts, demonstrating that both genders have compounds in their skin that inhibit ant attacks. No different behavior was observed in individuals impregnated with L. lineatus skin extracts or ultrapure water.
Fig. 3

Response (attack) of ants when exposed to Rhinella major individuals coated with skin extracts (N = 10) of frog, L. lineatus, and coated with ultrapure water (N = 10). The horizontal dark lines in the middle of the rectangles represent the median, the box quartiles, the vertical line range (maximum and minimum) and empty point outliers

Fig. 4

Response time of the ants when exposed to Rhinella major individuals coated with skin extracts (N = 10) of frog, L. lineatus, and coated with ultrapure water (N = 10). The horizontal dark lines in the middle of the rectangles represent the median, the box quartiles and the vertical line range (maximum and minimum)

Discussion

Coexistence of frogs and ants is unusual and the mechanisms that allow for this interaction are poorly understood (Rödel et al. 2013), but it has been suggested that the coexistence of frog Phrynomantis microps with ant Paltothyreus tarsatus is possibly due to two peptides produced by the frog.

We showed that species of frogs that do not usually associate with ants were aggressively attacked. Neither the phylogenetic proximity of Adenomera to L. lineatus, nor the phenotypic similarity of Allobates femoralis and Ameerega picta prevented the ants from attacking. All four anuran species tested, which are not associated with the ants, tried to escape by jumping and/or climbing the glass vessel wall after contact and attack of the ants. In contrast, no escape attempts were made by L. lineatus during the 10-min experiment, even when the frogs had contact with the ants. Substances on the L. lineatus skin prevented ant attacks on frog Rhinella major. Therefore, we suggest that frog L. lineatus has chemical compounds on their skin that can be recognized by the Atta laevigata and Atta sexdens soldiers. Analysis of the chemicals used by L. lineatus may shed light upon the chemical mechanisms used by Atta species to detect intruders, which are poorly understood (Hölldobler and Wilson 1986; Hernández et al. 20022006).

Our two experimental procedures clearly demonstrate that there is a chemical compound on L. lineatus skin that prevents the attack of ants of species A. laevigata and A. sexdens. We also showed that neither L. lineatus, nor R. major individuals coated with skin extracts from L. lineatus individuals were attacked by the A. sexdens soldiers. There are several reasons why it may be advantageous for frogs to live within leaf-cutting ant nests. Some authors have suggested that the anthill’s stable microclimate and higher humidity than the external environment, which are important features for the reproduction of frogs, may be favorable to frogs and their eggs (Schlüter and Regös 1981; Schlüter et al. 2009).

The Amazon has among the highest diversities of anuran reproductive strategies in the world. Due to this high diversity, Magnusson and Hero (1991) suggested that oviposition outside of the aquatic environment is a strategy used by several Amazonian anuran species to avoid predation. However, many terrestrial eggs suffer high predation from invertebrates, such as spiders (Luiz et al. 2013), flies (Vonesh 2000), beetles and wasps (Villa et al. 1982), as well as vertebrates, such as snakes (Lingnau and Di-Bernardo 2006) and turtles (Polo-Cavia et al. 2010). Leaf-cutting ants aggressively defend their nests against intruders, so we suggest that ants may help defend L. lineatus eggs and larvae, which builds its foam nests inside the ant nests.

In addition to the association with Atta ants, several aspects of the ecology and life history of L. lineatus are still very scarce. Our study provides evidence that frogs have some compounds that inhibit the defensive response of ants. This aspect opens discussions on the biotechnological applicability of the compounds present in the L. lineatus skin secretion, once the characterization and chemical structure of the substances have not yet been carried out. Similarly, the type of ecological relationship between L. lineatus and Atta ants is not well understood yet. As there are no known apparent benefits for the ants to have frogs as nestmates, this fact leads us to suspect that it is a commensal relationship; however, it is not known if L. lineatus feed on ants, which would make it a “parasite” in the nests. Here, we suggest some possible explanations for ants to tolerate frog presence, if it is a mutualistic relationship. Anthills are usually used by other invertebrates, thus, and as also suggested by Schlüter and Regös (1981), L. lineatus could be feeding on these individuals by controlling the number of co-inhabitants in the nest. Another possible explanation would be that L. lineatus would be feeding of the natural enemies of Atta ants, reducing competition between these organisms.

Conclusions

We have demonstrated that leaf-cutting ants of genus Atta do not attack frog Lithodytes lineatus, but aggressively attack other anuran species. Rhinella major individuals coated with L. lineatus skin extracts are not attacked when exposed to ants, but individuals of the same species coated with ultrapure water were attacked. Our results support the hypothesis that L. lineatus has chemical skin compounds that are recognized by ants of genus Atta, which may allow for coexistence between ants and frogs.

Acknowledgments

The Fundação de Amparo à Pesquisas do Estado do Amazonas (FAPEAM) provided funds through the Programa de Apoio a Núcleos de Excelência – PRONEX – 003/2009, process 653/2009. The Instituto Nacional de Pesquisas da Amazônia (INPA) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) provided a graduate scholarship to the first author. Santo Antônio Energia provided logistical support during field work in Rondônia. PhD. Fabricio Baccaro helped identify the ants. PhD. Adam Stow and PhD. William Magnusson made revisions and suggestions to the manuscript. Lucianne Cabral, Ingrid Guimarães, Diego Pires, Patrik Viana, Thais Dutra, Emerson Pontes, “Macuxi” and Taly Nayandra assisted in field work. PhD. Adrian Barnett helped with the English. The comments of two anonymous reviewers greatly improved this paper.

Compliance with ethical standards

Funding

This study was funded by the Fundação de Amparo a Pesquisas do Estado do Amazonas (FAPEAM) and provided funds through the Programa de Apoio a Núcleos de Excelência – PRONEX – 003/2009, process 653/2009 and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) provided a graduate scholarship to the first author.

Conflict of interest

The authors declare that they have no competing interests.

Ethical approval

The permissions that are required for the realization of experiments in this study were obtained through the Commission of Ethics in the Use of Animals (CEUA) of the Instituto Nacional de Pesquisas da Amazônia (INPA/CEUA, Protocol: 034/2014). All individuals of Adenomera spp., Allobates femoralis, Ameerega picta and Rhinella major used in the experiments were released after the experiment. They had no visible injuries and showed no signs of distress. These species naturally and frequently interact with Atta species in the wild.

Supplementary material

265_2016_2223_MOESM1_ESM.mp4 (19.6 mb)
ESM 1(MP4 20,037 kb)
265_2016_2223_MOESM2_ESM.mp4 (10.8 mb)
ESM 2(MP4 11,055 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2016

Authors and Affiliations

  • André de Lima Barros
    • 1
  • Jorge Luis López-Lozano
    • 2
  • Albertina Pimentel Lima
    • 1
  1. 1.Department of EcologyInstituto Nacional de Pesquisas da Amazônia–INPAManausBrazil
  2. 2.Core of poisonous animalsFundação de Medicina Tropical do Amazonas–FMT/AMManausBrazil