Insect fungiculture is practiced by ants, termites, beetles, and gall midges (Kaltenpoth 2009). The best-characterized examples are the attine ants, which are endemic to South and Central America and the southern USA. The ancestor of these ants evolved the ability to cultivate fungus as a food source around 50 million years ago, leading to the monophyletic tribe Attini, which number twelve genera with more than 230 species. The genera Acromyrmex and Atta (40 species) evolved 8–12 million years ago and form a branch of the higher attines, also known as leaf-cutter ants, which are characterized by large colonies of up to several million individuals (Schultz and Brady 2008). Like the other leaf-cutter ants, the well-studied species Acromyrmex octospinosus forms a mutualism with the fungus Leucoagaricus gongylophorus (Currie 2001). The fungal garden can be parasitized by a variety of fungi, which can destroy fungal gardens, leading to the collapse of the colony (Rodrigues et al. 2008; Reynolds and Currie 2004; Gerardo et al. 2004). The ants defend their fungal garden by a combination of grooming and weeding, in which contaminant spores are filtered out by the infrabuccal pocket, a specialized pouch in the mouth part of the ant, and the resulting infrabuccal pellets are deposited onto refuse piles (Little et al. 2006). Leaf-cutter ants also use antimicrobial metapleural gland secretions (Bot et al. 2002; Schildknecht and Koob 1971), and they host actinobacterial exosymbionts belonging to the genera Pseudonocardia, Streptomyces, and Amycolatopsis, that produce antifungal compounds which the ants use as ‘weedkillers’ (Sen et al. 2009; Oh et al. 2009; Haeder et al. 2009; Currie et al. 1999).

Another ant genus in the same subfamily, Myrmicinae, as the Attini, is Allomerus, which are specialist plant-symbionts (Heil and McKey 2003) that employ a unique mechanism of ambush hunting. Allomerus use a sooty mould of the order Chaetothyriales to build a galleried structure along the stems and branches of their host plants (Ruiz-Gonzalez et al. 2010; Dejean et al. 2005). They then hide in the galleries and await insect prey (Dejean et al. 2005). This specific interaction with a fungus in a non-attine ant is thought to be the result of fungiculture, but vertical transmission has not yet been demonstrated empirically. Fungiculture in ant-plant systems is predicted to be widespread (Defossez et al. 2009), but how Allomerus cultivate fungi is not well understood. Spores from as many as forty-four different fungal species can be detected within the fungal traps, but only mycelia from the one symbiotic sooty mould fungus have been observed, suggesting that the ants employ mechanisms to suppress the growth of alternative, and potentially parasitic, fungi (Ruiz-Gonzalez et al. 2010). Spores from non-cultivar fungi have also been found in the infrabuccal pellets of Allomerus workers, suggesting that, like the leaf-cutter ants, Allomerus are actively removing fungal contaminants through grooming and weeding (Ruiz-Gonzalez et al. 2010).

To determine if Allomerus utilize antifungal-producing actinobacteria to suppress the growth of fungal contaminants in their traps, we isolated bacteria from the cuticles of two Allomerus species, A. decemarticulatus, which live on the ant-plant Hirtella physophora and A. octoarticulatus, which live on Cordia nodosa, all in French Guiana. Ants were collected and stored in sterile solutions of 20% glycerol and phosphate-buffered saline for shipment to the UK, and then serial dilutions of the solutions were plated onto Lysogeny broth (LB-Lennox) agar. Actinobacterial colonies were selected by eye, purified by restreaking, further examined using light microscopy (Fig. 1) and identified by 16S rDNA sequencing as described previously (Barke et al. 2010). A total of seven actinobacterial strains belonging to known antibiotic-producing genera were isolated from workers of two Allomerus species taken from 19 trees, six strains belonged to the genus Streptomyces, and one strain belonged to Amycolatopsis (Table 1).

Fig. 1
figure 1

Light microscope images of the Streptomyces and Amycolatopsis strains isolated from Allomerus ant cuticles. See Table 1 for genus names and 16S rDNA accession numbers. White bar = 1 mm

Full size image
Table 1 16S rDNA identification of the actinobacteria isolated from Allomerus decemarticulatus (AD) and Allomerus octoarticulatus (AO)
Full size table

To test the hypothesis that actinobacterial symbionts of Allomerus ants produce antifungal compounds, we first performed bioassays using the human pathogen Candida albicans, which allows rapid testing of actinobacteria. Three Streptomyces strains (FG23, FG25 and FG26) and one Amycolatopsis strain (FG22) inhibited the growth of C. albicans in vitro (Table 1; Fig. 2). In order to investigate whether any of our isolates could inhibit fungi that they encounter in their natural habitat, we isolated fungi from traps collected from H. physophora trees and identified them by 18S rRNA gene sequencing (Table 2). The growth of two of these fungi, Annulohypoxylon spp. and Chaunopycnis spp., were suitable for agar plate bioassays. One of the Streptomyces isolates (FG26) and the Amycalotopsis isolate (FG22) inhibited the growth of the Annulohypoxylon fungus, and a second Streptomyces isolate (FG23) as well as Amycalotopsis FG22 inhibited the Chaunopycnis fungus (Fig. 2). Bioassays with the Chaetothyriales cultivar fungus and with other isolated trap fungi were not possible due to their patchy growth characteristics on agar plates. The data presented here confirm that actinobacteria are present on the cuticles of A. decemarticulatus ants, which can inhibit the growth of non-cultivar fungi present in the traps of their host plants.

Fig. 2
figure 2

Representative bioassays of the bioactive strains against Candida albicans and the nest parasites Annulohypoxylon spp. and Chaunopycnis spp. Red boxes indicate strains with antifungal activity

Full size image
Table 2 18S rDNA identification of the fungi isolated from Allomerus traps on their ant-plants. Genbank accession numbers are given for the 18S rDNA sequences
Full size table

Previous research has demonstrated that Allomerus grow a single strain of the sooty mould fungus from the order Chaetothyriales as a building material to bind together plant trichomes and form a trap for prey capture. This functions to protect the host plant from herbivores, to protect the ants, which move underneath the trap structure, and indirectly to provide the ants with food (Dejean et al. 2005; Ruiz-Gonzalez et al. 2010). Furthermore, the cultivar fungus has also been shown to mediate nutrient uptake by the plant (Leroy et al. 2011). In this study we set out to determine whether Allomerus ants also use mutualistic actinobacteria to produce antibiotic ‘weedkillers,’ as has been established previously for the attine ants (Barke et al. 2010; Currie 2001; Sen et al. 2009). Examples of insects using actinobacteria for protection against parasites are growing and are likely to be widespread in nature (Kaltenpoth 2009). However, this is the first study of non-attine, fungus-farming ants and we have provided evidence that these ants are associated with antifungal-producing actinobacteria. Interestingly, the actinobacterial strains isolated from Allomerus ants in this study have different spectra of activity against the human pathogen C. albicans, and the potential nest pathogens Annulohypoxylon spp. and Chaunopycnis spp., at least in our in vitro studies. We cannot rule out the possibility that these strains may have different bioactivities in the ant-plant microenvironment or that other isolates reported here produce antifungals under those conditions.

In summary, we identified actinobacteria belonging to the known antibiotic-producing genera Streptomyces and Amycolatopsis in seven out of 19 (37%) of the Allomerus ant-plants we studied and demonstrated that four of these actinobacteria have antifungal activity against non-symbiotic fungi under laboratory conditions. Although we were unable to culture actinobacteria from the remaining ant samples this does not prove their absence and is more likely due to loss of viability of bacteria during collection, storage and transport at room temperature. A wider survey and careful experimentation is required to determine if these actinobacteria are mutualists of Allomerus ants and what role their antifungal compounds play in the growth and maintenance of the Allomerus fungus. Nevertheless this is the first report of actinobacteria associated with non-attine fungus growing ants and offers preliminary evidence to support the hypothesis that such associations are widespread in systems of ant fungiculture.