Ascomycete fungi in the nests of ants inhabiting plants (= myrmecophytes) are very often cultivated by the ants in small patches and used as food source. Where these fungi come from is not known yet. Two scenarios of fungus recruitment are possible: (1) random infection through spores or hyphal fragments from the environment, or (2) transmission from mother to daughter colonies by the foundress queen. It is also not known at which stage of the colony life cycle fungiculture is initiated, and whether the- symbiont fungi serve as food for the ant queen. To clarify these questions, we investigated four Azteca ant species inhabiting three different Cecropia species (C. insignis, C. obtusifolia, and C. peltata). We analysed an rRNA gene fragment from 52 fungal patches produced by founding queens and compared them with those from established Azteca colonies (n = 54). The infrabuccal pockets of winged queens were dissected to investigate whether young queens carry fungi from their mother colony. Additionally, 15N labelling experiments were done to verify whether the queen feeds on the patches until she is nourished by her first worker offspring. We infer from the results that the fungi cultivated in hollow plant structures are transferred from the parental colony of the young queen. First, fungal genotypes/OTU diversity was not significantly different between foundress queen patches and established colonies, and second, hyphal parts were discovered in the infrabuccal pockets of female alates. We could show that fungiculture already starts before queens lay their eggs, and that the queens do not feed on fungal patch material but feed it to the larvae. Our findings suggest that fungiculture may be crucial for successful colony founding of arboreal ants in the tropics.
Citation: Mayer VE, Nepel M, Blatrix R, Oberhauser FB, Fiedler K, Schönenberger J, et al. (2018) Transmission of fungal partners to incipient Cecropia-tree ant colonies. PLoS ONE 13(2): e0192207. https://doi.org/10.1371/journal.pone.0192207
Editor: Renee M. Borges, Indian Institute of Science, INDIA
Received: July 11, 2017; Accepted: January 19, 2018; Published: February 21, 2018
Copyright: © 2018 Mayer et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The sequences are available at Genbank under the accession numbers MF066273-MF066356. Phylogenetic data, including alignments are available at TreeBASE, accession URL http://purl.org/phylo/treebase/phylows/study/TB2:S22034.
Funding: The study was supported by the Verein zur Förderung der Tropenstation La Gamba (to M.N.) and by the Faculty of Life Sciences.
Competing interests: The authors have declared that no competing interests exist.
Fungus farming in leaf cutter ants is famous; they grow basidiomycetes for food and manure the fungal gardens with various substrates. In other ant taxa, fungi have occasionally been reported to be present in the nests, but are generally regarded to be pathogens or merely commensals [1–7], or have simply remained overlooked as they form inconspicuously small dark patches inside the hollow stems, petioles, or leaf pouches (so-called “domatium”) of the host plants (myrmecophytes). Only in recent years has it become evident that these fungi are not accidental inhabitants of the ant nests and are actively cultivated and used by ants [8, 9], and represent a third symbiotic partner in many tropical ant-plant mutualisms [10, 11].
Most fungi cultivated inside the domatium belong to the order Chaetothyriales (Ascomycota), a group of”black yeasts” characterized by slow growing and melanized hyphae. Chaetothyriales are highly diverse with respect to their lifestyles. They may be endophytes in plant tissues , plant pathogens , epiphytes colonizing leaves of tropical trees , rock colonizers in extreme habitats , or colonizers in human-made habitats like washing machines, dish washers, or metro tunnels [16, 17]. Some are pathogens and are found to be infecting animal and human skin and central nervous system . Ants are the only animals known so far to live in a mutualistic relationship with chaetothyrialean fungi.
During the past couple of years, about 17 ant-plant associations that involve obligate chaetothyrialean fungi have been described [10, 11]. In these mutualistic associations, fungi are used for nest construction, sophisticated prey capture and defence mechanisms [19–22]. They are grown in small patches in the domatium and fed to the larvae [9, 23–25]. Some fungal strains isolated from domatia are ubiquitous and not specific to either the ant or the plant species, whereas other strains have been found to show some degree of ant-host specificity [23, 26].
Thus far, it is unknown how fungus-ant associations become established during the colony life cycle of obligate plant-ants. Until now, ant associated Chaetothyriales were described only from established colonies [5, 9–11, 23, 24]. When and how fungiculture is started, whether the queen carries the fungi along from her mother colony, transmitting the symbiont vertically; or whether inoculation occurs haphazardly when the young queen enters the domatium cavity or even later during patrolling of the worker ants (horizontal transmission), is so far not described from any ant-plant system with fungiculture.
To address these questions, we investigated fungiculture by Azteca queens founding new colonies in young Cecropia trees (Urticaceae). In the genus Cecropia, a group of neotropical pioneer trees, 46 of the 61 species are associated with ants . A recent multigene phylogeny inferred a single origin of the symbiotic relationship between Azteca ants and Cecropia plants, starting around 8 Mya ago . The Cecropia hosts provide hollow stem internodes for housing (domatia) and phyto-glycogen containing food bodies (Müllerian bodies)  for nutrition. The ants in return deter herbivores, prune their host trees from encroaching vegetation, and deposit extra nutrients within the hollow stem where they may be absorbed into the host tree's tissue [30–32]. In hollow stem internodes of Cecropia inhabited by Azteca colonies, we regularly found chaetothyrialean fungi in small, clearly delimited patches. Some of the fungus strains found were shared among different Azteca species, while others were ant-species specific . This pattern indicates two possible scenarios of fungus recruitment: (1) random infection through spores or hyphal fragments from the environment, or (2) transmission from mother to daughter colony by the foundress queen. If foundress queens carry along fungi from their mother colonies, the distribution and frequency of fungal strains in foundress queen colonized domatia should show the same pattern as observed in established colonies. If no such pattern is observable, the inoculation is suggested to originate from random infection.
In the present study, we observed the queens’ behaviour during colony foundation in the field, determined the fungal strains in patches of Azteca foundress queens with molecular methods, and compared the pattern of fungal strains with that of established Azteca colonies from the same sampling sites. We morphologically examined the infrabuccal pockets of alates before and after their nuptial flight for presence of fungi. Finally, we investigated the role of fungi as possible food source for the queen during the claustral colony founding (a stage in which queens do not forage but seal their nest during colony founding) using stable isotope analysis.
Material and methods
Study sites and species identification
Observations and sample collections were made in SW Costa Rica near the Tropical Research Station La Gamba (www.lagamba.at; N08°42’03”, W083°12’06”, 70 m asl) and near the Monteverde cloud forest between Guacimál (N10°12’57”, W084°50’46”) and Santa Elena (N10°19’12”, W084°49’30”) (345–1448 m asl). Patches of foundress queens and established colonies were sampled at the same sites. The Cecropia trees with colony founding Azteca queens were C. insignis Liebm., C. obtusifolia Bertol., and C. peltata L., between 0.5–2 m tall with a diameter of 1–3 cm, and grew along roadsides or in forest gaps. Ants (Azteca alfari Emery, A. coeruleipennis Emery, A. constructor Emery, A. xanthochroa Roger) were identified with the key provided by Longino .
Foundress queens’ behaviour
A foundress queen was defined as a non-physogastric dealate mated female observed in a recently colonised domatium, without any open entrance holes. Foundress queens either had no associated brood, eggs only, or brood at all stages; some were accompanied by a few dwarf workers.
We opened the stems of 64 young Cecropia plants (17 from Monteverde, 47 from La Gamba; in total 180 domatia with 212 living queens; see S2 Table), and documented the following parameters: the number of foundress queens and Azteca species per domatium (sometimes more than one species), the condition of the parenchyma on the inner domatia wall, the presence or absence of fungal patches, the presence or absence of eggs, larvae, pupae, and, in a few cases, newly hatched dwarf workers. Twice we were able to film a recently dealate gyne while biting open, entering through, and subsequently sealing the entrance hole to the domatium cavity with parenchyma scraped from the domatium wall (S1A–S1E Fig).
Fungal patch collection, DNA sequencing and analysis
Only fungal patches from domatia with a sealed entrance hole (see S1E Fig), which were big enough (diameter >3 mm) for sufficient yield of fungal DNA, were considered for molecular analysis (S1A and S1B Table). Whole patches were collected with sterile forceps, placed in plastic vials, closed with air-permeable cotton wool, and dried in a box with silica gel. In all but one case, we found only one patch per domatium, even though colony founding by multiple foundresses in one domatium (pleometrosis) was quite common (S2A Fig).
In total, 52 fungal samples from small piles made by the foundress queens (in the following named “foundress patches”) were sampled for sequencing. The majority (n = 45) were individual foundress patches with eighty-seven queens involved (12 Azteca alfari, 7 A. coeruleipennis, 33 A. constructor, 35 A. xanthochroa). In one domatium colonized by an A. coeruleipennis queen, two fungal patches occurred. These were separately collected and analyzed. Six samples were pooled from several rather small patches of A. alfari queens (S1A Table). Parenchyma from the inner wall of five uninhabited domatia was collected as control. In addition, 25 patch samples from established Azteca ant colonies defined by a distinct worker caste and re-opened domatia entrances were also analysed.
DNA extraction, PCR, and Sanger dideoxy sequencing was performed according to the protocols published earlier [10, 23]. For identification, we used the ITS1-5.8S-ITS2 (ITS) rRNA gene; details on primers, PCR and sequencing are described in . Mixed sequences were identified using a sequence comparison approach by comparing ambiguous alignment positions with the sequences of known genotypes as described earlier . All sequences are deposited in GenBank (http://www.ncbi.nlm.nih.gov/genbank/). Strains and GenBank accession numbers from this study are listed in S1A and S1B Table.
A matrix containing new sequences from the current study, representative sequences of Azteca associated operational taxonomic units (OTUs)  and selected sequences from other fungi representing the “domatia symbiont clade”  was aligned using Muscle version 3.8.31  and checked with BioEdit version 7.2.5. . The final data matrix for the analysis contained 468 nucleotide positions from 78 nucleotide sequences.
For phylogenetic reconstruction, a maximum likelihood analysis was done using MEGA7 , based on the Kimura 2-parameter (K2) model  with a discrete Gamma distribution (+G; 5 categories, parameter = 0.4209) and a proportion of evolutionarily invariable sites (+I; 41.37% sites). Initial trees for the heuristic search were obtained by applying the Neighbour-Joining method to a matrix of pairwise distances estimated using the Maximum Composite Likelihood (MCL) approach; heuristic search was done with extensive subtree-pruning-regrafting (SPR level 5) branch swapping.
Trees were rooted with Cladophialophora scillae [EU035412] and Cladophialophora hostae [EU035407] as outgroups.
Infrabuccal pocket content
From four female alates (two A. constructor, one A. alfari and A. coeruleipennis) infrabuccal pockets were investigated. After dissecting the heads, the infrabuccal pocket content was put on a glass slide and stained with 10 μL of calcofluor white M2R (1 g/L) (Sigma-Aldrich Co., USA) according to the manufacturers guidelines. Calcofluor white is a non-specific fluorochrome that binds to cellulose and chitin in cell walls. The samples were investigated with an epifluorescence microscope (Zeiss Axio Imager.M1) using UV excitation of 365 nm.
15N labelling of foundress patches
A small “window” was cut into domatia with sealed entrance holes, 4 μl of a 98 at% 15N amino acid mixture (Isotec Sigma-Aldrich, USA) were pipetted with a Hamilton syringe in four 1 μl droplets directly into the patch pile. After each droplet we waited for two minutes to be sure that the liquid was completely imbibed from the patch substrate and not scattered over the surface. Thereafter, the window was sealed again using duct tape. After 7 days of incubation, patches (n = 5), queens (n = 5), brood (larvae: n = 4, pupae: n = 5) and workers (n = 3) from domatia with queens still alive and free of mould were collected and dried. Queen (and–if present also the workers) were analysed only after having removed their legs to be sure that possible contamination on the tarsal did not adulterate the result. The larvae and pupae are not moving and the measured 15N must result from the consumption of fungal patch particles. The dried samples were weighed into tin capsules for isotope ratio mass spectrometric analysis (IRMS) in a continuous-flow IRMS system that consisted of an elemental analyser (EA 1110, CE Instruments, Milan, Italy) connected to an IRMS (DeltaPLUS, Finnigan MAT, Bremen, Germany) by a ConFlo II interface (for details see ). Samples of non-incubated patches (n = 13), queens (n = 5), larvae (n = 7), pupae (n = 7) and workers (n = 5) served as natural abundance controls.
With contingency tables, we tested (a) whether the distribution pattern of fungal OTUs is the same between foundress queen and established colony patches, and (b) whether significantly more OTUs occur if more than one queen is present in the same domatium. Due to the small sample numbers, Fisher’s exact test was used. In one exception, where all samples were pooled, a Monte Carlo Chi2 was applied .
Behaviour of foundress queens
Based on the observation of 180 domatia inhabited by 212 queens from 64 young Cecropia plants, we found internodes 2–6 (counted from the apex) preferably colonised. Queen number per plant ranged between 1 and 16 (data not shown) and was much higher in Monteverde with on average 6.7 (ranging from 1–16) foundress queens per plant compared to 1.9 (range 1–4) in La Gamba (S2 Table).
During initial colonization, queens not only started to immediately scrape the spongy white parenchyma from the inner domatia walls to seal the entrance hole (S1C–S1E Fig), but also amassed parenchyma into a small pile (the “foundress patch”) of 2–10 mm in diameter (Fig 1A and 1B). Even in very young plants with little developed parenchyma, pile-making was a priority task of the queens (S2B Fig). Eggs and larvae were only observed in domatia with such a pile of parenchyma tissue and they were usually deposited next to it (Fig 1B).
(a) Parenchyma (white tissue) of the inner domatium wall. The part from where it has been scraped off is marked with a circle; (b) An Azteca xanthochroa queen with a parenchyma pile inoculated with chaetothyrialean fungi (foundress patch). Eggs and larvae are deposited next to the fungal patch (circle); (c) Detail of an Azteca constructor foundress patch with hyphae and (d) conidiophores (arrowheads). Scale bars: (a) 2cm, (b) 2mm.
Fungal OTUs in foundress patches compared to established colonies
From parenchyma samples of uninhabited internodes we could not amplify any chaetothyrialean DNA. However, chaetothyrialean DNA was invariably found in foundress patches (Fig 1C and 1D). From 52 samples we got 64 different fungal sequences that could be assigned to 14 different genotypes and five OTUs (Fig 2; S1 Table). Nine samples contained more than one genotype. OTU1 was the most common one in foundress patches, followed by OTU3 and OTU2. OTU1 was not found in any of the A. alfari patches; instead OTU2 occurred in at least one foundress patch of each ant species (Table 1). With the extended sequence data, OTU6 from an earlier study  appeared within the OTU2 cluster and is no longer supported as a separate clade.
A maximum likelihood analysis of the ITS matrix was performed with MEGA7 , showing the best tree (-lnL = 1906.92) based on the Kimura 2-parameter (K2) model , applying a discrete Gamma distribution (+G; 5 categories, parameter = 0.4209) and a proportion of evolutionarily invariable sites (+I; 41.37% sites). The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. Bold branches indicate ML bootstrap support above 50%.
The number represents how often the genotype was found in the respective ant species (in %) relative to all samples of the respective ant species. In nine patches, more than one genotype was detected (n = 9 for queen and established patches). In eleven domatia, more than one founding queen occurred (up to 10). Genotype frequency is shown for each Azteca species separately. In some cases, the percentage exceeds 100% due to patches with more than one fungal OTU.
OTU distribution was not significantly different between foundress queens and established colonies for patches from domatia with single queens (Fisherʼs p = 0.666). However, if in addition to single queen domatia also those containing multiple queens were considered, a significant difference was found (Monte Carlo Chi2 p = 0.0153).
At the level of individual ant species, the OTU distribution was not significantly different between foundress queens and established colonies. In both cases A. alfari and A. coeruleipennis cultivated more often OTU2, A. xanthochroa OTU1, and A. constructor OTU1 and 3 (Table 1). OTU5 occurred only in A. coeruleipennis (14% of all patches).
Number and diversity of foundress queens and the range of fungal OTUs
At the founding stage, single-OTU patches (43) dominated clearly over multi-OTU ones (9), also in established colonies (43 single-OTU, 12 multi-OTU) (Table 2; Fig 3). Interestingly, five multi-OTU patches were from domatia with only one queen, and four from domatia with more than one queen. Twelve patches had only one fungal OTU although they were colonised with more than one queen—six of them even with queens from different Azteca species (Fig 3; Table 2). In the majority of cases, only one foundress queen was observed, and multiple queens occurred in 21.7% of the domatia in Monteverde and in 3% in La Gamba. They were either from the same species (up to 10 A. xanthochroa females in one domatium), or from different Azteca species (e.g., A. xanthochroa and A. constructor, A. coeruleipennis and A. constructor) (S2 Table).
In domatia with a single queen patches with only one OTU is common, whereas in multiqueen domatia with several queens of the same species as well as of different species multi-OTU patches are equal to those with only one OTU.
For foundress queen samples, it is indicated whether we found (i) a single queen (1 Q), (ii) more than one of the same Azteca species (>1 Q same sp.) or (ii) more queens of different Azteca species (>1 Q diff spp.) in one domatium.
A higher number of OTUs in the fungal patches was not related to the presence of more than one queen per domatium (Fisher's p = 0.618); this result remained robust in conspecific as well as mixed species groups for all four Azteca species. At the ant species level multi-OTU patches were significantly more often found in multi-queen domatia with A. xanthochroa involved (Fisher's p = 0.0458). For A. constructor (Fisher's p = 0.237) and A. coeruleipennis (Fisher's p = 0.4286) the relationship was not significant. A. alfari had single OTUs with only one exception. In contrast, most A. coeruleipennis patches with one queen had more than two fungal OTUs (Table 2).
Infrabuccal pocket content
In their infrabuccal pockets, alates carried hyphal fragments that had the same morphology and size as the fungi in the foundress patches. Additionally, spores, nematodes, nematode eggs, and many other unidentified particles were found (Fig 4).
Founding queens (a) carry in their infrabuccal pockets (b) hyphae, (c) a number of unclassified particles, and (d) nematode dauerlarvae. Infrabuccal pocket content was stained with calcofluor white and investigated with an excitation of 365 nm (b, c) and brightfield (d). Scale bars: (b, c) 10μm, (d) 20μm.
15N labelling of foundress patches
The incubation of foundress patches with 15N amino acids led to a significant accumulation of 15N in larvae (Mann-Whitney-U-Test: U < 0.001, p = 0.004). Ant pupae and workers only showed a trend towards elevated 15N ratios, whereas queens did not accumulate any 15N (Fig 5, Table 3).
δ15N values of founding queens, larvae and pupae after incubating the patches of the respective founding queen for 7 days with a 15N labelled amino acid mix. 15N accumulation could be shown for pupae and larvae, but only larvae were significantly enriched. The δ15N values of foundress queens from labelled patches did not differ from natural abundance. Hyphae were thus not eaten by the queens themselves but fed to the larvae.
Chaetothyrialean fungi as third party in numerous ant-plant mutualisms received increasing attention in the last years [5, 10, 11, 23, 24]. It was found that they are important for the ants’ nutrition and the system’s nutrient recycling [8, 25]. The present study is the first to focus on the establishment of the fungiculture at the beginning of a new colony cycle in an obligate ant-plant mutualism.
The role of the domatium parenchyma
After entering the host-plant provided nesting space, Azteca foundress queens scrape parenchyma from the inner nest wall, seal the entrance and produce small piles. We could show that the parenchyma itself is free of chaetothyrialean fungi, only after contact with the ant queens, the parenchyma piles contain fungal hyphae. Parenchyma scraping and sealing of entrance holes with masticated plant tissue is known in other claustrally founding plant-ant queens [3, 40–42]. Whether parenchyma is also accumulated into a pile has, however, not been described in these earlier accounts. But the presence of chaetothyrialean fungi in these other systems was discovered recently [9–11] and the onset of fungiculture is likely similar among these different tropical fungus-cultivating plant-ants.
As we did not find any hyphae in plant tissue of uninhabited domatia, we exclude an endophytic origin of the fungi in the parenchyma pile. Instead, vertical fungus transmission by the foundress queens seems plausible for the following reasons:
First, the broad overlap among genotypes and OTUs from established Azteca colonies (100% on OTU level, 82% on genotype level). Only two genotypes had not been found earlier in established colonies and three genotypes from established colonies  were not found in foundress patches (S1A Table). This is probably because we did not sample the respective mother colony or the respective reproductive ants.
Second, the overall diversity of chaetothyrialean genotypes/OTUs in the environment of the host plants may be at least ten times higher as shown by a study about chaetothyrialean fungi in carton galleries of Azteca brevis on the branches of various host trees . This earlier study took place in one of the collection sites of the present study and resulted in 128 genotypes and 62 OTUs of black yeasts compared to only 14 genotypes from five OTUs in foundress queen patches. Due to this low OTU number in domatia, de novo recruitment of the fungi is highly unlikely.
Third, the chaetothyrialean lineage containing the domatia symbionts has so far been exclusively detected in domatia and is yet unknown from any other substrate, indicating a close association with their ant symbionts. Notably, there is no overlap with the carton galleries of Azteca mentioned above, which contains a high species biodiversity of various chaetothyrialean lineages but not from the domatia symbiont clade.
Fourth, the presence of hyphal parts and fungal spores in infrabuccal pockets of alate queens points to an inoculation of the parenchyma pile with fungal pellet material brought from the fungiculture of their mother colonies. Unfortunately, we were not able to generate DNA sequences from the infrabuccal pocket content making it uncertain whether the hyphal parts and spores found are truly those of the later domatia fungal community. A vertical transmission mode is also known from Lasius fuliginosus cultivating fungi on their nest walls , or from leaf cutter ants [45, 46] and ensures the occurrence of the right partners in the next generation of the mutualism. For chaetothyrialean fungi with their low competitive ability  ant nests may be an important ecological niche.
In addition to fungal parts and spores, dauer stages of nematodes were also carried by the foundress queens (Fig 4D). Sclerorhabditis neotropicalis (Rhabditida) is frequently found in Cecropia colonizing Azteca nests  but it was not known until now that Sclerorhabditis neotropicalis is vertically transmitted and already present during colony founding.
Number and diversity of foundress queens and the range of fungal OTUs
Most of the foundress queen domatia showed haplometrotic colony founding with only one gyne present (76.6%). 86% of those single foundresses had parenchyma piles with one fungal OTU, 14% with more than one. In contrast, pleiometrotic colony founding was comparably rare, with multiple gynes of the same species in 12.8% and gynes from different ant species in 10.6% of the cases. Usually, in one Cecropia stem several independent founding events occur in the spatially segregated domatia. Multi-queen colonization of saplings and the survival of only one queen seems to be a common feature in ant-plant mutualisms and was described from several other ant-plant systems (e.g., Crematogaster-Macaranga ; Tetraponera-Barteria ; Ocotea-Myrmelachista ; Triplaris-Pseudomyrmex . Also in Cecropia only one of the incipient colonies becomes dominant occupying all other domatia and with a single queen only . This may lead to the acquisition of the fungal patches from the less successful queens by the ant colony that wins the race and may explain why multi-OTU fungal patches occurred in 22% of the established Cecropia inhabiting Azteca colonies (Fig 3; Table 2). However, this leads to another question: why do 78% of the established colonies have only single-OTU fungal patches although multi-queen colonization of saplings is common with different OTUs in each foundress queen patch?
Two scenarios may be possible. (1) The gyne with the most vigorously growing fungal strain may be the first with enough workers to take over the stem and outcompete other colony foundings. The most vital fungus may overgrow their strains and become dominant, replacing the original fungi. (2) Directed symbiont selection could have happened with selection of the most vigorously growing strain. In A. alfari, for example, OTU2 and 3 were found in foundress queen patches, but in 96% of the established colonies only OTU2 was cultivated. The other OTU became under-represented and patches with multiple OTUs were rare with this ant species. Azteca ants probably groom their fungus patches as described for fungus-cultivating Petalomyrmex phylax living in domatia of Leonardoxa africana , or like leaf cutter ants do in their fungus gardens to control growth and microbial infections [52, 53]. Ants may not only be able to distinguish between pathogenic fungi and fungal symbionts but may also recognize the most beneficial fungal strain. A good example for such screening is the rejection of leaves harmful for fungal growth in leaf-cutter ant colonies. The ants perceive the state of the fungus through olfactory cues released from the fungus itself [54, 55].
Chaetothyrialean fungi as food for the claustral queen?
Colony founding is a critical phase with high mortality rates: most ant queens with claustral colony founding deplete their flight muscles and fat deposits to maintain the metabolic needs and nurture their larvae, and most of them die from starvation [56–58] before the first workers emerge [49, 59, 60]. Therefore, various authors suggested that in Cecropia the foundress queen feeds on a nutrient rich callus tissue growing from the entrance hole or on the parenchyma of the inner domatia walls [61–63]. We did, however, not find a nutrient rich callus tissue in any of the domatia with young queens. Nor seemed the latex of the tissue plug used to seal the entrance hole to be suitable for queen nourishment and the fungal patch material was exclusively fed to the larvae. The availability of a fungal patch as food source may be a crucial factor for the development of the first workers–and indirectly for the survival of the queen as after depletion of her body’s own resources the queen is nurtured by the workers. The frequent occurrence of dead or moribund queens in very young Cecropia trees  may be due to the lack of a well-developed parenchyma layer, and abortive fungiculture. As a consequence, the larvae do not develop, no workers emerge, and the queen will die.
We infer from our data, that in Azteca/Cecropia system the transmission of the fungal symbiont for fungiculture is vertical. Compared to the chaetothyriales found in the environment, the genetic variability of the fungi from fungal patches is limited indicating a selection of strains beneficial to the ant partner. That egg laying only occurs after establishing a fungal inoculum and the fact that larvae are nourished with patch material points to a fundamental importance of fungus cultivation for successful ant colony founding in this ant-plant system.
Agriculture has for a long time been regarded a cultural achievement of humans. In truth this uniqueness does not exist. Fungus farming is found in ants, beetles, termites and snails, [65, 66] many of them arose long before humans started farming. Fish and sloth have algal farms [67, 68], crabs and mussels bacteria gardens [69, 70].
Future work investigating mechanisms of fungus selection by the ants and the functional role of the ubiquitous fungal patches in ant-plant interactions in the tropics will help to increase our understanding of the richness both in species diversity and number of individuals of canopy ants in an in fact nutrient poor environment.
DNA sequences: Genbank accession numbers are MF066273-MF066356. Phylogenetic data, including alignments are available at TreeBASE, accession URL http://purl.org/phylo/treebase/phylows/study/TB2:S22034.
S1 Table. Foundress patches (n = 52) and patches of established colonies (n = 25) analyzed in this study, GenBank accession numbers, Azteca species, collection site (Monteverde, La Gamba), and genotype and OTU assignment.
S2 Table. Distribution of the foundress queens in the trees investigated.
S1 Fig. Colonization of a young Cecropia peltata stem.
J.T. Longino (University of Utah, USA) is cordially thanked for providing two alate Azteca constructor females from Malaise traps. W. Sudhaus (FU Berlin, Germany) kindly indentified the nematode in Fig 4D as a dauer stage. G. Chomicki and an anonymous reviewer helped to improve the manuscript by their great reviews. N. McLaughlin-Walter (University of West Georgia, USA) is cordially thanked for the language revision. We thank the staff at the Estación Tropical La Gamba and the Biological Station Monteverde (both Costa Rica) for excellent working conditions and help. Javier Guevara (Sistema Nacional de Áreas de Conservación—SINAC) and Melania Muñoz García (Comisión Nacional para la Gestión de la Biodiversidad—CONAGEBIO) gratefully helped with the permits. All collections and exportations were under permission from SINAC and CONAGEBIO (No. 051-2012-SINAC, 142-2013-SINAC, R-046-2015-OT-CONAGEBIO).
- 1. Wheeler WM, Bailey IW. The feeding habits of the pseudomyrmine and other ants. Trans Am Phil Soc. 1920;22:235–79.
- 2. Wheeler WM. Studies of Neotropical ant-plants and their ants. Bull Mus Comp Zool. 1942;90:1–262.
- 3. Schremmer F. Untersuchungen und Beobachtungen zur Ökoethologie der Pflanzenameise Pseudomyrmex triplarinus, welche die Ameisenbäume der Gattung Triplaris bewohnt. Zoolo Jahrb Abt Syst, Geogr Biol Tiere. 1984;3:385–410.
- 4. Pearson HHW. On some Species of Dischidia with double pitchers. Bot J Linn Soc. 1902;35(245):375–90.
- 5. Moog J. The associations of the pant ant Cladomyrma with plants in Southeast Asia. Frankfurt, Germany: Johann Wolfgang Goethe University; 2009.
- 6. Bailey IW. Notes on neotropical ant-plants III. Cordia nodosa lam. Bot Gaz. 1924;77(1–4):32–49. PubMed PMID: WOS:000202864700002.
- 7. Janzen DH. Protection of Barteria (Passifloraceae) by Pachysima ants (Pseudomyrmicinae) in a Nigerian rainforest. Ecology. 1972;53:886–92.
- 8. Defossez E, Djieto-Lordon C, McKey D, Selosse MA, Blatrix R. Plant-ants feed their host plant, but above all a fungal symbiont to recycle nitrogen. Proc R Soc Lond B Biol Sci. 2011;278(1710):1419–26. PubMed PMID: ISI:000288851800019. pmid:20980297
- 9. Defossez E, Selosse MA, Dubois MP, Mondolot L, Faccio A, Djieto-Lordon C, et al. Ant-plants and fungi: a new threeway symbiosis. New Phytol. 2009;182(4):942–9. PubMed PMID: ISI:000265936500017. pmid:19383109
- 10. Voglmayr H, Mayer V, Maschwitz U, Moog J, Djieto-Lordon C, Blatrix R. The diversity of ant-associated black yeasts: insights into a newly discovered world of symbiotic interactions. Fungal Biol. 2011;115(10):1077–91. PubMed PMID: WOS:000296076300016. pmid:21944219
- 11. Vasse M, Voglmayr H, Mayer VE, Gueidan C, Nepel M, Moreno L, et al. A phylogenetic perspective on the association between ants (Hymenoptera: Formicidae) and black yeasts (Ascomycota: Chaetothyriales). Proc R Soc Lond B Biol Sci. 2017;284(1850). pmid:28298348
- 12. Cook D, Beaulieu WT, Mott IW, Riet-Correa F, Gardner DR, Grum D, et al. Production of the alkaloid Swainsonine by a fungal endosymbiont of the Ascomycete order Chaetothyriales in the host Ipomoea carnea. J Agric Food Chem. 2013;61(16):3797–803. pmid:23547913
- 13. Badali H, Gueidan C, Najafzadeh MJ, Bonifaz A, van den Ende AHGG, de Hoog GS. Biodiversity of the genus Cladophialophora. Stud Mycol. 2008;61:175–91. PubMed PMID: PMC2610306. pmid:19287540
- 14. Chomnunti P, Bhat DJ, Jones EBG, Chukeatirote E, Bahkali AH, Hyde KD. Trichomeriaceae, a new sooty mould family of Chaetothyriales. Fungal Divers. 2012;56(1):63–76.
- 15. Selbmann L, Zucconi L, Isola D, Onofri S. Rock black fungi: excellence in the extremes, from the Antarctic to space. Curr Genet. 2015;61(3):335–45. pmid:25381156
- 16. Réblová M, Hubka V, Thureborn O, Lundberg J, Sallstedt T, Wedin M, et al. From the tunnels into the treetops: New lineages of black yeasts from biofilm in the Stockholm Metro System and their relatives among ant-associated fungi in the Chaetothyriales. PLoS ONE. 2016;11(10):e0163396. pmid:27732675
- 17. Zupančič J, Novak Babi M, Zalar P, Gunde-Cimerman N. The Black Yeast Exophiala dermatitidis and other selected opportunistic human fungal pathogens spread from dishwashers to kitchens. PLoS ONE. 2016;11(2):e0148166. pmid:26867131
- 18. De Hoog GS, Queiroz-Telles F, Haase G, Fernandez-Zeppenfeldt G, Angelis DA, Van den Ende AHGG, et al. Black fungi: clinical and pathogenic approaches. Med Mycol. 2000;38(Supplement 1):243–50.
- 19. Mayer VE, Voglmayr H. Mycelial carton galleries of Azteca brevis (Formicidae) as a multi-species network. Proc R Soc Lond B Biol Sci. 2009;276(1671):3265–73. PubMed PMID: ISI:000268798600007.
- 20. Ruiz-Gonzalez MX, Male PJG, Leroy C, Dejean A, Gryta H, Jargeat P, et al. Specific, non-nutritional association between an ascomycete fungus and Allomerus plant-ants. Biol Lett. 2011;7(3):475–9. PubMed PMID: ISI:000290515100044. pmid:21084334
- 21. Dejean A, Solano PJ, Ayroles J, Corbara B, Orivel J. Insect behaviour: arboreal ants build traps to capture prey. Nature. 2005;434(7036):973. pmid:15846335
- 22. Mayer VE, Lauth J, Orivel J. Convergent structure and function of mycelial galleries in two unrelated Neotropical plant-ants. Insectes Soc. 2017:1–7. pmid:28757658
- 23. Nepel M, Voglmayr H, Blatrix R, Longino JT, Fiedler K, Schönenberger J, et al. Ant-cultivated Chaetothyriales in hollow stems of myrmecophytic Cecropia sp. trees–diversity and patterns. Fungal Ecol. 2016;23:131–40. http://dx.doi.org/10.1016/j.funeco.2016.07.007.
- 24. Kokolo B, Atteke C, Ibrahim B, Blatrix R. Pattern of specificity in the tripartite symbiosis between Barteria plants, ants and Chaetothyriales fungi. Symbiosis. 2016;69(3):169–74.
- 25. Blatrix R, Djiéto-Lordon C, Mondolot L, La Fisca P, Voglmayr H, McKey D. Plant-ants use symbiotic fungi as a food source: new insight into the nutritional ecology of ant–plant interactions. Proc R Soc Lond B Biol Sci. 2012;279 3940–7. pmid:22859596
- 26. Blatrix R, Debaud S, Salas-Lopez A, Born C, Benoit L, McKey DB, et al. Repeated evolution of fungal cultivar specificity in independently evolved ant-plant-fungus symbioses. PLoS ONE. 2013;8(7). doi: e6810110.1371/journal.pone.0068101. PubMed PMID: WOS:000322433300004.
- 27. Berg CC, Franco-Rosselli P. Cecropia: New York Botanical Garden; 2005.
- 28. Gutiérrez-Valencia J, Chomicki G, Renner SS. Recurrent breakdowns of mutualisms with ants in the neotropical ant-plant genus Cecropia (Urticaceae). Mol Phylogenet Evol. 2017;111(Supplement C):196–205. https://doi.org/10.1016/j.ympev.2017.04.009.
- 29. Bischof S, Umhang M, Eicke S, Streb S, Qi W, Zeeman SC. Cecropia peltata accumulates starch or soluble glycogen by differentially regulating starch biosynthetic genes. Plant Cell Online. 2013;25(4):1400–15. pmid:23632447
- 30. Janzen DH. Allelopathy by myrmecophytes: The ant Azteca is an allelopathic agent of Cecropia. Ecology. 1969;50:147–53.
- 31. Schupp EW. Azteca protection of Cecropia: ant occupation benefits juveniles trees. Oecologia. 1986;(70):379–85. pmid:28311924
- 32. Sagers CL, Ginger SM, Evans RD. Carbon and nitrogen isotopes trace nutrient exchange in an ant-plant mutualism. Oecologia. 2000;123(4):582–6. PubMed PMID: WOS:000088154300017. pmid:28308767
- 33. Longino JT. Ants of Costa Rica 2007. Available from: http://academic.evergreen.edu/projects/ants/AntsofCostaRica.html.
- 34. Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic acids res. 2004;32(5):1792–7. pmid:15034147
- 35. Hall T. BioEdit, version 7.2.5. Ibis Biosciences, Carlsbad, CA, USA. 2013.
- 36. Kumar S, Stecher G, Tamura K. MEGA7: Molecular Evolutionary Genetics Analysis version 7.0 for bigger datasets. Mol Biol Evol. 2016;33(7):1870–4. pmid:27004904
- 37. Kimura M. A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. J Mol Evol. 1980;16(2):111–20. pmid:7463489
- 38. Gegenbauer C, Mayer VE, Zotz G, Richter A. Uptake of ant-derived nitrogen in the myrmecophytic orchid Caularthron bilamellatum. Ann Bot. 2012;110(4):757–65. PubMed PMID: WOS:000308016100001. pmid:22778148
- 39. Hammer Ø. PAST—PAleontological STatistics. Version 3.12. Natural History Museum, University of Oslo. Available from: http://folk.uio.no/ohammer/past. Last accessed: 02 Feb 2017. [cited 2017].
- 40. Fiala B, Maschwitz U. Studies on the South East-Asian ant-plant association Crematogaster-borneensis Macaranga—adaptations of the ant partner. Insectes Soc. 1990;37(3):212–31.
- 41. Moog J, Saw LG, Hashim R, Maschwitz U. The triple alliance: how a plant-ant, living in an ant-plant, acquires the third partner, a scale insect. Insectes Soc. 2005;52(2):169–76.
- 42. Yumoto T, Maruhashi T. Pruning behavior and intercolony competition of Tetraponera (Pachysima) aethiops (Pseudomyrmecinae, Hymenoptera) in Barteria fistulosa in a tropical forest, Democratic Republic of Congo. Ecol Res. 1999;14(4):393–404.
- 43. Nepel M, Voglmayr H, Schönenberger J, Mayer VE. High diversity and low specificity of Chaetothyrialean fungi in carton galleries in a Neotropical ant–plant association. PLoS ONE. 2014;9(11):e112756. pmid:25398091
- 44. Schlick-Steiner BC, Steiner FM, Konrad H, Seifert B, Christian E, Moder K, et al. Specificity and transmission mosaic of ant nest-wall fungi. Proc Natl Acad Sci U S A. 2008;105(3):940–3. PubMed PMID: WOS:000252647900022. pmid:18195358
- 45. Quinlan RJ, Cherrett JM. Studies on the role of the infrabuccal pocket of the leaf-cutting ant Acromyrmex octospinosus (Reich) (Hym., Formicidae). Insectes Soc. 1978;25(3):237–45.
- 46. Hv Ihering. Die Anlagen neuer Colonien und Pilzgärten bei Atta sexdens. Zool Anz. 1898;21:238–45.
- 47. Esquivel A, Abolafia J, Hanson P, Pinto A. A new species of nematode, Sclerorhabditis neotropicalis SPN (Rhabditida), asociated with Azteca ants in Cecropia obtusifolia. Nematropica. 2012;42(1):163–9. PubMed PMID: WOS:000307992500024.
- 48. Kuhn KM. Colony founding by the ant Myrmelachista flavocotea. Insectes Soc. 2014;61(3):239–45.
- 49. Sanchez A. Establishing an ant-plant mutualism: foundress queen mortality and acquiring the third partner. Insectes Soc. 2016;63(1):155–62.
- 50. Longino JT. Geographic variation and community structure in an ant-plant mutualism: Azteca and Cecropia in Costa Rica. Biotropica. 1989;21:126–32.
- 51. Longino JT. Azteca ants in Cecropia trees: taxonomy, colony structure, and behaviour. In: Cutler D, Huxley C, editors. Ant-plant interactions. Oxford: Oxford University Press; 1991. pp. 271–88.
- 52. Currie CR, Stuart AE. Weeding and grooming of pathogens in agriculture by ants. Proc R Soc Lond B Biol Sci. 2001;268(1471):1033–9. pmid:11375087
- 53. Fernández-Marín H, Zimmerman JK, Rehner SA, Wcislo WT. Active use of the metapleural glands by ants in controlling fungal infection. Proc R Soc Lond B Biol Sci. 2006;273(1594):1689–95. pmid:16769642
- 54. Morath SU, Hung R, Bennett JW. Fungal volatile organic compounds: A review with emphasis on their biotechnological potential. Fungal Biol Rev. 2012;26(2–3):73–83. http://dx.doi.org/10.1016/j.fbr.2012.07.001.
- 55. Arenas A, Roces F. Learning through the waste: olfactory cues from the colony refuse influence plant preferences in foraging leaf-cutting ants. J Exp Biol. 2016. pmid:27284068
- 56. Keller LP, Luc. Size and fat content of gynes in relation to the mode of colony founding in ants (Hymenoptera; Formicidae). Oecologia. 1989;80(2):236–40. pmid:28313113
- 57. Hölldobler BWEO. The ants. Cambridge, Mass.: Belknap University Press; 1990.
- 58. Wheeler DE, Buck NA. Depletion of reserves in ant queens during claustral colony founding. Insectes Soc. 1996;43(3):297–302.
- 59. Marti HE, Carlson AL, Brown BV, Mueller UG. Foundress queen mortality and early colony growth of the leafcutter ant, Atta texana (Formicidae, Hymenoptera). Insectes Soc. 2015;62(3):357–63.
- 60. Wiernasz DC, Cole BJ. Queen size mediates queen survival and colony fitness in harvester ants. Evolution. 2003;57(9):2179–83. pmid:14575338
- 61. Valverde JP, Hanson P. Parenchyma: a neglected plant tissue in the Cecropia/ant mutualism. Symbiosis. 2011;55(1):47–51.
- 62. Müller F. Über das Haarkissen am Blattstiel der Imbauba (Cecropia), das Gemüsebeet der Imbauba-Ameise. Jena Z Med Naturwiss. 1876;10:281–6.
- 63. Hv Ihering. Die Cecropien und ihre Schutzameisen. Englers Bot Jahrb. 1907;39:666–714.
- 64. Nishi AH, Romero GQ. Colonization pattern of Cecropia by Azteca ants: Influence of plant ontogeny, environment and host plant choice by queens. Sociobiology. 2008;52(2):367–76. PubMed PMID: WOS:000259163300011.
- 65. Silliman BR, Newell SY. Fungal farming in a snail. Proc Natl Acad Sci U S A. 2003;100(26):15643–8. pmid:14657360
- 66. Mueller UG, Gerardo NM, Aanen DK, Six DL, Schultz TR. The evolution of agriculture in insects. Annu Rev Ecol Evol Syst. 2005;36:563–95.
- 67. Hata H, Kato M. A novel obligate cultivation mutualism between damselfish and Polysiphonia algae. Biol Lett. 2006;2(4):593–6. pmid:17148297
- 68. Pauli JN, Mendoza JE, Steffan SA, Carey CC, Weimer PJ, Peery MZ. A syndrome of mutualism reinforces the lifestyle of a sloth. Proc R Soc Lond B Biol Sci. 2014;281(1778). pmid:24452028
- 69. Thurber AR, Jones WJ, Schnabel K. Dancing for food in the deep sea: bacterial farming by a new species of Yeti crab. PLoS ONE. 2011;6(11):e26243. pmid:22140426
- 70. Ponnudurai R, Kleiner M, Sayavedra L, Petersen JM, Moche M, Otto A, et al. Metabolic and physiological interdependencies in the Bathymodiolus azoricus symbiosis. ISME J. 2016;11:463. pmid:27801908