Introduction

Ant–plant mutualisms are pervasive and important components of tropical ecosystems, and represent unique arboreal niches in which to study microbial ecology. In these symbioses, plant hosts generally provide their ant symbionts with two main resources, a nutrient-rich food source in the form of extrafloral nectar secretions or specialized plant structures (e.g., Müllerian bodies from Cecropia trees or Beltian bodies from Acacia trees), and nesting space in hollowed out plant structures termed domatia [1]. In return, the ants protect the plants from herbivores and remove encroaching vegetation. Deposition of waste materials by ants inside domatia may also provide nutrients to the host plant, particularly nitrogen which is often a limited resource for plants [2–9]. Both plant hosts and resident ants are phylogenetically diverse and are widespread throughout the tropics [10]. Fungi have long been observed in the domatia of many ant–plant mutualisms [11], but the mutualistic nature of this interaction has only recently been demonstrated [12]. The fungi in these patches were identified as black yeasts in the ascomycete order Chaetothyriales, observed to be fed to larvae, and shown to obtain nitrogen from ant activity [12–15]. Although fungal hyphae in these ant–plant systems were found in contact with plant material inside the domatia, they were not observed to grow into the plant tissue, suggesting that these fungal symbionts are likely not pathogenic to the host plant [12, 15, 16]. This recent work suggests that the mutualism extends beyond the ant–plant interaction and that these systems include aspects of insect–fungal agriculture [16, 17].

Insect fungiculture requires that the insect host protects both itself and its fungal mutualist from pathogens [18]. While it is likely that all fungus farmers are threatened by potentially virulent pathogens, attine ants remain the best studied example. Infections by the co-evolved mycoparasitic fungus Escovopsis have been documented in at least one third of attine ant colonies, with infection rates varying by genus, species, and age of the colony [19, 20]. Colonies control infection using at least three different tactics: mechanical removal of the pathogen by garden weeding, application of antifungal metabolites produced by Actinobacteria present on the ants’ cuticle, and potentially by application of secretions from the metapleural glands [21–24]. Likewise, antibiotic-producing Actinobacteria have been isolated from fungus-growing termites. These isolates displayed antifungal activity against the fungal pathogen Pseudoxylaria [25]. Other insects also use secondary metabolites produced by Actinobacteria. Several antimicrobials produced by the bacterium ‘Candidatus Streptomyces philanthi’ help protect overwintering beewolf brood and its food source from fungal attack [26–28]. Actinobacteria with potentially similar roles have been found in other insect systems, including bark beetles, ambrosia beetles, ants in the genus Allomerus, the woodwasp Sirex noctilio, and two species of solitary mud dauber wasps [29–33].

In addition to producing numerous secondary metabolites, Actinobacteria are also capable of degrading a variety of substrates, including recalcitrant plant material [34]. Numerous enzymes are needed to fully degrade plant compounds such as cellulose, hemicellulose, and lignin. Rarely do individual organisms produce this suite of enzymes, but some Actinobacteria, such as particular species of Streptomyces, have this ability. Actinobacteria associated with herbivorous arthropods have recently been suggested to fulfill host nutritive roles by degrading recalcitrant plant material. Plant-associated insects, such as S. noctilio, emerald ash borer (Agrilus planipennis), and higher termites, may use the metabolic potential of cellulose-degrading Actinobacteria to gain access to nutrients in difficult to digest plant material [32, 35–37]. Streptomyces associated with these three insects degrade cellulose in vitro and may serve this role in the host or in their feeding galleries [32, 35–37]. Likewise, firebugs and cotton stainers require two actinobacterial symbionts to exploit Malvale seeds as novel food sources [38].

We hypothesized that Actinobacteria are associated with ant–plant–fungi mutualisms and may similarly serve as symbionts by the production of antimicrobial metabolites or cellulose-degrading enzymes. In addition to potentially protecting the black yeast from pathogens as in the symbioses described above, Actinobacteria might also protect the ants, their brood, or the host plant from infection or aid in host nutrition. We assessed three ant–plant symbioses known to harbor fungi in their domatia: Pseudomyrmex penetrator/Tachigali sp. from French Guiana and Petalomyrmex phylax/Leonardoxa africana and Crematogaster margaritae/Keetia hispida, both from Cameroon. Bacterial isolations were performed on samples from each system and subjected to 16S rRNA gene sequencing. Antifungal bioassays and filter paper degradation assays were performed on a subset of isolates to assess their potential as protective or nutritional symbionts.

Methods

Colony Sampling

Domatia containing ants were haphazardly cut from trees, placed in sterile containers, and shipped to the laboratory at the University of Wisconsin-Madison. Pseudomyrmex penetrator workers, queens, pupae, larvae, and scale insects in Tachigali sp. were collected in French Guiana in October 2009 (Tables 1 and 2). Additional fungal patches were scraped from the inside of Tachigali domatia and placed into either empty microfuge tubes or tubes containing 1× phosphate-buffered saline (PBS) or 10 % glycerol (Table 2). Four Leonardoxa africana domatia inhabited by Petalomyrmex phylax were similarly collected in Cameroon in January 2010 including workers, alates, larvae, pupae, and fungal patch scrapings, along with two Keetia hispida domatia with Crematogaster margaritae including workers, larvae, pupae, and scale insects (Tables 1 and 2).

Table 1 Details of Actinobacteria population obtained from ant/plant associations
Full size table
Table 2 Summary of isolation rate of putative Actinobacteria from different biotic components of ant/plant sources
Full size table

Actinobacterial Isolation Procedure

For all samples, Actinobacteria were initially cultured using chitin medium supplemented with the antifungals nystatin and cyclohexamide (per liter: 5.33 g chitin, 0.767 g K2HPO4, 0.367 g KH2PO4, 0.244 g MgSO4, 0.01 g FeSO4·7H2O, 0.001 g ZnSO4·7H2O, 0.001 g MnCl2·4H2O, 20 g of agar, 0.0427 g nystatin, and 0.0667 g cyclohexamide). Chitin medium is selective for Actinobacteria [39].

Three to four plates were prepared for each sample and spread with 100 μl of inoculum. To prepare inocula, P. penetrator workers, pupae, and queens were washed in 1 ml 1× PBS, placed in 500 μl 1× PBS, and shaken for 4 min in a Mini-beadbeater (Biospec Products, Bartlesville, OK) with 0.25 g of 1:1 400:800 μm beads. Except for fungal patches, all remaining samples were placed in 500 μl 1× PBS with beads and shaken for 2 min twice in a beadbeater and used as inocula. Fungal patches originally placed in PBS and glycerol were homogenized with mini-pestles and used as inocula. Fungal patches without any solution for transport were shaken in a beadbeater for 2 min with 500 μl of 1 % Tween 20 in 1× PBS with beads, centrifuged for 3.5 min, homogenized with a minipestle, and used as inocula. Plates were incubated at 28 °C for up to 4 months and checked regularly for growth morphologically consistent with Actinobacteria growing on chitin medium [39]. Putative Actinobacteria colonies were counted and all morphologically distinct colonies were streaked for isolation first using chitin medium followed by yeast malt extract agar medium containing antifungals (YMEA) (per liter: 4 g yeast extract, 10 g malt extract, 4 g dextrose, 15 g agar, 40 mg nystatin, and 0.05 g cyclohexamide). To determine if colony morphology could serve as a way to group isolates, all isolates were grouped based on colony morphology on YMEA, and representatives of each morphotype from each ant–plant system were subjected to 16S rRNA gene sequencing and phylogenetic analysis. Sequence names correspond to the ant–plant system of origin, PsTa for P. penetrator/Tachigali, LaPp for P. phylax/L. africana, and KhCr for C. margaritae/K. hispida.

DNA Extraction, 16S rRNA Gene Amplification, and Sequencing

A portion of the isolates morphologically identified as Actinobacteria on YMEA were chosen for 16S rRNA gene sequencing: 60 out of 109 isolates from P. penetrator/Tachigali, 104 out of 353 isolates from P. phylax/L. africana, and 121 out of 399 isolates from C. margaritae/K. hispida. The remaining isolates were not sequenced or used for assays.

To extract genomic DNA from pure cultures, a loopful of actinobacterial growth was added to 250 μl of 2× cetyltrimethylammonium bromide extraction buffer and lysed by triplicate bead beating for 2 min followed by incubation at −80 °C for 2.5 min [40]. One volume of phenol-chloroform-isoamyl alcohol (25:24:1) was added, vortexed, and centrifuged for 5 min at 13,000×g. The upper phase was mixed with one volume of chloroform-isoamyl alcohol (24:1) in a new tube, vortexed, and centrifuged for 5 min at 13,000×g. The upper phase was then mixed with one volume of cold isopropanol, incubated at −20 °C overnight, and centrifuged for 15 min at 13,000×g at 4 °C. The resulting DNA pellet was washed once with 70 % ethanol, dried, and resuspended in 50 μl TE buffer.

The universal bacterial primers 27F (5′ AGAGTTTGATCNTGGCTCAG) and 1492R (5′ TACGGYTACCTTGTTACG) were used to amplify the near full-length 16S rRNA gene [41]. Each 25 μl reaction included the following: 12.5 μl EconoTaq Plus Green (Lucigen, Middleton, WI), 200 nM of each primer, and 10 ng of template DNA. The thermocycling program was as follows: initial denaturation at 95 °C for 2 min; 35 cycles of 95 °C for 20 s, 55 °C for 20 s, and 72 °C for 2 min; and a final extension of 72 °C for 3 min. PCR products were visualized on a 2 % agarose gel in 1× TAE buffer to confirm the presence of the expected single band at ~1,465 bp.

PCR products were sequenced using the BigDye Terminator mix v. 3.1 (Applied Biosystems, Carlsbad, CA) separately using primers 27F and 1492R. Each 10 μl reaction included 1 μl BigDye mix, 1.5 μl BigDye buffer, 100 nM 27F or 1492R, 0.5 μl DMSO, and 0.5 μl PCR product. The thermocycling program was as follows: initial denaturation at 96 °C for 2 min; 35 cycles of 96 °C for 10 s, 52 °C for 15 s, and 60 °C for 3 min; and a final extension of 72 °C for 1 min. PCR products were purified using CleanSeq magnetic beads (Agencourt Biosciences, Brea, CA) as per the manufacturer’s protocol and sequenced using an ABI 3730x sequencer at the University of Wisconsin-Madison Biotechnology Center. Sequences were deposited in Genbank under accession numbers KJ889020–KJ889240 (see also Supplemental Table 1).

Phylogenetic Tree Construction

16S rRNA gene sequence contigs were aligned and edited using Sequencher v 4.5 (Gene Codes Corporation, Ann Arbor, MI). Sequences with more than five ambiguous bases were discarded. Actinobacterial sequences were identified using SeqMatch from the Ribosomal Database Project (RDP), release 10, update 31 [42]. Non-Actinobacteria sequences were not used in further analyses. After removing non-Actinobacteria and low quality sequences, 46 P. penetrator/Tachigali, 68 P. phylax/L. africana, and 107 C. margaritae/K. hispida sequences were used for phylogenetic analysis. Reference actinobacterial 16S rRNA gene sequences, including taxa previously isolated from insects [25, 27, 30, 32, 43, 44], were downloaded from the Genbank and RefSeq databases [45, 46]. The 16S rRNA gene sequences of Slackia exigua and Atopobium fossor were used as a basal Actinobacteria outgroup, based on Gao and Gupta [47]. All reference sequences used here are listed in Supplemental Table 1. All sequences were aligned using the G-INS-i algorithm implemented in MAFFT v7.017b [48], trimmed at the 5′ and 3′ ends to remove trailing ends of longer sequences, and then realigned. Where individual samples had two or more identical sequences, only one sequence was used for the tree (henceforth referred to as duplicate sequences). This alignment was then used to generate a phylogenetic tree using RAxML 7.2.6 with the GTRGAMMA nucleotide substitution model and 1,000 rapid bootstraps [49, 50]. The tree was visualized using the web version of iTOL [51, 52].

Actinobacteria–Fungal Petri Plate Bioassay

Nineteen isolates were chosen for fungal bioassays based on initial morphological diversity. They were challenged with two entomopathogens, Fusarium oxysporum and Metarhizium anisopliae, in addition to the general fungi Aspergillus niger and Trichoderma reesei, as per Cafaro et al. and Poulsen et al. [53, 54]. A small loopful of each Actinobacteria isolate was point inoculated in quadruplicate into the center of 8.5 cm Petri dishes containing YMEA without antifungals and incubated for 10 days at 28 °C. Test fungi were then point inoculated 2.5 cm away from the edge of the bacterial colony, alongside control plates without Actinobacteria. Bioassay plates were incubated at 28 °C and inspected daily until the fungus in the fungus-only control plates had grown completely to the opposite side of the plate. Each bioassay plate was photographed and their zones of inhibition (ZOI) were measured.

Cellulose Degradation Assay

The same 19 isolates described above were also tested for their ability to degrade cellulose. Isolates were inoculated in duplicate into test tubes containing 5 ml of autoclaved M63 medium and 5 ml of trace elements solution (SPV-4) [55]. A strip of 1 cm × 10 cm Whatman #1 filter paper was added to each tube as a cellulose source. Tubes were incubated with constant agitation at 331 rpm at 28 °C for a maximum of 22 days and examined daily for visual evidence of filter paper degradation. Streptomyces sp SirexAA-E was used as a positive control, while a tube not inoculated with bacteria served as a negative control [55]. From the original culture collection, an additional 18 isolates with the same morphology on YMEA as Streptomyces sp KhCrAH244 were subsequently assayed for cellulose degradation, and their 16S rRNA genes sequenced. Likewise, seven additional isolates from clade III were retrospectively assayed for cellulose degradation.

Results

In total, we screened for Actinobacteria in 96 samples from Pseudomyrmex penetrator/Tachigali, 99 samples from Petalomyrmex phylax/L. africana, and 40 samples from C. margaritae/K. hispida. From all three ant–plant mutualisms, Actinobacteria were isolated from all types of insect samples, including workers, pupae, larvae, queens, and alates (Table 2). Actinobacteria were cultured from all fungal patches from P. phylax/L. africana including all three storage techniques and from P. penetrator/Tachigali fungal patches stored in PBS (Table 2). In total, Actinobacteria were isolated from 26 % of samples from P. penetrator/Tachigali, 67 % of P. phylax/L. africana samples, and 100 % of C. margaritae/K. hispida samples (Table 1). Average putative Actinobacteria colony forming units (CFUs) per sample varied across the three systems: P. penetrator/Tachigali samples contained 1.3 CFUs, P. phylax/L. africana samples contained 3.6 CFUs, and C. margaritae/K. hispida samples contained 10.0 CFUs (Table 1). Petalomyrmex phylax alates, P. penetrator workers, and C. margaritae pupae yielded the most CFUs, 8.3, 3.5, and 10.7, respectively (Table 2).

Five major clades of Actinobacteria were isolated from these ant–plant systems (Fig. 1, Supplemental Table 1). Clade I comprises 36 actinobacterial 16S rRNA gene sequences from P. penetrator/Tachigali, 14 from P. phylax/L. africana, and 68 from C. margaritae/K. hispida. These isolates form a clade with the Streptomyces albidoflavus species group (Fig. 1) [56]. Clade I contains the majority of ant–plant sequences in this tree and is the only well-supported clade associated with isolates from all three sampled ant–plant mutualisms. Clade II includes 9 isolates from P. phylax/L. africana, 21 from C. margaritae/K. hispida, the Streptomyces griseus species group [57], and the insect isolates Streptomyces griseus XylebKG1 (isolated from an ambrosia beetle), Streptomyces sp AV109 (from a fungus-farming termite), and Streptomyces sp SirexAA-E (from Sirex noctilio) [25, 30, 32]. Clade III comprises 2 isolates from P. phylax/L. africana, 17 isolates from C. margaritae/K. hispida, and the reference sequence Streptomyces drozdowiczii. Clade IV comprises 10 isolates from P. phylax/L. africana and the reference sequences Streptomyces recifensis, Streptomyces griseoluteus, and Streptomyces seoulensis. Lastly, clade V contains 20 P. phylax/L. africana isolates and is sister to Nocardioides luteus and Nocardioides albus. Clades I–V are supported by bootstrap values of 71, 78, 94, 60, and 97 %, respectively. The remaining 14 isolates represent singletons or minor clades. Only clade II contained Actinobacteria isolated from other insects (as described above), as did a minor clade containing two isolates from P. phylax/L. africana that were related to isolates from Dendroctonus frontalis and Sirex noctilio [32, 43].

Fig. 1
figure 1figure 1figure 1

Phylogenetic tree of Actinobacteria isolated from three ant–plant systems, related reference sequences, and other Actinobacteria isolated from arthropods. Leaves are color-coded based on isolate origin, including Pseudomyrmex penetrator/Tachigali (orange) (n = 46), Petalomyrmex phylax/L. africana (yellow) (n = 68), C. margaritae/K. hispida (green) (n = 107), other insects (pink), and reference sequences (gray). The outer color stripes represent the five major clades of isolates associated with the three ant–plant systems explored here, including reference sequences and other insect isolates that fall within these clades. From top to bottom, these include clade I (red), clade II (blue), clade III (green), clade IV (purple), and clade V (orange). a Top portion of the tree; b middle portion of the tree; c the bottom portion of the tree

Full size image

Of the 19 isolates tested for antifungal activity, only eight isolates exhibited distinct ZOI (Table 3; Supplemental Fig. 1). Streptomyces sp PsTaAH124, Streptomyces sp PsTaAH130, and Streptomyces sp LaPpAH322 inhibited all four fungi tested, while Streptomyces sp KhCrAH320 inhibited one of the entomopathogens, M. anisopliae, and one of the generalists, T. reesei. Streptomyces sp PsTaAH5 and Streptomyces sp KhCrAH316 inhibited only T. reesei, while Streptomyces sp KhCrAH308 and Streptomyces sp LaPpAH224 inhibited only M. anisopliae. Of the tested fungi, M. anisopliae was inhibited inconsistently with only one of four replicates from six isolates exhibiting a clear ZOI, while T. reesei was inhibited in all replicates from six isolates.

Table 3 Antifungal bioassay of selected Actinobacteria isolates against fungal test organisms
Full size table

Only 1 of the original 19 isolates tested, isolate Streptomyces sp KhCrAH244, degraded cellulose averaging 9 days until filter paper breakage (Fig. 2). Eighteen isolates with the same culture morphology as Streptomyces sp KhCrAH244 were subsequently chosen to further confirm this phenotype. Likewise, seven isolates from clade III were retrospectively chosen for this assay. All isolates tested from clade III degraded the filter paper, including eight of the isolates selected based on culture morphology. The remaining 10 isolates were not members of clade III and did not degrade the filter paper. Isolates took between 7 and 22 days to degrade filter paper, averaging 9.3 days, while our positive control, Streptomyces sp SirexAA-E, took 7.5 days.

Fig. 2
figure 2

Average number of days until evidence of filter paper degradation. Forty-four isolates were tested for their ability to degrade filter paper. Eleven isolates are included in this figure representing novel isolates capable of degrading cellulose. All the isolates represented are Streptomyces from clade III. Error bars represent standard deviation. Streptomyces sp SirexAA-E was included as a positive control

Full size image

Discussion

We investigated three ant–plant mutualisms as niches for Actinobacteria and their potential roles as nutritional or defensive symbionts within these systems. In total, we isolated 861 putative Actinobacteria from Pseudomyrmex penetrator/Tachigali sp., Petalomyrmex phylax/L. africana, and C. margaritae/K. hispida, suggesting that these ant–plant symbioses are indeed niches for Actinobacteria. Actinobacteria were obtained from the following: (1) three phylogenetically disparate ant species, (2) two geographically distant locations, (3) all three ant–plant symbioses explored, and (4) all types of samples of each system examined. All C. margaritae/K. hispida samples contained culturable Actinobacteria while a portion of P. penetrator/Tachigali and P. phylax/L. africana contained culturable Actinobacteria (Tables 1 and 2). Every life stage across all three systems were also associated with Actinobacteria, including workers, alates, larvae, pupae, and queens (Table 2). The range of Actinobacteria CFUs found here is consistent with those found in other insect–Actinobacteria associations examined. Plates generated from S. noctilio were found to have an average of 11.5 CFUs, which dropped to 3.1 when two outliers were removed [32]. Similarly, most D. frontalis were associated with at least 1 CFU of Streptomyces, with an average of 7.7 CFUs across 110 beetles sampled [43].

Isolation of Actinobacteria from fungal patches was more inconsistent than from other material types, though this may have been partially a factor of storage technique. All fungal patches from P. phylax/L. africana yielded Actinobacteria, regardless of how the fungal patch was prepared for transport. However, Actinobacteria were cultured only from fungal patches transported in PBS from P. penetrator/Tachigali. Similarly, Seipke et al. surmised that storage and transport conditions likely biased their results in looking for Actinobacteria associated with ants in the genus Allomerus [31]. The possibility remains that fungal patches from P. penetrator/Tachigali were more fastidious and only those stored in PBS for transport survived.

Our results suggest that the three ant–plant mutualisms explored here associate with only a few species of culturable Actinobacteria. Clades I–IV were identified as Streptomyces while clade V was identified as Nocardioides (Fig. 1). Clade I was the only clade isolated from all three ant–plant symbioses explored and could represent a Streptomyces symbiont important to the tripartite symbiosis between ants, plants, and fungi, regardless of geographic location [12]. Two clades (II and III) were associated with both of the systems from Cameroon, while the remaining two clades (IV and V) were specific for P. phylax/L. africana. This arrangement of clade association could indicate geographic (clades II and III), host (clades IV and V), and ant–plant–niche specificity (clade I).

All but one of the major clades presented here represent clades that were not previously known to include isolates from insects. Clade II in contrast includes isolates from three other insect systems that also associate with fungus including an ambrosia beetle, a fungus-farming termite, and S. noctilio (Fig. 1) [25, 30, 32]. As this clade also includes the Streptomyces griseus species group, this seeming ubiquity might be due to lack of resolution in the 16S rRNA gene and these isolates may still be novel species [57]. Likewise, other insect isolates might clade with our isolates, but the lack of near full-length 16S rRNA gene sequences from those isolates in public databases precluded their inclusion in our phylogenetic tree. Including those shorter sequences would likely have required trimming our sequences to a length where most of the phylogenetic resolution would have been lost. Additionally, while other genera from the phylum Actinobacteria are represented in our dataset, isolates were predominately identified as Streptomyces. Frequent isolation of Streptomyces may be due to bias in culturing technique and medium choice, and is unlikely to be indicative of the absence of other types of Actinobacteria from these systems. Indeed, ‘Candidatus Streptomyces philanthi,’ a known symbiont of beewolves, has yet to be cultured.

In general, little is currently known about ant pathogens, and specific pathogens have yet to be identified for the three systems studied here. We used A. niger, F. oxysporum, M. anisopliae, and T. reesei as proxies for potential fungal pathogens in these systems. A wide range of antifungal activity was found in bioassays challenging these potentially pathogenic fungi with phylogenetically diverse Actinobacteria isolated from all three systems (Table 3). Actinobacteria, particularly Streptomyces, are well-known producers of secondary metabolites, many of which demonstrate antifungal activity. All of the tested Actinobacteria isolates exhibited some level of antifungal capacity, with the Streptomyces isolates performing better than the Nocardioides isolates. However, only eight isolates produced measurable ZOIs (Table 3). T. reesei, a generalist fungal pathogen, was the most inhibited fungus, with three isolates from clade I and three singletons producing large ZOIs. M. anisopliae, a pathogen known to infect many insects, was the least inhibited, with one isolate each from clades I and II, and four singletons, producing small ZOIs. It is important to note that finding in vitro antifungal activity is not indicative of in vivo function. In vivo activity can be difficult to demonstrate. Likewise, antimicrobial activity can be dependent on a number of factors, including production of the antimicrobial compound, density of bacterial or fungal growth, sensitivity of the fungi to the compound, and temperature, among many others. To date, the only reports showing in vivo antifungal activity of Actinobacteria in insects include attine ants and beewolves [22, 58].

Unlike antifungal activity, the ability to degrade cellulose was found in only one clade of isolates from the ant–plant systems from Cameroon (clade III, Figs. 1 and 2). As these ants are arboreal, these cellulose-degrading bacteria could be playing any of several roles. While the fungi in these ant–plant systems do not appear to penetrate the plant tissue inside the domatia [12], cellulose-degrading bacteria could still co-occur with these ants and may be releasing by-products from cellulose degradation that are used by the ants. We should also consider that isolates clustered in clade III are pathogens of the Cameroonian systems, as the benefit they provide to either plant or ants has not been confirmed. Similar cellulose-degrading properties have been demonstrated with Actinobacteria from other plant-associated insects and are suggested to aid in nutrient acquisition [32, 35–37]. An isolate of Streptomyces cultured from emerald ash borer was capable of degrading carboxymethylcellulose in vitro [35]. Likewise, Streptomyces sp SirexAA-E may contribute nutritionally to S. noctilio by degrading recalcitrant plant material [32, 55].

Colony morphology was found to be a poor predictor of phylogenetic placement, particularly for Streptomyces isolates. This was particularly evident in the additional isolates screened for cellulose degradation, described above. Of the 18 chosen based on morphology, only eight were identified as members of clade III after 16S rRNA gene sequencing. The 10 remaining isolates, despite having similar colony morphology to Streptomyces sp KhCrAH244, did not degrade cellulose and were not members of clade III.

Taxonomically diverse insects have convergently evolved similar solutions for nutrient acquisition by associating with fungal symbionts [18]. Likewise, multiple insects have also evolved associations with Actinobacteria that provide protection of the host insect, brood, and their food source [59]. Consistent co-occurrence of isolates from clade I, and others, suggests that the ant–plant–fungus systems explored here associate with more than one clade of Actinobacteria and may also use the metabolic potential found in this phylum of bacteria. Our work describes the isolation of Actinobacteria from Pseudomyrmex penetrator/Tachigali, Petalomyrmex phylax/L. africana, and C. margaritae/K. hispida, and demonstrates the potential roles for these bacteria. Future work is needed to determine if the Actinobacteria isolated here are novel strains or species and to determine if these bioactive Actinobacteria isolates serve the roles described here in vivo.