Introduction

The tribe Attini is comprised of approximately 45 ant genera with worldwide distribution (Ward et al. 2015). Most notably in this tribe are the fungus-growing “attine” ants, which cultivate fungi for food (Schultz and Brady 2008; Ward et al. 2015). These ants are found only in the American continent and comprise 17 genera with more than 250 species (Brandão et al. 2011; Sosa-Calvo et al. 2013). Attine ants maintain a mutualistic interaction with basidiomycetous fungi in an obligatory association that originated about 50 million years ago (Schultz and Brady 2008). In this mutualism, workers collect various types of substrates (e.g. fresh leaves, seeds, and insect carcasses) to nourish the fungal cultivar, besides favoring fungal dispersion during the ants’ reproductive stage. In turn, the fungal partner is the primary food source for the brood and minor component in the nutrition of ant workers (Murakami and Higashi 1997; Silva et al. 2003).

Fungus-growing ant colonies represent a complex biological system that harbor a wide range of microorganisms in addition to the fungal cultivar, including bacteria, filamentous fungi and yeasts (Fisher et al. 1996; Rodrigues et al. 2008). Such microbes are found in the soil that composes the nest tunnels and chambers (Rodrigues et al. 2014) as well as in the fungus garden, structure that houses the substrate foraged by the ants and the mycelium of the fungal cultivar (Weber 1972).

Möller (1893) was the first to report the presence of alien fungi such as Aspergillus, Mucor, Penicillium and Rhizopus in the gardens of Acromyrmex disciger. Later, Kreisel (1972) reported Cunninghamella, Fusarium, Rhizopus and Trichoderma in ant-deprived gardens of the leafcutter ant Atta insularis. Although several studies have demonstrated the diversity of filamentous fungi found in gardens of attine ants, the origin of these fungi is still uncertain. Subsequent studies suggested such fungal diversity is mostly related to the type of plant substrate that workers use to nourish the fungal cultivar (Fisher et al. 1996; Van Bael et al. 2009).

Filamentous fungi and other microbes that enter the ant colony are derived from the surface of the plant material. Also, due to the complex ant behaviors associated with substrate preparation (scraping and cleaning the surface of leaves to eliminate microorganisms), the origin of some of the alien fungi in leaf-cutting ant colonies is also endophytic (Van Bael et al. 2009). Furthermore, Rodrigues et al. (2005, 2011) suggested that workers that are in direct contact with soil also carry fungi to their colonies. Although most studies focused on leaf-cutting ants, no research investigated the diversity of filamentous fungi in colonies of non-leaf-cutting attine ants. Because these ants use a variety of substrates such as seeds, plant sap, insect frass and parts of dried leaves and flowers to grow the fungal cultivar (Rodrigues et al. 2005, 2008; Mehdiabadi and Schultz 2010), it is possible those substrates contain several fungi that are carried into their colonies, including Trichoderma species.

Fungi in the genus Trichoderma occur in different parts of the world (Samuels et al. 2006; Druzhinina et al. 2011; Chaverri and Samuels 2013) on various substrates, including: (i) soil; (ii) decaying wood; (iii) on edible mushrooms and (iv) some as endophytes in plants (Samuels et al. 2006; Zhang et al. 2007; Hanada et al. 2008; Bae et al. 2009; Chaverri et al. 2011; Druzhinina et al. 2012; Bailey and Melnick 2013; Chaverri and Samuels 2013). Trichoderma species have a high adaptability to different substrates and environments (colonizing diverse ecological niches), probably due to the physicochemical mechanisms (production of enzymes and antibiotics) as tools for obtaining nutrients (Samuels et al. 2006; Jaklitsch 2009; Druzhinina et al. 2011; Druzhinina and Kubicek 2013; Atanasova et al. 2013). Due to the diversity of substrates collected by attine ants, and the cosmopolitan nature of Trichoderma, new Trichoderma species could enter the colonies of these insects.

Since the first report of Trichoderma in colonies of leaf-cutting ants (Kreisel 1972), several studies showed the occurrence of these fungi in colonies of other attine ant genera (Barbosa 2004; Poulsen and Currie 2006; Rodrigues et al. 2008, 2009; Augustin et al. 2011; Rodrigues et al. 2014). However, none of these studies used a detailed taxonomic approach to identify Trichoderma species that occur in this environment. Also, some of these studies sequenced the internal transcribed spacer 1 and 2 (ITS 1 and 2) of the rDNA region, which does not resolve species identification in several Trichoderma clades, especially for the Trichoderma harzianum species complex (Atanasova et al. 2013; Chaverri et al. 2015).

Systematic studies involving other microorganisms associated with leaf-cutting ants showed a high diversity in the fungus gardens or on the integument of ants, including many new yeast species (Middelhoven et al. 2003; Carreiro et al. 2004; Pagnocca et al. 2010; Attili-Angelis et al. 2014; Melo et al. 2014). However, studies focused on the taxonomy and systematics of filamentous fungi associated with these insects have only considered the genus Escovopsis, the specialized parasite of the mutualistic fungus (Muchovej and Della Lucia 1990; Seifert et al. 1995; Currie et al. 1999, 2003; Augustin et al. 2013; Masiulionis et al. 2015; Meirelles et al. 2015a, b). These studies revealed the presence of several new Escovopsis species, which suggests that attine ant colonies might also harbor a high diversity of filamentous fungi, including potential new species of the genus Trichoderma.

Currently, no study has focused on the taxonomic evaluation of Trichoderma species found in the microbial consortium of attine ant colonies. In this sense, this study systematically assessed Trichoderma species in colonies of 12 attine ant species (nine species of leaf-cutting and three of non-leaf-cutting ants) belonging to four genera (Atta, Acromyrmex, Trachymyrmex, and Cyphomyrmex). We used morphological and molecular markers to identify 62 strains isolated from fungus gardens, soil adjacent to the garden and soil distant from the ant colonies. Here, we report a high number of Trichoderma species found in the attine ant environment, including three new Trichoderma species as well as a new phylogenetic taxon. Such findings suggest that colonies of these insects may harbor more fungal species still unknown to science.

Materials and methods

Strains examined

The strains examined were obtained in several studies conducted over 10 years (2004–2013) by the research group of the Laboratory of Ecology and Fungal Systematics (LESF) in combination with other laboratory (Section of Integrative Biology, University of Texas at Austin, USA). Ant colonies were excavated to find the fungus gardens. Fragments from the top (new garden) and the bottom (old garden) sections of the fungus gardens were collected in sterile containers. Pieces (3 mm2) from both sections of this substrate were plated in different culture media supplemented with antibiotics (see Rodrigues et al. 2005, 2008, 2011). In addition, Trichoderma strains from soil samples were isolated by the spread plating technique (Rodrigues et al. 2014). Monosporic cultures were obtained for all isolates. All strains are preserved in 10 % glycerol at −80 °C at the UNESP—Microbial Resources Center (CRM-UNESP), Rio Claro, Brazil.

A total of 62 Trichoderma strains were revived and examined. From this total, 56 are derived from Atta and Acromyrmex leaf-cutting ants, comprehending 47 from fungus gardens, two from soil adjacent to garden and seven from soil 10 m away from ant colonies. In addition, we examined six strains derived from fungus gardens of non-leaf-cutting ants in the genera Trachymyrmex and Cyphomyrmex (Table S1).

Trichoderma strains here studied are derived from a total of 37 ant colonies, corresponding to 12 attine ant species (Table S1). Samples were collected in ten sites from four states in Brazil and two sites in Texas-USA (Fig. S1; Table S1), comprehending so far, the largest sampling of this taxon from attine ant colonies.

DNA extraction, PCR and sequencing

DNA extraction was performed directly from the mycelium grown in PDA medium for 5 days in the dark. Genomic DNA extraction followed the method adapted from Möller et al. (1992) and Gerardo et al. (2004). Three molecular markers were amplified: the ITS region (ca. 586–622 bp) with the primer pair ITS4 (5′TCCTCCGCTTATTGATATGC3′) and ITS5 (5′GGAAGTAAAAGTCGTAACAAGG3′ Schoch et al. 2012); the partial sequences (ca. 563–615 bp) of the gene coding for the elongation factor 1 alpha (tef1) (EF1-728F: 5′CATCGAGAAGTTCGAGAAGG3′and TEF1R: 5′ GCCATCCTTGGAGATACCAGC3′, Carbone and Kohn 1999; Samuels et al. 2002); and the gene coding for the second subunit (ca. 900–1130 bp) of the RNA polymerase II (rpb2) (fRPB2-5F: 5′GA(T/C)GA(T/C)(A/C)G(A/T)GATCA(T/C)TT(T/C)GG-3′ and fRPB2-7cR: 5′CCCAT(A/G)GCTTG(T/C)TT(A/G)CCCAT-3′, Liu et al. 1999). PCRs for the three markers were performed in a final volume of 25 µL (4 µL of dNTPs [1.25 mm each]; 5 µL of 5× buffer; 1 µL of BSA [1 mg mL−1]; 2 µL of MgCl2 [25 mm]; 1 µL of each primer [10 µm]; 0.2 µl of Taq polymerase [5 U µL−1], 2 µL of diluted genomic DNA [1:100] and 8.8 µL of sterile ultrapure water). PCR conditions for ITS were 94 °C/3 min followed by 35 cycles at 94 °C/1 min, 55 °C/1 min and 72 °C/2 min; for tef1 and rpb2, conditions were 94 °C/2 min followed by 15 cycles at 94 °C/30 s, 65 °C/30 s and 72 °C/1 min; followed by 35 cycles at 94 °C/30 s, 48 °C/30 s and 72 °C/1 min.

Amplicons were purified using Wizard® SV Gel and PCR Clean-Up System Kit (Promega) following the manufacturer’s protocol. Then, samples were quantified in NanoDrop® (Thermo Scientific) and subjected to cycle sequencing reaction using BigDye Terminator® v.3.1 Kit (Life Technologies), following the manufacturer’s instructions (primers used for sequencing were the same used in the amplification step). The sequencing conditions for all molecular markers were 95 °C/min followed by 28 cycles at 95 °C/15 s, 50 °C/45 s, and 60 °C/4 min. Bidirecional sequences (forward and reverse) were generated in ABI 3330xl sequencer (Life Technologies) and assembled into contigs in BioEdit v.7.0.5.3 (Hall 1999).

Phylogenetic analyses

ITS, tef1 and rpb2 contigs of all strains were compared to homologous sequences deposited in NCBI-GenBank. Due to the low intraspecific ITS resolution for Trichoderma (Druzhinina and Kubicek 2005; Druzhinina et al. 2006; Atanasova et al. 2013; Chaverri et al. 2015) and also detected in our study, only tef1 and rpb2 sequences were used for phylogenetic analyses. However, due to the importance of ITS as a barcode marker for fungi, such sequences are also provided for all strains. Sequences generated in the present study were deposited in the NCBI-GenBank database (see Table S2 for accessions).

Sequences generated in other studies (Table S2) were used in our phylogenetic analyses and were retrieved from the NCBI-GenBank database. These sequences were selected by inferring initial phylogenetic trees (not shown) for each Trichoderma clade in attempt to avoid sequences that were too divergent and could hinder the global analysis. After sequences selection, the final dataset comprised 195 partial tef1 sequences (833 bp) and 155 partial rpb2 sequences (1058 bp). Alignments were performed separately for each molecular marker using MAFFT v.7 (Katoh and Standley 2013). The tef1 and rpb2 alignments were concatenated using Winclada v.1.00.08 (Nixon, 2002), and the final file contained a total of 1891 bp.

We applied Bayesian inference in MrBayes v.3.2.2 to reconstruct our global phylogenetic tree (Ronquist et al. 2012) with the concatenated dataset (tef1 and rpb2). Nucleotide substitution models, GTR+I+G and SYM+I+G, were selected for tef1 and rpb2, respectively, using the Bayesian information criterion under 95 % of confidence interval in jModelTest 2 (Darriba et al. 2012). We set up two separate runs, each consisting of three hot chains and one cold chain, and a Markov Chain Monte Carlo (MCMC) sampling of seven million generations, which were sufficient to achieve values for standard deviation of split frequencies below 0.01. Finally, the initial 25 % of the generations of MCMC sampling were discarded as burn-in. The refinement of the phylogenetic tree was carried out in FigTree v.1.4 and Adobe Illustrator CC v.17.1.

Morphological characterization

A total of 30 out of 62 strains were selected for morphological examination based on the results of the phylogenetic analysis. We examined (i) macroscopic characteristics of the colony (radial growth, mycelium color, presence of diffusible pigment, concentric rings and pustules) as well as (ii) microscopic features (shape and size of conidiophores, supporting cells, phialides, conidia and chlamydospores) based on Samuels et al. (2002, 2012a), Chaverri et al. (2001, 2003a, b), Bissett et al. (2003), Druzhinina et al. (2005) and Jaklitsch et al. (2006).

Macroscopic characteristics were evaluated in PDA (Potato Dextrose Agar), CMD (Corn-meal Agar) and SNA (Synthetic Nutrient Deficient Agar) grown for 7 days at 25 °C for 12 h (light) and 12 h (dark). Growth tests were performed on these three media and incubated at four different temperatures (25, 30, 35 and 40 °C) in darkness. The strains were initially grown for 7 days in water-agar medium. Thereafter, mycelium fragments of 0.5 cm diameter were inoculated on the three culture media (one cm from the edge of the plate). When the mycelium reached 7.5 cm radius, this size was sufficient to cover the entire Petri dish. Growth tests were performed in triplicate on three separate time periods for one week. Measurements of growth radius were performed after 72 h and 96 h.

To assess the microscopic characteristics we carried out slide culture preparations on PDA and CMD. For this purpose, we placed a 5-mm2 block of each media on a microscopic slide and then inoculated Trichoderma spores. Finally, the preparations were covered with a cover slip and incubated at 25 °C with alternating light (12 h light/12 h dark) for a time period of 3–5 days. Moreover, we also prepared wet mounts using 3 % KOH for additional observations of conidia and chlamydospores. All preparations were documented using an optical microscope (DM750, Leica, Germany) and measurements of the structures were performed with 30 replicates of each structure using the program LAS v.4.0 (Leica, Germany).

Results

Phylogenetic analyses

The 62 Trichoderma strains grouped into 20 different phylogenetic species. Among these species, we found four new phylogenetic taxa (LESF 236, LESF 545, LESF 551 and LESF 552, Fig. 1). The other 16 Trichoderma species identified are species that have been described in previous studies not related to attine ants. The 20 Trichoderma phylogenetic species belong to eight clades according to the division proposed by Atanasova et al. (2013): i) Clade Harzianum [T. afroharzianum (n = 6 strains), T. guizhouense (n = 1), T. lentiforme (n = 1), Trichoderma sp. LESF 552 (n = 1) and T. velutinum (n = 1)]; ii) Clade Hypocreanum [T. eucorticioides (n = 1), and Trichoderma sp. LESF 545 (n = 1)]; iii) Clade Longibrachiatum [T. andinense (n = 2), T. longibrachiatum (n = 1), T. orientalis (n = 3)]; vi) Clade Lutea [T. deliquescens (n = 1)], v) Clade Spinulosa [Trichoderma sp. LESF 236, (n = 1)]; vi) Clade Stromaticum [T. spirale (n = 15)]; vii) Clade Trichoderma [T. asperellum (n = 1), T. atroviride (n = 1), T. hamatum (n = 3), T. koningiopsis (n = 6), Trichoderma sp. LESF 551 (n = 1, Fig. 3), T. viride (n = 1) and viii) Clade Virens [T. virens (n = 12)].

Fig. 1
figure 1figure 1

Phylogenetic tree displaying the position of different Trichoderma species isolated from the attine ant environment (fungus gardens, soil next to fungus gardens and soil 10 m away from ant colonies). The analysis was based on concatenated sequences of tef1 and rpb2 markers. The tree was built using the Bayesian inference algorithm and the numbers on branches indicate the posterior probabilities. The tree includes a total of 195 sequences of tef1 and 155 sequences of rpb2. The final alignment had 1891 base pairs. We also included Trichoderma sequences derived from previous studies obtained from GenBank. Species names are followed by the strains codes. The species identified in this study with its code (LESF) at the end of each name are highlighted in bold. For each Trichoderma obtained in this study, we show in parentheses the ant species from which fungi strains were isolated. The four new phylogenetic taxa found here (three of them described as new species) are highlighted in red. Dashed lines indicate the different Trichoderma clades sensu Atanasova et al. (2013). Protodrea pallida CBS 29978 was used as outgroup. Bars indicate nucleotide substitution per site

Full size image

Among the four new strains found, three are described here as new species: Trichoderma attinorum (LESF 236), Trichoderma texanum (LESF 551) and Trichoderma longifialidicum (LESF 552). The strain LESF 545 did not sporulate; therefore, it was not possible to describe this fungus using this key morphological characteristic. This strain showed 88 and 89 % (rpb2) similarity with T. austriacum and T. taxii (FJ860525 and DQ859030, respectively). Strain LESF 236 was found to form a separate branch into the Spinulosa clade with high posterior probability support (100 %, Fig. 1). Regarding the molecular markers used, LESF 236 (Fig. 2) showed 83 % (tef1) and 91 % (rpb2) similarity with T. sinuosum (KJ665729 and KJ665342, respectively), 87 % (tef1) and 92 % (rpb2) with T. thelephoricola (FJ860711 and FJ860600, respectively) and 81 % (tef1) and 91 % (rpb2) with T. spinulosum (FJ860699 and FJ860589, respectively). Strain LESF 551 clustered in the Trichoderma clade but was found to form a separate branch from the other species with a moderate posterior probability support (80 %, Fig. 1). This strain showed 94 % (tef1) and 98 % (rpb2) similarity with T. koningiopsis (FJ463288 and FJ442730, respectively) and T. ovalisporum (AY376037 and FJ442742, respectively); and 97 % (rpb2) similarity with T. petersenii (FJ442783). Regarding strain LESF 552, it clustered in the Trichoderma clade and was found to form a separate branch from the other species with a high posterior probability support (100 %, Fig. 1). This strain showed 93 % (tef1) and 97 % (rpb2) similarity with T. velutinum (AY937415 and JN133569, respectively), 90 % (tef1) and 95 % (rpb2) with T. guizhouense (KF134799 and KF134791, respectively). Finally, strain LESF 545 was 89 % (rpb2) similar to T. citrinum (FJ179603) and T. taxi (DQ859032).

Fig. 2
figure 2

Morphological characteristics of Trichoderma attinorum. a–c Cultures on PDA, CMD and SNA, respectively, after 7 days of growth at 25 °C. d Dense mycelium with small pustules. e, f Conidiophore branching pattern. g Phialides pattern. h Conidia: i, j Chlamydospores

Full size image

Morphological analyses

When we compared morphological aspects of the three new strains with their phylogenetic closest relatives, we observed distinct characteristics. For instance, T. attinorum (LESF 236) grouped close to T. sinuosum, T. thelephoricola and T. spinulosum on our phylogenetic analysis (Fig. 1). T. spinulosum is only known by the sexual phase (Chaverri et al. 2003a), different from T. attinorum that only formed the asexual phase (Fig. 2). T. sinuosum growth rate is relatively higher than T. attinorum (41–58 mm on PDA and 29–36 mm on SNA at 25 °C; 45–60 mm on PDA and 29–39 mm on SNA at 30 °C for T. sinuosum; 39–41 mm on PDA and 5–10 mm on SNA at 25 °C, 30–31 mm on PDA and 6–11 mm on SNA at 30 °C for T. attinorum). The growth rate of T. thelephoricola is lower than the growth rate of T. attinorum (6–16 mm on PDA and 4–10 mm on SNA at 25 °C, 13–15 mm on PDA and 4–7 mm on SNA 30 °C for T. thelephoricola). Moreover, T. sinuosum and T. thelephoricola did not form soluble pigment in artificial culture, while T. attinorum does on all three culture media (Fig. 2a–c). Conidiophores of T. attinorum showed significant differences when compared to T. sinuosum and T. thelephoricola. In T. sinuosum they are branched and sinuous, in T. attinorum they are usually straight and with short branches (Fig. 2e–g). When compared to T. thelephoricola, the difference is evident since this species has Gliocladium-like conidiophores. The phialides of T. thelephoricola have shape and size similar to T. attinorum (lageniform and cylindrical); however, T. sinuosum presents ampulliform phialides distinct from those of T. attinorum (Fig. 2g). The conidia of T. thelephoricola are longer than the other two species. Finally, T. attinorum presents more elongated chlamydospores (Fig. 2i, j) than the one found in T. sinuosum (they are globular-subglobose in T. sinuosum and subglobose-ovoid in T. attinorum); chlamydospores are not reported for T. thelephoricola.

Trichoderma texanum grouped close to T. koningiopsis, T. ovalisporum, and T. petersenii (Fig. 1). The three latter species have concentric rings on PDA, while T. texanum does not (Fig. 3a). The optimum growth temperature was between 25 and 30 °C for all four species (T. koningiopsis, T. ovalisporum, T. petersenii and T. texanum). Pustule formation in T. texanum occurred to a lower extent on CMD, but always on the edge of the plate (Fig. 3b), unlike T. koningiopsis, T. ovalisporum, and T. petersenii, which form abundant pustules in concentric rings. Conidia pigmentation was visible after 7 days in T. texanum, differently from T. koningiopsis (after 48 h), T. ovalisporum (after 72 h), and T. petersenii (after 96 h). Only T. texanum produces soluble pigment in culture. Conidiophores in T. texanum are similar to T. koningiopsis, T. ovalisporum, and T. petersenii regarding the branches arrangement, but with a few differences: the conidiophores in T. koningiopsis and T. ovalisporum are branched in the base with long and unbranched (or slightly branched) ends, where the phialides are formed directly from the central conidiophore hyphae; however, in T. petersenii and T. texanum the conidiophores do not have such terminations and phialides are formed from supporting cells and, with less frequency, directly from the conidiophore axis (Fig. 3e–g). Additionally, T. koningiopsis and T. ovalisporum have a central axis of the conidiophore at an angle close to 90°; however, in T. petersenii and T. texanum this angle is smaller than 90° (Fig. 3g). The phialide arrangement is similar in all four species, but T. koningiopsis, T. ovalisporum, and T. petersenii have lageniform phialides, while in T. texanum they are cylindrical (Fig. 3g). T. texanum has longer and thinner phialides when compared to T. koningiopsis, T. ovalisporum, and T. petersenii. There are also differences on the conidia shape: T. koningiopsis, T. ovalisporum, and T. petersenii have ellipsoidal conidia, while T. texanum presents globose-subglobose conidia (Fig. 3j). Chlamydospores of T. texanum (Fig. 3h, i) are longer than the other three species (in which chlamydospores are globose to subglobose).

Fig. 3
figure 3

Morphological characteristics of Trichoderma texanum. a–c Cultures on PDA, CMD and SNA, respectively, after 7 days of growth at 25 °C. d, e Pustules formed on CMD. f Conidiophore branching pattern. g Phialides pattern. h, i Chlamydospores. j Conidia

Full size image

Trichoderma longifialidicum grouped close to T. guizhouense (Chaverri et al. 2015) and T. velutinum (Bissett et al. 2003) on our phylogenetic analysis (Fig. 1). The growth rate of T. velutinum is smaller than T. longifialidicum at 30 °C on PDA (54–62 mm for the former and 65–68 mm for the latter). In comparison with T. guizhouense, T. longifialidicum growth was similar at 25 °C (45–70 mm on PDA and SNA for T. guizhouense; 67 mm on PDA, 60–63 mm on SNA for T. longifialidicum). T. longifialidicum often shows white or green-colored globose pustules with sterile ends, similar to those of T. velutinum (Fig. 4e) but distinct from T. guizhouense that presents irregular pustules without sterile ends. Moreover, T. longifialidicum and T. velutinum do not exhibit production of soluble pigment in artificial medium, different from T. guizhouense that exhibits yellowish to orange pigment at 30 and 35 °C. The conidiophores of T. longifialidicum are scarce, usually straight, little or not branched; also, formed mainly at the edge of pustules, rarely forming in the aerial mycelia and with three phialides in cross at the end of the conidiophore (Fig. 4e.). T. longifialidicum has a similar conidiophore when compared to T. guizhouense, but only T. velutinum has the main axis of the conidiophore ending in a elongate sterile hair. The phialide shape is a distinguishing feature of T. longifialidicum in comparison to other closely related species due to the elongated and cylindrical phialides (7–21 µm in length × 1.4–2.8 µm wide at the widest point and 1.1–2.3 µm wide at the base). Phialides in T. longifialidicum are found in whorls of two or three phialides (Fig. 4f–h). Nevertheless, T. guizhouense and T. velutinum present ampulliform and lageniform phialides (bottle shape), short and thick-walled (4.0–7.2 µm in length × 2.8–4.5 µm in width for T. velutinum and 4.7–7.5 µm in length × 3–4 µm in width for T. guizhouense). The conidia of T. longifialidicum (Fig. 4k) and T. velutinum are ellipsoidal with similar sizes; however, they differ in shape in comparison to T. guizhouense, which has globose to subglobose conidia. Chlamydospores of T. longifialidicum and T. velutinum are scarce in all culture media and the sizes are similar. Chlamydospores were not observed in T. guizhouense.

Fig. 4
figure 4

Morphological characteristics of Trichoderma longifialidicum. a–c Cultures on PDA, CMD and SNA, respectively, after 7 days of growth at 25 °C. d, e Pustules with sterile terminations. f Long phialides on the edge of the pustule. g, h Conidiophores. i, j Chlamydospores k Conidia

Full size image

Phylogenetic distribution of Trichoderma strains found in the attine ant environment

We found fourteen Trichoderma species in leaf-cutting ant colonies and six species in non-leaf-cutting ant colonies. The two ant groups (leafcutters and non-leafcutters) shared a total of three species (T. afroharzianum, T. koningiopsis and T. virens, Table S1). Regarding the substrate from where we isolated Trichoderma, all the 20 species were found in the fungus garden, three species were observed from the soil samples collected 10 m distant to the ant colonies (T. afroharzianum, T. spirale and T. virens) and only T. spirale was found in soil samples adjacent to the fungus garden (see Table S1 for substrate sources). All Trichoderma species found in soil samples were also found in the fungus garden (for a full comparison of Trichoderma species with ant species and geographic regions, see Fig. S1).

Finally, the 62 strains analyzed in the present work grouped into eight clades which include Trichoderma species previously found in several habitats unrelated with the attine ant environment. The four new phylogenetic taxa (T. attinorum, T. longifialidicum, T. texanum, and LESF 545) clustered into four different clades (Harzianum, Spinulosa, Hypocreanum and Trichoderma, Fig. 1).

Discussion

Fungus-growing attine ant colonies harbor a rich diversity of microorganisms in addition to the mutualistic fungal cultivar. The recent demonstration that such environments house unknown microbial species stimulated our systematic survey for Trichoderma in leaf-cutting ant colonies. In addition, since taxonomic studies investigating filamentous fungi in non-leaf-cutting attine ants are scarce (recent examples are: Masiulionis et al. (2015) and Meirelles et al. (2015a) which focused on Escovopsis), we also included Trichoderma strains isolated from gardens of non-leaf-cutting ants (genera Cyphomyrmex and Trachymyrmex) in our study. Even considering a limited number of isolates examined, this is the first systematic study to investigate Trichoderma species found in colonies of non-leafcutter attine ants.

Trichoderma species are commonly found in colonies of leaf-cutting ants (Augustin et al. 2011). Previous work demonstrated the occurrence of eight Trichoderma species: T. aureoviride, T. hamatum, T. harzianum, T. koningii, T. longibrachiatum, T. pseudokoningii, T. spirale, and T. virens (Fisher et al. 1996; Barbosa 2004; Rodrigues et al. 2005, 2008; Augustin et al. 2011; Rodrigues et al. 2011, 2014). Our results show that attine ant colonies harbor more Trichoderma species than previously reported, since the 62 strains corresponded to 20 species, including three new species described in this study, along with a new phylogenetic taxon.

Among the eight Trichoderma species previously reported in colonies of leaf-cutting ants, four of them were also found in our study (T. hamatum, T. longibrachiatum, T. spirale and T. virens), all isolated from leafcutter ant colonies. The remaining four previously reported species (T. aureoviride, T. harzianum, T. koningii and T. pseudokoningii) were not found in our sample set. It is important to note that previous studies used only classical taxonomic methods (morphological markers) or sequenced the ITS region to identify these isolates (Fisher et al. 1996; Barbosa 2004; Rodrigues et al. 2005, 2008, 2011, 2014). Several studies showed that the morphological species concept presents serious problems when used as the only tool for species identification in the genus Trichoderma (Chaverri et al. 2003a; Druzhinina and Kubicek 2005; Atanasova et al. 2013; Chaverri et al. 2015). On the other hand, using only the phylogenetic species concept for the identification of Trichoderma may eventually fall into the same problem, especially when a molecular marker with insufficient intraspecific variability (such as ITS region) is used alone (Druzhinina and Kubicek 2005; Atanasova et al. 2013; Chaverri et al. 2015). In that sense, it is more plausible to identify a species with a combination of morphological and phylogenetic data, using various molecular markers in a concatenated analysis [Genealogic Concordance Phylogenetic Species Recognition (GCPSR; Taylor et al. 2000; Druzhinina and Kubicek 2005; Jaklitsch 2009; Chaverri and Samuels 2013; Atanasova et al. 2013; Chaverri et al. 2015)]. Following this reasoning, we think it is necessary to sequence additional markers (as tef1 and rpb2) for the previously identified species (T. atroviride, T. harzianum, T. koningii and T. pseudokoningii) to indeed confirm their presence in the attine ant colonies.

Moreover, for a description of new taxon, it is desirable to find several isolates to understand the variability of the proposed taxon regarding its geographic range, substrate preferences, morphology and physiology of the sexual (if existent) and the asexual phases (Atanasova et al. 2013). Also, phylogenetic analysis using various molecular markers (minimum tef1 and rpb2) is essential (Atanasova et al. 2013; Chaverri et al. 2015). Despite the efforts to meet these recommendations, several studies showed that it is not always possible to isolate more than one strain, although this issue did not invalidate the description of several species: Trichoderma taiwanense (Samuels et al. 2006), T. neokoningii, Trichoderma scalesiae (Jaklitsch et al. 2006), Trichoderma aerugineum (Jaklitsch 2009), Trichoderma junci, Trichoderma valdunense (Jaklitsch 2011), Trichoderma caesareum, Trichoderma floccosum, Trichoderma ivoriense, Trichoderma vermipilum (Samuels et al. 2012b) and T. lixii (Chaverri et al. 2015).

The three taxa (LESF 236, 551 and 552) described here as new species have one strain each. Although this issue might hinder the description of the intraspecific variability of these taxa, our results show strong evidence of the distinct morphology and phylogenetic position (based with tef1 and rpb2 markers) that support those taxa as true taxonomic novelties. Also, when considering the biology of attine ants, these insects forage random substrates in different places; thus, it is possible that this foraged material (along with the different microorganisms found in that substrate) may not be collected again. Hence, the collection of a particular Trichoderma species will be influenced by how it is distributed in the environment and where the ants are foraging. Therefore, future works including more sampling of attine ant colonies and also from different substrates that surrounds the ant colonies (soil, leaves, etc.) might increase the chance to obtain more strains of the new species described here.

Fisher et al. (1996) showed the endophytic origin of some Trichoderma species when isolating T. hamatum and T. cf. harzianum from leaves offered to Atta cephalotes colonies. Several other studies with leaf-cutting ants suggested that Trichoderma fungi are associated with the plant substrate foraged by workers (Barbosa 2004; Rodrigues et al. 2008, 2009, 2011). Our results support this hypothesis, since 51 out 62 strains derived from leaf-cutting ants were isolated from plant material present in the fungus gardens. Moreover, T. atroviride, T. eucorticioides, T. hamatum, T. longibrachiatum, T. orientalis, T. spirale and T. viride found on fungus garden samples, have been reported as endophytes on substrates not related to leaf-cutting ants (Overton et al. 2006; Samuels et al. 2006; Bae et al. 2009; Chaverri et al. 2011; Druzhinina et al. 2012; Bailey and Melnick 2013). However, Trichoderma species are also parasites of other fungi (Bailey et al. 2008; John et al. 2010; Chaverri and Samuels 2013); in this sense, they could also enter the fungus garden together with the leaves collected by the ants, possibly as mycoparasites of plant pathogenic fungi. In fact, this route of entry for Trichoderma was not considered in previous studies and future efforts might look for evidences for this hypothesis.

Regarding T. viride, this species was considered to be restricted to the northern hemisphere and found in soil, trunks and decaying wood, with only two strains reported in the tropics, isolated as endophytes on Theobroma gileri (Ecuador) and Theobroma cacao (Brazil) (Lieckfeldt et al. 1999; Jaklitsch et al. 2006; Samuels et al. 2006). The present study identified a T. viride strain (LESF 115) in the fungus garden of Atta sexdens rubropilosa, which indicates this fungus could have been carried by the ants as an endophyte. Moreover, this result supports the hypothesis proposed by Samuels et al. (2006) suggesting this taxon also occurs in the tropics, but probably in less exploited environments, such as the attine ant colonies.

The non-leaf-cutting attine ants collect several substrates to feed their mutualistic fungus (Mehdiabadi and Schultz, 2010). Given the cosmopolitan characteristic of the genus Trichoderma, such substrates are likely sources of Trichoderma spores. In our study, we analyzed a total of six Trichoderma strains isolated from gardens of non-leaf-cutting ants, including Trachymyrmex spp. (n = 4) and Cyphomyrmex wheeleri (n = 2, Table S1). The strains were identified as T. koningiopsis, T. afroharzianum, T. virens, T. attinorum, T. texanum and T. longifialidicum; intriguingly, the three latter strains are new species. The fact of finding new species in gardens of non-leaf-cutting attine ants indicates that the colonies of such insects, less exploited when compared to leaf-cutting ants, might harbor a high richness of Trichoderma species, especially due to the wide range of substrates they collect.

On the other hand, Augustin et al. (2011) and Rodrigues et al. (2014) suggested that the soil could also be an alternative source for diversity of Trichoderma species present in colonies of leaf-cutting ants. For example, Rodrigues et al. (2014) showed that soil samples distant from the ant colonies have several fungal species such as T. spirale, which are also frequently isolated from the fungus gardens. The authors suggested that these fungi would be transported by foragers that are directly in contact with soil. Here, we identified T. spirale (LESF 549 and LESF 543) isolated from the soil adjacent to the chamber that holds the fungus garden of Atta sexdens rubropilosa and Atta bisphaerica (Table S1). Similarly, we identified T. afroharzianum (LESF 228, LESF 229) and T. spirale (LESF 557, LESF 558, LESF 559) isolated from soil 10 m far from Atta bisphaerica; T. virens (LESF 550) and T. spirale (LESF 562) was also isolated from soil 10 m distant from a colony of A. sexdens rubropilosa (Table S1). These three species (T. afroharzianum, T. spirale and T. virens) were also found in fungus garden samples (Table S1), reinforcing that the soil is another source for the diversity of Trichoderma in leaf-cutting ants colonies.

Our phylogenetic analysis also revealed that none of the 62 strains evaluated form a specific group of Trichoderma associated exclusively with attine ant colonies, since they grouped in clades comprehending Trichoderma strains isolated from various sources (Fig. 1; Table S2). Furthermore, we did not observe Trichoderma strains associated with a specific genus of leaf-cutting ants (i.e. the same Trichoderma species can occur in both Atta and Acromyrmex genera, Fig. 1; Table S1). Regarding the three new species isolated from gardens of non-leaf-cutting ants, the most parsimonious hypothesis is to consider that they may have been indirectly brought to the colonies, similarly to other Trichoderma species. In our phylogenetic analysis, the three new phylogenetic species grouped in different parts of the tree; if these species were exclusively associated with attine ants, such association would have originated multiples times over the evolutionary time. However, the fact that only few strains have been isolated indicates that they are not a common occurrence in this environment, which also supports the idea that they were brought in by ants suggesting the transient nature of Trichoderma in attine ant colonies (Poulsen and Currie 2006).

The genus Trichoderma has several ecological roles (Druzhinina and Kubicek 2005; Samuels et al. 2006; Jaklitsch 2009; Atanasova et al. 2013; Chaverri and Samuels 2013; Druzhinina and Kubicek 2013). Because ants feed their mutualistic fungus with the forage plant material, we can infer that some Trichoderma species (those who manage to resist the cleaning process of the ants) could use nutrients found in the fungus garden matrix as well as some species could act as an antagonist of the mutualistic fungus, especially those that are mycotrophic (for example: T. lixii, T. ceramicum, T. parestonicum, T. viride). However, so far there is no evidence that Trichoderma performs any of these roles in the attine ant colonies.

Finally, the presence of three new species (T. attinorum, T. longifialidicum and T. texanum) and a new phylogenetic taxon (LESF 545) in the attine ants’ gardens corroborates recent results indicating the potential of this environment to harbor high microbial diversity. Apparently, the dynamic foraging behavior of these insects (gathering several types of substrates) offers a unique environment for a rich and diverse microbiome. Future exploratory studies in this environment will certainly reveal even more unknown species for science, not only within the genus Trichoderma but for other fungal groups as well.

Taxonomy

Trichoderma attinorum Q.V. Montoya, L.A. Meirelles, P. Chaverri & A. Rodrigues sp. nov. (Fig. 2)

Mycobank: MB814486

Etymology: “attinorum” in relation to the name of the tribe Attini, the first source from where the species was isolated.

Typification: USA. Austin, Texas, Bull Creek Park, GPS 30°22′16″N; 97°47′08″W, Fungus garden, 10, 2006. A. Rodrigues. Ex-type strain LESF 236 (= CBS 139783, = CBMAI 1826). Holotype: CBS H-22180 (dried culture on PDA).

Sequences: ITS (HQ608035), tef1 (KT279039) and rpb2 (KT278971).

Teleomorph unknown.

Culture characters: Colonies showed a radius of 39–41; 25–30 and 5–10 mm on PDA, CMD and SNA, respectively, after 72 h at 25 °C in the dark; at 30 °C colonies exhibited a radius of 30–31, 29–38 and 6–11 mm on PDA, CMD and SNA, respectively. At 35 °C, the radius reached 10–15, 2–7 and 3–10 mm on PDA, CMD and SNA, respectively. The colony fills the plate after 7 days of incubation at 25 and 30 °C on PDA. After 72 h at 25 and 30 °C, concentric rings of alternating green and white colour were observed in cultures maintained on PDA. The conidia pigmentation starts from the inoculum region and continues instead of maintain to the edge of the colony after 48 h on PDA. The presence of yellow soluble pigmentation was observed in CMD and SNA after 48 h of incubation at all three temperatures. Hyphae are hyaline and smooth, usually forming pustules on PDA and CMD, no aerial mycelium on SNA. Pustules are globose to subglobose and cottony, connecting together forming dense masses and measuring 1–3 mm in diameter. Conidiophores are straight, with short branches, matched and in whorls, forming at the edge of pustules and in aerial mycelium. Phialides are formed usually in whorls, paired and rarely isolated, lageniform to cylindrical-shaped, usually straight and rarely curved measuring 5.2–11.2 µm in length, 1.8–2.8 µm in width at the widest point, 1.2–2.3 µm in width at the base. Phialides are formed from supporting cells measuring 4.5–14.6 µm in length × 1.7–3.4 µm wide; phialides rarely form from the central conidiophore hyphae. Conidia subglobose to ovoid, dark green-colored and smooth, measuring 3.2–4.3 × 2.5–3.4 µm. Chlamydospores hyaline and intercalated, measuring 5.3–11.4 × 4.5–6.8 µm, observed after 72 h at 25 °C on CMD and PDA.

Habitat: isolated from fungus garden of Cyphomyrmex wheeleri.

Notes: T. attinorum is closely related to T. sipulosum, T. sinuosum and T. thelephoricola and its most distinctive features are the dense pustules and yellow soluble pigment observed in the three culture media and at all temperatures. Moreover, differences in the shape of the conidiophores are present when compared to T. sipulosum, T. sinuosum and T. thelephoricola (the last one has a Gliocladium-like conidiophore, an even more remarkably-different conidiophore morphology when compared to T. attinorum).

Trichoderma texanum Q.V. Montoya, L.A. Meirelles, P. Chaverri & A. Rodrigues sp. nov. (Fig. 3)

Mycobank: MB814487

Etymology: “texanum” in relation to the state of “Texas” (EUA), from where the species was first isolated.

Typification: USA. Smithville, Texas, Stengl “Lost Pines” Biology Station, GPS 30°05′13″N; 97°10′15″W, Fungus garden, 10, 2006. A. Rodrigues. Ex-type strain LESF 551 (= CBS 139784, = CBMAI 1828). Holotype: CBS H-22182 (dried culture on PDA).

Sequences: ITS (HQ608136), tef1 (KT278988) and rpb2 (KT278920).

Teleomorph unknown.

Culture characters: Colonies showed a radius of 55–56; 45–46 and 8–10 mm on PDA, CMD and SNA, respectively after 72 h at 25 °C in the dark. At 30 °C colonies exhibited a radius of 37–42, 48–59 and 10–23 mm on PDA, CMD and SNA, respectively. No concentric rings or soluble pigmentation was observed. The colony fills the plate after 96 h on PDA and CMD and after 7 days on SNA. The conidia pigmentation starts from the inoculum region to the edge of the colony after 7 days on PDA. Hyphae are hyaline, smooth and sinuous, but not in a spiral. Pustules are small, globoid, cottony in masses, forming from the inoculum center on PDA and from the edge of the plate on CMD. Conidiophores usually straight, but rarely curved or sinuous, forming both from on the edge of the pustules and from aerial hyphae. The conidiophores have branches, usually paired, and alternated, but sometimes in whorls. Phialides are generally paired and whorls of three, forming usually from supporting cells, which measures 2.1–15.5 µm in length × 1.6–3.1 µm in width; or from the conidiophore apex. Phialides are cylindrical-shaped, usually arranged in whorls, isolated or paired, measuring 7.1–13 µm in length, 1.2–2.8 µm in width at the widest point, and 1.2–2.3 µm in width at the base. Conidia are globose to subglobose, dark green, smooth-walled and measuring 1–4 µm in length × 1.8–3.1 µm in width. Chlamydospores hyaline, intercalated, measuring 5.3–15.4 µm in length × 3.4–7.9 µm in width; chlamydospores were observed on CMD and PDA, after 5 days of growth at 25 °C.

Habitat: isolated from fungus garden of Trachymyrmex septentrionalis.

Notes: T. texanum is closely related to T. koningiopsis, T. petersenii and T. ovalisporum. T. texanum morphologically differs from these three species, due to (i) the pustules formation (rare in T. texanum and abundant and dense on the other species); (ii) conidia pigmentation (after seven days in T. texanum and from two to four days for the other species); (iii) the conidia shape (globose-subglobose for T. texanum, and ellipsoid for the other species); (iv) chlamydospores’ size and shape (larger and more elongated on T. texanum).

Trichoderma longifialidicum Q.V. Montoya, L.A. Meirelles, P. Chaverri & A. Rodrigues sp. nov. (Fig. 4)

Mycobank: MB814488

Etymology: “longifialidicum” relating to the long length of the phialides when compared to phylogenetic close species.

Typification: USA. Smithville, Texas, Stengl “Lost Pines” Biology Station, GPS 30°05′13″N; 97°10′15″W, Fungus garden, 10, 2006. A. Rodrigues. Ex-type strain LESF 552 (= CBS 139785, = CBMAI 1827). Holotype: CBS H-22181 (dried culture on PDA).

Sequences: ITS (KT278901), tef1 (KT279020) and rpb2 (KT278955).

Teleomorph unknown.

Culture characters: Colonies showed a radius of 67; 60–63 and 39–43 mm on PDA, CMD and SNA, respectively after 72 h at 25 °C in the dark. At 30 °C colonies exhibited a radius of 65–68, 58–67 and 40–47 mm on PDA, CMD and SNA, respectively. Concentric rings were observed on PDA and SNA. No soluble pigmentation was present. The colony filled all the plate after 96 h on all culture media. The conidia pigmentation starts from the center of the colony, after 7 days on PDA. Hyphae are hyaline, smooth and sinuous. Pustules are globoid, cottony, isolated on CMD and in dense masses on PDA and SNA, forming from the center of the plate on PDA and disperse on CMD and SNA. Conidiophores are usually straight, forming more often on the edge of pustules and fewer in the aerial hyphae. The conidiophores have few branches, usually alternated. Phialides generally paired and in whorls of three, forming usually directly from the central axis of the conidiophore and from supporting cells, which measures 6–16 × 1.7–3.2 µm. Phialides are long, cylindrical, usually paired or isolated, measuring 7–21 µm in length, 1.4–2.8 µm in width at the widest point, and 1.1–2.3 µm in width at the base. Conidia are ellipsoidal, green, smooth-walled and measuring 2.5–3.5 µm long × 1.6–2.4 µm in width. Chlamydospores are globose, hyaline and intercalated, measuring 7–8 × 6–6.2 µm; chlamydospores were observed on CMD and PDA, after 5 days of growth at 25 °C.

Habitat: isolated from fungus garden of Trachymyrmex septentrionalis.

Notes: T. longifialidicum is closely related to T. velutinum and T. guizhouense. T. longifialidicum has few morphological differences compared to the other two species; however, its most distinct morphological feature is the shape and size of its phialides (cylindrical and elongated in T. longifialidicum; short and ampuliform in the other species).