SÃO PAULO STATE UNIVERSITY INSTITUTE OF BIOSCIENCES AT RIO CLARO DEPARTMENT OF GENERAL AND APPLIED BIOLOGY GRADUATE PROGRAM IN BIOLOGICAL SCIENCES (APPLIED MICROBIOLOGY) MARIA JESUS SUTTA MARTIARENA FUNGAL-HELPER-BACTERIA COMMUNITIES IN HIGHER ATTINE ANT GARDENS Rio Claro - SP 2022 SÃO PAULO STATE UNIVERSITY INSTITUTE OF BIOSCIENCES AT RIO CLARO DEPARTMENT OF GENERAL AND APPLIED BIOLOGY GRADUATE PROGRAM IN BIOLOGICAL SCIENCES (APPLIED MICROBIOLOGY) MARIA JESUS SUTTA MARTIARENA FUNGAL-HELPER-BACTERIA COMMUNITIES IN HIGHER ATTINE ANT GARDENS Rio Claro - SP 2022 This thesis is presented to the Institute of Biosciences at Rio Claro, from the São Paulo State University (UNESP), as part of the requirements to obtain the title of Doctor of Philosophy in Biological Sciences (Applied Microbiology). Advisor: Dr. Andre Rodrigues Co-advisor: Dra. Laura Victoria Flórez Patino POTENTIAL IMPACT OF THIS RESEARCH This study was focused on the study of the leafcutter ant´s system. These insects are important for the economy of the American continent because they are pest of many agricultural crops. For science fungus-growing ants is a classic example of symbiosis and evolution. This system owes its evolutionary success to the symbiotic relationships they maintain with microorganisms. The tripartite mutualism between ants, mutualistic fungi and bacteria, was suggested as one of the key symbioses that led to the evolutionary success of this system. Recent studies have shown a great diversity of bacteria living within ant colonies. However, little was known about the symbiotic relationships between these microorganisms and the ants’ cultivars. Here we explored bacteria-fungus interaction inside of the fungus garden of Acromyrmex coronatus and Atta sexdens. Our hypothesis suggested the existence of bacterial communities closely associated with the hyphae of the attine cultivars. Using different approaches like molecular and microscopic techniques, we showed that bacteria can interact with the fungus mutualists on their surface (forming biofilms) and inside of the hyphae (as an endobacteria), comprising the fungus mutualist` microbiota. That supports the idea the microbiome of attine ants can be used as a model system of interkingdom interactions. The results of this study allow us to have a broader idea on the microbial symbiosis within the attine system, as well as the importance of these relationships in the evolutionary success of these social insects. In addition, this study may help researchers to discover new pathways for the biological control of these insects, as well as, for the discovery of new application of this intertwining net of microorganisms. Finally, this study proves that only multidisciplinary approaches like the one we develop in collaboration with L’Institut National de la Recherche Agronomique (INRAe, France) and the University of Copenhagen (Denmark), will help scientists to reveal the importance of the microorganisms and their symbiotic relationships in the evolutionary success of the life. IMPACTO POTENCIAL DA PESQUISA A presente tese de doutorado teve como alvo de estudo o sistema das formigas cortadeiras. Estes insetos são importantes para a economia do continente americano, por serem pragas de diversas culturas agrícolas. Para a Ciência, o sistema das formigas cultivadoras de fungos é um exemplo clássico para estudos de simbiose e evolução. Esse sistema deve seu sucesso evolutivo às relações simbióticas que as formigas mantêm com microrganismos. A interação tripartida entre formigas, fungos mutualísticos e bactérias, por exemplo, foi sugerida como uma das principais simbioses que levaram ao sucesso evolutivo do sistema. Estudos recentes revelaram uma grande diversidade de bactérias nas colônias dessas formigas. No entanto, nada se sabia sobre as relações simbióticas entre esses microrganismos e os cultivares das formigas. Nesta tese de doutorado exploramos a interação bactéria-fungo nos dos jardins de fungo de Acromyrmex coronatus e Atta sexdens. Nossa hipótese sugere a existência de comunidades bacterianas intimamente associadas às hifas dos cultivares das formigas atíneas. Utilizando diferentes abordagens, como técnicas moleculares e microscópicas, mostramos que as bactérias podem interagir com os mutualistas do fungo na superfície (formando biofilme) e dentro da hifa (como uma endobactéria), formando a microbiota do fungo mutualista. Isso apoia a ideia de que o microbioma das formigas atíneas pode ser usado como um sistema modelo de interações entre reinos. Nossos resultados permitiram ter uma ideia mais ampla sobre a simbiose microbiana no sistema das atíneas, bem como a importância destas relações no sucesso evolutivo desses insetos sociais. Além disso, este estudo ajudará a pesquisadores a descobrir novos caminhos para o controle biológico desses insetos, bem como, para a descoberta de novas aplicações dessa rede interconectada de microrganismos. Finalmente, este estudo prova que só abordagens multidisciplinares, como a que desenvolvemos em colaboração com o L'Institut National de la Recherche Agronomique (INRAe, França) e da Universidade de Copenhagen (Dinamarca), ajudarão os cientistas a revelar a importância dos microrganismos e suas relações simbióticas no sucesso evolutivo da vida. To my mother and my best friend, Josefina Sutta Martiarena. “People born with a seed of curiosity, however environmental condition allows few people keep the childish innocence to ask silly things, that makes the world works.” ACKNOWLEDGMENTS Acquiring and creating knowledge is a privileged of science. I must thank São Paulo State University (UNESP) to allow me to go deeper in the scientific path. In the past four years, I had the opportunity to improve my professional skills having UNESP as my working environment. I also thank the Laboratory of Fungal Ecology and Systematics (LESF), because I met people that taught me a lot about science and life. Specially, I would like to thank my advisor Dr. Andre Rodrigues, who supported my ideas and collaborated with the development of my research. Special thanks to my co-advisor Dra Laura Flórez, who helped me to have a more complete vision of science. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001. CAPES funding my visit at Institut national de recherche pour l'agriculture, l'alimentation et l'environnement (INRAE, France). Also thanks to Conselho Nacional de Desenvolvimento Científico e Tecnológico – CNPq (grant # 305269/2018-6) which funded my research, as well to the grant # 2019/03746-0 São Paulo Research Foundation (FAPESP). I would like to thank the Graduate Program in Biological Sciences (Applied Microbiology) and the INRAE for hosting me in my time in France. My stay at INRAe gave the opportunity to meet Dra Aurelie Deveau, a great scientist and person, to whom I am very grateful, because she taught me a lot and helped me to mature my ideas about science. Of course, I have to thank to my mother Josefina Sutta Martiarena, who gave me wings and always believe in me. Also, I must thank my fiancé Dr. Quimi Vidaurre Montoya for his constant professional and personal support. And also, my special thanks to Mrs. Rosa Montoya Castro. Finally, I have to thank my puppy “Rabito”, whose enthusiasm and joy encourage me every day. RESUMO As bactérias desempenham papéis metabólicos fundamentais para a sobrevivência de outros organismos, criando sistemas complexos. Os jardins de fungos das formigas attine (Hymenoptera: Attini: Attina) são exemplos didáticos dessa complexidade, pois esses insetos dependem obrigatoriamente de fungos basidiomicetos para sua alimentação. Embora esse mutualismo seja o cerne das colônias attinas, outros organismos estão envolvidos na manutenção desse microhabitat. Vários estudos mostraram comunidades bacterianas na matriz do jardín de fungo compostas por Pantoea, Klebsiella, Pseudonocardia, Streptomyces e outros gêneros considerados envolvidos na ciclagem de nutrientes e defesa contra fungos antagônicos que ameaçam o jardín. Por outro lado, a função de grupos bacterianos localizados na hifosfera, região que envolve a hifa fúngica e sob sua influência, não foi abordada em previos estudos. Nesta tese, objetivamos avaliar as comunidades bacterianas da hifosfera de Leucoagaricus gongylophorus e seus potenciais efeitos diretos sobre o cultivar das formigas. Isolamos bactérias da superfície da estafila (grupo de hifas inchadas onde os nutrientes do fungo são armazenados) com potenciais funções metabólicas para o desenvolvimento do cultivar e do jardim. Nossos resultados mostraram a presença de comunidades bacterianas como parte da microbiota do fungo. Ensaios in vitro mostraram que as bactérias que habitam a hifosfera podem ser benéficas para o fungo, aumentando a produção da biomassa e auxiliando na mineralização de compostos. Análise transcriptômica de L. gongylophorus na presença de Staphylococcus sp. mostraram vias antimicrobianas down-regulated permitindo o estabelecimento bacteriano nas hifas, bem como transcritos up-regualted relacionados à síntese de lipídios e carboidratos. Além disso, usando técnicas moleculares e microscópicas mostramos evidências de bactérias dentro da hifa de L. gongylophorus. Implicações ecológicas dessas bactérias para o fungo e para as formigas são discutidas. Nossos resultados coletivamente, mostram bactérias na superfície da hifa, bem como dentro da hifa, que constituem a microbiota fúngica que sustenta a vida microbiana no jardim de fungos. Palavras-chave: Bactéria, Leucoagaricus gongylophorus, formigas attine, interação. ABSTRACT Bacteria play fundamental metabolic roles for the survival of other organisms, creating complex systems. The fungus gardens of attine ants (Hymenoptera: Attini: Attina) are textbook examples of such complexity, as these insects obligatory depend on basidiomycetous fungi for food. Although this mutualism is the core of attine colonies, other organisms are involved in the maintenance of this microhabitat. Several studies showed bacterial communities in the fungus gardens matrix composed of Pantoea, Klebsiella, Pseudonocardia, Streptomyces, and other genera thought to be involved in nutrient cycling and defence against antagonistic fungi that threaten the gardens. In contrast, the function of bacterial groups located in the hyphosphere, the region surrounding the fungal hypha and under their influence, were not tackled in any study. In this thesis, we aimed to assess bacterial communities on and inside of the hyphosphere of Leucoagaricus gongylophorus and their direct potential effects on this ant fungal cultivar. We isolated bacteria from the staphyla surface (swollen hypha where fungal nutrients are stored) with putative metabolic functions for the development of the fungus cultivar and the garden. Our findings showed the presence of bacterial communities as part of the fungus microbiota. In vitro assays showed that bacteria inhabiting the hyphosphere can be beneficial for the fungus, increasing biomass production and helping in compound mineralization. Transcriptomics analysis of L. gongylophorus in presence of Staphylococcus sp. showed down- regulated antimicrobial pathways allowing the bacterial establishment on the hyphae, as well as upregulated transcripts related to the synthesis of lipids and carbohydrates. In addition, we showed evidence of bacteria inside L. gongylophorus hypha using molecular and microscopic techniques. Ecological implications of these bacteria for the fungus and the ants are discussed. Collectively, bacteria on the hypha surface as well as inside the hypha comprise the fungal microbiota that support microbial life in the fungus garden. Keywords: Bacteria, Leucoagaricus gongylophorus, attine ants, interaction. TABLE OF CONTENTS THESIS OVERVIEW .............................................................................................................. 14 Chapter I ................................................................................................................................... 16 The hyphosphere of leaf-cutting ant cultivars is enriched in helper bacteria .......................... 17 INTRODUCTION ............................................................................................................... 18 MATERIAL AND METHODS ........................................................................................... 20 Bacterial identification ..................................................................................................... 21 Screening of bacterial effects on the fungal cultivar development .................................. 23 Gongylidium size ......................................................................................................... 24 Biomass ........................................................................................................................ 24 Biofilm-like structure formation ...................................................................................... 25 Screening for putative metabolic functions beneficial for garden development ............. 26 Preparation of bacterial inoculum ................................................................................ 26 Inorganic phosphate solubilization .............................................................................. 27 Siderophore production ................................................................................................ 27 Cellulose degradation ................................................................................................... 28 Chitinase production .................................................................................................... 28 Bacteria pairwise interactions .......................................................................................... 29 In vitro bacterial antagonism against Escovopsis ............................................................ 30 Data visualization and statistical analyses ....................................................................... 30 RESULTS ............................................................................................................................ 31 Bacterial isolation and identification ............................................................................... 31 Screening of bacterial effects on the growth of Leucoagaricus gongylophorus ............. 32 Staphylae production ................................................................................................... 32 Gongylidium size ......................................................................................................... 32 Biomass production ..................................................................................................... 33 Biofilm-like structure formation ...................................................................................... 34 Screening for putative metabolic functions for garden development .............................. 34 Bacteria pairwise interactions .......................................................................................... 35 In vitro bacterial antagonism against Escovopsis ............................................................ 36 DISCUSSION ...................................................................................................................... 37 REFERENCES .................................................................................................................... 42 SUPLEMENTAL MATERIAL ............................................................................................... 52 Chapter II ................................................................................................................................. 59 Pairwise transcriptomic analysis reveals signatures of fungal growth promotion and biofilm formation by bacteria in the hyphosphere of Leucoagaricus gongylophorus .......................... 60 INTRODUCTION ............................................................................................................... 60 METHODS .......................................................................................................................... 61 Microbial cultures, growth conditions and inoculum preparation ................................... 61 Bacteria-fungi co-cultures ................................................................................................ 62 Biomass assessment ..................................................................................................... 62 Biofilm-like formation ................................................................................................. 63 RNA extraction ............................................................................................................ 63 Preprocessing reads and de novo assembly ................................................................. 64 Differential transcript expression analysis and functional annotation ......................... 64 RESULTS ............................................................................................................................ 65 Biomass assessment ......................................................................................................... 65 Biofilm-like structures ..................................................................................................... 66 Read preprocessing, de novo assembly, and functional annotation ................................. 66 Differential transcript expression analysis and functional annotation ............................. 67 DISCUSSION ...................................................................................................................... 71 REFERENCES .................................................................................................................... 73 Chapter III ................................................................................................................................ 75 A putative endohyphal bacterium of leaf-cutting ant fungal cultivars .................................... 76 INTRODUCTION ............................................................................................................... 76 METHODS .......................................................................................................................... 77 Fungal isolation ................................................................................................................ 77 DNA extraction ................................................................................................................ 78 PCR amplification and phylogenetic analysis ................................................................. 78 Fluorescence in situ hybridization (FISH) ....................................................................... 79 RESULTS ............................................................................................................................ 79 DISCUSSION ...................................................................................................................... 81 REFERENCES .................................................................................................................... 84 OVERALL THESIS CONCLUSION ...................................................................................... 85 14 THESIS OVERVIEW Bacteria are ubiquitous and play fundamental roles in the survival of almost all organisms. In complex systems bacteria can establish multidirectional interactions with different organisms as it occurs in the fungus gardens of fungus-farming ants (Hymenoptera: Formicidae: Attini: Attina, “the attines”). Considered the most derived fungiculture in the subtribe Attina (QUINLAN; CHERRETT, 1979), fungus gardens of higher attine ants comprise the obligatory symbiosis with mutualistic fungi (either Leucoagaricus gongylophorus or an unidentified Leucoagaricus species) as the main component (MARTIN; STADLER MARTIN, 1970), and a microbial community consisting of bacteria, filamentous fungi, yeasts, and other unknown microbes (CURRIE, 2001). Ants supply the fungus with plant material, and in exchange, they obtain from the cultivar nutrients, stored in a special fungal structure named “gongylidium” (POWELL ROY, 1984, Figure 1). While the symbiosis between the fungal cultivar and the ants is the core of the garden establishment and maintenance; other fundamental interspecific relationships take place in this habitat. Bacterial-fungal interactions are highly recognized in the attine ant colonies, especially those involving Actinobacteria in the genera Streptomyces and Pseudonocardia, which defend ant nests against antagonistic fungi, such as Escovopsis (LITTLE et al., 2006). Several studies focused in the fungus garden matrix, reported a great diversity of bacterial communities, such as Burkholderia, Klebsiella, Pantoea, and Pseudomonas (AYLWARD et al., 2012; BARCOTO et al., 2020; PINTO-TOMÁS et al., 2009). Nevertheless, no studies focused on the bacterial communities that are in direct contact with the cultivar hyphosphere, the surrounding area of the fungal hypha as well as its interior. In model systems like mycorrhiza-plants, lichens, corals and animal guts, bacteria use the hyphosphere as a source of nutrients, either by living on the surface or inside the hypha (WARGO; HOGAN, 2006). In exchange, specific bacteria influence spore germination, hyphal 15 growth, genetic regulation, nutrition, health, survival and virulence of the fungus (BAREA et al., 2005). Thus, bacteria and fungi can establish tight mutualistic relationships, which are often determinant for the survival of both organisms. Here we aimed to assess the bacteria present in the hyphosphere of the fungal cultivar and understand their putative roles. We assessed the putative metabolic functions of these bacterial groups, how they interact with the fungus in the hyphosphere, as well as determine the presence of intracellular bacterial groups in the hypha of the fungus cultivar. We structured this thesis in three chapters: Chapter I - “The hyphosphere of leaf-cutting ant cultivars is enriched in helper bacteria”, Chapter II - “Pairwise transcriptomic analysis reveals signatures of fungal growth promotion and biofilm formation by bacteria in the hyphosphere of Leucoagaricus gongylophorus”, and Chapter III - “Leafcutter ants’ story: In the ants’ colony has a garden, in the garden has a fungus, and in the fungus more life”. Figure 1. Mutualism between attines ants and their fungi. Ants cultivate fungi for food in structures named “fungus gardens”. Based on published data we suggest the term “Fungal- helper” for the bacterial community in the hyphosphere of higher attine ant gardens. The surface of swollen hyphae (gongylidium) is a nutrient resource for bacterial nutrition; bacterial communities may collaborate for the maintenance of healthy garden. 16 Chapter I (Article published in Microbial Ecology DOI: 10.1007/s00248-023-02187-w) 17 The hyphosphere of leaf-cutting ant cultivars is enriched in helper bacteria Martiarena MJS1, Deveau A2, Montoya QV1, Flórez LV3, and Rodrigues A1* 1 Department of General and Applied Biology, São Paulo State University (UNESP), Rio Claro, Brazil. 2 Université de Lorraine, INRAE, UMR IAM, 54280 Champenoux, France 3 Department of Plant and Environmental Sciences, University of Copenhagen, Denmark. *Corresponding author: São Paulo State University (UNESP), Department of General and Applied Biology Avenida 24-A, 1515, Bela Vista, Rio Claro, SP, Brazil, Zipcode: 13.506-900 andre.rodrigues@unesp.br (orcid: 0000-0002-4164-9362) ABSTRACT Bacteria can live in a variety of interkingdom communities playing key ecological roles. The microbiomes of leaf-cutting attine ant colonies are a remarkable example of such communities, as they support ants’ metabolic processes and the maintenance of ant fungus gardens. Currently, studies are focused on the bacterial community of the whole fungus garden, without discerning bacterial groups associated with the nutrient storage structures (gongylidia) of ant fungal cultivars. Here we studied bacteria isolated from the gongylidia surface of the mutualistic fungus of Atta sexdens and Acromyrmex coronatus, to assess whether the bacterial community influences the biology of the fungus. A total of 10 bacterial strains were isolated from gongylidia (Bacillus sp., Niallia sp., Lysinibacillus sp., Pantoea sp., Staphylococcus sp., Paenibacillus sp., and one Actinobacteria). Some bacterial isolates increased gongylidia production and fungal biomass while others had inhibitory effects. Eight bacterial strains were confirmed to form biofilm-like structures on the fungal cultivar hyphae. They also showed auxiliary metabolic functions useful for the development of the fungal garden such as phosphate solubilization, siderophore production, cellulose and chitin degradation, and antifungal activity against antagonists of the fungal cultivar. Bacteria-bacteria interaction assays revealed heterogeneous behaviors including synergism and competition, which might contribute to regulate the community structure inside the garden. Our results suggest that bacteria and the ant fungal cultivar interact directly, across a continuum of positive and negative interactions within the community. These complex relationships could ultimately contribute to the stability of the ant–fungus mutualism. Keywords: Bacteria-fungus interaction, Helper bacteria, Biofilm, Attine ants. 18 INTRODUCTION Bacteria very often share habitats with fungi and establish complex interkingdom interactions. Fungus-growing ants (Formicidae: Myrmycinae: Attina, “the attines”) and their fungal cultivar are an extraordinary example of such complex communities [1]. Leaf-cutting ants are considered a textbook case of symbiosis, as they maintain an obligatory mutualism with the basidiomycete fungus Leucoagaricus gongylophorus (Basidiomycota: Agaricales), which is cultivated for food [2]. The peculiarity that makes ants dependent on L. gongylophorus is a structure named gongylidium, a hyphal swelling at the tip of the fungal filament. The fungus stores in the gongylidium essential amino acids, lipids, carbohydrates, and proteins necessary for the ants’ nutrition [3]. In turn, ants collect plant material like leaf fragments, seeds, and flowers, as a nutritional fungal substrate. The fungal cultivar grows on this material as an intertwining net forming the “fungus garden” [4]. In addition to provide substrate, the ants propagate the fungus, which lost the ability to grow without the ants’ care [1]. Although the ant-fungus mutualism is the core of this association, many other microorganisms are present in the fungus garden such as bacteria, filamentous fungi, and yeasts [5]. The fungus garden is thus a habitat that supports an intricate microbial system where plant material is processed as a substrate by the fungus. Although ant fungal cultivars have the metabolic capacities for the degradation of plant material [4, 6], ants likely benefit from the presence of other microbes that enhance the production of their cultivar, similar to humans using “biostimulant microorganisms” to enhance the production of agricultural crops [7]. Several studies carried out with the garden matrix and ants describe the presence of diverse bacteria such as Streptomyces and Pseudonocardia, which support the defense against several fungal antagonists, including Escovopsis, an opportunistic fungus of the garden [8, 9]. Other bacteria, like Klebsiella and Burkholderia, have functions 19 such as nitrogen fixation and production of antifungal compounds [10, 11]. Also, metagenomic and metaproteomic analyses demonstrated that the fungus garden bacteriome might be able to provide essential amino acids to the mutualistic fungus [12]. These studies analyzed the garden bacterial community as a whole, without discerning bacterial groups associated to the fungal substrate or those physically attached to the fungus. They also did not consider those specifically associated to gongylidia (a nutritional storage structure, rich in lipids, carbohydrates and enzymes) vs. regular hyphae. The hyphosphere, i.e., the region surrounding the fungal hypha and under its metabolic influence, is a suitable habitat for hosting bacterial communities because it is rich in specific carbohydrates, proteins, lipids, and water [13]. Bacteria can use these conditions for establishing as biofilms (conferring physical and metabolic protection) or use hyphae as a means of transportation (“hyphal highways”), taking advantage of water film surrounding the hyphae [14]. These bacteria could influence the development of the fungus by improving nutrient availability and protecting it against competitors, ultimately enhancing fungal cultivar growth [15]. Concerning the fungus garden, there is scarce information on which bacterial groups colonize the hyphosphere of the fungus cultivar and whether these could promote its growth. We hypothesize that the hyphosphere of gongylidia is colonized by specific bacterial communities whose activities are beneficial to the fungus. To test this hypothesis, we isolated bacteria from staphyla (groups of gongylidia) of L. gongylophorus from three ant gardens and characterized the bacterial isolates at the taxonomic and functional levels using in vitro tests. Several potential bacterial contributions were tested: promotion of fungal growth and staphyla production, antagonism towards pathogens of the fungal garden, ability to degrade organic and inorganic matter, and biofilm-like structures formation on hyphae. Finally, we tested whether bacteria isolated from the same garden were more or less likely to be antagonistic to each other. 20 Collectively, the data suggest that potentially beneficial bacteria are present in the hyphosphere of fungus cultivar staphyla and these bacteria bear partially complementary beneficial activities. MATERIAL AND METHODS Sample collection and microbial isolation We sampled two Atta sexdens gardens, one from a colony maintained in laboratory conditions (further referred as laboratory fungus garden) and one collected from field; and one Acromyrmex coronatus garden also collected in the field from the São Paulo State University (UNESP) campus, Rio Claro, Brazil. After sampling, gardens were transported to the laboratory. To obtain pure cultures of the fungal mutualist, three fragments of approximately 1 cm2 of the fungus gardens were grown on Potato Dextrose Agar medium (PDA, NEOGEN® Culture Media, MI, USA) supplemented with 150 µg mL-1 of chloramphenicol (Sigma-Aldrich, MI, USA). After incubation at 25 °C, for 10 to 14 days, the fragments that remained uncontaminated by other fungi were transferred to new PDA plates. Axenic cultures of mutualistic fungi were preserved by replicating the cultures every 20 days on PDA medium. To isolate bacteria strictly associated with the mutualistic fungus, 500 staphylae were collected from each of the three gardens and washed with 1.7% NaCl solution for 1 minute to release bacteria that were not bound to the staphylae. Then, staphylae were washed three times with 0.8% NaCl and transferred into flasks containing 50 mL of Brain Heart Infusion broth (BHI, NEOGEN® Culture Media, MI, USA), supplemented with 3% methionine and 2% thioglycolate. Flasks were incubated in the dark at 25 °C for six days. For each garden, we set up 10 flasks (each containing 50 staphylae), totalizing 30 flasks and 1500 staphylae. Every 48 h an aliquot of 100 µL was removed and plated on BHI agar medium, totaling three inoculated plates per flask. The plates were incubated for 48 h at 25 °C. All the isolated bacteria were 21 purified by replicating cultures on a new BHI agar medium. Bacterial colonies were differentiated based on macroscopic features of the colony (i.e. color and shape). All the different bacterial isolates were stored in 30% glycerol at -20 °C. Bacterial identification Identification of bacteria isolated from the hyphosphere of fungal cultivar was carried out via Sanger sequencing. DNA extraction from the bacterial cultures was performed following the method described by Alvarado et al. [16]. The extracted DNA was quantified using a Nanodrop (Thermo Fischer, MA, USA) spectrophotometer. Two molecular markers, the 16S rRNA and the RNA polymerase subunit beta (rpoB) genes, were amplified using primers pairs: 8F/1544R (5’AGAGTTTGATCCTGGCTCAG3’ and 5’AGAAAGGAGGTGATCCAGCC3’) [17, 18]; and rpoB-F/R (5’ATCGAAACGCCTGAAGGTCCAAACAT3’ and 5’ACACCCTTGTTACCGTGACGACC3’), [19]. The amplification of these sequences was carried out in a final volume of 25 µL containing: 16.8 µL of ultrapure water, 2.5 µL of buffer (10x), 1.5 µL of MgCl2 (50 mM), 1 µL of BSA (1 mg mL-1), 1 µL of deoxynucleotide triphosphates (1 mM), 1 µL of primers (10 mM each), 0.2 µL of Taq polymerase (5 U µL-1) and 1 µL of DNA template (50 ng µL-1). The amplification conditions for the 16S rRNA gene were: initial denaturation at 94 °C for 5 min, followed by 35 cycles at: 94 °C for 45 s, 58 °C for 45 s and 72 °C for 1 min and a final extension step at 72 °C for 5 min. For RNA polymerase subunit beta (rpoB), amplification conditions were: initial denaturation at 95 °C for 3 min, followed by 35 cycles at: 95 °C for 20 s, 55 °C for 30s and 72 °C for 90 s and a final extension step at 72 °C for 5 min. Amplicons were purified using the PureLink® Quick Gel Extraction and PCR Purification Combo Kit (Invitrogen®, MA, USA), following the manufacturer's recommendations. Sequences (forward and reverse) were generated using the Sanger method in an ABI3500 capillary sequencer (Life Technologies, CA, USA), and the consensus 22 sequences were assembled in BioEdit v. 7.1.3 [20]. The obtained sequences were compared with homologous sequences deposited in the NCBI-GenBank database using the Blast tool (https://pubmed.ncbi.nlm.nih.gov/2231712/), to identify strains at the genus level for all bacterial isolates. We carried out phylogenetic analyses to confirm the position of all strains. To do so, we performed a multilocus analysis combining the 16S rRNA and rpoB genes for the strains that belong to the genera Bacillus, Pantoea, and Paenibacillus, and a single gene analysis using the 16S for those in the Lysinibacillus, Niallia and Staphylococcus genera. Multilocus phylogenetic analyses were constructed using a final data set for Bacillus with a total length of 2351 bp [35 16S sequences (1392 bp), and 30 rpoB sequences (959 bp)]. The final data set of Paenibacillus had a total length of 2457 bp [33 16S sequences (1372 bp), and 22 rpoB sequences (1085 bp)], and in the case of Pantoea it was 2138 bp long [23 16S sequences (1308 bp), and 21 rpoB sequences (830 bp)]. The final data set of Lysinibacillus, Niallia and Staphylococcus contained a total of 19 (1367 bp), 13 (1406 bp), and 24 (1533 bp) 16S sequences, respectively. In all cases, the sequences were aligned separately for each gene using MAFFT v.7 [21]. Then the non- informative regions were manually removed from the initial and terminal regions of each alignment, then nucleotide substitution models for each alignment were calculated in jModelTest 2 [23], using the Akaike Information Criterion (AIC) with 95% confidence intervals. In the case of Bacillus, Pantoea, and Paenibacillus, the sequences were concatenated using Winclada v.1.00.08 [22]. The final phylogenetic trees were reconstructed using Bayesian Inference (BI) in MrBayes v.3.2.2 [24] and Maximum likelihood (ML) in RAxML v.8 [25]. For BI analysis, we carried out two separate runs (each consisting of three hot chains and one cold chain) using the GTR + I +G model, and two million generations of the Markov Chain Monte Carlo (MCMC) were enough to reach convergence (standard deviation (SD) of split frequencies < 0.01). The first 23 25% of the trees were discarded as burn-in to generate the best BI tree. For the ML analysis, we estimated 1000 independent trees and performed 1000 bootstrap replicates using the same model as for the BI analysis. For the multilocus analysis we performed the BI and ML analyses considering the GTR + I +G model for each partition of the concatenated data, independently. The final tree was visualized in FigTree v.1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/) and edited using Adobe Illustrator CC v.17.1. Screening of bacterial effects on the fungal cultivar development Staphylae production Since staphylae are fundamental structures for the ant-fungus mutualism, we evaluated whether bacteria increase or decrease staphylae production by the cultivar. We carried out dual-culture assays, where 100 µL of fungal suspension (30 mg mL-1) from 20-day-old cultures were spread on plates containing PDA medium, supplemented with 2% peptone and 5% yeast extract (PDAPY). Plates were incubated at 25 °C for 5 days. Then, semi-liquid media were added on the plates to compare two conditions: (i) 6 mL of Potato Dextrose Broth medium supplemented with 5% agar, 2% peptone and 5% yeast extract (PDBPY), containing 200 µL of bacterial suspension 103 CFU (colony forming units) per mL; and (ii) 6 mL of PDBPY, containing 200 µL of 0.8% NaCl, as the control condition. Each treatment included six plates per bacterial isolate, which were incubated for additional 15 days at 25 °C. To measure the effect in the production of staphylae, we calculated the percentage of staphylae coverage per plate, by dividing it in 52 quadrants of 81 mm2. Quadrants filled with staphylae were then counted. Then, data were transformed to be expressed as a relative proportion from 0 to 1, where 1 is equivalent to the largest percentage of coverage. 24 Gongylidium size We sampled staphylae from each condition of the previous assay (Staphylae production) to 0.1% Tween 20 solution. Then, 20 µL of this suspension were transferred to mounting slides to examine the structures under a light microscope (DM2500 LED, Leica, Germany). We also directly picked staphylae from the fungus garden of a laboratory colony of Atta sexdens, and from fungus gardens of Atta sexdens and Acromyrmex coronatus collected from the field. Staphylae structures of each condition were photographed, and the area measured using LAS EZ v.4.0 (Leica Application Suite). We measured the area (2D) covered by 30 randomly chosen gongylidia, used as a proxy for size for each condition, hereafter referred as gongylidium size. Biomass We evaluated if bacteria could influence the fungal cultivar growth by measuring biomass production in dual-culture. Two conditions were compared: (i) 10.5 mg of fungal biomass from 20 day-old cultures inoculated in 40 mL of PDB medium containing 200 µL of axenic cultures in 0.8% NaCl; and (ii) 10.5 mg of fungal biomass inoculated in 40 mL of PDB medium containing 200 µL of bacterial suspension in different concentrations: 102, 103, 104, 105, and 106 CFU mL-1. Both treatments were incubated at 25 °C for 15 days at 120 rpm. Each treatment was carried out with six replicates. Then, fungal cultures were filtered using filter paper and the fungal biomass dried for 1 hour at 100 °C to obtain the dry mass. The dry biomass was weighed and both treatments were compared. For the bacterium that showed the highest fungal biomass stimulation (Bac10), we tested whether the biomass increases corresponds to an increment in fungal biomass and is not due to bacterial biomass, we measured the ergosterol concentration of axenic and the bacteria-fungi cocultures. Briefly, biomass of an axenic cultures and co-cultures previously grown for 8 days (six replicates per condition), were harvested by vacuum aspiration, and conserved in liquid 25 nitrogen. Samples were freeze-dried and weighed, then grounded to a fine powder in liquid nitrogen. Subsequently, the ergosterol was extracted following the method of He et al. [26], with some modification. Briefly, in a glass flask we added the fungal powder, glass beads and 1.5 mL of methanol. Flasks were shaked for 60 min at 37 °C, then the suspensions were filtered with 0.45µm filters and stored at -20 °C. The determination of ergosterol by HPLC-UV-DAD was performed using a UFLC chain (Shimadzu©) equipped with a Kinetex C18 column (100 mm x 4.6 mm x 2.6 µm) itself equipped with a pre-column (Phenomenex). The isocratic mobile phase consisted of 95% methanol with a flow rate of 1.6 mL/min. A range of commercial ergosterol standard (Sigma-Aldrich) was used at the following concentrations: 25, 50, 75, 100, 150, 175 and 200 µg/mL to build standard curve. Manual integration of the peaks at 282nm (retention time 4.2 min) was performed. Biofilm-like structure formation The biofilm-forming ability on fungal hyphae was tested for all bacterial isolates, following the method described by Guennoc et al. [27]. Bacterial suspensions were prepared as follows: for all bacteria (except for Actinobacteria), few colonies were picked from plate cultures and inoculated in tubes containing 6 mL of BHI and incubated for 24 h at 25 °C, at 120 rpm. Later, cultures were centrifugated at 7000 rpm at 4 °C for 10 min, then the supernatant was discharged, and the pellet was resuspended in 10 mL of 0.1 M potassium phosphate buffer (PPB, in g L-1: 25 KH2PO4 and 2.78 K2HPO4, pH 5.8). This procedure was repeated three times. Bacterial suspensions were adjusted to 108 CFU mL-1. For the Actinobacteria isolate, colonies were transferred to a new BHI agar plate and incubated for 48 h at 25 °C, then using a Drigalski rod, the colony surface was scrapped and spores collected in PPB to reach 108 spores mL-1. Then, 10 mL of bacterial suspension were inoculated in Petri plates (60 x 15 mm). 26 To evaluate biofilm directly on the fungal colonies: one staphylae was transferred to the center of a plate containing PDAPY and covered with a sterile cellophane membrane. Plates were incubated for 10 d at 25 °C. Using a scalpel, a square (4 cm2) was cut off from the plate and the cellophane with the fungal colony carefully removed. The fungal colony was transferred to the Petri dish containing the bacterial suspension, and the cellophane was removed. The dual- culture was incubated for 24 h at 25 °C at 60 rpm. After incubation, the co-culture was washed with PPB, and then 5 mL of 17 g L-1 NaCl was added and shaken for 1 min to remove non biofilm-forming bacteria. The fungal culture was transferred to a new plate containing 5 mL of sterile 0.1 M PPB, shaken for 2 min, and transferred to another plate containing 0.1 M PPB. Then, samples were fixed in osmium tetroxide vapor for 72 h, dehydrated by a gradient concentration of acetone (50, 75, 90, 95 and 100%) and dried to critical point using liquid CO2 (Balzers CPD030). Finally, the dried samples were sputtered with gold (Balzers SCD050) and examined under a scanning electron microscope (TM3000, Hitachi). Screening for putative metabolic functions beneficial for garden development Preparation of bacterial inoculum Stored bacteria were grown on BHI agar plates at 25 °C for 24 h (except for Actinobacteria that grew for 48 h). Then, for all bacteria (except for Actinobacteria), few colonies were picked and inoculated in tubes containing 6 mL of BHI and incubated for 24 h at 25 °C, and 120 rpm. Cultures were centrifuged at 7000 rpm at 4 °C for 10 min, then the supernatant was discharged, and the pellet was resuspended in 10 mL of 0.8% NaCl. This procedure was repeated three times. Bacterial suspensions were adjusted to reach optical density (OD) A600nm=0.6. Since the Actinobacteria isolate is filamentous, OD is not appropriate to produce standardized inoculum. Instead, standardized spore suspensions were used as inoculum for the Actinobacteria. For this 27 purpose, Actinobacteria colonies were inoculated in a new BHI agar plate and incubated for 48 h at 25 °C, then using a Drigalski rod, the colony surface was scrapped, and spores were collected in 0.8% NaCl. This suspension was adjusted to 108 spores mL-1. Inorganic phosphate solubilization Bacterial cultures were tested for in vitro inorganic phosphate solubilization potential. Briefly, 5 µL of each bacterial suspension with OD 0.6 (108 spores mL-1 for Actinobacteria) were inoculated to the center of plates containing Tricalcium phosphate growth agar medium (TCP) (in g L-1: glucose 10, Ca3(PO4)2 5, MgCl2.6H2O 5, MgSO4.7H2O 0.25, KCl 0.2, (NH4)2SO4 0.1 and agar 20) [28]. Plates were incubated at 25 °C for 7 days, and three replicates were made for each bacterial morphotype. Inorganic phosphate solubilization was evidenced by the formation of a halo around the colonies, whose diameter was measured. Siderophore production To test if the bacterial isolates can capture iron (Fe+3) through siderophore production, we used the universal method of Schwyn and Neilands [29]. Briefly, 5 µL of each bacterial culture at OD 0.6 (for Actinobacteria at 108 spores mL-1) were transferred to the center of plates containing Chrome Azurol S agar media (CAS) and incubated at 25 °C for 7 days. Three plates were made for each bacterial morphotype. Siderophore production was evidenced by the formation of a halo around the colonies, whose diameter was measured. A liquid assay was performed when bacteria were not able to grow directly on solid CAS medium. Bacterial cultures were incubated for 7 days in M9 liquid medium (g L-1: Na2HPO4 1.2 10-5, KH2HPO4 6 10-6, NaCl 1 10-6, NH4Cl 2 10-6, supplemented with 20 mL glucose 20%, 2 mL 1M MgSO4, 200 µL 1M CaCl2) without Fe+3 source. Then, the cultures were centrifuged, and the supernatant was mixed with 1:1 CAS liquid medium and incubated for 1 h. The 28 absorbance of the solution was measured at 655 nm with a Tecan spectrophotometer. The presence of siderophores was revealed by the reduction in absorbance. Cellulose degradation Since fungus gardens are mainly composed by plant material, we tested if bacteria isolates can help in the degradation of cellulose. Briefly, 5 µL of suspension of each bacterial culture at OD 0.6 (for Actinobacteria at 108 spores mL-1) was inoculated at three equidistant points on plates containing Carboxy Methyl Cellulose agar medium (CMC) (g L-1: K2HPO4 1, (NH4)2SO4 1, MgSO4.7H2O 0.5, NaCl 0.5, carboxy methyl cellulose 5, agar 20, adjusted pH to 5). Cultures were incubated at 25 °C for 7 days and two plates were used for each bacterial morphotype. After incubation, 3 mL of 1% congo red solution was added and incubated for 40 min at room temperature. Afterwards, the solution was removed and 6 mL of 1M NaCl was added and incubated for another 15 min. Cellulose degradation was evidenced by the formation of a halo around the colonies, whose diameter was measured. Chitinase production To assess if bacteria could help in the garden defense against antagonistic fungi, or to use the mutualistic fungus as carbon source, we tested for chitinase production in vitro. Briefly, 5 µL of suspension of each bacterial culture with OD 0.6 (for Actinobacteria at 108 spores mL-1) was inoculated to the center of plates containing minimum mineral agar (in g L-1: NaCl 5, KH2PO4 1, yeast extract 0.1, and agar 20) supplemented with 0.2% colloidal chitin [30]. Three replicates were made for each bacterial morphotype and incubated at 25 °C for 7 days. Chitinase production was evidenced by the formation of a halo around the colonies, whose diameter was measured. 29 Bacteria pairwise interactions To test the level of antagonism or potential cooperation between bacterial isolates, we performed a test of paired interaction. Thus, each bacterial isolate was tested against each other in plates containing PDAPY medium, following the method by Gonzalo et al. (2020) [31]. Briefly, 7 µL of suspension of each bacterial culture at OD 0.6 (for Actinobacteria at 108 spores mL-1) were streaked in a line 6 cm long at the center of the plate, corresponding to the “donor” culture. After 48 h, 6 µL of bacterial suspension of three different isolates (“recipient” cultures) at OD 0.6 were streaked (4.5 cm long) 0.5 cm apart from the donor culture (Fig. 5b). For the control, 7 µL of suspension of each bacterial culture at OD 0.6 were streaked in a line 6 cm long at the center of the plate without recipient cultures. All combinations were tested using three replicates. Plates were incubated for 7 days at 25 °C. Then, plates were evaluated by assessing three possible effects in the recipient cultures as described in Gonzalo et al. [31]: (i) growth inhibition (GI, reduction of the length in mm of the streak of the recipient in comparison with length of the inoculation streak); (ii) growth stimulation; overgrowth of the recipient bacteria on the donor bacteria (scored as +1, when overgrowth was visible), and (iii) no interaction (no inhibition and no overgrowth between them). The level of GI was scored as follows: 0: GI<2.5 mm; -1: 2.60.05). Gongylidium size We then tested if, in addition to changing the number of staphylae, the bacteria had an effect on the gongylidium size, using light microscopy. The gongylidia size from axenic cultures varied between 1488 µm2 ± 334.7 for lab-grown A. sexdens, 1123 µm2 ± 370.3 for A. sexdens 33 from field and 881 µm2 ± 124.4 for A. coronatus from field (Table S2). When in co-culture with Lysinibacillus sp. (Bac2), Pantoea sp. (BacB) and Actinobacteria (BacG), the fungal growth was inhibited (Fig. 1). While the presence of Pantoea sp. (Bac1) Paenibacillus sp. (Bac 11) and Bacillus spp. (BacA and BacC) had a negative effect on the gongylidia size (Dunnet test, p<0.05, Table S2). The other bacteria did not have a significant effect on the gongylidia size (Dunnet test, p >0.05, Table S2). When comparing gongylidia taken directly from the garden (in situ), the average area of the gongylidia in the A. sexdens field garden was larger (1256 µm2 ± 317) than gongylidia from A. sexdens of laboratory garden 1039.8 µm2 ± 329 (Dunnet test, p<0.05, Tabe S2). The gongylidia from A. coronatus was slightly smaller than the others with an average area of 885 µm2 ± 249.4 (Dunnet test, p<0.05, Table S2). Biomass production The bacterial strains had different effects in fungal cultivar biomass production after 15 days of growth. When in co-culture with Lysinibacillus sp. (Bac2) and Staphylococcus sp. (Bac10) at 102 CFU mL-1 the biomass of the culture was significantly increased (Fig. 2a, ANOVA, p<0.05). Paenibacillus sp. (Bac11) significantly increased the fungal biomass when in concentration up to 104 CFU (Fig. 2a, ANOVA, p<0.05). While in co-culture with Bacillus sp. (BacA) and Bacillus sp. (Bac C) there was a significant decrease of biomass production, inversely proportional to the bacterial concentration (Table S3, ANOVA, p<0.05). When in co- culture with Niallia sp. (BacD) and Bacillus sp. (BacE) there was no significant effect on biomass production (Fig. 2b, ANOVA, p>0.05). On the other hand, in the presence of Pantoea sp. (Bac1), Pantoea sp. (BacB) and Actinobacteria (BacG) fungal growth was inhibited (Fig. 2). 34 Measurement of ergosterol concentration in co-culture with Staphylococcus sp. (Bac10) confirmed that increase in biomass in this assay was due to an increase in fungal biomass, showing significant difference when compared with the axenic culture (T-test, p=0.011, Fig. S2). Biofilm-like structure formation Since the outcome of fungal-bacterial interactions can be influenced by physical contact [15], we assessed by SEM if bacteria are able to attach to the hyphae and which part of the fungal colony they colonized. After 24 hours of bacteria-fungus co-culture, bacterial colonization was evident on the surface of the hyphae (Fig. 3). All bacteria but Niallia sp. (BacD) and Actinobacteria (BacG) formed aggregates on the hyphae of the fungal cultivars (Fig. 3). Most interestingly, Lysinibacillus sp. (Bac2) grew in a streptobacillus-like shape surrounding the fungal hyphae (Fig. 3). For Actinobacteria (BacG) we did not observe biofilm formation, most likely due to the slow growth rate of this bacterium. Screening for putative metabolic functions for garden development We observed that bacteria isolated from the staphylae surface have different potential metabolic abilities that could be beneficial for garden development, such as phosphate solubilization, siderophore production, cellulose hydrolyzation and chitin degradation (Fig. 4). Briefly, half of the bacteria showed cellulose-degrading capability [Pantoea sp. (Bac1), Bacillus sp. (BacA), Bacillus sp. (BacC), Niallia sp. (BacD) and Actinobacteria (BacG), Fig. 4], with Bacillus sp. (BacA) showing the highest rate of degradation (index 1.79 – ratio between halo and colony size, Table S4). This result indicates the potential of hyphosphere bacteria to help the fungus 35 cultivar in the degradation of plant material, the main source of fungus nutrition. We found that Lysinibacillus sp. (Bac2), Staphylococcus sp. (Bac10) and Pantoea sp. (BacB) can produce siderophores (Table S4, Fig. 4), suggesting that these isolates could promote bioavailability of Fe+3. Additionally, Pantoea sp. (Bac1), Bacillus sp. (BacA), Pantoea sp. (BacB), Bacillus sp. (BacC) and Actinobacteria (BacG) were able to solubilize inorganic phosphates. Bacillus sp. (BacE) and Actinobacteria (BacG) showed a great ability to degrade chitin, with degradation indexes of 1.2 and 1.1, respectively (Table S4). Bacteria pairwise interactions To assess if bacteria found on the hyphosphere of the fungal mutualist form a synergistic community or compete with each other in the hyphosphere, we tested for interactions between them. In vitro bacteria-bacteria interaction assays revealed heterogeneous patterns including synergism and antagonism. All bacteria were inhibited by at least one other bacteria. Niallia sp. (BacD), Paenibacillus sp. (Bac11) and Lysinibacillus sp. (Bac2) had no inhibitory activity against the other bacteria (Fig. S3). Besides, Niallia sp. (BacD), Staphylococcus sp. (Bac10) and Lysinibacillus sp. (Bac2) were the most inhibited bacteria, as their growth was inhibited by six, five and six of the other bacteria, respectively (Fig. S3). By contrast, Bacillus sp. (BacA) and Bacillus sp. (BacC) showed the greatest inhibition activity (80% of the other bacteria was inhibited by them) (Fig. S3). Besides, Bacillus spp. (BacA, BacC and BacE) and Pantoea spp. (BacB and Bac1) showed the greatest level of inhibition against Niallia sp. (BacD) (score: -5), and Paenibacillus (Bac11). In contrast, Actinobacteria (BacG) had the lowest level of inhibition against Lysinibacillus sp. (Bac2) (score: -1) and Staphylococcus sp. (Bac10) (score: -1). Although less frequent, synergistic effects were also found: three strains showed stimulatory ability [Lysinibacillus sp. (Bac2), Pantoea sp. (BacB) and Actinobacteria (BacG)] and two got 36 promoted [Pantoea sp. (Bac1) and Bacillus sp. (BacE)] by other isolates (Fig. 5). The growth of Pantoea sp. (Bac1) was promoted by three different strains. Conversely, Lysinibacillus sp. (Bac2) promoted the growth of two different strains (Pantoea sp. (Bac1) and Bacillus sp. (BacE) (Fig. 5). Regarding the garden origin of the isolates, bacteria from A. sexdens field gardens had the most pronounced inhibition potential. Unlike isolates from A. coronatus gardens, which generally showed lower inhibition activity. Bacillus sp. (BacE) was an exception, as it inhibited Niallia sp. (BacD). Bacteria from the A. sexdens laboratory garden showed various interaction outcomes, including the highest growth-promotion potential by Lysinibacillus sp. (Bac2) (Fig. 5). In vitro bacterial antagonism against Escovopsis Finally, we tested the ability of the bacterial isolates for putative defense of the fungus garden against E. weberi. Lysinibacillus sp. (Bac2), Bacillus sp. (BacA), Pantoea sp. (BacB), and Bacillus sp. (BacC) isolates showed three patterns against E. weberi, inhibiting germination when inoculated earlier or simultaneously, and hindering mycelial growth when inoculated later. In the case of Bacillus spp. (BacA and BacC), they completely inhibited conidial germination and mycelial growth of E. weberi (Fig. 6, Table S5). Bacillus sp. (BacE) had two patterns, inhibiting conidial germination and mycelial growth when inoculated at the same time. Actinobacteria (BacG) had one pattern (inhibition of mycelial growth), allowed for conidial gemination, yet with minimal visible growth, in early and delayed bacterial inoculations; when inoculated at the same time, there were minimal effects on growth for either (Fig. 6). On the other hand, Pantoea sp. (Bac1), Staphylococcus sp. (Bac10), Paenibacillus sp. (Bac11), and Niallia sp. (BacD) did not impact the growth of E. weberi in the tested conditions (Fig. 6). 37 DISCUSSION We isolated different bacterial groups from the staphylae of ant fungal cultivars, including Actinobacteria, Bacillus, Lysinibacillus, Pantoea, and Staphylococcus, which were previously described in metagenomics analysis of the whole garden matrix [35, 36]. However, our study reported for the first time the presence of the genus Niallia that is mainly associated to leaves of plants [34]. We also provide evidence for intimate associations between bacteria and the fungus cultivar of two attine ant species. Bacteria had a positive effect on the fungus cultivar increasing the staphyla production [Staphylococcus sp. (Bac10) and Bacillus sp. (BacE)] and biomass production [Staphylococcus sp. (Bac10) and Paenibacillus sp. (Bac11)]. Also, bacteria isolated from gongylidia of fungus cultivars had the ability to form biofilm-like aggregates on the hyphal surface, and to perform useful metabolic functions (i.e., siderophore production, phosphate solubilization, cellulose degradation and chitinase production) for garden development and fungal growth. We propose a hypothetical scenario in which conditions in the fungus cultivar promote the formation of a microbial biofilm, housing a community of putatively beneficial associates on the hyphal surface (Fig. 7). Although the dominant microorganism in the attine garden is the fungal cultivar, other fungi and bacteria cohabit this environment [1]. Here we show that different bacteria are physically related to the staphyla from the cultivar. This suggest that, while ants eliminate some alien microbes from their cultures [37], cultivars may recruit beneficial microorganisms, as occurs in ectomycorrhizae [38]. However, when grown in co-culture with the fungus, we most often observed a negative relationship between bacterial concentration and fungal biomass (Fig. 2). Previous studies on the whole garden bacteriome pointed out that culturable microorganisms in the garden do not exceed 105 CFU [39], our results suggest that depending on the composition and abundance of bacteria in the garden, these microorganisms might be beneficial or harmful 38 to the ant’s system, as occurs in corals and human hosts [40, 41]. Most interestingly, we identified a Staphylococcus isolate (Bac10) which stimulated the staphylae production and increased fungal biomass, features that are interesting for fungal cultivar fitness in the context of leaf-cutting ant colonies. Besides, we observed that gongylidia from field gardens were slightly larger than the gongylidia from laboratory cultures suggesting that environmental conditions influence garden fitness. Additionally, most bacteria could form biofilm-like aggregations on the hyphae. This was especially evident for Lysinibacillus sp. (Bac2), which showed a prominent capability to grow on the hyphal surface (Fig. 3). These features could be useful to the mutualist (L. gongylophorus) to host additional bacterial communities, which might expand the metabolic repertoire that confer advantage over their partners. In this association, while the fungus can supply bacteria with sugars and polyols, bacteria could promote fungal growth by enhancing nutrient uptake, and/or suppressing pathogens by producing antibiotics, siderophores, and fungal cell wall-lysing enzymes (Fig. 7). Several bacteria examined in this study showed the ability to increase the availability of essential elements like phosphorus, iron and carbon. Here, Pantoea sp. (BacB) and Pantoea sp. (Bac1) were able to solubilize phosphate and produce siderophores. Thus, they could promote phosphorus availability to fungal cultivars. While the genus Pantoea was previously reported in attine ant’s garden as a potential fixator of nitrogen in the garden [11] , other putative function of these bacteria for the fungus had not been tested in vitro or in vivo in previous studies. Additionally, Actinobacteria (BacG), which have also been reported as part of the garden defence system [42], exhibited phosphate solubilization and cellulose degradation abilities. Previous studies showed that some Actinobacteria can solubilize phosphates in different systems [43], and have genes related to the production of different cellulose degrading enzymes [44]. While the main substrate for the fungal cultivar is plant material, and some studies showed that ant fungal cultivars have genes corresponding to 39 cellulose degradation [45], in vitro analyses suggest a different scenario [46, 47]. Our findings indicate that the fungus may not be the only cellulose degrader in the garden. We show that Bacillus spp. (BacA and BacC) and Niallia sp. (BacD) have cellulose degradation activity that can help to transform plant material. Thus, Bacillus spp. might support the fungus in the cellulose degradation process, as also reported by Moreira-Soto et al. [48]. Defense against antagonistic fungi by production of chitinase and antifungal compounds is another putative role of bacteria associated to the attine gardens. We observed that bacteria with [Bacillus sp. (BacE) and Actinobacteria (BacG)] and without [Lysinibacillus sp. (Bac2), Bacillus sp. (BacA), Bacillus sp. (BacC) and Pantoea (BacB)] chitinase activity were able to inhibit germination and mycelial growth of Escovopsis, the main antagonistic fungus of the garden. Previous studies have shown that bacteria including Bacillus, Enterobacter, and Lysinibacillus are present [48], in the initial pellet garden, when the cultivars are more susceptible to contaminants [49]. Therefore, these early bacterial colonizers may contribute to the successful development of the garden. Taken together, our findings suggest that several bacterial groups might contribute to garden defense in addition to Actinobacteria, which are well known for their antifungal role in the garden [42]. Beyond the effects on the fungal mutualist, bacterial pairwise interactions in the garden can be diverse, ranging from antagonistic to stimulatory. Notably, bacteria that had higher inhibitory activities against other bacteria also showed antagonistic behavior against Escovopsis [Pantoea (Bac1), Lysinibacillus sp. (Bac2), Bacillus sp. (BacA), Bacillus sp. (BacC), Bacillus sp. (BacE)]. Additionally, Bacillus sp. (BacE) and Lysinibacillus sp. (Bac2) did not have a significant effect on the growth of the mutualistic fungus in liquid co-culture, suggesting that they could cohabit with the fungus, but due to their antifungal activity they could help in the defense against antagonists. Considering our findings, the hyphosphere-associated bacteria may work as a helper community in maintaining the garden, while taking advantage of available 40 resources. In line with the high susceptibility of the cultivar against other fungi in in vitro assays and under controlled conditions [50, 51], the fungal cultivar might be more susceptible when the naturally-occurring microbial partners are absent. Here we highlight bacteria inhabiting the hyphosphere of L. gongylophorus with biofilm-like forming abilities, which potentially promote a microecosystem where nutrients are circulating and are accessible to the fungus. We focused on Lysinibacillus sp. (Bac2), Pantoea sp. (BacB) and Staphylococcus sp. (Bac10) isolates, which formed biofilm-like structures and showed putative metabolic functions including the production of siderophores, phosphate solubilization and antifungal activity. Remarkably, Staphylococcus sp. stimulated staphylae production and increased the biomass production of the cultivar, potentially through facilitation of Fe+ absorption by the fungus in a superficial biofilm-like matrix. Furthermore, we propose that Actinobacteria might have more than one function in the garden, given the cellulose degradation ability and phosphate solubilizing activity of the studied isolates, which may increase substrate accessibility for the fungus and bioavailability of Fe+3 for the microbial neighbors. Consequently, our findings show the metabolic capabilities of bacteria from the hyphosphere likely support garden maintenance. While our study reveals important potential from culturable bacteria (Fig. 4), further investigation including unculturable bacteria from the hyphosphere may reveal a complete picture of the complexity of bacteria-fungal interactions within the attine gardens. Acknowledgments We would like to thank the Laboratory of Fungal Ecology and Systematics (LESF - São Paulo State University, Rio Claro, SP, Brazil) and Ecogenomics of Interaction (INRAE, Champenoux, France) research teams for comments on early drafts of this manuscript. We 41 thank T. Dhalleine (IAM Lorraine University) for helping with HPLC analyses. We also thank three anonymous reviewers for their constructive comments on this manuscript. Funding We are grateful to Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and to the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for a fellowship (grant # 305269/2018-6). Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) provided financial support (grant # 2014/24298-1, # 2017/12689-4 and #2019/03746-0) to AR and (grant # 2021/04706-1) to QVM. The French National Research Agency (ANR) (ANR-11-LABX-0002-01, Lab of Excellence ARBRE) provided financial support to AD and the Novo Nordisk Foundation (Posdoctoral Fellowship NNF20OC0064385) to LVF. Competing interests The authors declare they have no financial interests. Authors' contributions MJSM, AD, LVF, and AR designed the study. MJSM and QVM carried out fieldwork. MJSM, AD and QVM carried out laboratory work. MJSM, AD and QVM analyzed the data. MJSM and AR wrote the first drafts of the manuscript. All authors revised and contributed the manuscript. Data availability Data supporting the results in the paper are available in the Supplementary Material. 42 Ethics approval No ethical approval was needed for this work. REFERENCES 1. Rodrigues A, Bacci M, Mueller UG, et al (2008) Microfungal “weeds” in the leafcutter ant symbiosis. Microb Ecol 56:604–614. https://doi.org/10.1007/s00248-008-9380-0 2. Quinlan RJ, Cherrett J (1979) The role of fungus in the diet of the leafcutting ant Atta cephulotes (L.). Ecol Entomol 4:151–160 3. Martin MM, Stadler Martin J (1970) THE BIOCHEMICAL BASIS FOR THE SYMBIOSIS BETWEEN THE ANT, ATTA COLOMBICA TONSIPES, AND ITS FOOD FUNGUS. J Insect Physiol 16:109–119 4. Khadempour L, Burnum-Johnson KE, Baker ES, et al (2016) The fungal cultivar of leaf- cutter ants produces specific enzymes in response to different plant substrates. Mol Ecol 25:5795–5805. https://doi.org/10.1111/mec.13872 5. Little AEF, Murakami T, Mueller UG, Currie CR (2006) Defending against parasites: Fungus-growing ants combine specialized behaviours and microbial symbionts to protect their fungus gardens. Biol Lett 2:12–16. https://doi.org/10.1098/rsbl.2005.0371 6. Huang EL, Aylward FO, Kim YM, et al (2014) The fungus gardens of leaf-cutter ants undergo a distinct physiological transition during biomass degradation. Environ Microbiol Rep 6:389–395. https://doi.org/10.1111/1758-2229.12163 7. Higa T (1991) Effective microorganisms: a biotechnology for mankind . First International Conference on Kyusei Nature Farming 8–14 8. Poulsen M, Currie CR (2010) Symbiont interactions in a tripartite mutualism: Exploring the presence and impact of antagonism between two fungus-growing ant mutualists. PLoS One 5:1–14. https://doi.org/10.1371/journal.pone.0008748 9. Jiménez-Gómez I, Barcoto MO, Montoya Q v., et al (2021) Host Susceptibility Modulates Escovopsis Pathogenic Potential in the Fungiculture of Higher Attine Ants. Front Microbiol 12:. https://doi.org/10.3389/fmicb.2021.673444 10. De Fine Licht HH, Boomsma JJ, Tunlid A (2014) Symbiotic adaptations in the fungal cultivar of leaf-cutting ants. Nat Commun 5:5675. https://doi.org/10.1038/ncomms6675 11. Pinto-Tomás AA, Anderson MA, Suen G, et al (2009) Symbiotic nitrogen fixation in the fungus gardens of leaf-cutter ants. Science (1979) 326:1120–1123. https://doi.org/10.1126/science.1173036 12. Aylward FO, Burnum KE, Scott JJ, et al (2012) Metagenomic and metaproteomic insights into bacterial communities in leaf-cutter ant fungus gardens. ISME Journal 6:1688–1701. https://doi.org/10.1038/ismej.2012.10 13. Barea JM, Pozo MJ, Azcón R, Azcón-Aguilar C (2005) Microbial co-operation in the rhizosphere. J Exp Bot 56:1761–1778. https://doi.org/10.1093/jxb/eri197 14. Barbieri E, Ceccaroli P, Palma F, et al (2012) Ectomycorrhizal Helper Bacteria: The Third Partner in the Symbiosis. In: Zambonelli A, Bonito GM (eds) Edible Ectomycorrhizal Mushrooms. Springer, Berlin 15. Frey-Klett P, Garbaye J (2005) Mycorrhiza Helper Bacteria: A Promising Model for the Genomic Analysis of Fungal 43 16. Alvarado PE, Barrios RMM, Xóchihua JAM, Hernández JFC (2017) Fast and reliable DNA extraction protocol for identification of species in raw and processed meat products sold on the commercial market. Open Agric 2:469–472. https://doi.org/10.1515/opag-2017-0051 17. Turner Sean, Pryer M Kathleen, Miao P W Vivian, Palmer D Jeffrey (1999) Investigating Deep Phylogenetic Relationships among Cyanobacteria and Plastids by Small Subunit rRNA Sequence Analysis. Eukaryotic Microbiology 46:327–338 18. Dorsch M, Stackebrandt E (1992) Some modifications in the procedure of direct sequencing of PCR amplified 16S rRNA. J Microbiol Methods 16:271–279 19. Ki JS, Zhang W, Qian PY (2009) Discovery of marine Bacillus species by 16S rRNA and rpoB comparisons and their usefulness for species identification. J Microbiol Methods 77:48– 57. https://doi.org/10.1016/j.mimet.2009.01.003 20. Hall TA (1999) Bioedit: A User-Friendly Biological Sequence Alignment Editor and Analysis Program for Windows 95/98/NT. Nucleic Acids Symp Ser 41:95–98 21. Katoh K, Standley DM (2013) MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability. Mol Biol Evol 30:772–780. https://doi.org/10.1093/molbev/mst010 22. Nixon KC (2002) WinClada ver. 1.0000. Published by the author, Ithaca, New York, USA 23. Darriba Diego, Taboada L. Guillermo, Doallo Ramon, Posada David (2012) jModelTest 2: more models, new heuristics and parallel computing. Nature Methods 9 772: 24. Ronquist F, Teslenko M, van der Mark P, et al (2012) MrBayes 3.2: Efficient Bayesian Phylogenetic Inference and Model Choice Across a Large Model Space. Syst Biol 61:539– 542. https://doi.org/10.1093/sysbio/sys029 25. Stamatakis A (2014) RAxML version 8: a tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30:1312–1313. https://doi.org/10.1093/bioinformatics/btu033 26. He L, He X, Liu X, et al (2020) A sensitive, precise and rapid LC–MS/MS method for determination of ergosterol peroxide in Paecilomyces cicadae mycelium. Steroids 164:108751. https://doi.org/10.1016/j.steroids.2020.108751 27. Guennoc CM, Rose C, Guinnet F, et al (2017) A new method for qualitative multi-scale analysis of bacterial biofilms on filamentous fungal colonies using confocal and electron microscopy. Journal of Visualized Experiments 2017:. https://doi.org/10.3791/54771 28. Nautiyal CS (1999) An efficient microbiological growth medium for screening phosphate solubilizing microorganisms. FEMS Microbiol Lett 170:265–270. https://doi.org/10.1111/j.1574-6968.1999.tb13383.x 29. Schwyn B, Neilands JB (1987) Universal chemical assay for the detection and determination of siderophores. Anal Biochem 160:47–56. https://doi.org/10.1016/0003-2697(87)90612-9 30. Hsu SC, Lockwood JL (1975) Powdered Chitin Agar as a Selective Medium for Enumeration of Actinomycetes in Water and Soil. Appl Microbiol 29:422–426 31. Gonzalo M, Deveau A, Aigle B (2020) Inhibitions Dominate but Stimulations and Growth Rescues Are Not Rare Among Bacterial Isolates from Grains of Forest Soil. Isolates from Grains of Forest Soil Microbial ecology 80:. https://doi.org/10.1007/s00248-020-01579-6ï 32. Shannon P, Markiel A, Ozier O, et al (2003) Cytoscape: A software Environment for integrated models of biomolecular interaction networks. Genome Res 13:2498–2504. https://doi.org/10.1101/gr.1239303 33. R Core Team 2020 R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. In: http://www.r-project.org/index.html 34. Gupta RS, Patel S, Saini N, Chen S (2020) Robust demarcation of 17 distinct bacillus species clades, proposed as novel bacillaceae genera, by phylogenomics and comparative genomic analyses: Description of robertmurraya kyonggiensis sp. nov. and proposal for an emended 44 genus bacillus limiting it only to the members of the subtilis and cereus clades of species. Int J Syst Evol Microbiol 70:5753–5798. https://doi.org/10.1099/ijsem.0.004475 35. Barcoto MO, Carlos-Shanley C, Fan H, et al (2020) Fungus-growing insects host a distinctive microbiota apparently adapted to the fungiculture environment. Sci Rep 10:. https://doi.org/10.1038/s41598-020-68448-7 36. Aylward FO, Suen G, Biedermann PHW, et al (2014) Convergent bacterial microbiotas in the fungal agricultural systems of insects. mBio 5:. https://doi.org/10.1128/mBio.02077-14 37. de Fine Licht HH, Boomsma JJ (2010) Forage collection, substrate preparation, and diet composition in fungus-growing ants. Ecol Entomol 35:259–269 38. de Boer W, Folman LB, Summerbell RC, Boddy L (2005) Living in a fungal world: Impact of fungi on soil bacterial niche development. FEMS Microbiol Rev 29:795–811 39. Pagnocca FC, Carreiro SC, Bueno OC, et al (1996) Microbiological changes in the nests of leaf-cutting ants fed on sesame leaves. Journal of Applied Entomology 120:317–320. https://doi.org/10.1111/j.1439-0418.1996.tb01612.x 40. Manavathu EK, Vager DL, Vazquez JA (2014) Development and antimicrobial susceptibility studies of in vitro monomicrobial and polymicrobial biofilm models with Aspergillus fumigatus and Pseudomonas aeruginosa. BMC Microbiol 14:. https://doi.org/10.1186/1471- 2180-14-53 41. Boilard A, Dubé CE, Gruet C, et al (2020) Defining coral bleaching as a microbial dysbiosis within the coral holobiont. Microorganisms 8:1–26. https://doi.org/10.3390/microorganisms8111682 42. Barke J, Seipke RF, Grüschow S, et al (2010) A mixed community of actinomycetes produce multiple antibiotics for the fungus farming ant Acromyrmex octospinosus. BMC Biol 8:. https://doi.org/10.1186/1741-7007-8-109 43. Vargas Hoyos HA, Chiaramonte JB, Barbosa-Casteliani AG, et al (2021) An Actinobacterium Strain From Soil of Cerrado Promotes Phosphorus Solubilization and Plant Growth in Soybean Plants. Front Bioeng Biotechnol 9:. https://doi.org/10.3389/fbioe.2021.579906 44. Anderson I, Abt B, Lykidis A, et al (2012) Genomics of aerobic cellulose utilization systems in actinobacteria. PLoS One 7:. https://doi.org/10.1371/journal.pone.0039331 45. Aylward FO, Burnum-Johnson KE, Tringe SG, et al (2013) Leucoagaricus gongylophorus Produces Diverse Enzymes for the Degradation of Recalcitrant Plant Polymers in Leaf-Cutter Ant Fungus Gardens. Appl Environ Microbiol 79:3770–3778. https://doi.org/10.1128/AEM.03833-12 46. Abril AB, Bucher EH (2002) Evidence that the fungus cultured by leaf-cutting ants does not metabolize cellulose. Ecol Lett 5:325–328 47. Vigueras G, Paredes-Hernández D, Revah S, et al (2017) Growth and enzymatic activity of Leucoagaricus gongylophorus, a mutualistic fungus isolated from the leaf-cutting ant Atta mexicana, on cellulose and lignocellulosic biomass. Lett Appl Microbiol 65:173–181. https://doi.org/10.1111/lam.12759 48. Moreira-Soto RD, Sanchez E, Currie CR, Pinto-Tomás AA (2017) Ultrastructural and microbial analyses of cellulose degradation in leaf-cutter ant colonies. Microbiology (United Kingdom) 163:1578–1589. https://doi.org/10.1099/mic.0.000546 49. Fernández-Marín H, Zimmerman JK, Wcislo WT (2007) Fungus garden platforms improve hygiene during nest establishment in Acromyrmex ants (Hymenoptera, Formicidae, Attini). Insectes Soc 54:64–69. https://doi.org/10.1007/s00040-007-0907-z 50. Barcoto MO, Pedrosa F, Bueno OC, Rodrigues A (2017) Pathogenic nature of Syncephalastrum in Atta sexdens rubropilosa fungus gardens. Pest Manag Sci 73:999–1009. https://doi.org/10.1002/ps.4416 45 51. Cafaro MJ, Poulsen M, Little AEF, et al (2011) Specificity in the symbiotic association between fungus-growing ants and protective Pseudonocardia bacteria. Proceedings of the Royal Society B: Biological Sciences 278:1814–1822. https://doi.org/10.1098/rspb.2010.2118 Figure legends Fig. 1 Gongylidium size (µm2) in bacteria-fungus (Leucoagaricus gongylophorys) dual-cultures after 15 days of interaction. a. Gongylidia area of fungal cultivar isolated from the laboratory garden of Atta sexdens. b. Gongylidia area of fungal cultivar isolated from the field garden of Atta sexdens. c. Gongylidia area of a fungal cultivar isolated from the field garden of Acromyrmex coronatus. d. Gongylidia area from staphyla obtained directly from fungus garden of A. sexdens (laboratory garden), A. sexdens (field) and A. coronatus (field). 46 Fig. 2 Fungal dry biomass after 15 days of bacteria-fungi dual-culture. a. Co-culture of bacteria and the fungal cultivar isolated from an Atta sexdens laboratory garden. b. Co-culture of bacteria and the fungal cultivar isolated from an A. sexdens field garden. c. Co-culture of bacteria and the fungal cultivar isolated from an Acromyrmex coronatus field garden. Boxplot colors represent bacteria concentration (CFU mL-1): Red 0; Yellow 102; Green 103; Blue 104; Purple 105; Pink 106. P values: *<0.05, **<0.01,***<0.001, ****<0.0001 47 Fig. 3 Bacterial biofilm-like structures formation in liquid co-culture with the fungus cultivar after 24 hours of incubation. Attachment of eight bacterial isolates on the hyphae surface of Leucoagaricus gongylophorus. Images were taken using Scanning Electron Microscopy (Bar=20 µm, 4000 magnification, Bar= 30 µm, 3000 magnification). 48 Fig. 4 Heatmap for the putative bacterial metabolic functions involved in ant fungal garden development. Columns indicate the metabolic functions (phosphate solubilization, siderophore production, cellulose degradation, chitinase activity, antifungal activity, staphylae stimulation and biofilm formation) and the rows the bacteria tested. Scale indicates the normalized intensity of the bacterial response to the tested functions from 0 to 1. 49 Fig. 5 Hierarchical network representing pairwise interactions. a. Bacteria-bacteria interaction assays revealed heterogeneous patterns including synergism, competition, and potential parasitism. Box colors refer to the attine ant host species associated to the garden where each bacterium was isolated. Red solid lines represent unidirectional growth inhibition; red dotted lines represent mutual growth inhibition; and green solid lines represent growth promotion. The color intensity of the lines represents interaction values from -4 (negative interaction) to 1 (Positive interaction). b. Representative plate of bacteria-bacteria interaction showing the inhibition zone. 50 Fig. 6 Several bacterial strains from the ant fungal cultivar hyphosphere inhibit Escovopsis weberi. a. Early bacterial inoculation. b. Delayed bacterial inoculation. c. Simultaneous inoculation of bacteria and fungus. d. Control (only bacteria). Rows represent the tested bacterial strains. 51 Fig. 7 Hypothetical scenario of helper bacterial communities on the hyphosphere of attine ants’ fungal cultivar: a. Pupae surrounded by staphyla (the main nutritional source for brood). b. Staplyla of fungus cultivar as a place where abiotic conditions in the garden, i. e. hypoxia and humidity, facilitate the establishment of bacterial biofilm patches promoting a close relationship with the cultivar. c. The hyphosphere has the conditions to support its own bacterial microbiota, which in turn could facilitate the bioavailability of some essential minerals like iron, phosphorus, and carbon. The hyphosphere could also contribute to defense against antagonists of the fungal cultivar, benefiting the attine ant-fungus mutualism. 52 SUPLEMENTAL MATERIAL The hyphosphere of leaf-cutting ant cultivars is enriched in helper bacteria Martiarena MS1, Deveau A2, Montoya QV1, Flórez LV3, and Rodrigues A1 1 Department of General and Applied Biology, São Paulo State University (UNESP), Rio Claro, Brazil. 2 Université de Lorraine, INRAE, UMR IAM, 54280 Champenoux, France 3 Department of Plant and Environmental Sciences, University of Copenhagen, Denmark. Table S1: Staphyla production in vitro test for bacteria-fungi co-cultures, test made by six replicates, Dunnet test comparation Staphylae Mean Percentage area covered (M) SE Multiple comparation (Dunnet test) 95% family-wise confidence level comparation diff lwr.ci upr.ci p value Atta sexdens garden from the Lab Control (Fungus) 0.9 1.6 Control-control 0 Fungus + Pantoea sp. (Bac1) 1.2 1.5 (Fungus + Pantoea sp. (Bac1))- control 0.32 -6.85 7.49 0.9999 Fungus + Lysinibacillus sp. (Bac2) 0,0 0 (Fungus + Lysinibacillus sp. (Bac2))-control -0.96 -8.13 6.21 0.9899 Fungus + Staphylococcus sp. (Bac10) 26.2 9.7 (Fungus + Staphylococcus sp. (Bac10))-control 25.32 18.14 32.49 2.3e-09 Fungus + Paenibacillus sp. (Bac11) 3.2 3.5 (Fungus + Paenibacillus sp. (Bac11))-control 2.24 -4.93 9.41 0.8303 Atta sexdens garden from field Control (Fungus) 1.9 1.7 Control-control 0 Fungus + Bacillus sp. (BacA) 8.0 4.6 (Fungus + Bacillus sp. (BacA))-control 6.08 1.35 10.82 0.0106 Fungus + Pantoea sp. (BacB) 0,0 0 (Fungus + Pantoea sp. (BacB))-control -1.92 -6.65 2.81 0.6112 Fungus + Bacillus sp. (BacC) 4.8 4.1 (Fungus + Bacillus sp. (BacC))-control 2.88 -1.84 7.61 0.3070 Acromyrmex coronatus garden from field Control (Fungus) 2.8 1 Control-control 0 Fungus + Niallia sp. (BacD) 5.1 2.6 (Fungus + Niallia sp. (BacD))- control 2.24 -0.49 4.98 0.1232 Fungus + Bacillus sp. (BacE) 9.6 2.4 (Fungus + Bacillus sp. (BacE))-control 6.73 3.99 9.46 1.3e-05 Fungus + Actinobacteria (BacG) 0,0 0 (Fungus + Actinobacteria (BacG))-control -2.88 -5.62 -0.14 0.0376 *Mean area: average percentage of staphylae covered, SD standard deviation. 53 Table S2: Gongylidia size in bacteria-fungi co-cultures, test made by six replicates, measures were made under microscopy light. Staphylae Mean area (M) SE Multiple comparation (Dunnet test) 95% family-wise confidence level (mm2) comparation diff lwr.ci upr.ci p value Atta sexdens garden from the Lab Control (Fungus) 1488 334.7 Control-control 0 Fungus + Pantoea sp. (Bac1) 946 284.6 (Fungus + Pantoea sp. (Bac1))- control -542.34 -704.72 -379.95 9.9e-14 Fungus + Lysinibacillus sp. (Bac2) 0 0 (Fungus + Lysinibacillus sp. (Bac2))-control - 1488.97 - 1651.36 - 1326.59 <2e-16 Fungus + Staphylococcus sp. (Bac10) 1615 269.1 (Fungus + Staphylococcus sp. (Bac10))-control 126.58 -35.79 288.97 0.17242 Fungus + Paenibacillus sp. (Bac11) 1206 242.8 (Fungus + Paenibacillus sp. (Bac11))-control -282.11 -444.49 -119.72 0.00015 Atta sexdens garden from field Control (Fungus) 1123 370.3 Control-control 0 Fungus + Bacillus sp. (BacA) 864 205.6 (Fungus + Bacillus sp. (BacA))- control -258.79 -400.58 -117.01 9.2e-05 Fungus + Pantoea sp. (BacB) 0 0 (Fungus + Pantoea sp. (BacB))- control - 1123.10 - 1264.89 -981.32 <2e-16 Fungus + Bacillus sp. (BacC) 858 182.8 (Fungus + Bacillus sp. (BacC))- control -264.25 -406.03 -122.46 7.3e-05 Acromyrmex coronatus garden from field Control (Fungus) 881 124.4 Control-control 0 Fungus + Niallia sp. (BacD) 860 165.4 (Fungus + Niallia sp. (BacD))- control -21.49 -100.35 57.35 0.8540 Fungus + Bacillus sp. (BacE) 851 151.6 (Fungus + Bacillus sp. (BacE))- control -30.75 -109.61 48.09 0.6758 Fungus + Actinobacteria (BacG) 0 0 (Fungus + Actinobacteria (BacG))-control -881.92 -960.78 -803.06 <2e-16 Atta sexdens garden from field 1256 317 0 Atta sexdens garden from lab 1039 329 -217.08 -391.66 -42.50 0.0121 Acromyrmex coronatus garden from field 885 249.4 -370.91 -545.49 -196.33 1.4e-05 *Mean area: average of gongylidia area, SD standard deviation. Table S3: Fungal biomass production in bacteria-fungi dual cultures, weight of dry mass made by six replicates. Mutualistic Fungus Mutualistic fungus + Bacteria (CFU mL-1) Statistic df p value p.adj p.adj.signif Atta sexdens garden from the Lab Control Lysinibacillus sp. (Bac2) (102) -4.076 7.425 4.00e-03 1.200000e-02 * Control Lysinibacillus sp. (Bac2) (103) -0.692 9.112 5.06e-01 5.692500e-01 ns Control Lysinibacillus sp. (Bac2) (104) 0.4632 8.234 6.55e-01 6.854651e-01 ns Control Lysinibacillus sp. (Bac2) (105) -1.881 9.841 9.00e-02 1.446429e-01 ns Control Lysinibacillus sp. (Bac2) (106) -0.734 8.405 4.82e-01 5.561538e-01 ns Control Staphylococcus sp. (Bac10) (102) -4.883 5.645 3.00e-03 1.125000e-02 * Control Staphylococcus sp. (Bac10) (103) -2.729 5.600 3.70e-02 7.568182e-02 ns Control Staphylococcus sp. (Bac10) (104) -0.097 8.118 9.25e-01 9.250000e-01 ns Control Staphylococcus sp. (Bac10) (105) 1.870 7.993 9.80e-02 1.515000e-01 ns Control Staphylococcus sp. (Bac10) (106) 8.718 5.047 3.13e-04 2.012143e-03 ** 54 Control Paenibacillus sp. (Bac11) (102) -1.216 9.620 2.53e-01 3.348529e-01 ns Control Paenibacillus sp. (Bac11) (103) -2.176 7.105 6.50e-02 1.218750e-01 ns Control Paenibacillus sp. (Bac11) (104) -3.011 6.988 2.00e-02 4.736842e-02 * Control Paenibacillus sp. (Bac11) (105) -3.031 9.600 1.30e-02 3.441176e-02 * Control Paenibacillus sp. (Bac11) (106) -5.597 8.488 4.14e-04 2.160000e-03 ** Atta sexdens garden from field Control Bacillus sp. (BacA) (102) -3.858 9.805 3.00e-03 0.0047368 ** Control Bacillus sp. (BacA) (103) -7.474 8.950 3.91e-05 0.0001127 *** Control Bacillus sp. (BacA) (104) -17.348 6.475 1.10e-06 0.0000342 **** Control Bacillus sp. (BacA) (105) -20.194 5.140 4.30e-06 0.0000368 **** Control Bacillus sp. (BacA) (106) -20.372 5.062 4.70e-06 0.0000368 **** Control Bacillus sp. (BacC) (102) -3.912 9.926 3.00e-03 0.0047368 ** Control Bacillus sp. (BacC) (103) -7.712 9.360 2.37e-05 0.0001016 *** Control Bacillus sp. (BacC) (104) -14.984 6.090 4.90e-06 0.0000368 **** Control Bacillus sp. (BacC) (105) -17.236 5.061 1.09e-05 0.0000580 **** Control Bacillus sp. (BacC) (106) -17.218 5.023 1.16e-05 0.0000580 **** Acromyrmex coronatus garden from field Control Niallia sp. (BacD) (102) -0.573 9.865 5.79e-01 0.6703846 ns Control Niallia sp. (BacD) (103) -0.187 6.594 8.57e-01 0.8865517 ns Control Niallia sp. (BacD) (104) 1.677 9.392 1.26e-01 0.1890000 ns Control Niallia sp. (BacD) (105) -0.019 7.588 9.85e-01 0.9850000 ns Control Niallia sp. (BacD) (106) -1.921 9.283 8.60e-02 0.1389474 ns Control Bacillus sp. (BacE) (102) -1.921 8.788 8.80e-02 0.1389474 ns Control Bacillus sp. (BacE) (103) -0.488 9.671 6.36e-01 0.7066667 ns Control Bacillus sp. (BacE) (104) -7.140 9.084 5.17e-05 0.0002216 *** Control Bacillus sp. (BacE) (105) -19.060 5.029 7.00e-06 0.0000361 **** Control Bacillus sp. (BacE) (106) -18.987 5.019 7.20e-06 0.0000361 **** Table S4. Bacterial putative metabolic function for the garden development Bacteria Phosphate solubilization Siderophore production Cellulase activity Chitinase activity Mean index (M) SD Mean index (M) SD Mean index (M) SD Mean index (M) SD Pantoea sp. (Bac1) 3.930 0.277 1.366 0.031 1.350 0.070 - - Lysinibacillus sp. (Bac2) - - 3.399 0.371 - - - - Staphylococcus sp. (Bac10) - - 3.666 0.281 - - - - Paenibacillus (Bac11) - - - - - - - - Bacillus sp. (BacA) 1.180 0.011 - - 1.794 0.052 - - Pantoea sp. (BacB) 1.134 0.018 3.956 0.130 - - - - Bacillus sp. (BacC) 1.206 0.039 - - 1.695 0.081 - - Niallia sp. (BacD) - - - - 1.781 0.081 - - Bacillus sp. (BacE) - - - - - - 1.246 0.099 Actinobacteria (BacG) 1.052 0.004 - - 1.770 0.010 1.094 0.013 *Mean: average of index; SD standard deviation; “-” no activity. Table S5: Bacterial activity against Escovopsis weberi. 55 Bacteria Bacteria early inoculation: Inhibition of conidia germination Bacteria delay inoculation: Mycelial growth inhibition Same time bacteria inoculation: inhibition by competition Value of activity inhibition Pantoea sp. (Bac1) - - - 0 Lysinibacillus sp. (Bac2) + + + 3 Staphylococcus sp. (Bac10) - - - 0 Paenibacillus (Bac11) - - - 0 Bacillus sp. (BacA) + + + 3 Pantoea sp. (BacB) + + + 3 Bacillus sp. (BacC) + + + 3 Niallia sp. (BacD) - - - 0 Bacillus sp. (BacE) + - + 2 Actinobacteria (BacG) + - - 1 +: Positive effect, -: negative effect. Figures Fig. S1. Phylogenies indicating the placement of the bacterial isolates from the gongylidia surface (highlighted in bold). The trees show the phylogenetic placement of: a. the isolates BacA, BacC, BacE in the Bacillus genus; b. the isolate Bac11 in the genus Paenibacillus, c. the isolates BacB and Bac1 in the genus Pantoea, d. the isolate Bac2 in the genus Lysinibacillus, e. the isolate BacD in the genus Niallia, and f. the isolate Bac10 in the genus Staphylococcus. The trees showed were inferred using Maximum Likelihood Inference. The analyses were based on concatenated sequences of 16S and the rpoB genes for the