The attine ants are a monophyletic lineage that switched to fungus-farming ca. 55-60 MYA. They have become a model for the study of complex symbioses after additional fungal and bacterial symbionts were discovered, but their abdominal endosymbiotic bacteria remain largely unknown. Here we present a comparative microbiome analysis of endosymbiotic bacteria spanning the entire phylogenetic tree. We show that, across 17 representative sympatric species from eight genera sampled in Panama, abdominal microbiomes are dominated by Mollicutes, α- and γ-Proteobacteria, and Actinobacteria. Bacterial abundances increase from basal to crown branches in the phylogeny reflecting a shift towards putative specialized and abundant abdominal microbiota after the ants domesticated gongylidia-bearing cultivars, but before the origin of industrial-scale farming based on leaf-cutting herbivory. This transition coincided with the ancestral single colonization event of Central/North America ca. 20 MYA, documented in a recent phylogenomic study showing that the entire crown-group of the higher attine ants, including the leaf-cutting ants, evolved there and not in South America. Several bacterial species are located in gut tissues or abdominal organs of the evolutionarily derived, but not the basal attine ants. The composition of abdominal microbiomes appears to be affected by the presence/absence of defensive antibiotic-producing actinobacterial biofilms on the worker ants’ cuticle, but the significance of this association remains unclear. The patterns of diversity, abundance, and sensitivity of the abdominal microbiomes that we obtained explore novel territory in the comparative analysis of attine fungus-farming symbioses and raise new questions for further in-depth research.
otu97 table pre-filtration blank_included
A: OTU table showing the 3334 OTUs identified in the MiSeq analysis of 107 samples analyzed by clustering 97% identical sequences in mothur before filtering. B: The same table rarefied at 8900 reads. C: OTU table showing the 65 OTUs identified in the three ‘blank’/water samples included in our MiSeq sequencing. To identify common OTUs between ‘blank’ and ant samples we used local blast analyses. For the OTUs that were present in both ant and ‘blank’ samples, we also show, at the bottom, the corresponding OTU numbers (IDs) in the ant sample table.
OTU97 table main ms filteredOTU97
A: OTU table showing the 2099 OTUs identified in the MiSeq analysis of 107 samples analyzed by clustering 97% identical sequences in Mothur. The table was rarefied at 8900 reads. To remove any potential water contaminants, we identified the OTUs (97% clustering) present in the blank ‘water’ samples that were sequenced along with the real samples, and removed these OTUs as contaminants if their relative abundances produced an average blank/sample ratio ≥ 0.2. B: The same OTU table presenting only the 21 abundant OTUs (Fig. S2).
OTU97_table_alternative_filtering0.2.unique
A rarefied OTU table (similar to the main OTU table used) using the same dataset but curated with an alternative filtering approach. Once again, to remove any potential water contaminants, we used the blank ‘water’ samples, but instead of creating ‘blank’ OTUs based on the 97% similarity we used the unique sequences. Similarly to the previous filtering approach, we then removed identical sequences that were common in the blank samples and the ant samples if their relative abundances produced an average blank/sample ratio ≥ 0.2.
unique seqs pre-filtration blank_included
A: Unique sequence table showing the distribution of the sequences identified in the MiSeq analysis of 107 ant samples before filtering. B: Unique sequence table showing the sequences identified in the three ‘blank’/water samples, included in the MiSeq sequencing. To identify common sequences between ‘blank’ and ant samples we used local blast analyses. For the unique sequences that were present in both ant and ‘blank’ samples, we also show, at the right, the corresponding unique sequence numbers (IDs) in the ant sample table.
unique table post-filtration
Unique sequence table showing the OTUs (after removing water contaminants; see Methods and Results) identified in the MiSeq analysis of 107 samples analyzed by clustering identical (100%) sequences in mothur.
unique_sequence_tabular_dryad
A table presenting all unique sequences identified in our study in both ant and ‘blank’ samples. From left to right (separated with tabs): a unique identifier, description of whether the sequence is a contaminant or ant specific (based on the 0.2 ratio), the number of times the sequence is present, the samples that the sequence is present (based on Table S1 in Supporting Information) and the nucleotide sequence.
OTU97 table main ms filteredOTU97_84samples
A: OTU table showing the 2099 OTUs after rarefaction and removing ‘water’ contaminants (see Methods and Results) identified in the MiSeq analysis of the 84 samples that were amplified with standard PCR, analyzed by clustering 97% identical sequences in Mothur. As before, the table was rarefied at 8900 reads. only The 18 abundant OTUs according to our criterion of ≥10% prevalence in at least two samples, are highlighted with Bold print. B: The same OTU table before rarefaction. To facilitate reading and interpretation we have named the 98 OTUs of highest overall abundance (Fig. S3) using five letters that refer to their bacterial order followed by a specific number that identifies their overall prevalence rank. Six of them had previously been identified in the guts of three species of Acromyrmex leaf-cutting ants from the same sampling site, and retained their names from that study (ActAcro1, EntAcro1, EntAcro2, EntAcro10, RhiAcro1 and WolAcro1: Sapountzis et al. 2015).
OTU97 table main ms filteredOTU97 84samples.xlsx
Supplementary Results 2
Examples of bacteria detected in various abdominal tissues, gut lumen and fat bodies of attine ant workers using confocal laser scanning FISH microscopy