Specificity and stability of the Acromyrmex-Pseudonocardia symbiosis

Abstract

The stability of mutualistic interactions is likely to be affected by the genetic diversity of symbionts that compete for the same functional niche. Fungus-growing (attine) ants have multiple complex symbioses and thus provide ample opportunities to address questions of symbiont specificity and diversity. Among the partners are Actinobacteria of the genus Pseudonocardia that are maintained on the ant cuticle to produce antibiotics, primarily against a fungal parasite of the mutualistic gardens. The symbiosis has been assumed to be a hallmark of evolutionary stability, but this notion has been challenged by culturing and sequencing data indicating an unpredictably high diversity. We used 454 pyrosequencing of 16S rRNA to estimate the diversity of the cuticular bacterial community of the leaf-cutting ant Acromyrmex echinatior and other fungus-growing ants from Gamboa, Panama. Both field and laboratory samples of the same colonies were collected, the latter after colonies had been kept under laboratory conditions for up to 10 years. We show that bacterial communities are highly colony-specific and stable over time. The majority of colonies (25/26) had a single dominant Pseudonocardia strain, and only two strains were found in the Gamboa population across 17 years, confirming an earlier study. The microbial community on newly hatched ants consisted almost exclusively of a single strain of Pseudonocardia while other Actinobacteria were identified on older, foraging ants in varying but usually much lower abundances. These findings are consistent with recent theory predicting that mixtures of antibiotic-producing bacteria can remain mutualistic when dominated by a single vertically transmitted and resource-demanding strain.

Keywords: 454 pyrosequencing; Pseudonocardia; attine ant mutualism; bacterial community.

Figure 1

Bacterial abundance and diversity data obtained by 454 sequencing. (A) Dorsal view (top) of a callow worker of Acromyrmex echinatior covered in Pseudonocardia bacteria and ventral view (bottom) of a mature worker with bacterial growth concentrated on the laterocervical plates. Yellow circles indicate the parts of the cuticle that was dissected and sequenced. (B) Rarefaction curves of sequencing depth, each representing an individual sample and showing the observed number of OTUs as a function of simulated sequencing effort. Red curves represent samples from cluster 1 and blue curves samples from cluster 2 (see Fig. 2). Dotted curves are callow laboratory ants, dashed curves mature laboratory ants and solid curves field-collected ants. Curves based on other attine ant samples than A. echinatior and samples of A. echinatior that could not be placed in either cluster 1 or 2 are plotted in grey.

Figure 2

Hierarchical clustering of the bacterial communities from the cuticle of Acromyrmex echinatior and selected other attine ants, based on beta diversity analyses (Yue & Clayton 2005). On the x-axis, are the 27 most common OTUs, with the different taxonomic groups highlighted in different colours. On the y-axis, the dendrogram illustrates clustering of the samples and not phylogenetic relationships between the OTUs. Sample IDs are followed by the number of the sequencing run (1 or 2), and the frequency of each OTU is illustrated by a colour scale from light yellow (=zero to low frequency) to red (=high frequency to complete dominance). The two main clusters of Acromyrmex samples that were identified by hierarchical clustering analysis are numbered 1 and 2 as written in the dendrogram, and were each dominated by a single Pseudonocardia OTU that was (with only one exception for A. echinatior) not found in the other cluster. The tentative placement of these two OTUs in the phylogenetic clades identified by Cafaro et al. (2011) is indicated by Roman numbers below our Arabic cluster numbers. At the top of the figure the two Trachymymex samples clustered together as they both harboured a unique Amycolatopsis OTU. Three of the A. echinatior field samples and two of the Cyphomymex samples were not placed in either of the two main clusters due to low Pseudonocardia prevalence. Both clusters also harboured samples with a higher bacterial diversity (less intense red towards the left and more yellow towards the right with variation increasing from top to bottom in each cluster). The horizontal order of the bacterial OTUs is based on the OTU classification numbers found in the DataS1.

Figure 3

Alpha diversity of bacterial communities. The bacterial diversity estimated with the inverse Simpson index after subsampling of 1023 sequences per sample so that mean ± SE estimates could be obtained. (A) A two-way anova showed that there was a significant diversity difference between clusters Ps1 (grey) and Ps2 (black) and ant-sampling categories (callow, mature, field), and that also the interaction term between these predictor variables was significant (clusters: F_1,45 = 14.89, P = 0.0004; ant categories: F_2,45 = 16.57, P < 0.0001; interaction: F_2,45 = 6.74, P = 0.003). The mature field worker diversity index of Pseudonocardia cluster 1 was significantly higher than the estimate obtained for cluster 2, and field-collected samples had consistently higher diversity than laboratory-collected callow and possibly also mature laboratory workers (see also Fig. 1B and Fig. 2), but the significant interaction term makes the latter result ambiguous. (B) Inverse Simpson diversity index plotted for each individual ant with samples from the same colony connected by lines. A one-way anova with matching columns found no significant effect of colony ID indicating that the differences across colonies are general. See text for further details.