FEMS Microbiology Ecology, 91, 2015, fiv073

doi: 10.1093/femsec/fiv073
Advance Access Publication Date: 25 June 2015
Research Article

RESEARCH ARTICLE

Bacterial community composition and diversity

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

in an ancestral ant fungus symbiosis
†
Katrin Kellner1,∗, , Heather D. Ishak1,2 , Timothy A. Linksvayer3
and Ulrich G. Mueller1
1
Section of Integrative Biology, Patterson Laboratories, University of Texas at Austin, 1 University Station
#C0930, Austin, TX 78712, USA, 2 Division of Infectious Disease and Geographic Medicine, Stanford University,
300 Pasteur Dr. MC 5107, Stanford, CA 94305, USA and 3 Department of Biology, University of Pennsylvania,
433 S. University Ave., Philadelphia, PA 19104, USA
∗ Corresponding author: Department of Biology, University of Texas at Tyler, 3900 University Blvd, Tyler, TX 75799, USA. Tel (903) 565-5931;
Fax (903) 566-7189; E-mail: [email protected]
†
Present address: Department of Biology, University of Texas at Tyler, 3900 University Blvd, Tyler, TX 75799, USA.
One sentence summary: Bacterial diversity associated with the basal fungus-farming ant Mycocepurus smithii colonies found along the Panama Canal.
Editor: Wietse de Boer

ABSTRACT
Fungus-farming ants (Hymenoptera: Formicidae, Attini) exhibit some of the most complex microbial symbioses because
both macroscopic partners (ants and fungus) are associated with a rich community of microorganisms. The ant and fungal
microbiomes are thought to serve important beneficial nutritional and defensive roles in these symbioses. While most
recent research has investigated the bacterial communities in the higher attines (e.g. the leaf-cutter ant genera Atta and
Acromyrmex), which are often associated with antibiotic-producing Actinobacteria, very little is known about the microbial
communities in basal lineages, labeled as ‘lower attines’, which retain the ancestral traits of smaller and more simple
societies. In this study, we used 16S amplicon pyrosequencing to characterize bacterial communities of the lower attine ant
Mycocepurus smithii among seven sampling sites in central Panama. We discovered that ant and fungus garden-associated
microbiota were distinct from surrounding soil, but unlike the situation in the derived fungus-gardening ants, which show
distinct ant and fungal microbiomes, microbial community structure of the ants and their fungi were similar. Another
surprising finding was that the abundance of actinomycete bacteria was low and instead, these symbioses were
characterized by an abundance of Lactobacillus and Pantoea bacteria. Furthermore, our data indicate that Lactobacillus strains
are acquired from the environment rather than acquired vertically.

Keywords: Host–microbe interaction; microbiota; coevolution; mutualism; Lactobacillus; Attines

INTRODUCTION 2015). Symbiotic bacteria allow specialization on a broad range

of diets (Douglas 2009; Feldhaar 2011; McFall-Ngai et al. 2013).
Insects are the most abundant and diverse taxon on earth, likely
Bacterial endosymbionts can help in digestion and detoxifica-
due to their ability to thrive in a wide range of ecological niches.
tion of food, provide essential nutrients (Cardoza, Klepzig and
The successful invasion of new habitats and feeding niches, ac-
Raffa 2006; Adams et al. 2009; Engel and Moran 2013; Dou-
companied by subsequent radiations, has been repeatedly facil-
glas 2014), contribute to nest hygiene (Currie et al. 1999, 2003;
itated through symbiosis with microorganisms (Douglas 2009,
Kaltenpoth et al. 2005; Feldhaar et al. 2007; Feldhaar and Gross

Received: 5 March 2015; Accepted: 19 June 2015


C FEMS 2015. All rights reserved. For permissions, please e-mail: [email protected]

1
 2 FEMS Microbiology Ecology, 2015, Vol. 91, No. 7

2008), and may play major roles in defense against pathogens (Kellner et al. 2013). In our study population in central Panama,
and parasites (Oliver et al. 2003; Hedges et al. 2008; Brownlie and we found a total of 11 ant clones and 9 fungus clone lineages in
Johnson 2009; Kaltenpoth 2009; Mattoso, Moreira and Samuels 52 colonies. Phylogenetic analyses suggested that ant colonies
2012). regularly exchange fungal cultivars or domesticate novel fungi
Microbiota of social insects (ants, termites, some bees and from free-living populations into symbiosis (Kellner et al. 2013).
some wasps) are of special interest because group living can in- Here we use 16S rRNA sequencing to characterize the bac-
crease the risk of disease spread, and social insects are thought terial communities associated with ants and fungus gardens of
to have evolved different techniques and associations with the lower attine ant M. smithii. We compare the ant- and fungus-
mutualistic bacteria to help with disease control (Koch and associated microbiota to bacterial communities from the sur-
Schmid-Hempel 2011; Konrad et al. 2012; Martinson, Moy and rounding soil, and explore the hypothesis that bacterial commu-
Moran 2012; Tragust et al. 2013a,b; Brütsch and Chapuisat 2014). nities are influenced by sample location rather than by either
Fungus-farming ants (tribe Attini) have become model systems ant or fungal genotype. This work provides the first quantita-
for studying microbial mutualisms, partly due to their impor- tive, culture-independent analyses of alpha- and beta-diversity

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

tance as dominant herbivores and agricultural pests (Hölldobler of bacterial communities of this basal attine ant, and the first
and Wilson 2011). Fungus-farming ants have to defend them- study to use soil samples from the ants’ immediate nest environ-
selves and their brood against diseases, but also their obligate ments as a non-symbiotic reference for comparison. We com-
fungus garden, the primary food source the ants feed on and pare the bacterial communities of M. smithii to those known from
which the ants protect from infections, parasitism and usurpa- more derived higher attines and from environmental sources,
tion by competing organisms. There has been extensive re- and conclude that many of the ant-associated bacteria are re-
search on higher attine ants (Atta, Acromyrmex, Trachymyrmex cruited into the symbiosis from external sources and do not ap-
and Sericomyrmex) to study the coevolution between mutualis- pear to be maintained through predominant vertical transmis-
tic ants, their fungi and associated bacterial communities (Sen sion between ant generations.
et al. 2009; Scott et al. 2010; Ishak et al. 2011a; Aylward et al. 2012,
2014). The best-studied bacterial symbionts in the higher attine
EXPERIMENTAL PROCEDURES
ants are Actinobacteria (actinomycetes), e.g. the genus Pseudono-
cardia, which is known to produce antibiotics against a diver- Sample collection
sity of bacterial and fungal microbes (Oh et al. 2009; Sen et al.
Field colonies
2009; Mueller et al. 2010). Recent studies have discovered several
Samples were collected during the excavation of whole M. smithii
other bacterial symbionts, which might be involved in defense
colonies in central Panama in April 2010, as described in Kellner
against parasites and diseases, among other functions, e.g. Strep-
et al. (2013). Single ants and fungus garden pieces (ca. 8 mm3 )
tomyces (Haeder et al. 2009; Barke et al. 2010; Seipke et al. 2011;
were collected with flame-sterilized forceps and stored in 100%
Seipke, Kaltenpoth and Hutchings 2012), Amycolatopsis (Sen et al.
ethanol. Two different types of soil samples were taken for each
2009) and Burkholderia (Santos et al. 2004). Additional bacterial
ant colony: one sample was from the ground surface (ca. 2 m
symbionts have been found which might play auxiliary roles
away from the colony entrance), in the following called ‘out-
in the ant-cultivated gardens, like lignocellulose-degrading and
side soil’. The other soil sample (‘nest wall soil’) was taken by
nitrogen-fixing bacteria, e.g. Pantoea, Klebsiella, Escherichia, Enter-
scraping soil from the walls of the fungus garden chambers. The
obacter, Citrobacter (Mueller et al. 2005; Pinto-Tomás et al. 2009;
samples analyzed and used in this study were 35 samples in to-
Scott et al. 2010; Aylward et al. 2012).
tal, collected from 16 different colonies: 4 outside soil samples,
Fungiculture in attine ants originated once approximately
14 chamber wall soil samples, 10 fungus garden samples and 7
50 million years ago (Schultz and Brady 2008) when the an-
worker ant samples. Due to soil contaminations of some fungus
cestral attine likely started feeding on nutrients produced by
gardens and ant samples, several samples had to be excluded,
leaf-litter-decomposing leucocoprinaceous fungi (Agaricaceae,
and amplification of DNA failed, leading to an unbalanced sam-
formerly Lepiotaceae) that the ants encountered in their nest
ple number (Table S1, Supporting Information). Collection loca-
environments and whose growth the ants began to sustain by
tions and genotypes of fungi and ants are described in Kellner
supplying such fungus with dead plant material (Mueller et al.
et al. (2013).
2001; De Fine Licht et al. 2010). A major transition from collect-
ing dead plant material to cutting fresh leaves as substrate for Garden fragments
fungal cultivation occurred 8–12 million years ago in the At- Because bacterial communities have been shown to vary spa-
tini (Schultz and Brady 2008). Extant leaf-cutter ants (Atta and tially in leaf-cutter ants (Suen et al. 2010) we examined the
Acromyrmex) cultivate a few closely related species of fungus spatial organization of a single M. smithii garden, which typi-
(Leucocoprinus sp.), while lower attine ants retained the more an- cally have a much different shape (long thin, hanging strands
cestral fungal associations with a great diversity of leucocopri- or curtains) than higher Attine gardens (oval-shaped; Seal and
naceous fungal species. Fungal strains cultivated by lower attine Tschinkel 2008). Accordingly, we took samples from a fun-
ants are still closely related or identical to free-living leucocopri- gus garden from colony RS100413–02 that was located approx-
naceous species, suggesting that some lower attine ants may oc- imately 30–40 cm below the surface (the chamber was approxi-
casionally adopt fungal symbionts from free-living populations mately 10 cm high). This garden was chosen because it appeared
(Mueller, Rehner and Schultz 1998; Vo, Mueller and Mikheyev healthy and above-average size. The garden was subdivided into
2009) and/or exchange symbionts through horizontal transmis- nine parts with flame-sterilized forceps, and a sample from each
sion (Green, Mueller and Adams 2002; Mueller et al. 2011; Kellner part (about 8 mm3 ) was preserved separately in 100% ethanol.
et al. 2013).
Mycocepurus smithii is an unusual attine ant since it is the only
DNA extraction and 454 sequencing
known asexually reproducing fungus-farming ant (Himler et al.
2009; Rabeling et al. 2009), such that each colony is comprised of Ethanol-preserved samples were stored at −80◦ C. In Jan-
a single ant clone tending a garden with a single fungus clone uary 2011, DNA extraction and sequencing was performed
 Kellner et al. 3

at the MRBI Biofilm Institute in Lubbock, Texas, using mission), we rather expect a strong influence of ant and fungus
bacterial tag-encoded FLX 454-pyrosequencing (bTEFAP) of genotypes on the bacterial community composition. Inclusion
16S amplicons from bacteria following previously described of soil samples in the analyses addresses the possibility that
standard protocols (Ishak et al. 2011b). 16S amplicons were specific bacterial lineages are contaminants derived from soil
amplified with primers Gray28F5 GAGTTTGATCNTGGCTCAG sources rather than an ant-engineered symbiont community.
and Gray519R 5 GTNTTACNGGGCKGCTG, spanning the V1–V3
hypervariable regions of the 16S rRNA gene. For ant samples, Sequence-based analyses
DNA was extracted from the entire insect; therefore, bacterial The Cd-hit output containing 7708 clusters was further pro-
communities represent both intestinal and surface bacteria. cessed in the mothur pipeline (Schloss et al. 2009) to gener-
Although we realize that integumental bacteria are likely ate an alignment using the SILVA database (Quast et al. 2013)
distinct from those found in the intestine, the small body size as template. The alignment was then filtered (soft filter =
of these ants (abdominal length <0.4 mm) and the field location 5%), and poorly aligned sequences were removed (screen.seqs,
prevented us from dissecting out guts under sterile conditions. >250 bp). The final alignment contained 7521 unique sequences.

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

Using B2C2 (Gontcharova et al. 2010), short reads (<300 bp) and An approximate maximum-likelihood tree was constructed us-
chimeras were depleted. Sequences were assigned to bacterial ing FastTree (Price, Dehal and Arkin 2010), which served as input
types via BLAST at a minimum identity of 75% using a reference for Fast UniFrac (Hamady, Lozupone and Knight 2009). Pairwise
database curated by MRBI, and compared to hits obtained from unweighted and weighted UniFrac distances were investigated
SILVA database (Quast et al. 2013). Using taxonomic identities by NMDS ordination in PAST (Hammer, Harper and Ryan 2001).
at a 97% sequence shared identity, we used simple resampling Using vegan (Oksanen et al. 2015) in R 2.13.0, we ran MRPP us-
scripts to obtain species accumulation (rarefaction) curves for ing unweighted UniFrac distances to test for significant differ-
each sample (Fig. S1, Supporting Information). Sequence data ences among the four habitats (outside soil, nest wall soils, ants
is deposited in the GenBank/NCBI short read archive under and fungus gardens; P values were obtained with 1000 permuta-
accession number SRA055805. tions). Because we were investigating community compositions
and phylogenetic relationships between very different habitats
Data analyses (soils, fungus gardens, worker ants), we expected qualitative dif-
ferences rather than abundance differences in bacterial commu-
OTU-based analyses nity composition, and we therefore chose unweighted UniFrac
The total number of sequence reads generated from 35 sam- distances and Jaccard distances based on presence/absence ma-
ples was 248 358. To standardize the sequence reads between trices as appropriate distance measure (for a quantitative anal-
samples (ranging from 1257 sequences in one worker sample yses based on OTU abundances using Bray–Curtis distances and
to 16 699 sequences in one soil sample, Table S1, Supporting weighted UniFrac distances, see Supplementary material).
Information), we randomly subsampled 2000 sequences from
each sample or the total number of sequence reads available if
less than 2000 were present, resulting in a total of 85 900 se-
Spatial variation in the bacterial community of a
quence reads from 35 samples. Using the software Cd-hit (Li fungus garden
and Godzik 2006; Schmidt, Matias Rodrigues and von Mering The total amount of sequence reads after initial quality screen
2014), the sequences were clustered into unique OTUs (opera- was 45 884 summed over nine subsamples from one garden,
tional taxonomic units) at a 97% shared identity level, yielding with 2898 sequences the lowest and 7248 sequences the high-
a total of 7708 OTUs. OTUs with less than four sequences were est count. The raw sequences of the nine garden fragments
excluded to parse out rare OTUs which could be due to contami- were processed as described above. Briefly, from each sample,
nation, reducing the number of OTUs to 860. These OTUs served 3000 sequences were drawn randomly (except 2898 sequences
as initial response abundance matrix in PC-ORD (McCune and were used for the smallest sample) and clustered using Cd-hit
Mefford 2006). The response abundance matrix was transformed into 2432 unique OTUs (97% shared identity threshold). Single-
into a presence/absence matrix. To investigate relationships of tons were removed, leaving 552 OTUs. These OTUs were used
the fungus gardens, workers, chamber wall soil and outside soil as abundance response matrix in PC-ORD. All OTUs which were
to each other, we used non-metric multidimensional scaling counted less than four times were removed, leaving 256 OTUs
(NMDS) as ordination method, choosing Jaccard distances as ap- in the abundance response matrix. Cluster Analyses (UPGMA)
propriate distance measurement. The same analyses were run was performed using Bray–Curtis distances and weighted (nor-
on an OTU abundance matrix based on Bray–Curtis distances. malized) UniFrac distances (obtained through FastUniFrac as de-
An indicator species analysis was performed to determine scribed above). Because OTU composition and phylogenetic re-
the most influential OTUs based on their relative abundance lationships of very similar communities (fragments from one
and frequency within the four defined groups (Dufrene and garden) were investigated, quantitative rather than qualitative
Legendre 1997). We used a multiresponse permutation proce- differences were expected. Therefore, weighted (normalized)
dure (MRPP, as implemented in PC-ORD) to test for differences UniFrac distances and Bray–Curtis distances (based on an abun-
between the sample types (outside soil, nest wall soil, fungus dance matrix) were chosen as appropriate distance measures.
gardens, ants), and between samples from different sampling Cluster analyses grouped single samples based on similarities
locations in central Panama (Gamboa/Apartment 183, Achiote, to each other without knowledge of predefined group member-
Casa Verde, Gatun, Gamboa/Greenhouse and Gamboa/Resort). ships.
MRPP was also used to test for differences between bacterial
communities grouped by ant and fungus genotypes (ant and
fungus lineages were previously genotyped in Kellner et al. 2013). RESULTS
If symbiont recruitment (horizontal transmission) is the main
Bacterial taxonomy and distribution
driver of bacterial community composition, we expect a strong
effect of sampling location. If bacterial symbionts are inherited A total of 304 152 bacterial 16S sequence reads were col-
and passed on from mother to daughter colonies (vertical trans- lected and classified (total number of sequences for soil: 34 235;
 4 FEMS Microbiology Ecology, 2015, Vol. 91, No. 7

Table 1. Core microbiota of M. smithii worker ants, fungus gardens, nest wall soil and outside soil shown as the percentage of bacterial read abun-
dance. Only genera found in more than 1% abundance are shown. (For a complete list of bacterial genera, see Table S2, Supporting Information.)

Sample type Bacteria genera % Sample type Bacteria genera %

Garden Lactobacillus 19.3231 Worker ants Lactobacillus 56.8517

Pantoea 15.9296 Lysobacter 3.8097
Pseudonocardia 4.8448 Pseudonocardia 3.1987
Stenotrophomonas 4.5561 Ruminococcus 3.0402
Erwinia 4.4588 Xanthomonas 2.6329
Escherichia 4.0075 Staphylococcus 2.3537
Amycolatopsis 3.4163 Brevibacterium 2.1274
Serratia 3.0668 Escherichia 1.6521
Enterobacter 2.6747 Enterococcus 1.5880

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

Microlunatus 2.6549 Bacteroides 1.4975
Ruminococcus 2.0030 Aminobacter 1.4711
Cronobacter 1.7796 Stenotrophomonas 1.4183
Pectobacterium 1.7082 Rhodoferax 1.0826
Pseudomonas 1.3556 Microbacterium 1.0411
Derxia 1.1139 Total unknown 4.6471
Clostridium 1.0699
Lysobacter 1.0684
Total unknown 4.9801

Sample type Bacteria genera % Sample type Bacteria genera %

Outside soil Patulibacter 10.1942 Nest wall soil Nitrosovibrio 6.7023

Acidobacterium 8.0852 Acidobacterium 5.3395
Nitrosovibrio 6.8438 Pseudonocardia 4.0251
Solirubrobacter 3.0407 Streptomyces 3.8248
Rhodoplanes 2.8976 Mycobacterium 3.2674
Caldilinea 2.5792 Patulibacter 2.4752
Brevibacillus 2.2725 Caldilinea 2.1246
Ruminococcus 2.0272 Rhizobium 2.1131
Stenotrophomonas 1.9308 Nitrospira 2.0992
Rhizobium 1.9308 Nocardioides 1.8299
Bacillus 1.8402 Solirubrobacter 1.5926
Clostridium 1.6241 Derxia 1.4843
Paenibacillus 1.5598 Bradyrhizobium 1.4753
Holophaga 1.4430 Rhodomicrobium 1.3217
Anaeromyxobacter 1.3699 Rhodoplanes 1.2979
Nitrospira 1.2414 Holophaga 1.2979
Actinocatenispora 1.0282 Kribbella 1.0763
Total unknown 29.7647 Staphylococcus 1.0484
Total unknown 18.8261

chamber walls: 121 810; worker: 26 511, garden: 65 802, garden samples (complete taxonomical listing in Table S1, Supporting
fragments: 55 794). Outside and chamber wall soils contained Information).
the highest proportions of unclassified sequences (soil: 29.76%, Rarefaction analyses (at 97% identity threshold) indicated
chamber wall soil: 18.83%), whereas gardens and ant work- that most of the samples were adequately sampled (Fig. S1,
ers had the lowest proportions (gardens: 4.98%, worker ants: Supporting Information). Across all samples, nest wall soil con-
4.65%). tained the highest number of bacteria genera (534 bacteria gen-
The most common bacteria orders found were Lactobacil- era), followed by fungus gardens (419), outside soil (281) and ant
lales (ants: 56%; gardens: 26%), Enterobacteriales (ants: 6%; gar- workers (123). Table 1 contains the most common bacteria gen-
dens: 30%), Actinomycetales (ants: 10%; gardens: 15%), Xan- era (the core microbiota) found in the four sample types.
thomonadales (ants: 7%; gardens: 6%), Clostridiales (ants: 8%;
gardens: 6%), Rhizobiales (ants: 2%; garden: 5%) and Burkholde- Comparison of ants and fungus gardens vs their
riales (ants: 3%; gardens: 2%). Some of these orders were also surroundings
represented at significant proportions in the two soil types, e.g.
Rhizobiales (nest walls: 11%; soil: 16%), Xanthomonadales (nest Soil types (outside soil and nest walls) had a significantly higher
walls: 2%; soil: 2%), Actinomycetales (nest walls: 24%; not found average richness of OTUs (Kruskal–Wallis test: H (3, N = 35) =
in soil). The two soil types contained their own core microbiota, 18.35, P = 0.004) and Shannon’s diversity (Kruskal–Wallis test:
e.g. Nitrosomonadales (nest walls: 9%; soil:12%), Acidobacteri- H (3, N = 35) = 19.52, P = 0.002) than fungus gardens and
ales (nest walls: 7%; soils: 12%) and Solirubrobacteriales (nest workers. On average, soil samples (202.7 OTUs) contained four
walls: 5%; soils:16%), which were not found in ants or garden times the number of OTUs of workers (52.1 OTUs) and gardens
 Kellner et al. 5

Table 2. MRPP test of significance of pairwise comparisons of bacte- Garden fragments

rial communities associated with M. smithii worker ants, fungus gar-
dens, nest wall soil and outside soil. Upper matrix: A values, lower Fungus gardens of the ant M. smithii are not homogeneous
matrix P values. Significant P values are denoted in bold (α level 0.05, in bacterial community composition. Species richness of frag-
sequential Bonferroni correction).
ments ranged from 62.0 to 168.0, and diversity (Shannon’s in-
dex) ranged from 3.11 to 4.16, with a tendency of fungus frag-
(a) OTU-based Jaccard distances, (b) Sequence-based UniFrac
ments surrounded by others having a lower richness and diver-
distances;
sity than garden pieces on the edge, facing the nest walls (‘Side 3’
Outside Nest wall Fungus piece had both lowest richness and diversity, while ‘Edge 1’ had
a) soil soil gardens Ants the highest values). Cluster analyses of garden fragments based
on OTU abundances and UniFrac distances resulted both in five
Outside soil −0.001 0.360 0.397 clusters (grouping the fragments into five clusters yielded the
Nest wall soil 0.448 0.459 0.456 highest number of significant indicator OTUs for each group)
Fungus gardens <0.00001 <0.00001 0.030

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

(see Fig. 2). As expected, differences in microbial communi-
Ants <0.00001 <0.00001 0.200
ties of garden fragments were more quantitative than quali-
Outside Nest wall Fungus tative, meaning differences between fragments were based on
b) soil soil gardens Ants the abundances of OTUs rather than presence or absence of
OTUs. Accordingly, diversity indices were not significantly dif-
Outside soil 0.003 0.067 0.121 ferent among the five groups (Kruskall–Wallis, richness: H(4,
Nest wall soils 0.239 0.069 0.103
N = 9) = 3,47, P = 0.48, Shannon’s diversity: H(4, N = 9) = 7,47,
Fungus gardens <0.001 <0.001 0.009
P = 0.11). The most common bacteria genera in all five groups of
Ants <0.01 <0.001 0.099
fragments found were Staphylococcus, Bacteriodes, Pseudomonas,
Clostridium, Ruminococcus, Propionibacterium, Pantoea and Lacto-
bacillus (genera abundances in Table S6, Supporting Informa-
tion).

(57.7 OTUs). The bacterial communities associated with ants

and fungus gardens were clearly distinct from outside soils and
DISCUSSION
nest walls. Overall differences between the four groups were
statistically significant (Jaccard distances, MRPP, A = 0.486, P Our analysis of bacterial communities of M. smithii colonies from
< 0.0001), post-hoc analysis revealed significant differences be- the field shows that colonies of ants and fungus gardens are dis-
tween soil and garden, and between soil and ants, but no differ- tinct from their surroundings, and that the microbiome of ants
ences between nest walls and outside soil, and no differences and fungus gardens is mostly dominated by the genera Lacto-
between ants and gardens. UniFrac analyses yielded a similar re- bacillus and Pantoea. Another discovery was the lack of signifi-
sult (unweighted UniFrac analysis, MRPP, A = 0.092, P < 0.001; see cant differentiation between ant- and garden-associated bacte-
Table 2 for details). Ordination of samples based on UniFrac and rial communities, which contrasts with the findings of similar
Jaccard distances supports these results (Fig. 1a and b) with out- analyses on the higher derived attine Trachymyrmex septentrion-
side soil and nest wall soil clustering together while separated alis (Ishak et al. 2011a), where ant- and garden-associated bacte-
from a second cluster containing fungus gardens and worker ria formed distinct communities. These results suggest that M.
ants. Analyses with Bray–Curtis distances based on OTU abun- smithii ants and symbiotic fungi may exchange bacteria when
dances and weighted UniFrac distances revealed a similar pat- the ants move through their gardens or ingest garden material
tern (Fig. S2 and Table S3, Supporting Information). with the result that the bacterial communities are homogenized
Indicator species analyses of OTU composition revealed a to- to a greater extent than in the higher attine T. septentrionalis. My-
tal of 206 significant indicator values for the soil samples, 23 for cocepurus smithii bacterial communities were also clearly sepa-
chamber walls, 10 for fungus gardens and 21 for ants. According rated from the surrounding nest walls and adjunct soil from out-
to indicator species analyses (Dufrene and Legendre 1997), these side the nests; thus, the bacteria found on ants and gardens are
are the OTUs that contribute most to the separation of the test not the result of accidental contamination from environmen-
groups. A total of 15 of the 21 significant ant-associated indica- tal sources such as the surrounding soil. This suggests that the
tor species were different Lactobacillus OTUs. Among the 10 sig- ants are able to shape their own microbial environment, and
nificant fungus gardens OTUs, one Lactobacillus OTU was found, most likely control the microbial communities on their cuticle
together with OTUs of Pantoea, Stenotrophomonas, Erwinia and Ser- and on the gardens by grooming off unwanted harmful bacteria
ratia (Table S4, Supporting Information). (Fernández-Marı́n et al. 2009).
OTUs of ants, fungus gardens, and the two soil types are de- The finding that sample location rather than genotype has
termined by collection location (MRPP, six locations, A = 0.271, an influence on the bacterial communities of colonies indicates
P = 0.0002) because significant differences in bacterial com- that horizontal transmission of bacterial symbionts might play
munity composition were found between samples from Gam- a more important role than vertical transmission. No evidence
boa/Apartment 183 and several other locations (Table S5, Sup- of correlation between bacteria strains and ant or fungus ge-
porting Information). When soil types were excluded, fungus netic lineages was found, although vertical transmission of fun-
garden and ant samples differed significantly between two of the gal strains among ant generations undoubtedly occurs (Kellner
three compared locations. The genetic background of ant and et al. 2013), and bacteria might also be cotransmitted. The bac-
fungus gardens (genotypes analyzed in Kellner et al. 2013) did terial communities of M. smithii were rather diverse in some
not significantly influence the OTU/bacterial community that genera but depauperate in others. Many genera such as Pantoea,
they harbored (MRPP, comparing four ant genotypes, A = –0.022, Klebsiella, Stenotrophomonas and Serratia have also been found in
P = 0.567; comparing four fungus genotypes, A = 0.061, P = 0.190) fungus gardens of leaf-cutter ants (Atta and Acromyrmex; Scott
(Details in Table S5, Supporting Information). et al. 2010; Suen et al. 2010; Aylward et al. 2012). In particular,
 6 FEMS Microbiology Ecology, 2015, Vol. 91, No. 7

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

Figure 1. NMDS representation of bacterial communities of M. smithii fungus gardens, worker ants, nest wall soils and outside soils. Distances are based on dissimilarity
matrices of (a) OTU-based Jaccard distances and (b) sequence-based unweighted UniFrac distances.

Figure 2. Dendrogram of bacterial communities of fungus garden fragments (UPGMA: unweighted pair group method with arithmetic mean) using weighted UniFrac
(above) and OTU-based Bray–Curtis distances (below). The scale bar represents percentage of information. Clustering the fragments into five groups at 60% information
(indicated with the horizontal dashed line) yielded the highest number of significant indicator values (27 significant indicators, P < 0.05, in five or less clusters). The
drawing depicts the relative positions of fungus garden fragments (depicted by polygons) as they were collected in the field.
 Kellner et al. 7

Pantoea and Klebsiella are thought to serve as symbiotic nitrogen- symbionts from their environment rather than through vertical
fixing agents in gardens of Atta cephalotes (Pinto-Tomás et al. transmission between honeybee generations (McFrederick et al.
2009), and they might play a similar role in gardens of M. smithii. 2012, 2013). The same is probably true for M. smithii, because
One of the hallmark features of the attine symbiosis that ap- Lactobacillus strains associated with M. smithii do not form a
pears to be relatively rare in the M. smithii symbiosis are mu- host-specific clade (McFrederick et al. 2013). Likewise, Lactobacilli
tualistic Actinobacteria (actinomycetes), which are thought to might serve an important role as defensive agents in M. smithii,
serve important defensive roles [Pseudonocardia: (Currie et al. as it has been shown in fungus-growing termites (Long et al.
1999; Oh et al. 2009); Streptomyces: (Haeder et al. 2009; Barke et al. 2010; Mathew et al. 2012), where they suppress the growth of the
2010; Seipke et al. 2011)]. Actinobacteria are well known from parasitic Trichoderma fungus and thus protect the termites’ fun-
other insect systems, including wasps (Madden et al. 2013), bee gus combs (Termitomyces). As a result, it is possible that the usage
wolves (Kaltenpoth et al. 2010; Kroiss et al. 2010) and bark beetles of insect feces and other sources of Lactobacilli not only serve as
(Dendroctonus frontalis—Streptomyces; Cardoza, Klepzig and Raffa substrate or fertilizer for the fungus gardens, but the application
2006; Scott et al. 2008). The Actinobacteria Pseudonocardia and of Lactobacilli might represent a defensive strategy alternative to

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

Streptomyces, and the Proteobacterium Burkholderia, were found actinomycetes employed by most attines (Fernández-Marı́n et al.
only in small quantities in M. smithii ants and fungus gardens, 2009).
which suggests that these bacteria likely play less important Our analysis of fragments of a single fungus garden shows
roles than in the well-studied Trachymyrmex and Acromyrmex that microbial communities within M. smithii gardens are not
ants, which are known to use antibiotics produced by these homogeneous. Gardens of M. smithii have a hanging structure,
bacteria (Santos et al. 2004; Seipke et al. 2011). Because DNA with curtain-like strands hanging from the ceiling of the nest
was extracted from whole ants, we were not able to determine cave, which is unlike the relatively solid globule-shaped gardens
from which parts of the ant body the bacterial OTUs reside and typical of most leaf-cutting ants and other higher derived at-
screening of separate ant integument and ant intestines might tines (Seal and Tschinkel 2008). Within a single strand, the fun-
reveal spatial partitioning of the microbiomes. gus grows from top to bottom (the ants elongate the strand as
Different strains of Pseudonocardia were discovered in a previ- it grows). Because M. smithii in the field and in the lab start new
ous study on M. smithii (Sen et al. 2009), but samples in that study gardens in the center of a chamber ceiling rather than near the
were taken from colonies maintained in the laboratory for sev- ceiling periphery, it is likely that each garden has an age profile
eral years, unlike the data in the present study which were taken from the inside to the outside (central older parts, peripheral
from freshly collected field colonies. This might also explain the younger parts). The nine garden fragments analyzed clustered
discrepancies between the current dataset and that from Sen into five likely groups, with the central (presumably older) parts
et al. (2009) who reported high Pseudonocardia but low Lactobac- forming one of these groups. The non-homogenous nature of
cillus abundance which is the opposite of what we report here. Mycocepurus gardens parallels the nature of derived leaf-cutter
Nevertheless, we raise the possibility that M. smithii along with ant gardens, which also are partitioned into younger (top), ma-
Atta and Sericomyrmex represent ant lineages that may be less ture (center) and older portions (bottom) (Suen et al. 2010). Other
reliant on Pseudonocardia bacteria as a form of biological control factors likely to influence garden spatial microbial communities
(Hughes et al. 2008; Fernández-Marı́n et al. 2009). are the locations where the ants add new substrate (the out-
The strong presence of different strains from the genus Lac- side), and where they place their brood (toward the inside, and
tobacillus might be explained by M. smithii collecting vertebrate towards the top of gardens). Pantoea was found in significant
or insect feces as substrate for fungus cultivation, which many amounts only in two fragments facing the outside of the gar-
attine species are known to collect (Leal and Oliveira 2000; Seal den suggesting that this garden region might have been where
and Tschinkel 2007, 2008). The Lactobacillus strains found in the ants recently added garden substrate. Similarly, while not
M. smithii were identified as well-known strains from human very common (relative abundances <5% in all nine garden frag-
and bird intestinal tracts (e.g. Lactobacillus gallinarium) rather ments), Lactobacillus was found in the highest amounts near the
than ant or attine specific (McFrederick et al. 2013). Lactobacillus edges. The importance of the different bacteria genera and their
generally have anti-microbial properties, which are well-known roles in digestion and nutrient processing, nitrogen fixation, lar-
commensals of humans, insects and other animals, and are val nutrition and disease defense remains unknown, and gives
important for the food industry and the fermentation of dairy opportunity to further investigations.
products (Hammes and Vogel 1995). Lactobacillus exerts their an-
timicrobial properties by secreting lactic acid to acidify their
environment—conditions that few other bacteria (and fungi) can SUPPLEMENTARY DATA
tolerate. The appearance of Lactobacillus is well studied in wild
bees (McFrederick et al. 2012, 2013, 2014; McFrederick, Mueller Supplementary data is available at FEMSEC online.
and James 2014) and honey bees (Jeyaprakash, Hoy and All-
sopp 2003; Vásquez et al. 2012) where it plays a key role in
honey production (Olofsson and Vásquez 2008) and beebread
ACKNOWLEDGEMENTS
(Vásquez and Olofsson 2009). For example, Apis mellifera colonies
treated with L. johnsonii increased honey production while low- We thank the Autoridad Nacional del Ambiente de Panamá,
ering Varroa and Nosema infection, and induced swarming (Audi- the Smithsonian Tropical Research Institute, and Orelis Arose-
sio, Sabate and Benitez-Ahrendts 2015). Lactobacillus bacteria are mena and Annette Aiello for collecting and export permits.
also thought to be major candidates inhibiting the bacteria Paeni- We thank Jon N Seal, Quinn McFrederick and three anony-
bacillus larvae, which is the cause of American foulbrood disease mous reviewers whose comments and criticism improved the
infecting honeybee colonies (Forsgren et al. 2010). Sweat bees manuscript. Ruchira Sen helped with field collections. DNA ex-
also associate with Lactobacillus, but unlike honeybees do not traction and 454-sequencing was performed by Scot Dowd at the
have a bee-specific Lactobacillus clade, but acquire acidophilic MRBI Biofilm Institute in Lubbock, Texas.
 8 FEMS Microbiology Ecology, 2015, Vol. 91, No. 7

FUNDING Fernández-Marı́n H, Zimmermann JK, Nash DR, et al. Reduced

biological control and enhanced chemical pest management
This work was supported by the National Science Foundation
in the evolution of fungus farming ants. P R Soc B 2009;
[DEB-0919519] to TAL and UGM, and by the W.M. Wheeler Lost
276:2263–9.
Pines Endowment.
Forsgren E, Olofsson TC, Vásquez A, et al. Novel lactic acid bac-
teria inhibiting Paenibacillus larvae in honey bee larvae. Api-
Conflict of interest. None declared.
dologie 2010;41:99–108.
Gontcharova V, Youn E, Wolcott RD, et al. Black box chimera
REFERENCES check (B2C2): a windows-based software for batch depletion
of chimeras from bacterial 16S rRNA gene datasets. Open Mi-
Adams AS, Currie CR, Cardoza Y, et al. Effects of symbiotic crobiol J 2010;4:47–52.
bacteria and tree chemistry on the growth and reproduc- Green AM, Mueller UG, Adams RMM. Extensive exchange of fun-
tion of bark beetle fungal symbionts. Can J For Res 2009;39: gal cultivars between sympatric species of fungus-growing

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

1133–47. ants. Mol Ecol 2002;11:191–5.
Audisio MC, Sabate DC, Benitez-Ahrendts MR. Effect of Lacto- Haeder S, Wirth R, Herz H, et al. Candicidin-producing Strepto-
bacillus johnsonii CRL1647 on different parameters of honey- myces support leaf-cutting ants to protect their fungus gar-
bee colonies and bacterial populations of the bee gut. Benef den against the pathogenic fungus Escovopsis. P Natl Acad Sci
Microbes 2015;1:1–10, DOI: 10.3920/bm2014.0155:1-10. USA 2009;106:4742–6.
Aylward FO, Burnum KE, Scott JJ, et al. Metagenomic and Hamady M, Lozupone C, Knight R. Fast UniFrac: facilitating high-
metaproteomic insights into bacterial communities in leaf- troughput phylogenetic analyses of microbial communities
cutter ant fungus gardens. ISME J 2012;6:1688–701. including analysis of pyrosequencing and PhyloChip data.
Aylward FO, Suen G, Biedermann PHW, et al. Convergent bacte- ISME J 2009;4:17–27.
rial microbiotas in the fungal agricultural systems of insects. Hammer Ø, Harper DAT, Ryan PD. PAST: Palaeontological Statis-
mBio 2014;5:e02077–14. tics software package for education and data analysis.
Barke J, Seipke R, Gruschow S, et al. A mixed community Palaeontol Electron 2001;4:9.
of actinomycetes produce multiple antibiotics for the fun- Hammes WP, Vogel RF. The genus Lactobacillus. In: Wood BJB,
gus farming ant Acromyrmex octospinosus. BMC Biol 2010;8: Holzapfel WH (eds). The Genera of Lactic Acid Bacteria. London:
109. Chapman & Hall, 1995, 19–54.
Brownlie JC, Johnson KN. Symbiont-mediated protection in in- Hedges LM, Brownlie JC, O’Neill SL, et al. Wolbachia and virus pro-
sect hosts. Trends Microbiol 2009;17:348–54. tection in insects. Science 2008;322:702.
Brütsch T, Chapuisat M. Wood ants protect their brood with tree Himler AG, Caldera EJ, Baer B, et al. No sex in fungus-
resin. Anim Behav 2014;93:157–61. farming ants or their crops. P Roy Soc B 2009;276:
Cardoza YJ, Klepzig KD, Raffa KF. Bacteria in oral secretions of 2611–6.
an endophytic insect inhibit antagonistic fungi. Ecol Entomol Hölldobler B, Wilson EO. The Leafcutter Ants. New York, London:
2006;31:636–45. WW Norton, 2011.
Currie C, Scott J, Summerbell R, et al. Fungus-growing ants use Hughes WOH, Pagliarini R, Madsen HB, et al. Antimicrobial de-
antibiotic-producing bacteria to control garden parasites. fense shows an abrupt evolutionary transition in the fungus-
Nature 1999;398:701–5. growing ants. Evolution 2008;62:1252–7.
Currie CR, Wong B, Stuart AE, et al. Ancient tripartite coevolu- Ishak HD, Miller JL, Sen R, et al. Microbiomes of ant
tion in the attine ant-microbe symbiosis. Science 2003;299: castes implicate new microbial roles in the fungus-
386–8. growing ant Trachymyrmex septentrionalis. Sci Rep 2011a;
De Fine Licht HH, Schiøtt M, Mueller UG, et al. Evolutionary tran- 1:204.
sitions in enzyme activity of ant fungus gardens. Evolution Ishak HD, Plowes R, Sen R, et al. Bacterial diversity in Solenop-
2010;64:2055–69. sis invicta and Solenopsis geminata ant colonies character-
Douglas AE. The microbial dimension of insect nutritional ecol- ized by 16S amplicon 454 pyrosequencing. Microbiol Ecol
ogy. Funct Ecol 2009;23:38–47. 2011b;61:821–31.
Douglas AE. The molecular basis of bacterial-insect symbiosis. J Jeyaprakash A, Hoy MA, Allsopp MH. Bacterial diversity in
Mol Biol 2014;426:3830–7. worker adults of Apis mellifera capensis and Apis mellifera
Douglas AE. Multiorganismal insects: diversity and function scutellata (Insecta: Hymenoptera) assessed using 16S rRNA
of resident microorganisms. Annu Rev Entomol 2015;60: sequences. J Invertebr Pathol 2003;84:96–103.
17–34. Kaltenpoth M. Actinobacteria as mutualists: general
Dufrene M, Legendre P. Species assemblages and indicator healthcare for insects? Trends Microbiol 2009;17:
species: the need for a flexible asymmetrical approach. Ecol 529–35.
Monogr 1997;67:345–66. Kaltenpoth M, Göttler W, Herzner G, et al. Symbiotic bacte-
Engel P, Moran NA. The gut microbiota of insects - diversity in ria protect wasp larvae from fungal infestation. Curr Biol
structure and function. FEMS Microbiol Rev 2013;37:699–735. 2005;15:475–9.
Feldhaar H. Bacterial symbionts as mediators of ecologi- Kaltenpoth M, Schmitt T, Polidori C, et al. Symbiotic Strepto-
cally important traits of insect hosts. Ecol Entomol 2011;36: mycetes in antennal glands of the South American digger
533–43. wasp genus Trachypus (Hymenoptera, Crabronidae). Physiol
Feldhaar H, Gross R. Immune reactions of insects on bacterial Entomol 2010;35:196–200.
pathogens and mutualists. Microbes Infect 2008;10:1082–8. Kellner K, Fernandez-Marin H, Ishak HD, et al. Co-evolutionary
Feldhaar H, Straka J, Krischke M, et al. Nutritional upgrading for patterns and diversification of ant-fungus associations in the
omnivorous carpenter ants by the endosymbiont Blochman- asexual fungus-farming ant Mycocepurus smithii in Panama. J
nia. BMC Biol 2007;5:48. Evol Biol 2013;26:1353–62.
 Kellner et al. 9

Koch H, Schmid-Hempel P. Socially transmitted gut microbiota Mueller UG, Scott JJ, Ishak HD, et al. Monoculture of leafcutter
protect bumble bees against an intestinal parasite. P Natl ant gardens. PLoS One 2010;5:e12668.
Acad Sci USA 2011;108:19288–92. Oh D, Poulsen M, Currie C, et al. Dentigerumycin: a bacterial
Konrad M, Vyleta ML, Theis FJ, et al. Social transfer of pathogenic mediator of an ant-fungus symbiosis. Nat Chem Biol 2009;
fungus promotes active immunisation in ant colonies. PLoS 5:391–3.
Biol 2012;10:e1001300. Oksanen J, Guillaume Blanchet F, Kindt R, et al. vegan: Community
Kroiss J, Kaltenpoth M, Schneider B, et al. Symbiotic Strep- Ecology Package. R package version 2.2-1, 2015, http://CRAN.
tomycetes provide antibiotic combination prophylaxis for R-project.org/package=vegan.
wasp offspring. Nat Chem Biol 2010;6:261–3. Oliver KM, Russell JA, Moran NA, et al. Facultative bacterial sym-
Leal IR, Oliveira PS. Foraging ecology of attine ants in a Neotrop- bionts in aphids confer resistance to parasitic wasps. P Natl
ical savanna: seasonal use of fungal substrate in the cerrado Acad Sci USA 2003;100:1803–7.
vegetation of Brazil. Insect Soc 2000;47:376–82. Olofsson TC, Vásquez A. Detection and identification of a
Li W, Godzik A. Cd-hit: a fast program for clustering and compar- novel lactic acid bacterial flora within the honey stom-

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020

ing large sets of protein or nucleotid sequences. Bioinformatics ach of the honeybee Apis mellifera. Curr Microbiol 2008;57:
2006;22:1658–9. 356–63.
Long YH, Xie L, Liu N, et al. Comparison of gut-associated Pinto-Tomás AA, Anderson MA, Suen G, et al. Symbiotic nitro-
and nest-associated microbial communities of a fungus- gen fixation in the fungus gardens of feaf-cutter ants. Science
growing termite (Odontotermes yunnanensis). Insect Sci 2010;17: 2009;326:1120–3.
265–76. Price MN, Dehal PS, Arkin AP. FastTree2 - approximately
McCune B, Mefford MJ. PC-ORD. Multivariate Analysis of Ecological maximum-likelihood trees for large alignments. PloS One
Data. Version 5.15. Gleneden Beach, Oregon, USA: MjM Soft- 2010;5:e9490.
ware, 2006. Quast C, Pruesse E, Yilmaz P, et al. The SILVA ribosomal RNA gene
McFall-Ngai M, Hadfield MG, Bosch TCG, et al.. Animals in a bac- database project: improved data processing and web-based
terial world, a new imperative for the life sciences. P Natl tools. Nucleic Acids Res 2013;41:D590–6.
Acad Sci USA 2013;110:3229–36. Rabeling C, Lino-Neto J, Cappellari SC, et al. Thelytokous
McFrederick QS, Cannone JJ, Gutell RR, et al. Specificity be- parthenogenesis in the fungus-gardening ant Mycocepurus
tween Lactobacilli and Hymenopteran hosts is the ex- smithii (Hymenoptera: Formicidae). PLos One 2009;4:e6781.
ception rather than the rule. Appl Environ Microb 2013;79: Santos AV, Dillon RJ, Dillon VM, et al. Occurrence of the antibi-
1803–12. otic producing bacterium Burkholderia sp. in colonies of the
McFrederick QS, Mueller UG, James RR. Interactions between leaf-cutting ant Atta sexdens rubropilosa. FEMS Microbiol Lett
fungi and bacteria influence microbial community struc- 2004;239:319–23.
ture in the Megachile rotundata larval gut. P R Soc B Schloss PD, Westcott SL, Ryabin T, et al. Introducing mothur:
2014;281:20132653. open-source, platform-independent, community-supported
McFrederick QS, Wcislo WT, Hout M, et al. Host species and software for describing and comparing microbial communi-
developmental stage, but not host social-structure, affects ties. Appl Environ Microb 2009;75:7537–41.
bacterial community structure in socially polymorphic bees. Schmidt TSB, Matias Rodrigues JF, von Mering C. Limits
FEMS Microbiol Ecol 2014;2:398–406. to robustness and reproducibility in the demarcation of
McFrederick QS, Wcislo WT, Taylor DR, et al. Environment or operational taxonomic units. Environ Microbiol 2014;17:
kin: whence do bees obtain acidophilic bacteria? Mol Ecol 1689–706.
2012;21:1754–68. Schultz TR, Brady SG. Major evolutionary transitions in ant agri-
Madden AA, Grassetti A, Soriano JA, et al. Actinomycetes culture. P Natl Acad Sci USA 2008;105:5435–40.
with antimicrobial activity isolated from paper wasp (Hy- Scott JJ, Budsberg KJ, Suen G, et al. Microbial community struc-
menoptera: Vespidae: Polistinae) nests. Environ Entomol ture of leaf-cutter ant fungus gardens and refuse dumps. Plos
2013;42:703–10. One 2010;5:e9922.
Martinson VG, Moy J, Moran NA. Establishment of characteris- Scott JJ, Oh D-C, Yuceer MC, et al. Bacterial protection of beetle-
tic gut bacteria during development of the honeybee worker. fungus mutualism. Science 2008;322:63.
Appl Environ Microb 2012;78:2830–40. Seal JN, Tschinkel WR. Energetics of newly-mated queens and
Mathew GM, Ju YM, Lai CY, et al. Microbial community analy- colony founding in the fungus-gardening ants Cyphomyrmex
sis in the termite gut and fungus comb of Odontotermes for- rimosus and Trachymyrmex septentrionalis (Hymenoptera:
mosanus: the implication of Bacillus as mutualists. FEMS Mi- Formicidae). Physiol Entomol 2007;32:8–15.
crobiol Ecol 2012;79:504–17. Seal JN, Tschinkel WR. Food limitation in the fungus-gardening
Mattoso TC, Moreira DDO, Samuels RI. Symbiotic bacteria on the ant, Trachymyrmex septentrionalis. Ecol Entomol 2008;33:597–
cuticle of the leaf-cutting ant Acromyrmex subterraneus sub- 607.
terraneus protect workers from attack by entomopathogenic Seipke RF, Barke J, Brearley C, et al. A single Streptomyces
fungi. Biol Lett 2012;8:461–4. symbiont makes multiple antifungals to support the fun-
Mueller UG, Gerardo NM, Aanen DK, et al. The evolution of agri- gus farming ant Acromyrmex octospinosus. PLoS One 2011;6:
culture in insects. Annu Rev Ecol Evol S 2005;36:563–95. e22028.
Mueller UG, Mikheyev AS, Solomon SE, et al. Frontier mutualism: Seipke RF, Kaltenpoth M, Hutchings MI. Streptomyces as sym-
coevolutionary patterns at the northern range limit of the bionts: an emerging and widespread theme? FEMS Microbiol
leaf-cutter ant–fungus symbiosis. P R Soc B 2011;278:3050–9. Rev 2012;36:862–76.
Mueller UG, Rehner SA, Schultz TR. The evolution of agriculture Sen R, Ishak H, Estrada D, et al. Generalized antifungal activ-
in ants. Science 1998;281:2034–8. ity and screening of Pseudonocardia and Amycolatopsis bac-
Mueller UG, Schultz TR, Currie CR, et al. The origin of the attine teria in nests of fungus-growing ants. P Natl Acad Sci USA
ant-fungus mutualism. Q Rev Biol 2001;76:169–97. 2009;106:17805–10.
 10 FEMS Microbiology Ecology, 2015, Vol. 91, No. 7

Suen G, Scott JJ, Aylward FO, et al. An insect herbivore mi- Vásquez A, Forsgren E, Fries I, et al. Symbionts as major modula-
crobiome with high plant biomass-degrading capacity. PLoS tors of insect health: lactic acid bacteria and honeybees. PLoS
Genet 2010;6:e1001129. One 2012;7:e33188.
Tragust S, Mitteregger B, Barone V, et al. Ants disinfect fungus- Vásquez A, Olofsson TC. The lactic acid bacteria involved in
exposed brood by oral uptake and spread of their poison. Curr the production of bee pollen and bee bread. J Apicult Res
Biol 2013a;23:76–82. 2009;48:189–95.
Tragust S, Ugelvig LV, Chapuisat M, et al. Pupal cocoons af- Vo TL, Mueller UG, Mikheyev AS. Free-living fungal symbionts
fect sanitary brood care and limit fungal infections in ant (Lepiotaceae) of fungus-growing ants (Attini: Formicidae).
colonies. BMC Evol Biol 2013b;13:225. Mycologia 2009;101:206–10.

Downloaded from https://academic.oup.com/femsec/article-abstract/91/7/fiv073/604805 by guest on 14 June 2020