Diet is the primary determinant of bacterial community
structure in the guts of higher termites
ARAM MIKAELYAN,*†CARSTEN DIETRICH,* TIM K
OHLER,* MICHAEL POULSEN,‡DAVID
ES§¶ and ANDREAS BRUNE*†
*Department of Biogeochemistry, Max Planck Institute for Terrestrial Microbiology, Karl-von-Frisch-Str. 10, 35043, Marburg,
Germany, †LOEWE Center for Synthetic Microbiology, SYNMIKRO, Philipps-Universit€
at Marburg, Marburg, Germany,
‡Section for Ecology and Evolution, Department of Biology, Centre for Social Evolution, University of Copenhagen, Copenhagen
East, Denmark, §Laboratory of Experimental and Comparative Ethology (LEEC), University of Paris 13, Sorbonne Paris Cit
Villetaneuse, France, ¶Institute of Ecology and Environmental Sciences –Paris (iEES-Paris), Institute of Research for
Development, Sorbonne Universit
es, Bondy, France
The gut microbiota of termites plays critical roles in the symbiotic digestion of ligno-
cellulose. While phylogenetically ‘lower termites’ are characterized by a unique associ-
ation with cellulolytic ﬂagellates, higher termites (family Termitidae) harbour
exclusively prokaryotic communities in their dilated hindguts. Unlike the more primi-
tive termite families, which primarily feed on wood, they have adapted to a variety of
lignocellulosic food sources in different stages of humiﬁcation, ranging from sound
wood to soil organic matter. In this study, we comparatively analysed representatives
of different taxonomic lineages and feeding groups of higher termites to identify the
major drivers of bacterial community structure in the termite gut, using amplicon
libraries of 16S rRNA genes from 18 species of higher termites. In all analyses, the
wood-feeding species were clearly separated from humus and soil feeders, irrespective
of their taxonomic afﬁliation, offering compelling evidence that diet is the primary
determinant of bacterial community structure. Within each diet group, however, gut
communities of termites from the same subfamily were more similar than those of dis-
tantly related species. A highly resolved classiﬁcation using a curated reference data-
base revealed only few genus-level taxa whose distribution patterns indicated
speciﬁcity for certain host lineages, limiting any possible cospeciation between the gut
microbiota and host to short evolutionary timescales. Rather, the observed patterns in
the host-speciﬁc distribution of the bacterial lineages in termite guts are best explained
by diet-related differences in the availability of microhabitats and functional niches.
Keywords: gut microbiota, insects, pyrosequencing, termites
Received 25 June 2015; revision received 20 August 2015; accepted 1 September 2015
Termites are eusocial cockroaches that have specialized
on a diet of lignocellulose in various stages of humiﬁca-
tion (Brune 2014). The evolutionary transition from an
omnivorous to wood-feeding lifestyle, which occurred
more than 150 million years ago, was accompanied by
several digestive modiﬁcations, including a distinctive
enlargement of the hindgut and the acquisition of cellu-
lolytic ﬂagellates (Brune & Dietrich 2015). The ﬂagel-
lates play a critical role in symbiotic digestion in the
hindgut of all evolutionary ‘lower’ termite families, but
were lost in the most derived family, the Termitidae or
‘higher’ termites, which appeared about 50 million
years ago and whose gut microbiota is entirely prokary-
otic (Brune & Ohkuma 2011; Bourguignon et al. 2015).
Representing over 85% of all termite genera, the higher
termites are the most diverse of all termite families
(Krishna et al. 2013). While higher termites have
Correspondence: Andreas Brune, Fax: +49 6421 178999; E-mail:
©2015 John Wiley & Sons Ltd
Molecular Ecology (2015) doi: 10.1111/mec.13376
diversiﬁed in many aspects, the most important special-
ization concerns their diet, which extends far beyond the
wood-feeding lifestyle of ‘lower termites’ (Eggleton &
Tayasu 2001). Some higher termites feed on sound ligno-
cellulose such as wood or dry grass, others on partially
degraded material such as leaf litter, herbivore dung or
humus, or in the case of the true soil feeders, on soil
organic matter in an advanced stage of humiﬁcation
(Donovan et al. 2001; Eggleton & Tayasu 2001). A special
case are the fungus-cultivating termites, which grow
basidiomycete fungi (Termitomyces spp.) to predigest
their lignocellulosic diet and, in addition, also consume
the mycelia of the fungal garden (Poulsen 2015). While
fungus feeders consist exclusively of Macrotermitinae,
other feeding groups usually comprise representatives
from various subfamilies (Donovan et al. 2001).
Higher termites offer the unique opportunity to study
the co-evolution of gut microbiota and host. The
tremendous diversity in their dietary specializations
and corresponding digestive adaptations in gut anat-
omy and physiology are reﬂected in the composition of
their bacterial communities (Brune & Ohkuma 2011;
Brune 2014). Recent studies using deep-sequencing of
16S rRNA genes to examine the gut microbiota across a
wide range of higher termites have detected obvious
patterns in community structure, which were
interpreted to reﬂect both phylogeny and diet of the
respective host groups (Dietrich et al. 2014; Otani et al.
2014; Rahman et al. 2015). Nevertheless, limited taxon
sampling within the higher termites has so far pre-
vented an effective identiﬁcation of the major drivers of
community structure of their gut microbiota.
Therefore, to investigate potential drivers of community
structure in higher termites, it is essential that the analysis
include a critical number of representatives from the same
feeding groups that belong to evolutionarily independent
host lineages. Building on our previous analysis of bacte-
rial community structure in a wide range of termites and
cockroaches, which already included 10 species of higher
termites (Dietrich et al. 2014), we investigated eight addi-
tional species from previously undersampled feeding
groups. We sequenced the ampliﬁed 16S rRNA (V3–V4
region) genes using the Illumina MiSeq platform and
analysed the communities across all samples including
both similarity-based metrics and the identiﬁcation of
individual lineages using a highly resolved taxonomic
framework (DictDb; Mikaelyan et al. 2015).
Materials and methods
Termites: collection, identiﬁcation and dissection
Atlantitermes sp., Cornitermes sp., Neocapritermes taracua,
Termes fatalis and Velocitermes sp. were collected near
Petit-Saut dam in French Guiana (5°40N, 52°590W), Ter-
mes hospes and Microcerotermes parvus were collected
near Pointe-Noire in the Democratic Republic of the
Congo (4°410S, 11°510E), and Promirotermes sp. was col-
lected in ARC-PPRI Rietondale Research Station, Preto-
ria in South Africa, in the year 2013. The origin of the
remaining species has been previously described
(Dietrich et al. 2014). Species identity was conﬁrmed by
sequence analysis of the gene encoding cytochrome oxi-
dase subunit 2 (COII; for accession numbers, see
Table 1). A maximum-likelihood tree was calculated for
the COII gene sequences using FASTTREE (Price et al.
2009) under the general time-reversible (GTR) model of
evolution to ascertain the phylogenetic position of the
sequences with respect to each other and to close rela-
tives in GenBank (Fig. S1, Supporting information). All
termites (worker caste only) were dissected with ﬁne-
tipped forceps within a few days of arrival in the labo-
ratory in Marburg (for details, see Table 1). Hindguts
(10–20 per sample) were pooled in 2-mL tubes contain-
ing 750 lL sodium phosphate buffer (120 mM; pH 8.0)
and homogenized. DNA was extracted and puriﬁed
using a bead-beating protocol as previously described
(Paul et al. 2012).
Library preparation and sequencing
The V3–V4 region of the 16S rRNA gene was ampliﬁed
from each sample using the primers M13-343Fmod and
M13-784Rmod, which were based on the universal
bacterial primers 343Fmod and 784Rmod (K€
ohler et al.
2012) respectively, but additionally included universal
M13-speciﬁc priming sites on their 50ends (Daigle et al.
2011). The cycling conditions for this PCR step were as
described previously (K€
ohler et al. 2012). The resulting
amplicons, tagged with the M13 tails, were used as tem-
plate for a subsequent PCR step using the Herculase II
Fusion DNA Polymerase Kit (Agilent Technologies,
USA). The following PCR conditions were used: 94 °C for
3 min, followed by 28 cycles (94 °C for 20 s, 58 °Cfor
20 s and 72 °C for 50 s), and ﬁnally 72 °C for 2 min. Puri-
ﬁed PCR products were mixed in equimolar amounts
and sequenced commercially (paired-end; 2 9350 nt;
Illumina MiSeq; GATC Biotech, Konstanz, Germany). The
quality of the ﬁnal products was checked by gel elec-
trophoresis. The quality-checked MiSeq data sets were
submitted to the MG-RAST server (Meyer et al. 2008) and
can be accessed under the numbers 4639074.3–4639081.3.
Processing of sequence data
Both iTag (Degnan & Ochman 2012) libraries (this
study) and pyrotag libraries (Dietrich et al. 2014;
GenBank accession nos SAMN02228091–101) were
©2015 John Wiley & Sons Ltd
2A. MIKAELYAN ET AL.
processed using the UPARSE pipeline (Edgar 2013).
Only reads with a minimum length of 250 bp and a
maximum expected error of 0.5 were selected and sepa-
rated by sample into different fastq ﬁles using the sam-
ple-speciﬁc barcodes included in the sequences. For the
iTag libraries, the UPARSE pipeline was used to merge
paired reads; pairs with mismatches in the overlapping
region were discarded. After removal of barcodes and
primers, the reads from each sample were clustered at
a threshold of 99% sequence similarity to form opera-
tional taxonomic units (OTUs) using UPARSE. A repre-
sentative sequence from each OTU was selected and
aligned with the MOTHUR aligner, using the Silva refer-
ence alignment (SSURef release 119) as a template.
Analysis of community structure
For the analysis of taxonomic composition of each sam-
ple, the OTUs in the de-replicated data sets were classi-
ﬁed using the RDP classiﬁer (Wang et al. 2007)
implemented in the MOTHUR software suite (Schloss et al.
2009), using a conﬁdence cut-off of 80% and the Dicty-
optera taxonomic reference database (DICTDB) v. 3.0
(Mikaelyan et al. 2015). Genus-level lineages were
ranked by determining their cumulative contribution to
the principal component analysis (PCA) of the gut com-
munities as previously described (Dietrich et al. 2014;
Otani et al. 2014). Bray–Curtis distances between all
communities at the genus level were calculated using
the VEGAN package (Oksanen et al. 2015) in the Rstatisti-
cal software suite (R Core Team 2015). Data sets were
subsampled to a thousand sequences, and distances
were visualized using principal coordinate analysis
(PCoA) implemented in vegan. The Bray–Curtis dis-
tances were additionally subjected to hierarchical clus-
ter analysis and visualized as a dendrogram with the
PVCLUST package (Suzuki & Shimodaira 2006) of R.
Similarities in the phylogenetic structure of the com-
munities were determined using the taxonomy-indepen-
dent weighted UniFrac metric (Lozupone et al. 2007)
implemented in MOTHUR. A maximum-likelihood tree
was constructed using FASTTREE (Price et al. 2009) with
subsamples of 1100 sequences per sample (correspond-
ing to the number of sequences in the smallest library).
This tree served as input for the calculation of pairwise
distances between all 18 samples with UniFrac, which
were ordinated using PCoA.
The signiﬁcance of clustering at the community level
was tested for both metrics using the adonis function
implemented in the VEGAN package (Oksanen et al.
2015), including taxonomy, diet and geography as pos-
sible variables. Additionally, the Kruskal–Wallis non-
parametric test implemented in Rwas used to assess
whether differences in the relative abundance of indi-
vidual bacterial genera can be explained by differences
in diet or host taxonomy.
Phylogenetic analysis of short reads
Representatives of each OTU from each sample were
subjected to phylogenetic analysis using FASTTREE (Price
et al. 2009) under maximum-likelihood criteria and the
GTR model. The relative abundance of each OTU in the
tree was annotated as circles using the APE package
(Paradis et al. 2004) written for the Rsoftware suite
(R Core Team 2015).
Table 1 Higher termites used in the current study, their phylo-
genetic afﬁliations and dietary preferences
ID Host species COII gene
1Macrotermes sp. KT184483 F Lignocellulose,
KT184482 F Lignocellulose,
3Odontotermes sp. KT184484 F Lignocellulose,
KT184481 S Soil organic
5Cornitermes sp. AIZ68247
KT184480 W Wood
S Soil organic
10 Ophiotermes sp. KT184477 S Soil organic
11 Termes hospes AIZ68312
12 Termes fatalis KT184478 H Humus
13 Promirotermes sp. KT184479 H Humus
14 Atlantitermes sp. KT184476 H Humus
15 Velocitermes sp. KT184475 L Litter
16 Trinervitermes sp. KT184474 W Dry grass
KT184473 W Wood
*Feeding groups: F, fungus feeders; W, wood or grass feeders;
L, litter feeders; H, humus feeders; S, ‘true’ soil feeders.
Based on dietary information in Bignell et al. (2011).
COII genes annotated in the mitochondrial genomes recon-
structed by Dietrich & Brune (2014).
Based on dietary information in Gontijo & Domingos (1991).
©2015 John Wiley & Sons Ltd
DIET SHAPES TERMITE GUT COMMUNITY STRUCTURE 3
Quality processing of the amplicon libraries yielded
1097–21 828 sequence reads for each pyrotag and
57 028–222 116 sequence reads for each iTag library.
Classiﬁcation with the taxonomic framework of DictDb
successfully assigned 99% of the reads at the phylum
level. Classiﬁcation success decreased with taxonomic
depth (48–90% at the genus level; for details, see
Table S1, Supporting information). The lowest classiﬁca-
tion success was observed with some humus-feeding
taxa, which is indicative of the bacterial diversity that
still remains to be explored in this diet group, and the
need to add more full-length 16S rRNA reference
sequences to DictDb. For a detailed overview of the
classiﬁcation results, see Tables S2 and S3 (Supporting
At the phylum level, termites within a subfamily dif-
fered considerably in community structure, with the
distribution of dominant phyla being strongly reﬂective
of the termite feeding group. Irrespective of their phylo-
genetic afﬁliation, termites belonging to the same
feeding group showed striking similarities in the distri-
bution of dominant bacterial phyla (Fig. 1). Members of
Spirochaetes, Fibrobacteres and/or the TG3 phylum
were more prevalent among termites feeding on sound
wood or grass than among termites of other feeding
groups, although the proportion of each phylum dif-
fered among host species. Large proportions of Spiro-
chaetes, Fibrobacteres and TG3 phylum were also
found among litter-feeding termites. However, with the
prevalence of Firmicutes and Bacteroidetes encountered
in these taxa, the overall pattern among the litter
feeders resembled that of the humus feeders. Humus
feeders, soil feeders and fungus feeders also shared
similarities in community structure, particularly in the
large proportions of Firmicutes, Bacteroidetes and Pro-
teobacteria, but differed in the abundance of Spiro-
chaetes, which was lower in soil feeders and almost
absent from fungus feeders (Fig. 1). Soil feeders also
harboured a larger proportion of Actinobacteria than
most other species. Fibrobacteres and TG3 phylum,
which were characteristic of wood-feeding and litter-
feeding termites, were virtually absent from most repre-
sentatives of the other feeding groups.
The diet-speciﬁc pattern in gut community structure
observed at the phylum level was even more evident at
higher taxonomic or phylogenetic resolution. PCoA of
community structure showed a clustering of samples by
host diet with both Bray–Curtis (at the genus level) and
UniFrac metrics, with the ﬁrst two axes explaining
approximately half of the total variability in the data
sets (Fig. 2). In addition, hierarchical cluster analysis
revealed a signal of host taxonomy within each diet
group; that is, phylogenetically related termites are
more similar in community structure than termites from
different subfamilies (Fig. 3). Both Bray–Curtis and
UniFrac dissimilarities in community structure were
explained by host diet and taxonomy (considering a sig-
niﬁcance threshold at P≤0.05), and not by geographi-
cal location (see Table S5, Supporting information for
details). Typically, differences in the relative abundance
of individual bacterial genera were better explained by
differences in diet in comparison with host taxonomy
(Tables S6 and S7, Supporting information).
To identify the genus-level taxa that contribute most
to the observed clustering pattern shown in Fig. 2, we
conducted a PCA and ranked the genus-level taxa in
descending order of their cumulative contribution to
the loading factors. The relative abundances of the 20
taxa contributing the most to the clustering observed in
PCA illustrate a preferential enrichment of different
Fig. 1 Abundance of dominant bacterial
phyla in the gut microbiota of higher ter-
mites. Numbers below the heatmap cor-
respond to hosts in Table 1. The
cladogram represents the phylogenetic
relationship among the hosts and is
based on the latest multilocus analysis of
concatenated mitochondrial genes (Bour-
guignon et al. 2015). The subfamily
assignment is based on the taxonomy
proposed by Inward et al. (2007). The
letters above the heatmap indicate the
feeding groups: F, fungus-feeding; S,
soil-feeding; W, wood or grass; H,
humus; L, litter (see Table 1 for more
information on the hosts’ dietary prefer-
©2015 John Wiley & Sons Ltd
4A. MIKAELYAN ET AL.
bacterial lineages among certain termite feeding groups
(Fig. 4). With only a few exceptions, the reads classiﬁed
as Spirochaetes (assigned to the genus-level lineages
Treponema Ia, Ic or If) decreased in abundance with
increasingly humiﬁed host diets. Treponema Ia was most
abundant in wood/grass feeders, and more abundant
in soil feeders than in humus/litter feeders (Fig. 4;
Table S2, Supporting information). The relative abun-
dance of Treponema Ic rose from 0.1% to 1.3% in the
fungus-feeding termites, through 0.25–2.1% in the soil
feeders, to as high as 16.0% in humus/litter feeders.
The highest abundance of Treponema Ic, however, was
observed among termites feeding on sound lignocellu-
lose, where the reads accounted for 25.1–48.7% of the
communities; an exception is the grass-feeding
Trinervitermes sp. (4.2% of the community). Also Tre-
ponema If was signiﬁcantly more abundant among the
wood/grass feeders than the other diet groups, with
the exception of the wood-feeding Nasutitermes takasa-
Our assessment of the gut communities at greater tax-
onomic depth revealed patterns of preferential associa-
tion of genus-level groups in Fibrobacteres and TG3
phylum with particular host lineages (Fig. 4). Subcluster
Ia of Fibrobacteres and Subcluster IIIb of TG3 phylum
were preferentially associated with Microcerotermes spp.
and Termes spp., both members of Termitinae. Contrast-
ingly, Subcluster Ib (of Fibrobacteres) and Subcluster
IIIa (of TG3 phylum) were found associated with Nasu-
titermes corniger and Trinervitermes sp., both members of
Nasutitermitinae. Reads from Cornitermes sp., a member
of Syntermitinae, were classiﬁed to Subcluster IIIb,
while those from Macrotermes sp. that were assigned to
Fibrobacteres could be binned to genus-level clusters
within ‘Cockroach cluster II’.
One of the most predominant genus-level taxa among
fungus-feeding termites was ‘Alistipes II’ (family:
Rikenellaceae), which represented 6–19% of the reads
obtained from the three Macrotermitinae but was absent
from most other host species (Fig. 4; Table S2,
Fig. 2 Principal coordinate analysis (PCoA) of pairwise distances among bacterial communities based on the Bray–Curtis and Uni-
Frac metrics. Each data point in the plots represents the community harboured by a given termite species; colours indicate the differ-
ent host subfamilies. For more details on the samples, see Table 1.
Fig. 3 Hierarchical cluster analysis of pairwise distances
between bacterial communities based on the Bray–Curtis met-
ric. Nodes in the dendrogram that are strongly supported by
high approximately uniform probability values are marked by
open (>70%) and closed (>90%) circles.
©2015 John Wiley & Sons Ltd
DIET SHAPES TERMITE GUT COMMUNITY STRUCTURE 5
Supporting information). In addition, the fungus feeders
were characterized by the abundant presence of several
genus-level groups within Lachnospiraceae and
Ruminococcaceae. The former include the genus-level
‘Gut cluster 13’ and ‘Ca. Arthromitus’, which were rep-
resented also in other feeding groups and most abun-
dant among the soil feeders (Fig. 4).
An analysis of the representative phylotypes from
selected genus-level lineages in the Treponema I clade
revealed that the highly divergent phylotypes clustered
according to the subfamilies of their respective host
species (e.g. Treponema Ia, Ic; Fig. 5A). A tendency of
the phylotypes to cluster by host subfamily was
observed also in the Fibrobacteres (Cluster I) and the
related TG3 phylum (Subcluster III) (Fig. 5B). However,
such host-speciﬁc clustering was observed mostly with
the dominant phylotypes, whereas many of the rarer
phylotypes clustered with sequences from other
subfamilies (Fig. S2, Supporting information).
Previous studies have attempted to assess in how far
host phylogeny or dietary specialization of termites
determines the composition of their gut microbiota
(Colman et al. 2012; Dietrich et al. 2014; Rahman et al.
2015). However, because dietary diversiﬁcation of
higher termites is part of their evolutionary history, the
inﬂuence of diet on gut community structure cannot be
easily disentangled from that of host phylogeny (Brune
& Dietrich 2015). Dietary diversiﬁcation apparently
involved concerted adaptations in mandible morphol-
ogy (Donovan & Jones 2000), intestinal anatomy (Noirot
2001) and physicochemical gut conditions (Bignell &
Eggleton 1995; Brune et al. 1995; Brune & K€
At the same time, there is considerable phylogenetic
conservatism in the dietary habits of closely related spe-
cies, for example wood-feeding within the genus Micro-
cerotermes, and soil-feeding within the ‘Cubitermes
group’ (Bourguignon et al. 2015).
Termite guts are small but complex ecosystems with
a wide range of physicochemically distinct microhabi-
tats, the distribution of which strongly reﬂects adapta-
tions to different diets (Brune & Dietrich 2015). In this
section, we discuss how differences in the availability
of these microhabitats and the dynamics of ecological
niches for the gut microbiota in the course of dietary
diversiﬁcation can explain the differences in the distri-
bution and abundance of many bacterial lineages and
provide tenable ecological and evolutionary explanations
Fig. 4 Abundance of selected bacterial genus-level groups in higher termites. Numbers below the heatmap correspond to hosts in
Table 1; colours of circles represent the different subfamilies of higher termites (see Fig. 1 legend for colour key). The letters above
the heatmap indicate the different feeding groups: F, fungus-feeding; S, soil-feeding; W, wood or grass; H, humus; L, litter. See
Tables S2 and S3 (Supporting information) for a more detailed taxonomic overview of the communities.
©2015 John Wiley & Sons Ltd
6A. MIKAELYAN ET AL.
that reconcile the artiﬁcial dichotomy in the effects of
diet and host phylogeny on gut community structure in
Bacterial communities in wood feeders
The remarkable convergence of bacterial community
structure among wood-feeding Nasutitermitinae and
Termitinae serves to illustrate how diet can engineer
major changes in the gut microbiota through a redistri-
bution of niches and microhabitats in the gut. In the gut
of wood-feeding ‘lower termites’, microbial niches con-
nected to lignocellulose digestion are mostly occupied
by gut ﬂagellates, which sequester wood particles in
their digestive vacuoles (Brugerolle & Radek 2006; Ni &
Tokuda 2013). In higher termites, the loss of ﬂagellates
opened up these niches to ﬁbre-digesting bacteria
(Brune & Dietrich 2015). This hypothesis is sparked by
the upshift in abundance in wood-feeding higher ter-
mites of bacterial lineages from the Fibrobacteres and
the TG3 phylum, which are extremely rare in the gut
microbiota of ‘lower termites’ and in higher termites of
other feeding groups (Hongoh et al. 2006; Dietrich et al.
2014; Rahman et al. 2015).
The deep taxon sampling of higher termites and the
highly resolved classiﬁcation of the present study
revealed that bacterial lineages from Termite cluster I
(phylum: Fibrobacteres), Termite cluster III (TG3 phy-
lum) and the Treponema clusters Ic and If (phylum: Spir-
ochaetes) are more abundant in wood-feeding
representatives than in other feeding groups. Members
of these bacterial lineages dominate the cellulolytic
community that is speciﬁcally associated with the wood
particles in the hindgut of Nasutitermes spp. (Mikaelyan
et al. 2014), and are also abundantly represented in the
guts of other wood-feeding termites (Hongoh et al.
2006; Warnecke et al. 2007; He et al. 2013). The high
abundance of these cellulose-digesting, ﬁbre-associated
bacterial lineages among wood and litter feeders identi-
ﬁes the important structuring role played by both niche
(cellulose degradation) and microhabitat (wood ﬁbres)
selection in the guts of higher termites.
Bacterial communities in litter, humus and soil feeders
One of the most notable adaptations to humus or soil
based diet is the increased alkalinity of the gut ﬂuid,
particularly in the ﬁrst-proctodeal compartment of the
Fig. 5 Maximum-likelihood tree of 16S rRNA phylotypes from the different subfamilies of higher termites (see Fig. 1 legend for col-
our key and Fig. S2, Supporting information for tip labels). (A) Treponema Ia–c genus-level groups. (B) Phylum Fibrobacteres and
TG3. Size of the circles at the tips of the tree is proportional to the relative abundance of the phylotypes in the respective hosts. Col-
ours of the circles correspond to different subfamilies of higher termites.
©2015 John Wiley & Sons Ltd
DIET SHAPES TERMITE GUT COMMUNITY STRUCTURE 7
hindgut (Bignell & Eggleton 1995; Brune & K€
We observed the association of several genus-level lin-
eages among Ruminococcaceae and Lachnospiraceae
(phylum Firmicutes) that are speciﬁcally enriched in the
alkaline ﬁrst-proctodeal (P1) compartments of wood
feeders (Thongaram et al. 2005; K€
ohler et al. 2012),
humus feeders (Thongaram et al. 2005) and soil feeders
(Schmitt-Wagner et al. 2003), clearly suggesting a prefer-
ence of these clostridial lineages for alkaline habitats in
the gut. Moreover, in comparison with the generally
tubular compartments of wood feeders, humus- and
soil-feeding termites are characterized by dilated P1
compartments that probably further magnify the rela-
tive abundance of these alkali-adapted clostridia in their
hindguts. Interestingly, guts of humivorous Pachnoda
scarab beetle larvae possess alkaline midguts (Lemke
et al. 2003), and many phylotypes in clone libraries from
Pachnoda spp. (Egert et al. 2003; Andert et al. 2010) have
been found to cluster in close phylogenetic neighbour-
hood of clones from P1 hindgut compartments of
humivorous higher termites (Schmitt-Wagner et al.
2003; Thongaram et al. 2005). The convergent enrich-
ment of related bacterial lineages in distantly related
orders of insects further underscores the importance of
habitat selection in the structuring of gut communities
in analogous gut environments, both characterized by
elevated intestinal pH.
Although the exact role of these alkali-adapted clos-
tridia remains unknown, their presence in humus- and
soil-feeding termite species may suggest an important
role in the digestion of peptides. This hypothesis is sup-
ported by the presence of alkaline proteolytic activity
(Ji & Brune 2005) and the accumulation of large
amounts of ammonia (Ngugi & Brune 2012) in the
anterior hindgut of soil feeders. Several genes encoding
proteases in the metagenome of the humivorous (dung-
feeding) Amitermes wheeleri have been binned to the
Clostridiales (He et al. 2013). Interestingly, the distribu-
tion patterns of genus-level groups within the Lach-
nospiraceae and Ruminococcaceae observed in
A. wheeleri are quite similar to those in termites feeding
on litter or humus (see Table S4, Supporting informa-
tion), suggesting important roles of these lineages in the
digestion of peptides in dung and humus.
In addition to the proteolytic activities (Ji & Brune
2005), the hindgut content of soil-feeding termites also
shows small amounts of xylanolytic and cellulolytic
activities (Rouland & Chararas 1986), which have been
suggested to be of bacterial origin (Brauman 2000).
Again, the abundance of clostridial genes encoding
xylanases in the hindgut metagenome and metatran-
scriptome of A. wheeleri (He et al. 2013) suggests that
the clostridial lineages in the soil-feeding and humus-
feeding termites investigated in the current study play
a role also in the digestion of residual polysaccharides
Gut content analyses have also shown that members
of the humus-feeding termites consume a greater pro-
portion of plant material and/or wood ﬁbres than the
true soil feeders (Sleaford et al. 1996; Donovan et al.
2001), and the composition of artiﬁcial lignocellulosic
diets strongly affects community structure in wood-
feeding N. takasagoensis, particularly the abundance of
the phyla Fibrobacteres, Spirochaetes and TG3 (Miyata
et al. 2007). Therefore, differences in the proportion of
plant material in the diets of higher termites could
explain the higher abundance of ﬁbre-associated bacte-
rial lineages in humus feeders compared to soil feeders,
that is Treponema clusters Ic and If (Spirochaetes), Ter-
mite cluster I (Fibrobacteres) and Termite cluster III
(TG3 phylum). Considerable differences in the distribu-
tion of these lineages even among the humus feeders
agree with differences in the degree of humiﬁcation in
their lignocellulosic diets (Bourguignon et al. 2011).
Bacterial communities in fungus feeders
The similarity in gut community structure we observed
among fungus-feeding termites from different genera
conﬁrms the presence of a core set of bacterial lineages
that are either absent or rare in the communities of
other higher termites (Dietrich et al. 2014; Otani et al.
2014). This similarity in community structure among
the fungus feeders is driven by their highly specialized
diet, which has been shown to include fungal mycelia,
in addition to wood or litter (Donovan et al. 2001). The
shift to a relatively proteinaceous diet among the fun-
gus feeders could be also responsible for the conver-
gence of community structure with that of omnivorous
cockroaches (Dietrich et al. 2014), particularly in the
preferential enrichment of genus-level lineages such as
Alistipes clusters II, III and IV in both distantly related
insect groups (Mikaelyan et al. 2015). However, the role
of the Alistipes clusters in the guts of the termites is still
unknown (Poulsen et al. 2014). As another result of pos-
sible habitat selection, Macrotermes sp. harbours Cock-
roach cluster II of Fibrobacteres, a lineage so far
detected only among litter-feeding cockroaches (Mikael-
yan et al. 2015), which could again be a consequence of
dietary similarities between the two insect groups.
Evidence for co-evolution or cospeciation?
Our analyses clearly identify diet as the primary deter-
minant of community structure in higher termites. Nev-
ertheless, each diet group conceals a signiﬁcant signal
of host taxonomy, which is reﬂected in the clustering of
bacterial phylotypes by termite subfamily (Mikaelyan
©2015 John Wiley & Sons Ltd
8A. MIKAELYAN ET AL.
et al. 2015; Fig. 5). This explains the previously reported
patterns of host phylogeny in the bacterial gut micro-
biota in a broader selection of termite and cockroach
hosts (Dietrich et al. 2014; Rahman et al. 2015), which
led to the suggestion of vertical transmission being the
primary force shaping the composition of the termite
gut microbiota (Rahman et al. 2015). However, even a
thorough assessment of phylogenetic similarity of gut
communities in different hosts (e.g. by the unweighted
UniFrac metric) may be misleading, because not all
members of the gut community are necessarily co-
evolving with the host (Brune & Dietrich 2015).
Using a wider selection of host taxa that comprises
members of the same feeding groups from different
subfamilies, we were able to demonstrate that the co-
evolutionary signal in community structure is masked
by an overwhelming signal of host diet. Although a
preferential association of certain bacterial lineages with
particular host groups is evident already in the classiﬁ-
cation (Fig. 4) and phylogenetic analyses of short reads
(Fig. 5), the co-evolutionary signal is better resolved in
phylogenetic analyses of full-length 16S rRNA
sequences. For instance, detailed analyses of Fibrobac-
teres and the TG3 phylum revealed several mono-
phyletic, genus-level lineages that are speciﬁcally
associated with termites of the same subfamilies
(Hongoh et al. 2006; Mikaelyan et al. 2015).
Although co-evolution may lead to cospeciation, the
degree of cocladogenesis will depend on the intimacy
of interaction among the partners. In contrast to the
hereditary symbioses between insects and their intracel-
lular bacteria (e.g. Buchnera,Blattabacterium), which have
resulted in perfectly congruent phylogenies of symbiont
and host (Clark et al. 2000; Lo et al. 2003), the vertical
inheritance of gut microbiota is less reliable (Brune &
Dietrich 2015). The exchange of stomodeal or proc-
todeal ﬂuids among nestmates (trophallaxis) and the
consumption of faeces (coprophagy) or nest material
(which is composed of faeces) increase the possibility of
an uptake of bacteria from the environment, including
gut microbiota of other termite species (Nalepa et al.
However, the contribution of horizontal transfer of
gut microbiota among different termite lineages is
apparently not very large. Already an early study of
several Reticulitermes and Microcerotermes species indi-
cated that the bacterial communities in the guts of con-
generic termites are very similar, irrespective of
geographical location (Hongoh et al. 2005). Also the
clustering patterns in our larger and more diverse data
set (Fig. 2) cannot be explained by the geographical ori-
gin of the samples (Table S5, Supporting information).
Nonetheless, it would be important to investigate
whether sympatric termite species with more frequent
opportunities to exchange gut microbiota, for example
inquilines such as Velocitermes spp. that cohabit the ter-
mitaria of other termites (Florencio et al. 2013), share
gut microbiota with their respective host. A horizontal
transmission of microbial symbionts, for example dur-
ing territory defence, has been used to explain high
similarities in the ﬂagellate assemblages of Termopsidae
and certain Rhinotermitidae (Reticulitermes spp.), which
sometimes occur even within the same log of wood
Although the phylogenetic continuity in the identity
of other microbial lineages that occupy important niches
in the gut is in agreement with the vertical transmission
of symbionts, some of the co-evolutionary patterns
observed at higher taxonomic levels may merely be the
result of a diffuse selective pressure that the gut habitat
exerts on microbial lineages that are transferred via the
environment (Won et al. 2003; Fontanez & Cavanaugh
2014). Convincing evidence for cospeciation cannot be
obtained from the analysis of short reads but will
require the highly resolved phylogenetic analysis of
candidate symbionts from a sufﬁcient number of closely
We thank Katja Meuser for excellent laboratory assistance. This
study was funded by a grant of the Deutsche Forschungsge-
meinschaft (DFG) in the Collaborative Research Center SFB 987
and the LOEWE Programme of the State of Hessen (Synmikro).
AM was supported by the Synmikro Post-Doc Programme. CD
received a scholarship from the International Max Planck
Research School for Environmental, Cellular and Molecular
Microbiology (IMPRS-MIC). MP was supported by a STENO
grant from the Danish Agency for Science, Technology and
Andert J, Marten A, Brandl R, Brune A (2010) Inter- and
intraspeciﬁc comparison of the bacterial assemblages in the
hindgut of humivorous scarab beetle larvae (Pachnoda spp.).
FEMS Microbiology Ecology,74, 439–449.
Bignell DE, Eggleton P (1995) On the elevated intestinal pH of
higher termites (Isoptera: Termitidae). Insectes Sociaux,42,
Bignell DE, Roisin Y, Lo N (2011) Global biogeography of ter-
mites. In: Biology of Termites: A Modern Synthesis(eds Bignell
DE, Roisin Y, Lo N), pp. 477–498. Springer Netherlands, Do-
ık J, Lepoint G et al. (2011) Feeding ecol-
ogy and phylogenetic structure of a complex neotropical ter-
mite assemblage, revealed by nitrogen stable isotope ratios.
Ecological Entomology,36, 261–269.
Bourguignon T, Lo N, Cameron SL et al. (2015) The evolution-
ary history of termites as inferred from 66 mitochondrial
genomes. Molecular Biology and Evolution,32, 406–421.
©2015 John Wiley & Sons Ltd
DIET SHAPES TERMITE GUT COMMUNITY STRUCTURE 9
Brauman A (2000) Effect of gut transit and mound deposit
on soil organic matter transformations in the soil feeding
termite: a review. European Journal of Soil Biology,36, 117–
Brugerolle G, Radek R (2006) Symbiotic protozoa of termites.
In: Intestinal Microorganisms of Termites and Other Invertebrates
onig H, Varma A), pp. 243–269. Springer, Berlin, Hei-
Brune A (2014) Symbiotic digestion of lignocellulose in termite
guts. Nature Reviews Microbiology,12, 168–180.
Brune A, Dietrich C (2015) The gut microbiota of termites:
digesting the diversity in the light of ecology and evolution.
Annual Review of Microbiology,69, in press. doi: 10.1146/an-
Brune A, K€
uhl M (1996) pH proﬁles of the extremely alkaline
hindguts of soil-feeding termites (Isoptera: Termitidae) deter-
mined with microelectrodes. Journal of Insect Physiology,42,
Brune A, Ohkuma M (2011) Role of the termite gut microbiota
in symbiotic digestion. In: Biology of Termites(eds Bignell DE,
Roisin Y, Lo N), pp. 439–475. Springer, Dodrecht.
Brune A, Emerson D, Breznak JA (1995) The termite gut micro-
ﬂora as an oxygen sink: microelectrode determination of oxy-
gen and pH gradients in guts of lower and higher termites.
Applied and Environmental Microbiology,61, 2681–2687.
Clark MA, Moran NA, Baumann P et al. (2000) Cospeciation
between bacterial endosymbionts (Buchnera) and a recent
radiation of aphids (Uroleucon) and pitfalls of testing for
phylogenetic congruence. Evolution,54, 517–525.
Colman DR, Toolson EC, Takacs-Vesbach CD (2012) Do diet
and taxonomy inﬂuence insect gut bacterial communities?
Molecular Ecology,21, 5124–5137.
Daigle D, Simen BB, Pochart P (2011) High-throughput
sequencing of PCR products tagged with universal primers
using 454 life sciences systems. Current Protocols in Molecular
Degnan P, Ochman H (2012) Illumina-based analysis of micro-
bial community diversity. The ISME Journal,6, 183–194.
Dietrich C, Brune A (2014) The complete mitogenomes of six
higher termite species reconstructed from metagenomic data-
sets (Cornitermes sp., Cubitermes ugandensis,Microcerotermes par-
vus,Nasutitermes corniger,Neocapritermes taracua, and Termes
hospes). Mitochondrial DNA. doi: 10.3109/19401736.2014.987257.
Dietrich C, K€
ohler T, Brune A (2014) The cockroach origin of
the termite gut microbiota: patterns in bacterial community
structure reﬂect major evolutionary events. Applied and Envi-
ronmental Microbiology,80, 2261–2269.
Donovan SE, Jones DT (2000) Morphological phylogenetics of
termites (Isoptera). Biological Journal of the Linnean Society,70,
Donovan SE, Eggleton P, Bignell DE (2001) Gut content analy-
sis and a new feeding group classiﬁcation of termites. Ecolog-
ical Entomology,26, 356–366.
Edgar RC (2013) UPARSE: highly accurate OTU sequences
from microbial amplicon reads. Nature Methods,10, 996–998.
Egert M, Wagner B, Lemke T et al. (2003) Microbial community
structure in midgut and hindgut of the humus-feeding larva
of Pachnoda ephippiata (Coleoptera: Scarabaeidae). Applied and
Environmental microbiology,69, 6659–6668.
Eggleton P, Tayasu I (2001) Feeding groups, lifetypes and the
global ecology of termites. Ecological Research,16, 941–960.
Florencio DF, Marins A, Rosa CS et al. (2013) Diet segregation
between cohabiting builder and inquiline termite species.
PLoS ONE,8, e66535.
Fontanez KM, Cavanaugh CM (2014) Evidence for horizontal
transmission from multilocus phylogeny of deep-sea mussel
(Mytilidae) symbionts. Environmental Microbiology,16,
Gontijo TA, Domingos DJ (1991) Guild distribution of some
termites from cerrado vegetation in south-east Brazil. Journal
of Tropical Ecology,7, 523–529.
He S, Ivanova N, Kirton E et al. (2013) Comparative metage-
nomic and metatranscriptomic analysis of hindgut paunch
microbiota in wood- and dung-feeding higher termites. PLoS
Hongoh Y, Deevong P, Inoue T et al. (2005) Intra- and inter-
speciﬁc comparisons of bacterial diversity and community
structure support coevolution of gut microbiota and termite
host. Applied and Environmental Microbiology,71, 6590–6599.
Hongoh Y, Deevong P, Hattori S et al. (2006) Phylogenetic
diversity, localization, and cell morphologies of members of
the candidate phylum TG3 and a subphylum in the phylum
Fibrobacteres, recently discovered bacterial groups dominant
in termite guts. Applied and Environmental Microbiology,72,
Inward DJG, Vogler AP, Eggleton P (2007) A comprehensive
phylogenetic analysis of termites (Isoptera) illuminates key
aspects of their evolutionary biology. Molecular Phylogenetics
and Evolution,44, 953–967.
Ji R, Brune A (2005) Digestion of peptidic residues in humic
substances by an alkali-stable and humic-acid-tolerant prote-
olytic activity in the gut of soil-feeding termites. Soil Biology
and Biochemistry,37, 1648–1655.
Kitade O (2004) Comparison of symbiotic ﬂagellate faunae
between termites and a wood-feeding cockroach of the genus
Cryptocercus.Microbes and Environments,19, 215–220.
ohler T, Dietrich C, Scheffrahn RH, Brune A (2012) High-res-
olution analysis of gut environment and bacterial microbiota
reveals functional compartmentation of the gut in wood-
feeding higher termites (Nasutitermes spp.). Applied and Envi-
ronmental Microbiology,78, 4691–4701.
Krishna K, Grimaldi DA, Krishna V, Engel MS (2013) Treatise
on the Isoptera of the world. Bulletin of the American Museum
of Natural History. American Museum of Natural History,
Lemke T, Stingl U, Egert M, Friedrich MW, Brune A (2003)
Physicochemical conditions and microbial activities in the
highly alkaline gut of the humus-feeding larva of Pachnoda
ephippiata (Coleoptera: Scarabaeidae). Applied and Environmen-
tal Microbiology,69, 6650–6658.
Lo N, Bandi C, Watanabe H, Nalepa C, Beninati T (2003) Evi-
dence for cocladogenesis between diverse dictyopteran lin-
eages and their intracellular endosymbionts. Molecular
Biology and Evolution,20, 907–913.
Lozupone CA, Hamady M, Kelley ST, Knight R (2007) Quanti-
tative and qualitative bdiversity measures lead to different
insights into factors that structure microbial communities.
Applied and Environmental Microbiology,73, 1576–1585.
Meyer F, Paarmann D, D’Souza M et al. (2008) The metage-
nomics RAST server—a public resource for the automatic
phylogenetic and functional analysis of metagenomes. BMC
©2015 John Wiley & Sons Ltd
10 A. MIKAELYAN ET AL.
Mikaelyan A, Strassert JFH, Tokuda G, Brune A (2014) The
ﬁbre-associated cellulolytic bacterial community in the hind-
gut of wood-feeding higher termites (Nasutitermes spp.). En-
vironmental Microbiology,16, 2711–2722.
Mikaelyan A, K€
ohler T, Lampert N et al. (2015) Classifying the
bacterial gut microbiota of termites and cockroaches: a
curated phylogenetic reference database (DictDb). Systematic
and Applied Microbiology, in press. doi: 10.1016/
Miyata R, Noda N, Tamaki H et al. (2007) Inﬂuence of feed
components on symbiotic bacterial community structure in
the gut of the wood-feeding higher termite Nasutitermes
takasagoensis.Bioscience, Biotechnology, and Biochemistry,71,
Nalepa CA, Bignell DE, Bandi C (2001) Detritivory, copro-
phagy, and the evolution of digestive mutualisms in Dicty-
optera. Insectes Sociaux,48, 194–201.
Ngugi DK, Brune A (2012) Nitrate reduction, nitrous oxide for-
mation, and anaerobic ammonia oxidation to nitrite in the
gut of soil-feeding termites (Cubitermes and Ophiotermes
spp.). Environmental Microbiology,14, 860–871.
Ni J, Tokuda G (2013) Lignocellulose-degrading enzymes from
termites and their symbiotic microbiota. Biotechnology
Noirot C (2001) The gut of termites (Isoptera). Annales de la
e Entomologique de France,37, 431–471.
Oksanen JF, Blanchet G, Kindt R et al. (2015) vegan: Community
ecology package. R package version 2.3-0. Available from
Otani S, Mikaelyan A, Nobre T et al. (2014) Identifying the core
microbial community in the gut of fungus-growing termites.
Molecular Ecology,23, 4631–4644.
Paradis E, Claude J, Strimmer K (2004) APE: analyses of phylo-
genetics and evolution in R language. Bioinformatics,20,
Paul K, Nonoh JO, Mikulski L, Brune A (2012) “Methanoplas-
matales”, Thermoplasmatales-related archaea in termite guts
and other environments, are the seventh order of methano-
gens. Applied and Environmental Microbiology,78, 8245–8253.
Poulsen M (2015) Towards an integrated understanding of the
consequences of fungus domestication on the fungus-grow-
ing termite gut microbiota. Environmental Microbiology. doi:
Poulsen M, Hu H, Li C et al. (2014) Complementary symbiont
contributions to plant decomposition in a fungus-farming
termite. Proceedings of the National Academy of Sciences of the
Price MN, Dehal PS, Arkin AP (2009) FastTree: computing
large minimum evolution trees with proﬁles instead of a dis-
tance matrix. Molecular Biology and Evolution,26, 1641–1650.
R Core Team (2015) R: A Language and Environment for Statisti-
cal Computing. R Foundation for Statistical Computing, Vi-
enna, Austria. Available from http://www.R-project.org/.
Rahman NA, Parks DH, Willner DL et al. (2015) A molecular
survey of Australian and North American termite genera
indicates that vertical inheritance is the primary force shap-
ing termite gut microbiomes. Microbiome,3,1–16.
Rouland C, Chararas C (1986)
ee des osidases de
eces de termites africains
erent. Comptes rendus de l’Acad
emie des sciences,302, 341–
Schloss PD, Westcott SL, Ryabin T et al. (2009) Introducing
mothur: open-source, platform-independent, community-
supported software for describing and comparing microbial
communities. Applied and Environmental Microbiology,75,
Schmitt-Wagner D, Friedrich MW, Wagner B, Brune A (2003)
Phylogenetic diversity, abundance, and axial distribution of
Bacteria in the intestinal tract of two soil-feeding termites
(Cubitermes spp.). Applied and Environmental Microbiology,69,
Sleaford F, Bignell DE, Eggleton P (1996) A pilot analysis of
gut contents in termites from the Mbalmayo Forest Reserve,
Cameroon. Ecological Entomology,21, 279–288.
Suzuki R, Shimodaira H (2006) Pvclust: an R package for
assessing the uncertainty in hierarchical clustering. Bioinfor-
Thongaram T, Hongoh Y, Kosono S et al. (2005) Comparison
of bacterial communities in the alkaline gut segment
among various species of higher termites. Extremophiles,9,
Wang Q, Garrity GM, Tiedje JM, Cole JR (2007) Na€
classiﬁer for rapid assignment of rRNA sequences into the
new bacterial taxonomy. Applied and Environmental Microbiol-
Warnecke F, Luginb€
uhl P, Ivanova N et al. (2007) Metagenomic
and functional analysis of hindgut microbiota of a wood-
feeding higher termite. Nature,450, 560–565.
Won YJ, Hallam SJ, O’Mullan GD et al. (2003) Environmental
acquisition of thiotrophic endosymbionts by deep-sea mus-
sels of the genus Bathymodiolus.Applied and Environmental
A.M., T.K. and A.B. conceived and designed the experi-
ments. M.P. and D.S.-D. collected and identiﬁed termites.
A.M. and C.D. performed the experiments and analysed
the data. C.D. and D.S.-D. contributed to the writing the
manuscript. A.M. and A.B. wrote the manuscript.
The de-replicated, quality-checked MiSeq data sets have
been archived at MG-RAST under the accession num-
bers 4639074.3–4639081.3. The COII nucleotide align-
ment and tree, the phylogenetic tree of OTU
representatives (encoding abundance and sample infor-
mation in the labels), the OTU table for each sample,
and the Bray–Curtis and UniFrac dissimilarity matrices
are archived at Dryad (http://dx.doi.org/10.5061/
Additional supporting information may be found in the online ver-
sion of this article.
©2015 John Wiley & Sons Ltd
DIET SHAPES TERMITE GUT COMMUNITY STRUCTURE 11
Fig. S1 Maximum-likelihood tree constructed using COII DNA
sequences from 18 termite hosts included in this study in the
context of close relatives in public databases.
Fig. S2 Maximum-likelihood tree of 16S rRNA phylotypes from
the different subfamilies of higher termites.
Table S1 Characteristics of the 16S rRNA amplicon libraries
and the classiﬁcation success obtained for each library with
DictDb v 3.0 at different taxonomic levels.
Table S2 Relative abundance of bacterial taxa in iTag and
pyrotag libraries of 16S rRNA amplicons from different higher
termites (arranged by host phylogeny).
Table S3 Relative abundance of bacterial taxa in the iTag and
pyrotag amplicon libraries of 16S rRNA genes from the guts of
different higher termites (arranged by host diet).
Table S4 Comparison of the distribution of genus-level clusters
in Lachnospiraceae and Ruminococcaceae (order Clostridiales)
in a library from Amitermes wheeleri (He et al. 2013) and the ter-
mites investigated in the current study.
Table S5 The correlation of the diet, host taxonomy and geo-
graphical origin with Bray–Curtis and UniFrac distances tested
for signiﬁcance tested with adonis.
Table S6 Pairwise Kruskal–Wallis tests for signiﬁcant differ-
ences in the distribution of bacterial lineages for different diet
groups of termites (F, fungus feeders; S, soil feeders; H,
humus/litter feeders; W, wood/grass feeders).
Table S7 Pairwise Kruskal–Wallis tests for signiﬁcant differ-
ences in the distribution of bacterial lineages for different com-
binations of taxonomic groups of termites (At, Apicotermitinae;
Mt, Macrotermitinae; Nt, Nasutitermitinae; Tt, Termitinae; St,
©2015 John Wiley & Sons Ltd
12 A. MIKAELYAN ET AL.