Introduction

Termites play diverse roles in semi-arid and humid ecosystems and have an impact on the turnover of organic matter in tropical and subtropical regions. The microbiota associated with termites are crucial for the degradation of recalcitrant components of plant biomass and this has a major influence on soil structure and carbon mineralization (Brune and Friedrich 2000; Ohkuma et al. 2003). Termites rely on microbes in their guts that are exchanged between colony members and transferred to the next generation through trophallaxis. These microbes assist termites in breaking down lignocellulose (Brune and Ohkuma 2011). The isolation and cultivation of several bacterial strains from termite guts has partially enabled their classification as decomposers of lignocellulose, uric acid and/or other aromatic compounds; as nitrogen-fixers; and/or as H2/CO2-acetogens (Breznak 2000). However, the majority of the microbial species are difficult or not currently possible to cultivate, thus limiting our understanding of their role in the gut ecosystem (Breznak 2000).

Molecular approaches based on 16S rRNA gene analyses have been helpful for assessing the microbial diversity in termites without cultivation (Schmitt-Wagner et al. 2003; Hongoh et al. 2003, 2006; Shinzato et al. 2005, 2007; Fisher et al. 2007; Long et al. 2010; Mackenzie et al. 2007; Mathew et al. 2012; Schauer et al. 2012; Köhler et al. 2012). These studies not only revealed high bacterial diversity in the guts, but also termite-specific bacterial lineages (Hongoh et al. 2003; Shinzato et al. 2005). Moreover, metagenomic studies yielded key information on bacterial genes required for fermentation, lignocellulosic digestion, reductive acetogenesis, and nitrogen fixation within the host-symbiont association (Warnecke et al. 2007; Tartar et al. 2009; Mattéotti et al. 2011; Liu et al. 2011; Köhler et al. 2012). Although such studies have greatly expanded our knowledge of gut symbiosis, the evolutionary relationships between symbionts and host termites is still inadequately addressed owing to their enormous diversity (Shinzato et al. 2007).

The Macrotermitinae originate from Africa and comprise many of the economically important termite species (Ahmed et al. 2011). Their high abundance and cryptic and unpredictable foraging patterns, coupled with the diversity of their gut symbionts, challenge the development of termite control strategies (Ahmed et al. 2011). Although comprehensive studies have been conducted on these termites (Hongoh et al. 2006; Shinzato et al. 2007; Long et al. 2010; Mackenzie et al. 2007; Mathew et al. 2012; Zhu et al. 2012), information on the gut microbial diversity of the members of the genus Odontotermes is still inadequate. Furthermore, such information is lacking entirely for the members of the genus Microtermes on account of its rarity and lack of adequate termite experts in Africa (Ahmed et al. 2011). Therefore, in this study, we characterize the gut bacterial diversity and compare its community structure and specificity in three termites representing these genera. The findings will extend the clone-based inventories of the termite gut microbiota and contribute to understanding of the specificity and mutualistic relationship of gut symbionts with their hosts.

Materials and methods

Collection and identification of termites

Samples were collected in March, 2011 from Thika district, Kenya (latitude 1°5′54.68″N, longitude 37°1′1.10″W). Termite mounds [C and C1, approximately 0.5 km far apart, were colonized by Odontotermes sp. (JQ247986) and O. somaliensis (JQ247985), respectively; mound B, located approximately 2 km from mound C, was colonized by Microtermes sp. (JQ247991)] were excavated to a depth of 0.5–1.0 m as described elsewhere (Makonde et al. 2013). Termites (n = 200 workers and 50 soldiers) were sampled into sterile plastic boxes. Worker-caste termites were used in the experiments due to their foraging behaviour during establishment and renewal of the fungus gardens. The identity of the termites was confirmed by sequencing the mitochondrial cytochrome oxidase II gene in DNA extracted from the heads of soldiers (Austin et al. 2004; Makonde et al. 2013) and comparing it to the sequences of previously identified specimens (Inward et al. 2007).

DNA extraction and PCR amplification

The exterior surfaces of the termites were washed with 70 % ethanol and then rinsed with sterile distilled water. The guts were aseptically removed with forceps (Schmitt-Wagner et al. 2003). A total of 26 guts (approximately 144 mg) from O. somaliensis and Odontotermes sp. (JQ247986) and 74 guts (approximately 143 mg) from Microtermes sp. (JQ247990) were put separately into three sterile micro tubes containing 0.2 ml of TE buffer (10 mM Tris–HCl, 1 mM EDTA, pH 8.0). They were then homogenized using a sterile glass rod. The corresponding homogenates were then transferred into sterile tubes and used for total DNA extraction using the UltraClean®. Mega soil DNA isolation kit (MO BIO Laboratories, Inc.) according to the manufacturer’s protocol.

Clone libraries, construction and sequencing

Purified DNA was quantified photometrically (NanoDrop; Thermo Fisher Scientific, Schwerte, Germany) and used as a template for amplification of 16S rRNA genes using the universal bacterial primers (27F 5′-TAG AGT TTG ATC CTG GCT CAG-3′ forward primer and 1392R 5′-GAC GGG CGG TGT GTA CA-3′ reverse primer) according to their position in relation to the Escherichia coli gene sequence (Lane 1991). For each PCR, 1 μl (25 ng/μl) of the template was mixed with TaKaRa Ex Taq™ HS (5 units/μl) according to the manufacturer’s protocol. The PCR conditions were as described by Mackenzie et al. 2007 except the final extension was at 72 °C for 10 min. PCR product size was checked on 1 % agarose gels stained with ethidium bromide. The amplicons were gel purified using the Macherey-Nagel NucleoSpin extract II kit (740609.50) and eluted in 30 μl of TE Buffer (5 mM, pH 8.0). The quality of the purified PCR product was determined through electrophoresis on a 1 % agarose gel.

Purified PCR products were ligated into pGEM-T® Easy vector system II (Promega) and transfected through heat shock into E. coli JM109 high efficiency competent cells (Promega). Selection of transformants and extraction of plasmid DNA followed described protocols (Ausubel 1995). Further screening was done via restriction fragment length polymorphism (RFLP) to select representative clones for sequencing. Restriction digests of cloned PCR products were performed using the restriction enzyme HaeIII (New England Biolabs). Representative clones from the three termites were selected for sequencing at Helmholtz Centre for Infection Research (HZI), Braunschweig, Germany. Trace files were manually edited and assembled using Invitrogen vector NTI 11.5 software. The obtained sets of sequences were deposited in GenBank under the accession numbers JQ710341 to JQ710433 (for the O. somaliensis sequences), JX421772–JX421863 (for the Odontotermes sp. sequences), JX421864–JX421955 (for the Microtermes sp. sequences).

Phylogenetic analysis

We checked all sequences for chimeric structures using the Mallard program (Ashelford et al. 2006). A search for similar sequences using BLASTN (Altschul et al. 1990) against the National Center for Biotechnology Information (NCBI) database was performed, and sequence alignment between the query sequences and the identified nearest neighbours was performed using the CLUSTAL Omega program (http://www.clustal.org). A neighbour-joining tree of the aligned sequences was constructed (Saitou and Nei 1987) using MEGA V5.10 (Tamura et al. 2011). Evolutionary distances were computed using the Maximum Composite Likelihood method (Tamura et al. 2004). To obtain support values for the branches, bootstrapping (Felsenstein 1985) was conducted with 1,000 replicates. All sites, including gaps in the sequence alignment, were excluded pairwise in the phylogenetic analysis. Using the resultant neighbour-joining tree, we assigned each clone to a taxonomic group. We defined ribotypes as those sequences sharing at least 98 % sequence identity with each other (Hongoh et al. 2003). The taxonomic assignment was confirmed at an 80 % confidence level using the naïve Bayesian rRNA classifier on the RDP website (Cole et al. 2005).

Comparison of bacterial communities in different termites

To compare the bacterial community structures in different termite species based on the relative abundances of different bacterial phyla in the guts, 16S rRNA gene sequences obtained from clone libraries of different termites were downloaded from the NCBI Genbank database. Details of the information on the downloaded sequences can be found in the Online Supplementary Table S1. The retrieved set of sequences downloaded, together with our sequences, were separately subjected to the RDP naïve Bayesian rRNA Classifier Version 2.5 program on the RDP website and then compared against the Genbank 16S rRNA gene sequence database (www.ncbi.nlm.nih.gov/BLAST/). The relative abundances (%) of the bacterial groups (Supplementary Table S1) were then used for correlation analysis using principal components analysis (PCA) as implemented in XLSTAT version 2013.2.

Results

Affiliation of 16S rRNA gene sequences from termite guts

A total of 689 clones from O. somaliensis and Odontotermes sp., and 336 clones from Microtermes sp. were analysed by RFLP (Supplementary Figure S1). 296 unique RFLP patterns from O. somaliensis (100 clones), Odontotermes sp. (100 clones) and Microtermes sp. (96 clones) were sequenced. Of the sequenced clones, 277 were non-chimeric and were included in the subsequent analysis. The clones (prefixed as OTG in O. somaliensis, OGH in Odontotermes sp. and MIGH in Microtermes sp.) were affiliated with 151 different phylotypes. Phylogenetic analyses showed the clones corresponded to diverse bacterial species affiliated with seven bacterial phyla (Bacteroidetes, Spirochaetes, Firmicutes, Proteobacteria, Synergistetes, Planctomycetes and Actinobacteria) (Figs. 1, 2, 3, 4). The percentage numbers indicate the relative abundances of the bacterial groups from the total number of clones.

Fig. 1
figure 1

Percentage of the representative phyla in the gut of Odontotermes and Microtermes species

Full size image
Fig. 2
figure 2

Evolutionary relationships between partial 16S rRNA gene clone sequences and selected taxa in the phylum Bacteroidetes. OTG denotes clone sequences from O. somaliensis. OGH denotes clone sequences from Odontotermes sp., MIGH denotes clone sequences from Microtermes species. Methanoculleus thermophiles (AB065297) was used to root the tree

Full size image
Fig. 3
figure 3

Continuation of Fig. 2 showing evolutionary relationships between partial 16S rRNA gene clone sequences and selected taxa in the phylum Bacteroidetes. OTG denotes clone sequences from O. somaliensis. OGH denotes clone sequences from Odontotermes sp., MIGH denotes clone sequences from Microtermes species

Full size image
Fig. 4
figure 4

Evolutionary relationships between partial 16S rRNA gene clone sequences and selected taxa in the phyla Spirochaetes, Firmicutes, Synergistetes, Planctomycetes, Proteobacteria and Actinobacteria. OTG denotes clone sequences from O. somaliensis. OGH denotes clone sequences from Odontotermes sp., MIGH denotes clone sequences from Microtermes species. Methanoculleus thermophiles (AB065297) was used to root the tree

Full size image

Gut bacterial community structure in Odontotermes spp.

The gut bacterial community structure of the two Odontotermes species was almost identical at the phylum level, as shown in Fig. 1. The phylum Bacteroidetes accounted for 68 and 64 % of the total clones in O. somaliensis and Odontotermes sp., respectively (Fig. 1). The majority of the phylotypes (accounting for 34 % clones in O. somaliensis and 28 % clones in Odototermes sp.) belonged to the genus Alistipes and often clustered with clones previously obtained from guts of termites, humans, cockroach and mouse faeces (Fig. 2). Several phylotypes, among them OGH174 (4.2 %) and OTG026 (8.8 %), formed a cluster with clones from Odontotermes formosanus and Macrotermes gilvus and were distantly related to those of bacteria in the genus Alistipes (<94 % sequence identity to Alistipes finegoldii [AB554230] and Alistipes sp. JC136 [JF824799]) (Fig. 2). Phylotypes OGH158 and OTG019 clustered with clones from M. gilvus and mouse faeces. These were distantly affiliated with bacteria in the genus Alistipes (90–95 % sequence identity to Alistipes sp. JC136 [JF824799], Alistipes shahii [AB554233] and Alistipes putredinis [AB554232]), indicating that they are part of the gut intestinal tract microbiota. Other phylotypes including OGH182 and OTG004 were often closely affiliated with clones from M. gilvus and Macrotermes michaelseni but were only distantly related to Alistipes sp. JC136 and A. finegoldii (<94 % sequence similarity) (Fig. 2). Phylotype OTG053 (5.5 %) and several others including phylotypes OGH183 and OTG003 belonged to the order Bacteroidales and had between 89 and 98 % sequence similarities with other clones from different environments (termite gut, anaerobic reactor, industrial sediments, sludge and soil) (Fig. 2). Phylotypes OGH177, OGH41 and several others were assigned to the order Bacteroidales and formed a cluster with clones from Nasutitermes species (Fig. 2). Eight phylotypes (OTG024, OTG044, OTG066, OTG073, OGH45, OGH119, OGH144 and OGH176) were affiliated with the genus Bacteroides (Fig. 3). Phylotypes OGH176 and OTG066 formed a subcluster with some Bacteroides species isolated from human stools and mouse guts. Phylotypes OGH45 and OTG073 clustered together with clone ImMB5 from Incisitermes minor (Fig. 3). Phylotypes OGH58 and OTG069 were distantly related to bacteria in the genus Prevotella (89–90 % sequence identity with Prevotella nanceiensis and Prevotella sp. 310-5) as the nearest cultivated neighbours (Fig. 3). Two phylotypes, OTG012 and OGH165, formed a large cluster with several clones from O. formosanus, Reticulitermes speratus and Nasutitermes sp. and were distantly related to Bacteroides species (90 % sequence identity). Phylotypes OGH20 and OTG096 formed a separate cluster and were distantly related to Dysgonomonas mossii (AB548676) and Dysgonomonas oryzarvi (AB547446) (90 % sequence similarity). Cluster one (Fig. 3) consisted of four phylotypes OTG085 (two clones [JQ710351]), OGH23 (three clones [JX421796]), OGH60 (three clones [JX421797]) and OTG075 (two clones [JQ710361]) together with clone Cf2-02 (GQ502485) from Coptotermes formosanus. These were related to bacteria in the genus Dysgonomonas (92–94 % sequence identity with Dysgonomonas hofstadii [FN356023] and Dysgonomonas gadei [AB548675]). Members of the genus Dysgonomonas have been reported to inhabit intestinal guts, blood, wounds, urine, peritoneal fluid, stools and gallbladder (Hansen et al. 2005). Several other phylotypes i.e. OTG043 (12 clones [JQ710392]), OTG047 (three clones [JQ710388]), OTG060 (three clones [JQ710376]), OTG084 (three clones [JQ710352]), OGH27 (seven clones [JX421793-94]), OGH76 (four clones [JX421791-92]), OGH99 (eight clones [JX421789-90]) and OGH155 (nine clones [JX421809-11]) formed cluster two together with other clones often obtained from O. formosanus and M. gilvus (Fig. 3). These clones belonged to the genus Parabacteroides and were distantly affiliated with Parabacteroides gordonii (AB4703344) and Parabacteroides johnsonii (NR_041464) (89–91 % sequence identity) (Fig. 3). Phylotypes OGH184 and OTG002 were assigned to the genus Tennerella (93 % sequence identity with Tannerella forsythia) (JN713185) and Tannerella forsythia ATCC 43037 (NR_074140) and formed a subcluster with clones from insects (cockroach and Pachnoda ephippiata).

The phylum Synergistetes accounted for ten phylotypes. Phylotypes OGH147 (seven clones [JX421858]), OGH135 (two clones [JX421944]), OTG011 (eight clones [JQ710423, JQ710404]) and OTG020 (six clones [JQ710432, JQ710385]) were assigned to the family Synergistaceae and formed cluster six (Fig. 4) together with Candidatus Tammella caduceiae (AB299516) and some clones often obtained from M. gilvus. Phylotypes OTG055, OGH101, OGH128 and OTG029 belonged to the genus Thermovirga (<80 % sequence identity with Thermovirga lienii [DQ071273] and were distantly related to members in the genera Deferribacter and Flexistipes (93–94 % sequence identities with Flexistipes-like sp. [AY005447], Deferribacteres sp. [AY349370] and Synergistales bacterium [JN713409]) and clustered with some clones from O. formosanus (Shinzato et al. 2007). Two phylotypes OGH181 and OTG013 belonged to the family Synergistaceae and were distantly affiliated with clones obtained from soil (90–93 % sequence identity). A total of eight phylotypes were related to the phylum Spirochaetes (Fig. 4). Phylotypes OTG057 and OGH173 clustered with treponemal clones from M. gilvus. Four phylotypes OGH15 (three clones [JX421837]), OGH39 (12 clones [JX421838, JX421843, JX421844]), OTG063 (17 clones [JQ710373, JQ710360, JQ710357]) and OTG087 (three clones [JQ710349]) formed cluster seven (Fig. 4) together with Spirochaeta sp. (AJ419819), Treponema azotonutricium ZAS-9 (CP001841), Treponema sp. ZAS-1 (AF09325) (92–94 % sequence identity) and other treponemal clones often from Nasutitermes species and M. gilvus. Two phylotypes OGH185 and OTG001 formed a cluster with other treponemal clones obtained from M. gilvus and were related to Spirochaeta sp. (AJ419819) from Neotermes castaneus and Treponema primitia ZAS-2 (CP001843) (93–94 % sequence identity) (Fig. 4). Eight phylotypes were distributed in three subdivisions of the Proteobacteria. Two phylotypes, OTG038 and OGH97, were assigned to the genus Propionivibrio (ß-Proteobacteria) and clustered with Propionivibrio limicola (NR_025455) (<94 % sequence identity) and clone Cf6-10 (GQ502594) from C. formosanus (Fig. 4). Phylotypes OGH07 and OTG089 belonged to the subdivision α-Proteobacteria, with 98–99 % sequence affiliation to Wolbachia species from guts of termites and insects. Four phylotypes, OGH145 (13 clones [JX421846, JX421849]), OGH148 (seven clones [JX421845, JX421850]), OTG017 (nine clones [JQ710418, JQ421412]), and OTG092 (three clones [JQ710346]) formed cluster five (Fig. 4) together with clones from M. michaelseni. These were assigned to the subdivision δ-Proteobacteria.

Phylotype OTG071 (14 clones [JQ710350, JQ710365, JQ710371]) was the most abundant in the phylum Firmicutes and formed cluster four (Fig. 4) together with clones from O. formosanus and C. formosanus. Phylotype OTG071 was related to bacteria in the genus Lactovum (94 % sequence identity with Lactovum miscens [AJ439543]). Phylotype OGH22 partly clustered with Lactococcus species while phylotype OTG032 was assigned to the family Ruminococcaceae and was distantly affiliated with members in the genus Ruminococcus (92 % sequence similarity with Ruminococcus sp. 14565 [AJ315980] and Ruminococcaceae bacterium AP7 [JX101688]). The genus Ruminococcus includes Ruminococcus flavefaciens, an anaerobic cellulolytic bacterium found in the rumen and in the hindgut of monogastric domestic and wild mammals (Bayer et al. 2008). These and other cellulolytic bacteria play an important role in the digestion of hemicellulose and cellulose plant cell walls (Bayer et al. 2008). Phylotype OGH102 was assigned to the order Clostridiales and was 92 % related to Ruminococcus sp. 14565. Planctomycetes accounted for a single phylotype in each termite (OTG070 and OGH02) and formed a subcluster with other clones from M. gilvus (Fig. 4). The phylum Actinobacteria was represented by a single phylotype OGH201 from Odontotermes sp. that was affiliated with Streptomyces sanglieri (JQ342914) isolated from soil.

Gut bacterial community structure in Microtermes sp.

The gut bacterial community structure of Microtermes sp. showed differences from that of Odontotermes species (Fig. 1). Phylotypes (27 % of the clones) belonging to the phylum Bacteroidetes were assigned to different genera. Phylotypes MIGH04 (five clones [JX421895]), MIGH32 (two clones [JX421896]), MIGH45 (one clone [JX421894]), MIGH50 (four clones [JX421893]), MIGH86 (five clones [JX421892]), MIGH88 (four clones [JX421897]), MIGH142 (five clones [JX421891]) and MIGH150 (three clones [JX421898]) were assigned to the genus Tannerella and formed large cluster three (Fig. 3) with clones often from guts of termites (R. speratus, Hodotermopsis sjoestedti, Cubitermes orthognathus) and insects. Phylotype MIGH143 formed a cluster with clones from other termites and was affiliated with Bacteroidetes bacterial clone from water (95 % sequence similarity). Only a single phylotype MIGH10 (5.1 %) was related to bacteria in the genus Alistipes (95 % sequence identity with A. finegoldii [AB554230]) and several other clones from M. gilvus, Odontotermes species and mouse gut (Fig. 2). Five phylotypes (MIGH70, MIGH134, MIGH139, MIGH117 and MIGH157) belonged to the order Bacteroidales and were distantly related (94–96 % sequence identity) to other clones obtained from different environments (termite gut, anaerobic reactor, industrial sediments, sludge and soil) (Fig. 2). Notably, seventeen phylotypes (29 % of the clones) were affiliated with the phylum Spirochaetes and formed a large cluster with treponemal clones and some Spirochaeta species from several termite guts, particularly wood feeding termites (Fig. 4). Phylotype MIGH131 (6.8 %), which was the most abundant in this group, formed part of cluster seven (Fig. 4) together with other clones from Odontotermes species, M. gilvus and Nasutitermes species. Phylotypes MIGH99 and MIGH128 were affiliated with bacteria in the genus Spirochaeta (95–96 % sequence similarity with Spirochaeta sp. AJ419823 from Zootermopsis angusticollis) and formed a subcluster with clones from Neotermes koshunensis and Reticulitermes species (Fig. 4). Several other phylotypes i.e. MIGH115 (16 clones [JX421877-79]), MIGH47 (three clones [JX421884]), MIGH89 (three clones [JX421885]), MIGH39 (two clones [JX421874]), MIGH52 (four clones [JX421875]), MIGH36 (three clones [JX421876]), MIGH69 (two clones [JX421881]), MIGH121 (four clones [JX421882]), MIGH138 (three clones [JX421883]), MIGH141 (four clones [JX421886]), MIGH60 (one clone [JX421887]), MIGH73 (nine clones [JX421888-89]), MIGH53 (one clone [JX421890]) formed a large monophyletic group (cluster eight) together with Spirochaeta species (AJ419817, AB015812) and other clones from termites (Nasutitermes sp., C. orthognathus and Termes comis) (Fig. 4).

The phylum Firmicutes accounted for 23 % of the total clones. Phylotypes MIGH123 (four clones [JX421936]), MIGH80 (eight clones [JX421937]), MIGH133 (three clones [JX421940]) and MIGH100 (one clone [JX421939]) formed part of cluster four (Fig. 4) together with other clones from Odontotermes species. These were affiliated with the genus Lactovum and clustered with L. miscens. Phylotypes MIGH59 and MIGH154 belonged to the family Ruminococcaceae and formed a subcluster with clones from Microcerotermes species (Fig. 4). These two phylotypes (MIGH59 and MIGH154) were affiliated with Clostridium sp. K13-19 (HE862234) and Bacteroides cellulosolvens (NR_025918) (94–95 % sequence identity) (Fig. 4). Clostridium sp. K13-19 and Bacteroides cellulosolvens are anaerobic cellulose-degrading bacteria that may have a role in degradation of plant biomass. For instance, Bacteroides cellulosolvens is known to bind tightly and degrade crystalline forms of cellulose (Giuliano and Khan 1984). Three phylotypes (MIGH159, MIGH135 and MIGH21) had <93 % sequence similarity with Ruminococcus sp. 14565 (AJ315980) and clustered with clones from Odontotermes and Reticulitermes species (Fig. 4). Phylotype MIGH28 belonged to the genus Anaerotruncus and was distantly related to Anaerotruncus sp. NML 070203 [EU815226] (94 % sequence identity). Phylotype MIGH84 (5.4 %) belonged to the order Clostridiales and was the most abundant in this phylum, forming a cluster with Clostridiales bacterium (JN713425) and two clones from M. gilvus (Fig. 4). One phylotype MIGH51 belonged to δ-Proteobacteria and clustered with clones from other termites. Phylotype MIGH151 (11.6 %), the most abundant in the clone library belonged to the subdivision α-Proteobacteria and was closely affiliated with Wolbachia species (>97 % sequence identity) (Fig. 4). The phyla Planctomycetes and Actinobacteria were the least represented, with a single phylotype from each phylum (MIGH219 and MIGH136, respectively). Phylotype MIGH136 formed a cluster with Actinobacterium P27 (JQ411258) and Propionicimonas paludicola (FR733712) (Fig. 4). Phylotype MIGH219 was closely affiliated with clones from M. gilvus and those from Odontotermes species.

The PCA (Fig. 5), indicated that the relative abundances of Alistipes and Treponema are the major effect determining the overall variance of the genus compositions in the samples, followed by the relative abundance of genus Hespellia. Differences regarding the other genera detected in the termite gut are negligible. Hespellia abundance increases in the direction of the soil feeder, Cubitermes sp., while abundance of Alistipes increases towards M. gilvus. Treponema, however, increases towards the wood feeder, R. flavipes. The relative abundance of the bacterial genera in the guts of fungus-cultivating termites, which are represented by more than one sample, showed considerable divergence but no genus-specific pattern.

Fig. 5
figure 5

Principal-component analysis of bacterial communities based on the relative abundances of the bacterial genera. The vectors indicate the direction and impact of each detected bacterial genus on the overall variance. Those with a negligible influence were not annotated. Abbreviations in figure: CuX Cubitermes sp., OdF O. formosanus, OdS O. somaliensis, OdX Odontotermes sp., ReF R. flavipes, MiX Microtermes sp.

Full size image

Discussion

The phylogenetic analysis of the 16S rRNA gene sequences from the three termites revealed diverse gut bacterial communities that are still uncultured. The affiliation of the cloned sequences with others previously obtained from termite guts underlines the existence of termite-specific bacterial lineages (Hongoh et al. 2003, 2005, 2006; Schmitt-Wagner et al. 2003; Shinzato et al. 2007; Warnecke et al. 2007) (Figs. 2, 3, 4). Notably, the results indicated a higher gut bacterial diversity in Odontotermes than in Microtermes. The bacterial community structure within the two here examined Odontotermes species was almost identical at the phylum level (Fig. 1). The phylum Synergistetes was exclusively detected in Odontotermes species (Fig. 4). Previously, members of Synergistetes were reported in guts of M. gilvus (Hongoh et al. 2006), M. michaelseni (Mackenzie et al. 2007) and O. formosanus (Shinzato et al. 2007), wood-feeding termites (Hongoh et al. 2005; Köhler et al. 2012) and in cockroaches (Schauer et al. 2012). However, they were not detected in the guts of soil-feeding termites (Schmitt-Wagner et al. 2003) and Macrotermes barneyi (Zhu et al. 2012). Members of the phylum Synergistetes mainly inhabit anaerobic environments, including gastrointestinal tracts, human clinical specimens, soil, oil wells and wastewater treatment plants (Vartoukian et al. 2007). Their presence at pathology related sites suggests they may be opportunistic pathogens; however, other species of Synergistetes are significant contributors in the degradation of sludge for production of biogas in anaerobic digesters (Riviere et al. 2009). Though their precise role in termites remains unclear, some Synergistetes species (e.g. Thermovirga lienii) have been implicated in amino-acid-degradation (Dahle and Birkeland 2006), which is an important process in the gut of the termites.

The PCA indicated that the relative similarity in the bacterial communities across the termites is mainly impacted by the genera Alistipes, Treponema, Tannerella, Dysgonomonus and Hespellia (Fig. 5). Their relative abundances considerably varied across the different termite samples (Fig. 5). Members of Treponema seem to be more predominant in wood feeders such as R. flavipes and in the Microtermes sp. (see Supplementary Table S1) while Hespellia appear more dominant in the soil-feeders such as Cubitermes species. However, it cannot be generalized at this point because these two guilds (wood feeders and soil-feeders) were represented by a single sample only, respectively. Rather, the considerable divergence in the relative abundances of these genera within the fungus-cultivating termites, which were represented by many samples, indicates that other factors besides feeding habits influence the abundances of the phyla within the gut bacterial communities (Sanyika et al. 2012).

The phylum Bacteroidetes showed differences in group abundance; for instance, the genera Parabacteroides, Bacteroides and Dysgonomonas were detected in Odontotermes but undetected in Microtermes (Supplementary Table S1). The genus Alistipes was represented by more phylotypes from Odontotermes (Fig. 2) than in Microtermes, which had more phylotypes affiliated with the genus Tannerella (Fig. 3) than with Odontotermes. The described species from the genus Alistipes (A. finegoldii, A. onderdonkii, A. shahii, A. putredinis and A. indistinctus) (Könönen et al. 2010; Nagai et al. 2010) have also been isolated from the intestines of healthy humans but their precise role is not yet known. The high abundance of Bacteroidetes in the termite guts (Figs. 2, 3) is in agreement with findings in other fungus-cultivating termites (Hongoh et al. 2006; Mackenzie et al. 2007; Zhu et al. 2012), suggesting that they play significant roles in the termites; for examples, members of the genera Tannerella, and Prevotella (Prevotella ruminicola 23, Prevotella intermedia 17) can ferment both xylan and cellulose through carbohydrate-active enzymes such as xylanase, carboxymethylcellulase and endoglucanase (http://www.cazy.org). Recently, D. oryzarvi was isolated from a microbial fuel cell (Kodama et al. 2012), which implicates it in cellulose degradation. Moreover, genome analysis has shown Bacteroides involvement in breaking down polysaccharides and metabolizing their sugars (Xu et al. 2003; Sonnenburg et al. 2010) by contributing glycosyl hydrolases for their hosts’ digestion (Liu et al. 2011). Bacteroidetes also benefit their host by excluding potential pathogens from colonizing the gut (Wexler 2007). However, some species such as Bacteroidetes fragilis have been implicated in diseases (Wexler 2007; Saulnier et al. 2011). Although the involvement of Bacteroidetes in degradation and fermentation of plant biomass partly implicates them in termite nutrition, it remains to be specifically determined.

Interestingly, Spirochaetes were more predominant in Microtermes than in Odontotermes (Fig. 4). This finding contrasts the literature (Shinzato et al. 2007; Zhu et al. 2012; Mathew et al. 2012), which shows infrequent detection of Spirochaetes in the guts of other fungus-cultivating termites. Notably, many members of Spirochaetes are host-associated and inhabit the oral cavity, intestinal tract and genital areas of humans or other mammals, as well as the gut contents of wood-feeding insects (Norris et al. 2010). They form an abundant group in the guts of most termites (Breznak 1984) (Fig. 4), especially wood-feeding termites (Hongoh et al. 2005; Köhler et al. 2012); this concurs with our results regarding Microtermes sp. (Fig. 4), where majority of the phylotypes were affiliated with treponemal clones from wood-feeding termites. The relative abundance of treponemal species in the gut substantiates their importance in termites. For instance, Ballor et al. (2012) revealed the presence of termite gut treponemes with multiple iron-hydrogenases and Treponema azotonutricium (an isolate from the lower termite Z. angusticollis) produces molecular hydrogen as a by-product of carbohydrate fermentation (Graber et al. 2004), which validates the suggestion (Köhler et al. 2012) that Spirochaetes are partly responsible for hydrogen production in termites. In addition, examination of Treponema strains ZAS-1, ZAS-2 and ZAS-9 revealed that they possess two homologues of nifH and each exhibited nitrogenase activity, demonstrating their involvement in nitrogen fixation (Lilburn et al. 2001).

It should be noted that members of the phyla Firmicutes and Proteobacteria are regularly encountered in termite guts (Fig. 4) and may play key functions. Two phylotypes (MIGH59 and MIGH154) detected in Microtermes sp. (Fig. 4) were affiliated with Clostridium sp. K13-19 (HE862234) and Bacteroides cellulosolvens (NR_025918) (94–95 % sequence identity). Clostridium sp. K13-19 and Bacteroides cellulosolvens are anaerobic cellulose-degrading bacteria that may have a role in degradation of plant biomass. For instance, Bacteroides cellulosolvens is known to bind and degrade crystalline forms of cellulose (Giuliano and Khan 1984). Some clones that were detected had affiliation with the genus Lactovum (Fig. 4) and clustered with L. miscens (isolated from acidic forest soil) that is reported to be involved in mixed fermentative metabolism (Matthies et al. 2004). Moreover, members of the subdivision δ-proteobacteria such as Desulfovibrio spp. isolated from termite guts display high rates of hydrogen-dependent oxygen reduction (Kuhnigk et al. 1996). Members of the genus Propionivibrio (ß-Proteobacteria) (represented by phylotypes OTG038 and OGH97) that were detected in the clone libraries were related to P. limicola (NR_025455) (Fig. 4). P. limicola has been shown to be fermentative and specialize in the degradation of hydroaromatic compounds (Brune et al. 2002). Members of α-Proteobacteria such as Wolbachia species that were also present in our clone libraries (Fig. 4) are associated with four distinct reproductive phenotypes in a wide range of Arthropoda: parthenogenesis, male killing, feminization and cytoplasmic incompatibility; nonetheless, little is known about possible phenotypes linked to Wolbachia in Isoptera (Werren et al. 2008).

There was low detection of Planctomycetes and Actinobacteria in all termites (Fig. 4). This trend has been reported in other termites (Shinzato et al. 2005; Fisher et al. 2007; Mackenzie et al. 2007; Long et al. 2010; Zhu et al. 2012). Nevertheless, Actinobacteria have diverse metabolic capabilities and some isolates from termite guts (Pasti and Belli 1985; Watanabe et al. 2003; Mackenzie et al. 2007) have cellulolytic activity (Pasti and Belli 1985; le Roes-Hill et al. 2011) and lignin-solubilizing activity (Pasti and Belli 1985; Pasti et al. 1990). In addition, Actinobacteria excrete antimicrobial peptides (Bulmer and Crozier 2004), which have been shown to inhibit the growth of some Pseudoxylaria and Termitomyces (Visser et al. 2012), hence preventing contamination in the farming of fungus gardens (Moriya et al. 2005).

Notably, the majority of the representative phylotypes found in our clone libraries are affiliated with sequences previously isolated from termites gut (Figs. 2, 3, 4). For instance, in O. somaliensis a total of forty-three phylotypes out 53 were affiliated with termite-related clones. Similarly, in Odontotermes sp. 27 phylotypes out of the 51 clustered with clones from fungus-cultivating termites. This trend has been reported elsewhere (Shinzato et al. 2007). However, in the Microtermes sp. several phylotypes were closely affiliated with clones previously obtained from non-fungus cultivating termites (Figs. 2, 3, 4). Previously, Hongoh et al. (2005) observed consistency in the bacterial phylogeny and the community structure within a genus of termites, which is in agreement with our findings regarding the two Odontotermes species (Fig. 1). Such observations suggest that majority of termite gut bacteria are specific symbionts that have coevolved with their hosts (Shinzato et al. 2005). The vertical mode of transmission of such gut microbes could be one of the major factors contributing to formation of the termite-specific bacterial lineages observed. The presence of clones affiliated with those originating from non-termite environments could emanate from either random acquisition of microorganisms from the environment (Curtis and Sloan 2004) or variation in the hosts’ diets (Tanaka et al. 2006). However, the factors determining the community structure of termite guts are still unclear (Shinzato et al. 2005) and further research is needed to address the mechanisms establishing the microbial community structures in the different termite species. It should be noted that there could be an underestimation of the diversity since individual taxa present in smaller numbers are difficult to detect due to PCR bias (von Wintzingerode et al. 1997; Farris and Olson 2007) and the RFLP screening method used (Supplementary Figure 1).

Conclusion

The findings of this study reveal a high level of bacterial diversity in the guts of fungus-cultivating termites, the majority of which are still uncultivated. This fact, coupled with the great diversity of termite species (Ahmed et al. 2011), challenge our ability to resolve the physiology and metabolic functions of the bacteria in the gut ecosystem. Nonetheless, the affiliation of the clones with those from guts of other termites demonstrates that the majority of the gut bacteria are autochthonous and have mutualistic relationship with their hosts (Hongoh et al. 2006; Shinzato et al. 2007).

Therefore, combined efforts using both culture and culture-independent methods are needed to comprehensively characterize the microbial species’ richness and their specific roles in the termite gut. Although the approaches used in this study cannot help infer physiological roles for the uncultured bacteria in the termites, the results provide key insights into bacterial community structure in the guts of fungus-cultivating termites and contribute to understanding gut bacterial diversity and their mutualism with termites.