Thermophilic enrichment of microbial communities in the presence of the ionic liquid 1-ethyl-3-methylimidazolium acetate


  • A.P. Reddy,

    1. Joint BioEnergy Institute, Emeryville, CA, USA
    2. Biological and Agricultural Engineering, University of California-Davis, Davis, CA, USA
    Search for more papers by this author
  • C.W. Simmons,

    1. Joint BioEnergy Institute, Emeryville, CA, USA
    2. Biological and Agricultural Engineering, University of California-Davis, Davis, CA, USA
    Search for more papers by this author
  • J. Claypool,

    1. Biological and Agricultural Engineering, University of California-Davis, Davis, CA, USA
    Current affiliation:
    1. Agricultural and Biosystems Engineering, Iowa State University, Ames, IA, USA
    Search for more papers by this author
  • L. Jabusch,

    1. Biological and Agricultural Engineering, University of California-Davis, Davis, CA, USA
    Search for more papers by this author
  • H. Burd,

    1. Joint BioEnergy Institute, Emeryville, CA, USA
    2. Earth Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA
    Search for more papers by this author
  • M.Z. Hadi,

    1. Joint BioEnergy Institute, Emeryville, CA, USA
    2. Physical and Life Sciences Directorate, Lawrence Livermore National Laboratory, Livermore, CA, USA
    Search for more papers by this author
  • B.A. Simmons,

    1. Joint BioEnergy Institute, Emeryville, CA, USA
    2. Physical and Life Sciences Directorate, Lawrence Livermore National Laboratory, Livermore, CA, USA
    Search for more papers by this author
  • S.W. Singer,

    1. Joint BioEnergy Institute, Emeryville, CA, USA
    2. Department of Biomass Science and Conversion Technology, Sandia National Laboratories, Livermore, CA, USA
    Search for more papers by this author
  • J.S. VanderGheynst

    Corresponding author
    1. Biological and Agricultural Engineering, University of California-Davis, Davis, CA, USA
    • Joint BioEnergy Institute, Emeryville, CA, USA
    Search for more papers by this author


Jean S. VanderGheynst, Department of Biological and Agricultural Engineering, University of California, Davis, CA 95616, USA. E-mail:



The aim of the study was to develop an approach to enrich ionic liquid tolerant micro-organisms that efficiently decompose lignocellulose in a thermophilic and high-solids environment.

Methods and Results

High-solids incubations were conducted, using compost as an inoculum source, to enrich for thermophilic communities that decompose switchgrass in the presence of the ionic liquid 1-ethyl-3-methylimidazolium acetate ([C2mim][OAc]). Ionic liquid levels were increased from 0 to 6% on a total weight basis incrementally. Successful enrichment of a community that decomposed lignocellulose at 55°C in the presence of 6% [C2mim][OAc] was achieved, when the [C2mim][OAc] level was increased stepwise from 2% to 4% to 5% to 6%. Pyrosequencing results revealed a shift in the community and a sharp decrease in richness, when thermophilic conditions were applied.


A community tolerant to a thermophilic, high-solids environment containing 6% [C2mim][OAc] was enriched from compost. Gradually increasing [C2mim][OAc] concentrations allowed the community to adapt to [C2mim][OAc].

Significance and Impact of the Study

A successful approach to enrich communities that decompose lignocellulose under thermophilic high-solids conditions in the presence of elevated levels of [C2mim][OAc] has been developed. Communities yielded from this approach will provide resources for the discovery of enzymes and metabolic pathways relevant to biomass pretreatment and fuel production.


The economical production of biofuels from lignocellulose will require major advances in the pretreatment and hydrolysis of plant cell walls (Richard 2010; Rubin 2008). One pretreatment approach that has received significant attention recently involves the use of ionic liquids that behave as biomass solvents (Mora-Pale et al. 2011). Studies have demonstrated disruption of the interactions between plant cell wall polymers and improvements in enzymatic hydrolysis of cellulose upon pretreatment with a variety of ionic liquids (Zakrzewska et al. 2010; Zavrel et al. 2009).

One challenge facing the use of ionic liquids for lignocellulose pretreatment is the inhibition of hydrolytic enzymes and fermentative organisms by residual ionic liquids remaining in the pretreated biomass (Matsumoto et al. 2004; Docherty and Kulpa 2005; Lee et al. 2005; Romero et al. 2008). Identifying metabolic pathways that confer tolerance of micro-organisms to ionic liquids and enzymes that remain active in ionic liquids is an active area of research (Park et al. 2012; Khudyakov et al. 2012). Because pretreatment processes are carried out at high temperatures, an additional requirement is enzymes and micro-organisms that function in thermophilic environments. Finally, because water increases the energy costs of processing lignocellulose to fuel, high-solids operation is an equally important trait for industrially relevant enzymes and organisms (Gerbens-Leenes et al. 2009).

The aim of this study was to develop an enrichment method that yielded a microbial community that was tolerant to the ionic liquid 1-ethyl-3-methylimidazolium acetate ([C2mim][OAc]) and that efficiently decomposed lignocellulose in a thermophilic and high-solids environment. [C2mim][OAc] was chosen because it is currently one of the most promising ionic liquids for lignocellulose pretreatment (Zavrel et al. 2009). Switchgrass was selected because of its potential for bioenergy production and reducing greenhouse gas emissions (Schmer et al. 2008; Spatari et al. 2005).

Materials and methods

High-solids incubations

Finished green waste compost was obtained from a commercial facility that composts agricultural residues including tree and vine prunings (Grover Soil Solutions, Zamora, CA, USA). Compost was solar-dried and stored at 4°C until applied as inocula. Switchgrass stem pieces were obtained from the laboratory of Dr Ken Vogel of USDA-ARS, University of Nebraska, Lincoln, NE, USA.

The switchgrass was size reduced using a leaf shredder and air- dried until the moisture was <10%. Then, it was further size reduced using a Wiley mill with a 10-mm screen. Switchgrass was stored in airtight containers at 4°C until needed.

Prior to incubation, switchgrass was wetted with minimal media with no additional carbon source (DeAngelis et al. 2010) to a target moisture content of 400 wt% dry basis (g water per g dry solid) (80 wt% wet basis (g water per g total) and allowed to equilibrate at 4°C overnight. Wetted switchgrass was inoculated with 10 wt% (g dry compost per g dry solid) compost immediately before incubation.

High-solids incubations were conducted as described previously with minor modifications (Reddy et al. 2011). Briefly, bioreactors with a 0·2 l working volume were loaded with 5–7 g dry weight of feedstock and compost mixture. Air was supplied to each bioreactor at 15 ml min−1 to maintain aerobic conditions. Incubator temperature was maintained at 35°C for 1 day, ramped to 55°C over 1 day, and held at 55°C for the duration of the experiment. To maintain a moisture content ideal for microbial activity, water lost during incubation was replaced on a weight basis and each bioreactor was mixed every 3 to 5 days.

The respiration rate of the microbial community, represented as CO2 evolution rate (CER), was measured for all incubated samples (Reddy et al. 2009). Carbon dioxide concentration was measured on the influent and effluent air of the bioreactors using an infrared CO2 sensor (Vaisala, Woburn, MA, USA) and flow was measured with a thermal mass flow meter (Aalborg, Orangeburg, NY, USA). Carbon dioxide and flow data were recorded every 20 min using a data acquisition system; carbon dioxide evolution rate (CER) and cumulative respiration (cCER) were calculated as described previously (Reddy et al. 2009).

High-solids enrichments were completed with stepwise increases in the concentration of 1-ethyl-3-methylimidazolium acetate ([C2mim][OAc]) on a wt% basis. The enrichment pathways are presented in Fig. 1. Every 5–7 days, respiration data from incubated treatments were examined and incubated switchgrass samples containing each enriched community were collected. Sources of inoculum for the next enrichment were selected based on [C2mim][OAc] levels and respiration data to facilitate further enrichment of communities active in the presence of increasing levels of [C2mim][OAc]. Fresh switchgrass was inoculated with 10 wt% (g dry enriched sample per g total dry weight) of the enriched community, the [C2mim][OAc] level was adjusted and transferred to a new bioreactor. The enrichment experiment ran for 7 weeks yielding a total of seven time points (T0, T1, T2, T3, T4, T5 and T6) for microbial community analyses.

Figure 1.

Treatment map showing [C2mim][OAc] treatment lineage applied to reactors. Text in boxes indicates [C2mim][OAc] concentration applied at time point designated in leftmost column. Treatments with asterisks designate samples for which extracted DNA underwent pyrosequencing. The bolded arrow indicates where in the lineage the temperature ramp from 35 to 55°C occurred.

At the end of the incubation, samples were collected for the measurement of microbial community structure and moisture content. Methods for microbial community structure analysis are described in detail in later sections of this article. Moisture content was measured gravimetrically after drying samples at 105°C for 24 h.

DNA extraction and small-subunit (SSU) rRNA amplicon pyrosequencing

Samples from bioreactors at each time point were snap-frozen in liquid nitrogen and homogenized with an oscillating ball mill (MM400; Retsch Inc., Newtown, PA, USA). DNA was extracted with a CTAB protocol described elsewhere (Allgaier et al. 2010) except that samples were subjected to beating using a horizontal vortex tube rack for five minutes rather than using a bead beater.

Small-subunit rRNA amplicon pyrosequencing was used to study adaptation of the enriched community to thermophilic conditions and increasing [C2mim][OAc] concentrations. A ~450 bp fragment of the small-subunit (SSU) rRNA gene was amplified using the broadly conserved primer pair 926F-1392R as described by Engelbrektson et al. (2010). The reverse primer included a 5 bp barcode for multiplexing of samples during sequencing. Barcoded amplicons were mixed in equal proportions prior to emulsion PCR following manufacturer's instructions and sequenced using the Roche 454 GS FLX Titanium technology (Banford, CT). Sequencing tags were quality trimmed and analysed using the pyroclust version of the software tool PyroTagger (Kunin et al. 2010) with a 220-bp sequence length threshold and an accuracy of 10% for low-quality bases.

Analysis of pyrosequencing data

Common ecological analyses were performed to characterize enriched microbial communities. Bray–Curtis dissimilarity values were calculated between each community in a pairwise fashion to quantify differences in community structure as a function of differences in richness (i.e. the number of species in a community) and evenness (i.e. how similar the abundance values are for each species in a community) (Bray and Curtis 1957). Specifically, Bray–Curtis dissimilarity, b, was calculated as:

display math(1)

where S is the total number of species among the two communities i and j, nik is the abundance of species k in population i, and Ni is the size of population i. Bray–Curtis dissimilarity values fall between 0 and 1, where 0 indicates that the two communities are identical and 1 indicates that the two communities are completely dissimilar and share no common species. Furthermore, Shannon index values (Shannon 1948), H′, were calculated for each community as follows:

display math(2)

where R is the richness of the community and pi is the relative abundance of species i in the community. Shannon indices (H′) provided a measure of diversity in each community, where larger values of H′ indicated greater diversity. Community evenness was quantified using Pielou's evenness index (Pielou 1969), J, calculated as:

display math(3)

J is constrained between 0 (highly uneven) and 1 (perfect evenness among all species).

Two-dimensional nonmetric multidimensional scaling (NMDS) (Kruskal 1964), in which pairwise differences in community composition (as quantified by Bray–Curtis dissimilarity values) are visualized as two-dimensional spatial relationships (i.e. communities that are similar appear clustered closer together than those that are dissimilar), was used for ordination of enriched communities over time.

As the source organism of a particular SSU rRNA gene sequence cannot always be identified to the species level during clustering of pyrosequencing reads to phylogenetic bins, the term operational taxonomic unit (OTU) is used instead of species when referring to micro-organisms identified through pyrosequencing. OTUs encompass both sequence clusters resolved to the species level and those that can only be identified at a more general taxonomic level (e.g. genus level). To reduce noise in statistical analysis, all singleton OTUs were removed from the data set. NMDS analyses and calculation of Shannon's diversity indices were performed using the vegan package in the R software environment ( The metaMDS function of vegan was used to find a stable two-dimensional NMDS based on Bray–Curtis distances from OTU relative abundances from 1000 random starts. Similarity percentage (simper) analysis was performed using Bray–Curtis dissimilarity values as described previously (Clark 1993). simper measures the contribution of each species to the overall Bray–Curtis dissimilarity between two communities with the goal of identifying the species most responsible for dissimilarity. Bray–Curtis dissimilarity values from the T5 time point of the control reactor and the reactor containing the most active microbial community in the presence of [C2mim][OAc] were used for simper calculations.


Microbial activity response to increasing ionic liquid concentration

After inoculation with compost, the microbial community was allowed to adapt to thermophilic conditions through two transfers T0 and T1. The average 5-day cumulative respiration from these two transfers was 130 mg CO2 g per dw. For the T2 through T6 enrichments without [C2mim][OAc], 7-day cumulative respiration (cCER) varied between 170 and 206 mg CO2 per g dw.

The community from T1 was used to inoculate a control sample (no [C2mim][OAc] added) and treatments containing 1% and 2% [C2mim][OAc]. While cCER in the treatments with [C2mim][OAc] varied between 37 and 109 mg CO2 per g dw, the respiration was lower compared with the control which had a cCER of 176 mg CO2 per g dw (data not presented).

The communities from the 1 and 2% T2 enrichments were used as inocula for further adaptation studies in the T3 enrichments. Cumulative respiration decreased with increasing [C2mim][OAc] for both the 1% and 2% T2 lineages (Fig. 2a). In particular, activity decreased from 41 to 2·5 mg CO2 per g dw when [C2mim][OAc] concentration increased from 4 to 6%.

Figure 2.

Changes in microbial activity with increasing ionic liquid levels and enrichment. (a) T3 cumulative respiration for the control and samples inoculated with biomass from 0% [C2mim][OAc] enrichment (▲), 1% [C2mim][OAc] enrichment (●) and 2% [C2mim][OAc] enrichment (■). (b) T4 cumulative respiration for the control and samples inoculated with biomass from 0% [C2mim][OAc] enrichment (△), 2% [C2mim][OAc] enrichment (○), and 3% [C2mim][OAc] enrichment (☐), 4% [C2mim][OAc] enrichment (♢).

T4 enrichments were completed using inocula from the 2 and 3% T3 communities from the 1% T2 lineage and the 4% T3 community from the 2% T2 lineage. This subset of inocula was selected based on cCER measured during the T3 enrichment and [C2mim][OAc] levels. As observed with T3 enrichments, cCER decreased with increasing [C2mim][OAc] for two of the inoculum sources tested (Fig. 2b); however, the inoculum derived from the T3 enrichment containing 4% [C2mim][OAc] appeared to tolerate increasing levels of [C2mim][OAc]; the cCER for all three samples only varied between 86 and 89 mg CO2 per g dw despite an increase in [C2mim][OAc] from 4% to 5%.

Additional enrichment studies (T5 enrichments) were completed using selected communities from the T4 enrichments as inoculum. The subset of inocula was selected based on cCER measured during the T4 enrichment and [C2mim][OAc] levels. Cumulative respiration was not calculated for T5 samples due to temporary equipment limitations at the beginning of the experiment. However, respiration rates measured at day 3 varied between 0·3 and 30 mg CO2 per day per g dw for the T5 enrichments containing [C2mim][OAc] indicating the community was still active (data not presented). A final enrichment (T6 enrichment) was completed using the 5% and 6% [C2mim][OAc] T5 communities as inocula. The T6 enrichment, which included treatments containing 5% and 6% [C2mim][OAc], yielded a community with cumulative respiration levels ranging between 5·5 and 15 mg CO2 per g dw (data not presented).

Changes in microbial community structure with increasing ionic liquid concentration

The community lineage demonstrating the highest activity in the presence of the greatest levels of [C2mim][OAc] was selected for SSU rDNA amplicon pyrosequencing. The selected community was one in which [C2mim][OAc] was 2% in T2, 4% in T3, 5% in T4 and 6% in T5 (Fig. 1). Treatments without [C2mim][OAc] were also analysed for comparative purposes. Ordination analysis of the control and 2% to 4% to 5% to 6% communities revealed a distinct shift when thermophilic conditions were applied (Fig. 3). NMDS of communities showed that the community structure at T1, which is common to both lineages sequenced and lacked [C2min][OAc], is distanced from all other communities at subsequent time points and ionic liquid levels. Communities with increasing levels of [C2min][OAc] beyond 2% clustered together separate from communities without ionic liquid. These data suggest that the early thermophilic incubation period in the absence of ionic liquid resulted in a change in microbial structure away from that initially present in the switchgrass and compost inoculum. The community adapted to 2% [C2min][OAc] was more similar to those cultured without ionic liquid compared with those adapted to higher levels of ionic liquid. Differences in community structure were less pronounced for communities at 4, 5 and 6% [C2mim][OAc], suggesting a lower degree of community restructuring as ionic liquid was added beyond 2%. Shannon diversity indices for each community revealed that the initial thermophilic incubation step common to both lineages led to a drop in community diversity (Table 1). Loss of diversity was due to decreases in both community richness and evenness. In communities lacking ionic liquid, richness was nearly lowered 90% following two weeks of thermophilic incubation, dropping from approximately 1000 OTUs at T1 to approximately 100 OTUs at T3. Community evenness decreased to approximately one-third of the initial value, from 0·85 to 0·27, over the same period. For communities cultured without [C2mim][OAc], Shannon indices and evenness values remained nearly constant for subsequent time points, although richness ultimately decreased to 67 at T5. Similarly, community diversity decreased with the addition of 2% [C2mim][OAc] (Table 1). However, there was a greater decrease in richness in the 2% ionic liquid culture compared to the community without ionic liquid at the same time point. Unlike communities without ionic liquid, community diversity continued to fall following T3 as ionic liquid levels increased to 4, 5 and 6%. Decreases in the Shannon indices at these ionic liquid levels were primarily due to decreased community evenness. Changes in community diversity corresponded to enrichment of specific OTUs. OTU relative abundance data revealed enrichment of certain taxa in both ionic liquid treatment lineages. Specifically, for both lineages, there was selection against the proteobacteria and chlamydiae present in the T1 community (Table 2 and Fig. 4). Lineages with and without ionic liquid both exhibited enrichment primarily of Firmicutes following the initial thermophilic incubation period. Within the Firmicutes phylum, there was enrichment of three OTUs for both lineages (Fig. 5). However, the evenness of the dominant OTUs was different for each lineage. Geobacillus dominated the [C2mim][OAc] treatment from the second time point onward. In contrast, Bacillus and Ureibacillus dominated the control community until the last time point, when the Geobacillus population increased. SIMPER analysis revealed that over 90% of the Bray–Curtis dissimilarity between the T5-enriched communities for each lineage could be explained by differences in Geobacillus, Bacillus and Ureibacillus levels (Table 3).

Figure 3.

Nonmetric multidimensional scaling of microbial communities enriched on switchgrass under thermophilic, high-solids conditions with and without addition of ionic liquid.

Figure 4.

Relative abundance of phyla in thermophillic and IL-tolerant microbial communities over time. (image) Firmicutes, (image) Proteobacteria, (image) Viridiplantae, (image) Chlamydiae and (□) all other phyla.

Figure 5.

Relative abundance of operational taxonomic units in thermophillic and IL-tolerant microbial communities over time. (image) Ureibacillus (Firmicutes), (image) Geobacillus (Firmicutes), (image) Bacillus (Firmicutes) and (□) all other taxa.

Table 1. Biodiversity measures for microbial communities enriched on switchgrass under thermophilic, high-solids conditions with and without addition of ionic liquid
IL Concentration of inoculum (%)Time pointIL (%)Shannon indexPielou indexRichness
Table 2. Changes in relative abundance (%) of selected operational taxonomic units (OTUs) in thermophilic and IL-tolerant microbial communities over time
OTUT1T2T3T4T5Taxonomy; greengenes ID (
  1. For each OTU cluster, relative abundance values within the community are presented for time points T1–T5. The identity of the OTU is given in the final column as genus (phylum).

  2. a

    (DeSantis et al. 2006).

No ionic liquid
Cluster 10 0·40·525·5Geobacillus (Firmicutes); 364707
Cluster 20 41·940·738·8Bacillus (Firmicutes); 367166
Cluster 30 48·648·028·1Ureibacillus (Firmicutes); 30329
Increasing ionicliquid
Cluster 1022·972·167·485·5Geobacillus (Firmicutes); 364707
Cluster 2040·314·715·21·7Bacillus (Firmicutes); 367166
Cluster 3032·31·21·10·1Ureibacillus (Firmicutes); 30329
Table 3. simper analysis between T5 microbial communities of higher-contributing (>90% cumulative contribution) operational taxonomic units (OTUs)
OTURelative abundance without ILRelative abundance with ILContribution (%)Taxonomy; greengenes ID (
  1. a

    DeSantis et al. 2006.

Cluster 125·585·544·8Geobacillus (Firmicutes); 364707
Cluster 238·8 1·727·7Bacillus (Firmicutes); 367166
Cluster 328·1 0·120·9Ureibacillus (Firmicutes); 30329


Respiration results demonstrated that a thermophilic, switchgrass-decomposing community was enriched during the high-solids incubation studies. Establishing an active microbial community on switchgrass under thermophilic conditions followed by gradual increases in [C2min][OAc] level yielded a community that colonizes switchgrass in a high-solids, thermophilic environment in the presence of ionic liquid. Decreases in respiration rate with increasing [C2min][OAc] levels indicated microbial activity was negatively influenced by [C2min][OAc]. In previous high-solids incubation studies in which compost-inoculated switchgrass was treated with [C2min][OAc] varying between 0 and 8% and temperature was ramped from 35 to 55°C, very little microbial activity was detected when temperature and [C2min][OAc] exceeded 45°C and 4%, respectively (Singer et al. 2011). Not allowing the community to adapt to [C2min][OAc] in a gradual manner may have prevented microbial growth in this earlier study. Previous high-solids enrichment studies with [C2min][OAc] also tended to enrich for fungi (Singer et al. 2011). The combination of an initial mesophilic temperature and immediate exposure of the community to ionic liquid may have selected for fungal growth.

Applying thermophilic conditions greatly shifted and reduced the complexity of the microbial community even in the absence of [C2min][OAc]. This observation was consistent with a similar enrichment performed on switchgrass (Reddy et al. 2011). In this enrichment, simplification of an active microbial community in thermophilic switchgrass enrichments was observed, indicating selection for a few dominant community members adapted to decompose switchgrass in a high-solids, thermophilic environment. The dominant members of the adapted community in the previous study were members of the genera Microbispora, Cohnella and Chelatococcus. The dominant members in the final enrichment in the present study were from the genus Geobacillus, Bacillus and Ureibacillus. Differences between the two enrichments on switchgrass may have been due to differences in feedstock preparation and enrichment time. The previous study used ethanol-extracted switchgrass, while the present study used switchgrass that was not extracted. The non-extracted switchgrass likely had higher levels of water-soluble organic compounds that contributed to differences in enrichment. The previous studies also employed a longer enrichment time of 2 weeks compared with 5–7 days for the current study. The shorter time may have favoured a community that decomposed the less recalcitrant polysaccharides in switchgrass. The differences in the composition of the adapted community from the two different enrichments approaches demonstrate the importance of incubation variables in the selection of microbial communities that decompose switchgrass.

Simplification of the community was also observed in enrichments that included [C2min][OAc]. The main difference between enrichments that included [C2min][OAc] and those that did not was the dominance of members of the genus Geobacillus. The community for the final enrichment without [C2min][OAc] had a relatively uniform distribution of Geobacillus, Bacillus and Ureibacillus, while the final enrichment with [C2min][OAc] was predominantly Geobacillus. Identification of the exact role of each organism in switchgrass decomposition requires further studies. Members of the genus Geobacillus have been observed in several types of compost operations including one compost based on hay (Sung et al. 2002) and another based on food waste (Ryckeboer et al. 2003) and have been isolated from other extreme environments (Poli et al. 2006; Nazina et al. 2005). Of particular relevance to biofuel production, certain species appear to be a promising source of cellulases (Rastogi et al. 2010) and some have the ability to produce ethanol under thermophilic conditions (Fong et al. 2006). This is the first report of tolerance of Geobacillus to an ionic liquid.

The results demonstrate that a community tolerant to a thermophilic, high-solids environment containing 6% [C2mim][OAc] can be enriched from compost. The rates at which both temperature and [C2mim][OAc] concentration increased had an effect on adaptation by the community. A gradual increase in [C2mim][OAc] concentration permitted the adaption of the microbial community to thermophilic, high-solids decomposition of switchgrass in the presence of [C2mim][OAc]. Further studies are needed to optimize this community and to explore the ability of the Geobacillus population to tolerate ILs.


We thank Tijana Glavina del Rio, Susannah Tringe and Stephanie Malfatti of the DOE Joint Genome Institute for their assistance in obtaining pyrotag sequencing data. This manuscript was performed as part of the DOE Joint BioEnergy Institute ( supported by the US Department of Energy, Office of Science, Office of Biological and Environmental Research, through contract DE-AC02-05CH11231 between Lawrence Berkeley National Laboratory and the US Department of Energy. Pyrotag sequencing was conducted by the Joint Genome Institute which is supported by the Office of Science of the US Department of Energy under Contract no. DE-AC02-05CH11231.