You are on page 1of 7

bs_bs_banner

Environmental Microbiology Reports (2014) doi:10.1111/1758-2229.12187

Functional congruence of rhizosphere microbial


communities associated to leguminous tree from
Brazilian semiarid region

Rodrigo Gouvêa Taketani,* distribution of rain (Pointing and Belnap, 2012). Hence, its
Vanessa Nessner Kavamura, Rodrigo Mendes and inhabitants have the ability to withstand the contrast
Itamar Soares Melo between dry and rainy seasons (Kavamura et al., 2013a).
Laboratory of Environmental Microbiology, Embrapa The Caatinga biome is a semiarid environment located in
Environment, Brazilian Agricultural Research Brazilian Northeast covering approximately 12% of its
Corporation-EMBRAPA, Jaguariuna, São Paulo, Brazil. territory (Trovão et al., 2007). The precipitation distribution
is predominantly concentrated in a period of 3–4 months
over the year, resulting in two contrasting seasons, i.e. dry
Summary
and wet; eventually, the dry season can last for more than
Semiarid environments are characterized by the 1 year (Trovão et al., 2007). The hydric stress caused by
uneven spread of rain throughout the year. This leads the lack of rainfall associated with a significant increase in
to the establishment of a biota that can go through temperature and ultraviolet exposure for surface dwellers
long periods without rain. In order to understand (Kavamura et al., 2013a) in combination with human
the dynamics of rhizosphere microbial communities predatory activity makes the Caatinga prone to desertifi-
across these contrasting seasons in Caatinga, we cation (Silva et al., 2004).
used the Ion Torrent platform to sequence the While many studies have addressed above-ground
metagenome of the rhizosphere of a native legumi- seasonal changes, effects on the below-ground micro-
nous plant (Mimosa tenuiflora). The annotation indi- flora have being neglected. In this context, we hypoth-
cated that most abundant groups detected were the esized that differences in climatic conditions between
Actinobacteria and Proteobacteria, and the dominant seasons would select for some key features in the
functional groups were carbohydrate and protein rhizosphere. In order to test this hypothesis, we evaluate
metabolisms, and that in the wet season, the commu- the rhizosphere microbiome of one of the most com-
nities carried carbohydrate and amino acid meta- mon leguminous tree found in the Caatinga, Mimosa
bolisms. The major differences observed between tenuiflora. Leguminous trees can represent 70% of the
seasons were higher abundance of genes related to number of plant individuals in the caatinga landscape
carbohydrate and amino acid metabolisms in the (Queiroz, 2006). This high prevalence can be attributed
rainy season, indicating that the populations present to the low amount of nitrogen often observed in these
might be better adapted to a higher abundance of soils (Queiroz, 2006). Among these plants, M. tenuiflora
organic matter. Besides, no clear separation of is one of the most abundant and widespread leguminous
samples was detected based on their taxonomic com- trees in this biome (Queiroz, 2006). It is also intensively
position whereas the functional composition indi- used as firewood and in constructions by the local
cates that samples from the rain season are more farmers (Queiroz, 2006). Hence, we have chosen this
related. Altogether, our results indicate that there is a species as our model because of its distribution, ecologi-
large functional stability in these communities mostly cal and social roles.
due to the selection of features that aid the biota to The interaction between the plant, the soil and its micro-
endure the dry season and blossom during rain. bial inhabitants shapes what we call the rhizosphere envi-
ronment (Bais et al., 2006). Many studies have shown the
participation of each of these as drivers of selection in this
Introduction
environment (Fierer and Jackson, 2006; Martiny et al.,
Semiarid environments are scattered around the globe 2006; Ramette and Tiedje, 2007). In some cases, this
and are characterized by the uneven spatial and seasonal interaction between plant and rhizospheric community
even allows the plant to withstand abiotic or biotic
Received 12 February, 2014; revised 4 June, 2014; accepted 7 June
2014. *For correspondence. E-mail rgtaketani@yahoo.com.br; Tel. stresses, such as desiccation (Kavamura et al., 2013a) or
+55-19-33112665; Fax +55-10-33112640. diseases (Philippot et al., 2013). In turn, the rhizosphere
© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd
2 R. G. Taketani, V. N. Kavamura, R. Mendes and I. S. Melo

environment created by the plant is a nutrient hotspot that lysed characteristics, while site 3 did, which makes site 1
allows for a high abundance of microorganisms (Bais and 2 much more similar between them than with site 3.
et al., 2006). These data were used as environmental data on the mul-
Recent advances in high throughput sequencing tech- tivariate statistics analysis.
nology have provided the ability to sequence metagenomic Total community DNA from soil and rhizosphere of
samples at much lower costs. This, together with the M. tenuiflora samples were extracted using Power Soil
availability of computational tools to analyse these data, DNA Isolation Kit (MoBio Laboratories, EUA) according to
has led many researchers to invest in direct metagenomic the manufacturer’s protocol. Total community DNA was
sequencing as a tool to investigate the effects of several checked in a NanoDrop spectrophotometer for quantity
factors over microbial communities (Venter et al., 2004; and quality (A260/280 was ~ 1.8) and in a Qubit® 2.0
Jones, 2009; Gilbert et al., 2010; Fierer et al., 2012). This fluorometer using Qubit® dsDNA BR Assay Kit to confirm
approach allows the description of the components of a DNA quantity. Six to eight independent DNA extractions
community without any selective procedure (besides DNA were pooled to obtain 1 µg of metagenomic DNA, which
extraction) as opposed to amplicon sequencing which is, was used to construct a library. In total, six libraries were
more often than not, specific to certain phylogenetic or obtained for sequencing, three from the dry season and
functional groups, displaying some limitations (Petrosino three from the rainy season.
et al., 2009). Besides that, the direct sequencing of a
metagenome provides information about the genes and
Metagenomic DNA sequencing using Ion Torrent
metabolic capabilities of a community (Hugenholtz and
Personal Genome Machine (PGM) platform
Tyson, 2008); this allows researchers to estimate how a
factor affects the community from two different and com- The total community DNA obtained as described above
plementary perspectives. Here, we used metagenomic was enzymatically fragmented and ligated into the appro-
techniques to analyse M. tenuiflora rhizosphere samples priate adapters using the Ion Xpress Plus Fragment
collected during the dry and rainy seasons in three different Library Kit following the manufacturer’s protocol. The
sites of Caatinga biome to gain further insight into meta- obtained fragments were then bound into the ion spheres
bolic and taxonomic shifts that occur in this plant- and amplified by emulsion polymerase chain reaction
associated microbial communities across seasons. using the Ion Xpress Template Kit. The enriched beads
were then sequenced in an Ion Torrent (PGM) using 314
chips and Ion PGM sequencing kit. Raw data were ana-
Materials and methods lysed on a torrent server for base calling. Sequence files
Ethics statement (sff) were exported and analysed as described later.

This project was conducted with the authorization of


the Institute of Environment and Renewable Natural Sequence analysis and statistical treatment
Resources process number 02001.004527/2011-90 and it The sff were uploaded to the MG-RAST server (Meyer
did not involve endangered or protected species. et al., 2008) for automatic annotation using the default
settings and are stored under the accession num-
bers: 4487345.3, 4487344.3, 4487567.3, 4487653.3,
Sampling, DNA extraction quality and yield
4487639.3 and 448787638.3. All phylogenetic analysis
Samples from the M. tenuiflora rhizosphere were col- presented here are the result of the Best Hit Classification
lected in three different sites of typical Caatinga vegeta- against the M5NR database using an E-value cut-off of
tion. Coordinates were: site 1 – 09° 13′ 24.8″ S, 41° 05′ 10−5, minimum identity of 60% and a minimum alignment
11.4″ W; site 2 – 08° 50′ 01.6″ S, 42° 33′ 13.3″ W; site 3 of 50 bp (Delmont et al., 2011). The functional annotation
– 06° 42′ 44.2″ S, 38° 15′ 08.2″ W. Samples were col- was performed by Hierarchical Classification against the
lected during the spring (October 2010) and winter (May Subsystems database using an E-value cut-off of 10−5,
2011) which corresponds to the peak of the dry and wet minimum identity of 60% and a minimum alignment of 15
seasons respectively. Samples were taken from three dif- amino acids (for a summary, see Table S1). Both annota-
ferent plants of similar sizes (height and trunk width) at tion tables were then exported to Statistical Analysis of
each site and were mixed in equal amounts. Samples Metagenomic Profiles (STAMP) (Parks and Beiko, 2010)
were kept in plastic bags and stored at room temperature to identify the differentially abundant features using Fish-
until processing. Physical and chemical analyses were er’s exact test; confidence intervals were calculated by
performed in each sample and were published previously Newcombe–Wilson’s method.
(Lançoni et al., 2013). According to this study, sites 1 and The phylogenetic and functional annotation tables were
2 did not present many significant differences in the ana- used in canonical correspondence analysis (CCA) fol-

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports
Functional congruence in semiarid 3

lowed by forward selection (FS) within the CANOCO v.4.5 functions. Only the minor groups represented ecological
software package. Also, analysis of similarity (ANOSIM) functions such as nutrient cycles, including those related
and Mantel tests were performed within Past software to the nitrogen cycle that we were expecting to find in
(Hammer et al., 2001); for this tests, all annotation tables higher abundance due to the ecological role of these
were converted into relative abundance tables and envi- plants. Such distribution was expected and has been
ronmental tables which were normalized by the Box–Cox observed in several other metagenomes from natural
method. ANOSIM was performed using a Bray–Curtis environments (Delmont et al., 2012; Fierer et al., 2012). In
similarity matrix. For the Mantel test, Bray–Curtis similarity dry areas such as deserts, function related to nutrient
matrices were constructed from annotation tables and cycling metabolism (nitrogen, potassium and sulphur)
Euclidian similarity matrixes were constructed from the tend to be lower than in non-desert environments and the
environmental data. opposite is observed for genes related to the metabolism
of carbohydrates, amino acids, proteins and dormancy/
sporulation genes (Fierer et al., 2012). The presence of
Results and discussion
higher numbers of carbon and protein metabolism genes
Rhizosphere soils collected from the three different sites might be linked to the ability of the soil microbiota of the
during the dry and rainy seasons along the Brazilian semi- semiarid biome to rapidly respond to the rewetting
arid environment were sequenced by Ion Torrent (PGM) (Borken and Matzner, 2009; Inglima et al., 2009). These
using a shotgun metagenomic approach. Sequences organisms increase their biological activity in a matter of
classified as bacterial represented about 83–98% of the hours and increase in numbers by several fold during the
annotated sequences. This ratio was similar to those pre- rainy period (Kavamura et al., 2013a). Also, since this
viously observed for other soil metagenomic studies environment is subjected to high heat, osmotic and oxi-
(Delmont et al., 2012; Fierer et al., 2012). The most abun- dative stresses, the large abundance of stress response
dant phyla in both small subunit (SSU) rRNA libraries and genes is not surprising and this has also been observed
gene annotation were Actinobacteria and Proteobacteria; by Fierer and colleagues (2012) when comparing desert
however, there were some discrepancies between the soils to non-desert soil samples.
abundance observed for the gene and SSU classifi- The alpha diversity indexes varied greatly across
cation. The 16S rRNA-based classification revealed samples. In terms of diversity, dry samples presented a
Actinobacteria as the dominant phylum in all libraries higher Shannon diversity index (H′) (based on class com-
(data not shown), whereas the gene classification placed position) and evenness [Buza and Gibson’s evenness
the Proteobacteria as dominant (Supporting Information (Buzas and Gibson, 1969)] (Fig. 1A). However, although
Fig. S1A). These differences must be due to the more the average of these indexes were different, this was not
comprehensive nature of Silva SSU database and/or dif- significant due to the large variation observed. The rank
ferent number of rRNA copies per functional gene ratio abundance curve (RAC) (Fig. 1B) of the taxonomic ranks
between these phyla. These groups have been previously shows that there is no particular pattern of dominance or
identified as major components of the microbial commu- evenness when comparing dry versus rainy samples.
nity in arid and semiarid environments (Barnard et al., Nevertheless, these results indicate that the diversity is
2013; Kavamura et al., 2013b) which are abundant in the not specially affected by lack of rainfall. Although previous
majority of the soils studied up to now (Janssen, 2006; studies have shown that the dry season leads to a
Fierer and Ladau, 2012; Fierer et al., 2012). Sequences decrease in microbial counts, diversity and richness
from archaeal origin were rare in all samples from 0.05% (Kavamura et al., 2013b), in our study this pattern was not
to 1.75% according to functional classification and from observed; however, it is important to note that the three
undetected to 1.89% on SSU libraries, which is most likely selected sites are not homogeneous in terms of soil nor
due to the lower coverage of the 16S rRNA of sequences. vegetation. The presence of a rich and diverse plant com-
The high variability of the communities observed in differ- munity is considered an important factor in the resilience
ent sites of semiarid soils have been observed in previous to desiccation (Bloor and Bardgett, 2012; Vogel et al.,
studies (Fierer et al., 2012; Barnard et al., 2013; 2012); this might also be true to the rhizosphere commu-
Kavamura et al., 2013b) and was highlighted by Barnard nity. The presence of a large number of species might
and colleagues (2013). benefit the functional redundancy, which would aid the
In terms of functional categories, all libraries have activity resumption upon the return of rain.
shown similar patterns among them (Supporting Informa- This functional redundancy can be accessed by the
tion Fig. S1B). The most abundant category observed means of the functional diversity. The observed H′, based
was carbohydrate metabolism followed by clustering on the functional diversity, was higher during rain as
based subsystems and protein metabolism. All of the 11 opposed to what was observed for the class composition.
most abundant categories were related to basic cellular However, the evenness was not different between the

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports
4 R. G. Taketani, V. N. Kavamura, R. Mendes and I. S. Melo

Fig. 1. Alpha diversity analysis of the Caatinga metagenomes.


A. Diversity indexes, Buza and Gibson’s Evenness and Shannon’s H′ based on the annotation by MG-RAST, i.e. function annotation by
hierarchical classification and taxonomic assignment by best hit classification.
B. Rank abundance curve of the taxonomic classification of the Caatinga metagenomes.
C. Rank abundance curve of the functional classification of the Caatinga metagenomes.

seasons, indicating that rain might have a positive effect distance between them and this site variation in samples
over the functional diversity. The RAC (Fig. 1C) also collected across large distances and that present physical
shows that there is a very different profile between and chemical differences are expected and has been
seasons. Also, the present result indicates that there is a seen previously in studies that used 16S gene libraries
contradiction between the effect of rain/dry over commu- (Barnard et al., 2013; Kavamura et al., 2013b). Therefore,
nity structure and function; these community characteris- this observation suggests that if we had a larger number
tics may respond differently when challenged by the of libraries (like in the mentioned studies), we could be
changes in the environment. This also led us to believe able to detect such features as seen previously (Barnard
that there might be an important role of functional re- et al., 2013; Kavamura et al., 2013b).
dundancy among different taxa (even different phyla) On the other hand, since the variation between samples
that allows the discrepancy between the functional/ in the functional assignment of the reads was less pro-
phylogenetic results. nounced, the Welch’s t-test (implemented in STAMP)
Due to the large variation observed between samples could identify different functional groups that were
from the same season and low correlation between them, selected by each season (Fig. 2). The central carbohy-
the comparisons based on the phylogenetic assignment drate metabolism cluster was the most significantly abun-
of the sequences did not produce any taxon that could be dant function to be negatively affected by desiccation.
marked as significantly different. However, it does not Since during this period, the organic matter and the micro-
mean that there is no taxon that is selected by either bial activity are decreased (Inglima et al., 2009) and there
season; it just means that the variation was large enough is a paucity of plant biomass that reduces the inputs of
to prevent the identification of such taxa. It is also impor- organic carbon (Fierer et al., 2012) due to the loss of
tant to note that the samples were collected with a long leaves by the majority of plants; the low abundance

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports
Functional congruence in semiarid 5

Fig. 2. Differentially abundant taxa as identified by STAMP of the comparison of dry (black bars) and rainy season (white bars) metagenomes
error bars indicate the 95% confidence intervals. ammon, ammonia assimilation.

of microorganisms carrying a high genomic content diversity in the Caatinga based on gene markers (SSU
related to the carbohydrate metabolism is expected. Also, and amoA gene) (Kavamura et al., 2013b; Lançoni et al.,
three other groups differentially abundant between 2013).
seasons were associated with amino acids and deriva- Functional redundancy has been often seen as a
tives, i.e. arginine; urea and polyamines; glutamine, crucial factor on the resilience of an ecosystem to any
glutamate, aspartate, asparagine and ammonia assimila- disturbance. In the case of microbial communities, due to
tion; and proline and 4-hydroxyproline. The first cluster is its overwhelming magnitude and richness, taxon diversity
related to the production of urea that would lead to the has always been used as a proxy to functional diversity.
loss of nitrogen, and the remaining two clusters are within This study represents one of the first analyses of Brazilian
the central amino acid metabolism and the assimilation of semiarid metagenomes conducted to date with over
ammonia. Interestingly, the first cluster is increased in the 250 Mb of metagenomic data obtained from six different
dry season while the other two are increased during rain, libraries. Nevertheless, we could detect some patterns in
which is concordant to the pattern of soil’s nitrogen the datasets that highlight the same trend in the shifts of
content in this forest (Trovão et al., 2007), i.e. nitrogen is communities when compared with studies of the same
lost during the dry season and accumulated in rain. biome (Kavamura et al., 2013b; Lançoni et al., 2013).
Furthermore, the higher abundance of cell division, Furthermore, microbial communities found in semiarid
polyhydroxybutyrate metabolism, purines and folate and soils are known to display two different life moments, one
pterines reads during rain also indicates that population of high activity (Barnard et al., 2013) and increased diver-
with higher metabolic activity are selected during this sity characterized by the presence of Proteobacteria due
season. to the presence of rain, and another one of lower activity
The application of CCA (a direct gradient analysis, and abundance of Actinobacteria, in a strategy of prepar-
i.e. uses only operational taxonomic unit (OTU) data that edness. However, despite the differences observed on
correlates with environmental data) also showed that dry the studied taxa, due to the variability of the sites, in
samples had lower similarity among them compared with general, the functional traits detected were generally
samples obtained from the rainy season (Supporting stable between sites and seasons suggesting that the
Information Fig. S2A and B). Although CCA allows the native microbial communities of semiarid rhizosphere
correlation between the OTU tables and environmental soils of the Caatinga might have selected soil organisms
variables, automatic FS with Monte Carlo permutation to display the same functional roles to cope with stress
(999 permutations) indicates that none of them had a despite their phylogenetic assignment. Therefore, we
significant effect over the community structure. Despite believe that this might be responsible for the resilience
the season, phylogenetic differences had been observed expected from these communities.
among samples from the three sites. Although there has
not been a significant effect of environmental variables Acknowledgements
over the community structure through CCA, soil variabil-
The authors acknowledge the support of Cosme Corrêa dos
ity might have influenced this variation, since some of
Santos (UEFS) on identifying the plant species Mimosa
these characteristics (i.e. temperature, moisture content,
tenuiflora. RGT was a recipient of a postdoctoral grant from
nutrient availability) have been pointed out as the main FAPESP (2010/50799-7). This study was supported by
drivers of soil bacterial communities variability (Fierer Embrapa. The authors also thank João Luiz da Silva and
et al., 2012; Pasternak et al., 2013) as observed for the Suikinai Nobre Santos for the support during sampling.

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports
6 R. G. Taketani, V. N. Kavamura, R. Mendes and I. S. Melo

References Kavamura, V.N., Taketani, R.G., Lançoni, M.D., Andreote,


F.D., Mendes, R., and Soares de Melo, I. (2013b) Water
Bais, H.P., Weir, T.L., Perry, L.G., Gilroy, S., and Vivanco, regime influences bulk soil and rhizosphere of cereus
J.M. (2006) The role of root exudates in rhizosphere inter- jamacaru bacterial communities in the Brazilian Caatinga
actions with plants and other organisms. Annu Rev Plant biome. PLoS ONE 8: e73606.
Biol 57: 233–266. Lançoni, M.D., Taketani, R.G., Kavamura, V.N., and de Melo,
Barnard, R.L., Osborne, C.A., and Firestone, M.K. (2013) I.S. (2013) Microbial community biogeographic patterns in
Responses of soil bacterial and fungal communities the rhizosphere of two Brazilian semi-arid leguminous
to extreme desiccation and rewetting. ISME J 7: 2229– trees. World J Microbiol Biotechnol 29: 1233–1241.
2241. Martiny, J.B.H., Bohannan, B.J.M., Brown, J.H., Colwell,
Bloor, J.M.G., and Bardgett, R.D. (2012) Stability of above- R.K., Fuhrman, J.A., Green, J.L., et al. (2006) Microbial
ground and below-ground processes to extreme drought in biogeography: putting microorganisms on the map. Nat
model grassland ecosystems: interactions with plant Rev Microbiol 4: 102–112.
species diversity and soil nitrogen availability. Perspect Meyer, F., Paarmann, D., D’Souza, M., Olson, R., Glass,
Plant Ecol Evol Syst 14: 193–204. E.M.M., Kubal, M., et al. (2008) The metagenomics RAST
Borken, W., and Matzner, E. (2009) Reappraisal of drying server – a public resource for the automatic phylogenetic
and wetting effects on C and N mineralization and fluxes in and functional analysis of metagenomes. BMC
soils. Glob Chang Biol 15: 808–824. Bioinformatics 9: 386.
Buzas, M.A., and Gibson, T.G. (1969) Species diversity: Parks, D.H.D., and Beiko, R.R.G. (2010) Identifying biologi-
benthonic foraminifera in Western North Atlantic. Science cally relevant differences between metagenomic commu-
163: 72–75. nities. Bioinformatics 26: 715–721.
Delmont, T.O., Malandain, C., Prestat, E., Larose, C., Monier, Pasternak, Z., Al-Ashhab, A., Gatica, J., Gafny, R., Avraham,
J., Simonet, P., et al. (2011) Metagenomic mining for micro- S., Minz, D., et al. (2013) Spatial and temporal biogeogra-
biologists. ISME J 5: 1837–1843. phy of soil microbial communities in arid and semiarid
Delmont, T.O., Prestat, E., Keegan, K.P., Faubladier, M., regions. PLoS ONE 8: e69705.
Robe, P., Clark, I.M., et al. (2012) Structure, fluctuation Petrosino, J.F., Highlander, S., Luna, R.A., Gibbs, R.A., and
and magnitude of a natural grassland soil metagenome. Versalovic, J. (2009) Metagenomic pyrosequencing and
ISME J 6: 1677–1687. microbial identification. Clin Chem 55: 856–866.
Fierer, N., and Jackson, R.B. (2006) The diversity and bio- Philippot, L., Raaijmakers, J.M., Lemanceau, P., and van der
geography of soil bacterial communities. Proc Natl Acad Putten, W.H. (2013) Going back to the roots: the microbial
Sci USA 103: 626–631. ecology of the rhizosphere. Nat Rev Microbiol 11: 789–
Fierer, N., and Ladau, J. (2012) Predicting microbial distribu- 799.
tions in space and time. Nat Methods 9: 549–551. Pointing, S.B., and Belnap, J. (2012) Microbial colonization
Fierer, N., Leff, J.W., Adams, B.J., Nielsen, U.N., Bates, and controls in dryland systems. Nat Rev Microbiol 10:
S.T., Lauber, C.L., et al. (2012) Cross-biome meta- 551–562.
genomic analyses of soil microbial communities and their Queiroz, L.P. (2006) The Brazilian Caatinga: phytogeo-
functional attributes. Proc Natl Acad Sci USA 109: graphical patterns inferred from distribution data of the
21390–21395. Leguminosae. In Neotropical Savannas and Dry Forests:
Gilbert, J.A., Meyer, F., and Bailey, M.J. (2010) The future of Plant Diversity, Biogeography, and Conservation.
microbial metagenomics (or is ignorance bliss?). ISME J 5: Penninigton, R.T., Lewis, G.P., and Ratter, J.A. (eds).
777–779. Oxford, UK: Taylor & Francis CRC Press, pp. 113–149.
Hammer, Ø., Harper, D.A.T., and Ryan, P.D. (2001) Ramette, A., and Tiedje, J.M. (2007) Multiscale responses of
Paleontological statistics software package for education microbial life to spatial distance and environmental hetero-
and data analysis. Palaeontol Electron 4: 1–9. geneity in a patchy ecosystem. Proc Natl Acad Sci USA
Hugenholtz, P., and Tyson, G.W. (2008) Metagenomics. 104: 2761–2766.
Nature 455: 481–483. Silva, J.M.C., Tabarelli, M.T., Fonseca, M.T., and Lins, L.V.
Inglima, I., Alberti, G., Bertolini, T., Vaccari, F.P., Gioli, (2004) Biodiversidade da caatinga: áreas e ações
B., Miglietta, F., et al. (2009) Precipitation pulses prioritárias para conservação. Brasília, UK: Ministério do
enhance respiration of Mediterranean ecosystems: the Meio Ambiente.
balance between organic and inorganic components of Trovão, D.M.B., Fernandes, P.D., de Andrade, L.A., and
increased soil CO 2 efflux. Glob Chang Biol 15: 1289– Neto, J.D. (2007) Variações sazonais de aspectos
1301. fisiológicos de espécies da Caatinga. Rev. Bras.
Janssen, P.H. (2006) Identifying the dominant soil bacterial deEngenharia Agrícola e Ambient 11: 307–311.
taxa in libraries of 16S rRNA and 16S rRNA genes. Appl Venter, J.C., Remington, K., Heidelberg, J.F., Halpern, A.L.,
Environ Microbiol 72: 1719–1728. Rusch, D., Eisen, J.A., et al. (2004) Environmental genome
Jones, W.J. (2009) High-throughout sequencing and shotgun sequencing of the Sargasso Sea. Science 304:
metagenomics. Estuaries and Coasts 33: 944–952. 66–74.
Kavamura, V.N., Santos, S.N., Silva, J.L., Parma, M.M., Vogel, A., Scherer-Lorenzen, M., and Weigelt, A. (2012)
Avila, L.A., Visconti, A., et al. (2013a) Screening of Brazil- Grassland resistance and resilience after drought depends
ian cacti rhizobacteria for plant growth promotion under on management intensity and species richness. PLoS
drought. Microbiol Res 168: 183–191. ONE 7: e36992.

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports
Functional congruence in semiarid 7

Supporting information Fig. S2. Multivariate statistics analysis of the classification


of metagenomic reads from the rhizosphere of Mimosa
Additional Supporting Information may be found in the online tenuiflora.
version of this article at the publisher’s web-site: A. Canonical correspondence analysis of the class-level
Fig. S1. Metagenomic community profiles obtained by the phylogenetic classification of metagenomic reads.
phylogenetic and functional classification of the caatinga B. Canonical correspondence analysis of the functional
metagenomes. classification.
A. Phylogenetic classification represents the affiliation up to Table S1. Overview of the Mimosa tenuiflora rhizosphere
the level of phylum by the best hit classification tool. metagenomes.
B. Functional classification represents the MG-RAST classi-
fication as assigned by the hierarchical classification tool.

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Reports

You might also like