Comparative Analysis of Prokaryotic Communities Associated with Organic and Conventional Farming Systems

RESEARCH ARTICLE Comparative Analysis of Prokaryotic Communities Associated with Organic and Conventional Farming Systems Elizaveta Pershina 1,2 *, Jari Valkonen 3 , Päivi Kurki 4 , Ekaterina Ivanova 1,5 , Evgeny Chirak 1 , Ilia Korvigo 1 , Nykolay Provorov 1 , Evgeny Andronov 1,2 1Laboratory of microbiological monitoring and bioremediation of soils, All-Russia Research Institute for Agricultural Microbiology, Saint-Petersburg, Russia, 2Saint-Petersburg State University, Saint-Petersburg, Russia, 3Department of Agricultural Sciences, University of Helsinki, Helsinki, Finland, 4Natural Resources Institute Finland, Mikkeli, Finland, 5Laboratory of biology and biochemistry of soils, V.V. Dokuchaev Soil Science Institute, Moscow, Russia *[email protected] Abstract One of the most important challenges in agriculture is to determine the effectiveness and environmental impact of certain farming practices. The aim of present study was to determine and compare the taxonomic composition of the microbiomes established in soil following long-term exposure (14 years) to a conventional and organic farming systems (CFS and OFS accordingly). Soil from unclared forest next to the fields was used as a control. The analysis was based on RT-PCR and pyrosequencing of 16S rRNA genes of bacteria and archaea. The number of bacteria was significantly lower in CFS than in OFS and woodland. The highest amount of archaea was detected in woodland, whereas the amounts in CFS and OFS were lower and similar. The most common phyla in the soil microbial communities analyzed were Proteobacteria (57.9%), Acidobacteria (16.1%), Actinobacteria (7.9%), Verrucomicrobia (2.0%), Bacteroidetes (2.7%) and Firmicutes (4.8%). Woodland soil differed from croplands in the taxonomic composition of microbial phyla. Croplands were enriched with Proteobacteria (mainly the genus Pseudomonas), while Acidobacteria were detected almost exclusively in woodland soil. The most pronounced differences between the CFS and OFS microbiomes were found within the genus Pseudomonas, which significantly (p<0,05) increased its number in CFS soil compared to OFS. Other differences in microbiomes of cropping systems concerned minor taxa. A higher relative abundance of bacteria belonging to the families Oxalobacteriaceae,Koribacteriaceae,Nakamurellaceae and genera Ralstonia,Paenibacillus and Pedobacter was found in CFS as compared with OFS. On the other hand, microbiomes of OFS were enriched with proteobacteria of the family Comamonadaceae (genera Hylemonella) and Hyphomicrobiaceae, actinobacteria from the family Micrococcaceae, and bacteria of the genera Geobacter,Methylotenera,Rhizobium (mainly Rhizobium leguminosarum) and Clostridium. Thus, the fields under OFS and CFS did not differ greatly for the composition of the microbiome. These results, which were also PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 1/16 OPEN ACCESS Citation: Pershina E, Valkonen J, Kurki P, Ivanova E, Chirak E, Korvigo I, et al. (2015) Comparative Analysis of Prokaryotic Communities Associated with Organic and Conventional Farming Systems. PLoS ONE 10(12): e0145072. doi:10.1371/journal. pone.0145072 Editor: A. Mark Ibekwe, U. S. Salinity Lab, UNITED STATES Received: July 25, 2015 Accepted: November 28, 2015 Published: December 18, 2015 Copyright: © 2015 Pershina et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information file. Funding: Financial support to the AgroBiotechnology focused on Root-Microbe Systems Baltic Sea region network (AB-RMS) was received from Norden/NordForsk (Authors: Jari Valkonen, Päivi Kurki):http://www.nordforsk.org/en. Financial support for bioinformatics data analysis was obtained from Russian Science Foundation (RSF) grant 14-2600094: "Analysis of the genetic and evolutionary potential of the soil microbiome to improve plant productivity and soil fertility" (Authors: Evgeny Chirak, confirmed by cluster analysis, indicated that microbial communities in the field soil do not necessarily differ largely between conventional and organic farming systems. Introduction Soil microorganisms can serve as bioindicators of anthropogenic stress experienced by the soil during agricultural use [1]. For a long period of time, biologically valuable soil microorganisms have been studied by isolation and cultivation in laboratory [2]. The next-generation sequencing technologies have intensified exploration of soil microbial diversity and allowed to identify biological indicators, not only among the microbes that can be cultured in vitro, but also among the bacteria and archaea which cannot be cultured [3]. One of the most important challenges of modern agriculture is to determine the effectiveness and environmental impact of systems based on organic or conventional farming (OFS and CFS respectively). Organic farming is considered ecologically friendly and to have less damaging effects on the ecosystem, whereas conventional agriculture is thought to cause significant changes in biocenoses due to the intensive inputs of synthetic fertilizers [4,5,6]. Productivity in CFS is generally higher than in OFS, but the negative impact on the environment associated with the use of a particular type of farming system is debated [7,8,9]. The DOK (short for the German words dynamic, organic and conventional, respectively) experiment is one of the most comprehensive studies on the long-term effects of diverse agricultural techniques on the ecosystem [10]. Studying the soil microbial diversity by pyrosequencing and analysis of the taxonomic markers for bacteria and fungi Hartmann et al. (2014) found that organic fertilizer amendments had a positive effect on the composition of microbial communities and on the α-diversity parameters. Organic matter inputs increased the richness and decreased evenness indices [10]. This effect has been found also in other studies [11,12]. On the other hand, significant increase in richness can lead to positive or neutral effects on evenness in systems with organic fertilizer amendment [13,14,15]. The fluctuations in α-diversity parameters are often explained by predominance of the copiotrophic microorganisms, whose growth is stimulated by organic fertilizers [10,16]. But this statement is equitable only in the short-time experiments, particularly it was shown that copiotrophic bacteria are temporarily stimulated by the addition of organic fertilizers to soil. In the long-run, under stable conditions, the ratio of oligotrophic to copiotrophic bacteria may be greater in OFS than in CFS [17]. Indeed, taxonomic analyses indicate at the phylum level that Proteobacteria and Firmicutes tend to dominate in the organic farming systems, while Actinobacteria, and to a lesser extent Acidobacteria, predominate in conventionally managed croplands and natural environments [18,19]. High abundance of the plant growth-promoting bacteria (PGPB), mainly the genera Rhizobium,Bradyrhizobium,Mesorhizobium,Burkholderia,Stenotrophomonas,Pseudomonas, Sphingomonas and Rhodoplanes, has been documented among the proteobacteria in OFS [13,16,18,2]. Firmicutes in croplands are represented by bacteria capable of degrading various complex organic materials and include, e.g., the genera Bacillus,Clostridium,Epulopiscium, Paenibacillus and Solibacillus [10]. These data obtained by the modern molecular techniques are partially consistent with the data obtained using bacterial cultivation techniques [20]. Microbial dynamics associated with certain land-use practices must be considered together with the spatial and temporal variations in microbial composition, occurring in soil as a result of plant growth and seasonal changes. Spatial fluctuations in soil microbial communities derive from the unequal distribution of the organic compounds within individual soil aggregates or Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 2/16 Ilia Korvigo, Elizaveta Pershina, Evgeny Andronov): http://grant.rscf.ru/forms?rid=2Wv-G10JrV83i8XBU2cLb00. Discussion chapter was prepared together with the research scientists participating in Russian Science Foundation (RSF) grant 14-2600079: "Bio-physical and chemical diagnosis of soil organic matter quality for the development of scientific and theoretical basis of Agrobiotechnology" (Author: Ekaterina Ivanova, http://grant.rscf.ru/forms? rid=1IgCW10JqtDX3i8WD72cLb00.) The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. horizons [21,22,23] and on the different distances from the plant roots [24,25]. As it was shown by van Diepeningen and co-workers the composition of soil microbial community oscillated, depending on the distance remaining from the root. These wavelike patterns were detected both for oligotrophic and copiotrophic bacteria both in OFS and CFS soils but were significantly stronger in conventional croplands [24]. One of the main advantages of employing the pyrosequencing techniques in biodiversity studies is the improvement of knowledge about the impact of agriculture on unculturable microorganisms. The most pronounced effect on soil microbiome revealed in the DOK experiment was the impact of organic fertilizers to the abundance of Acidobacteria in soil. In vitro cultivation methods for this bacterial phylum are lacking for most of its members, except in rare attempts to define the role of these bacteria in the agricultural systems managed with organic fertilizers amendments [26]. Among acidobacteria, genera Cand.Solibacter and Cand. Koribacter have been found exclusively associate with CFS, whereas Chloracidobacteria and the RB25 group have been found associated with OFS in previous studies [10]. The aim of this study was to compare long-term impacts of OFS and CFS on microbial diversity in soil. In the experimental station Karila (Mikkeli, Finland) where OFS and CFS have been carried out in adjacent fields for 14 years. The aim was to compare the taxonomic structure of microbiomes in OFS and CFS and to identify microbes specifically inhabiting these ecosystems. Materials and Methods Soil sampling Field experiments were carried out under permission of the Natural Resources Institute Finland (formerly MTT AgriFood Research Finland). The field studies did not involve endangered or protected species. Sampling was done at once from CFS and OFS fields in the experimental station Karila (Mikkeli, Finland) during the season of active plant growth in July 2011. The fields had been cleared from pine-spruce forest in the beginning of 20th century and therefore soil samples were collected from the pine-spruce forest next to the fields included for comparison (Table 1). The soil type was a coarser fine sand in both sampling fields. According to US soil taxonomy soil was sandy Aquic Haplocryod. Soil samples were taken from the top soil layer (10 cm) using soil drill (Ø 1 cm). At each sampling site three circles (Ø 1 m) were marked and 10 soil subsamples were taken from inside each circle and combined. Hence, three samples (replicates) were obtained for analysis from each type of soil (woodland, CFS and OFS). The distance between the sampling sites was 45 m in average. All samples were immediately transported to the laboratory and Table 1. Summary of the cultivation history of the fields sampled in the experimental station “Karila” (Mikkeli, Finland). For details, see S1 Table. Year CFS OFS 1928 The forest was cut down 1997–2010 Application of the CFS, regular input of mineral fertilizers Application of the OFS, regular organic fertilization with cow slurry and green manure 1997–2006 Sowing of spring cereals, black currant (in one part of the field) Crop rotation in 4 steps (1997–2010): 1)spring cereal with ley 2) 3 years of clover-grass ley 3) spring cereal 4)vetch-oats 2007 Bare fallow, glyphosate was used 2008–2010 Ley with oats doi:10.1371/journal.pone.0145072.t001 Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 3/16 stored at -70°C. Coordinates of the sampling sites were the following: woodland soil sample 1 (N61°40'32.46", E27°13'53.70"), 2 (N61°40'32.04", E27°13'55.80") and 3 (N61°40'31.86", E27° 13'57.54"); OFS soil sample 1 (N61°40'29.64", E27°13'40.44"), 2 (N61°40'30.00", E27°13'44.40") and 3 (N61°40'30.42", E27°13'48.96"); and CFS soil sample 1 (N61°40'38.22", E27°13'50.04"), 2 (N61°40'37.50", E27°13'51.24") and 3 (N61°40'36.30", E27°13'53.16"). The cultivation history of the fields in Karila is presented in Table 1. Details of the cultivation practices during the last three growing seasons prior to sampling are provided in S1 Table. At the time of sample collection timothy grass (Phleum pratense) and meadow fescue (Festuca pratensis) were grown as a mixture in both sampled fields (OFS and CFS). Besides analysis of the microbiome, the soil samples from OFS, CFS and woodland were subjected to agrochemical analyses (Table 2). DNA extraction DNA was extracted from 0.2 g of soil using PowerSoil DNA Isolation Kit (Mobio Laboratories, Solana Beach, CA, USA), which included a bead-beating step, according to the manufacturer’s specifications. Homogenization of the samples was performed using FatsPrep (MP Biomedicals, Santa Ana, CA, USA). The purity and quantity of DNA were tested by electrophoresis in 0.5× TAE buffer on 1% agarose. DNA concentrations were measured at 260 nm using SPECTROStar Nano (BMG LABTECH, Ortenberg, Germany). The average DNA yield was 2–5μg DNA with the concentration of 10–50 ng/μl. Quantitative PCR analyses Relative abundances of bacterial and fungal small subunit rRNA gene copies were analyzed by quantitative PCR (qPCR) (reaction volume 25 μl) using iQ™SYBR Green Supermix (BIO RAD, Hercules, USA) and 10 ng of sample DNA. For bacteria, the forward primer Eub338 and reverse primer Eub518 were used [27]. The forward primer arc915 and the reverse primer arc1059r were used for archaea [28]. To estimate bacterial and archaeal small-subunit rRNA gene abundances, standard curves were generated using a 10-fold serial dilution of a plasmid containing a full-length copy of 16S rRNA gene belonging either to the Escherichia coli or FG-07 strain of Halobacterium salinarum (courtesy of G. Jurgens, University of Helsinki). All qPCR reactions were run in triplicate. The reaction was carried out in iCycler (BIO RAD, Hercules, USA) using the following: 94°C for 15 min, followed by 40 cycles of 94°C for 30 s, 50°C for 30 s and 72°C for 30 s. Melting curve analyses were done to verify that the amplified products were of the expected size. Fungal and bacterial gene copy numbers were Table 2. Agrochemical properties of soil samples. CFS OFS WOOD Ca (mg/kg) 1247,33±129,90 1087,00±86,31 242,33±23,13 P (mg/kg) 14,70±1,79 10,50±1,22 3,33±0,26 K (mg/kg) 135,73±17,25 69,76±8,92 72,70±9,67 Mg (mg/kg) 146,67±28,18 117,33±11,21 45,13±5,07 pH 6,67±0,09 5,77±0,07 4,43±0,03 Conductivity 0,83±0,07 0,63±0,15 0,37±0,03 Total N 0,27±0,02 0,34±0,03 0,21±0,01 Total C 4,80±0,05 5,49±0,09 5,63±0,12 CFS–conventional farming system, OFS–organic farming system, WOOD–woodland. doi:10.1371/journal.pone.0145072.t002 Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 4/16 estimated using a regression equation for each assay relating the cycle threshold (Ct) value to the known number of copies in the standards. Statistical analysis of the qPCR data was carried out using one-way ANOVA in STATISTICA10 Enterprise (www.statsoft.com). Statistical significance was tested by Fischer’s least significant difference (LSD) and Bonferroni adjusted p-values. Bar-coded pyrosequencing of bacterial and archaeal communities The purified DNA templates were amplified with universal multiplex primers F515 5’- GTGCCAGCMGCCGCGGTAA-3’and R806 5’-GGACTACVSGGGTATCTAAT-3’[29] targeting the variable region V4 of bacterial and archaeal 16S rRNA genes. Each multiplex primer contained the adapter, 4-bp key (TCAG), 10-bp barcode and primer sequences. The expected length of the amplification product was 400 bp. Purification, pooling and pyrosequencing of the amplicons were performed with reagents according to manufacturer’s instructions (Roche, Branford, USA). Pyrosequencing was carried out using GS Junior system (Roche). Bioinformatics of the pyrosequencing-derived dataset The raw sequences were processed using QIIME ver. 1.8.0 [30]. To reduce sequencing errors, the multiplexed reads were first filtered for quality and grouped according to barcode sequences. Sequences were omitted from the analysis if they were less than 200 bp, had a quality score less than 25, contained uncorrectable barcodes, primers, ambiguous characters or a homopolymer length equal or greater than 8 bp. All non-bacterial ribosomal sequences and chimeras were also removed from the database. In total, 17 311 sequences were obtained with an average of 1923 sequences per library. The dataset was subjected to the normalization procedure resulting in 1100 sequences per sample. The minimum, median and maximum lengths of sequences were 200, 355 and 313 bp, respectively. Similar sequences were clustered into operational taxonomic units (OTUs) with a minimum identity of 97% using de novo and closed reference algorithms. A representative set of sequences was chosen by selecting the most abundant sequence from each OTU. Representative sequences from each OTU were subjected to RDP naïve Bayesian rRNA Classifier [31] with a confidence level of 80% and aligned using PyNast [32] and Greengenes database [33]. Aligned sequences were used to build a distance matrix with a distance threshold of 0.1 and phylogenetic tree necessary for downstream analysis. Sequence data were archived in SRA database with accession SUB473223. To compare microbial communities the alpha and beta diversity analyses were performed. To estimate alpha diversity, the indices for richness (observed species, ChaoI) and evenness (PD_whole tree, Shannon evenness, Simpson index) were calculated. The t-test was performed to verify the observed differences. For beta diversity the weighted Unifrac metrics [34] was used to calculate the amount of dissimilarity (distance) between the compared bacterial communities. The results were presented in PCoA analysis using QIIME ver. 1.8.0 [30]. All estimates were measured for the normalized data (normalization was carried out up to the smallest number of sequences present in the sample). The multiple matrix regression based on Mantel permutations [35] implemented in the phytools R package (http://www.phytools.org) was conducted to reveal the relationships between community composition and different agrochemical properties of soil. To reduce factor space dimensionality (by removing redundant variables) we performed multiple pairwise tests for Spearman rank-order correlation. Significant dependency observed between pH and P allowed us to remove the latter from our feature set. The abundances of OTUs were compared between samples by calculating the median relative change values for all groups of triplicates. A positive median indicated an increase in Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 5/16 abundance, whereas a negative median was taken as evidence for decline of abundance. A basic permutation test was used to infer significance, whereas a jackknife-like resampling approach was applied to test the stability of median estimates. Results Land use effects on edaphic soil properties The agrochemical properties of cropland soils managed according to the two different farming systems were rather similar, but differed from the woodland soil despite of the similar soil type (Table 2). The woodland soil had the highest content of organic matter and C/N index, whereas the lowest C/N value was observed in CFS. Soil pH was lowest in the woodland. As for the main biogenic elements, woodland soil was rich in sulfur and manganese, while the croplands were higher in magnesium, calcium and phosphorus (Table 2). Relative quantities of bacteria and archaea estimated by qPCR The amounts of the bacterial and archaeal biomass estimated by qPCR were expressed as the copy number of rRNA operons per gram of soil and used for comparing the relative abundances of microorganisms in the soil samples. The copy number of ribosomal operons in the genomes of microorganisms varies and is, in average, 4.09 for bacteria and 1.76 for archaea according to the rrnDB database [36]. The experimental data on the average copy numbers of E.coli and H.salinarum rRNA operons in soil samples were used to calculate the abundance of bacterial and archaeal communities, respectively. The average number of bacteria in soil was 8.37·10 8 for CFS, 1.56·10 9 for OFS and 2.19·10 9 for woodland (Fig 1). Archaea were about three folds of magnitude less abundant and their average numbers were 8.15·10 5 for CFS, 2.41·10 6 for OFS, and 1.37·10 7 for woodland (Fig 1). These results showed that the population densities of bacteria and archaea were lowest in CFS and highest in woodland (p<0.05). This tendency was particularly noticeable for archaea, whose numbers in the woodland were 2 orders of magnitude higher than in the croplands. The number of bacteria in OFS was significantly higher than in CFS (p <0.05), whereas the total counts of archaea did not vary between CFS and OFS (Fig 1). α-biodiversity of the soil microbial communities The biodiversity within each individual sample was estimated using richness (number of observed species, Chao1) and evenness (Shannon evenness, Simpson) indices (Table 3). Woodland samples had the highest percent of coverage (the number of OTUs to chao1 ratio expressed as a percentage) per library (82.7% in average). The coverage values for OFS and CFS samples were 65.1% and 65.8%, respectively. The observed species richness and Simpson index of dominance were not significantly different between the samples (Table 3). Microbial community composition At the phylum level there were 22 major bacterial taxa present in most of the soils Proteobacteria (57.9% in average), Acidobacteria (16,1%), Actinobacteria (7,9%), Verrucomicrobia (2,0%), Bacteroidetes (2,7%) and Firmicutes (4,8%). The phyla with relative abundance less than 1% were considered rare. They included Crenarchaeota,Armatimonadetes, BHI80-139, Chlamydiae,Elusimicrobia,Fibrobacteres, GAL15, Nitrospirae, TM6, TM7 and WPS-2. Some phyla, such as Fibrobacteres, BHI80-139, TM6 and TM7, were found only in croplands. The portion of organisms with unknown taxonomy ranged from 0.6 to 2.1% and was the highest in CFS. Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 6/16 At the phylum level, only minor differences were found between the bacterial communities of CFS and OFS, whereas the differences between woodland soil and croplands were more apparent (Fig 2). Proteobacteria were among the most abundant phyla in croplands, whereas Acidobacteria dominated in the woodland soil (Fig 2). In general, different microbial taxa in woodland soil were more evenly represented, including Firmicutes,Actinobacteria,Nitrospira, Gemmatimonadetes and Chloroflexi. The microbiomes of croplands and woodland soils differed markedly in the composition of Proteobacteria (Fig 2). Woodland soil was dominated by Alphaproteobacteria and Gammaproteobacteria, while in croplands Betaproteobacteria and Gammaproteobacteria were substituting Alphaproteobacteria. Bacteria from the families Pseudomonadaceae and Enterobacteriaceae accounted for more than 80% of the gammaproteobacteria. The family Pseudomonadaceae was almost exclusively represented by the genus Pseudomonas (Fig 3). The abundance of bacteria of this genus varied between the croplands (16.0% in CFS and 13.2% in OFS). In the woodland soil the number of pseudomonads was only 5.3%. On the other hand, the proteobacterial family Sinobacteriaceae counted for more than 7.2% in the woodland soil, as compared with only 1.7% in OFS and CFS. Similarly, betaproteobacteria of the genus Burkholderia and alphaproteobacteria of the family Bradyrhizobiaceae and the genus Rhodoplanes were substantially more common in the woodland soil than in cropland soils (Fig 4). Among the most notable differences in the microbial taxonomic composition between woodland and croplands was the much higher prevalence of the phylum Acidobacteria in woodland, in particular, bacteria of the Fig 1. The number of bacteria and archaea per gram of soil, estimated by quantitative PCR. The raw data on the number of 16S rRNA genes per gram of soil, calibrated to the E.coli and H.salinarum 16S rDNA copy number, were translated to the number of prokaryotic cells per gram of soil by use of the information on the average number of 16S rRNA copies in bacterial and archaeal genomes deposited in rrnDB database [36]. Error bars indicate standard deviation (n = 3). doi:10.1371/journal.pone.0145072.g001 Table 3. Alpha-diversity parameters of soil microbiomes. Sample ID Farming system a CFS OFS WOOD PD_whole_tree 30,25±0,87 32,39±0,87 22,24±0,57 Shannon 6,58±0,19 6,83±0,19 6,72±0,08 Simpson 0,96±0,01 0,97±0,01 0,98±0 Chao1 b 406,53±39,8 450,97±39,8 264,57±9,05 Number of OTUs 267,33±11,67 293,67±11,67 218,67±3,71 Shannon evenness b 0,82±0,02 0,83±0,02 0,86±0,01 a CFS, conventional farming system; OFS, organic farming system; WOOD, woodland. b The alpha-diversity parameters indicated significant differences (p <0,05). doi:10.1371/journal.pone.0145072.t003 Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 7/16 family Koribacteraceae (mainly Candidatus Koribacter, Figs 3and 4), the order Ellin6513 and Solibacterales (Fig 4). The main bacterial genera found in soil microbiomes are shown in Fig 3. The most pronounced differences between the CFS and OFS microbiomes were within the genus Pseudomonas, which were significantly (p<0,05) more abundant in CFS soil, as compared with OFS. Other statistically significant (p<0,05) differences in the taxonomic composition between CFS and OFS microbiomes were minor and found among the bacterial genera with frequencies rarely exceeding 1% of all taxa. Compared with OFS, CFS had higher relative abundances of the actinobacteria belonging to the family Nakamurellaceae, acidobacteria of the family Fig 2. Abundance ratios of the most common bacterial phyla in the soil in organic (OFS) vs. conventional (CFS) farming systems, and the woodland vs. farmland systems (wood vs. FS; FS combines OFS and CFS samples). Circle size indicates the average abundance of the phylum. doi:10.1371/journal.pone.0145072.g002 Fig 3. Heatmap comparison of the microbiomes in croplands (CFS and OFS) and the woodland. Colors mark the average relative abundance (in number of sequences per sample) of each bacterial genus within the sample. Only identified genera with total counts exceeding 5 sequences per library are presented. doi:10.1371/journal.pone.0145072.g003 Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 8/16 Koribacteraceae, proteobacteria of the groups SC-I-84, Ellin6067, Oxalobacteriaceae (particularly the genus Janthinobacterium) and Ralstonia and bacteria, belonging to the genera Paenibacillus and Pedobacter. Microbial community in OFS was enriched with proteobacteria of the Fig 4. OTUs analyzed in a bootstrapped maximum likelihood phylogenetic tree and their abundance presented in a table. Pairwise tests indicated either an increase (+) or a decrease (–) in abundance between samples of the organic farming system (OFS), the conventional farming system (CFS) and the woodland (Wood). Blank cells indicate insufficient data. The significance of difference was assessed using a permutation test, INS indicates insignificant difference. doi:10.1371/journal.pone.0145072.g004 Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 9/16 50. Sudhakaran M, Ramamoorthy D, Rajesh kumar S. Impacts of conventional, sustainable and organic farming system on soil microbial population and soil biochemical properties, Puducherry, India. International Journal of Environmental Sciences. 2013; 4(1): 28–41. 51. Laws MT, Graves WR. Nitrogen inhibits nodulation and reversibly suppresses nitrogen fixation in nodules of Alnus maritime J. Amer. Soc. Hort. Sci. 2005; 130(4): 496–499. 52. Lombard N, Prestat E, van Elsas JD, Simonet P. Soil-specific limitations for access and analysis of soil microbial communities by metagenomics. FEMS Microbiol Ecol. 2011; 78(1): 31–49. doi: 10.1111/j. 1574-6941.2011.01140.x PMID: 21631545 Soil Prokaryotic Communities of Farming Systems PLOS ONE | DOI:10.1371/journal.pone.0145072 December 18, 2015 16 / 16