soil bacterial and fungal communities across a ph

7/28/2019 Soil Bacterial and Fungal Communities Across a pH

1/12

ORIGINAL ARTICLE

Soil bacterial and fungal communities across a pHgradient in an arable soil

Johannes Rousk1,7, Erland Baath1, Philip C Brookes2, Christian L Lauber3, CatherineLozupone4, J Gregory Caporaso4, Rob Knight4,5 and Noah Fierer3,61Department of Microbial Ecology, Lund University, Ecology Building, Lund, Sweden; 2Soil ScienceDepartment, Rothamsted Research, Harpenden, Herts, UK; 3Cooperative Institute for Research inEnvironmental Sciences, University of Colorado, Boulder, CO, USA; 4Department of Chemistry andBiochemistry, University of Colorado, Boulder, CO, USA; 5Howard Hughes Medical Institute, University ofColorado, Boulder, CO, USA and 6Department of Ecology and Evolutionary Biology, University of Colorado,Boulder, CO, USA

Soils collected across a long-term liming experiment (pH 4.08.3), in which variation in factors otherthan pH have been minimized, were used to investigate the direct influence of pH on the abundanceand composition of the two major soil microbial taxa, fungi and bacteria. We hypothesized thatbacterial communities would be more strongly influenced by pH than fungal communities. Todetermine the relative abundance of bacteria and fungi, we used quantitative PCR (qPCR), and toanalyze the composition and diversity of the bacterial and fungal communities, we used a bar-codedpyrosequencing technique. Both the relative abundance and diversity of bacteria were positivelyrelated to pH, the latter nearly doubling between pH 4 and 8. In contrast, the relative abundance offungi was unaffected by pH and fungal diversity was only weakly related with pH. The composition ofthe bacterial communities was closely defined by soil pH; there was as much variability in bacterialcommunity composition across the 180-m distance of this liming experiment as across soilscollected from a wide range of biomes in North and South America, emphasizing the dominance ofpH in structuring bacterial communities. The apparent direct influence of pH on bacterial communitycomposition is probably due to the narrow pH ranges for optimal growth of bacteria. Fungalcommunity composition was less strongly affected by pH, which is consistent with pure culturestudies, demonstrating that fungi generally exhibit wider pH ranges for optimal growth.The ISME Journal advance online publication, 6 May 2010; doi:10.1038/ismej.2010.58Subject Category: microbial ecology and functional diversity of natural habitatsKeywords: bacterial community; fungal community; pyrosequencing; quantitative PCR; RothamstedHoosfield Acid Strip; soil pH

Introduction

Environmental factors controlling the distributionand abundance of plants and animals acrossterrestrial biomes have been studied for centuries,but the environmental factors controlling the dis-tribution and abundance of soil microorganisms are

still poorly understood. Recent study conductedusing a variety of molecular or biochemical ap-proaches has started to explore the distributionalpatterns exhibited by soil microbial communitiesand the biotic or abiotic factors driving these

patterns. However, much of this previous studyhas been conducted using techniques that do notpermit detailed and comprehensive phylogenetic ortaxonomic surveys of microbial communities (forexample, DNA fingerprinting-based approaches,(Fierer and Jackson, 2006), phospholipid fatty acid(PLFA) analyses, (Baath and Anderson, 2003).Similarly, much of this previous study has focusedon the distribution patterns exhibited by singletaxonomic groups, for example, Acidobacteria,(Jones et al., 2009, fungi: Bennett et al. (2009) andBuee et al. (2009), or SR1 bacteria: Davis et al.(2009)), making it difficult to directly compare thepatterns exhibited by common microorganisms.

Recent study has demonstrated that changes insoil microbial communities across space are oftenstrongly correlated with differences in soil chemis-try (Frey et al., 2004; Nilsson et al., 2007; Lauberet al., 2008; Jenkins et al., 2009). In particular, it hasbeen shown that the composition, and in some cases

Received 17 December 2009; revised 26 March 2010; accepted 27March 2010

Correspondence: J Rousk, Department of Microbial Ecology, LundUniversity, Ecology Building, Solvegatan 37, Lund 22362,Sweden.E-mail: [email protected] address: School of the Environment, Natural Resourcesand Geography, Bangor University, Bangor, Gwynedd, UK.

The ISME Journal (2010), 112

& 2010 International Society for Microbial Ecology All rights reserved 1751-7362/10 $32.00

www.nature.com/ismej
http://dx.doi.org/10.1038/ismej.2010.58mailto:[email protected]://www.nature.com/ismejhttp://www.nature.com/ismejmailto:[email protected]://dx.doi.org/10.1038/ismej.2010.58


2/12

diversity, of soil bacterial communities is oftenstrongly correlated with soil pH (Fierer and Jackson,2006; Hartman et al., 2008; Jenkins et al., 2009;Lauber et al., 2009). This pattern holds both foroverall bacterial community composition (Fiererand Jackson, 2006; Baker et al., 2009; Lauber et al.,2009) and for the composition of individual bacter-

ial groups (Nicol et al., 2008; Davis et al., 2009;Jenkins et al., 2009; Jones et al., 2009). It also seemsto hold across a variety of spatial scales, includingcontinental scales (Fierer and Jackson, 2006; Lauberet al., 2009), across land-use types at a givenlocation (Lauber et al., 2008; Jenkins et al., 2009),and across small, submeter scales (Baker et al., 2009;Philippot et al., 2009). However, one limitation ofthese observational studies is that it is impossible todetermine whether the communities are structureddirectly or indirectly by pH. In other words, we donot know whether pH itself is the factor shapingthese communities, or whether pH may be indirectlyrelated to the observed community changes throughmany environmental factors (for example, nutrientavailability, organic C characteristics, soil moistureregime and vegetation type), which often co-varywith changes in soil pH. Similarly, we do notknow whether soil pH is also correlated with thecommunity composition of fungi, another dominantmicrobial group in soil. There is some indicationthat the spatial patterns exhibited by soil fungalcommunities are not correlated, or less stronglycorrelated, with soil pH than has been observed forbacterial communities (for example, Lauber et al.,2008). However, the paucity of detailed studiesdirectly comparing the spatial patterns exhibited

by soil bacteria and fungi make it difficult todraw any robust conclusions regarding the simila-rities or differences in the environmental factorsthat shape the composition of these communitiesin soil.

The Hoosfield acid strip, at Rothamsted Research,Harpenden, UK, is a long-term field experimentideal for examining the effects of pH on soilmicrobial communities (see Materials and methods).There are minimal differences in microbial CO2production rates across this pH gradient, with basalrespiration decreasing by only 30% from pH 8.3 to4.5. Fungal and bacterial biomass (estimated usingPLFA and ergosterol; Rousk et al., 2009) and total

microbial biomass C and N (measured using avariety of methods; Aciego Pietri and Brookes,2007b, 2009; Rousk et al ., 2009, 2010a) alsoremained very stable across the gradient. However,bacterial and fungal growth, estimated by leucineincorporation and acetate incorporation into ergo-sterol, respectively, revealed dramatic differences inthe activity of these microbial decomposer commu-nities (Rousk et al., 2009). Bacterial growth washighest at the highest pH values of the gradient anddeclined by a factor of 5 toward the lower pH values.In contrast, fungal growth was maximal at pH 4.5,and decreased by a factor of more than 5 toward the

high pH end. Across the same gradient, PLFA-basedanalyses suggested that the composition of themicrobial community was highly influenced by thechanging soil pH across this gradient (Aciego-Pietriand Brookes, 2009; Rousk et al., 2010a). However,the limited taxonomic resolution of this techniquedid not allow us to identify the specific microbial

groups that shift in abundance across the pHgradient nor could we specifically identify changesin bacterial versus fungal community compositionand diversity. The application of bar-coded pyrose-quencing (Fierer et al., 2008; Hamady et al., 2008)allows for more detailed phylogenetic and taxo-nomic surveys of microbial communities thanprovided with other techniques, including PLFAand DNA fingerprinting, while also making itpossible to survey a relatively large number ofindividual samples simultaneously.

Our objectives for this study were (i) to determinethe abundance, taxonomic diversity and composi-tion of the bacterial community across the 180-mdistance of the Hoosfield acid strip (in which soilcharacteristics other than pH have largely been heldconstant), (ii) to directly compare variability inbacterial communities across this 180-m gradientwith the variability across soils collected from awide range of biomes in North and South Americacovering a similar pH range, and (iii) to test whetherbacterial communities were more affected by soilpH than the fungal communities, as suggested by(Fierer et al., 2009; Lauber et al., 2009). To do this,we collected soils from across the Hoosfield acidstrip to investigate the direct influence of soil pH onthe abundance, taxonomic diversity and composi-

tion of the two major soil microbial taxa, fungi andbacteria. We used quantitative PCR (qPCR) todetermine the relative abundance of bacteria andfungi across this gradient and a bar-coded pyrose-quencing technique to analyze the composition anddiversity of the bacterial and fungal communities.Finally, we directly compared the communitypatterns across the two very distinct sets ofsamples by combining the obtained data set for thebacterial community composition with that of across-continental study (Lauber et al., 2009).

Materials and methodsSample collection, DNA-extraction and soilcharacterizationThe Hoosfield acid strip is a pH gradient rangingfrom 4.0 to 8.3 within 200 m that resulted from aone-time uneven application of chalk in the mid19th century. Since the chalk application, nofertilizers have been added to the soil, and winterwheat has been continuously grown. Other environ-mental factors vary minimally along the gradient.For instance, the organic carbon (C) content of thesoil is nearly constant at 0.90.01% and the C:Nratio is stable at about 9 between pH 4.5 and 8.3

Soil pH influence on bacterial and fungal communitiesJ Rousk et al

2

The ISME Journal


3/12

(Rousk et al., 2009). The soil from the HoosfieldAcid strip at Rothamsted Research, UK is classifiedas Typic Paleudalf (USDA, United States Depart-ment of Agriculture Soil Survey Staff, 1992). Suchsoils are moderately well drained and developed ina relatively silty (loess containing) superficialdeposit overlaying, and mixed with clay-with-flints.

The plough layer is a flinty, silty clay loam,with 1827% clay. For a full description, seeAciego-Pietri and Brookes, (2007a, b).

In April 2008, we sampled along the first 180 m ofthe strip taking 5 cm diameter, 023 cm depth coresat each sampling position along the gradient. Thegradient was sampled every 10 m between 040 m,then every 5 m between 40 and 120 m and, thenevery 10m between the final 120180 m of thegradient (Rousk et al., 2009). The greater samplingintensity between 40120 m was based on the fasterpH changes observed previously within that interval(Aciego Pietri and Brookes, 2007a). Twenty-sevensoil samples were sieved (o2.8 mm) in the labora-tory, removing apparent roots and stones, and watercontents determined (105 1C; 24 h). The soil sampleswere stored frozen until molecular analyses wereperformed in January 2009. The DNA was extractedfrom 0.5 g subsamples using the MoBio PowerSoilDNA extraction kit (Carlsbad, CA, USA) followingthe manufacturers instructions.

Quantitative PCR analysesRelative abundances of bacterial and fungal small-subunit rRNA gene copies were quantified using themethod described by Fierer et al. (2005). For fungi,

the forward primer used was 5.8 s combined withthe reverse primer, ITS1f, while the forward primerEub338 and reverse primer Eub518 was used forbacteria (Fierer et al., 2005). To estimate bacterialand fungal small-subunit rRNA gene abundances,standard curves were generated using a 10-foldserial dilution of a plasmid containing a full-lengthcopy of either the Escherichia coli 16S rRNA geneor the Saccharomyces cerevisiae 18S rRNA gene.The 25 mg qPCR reactions contained 12.5ml ABgeneSYBR MasterMix (ABgene, Rochester, NY, USA),0.5ml of each 10-mM forward and reverse primers,and 9.5ml sterile, DNA-free water. Standard andenvironmental DNA samples were added at 2.0 ml

per reaction. The reaction was carried out on anEppendorf Master-cycler ep Realplex thermocycler(Eppendorf, Westbury, NY, USA) using a programof 94 1C for 15 min followed by 40 cycles of 94 1C for30s, 50 1C for 30s and 72 1C for 30 s. Melting curveand gel electrophoresis analyses were performed toconfirm that the amplified products were of theappropriate size. Fungal and bacterial gene copynumbers were generated using a regression equationfor each assay relating the cycle threshold (Ct) valueto the known number of copies in the standards.All qPCR reactions were run in quadruplicate withthe DNA extracted from each soil sample.

Bar-coded pyrosequencing of fungal and bacterialcommunitiesAmplification, purification, pooling and pyrose-quencing of a region of the bacterial 16S rRNA genewere performed as described by Fierer et al. (2008).The fungal community was analyzed using anidentical approach, except that we substituted a

primer set adapted for pyrosequencing as describedby Borneman and Hartin (2000). The forward primerconsists of the 454 adapter B, a 2-bp linker sequence(AG) and the 817f 18S rRNA-specific primer (50-TTAGCATGGAATAATRRAATAGGA-30). The reverseprimer consists of the 454 adapter A, a 12-bpbarcode, a 2-bp linker sequence (AC) and the 1196rprimer (50-TCTGGACCTGGTGAGTTTCC-3 0). Thisprimer set was selected because it has been shownto be fungal-specific and targets a region of the 18SrRNA gene, which is variable between major taxaand can be aligned (permitting phylogenetic ana-lyses such as UniFrac). However, it shouldbe noted that this gene region is not sufficientlyvariable to permit detailed taxonomic identificationof fungal communities and thus community ana-lyses are restricted to the family level or above.

Processing of pyrosequencing dataCommunity level diversity across the pH gradientwas analyzed by comparing the number of opera-tional taxonomic units (OTUs) with an OTU definedat the 97% similarity level for both bacteria andfungi (Kunin et al., 2010). To correct for survey effort(number of sequences analyzed per sample), weused a randomly selected subset of sequences per

soil sample to compare relative differences in OTU-level diversity across the gradient. Of the 27 soilsamples that were collected, we only included thosesamples for which we obtained at least 600 bacterialsequences per sample or at least 85 fungal sequencesper sample for subsequent community analyses(22 and 16 samples from the bacterial and fungalanalyses, respectively; there were no apparentbiases with distribution of the omitted samplesand high and low pH). These levels of per-samplesequencing effort are not sufficient to characterizethe full extent of bacterial and fungal diversity inthe collected soils. However, previous study sug-gests that we can quantitatively compare overall

community composition and relative differences indiversity with these levels of sequencing effort(Shaw et al., 2008).

Data were processed as described by Fierer et al.(2008) using the Quantitative Insights Into MicrobialEcology (QIIME) toolkit (Caporaso et al., 2010). Inbrief, bacterial and fungal sequences were qualitytrimmed, and assigned to soil samples based ontheir barcodes. Bacterial sequences were binned intoOTUs using a 97% identity threshold with cd-hit(Li and Godzik, 2006), and the most abundantsequence from each OTU was selected as a repre-sentative sequence for that OTU. Taxonomy was


3

The ISME Journal


4/12

assigned to bacterial OTUs by using the Basic LocalAlignment Search Tool (BLAST) for each represen-tative sequence against a subset of the Silva database(the full database filtered at 97% sequence identityusing cd-hit). Phylogenetic trees were then built fromall representative sequences using the FastTree algo-rithm (Price et al., 2009). The differences in overall

community composition between each pair of sam-ples was determined using the unweighted UniFracmetric (Lozupone and Knight, 2005; Lozupone et al.,2006), which calculates the distance between any pairof communities based on the fraction of unique(nonoverlapping) branch length of their sequencesin the tree (Lozupone and Knight, 2005).

Taxonomic classification of the fungal pyrosequen-cing reads and annotation of a reference phylogenetictree for the UniFrac analysis was carried out byBLASTing to the fungal component of the Silvadatabase (Pruesse et al., 2007). Specifically, the fungalpyrosequencing reads were BLASTed against adatabase of the 18S rRNA sequences from Silva, andthe sequences were assigned to their closest BLASThit. A total of 118 of the sequences had no hit below aPo1030 threshold and were excluded from theUniFrac and taxonomic analysis. These likely repre-sent fungal sequences that are highly divergent fromthose for which near full-length sequences have beenpreviously recovered. The remaining sequencesmapped to a total of 725 reference sequences in theSilva database. The % identity similarity to theassigned sequence ranged from 84.4100% with anaverage value of 98.3%. UniFrac analysis wasperformed using the Fast UniFrac Web interface, theSilva reference tree and an environment file mapping

pyrosequencing reads to Silva reference sequences foreach sample as described previously (Hamady et al.,2010). All UniFrac analyses were performed on a setsurveying effort, at 600 randomly selected sequencesper sample for bacteria and 85 for fungi, thusincluding 22 samples for the bacterial analysis and16 for the fungal analysis.

To compare the variability in bacterial communitycomposition across this pH gradient to variabilityacross soils collected from a wide range of biomesfrom North and South America, we combined thedata set from this study and that previously reportedby Lauber et al. (2009), calculating pairwise un-weighted UniFrac distances between each of the 110

soils. The UniFrac analyses were performed at a setsurveying effort of 600 randomly selected sequencesper sample. Although identical molecular methodswere used to analyze both sets of samples, the soilswere not collected in an identical manner. Forexample, the soils from this study were collectedfrom the top 23 cm of mineral soil, whereas thosefrom Lauber et al. (2009), see also (Fierer andJackson, 2006), were collected from the top 5 cm ofmineral soil. Nevertheless, we conducted this directcomparison to qualitatively compare patterns inbacterial community composition across a range ofspatial scales.

Statistical analysesRarefaction curves generated using Primer v6 wereused to compare relative levels of fungal andbacterial OTU diversity across the gradient. Therelationships between the copy number of bacteriaand fungi with soil pH, and between the taxonomicdiversity for the two groups with pH, were tested

with linear regression analyses. Pairwise UniFracdistances calculated for the bacterial and fungalcommunities were visualized using nonmetric mul-tidimensional scaling plots (Primer v6). Principalcoordinate analyses (Multi-Variate Statistical Pack-age for Windows) were used to determine therelationship between pH and UniFrac distances byregressing scores for the first principal coordinateagainst pH. Inspection of the residuals of a linearcurve fitted to the bacterial PCo1 and soil pHrevealed this not to be randomly distributed roundthe line, whereas a second degree polynomialfunction gave a better fit. In addition, we usedMantel tests to test for statistical significancebetween the UniFrac composition of the fungaland bacterial communities and soil pH (Primer v6).Linear regressions were used to test the relation-ships between the relative abundances of bacterialand fungal taxa with pH. To avoid the potentialsubstrate effects below pH 4.5, due to the confound-ing factor of impeded plant growth below this pH(see Discussion section), regressions were per-formed for data points between pH 4.5 and 8.3 only.

Results

qPCR estimates of bacterial and fungal abundance

The abundance of bacteria, as determined usingqPCR, increased nearly fourfold across the pHgradient from low to high pH (Figure 1a, P0.002; r20.33). The abundance of fungi did not showa consistent relationship with pH across the gradient(Figure 1b). This resulted in no significant relation-ship between pH and fungal-to-bacterial small-subunit rRNA gene copy ratios across the gradient(P0.40, r20.03).

Taxonomic diversityFor the bacterial community analyses in the 22 soilsamples used from across the gradient, we obtained

a total of 32 900 (89% of the total 37000) thatcould be classified for a mean of 1500 classifiablesequences per sample used in the subsequentanalyses (range 6002000). The number of OTUsper bacterial community determined at the samelevel of surveying effort (600 sequences per sample)was positively related to soil pH (Po0.0001;r20.75), effectively doubling between pH 4.0 andpH 8.3 (Figure 2a).

For the fungal community analyses, we obtained atotal of 4700 classifiable sequences among the 16samples used for an average of 290 fungal sequencesper soil sample. The number of fungal OTUs per


4

The ISME Journal


5/12

sample, determined at the same level of surveyingeffort (85 sequences per sample), showed a weak

and nonsignificant relationship to soil pH (P

0.08,r20.21), with OTU numbers increasing marginallyas soil pH increased (Figure 2b).

Bacterial community compositionThere was a strong influence of soil pH on thecomposition of the bacterial communities acrossthe gradient. This is evident from the nonmetricmultidimensional scaling plot ordination of thepairwise UniFrac distances that illustrate howbacterial communities of different pH align alongthe first dimension of the plot (Figure 3a). Manteltests corroborated this finding, indicating a strong

relationship between the bacterial community com-position and soil pH (Spearman r0.96; Po0.001),as did a regression between the scores on thefirst principal coordinate axis (PCo1) and soil pH(Figure 3c; Po0.0001; r20.994). Including organicC, total N or the C:N ratio of the soil samples did notsignificantly improve the model over soil pH alone.Although changes in the relative sequence abun-dance of acidobacterial groups were pronounced(see below), excluding all acidobacteria from theUniFrac analysis of the bacterial community did notdiminish the correlation with soil pH (Mantel test:Spearman r0.94, Po0.001). The second-degree

polynomial function used to describe the relation-ship between PCo1 and pH indicated that thedissimilarity in the bacterial community composi-tion (PCo1 difference) was smaller at the higher pHend than at the lower end, and that incrementaldifferences in bacterial community compositionwith pH were insignificant above pH 6.8. Thiswas also corroborated by a direct comparison ofthe UniFrac dissimilarity values. The dissimilaritybetween the bacterial community in soils at pH6.820.16 (n3; averages.e.) was on average notappreciably different from the bacterial communityin soils of pH 8.020.06 (n6); The UniFracdissimilarity within pH 8.02 was 0.720.005,

whereas UniFrac dissimilarity between pH 6.82and 8.02 was 0.730.02.

The phylogenetic distances between soil bacterialcommunities across the pH gradient (Figure 3) wereclearly related to shifts in the relative abundances ofbacterial taxa at the phylum and subphylum levels(Figure 4). The relative abundances ofAcidobacteriasubgroups 1, 2 and 3 decreased with increasing soilpH. The relative abundance of Acidobacteria sub-groups 5, 6 and 7 increased with soil pH (Figures4ah). Except for high values at the extreme low endof the pH scale (opH 4.1), the relative abundance ofActinobacteria was largely unaffected by soil pH

Figure 2 The number of operational taxonomic units (OTUs) ofbacteria (a) and fungi (b) across the Hoosfield acid strip. Thebacterial community was sampled at the 600 sequences level, andthe fungal community was sampled at the 85 sequences level.Linear functions were used to describe the OTU richness to pHrelationships.

Figure 1 The abundance of bacteria (a) and fungi (b) , asindicated by the number of 16S or 18S ribosomal DNA (rDNA)copies measured using quantitative PCR (qPCR). Linear functionswere used to describe the relationship between bacterial andfungal abundance and pH.


5

The ISME Journal


6/12

(Figure 4i). Similarly, the relative abundance ofBacteriodetes (Figure 4j) did not show clear relation-ships with soil pH. Nitrospira was relatively moreabundant at higher soil pH but the trend was notsignificant (Figure 4k). Above pH 4.5, the a-, b-,g- and d-Proteobacteria all increased in relative

abundance with increasing soil pH, but these trendswere only significant for the g-Proteobacteria.The single largest bacterial group in the low pHsoils was Acidobacteria subgroup 1 (at almost 40%of the sequences), whereas a-Proteobacteria weremost abundant in the high pH soils in which thisgroup represented about 40% of all sequences.

Fungal community compositionAs with the bacterial communities, the compositionof the fungal communities was also related to soilpH, although the relationship was far weaker(Figure 3b). Mantel tests corroborated this pattern,

indicating a significant, albeit weaker relationshipbetween fungal community composition and soil pHthan was observed with the bacterial community(Spearman r0.37; Po0.001), a pattern corrobo-rated by a regression between the scores on PCo1and soil pH (P0.001; r20.55, Figure 3d). Includ-ing organic C, total N or the C:N ratio of the soil samples did not significantly improve themodel over soil pH alone.

The phylogenetic distances between soil fungalcommunities across the pH gradient (Figure 3) werenot clearly related to shifts in the relativeabundances of major fungal groups (ascomycete,

basidiomycete or chytridiomycete) as these fungalgroups made up approximately 45, 30 and 25% ofthe sequences in each sample, respectively, regard-less of soil pH. However, if we examine the fungalcommunities in more detailed levels of taxonomicresolution, some patterns with pH were evident

(Figure 5). For instance, the relative abundanceof a sequence with an average of 99.6% sequencesimilarity to the ascomycete group Hypocreales(Silva accession number AY526489) increased frombeing absent below pH 5.5 to making up nearly 5%of the sequences at the high pH end (Figure 5b).Conversely, a sequence with an average of 99.8%sequence similarity to the ascomycete groupHelotiales (Silva accession number DQ471005) andone with 99.0% sequence similarity to the mito-sporic basidiomycete group (Silva accession numberAJ319755) both decreased from making up nearly6 and 30% of the sequences at the low pH end,respectively, to very low relative abundances at the

high pH end (Figures 5a and c).

Comparison with a cross-continental pH gradientTo compare the variability in bacterial communitycomposition across this pH gradient with thevariability observed in communities collected fromacross North and South America, representing awide range of biomes and soils with pH valuesranging from below 3.5 to nearly 9.0, we directlycompared the bacterial communities in the soilsstudied here with those described by Lauber et al.(2009). The variation in bacterial community

Figure 3 Nonmetric multidimensional scaling (NMDS) plots derived from pairwise unweighted UniFrac distances between soilsamples with symbols coded by pH category (a and b), and the first component from a principal coordinate analysis (which explained29.1% of the bacterial community composition variation and 17.3% of the fungal community composition variation, respectively) of thepairwise unweighted UniFrac distances regressed against soil pH using a second-degree polynomial function for the bacterial community ( c)and a linear function for the fungal community (d).


6

The ISME Journal


7/12

Figure 4 The relationships between relative abundances of 15 dominant bacterial groups and soil pH. Linear regressions were used totest the relationship between the taxas relative abundances and pH. To avoid the possible confounding substrate influence below pH 4.5,due to impeded plant growth below this pH (see Discussion section), only data points between pH 4.5 and 8.3 (filled circles) were usedfor the analyses, and data points below pH 4.5 (open circles) were excluded. Panels ( a)(o) depict the relationships between specificbacterial taxa as defined in each panel.


7

The ISME Journal


8/12

composition across the single soil type examinedhere was shown to be of nearly identical magnitudeto the cross-biome variation in bacterial commu-nities reported by Lauber et al. (2009) (Figure 6). Inother words, there was as much variability in thesoil bacterial communities across the Hoosfield acidstrip (that is, across a distance of only 180m) asacross a wide range of biomes separated by many

thousands of kilometers with a qualitatively similarinfluence of pH on bacterial community composi-tion evident across both sets of samples.

Discussion

Soil pH has a strong influence on the diversity andcomposition of soil bacterial communities across theentire gradient. This corroborates the positiverelationship between bacterial diversity and highersoil pH (in the interval pH 47) observed in, forexample, Lauber et al. (2009), confirming that this

pattern is robust across different spatial scales andsoil types. This is not the first study to show that soilpH has a strong influence on bacterial communitycomposition. Similar results have been reportedusing other techniques with more limited resolutionthan pyrosequencing. For instance, the microbialPLFA composition, which primarily reflects thestructure of the bacterial community (as few markersare associated with fungi), has been shown to behighly influenced by soil pH (Nilsson et al., 2007;Baath and Anderson, 2003). Recently, this has alsobeen reported in the pH gradient studied here(Aciego Pietri and Brookes, 2009; Rousk et al.,2010a). In addition, characterizations of soil bacter-

ial communities across a wide range of sample setshave consistently indicated the same powerful corre-lation between soil pH and bacterial communitycomposition, regardless of the technique used, includ-ing DNA fingerprinting-based techniques (Fierer andJackson, 2006; Mannisto et al., 2007; Jenkins et al.,2009), clone library-based techniques (Lauber et al.,2008; Jesus et al., 2009) and pyrosequencing-basedtechniques (Lauber et al., 2009).

The results from previous studies are thuscorroborated by our results, and the strong connec-tion between pH and the bacterial communitycomposition is consequently consistent both across

Figure 5 The relationships between relative abundances of threeabundant fungal sequences and soil pH. The fungal sequencesinclude one with an average sequence similarity of 99.8% toaccession number DQ471005 of the Silva database, a Helotiales (a),one with an average similarity of 99.6% to accession numberAY526489, a Hypocreales (b), and one with an average similarity of99.0% to accession number AJ319755, a mitosporic basidiomycete (c).Linear regressions were used to test the relationship betweenrelative abundances and pH. To avoid the possible confoundingsubstrate influence below pH 4.5, due to impeded plant growthbelow this pH (see Discussion section), only data points betweenpH 4.5 and 8.3 (filled circles) were used for the analyses, and datapoints below pH 4.5 (open circles) were excluded.

Figure 6 Nonmetric multidimensional scaling (NMDS) plotsderived from pairwise unweighted UniFrac distances betweensoil samples of the Hoosfield soils (filled circles) and the 88-soilcross-biome study by Lauber et al. (2009) (filled diamonds) withsymbols color-coded by pH category.


8

The ISME Journal


9/12

biomes, in which many soil and site characteristicsother than pH also change (the listed studies above),and across an experimental pH gradient establishedon a single soil type in which the other soil charac-teristics (for example, vegetation type, texture,moisture and carbon concentrations) vary minimally(this study). We also directly related the data set

from an inter-biome collection of very different soils(Lauber et al., 2009) to the relationship betweenbacterial community composition and soil pH in thethis study (Figure 6). This comparison emphasizedthe canonical influence of soil pH on the bacterialcommunity: bacterial communities separated byless than 180 m were often as different as bacterialcommunities in soils collected from very differentbiomes across North and South America. Thisimplies that pH is greater driver of bacterialcommunity composition than dispersal limitationand other environmental factors that vary betweenbiomes.

These results raise the question: why is soil pH sooften such a good predictor of bacterial communitycomposition, both across biomes and within anindividual soil type? One hypothesis is that thecommunities are directly influenced by soil pH dueto most bacterial taxa exhibiting relatively narrowgrowth tolerances. This is in close accordance withthe usually narrow pH range permissive for growthof individual bacterial species in pure culture,usually varying between 34 pH units betweenminimum and maximum (Rosso et al., 1995). Inaddition, pH optima for natural bacterial commu-nities of different soils are closely constrained totheir in situ pH (Baath, 1996). Deviations of 1.5 pH

units from in situ pH of bacterial communitiesconsistently reduce their activity by 50% (Fernan-dez-Calvino and Baath, 2010). A growth decrease ofonly 25% compared with optimum growth wouldlead to a population being rapidly outcompeted byother bacteria that were not impeded. These narrowpH optima for bacterial strains would explain thestrong relationship between bacterial communitycomposition and soil pH.

The shifts in the relative abundances of specifictaxonomic groups across this pH gradient are similarto the pH responses observed in other studies. Forinstance, the relative abundance of Acidobacteriahas been shown to increase toward lower pH

(Mannisto et al., 2007; Jones et al., 2009; Dimitriuand Grayston, 2010). In addition, subgroups withinthe Acidobacteria phylum have also been shown tobe strongly, and differentially, affected by pH. Forinstance, Acidobacteria subgroups 1 and 2, the mostabundant groups, have previously been found to benegatively dependent on pH (Jones et al., 2009), aresult that is corroborated in the results of this study.However, Acidobacteria subgroup 3 demonstrated asimilar positive trend with lower pH in this study,which has not been discerned previously. Conver-sely, Acidobacteria subgroup 4, but also subgroups6, 7 and 16, have been found to positively correlate

with higher pH (Jones et al., 2009), which ispartially corroborated by our results. The strongpositive correlation between the relative abundanceof both Actinobacteria and Bacteriodetes withhigher pH previously reported for a wide range ofsoils from different biomes (Lauber et al., 2009)was, however, not seen in the Hoosfield acid strip.

This may indicate that the previously observedconnection between these phyla and soil pH mayin fact have been related to another soil or sitecharacteristic (for example, C availability or soilmoisture) that may have co-varied with soil pH.

The relative abundances of proteobacterial groupstended to be positively related to pH across theHoosfield acid strip, a pattern that was not evidentin the cross-continental study in which manysoil types were included (Lauber et al., 2009).High abundances of proteobacterial groups havepreviously been connected with higher availabilityo f C (McCaig et al., 1999; Axelrood et al., 2002;Kirchman, 2002; Fazi et al., 2005: Fierer et al., 2007).Plant growth (sown wheat) along the Hoosfieldacid strip has been noted to decrease sharplybelow pH 4.5 (JC Aciego-Pietri, unpublished data),resulting in correspondingly reduced plant-C inputavailable to the microorganisms. This decrease inthe availability of C at low pH might have con-tributed to the change in relative abundance ofProteobacteria with increasing pH. Low plantgrowth at low pH may also explain the differenttrend for abundance of some bacterial groupsbelow pH 4.5, including, for example, Actino-bacteria (Figure 4i) and Acidobacteria subgroup 2(Figure 4b).

This study is the one of the few pyrosequencing-based assessments of the fungal community in soils,and it is one of only a few studies to combinesequence-based assessments of both the bacterialand fungal communities together. A consequence ofthis is that the influence of soil pH on the fungalcommunity has been less explored than effects ofsoil pH on the bacterial community. We found thatthe composition of the fungal community wasrelated to soil pH, but the influence was far weakerthan for the bacterial community. This may berelated, in part, to the ability of the applied methodto document the phylogenetic differences betweenthe fungal communities across this gradient. How-

ever, these results do agree with physiologicalstudies of bacterial and fungal pure cultures,indicating that these microbial groups differ intheir responses to pH. Fungal species typically havea wide pH optimum, often covering 59 pH unitswithout significant inhibition of their growth(Wheeler et al., 1991; Nevarez et al., 2009). Theavailable evidence for the fungal pH relationshipthus indicates a significantly weaker direct connec-tion with pH than bacteria (Beales, 2004), althoughpure culture studies have shown preference forcertain pH values for different taxa of soil fungi(Domsch et al., 1980). For instance, the fungal genus


9

The ISME Journal


10/12

Mariannaea, with 100% sequence similarity to anabundant sequence of the ascomycete Hypocrealesgroup (Figure 5b), has been determined to growbetween pH 5.5 and 8.5 in culture-based studies(Domsch et al., 1980), which matches the resultsobtained in the present study. Alternatively, it ispossible that competitive interactions between fungi

and bacteria could, in part, be driving the observedfungal community shifts across a steep gradient ofbacterial community change. For instance, removalexperiments revealed intensive fungalbacterialinteraction in soil in which inhibition of thebacterial community immediately stimulated fungi(Rousk et al., 2008). Furthermore, when similarexperiments were recently used to inhibit bacterialgrowth along the pH gradient in Hoosfield, therewere indications of high fungal growth at all pHs(Rousk et al., 2010b). This could suggest that theobserved pattern between the fungal communitycomposition and soil pH is an indirect effect,mediated by the competitive influence from thehighly dynamic bacterial community along the pHgradient.

The soil pH effect on the abundance (qPCR copynumber) of both bacteria and fungi was relativelymodest. There was a small positive relationshipbetween the abundance of bacteria and higher pH,increasing nearly twofold, whereas fungal abun-dance was unaffected. It should be noted that qPCRdoes not provide an estimate of biomass ratiosbecause ribosomal gene copy number and cellularnucleic acid concentrations can vary between taxa.However, it has been shown to provide a reprodu-cible metric to track shifts in the relative abun-

dances of bacteria and fungi across a landscape(Fierer et al., 2005). Corroborating this finding, ourresults indicate that the qPCR results are largely inline with alternative biomass measurements pre-viously made on the same gradient (for example,Rousk et al., 2009, 2010a). Thus qPCR, supported bythe previously applied biomass methods, mayprovide reasonable indices of standing biomass poolsizes, but such methods do not resolve the dynamicsin the active fungal and bacterial communities thatwere demonstrated previously using growth-basedmeasurements (Rousk et al., 2009).

In conclusion, the composition of the bacterialcommunity in the Hoosfield acid strip was sharply

defined by soil pH. This corroborates the resultsfrom the inter-biome pH gradient studied by Lauberet al. (2009), and many others, which included awide range of soil types, and suggests that pH ismore important overall for structuring soil commu-nities than is biome type. When the results from thestudies were combined, there was as much varia-bility in bacterial community composition acrossthe 180 m distance of this liming experiment asacross soils collected from a wide range of biomes inNorth and South America. The apparent strongerinfluence by pH on the bacterial communitycomposition is probably due to the narrow pH

ranges for optimal growth of bacteria, whereas theweaker influence by pH on the fungal communitycomposition is consistent with pure culture studies,demonstrating that fungi generally exhibit wider pHranges for optimal growth.

AcknowledgementsWe thank D Ahren for helpful discussions and M Hamadyfor assistance with bioinformatics. This study wassupported by grants from the Swedish Research Council(VR) and from the Royal Physiographic Society of Lund(Kgl Fysiografen) (to JR). This study was also supported bygrants from the US National Science Foundation, the USDepartment of Agriculture and the AW Mellon Founda-tion to NF and by a grant from the Swedish ResearchCouncil (VR) to EB

References

Aciego Pietri JC, Brookes PC. (2007a). Nitrogen minerali-sation along a pH gradient of a silty loam UK soil.Soil Biol Biochem 40: 797802.

Aciego Pietri JC, Brookes PC. (2007b). Relationshipsbetween soil pH and microbial properties in a UKarable soil. Soil Biol Biochem 40: 18561861.

Aciego Pietri JC, Brookes PC. (2009). Substrate inputs andpH as factors controlling microbial biomass, activityand community structure in an arable soil. Soil BiolBiochem 41: 13961405.

Axelrood PE, Chow ML, Radomski CC, McDermott JM,Davies J. (2002). Molecular characterization of bacter-ial diversity from British Columbia forest soilssubjected to disturbance. Can J Microbiol 48: 655674.

Baath E. (1996). Adaptation of soil bacterial communitiesto prevailing pH in different soils. FEMS Microb Ecol19: 227237.

Baath E, Anderson TH. (2003). Comparison of soil fungal/bacterial ratios in a pH gradient using physiologicaland PLFA-based techniques. Soil Biol Biochem 35:955963.

Baker KL, Langenheder S, Nicol GW, Ricketts D, Killham K,Campbell CD et al. (2009). Environmental and spatialcharacterisation of bacterial community composition insoil to inform sampling strategies. Soil Biol Biochem 41:22922298.

Beales N. (2004). Adaptation of microorganisms to coldtemperature, weak acid preservatives, low pH, andosmotic stress: a review. Comp Rev Food Sci F 3:

120.Bennett LT, Kasel S, Tibbits J. (2009). Woodland treesmodulate soil resources and conserve fungal diversityin fragmented landscapes. Soil Biol Biochem 41:21622169.

Borneman J, Hartin REJ. (2000). PCR primers that amplifyfungal rRNA genes from environmental samples.Appl Environ Microbiol 66: 43564360.

Buee M, Reich M, Murat C, Morin E, Nilsson RH, Uroz Set al. (2009). 454 pyrosequencing analyses of forestsoils reveal an unexpectedly high fungal diversity.New Phytol 184: 449456.

Caporaso JS, Kuczynski J, Stombaugh J, Bittinger K,Bushman FD, Costello EK et al. (2010). QIIME allowsintegration and analysis of high-throughput


10

The ISME Journal


11/12

community sequencing data. Nat Methods (in press),doi:10.1038/nmeth.f.303.

Davis JP, Youssef NH, Elshahed MS. (2009). Assessment ofthe diversity, abundance, and ecological distributionof candidate division SR1 reveals a high level ofphylogenetic diversity but limited morphotypic diver-sity. Appl Environ Microbiol 75: 41394148.

Dimitriu PA, Grayston SJ. (2010). Relationship between

soil properties and patterns of bacterial b-diversityacross reclaimed and natural boreal forest soils.Microb Ecol 59: 563573.

Domsch KH, Gams W, Anderson TH. (1980). Compendiumof Soil Fungi Volume 1 Academic Press: London.

Fazi S, Amalfitano S, Pernthaler J, Puddu A. (2005).Bacterial communities associated with benthic organ-ic matter in headwater stream microhabitats. EnvironMicrobiol 7: 16331640.

Fernandez-Calvino D, Baath E. (2010). Growth responseof the bacterial community to pH in soils differingin pH. FEMS Microbiol Ecol (in press), doi: 10.1111/j.1574-6941.2010.00873.x.

Fierer N, Bradford MA, Jackson RB. (2007). Toward anecological classification of soil bacteria. Ecology 88:

13541364.Fierer N, Hamady M, Lauber CL, Jackson RB. (2008). The

influence of sex, handedness, and washing on thediversity of hand surface bacteria. Proc Natl Acad SciUSA 105: 1799417999.

Fierer N, Jackson RB. (2006). The diversity and biogeo-graphy of soil bacterial communities. Proc Natl AcadSci USA 103: 626631.

Fierer N, Jackson JA, Vilgalys R, Jackson RB. (2005).Assessment of soil microbial community structure byuse of taxon-specific quantitative PCR assays.Appl Environ Microbiol 71: 41174120.

Fierer N, Strickland MS, Lipzin D, Bradford MA, Cleve-land CC. (2009). Global patterns in belowgroundcommunities. Ecol Lett 12: 12381249.

Frey SD, Knorr M, Parrent JL, Simpson RT. (2004). Chronicnitrogen enrichment affects the structure andfunction of the soil microbial community in temperatehardwood and pine forests. For Ecol Manag 196:159171.

Hamady M, Lozupone C, Knight R. (2010). Fast UniFrac:facilitating high-throughput phylogenetic analyses ofmicrobial communities including analysis of pyrose-quencing and PhyloChip data. ISME J 4: 1727.

Hamady M, Walker JJ, Harris JK, Gold NJ, Knight R. (2008).Error-correcting barcoded primers for pyrosequencinghundreds of samples in multiplex. Nat Methods 5:235237.

Hartman WH, Richardson CJ, Vilgalys R, Bruland GL.(2008). Environmental and anthropogenic control of

bacterial communities in wetland soils. Proc NatlAcad Sci USA 105: 1784217847.

Jenkins SN, Waite IS, Blackburn A, Husband R, RushtonSP, Manning DC et al. (2009). Actinobacterial commu-nity dynamics in long term managed grasslands. 95:319334.

Jesus ED, Marsh TL, Tiedje JM, Moreira FMD. (2009).Changes in lands use alter the structure of bacterialcommunities in Western Amazon soils. ISME J 3:10041011.

Jones RT, Robeson MS, Lauber CL, Hamady M, Knight R,Fierer N. (2009). A comprehensive survey of soilacidobacterial diversity using pyrosequencing andclone library analyses. ISME J 3: 442453.

Kirchman DL. (2002). The ecology of CytophagaFlavo-bacteria in aquatic environments. FEMS MicrobiolEcol 39: 91100.

Kunin V, Engelbrektson A, Ochman H, Hugenholtz P.(2010). Wrinkles in the rare biosphere: pyrosequen-cing errors can lead to artificial inflation of diversityestimates. Environ Microbiol 12: 118123.

Lauber CL, Hamady M, Knight R, Fierer N. (2009). Pyro-

sequencing-based assessment of soil pH as a predictorof soil bacterial community composition at the conti-nental scale. Appl Environ Microbiol 75: 51115120.

Lauber CL, Strickland MS, Bradford MA, Fierer N. (2008).The influence of soil properties on the structure of

bacterial and fungal communities across land-usetypes. Soil Biol Biochem 40: 24072415.

Li W, Godzik A. (2006). Cd-hit: a fast program forclustering and comparing large sets of protein ornucleotide sequences. Bioinformatics 22: 16581659.

Lozupone C, Hamady M, Knight R. (2006). UniFracanonline tool for comparing microbial communitydiversity in a phylogenetic context. BMC Bioinfor-matics 7: 731.

Lozupone C, Knight R. (2005). UniFrac: a new phyloge-

netic method for comparing microbial communities.Appl Environ Microbiol 71: 82288235.

Mannisto MK, Tiirola M, Haggblom MM. (2007). Bacterialcommunities in Arctic fjelds of Finnish Lapland arestable but highly pH dependent. FEMS Microbiol Ecol59: 452465.

McCaig AE, Glover LA, Prosser JI. (1999). Molecularanalysis of bacterial community structure and diver-sity in unimproved and improved upland grasspastures. Appl Environ Microbiol 65: 17211730.

Nevarez L, Vasseur V, Le Madec L, Le Bras L, Coroller L,Leguerinel I et al. (2009). Physiological traits ofPenicillium glabrum strain LCP 08.5568, a filamentousfungus isolated from bottled aromatised mineralwater. Int J Food Microbiol 130: 166171.

Nicol GW, Leininger S, Schleper C, Prosser JI. (2008). Theinfluence of soil pH on the diversity, abundance andtranscriptional activity of ammonia oxidizing archaeaand bacteria. Environ Microbiol 10: 29662978.

Nilsson LO, Baath E, Falkengren-Grerup U, Wallander H.(2007). Growth of ectomycorrhizal mycelia and com-position of soil microbial communities in oak forestsoils along a nitrogen deposition gradient. Oecologia153: 375384.

Philippot L, Cuhel J, Saby NPA, Cheneby D, ChronakovaA, Bru D et al. (2009). Mapping fine-scale spatialpatterns of size and activity of the denitrifier commu-nity. Environ Microbiol 11: 15181526.

Price MN, Dehal PS, Arkin AP. (2009). Fasttree: computinglarge minimum-evolution trees with profiles insteadof a distance matrix. Mol Biol Evol 26: 16411650.

Pruesse E, Quast C, Knittel K, Fuchs B, Ludwig W, PepliesJ et al. (2007). SILVA: a comprehensive online resourcefor quality checked and aligned ribosomal RNAsequence data compatible with ARB. Nucleic AcidsRes 35: 71887196.

Rosso L, Lobry JR, Bajard S, Flandrois JP. (1995).Convenient model to describe the combined effectsof temperature and pH on microbial growth.Appl Environ Microbiol 61: 610616.

Rousk J, Alden Demoling L, Bahr A, Baath E. (2008).Examining the fungal and bacterial niche overlapusing selective inhibitors in soil. FEMS Microbiol Ecol63: 350358.


11

The ISME Journal


12/12

Rousk J, Brookes PC, Baath E. (2009). Contrasting soil pHeffects on fungal and bacterial growth suggestsfunctional redundancy in carbon mineralisation.Appl Environ Microbiol 75: 15891596.

Rousk J, Brookes PC, Baath E. (2010a). The microbialPLFA composition as affected by pH in an arable soil.Soil Biol Biochem 42: 516520.

Rousk J, Brookes PC, Baath E. (2010b). Investigating the

mechanisms for the opposing pH-relationships offungal and bacterial growth in soil. Soil Biol Biochem42: 926934.

Shaw AK, Halpern AL, Beeson K, Tran B, Venter JC,Martiny JBH. (2008). Its all relative: ranking thediversity of aquatic bacterial communities. EnvironMicrobiol 10: 22002210.

USDA (United States Department of Agriculture Soil SurveyStaff (1992). Keys to soil taxonomy. SMSS TechnicalMonograph 19. Pocahontas Press: Blacksburg, VA, 541.

Wheeler KA, Hurdman BF, Pitt JI. (1991). Influence

of pH on the growth of some toxigenic species ofAspergillus, Penicillium and Fusarium. Int J FoodMicrobiol 12: 141150.


12

The ISME Journal

soil bacterial and fungal communities across a ph

Documents