Heterotrophy-coordinated diazotrophy is associated with significant changes of rare taxa in soil microbiome

Wei GAOLianfeng WANG and Zhongjun JIA∗

1State KeyLaboratoryof Soil and Sustainable Agriculture,Institute of Soil Science,Chinese Academyof Sciences,Nanjing 210008(China)2Universityof Chinese Academyof Sciences,Beijing 100049(China)3College of Environmental and Chemical Engineering,Dalian Jiaotong University,Dalian 116028(China)

ABSTRACT Soil heterotrophic respiration during decomposition of carbon(C)-rich organic matter plays a vital role in sustaining soil fertility.However,it remains poorly understood whether dinitrogen(N2)fixation occurs in support of soil heterotrophic respiration.In this study, 15N2-tracing indicated that strong N2 fixation occurred during heterotrophic respiration of carbon-rich glucose.Soil organic 15N increased from 0.37 atom%to 2.50 atom%under aerobic conditions and to 4.23 atom%under anaerobic conditions,while the concomitant CO2 flux increased by 12.0-fold under aerobic conditions and 5.18-fold under anaerobic conditions.Soil N2 fixation was completely absent in soils replete with inorganic N, although soil N bioavailability did not alter soil respiration.High-throughput sequencing of the 16S rRNA gene further indicated that:i)under aerobic conditions,only 15.2%of soil microbiome responded positively to glucose addition,and these responses were significantly associated with soil respiration and N2 fixation and ii)under anaerobic conditions,the percentage of responses was even lower at 5.70%.Intriguingly,more than 95%of these responses were originally rare with<0.5%relative abundance in background soils,including typical N2-fixing heterotrophs such as Azotobacter and Clostridium and well-recognized non-N2-fixing heterotrophs such as Sporosarcina, Agromyces,and Sedimentibacter.These results suggest that only a small portion of the soil microbiome could respond quickly to the amendment of readily accessible organic C in a fluvo-aquic soil and highlighted that rare phylotypes might have played more important roles than previously appreciated in catalyzing soil C and nitrogen turnovers.Our study indicates that N2 fixation could be closely associated with microbial turnover of soil organic C when available in excess.

KeyWords: rare phylotypes,diazotrophs,heterotrophic respiration,N2 fixation, 15N2-tracing

INTRODUCTION

Soil fertility is crucial to sustain both agricultural production and soil quality(Suet al.,2006).Agricultural practices such as incorporation of carbon (C)-rich straw into fields have long been employed to improve soil fertility and sustainability (Dossou-Yovoet al., 2016).For instance, it is estimated that approximately 630 million tons of crop straw were generated per year during the period from 1995 to 2005,and that more than 30.5%of this is returned to fields(Liuet al.,2008).The input of this C-rich straw can lead to excessively high soil C/N ratios for a short time,leading to severe nitrogen(N)limitation for microorganisms involved in straw decomposition(Bond-Lamberty and Thomson,2010).It has long been assumed that soil heterotrophic respiration chain is progressively limited with increasing input of C-rich organic matter (Hessenet al., 2013), and dinitrogen (N2)fixation might have occurred to overcome N stoichiometric imbalance of microorganisms in soils to support microbial decomposition process(Rao,1978;Vitousek and Hobbie,2000;Mooshammeret al.,2014).However,the occurrence of soil N2fixation associated with decomposition of C-rich organic matter is poorly understood owing to technical limitation such as culture-dependent approaches and indirect activity measurement by acetylene reduction assay(ARA).

The ARA is considered a classic method for assessment of nitrogenase activity in complex soil environments(Stewartet al.,1967;Saizet al.,2019).Numerous studies have revealed that straw incorporation stimulates N2fixation,which is often explained by the lack of bioavailable N in support of rapid propagation of microorganisms responsible for plant litter decomposition(Roper and Smith,1991;Tanakaet al.,2006).For instance,the ARA indicated that N2-fixing activity could have been stimulated upon input of 1.0%rice straw in waterlogged paddy soils(Yoneyamaet al.,1977),and the fixed N was estimated to be up to 2.2 mg N per gram straw(Rice and Paul,1972).However,significant bias associated with the ARA has been extensively reportedbecause the ratio of C2H2reduced for each mole of N2fixed varies widely from 2.2 to 6.9 among phylogenetically distinct diazotrophs(Hardyet al.,1973).In addition,C2H2may inhibit the growth of diazotroph itself,leading to underestimation of N2fixation activity(Raghoebarsinget al.,2005; Buckleyet al.,2008).For example,a large number of methanotrophs are capable of N2fixation,but methane monooxygenase as the key catalyst for energy production can be inhibited by C2H2(Dalton and Whittenbury,1976).

The rapid advance of stable isotope mass spectrometry technology allows the detection of trace amounts of15N enrichment in soils at unprecedented resolution, thus providing direct evidence for N2-fixing activity with great advantages over ARA for measurements of N2fixation in complex environments(Chalket al.,2017;Saizet al.,2019).In fact,Morriset al.(1985)compared the N2-fixing activities in 25 pasture sites in central Texas of USA by15N2tracing and ARA assay,and the results showed that the ratio of C2H2reduced to N2fixed was highly variable ranging from 0 to 12.It has also been speculated that anaerobic diazotrophs were responsible for N2fixation associated with microbial decomposition of cellulose or straw(Adachiet al.,1989).N2-fixing activity was indeed favored by waterlogged anaerobic rather than aerobic conditions(Yoneyamaet al.,1977).Nonetheless,the taxonomic identities of diazotrophs in complex soils remain poorly understood because it is challenging to establish a direct link between diazotrophic activity in natural settings and the taxonomic identity of active diazotrophs such asAzotobacter vinelandii(Burris and Miller,1941),Azotobacter chroococcum(Yates,1970),andChlorogloea fritschii(Fay,1965)using culture-dependent techniques.Rapid advances in culture-independent techniques have fundamentally changed our perception of soil microbial diversity and provided a means to track the shifts of key diazotrophs in complex environments.

In this study,we investigated the taxonomic identities of potential diazotrophs in an upland agricultural soil amended with C-rich organic matter(glucose)by using15N2-based stable isotope probing and high-throughput sequencing over a 20-d incubation period.We hypothesized that a wide diversity of phylogenetic distinct microorganisms might have catalyzed soil C and N turnover.The objective of this study was to identify the phylogenetic identities of bacterial taxa associated with glucose decomposition under different N limitations.The responsive phylotypes in this study are defined as those with statistically significant increases in relative abundance in response to glucose with or without urea addition.To the best of our understanding,this study represents the first attempt to decipher the taxonomic identities of diazotrophs and heterotrophs in complex soil environments by the combination of high-throughput sequencing and15N2-stable isotope probing.

MATERIALS AND METHODS

Site description

Soil samples were collected from Fengqiu State Key Experimental Station for Ecological Agriculture,Fengqiu County,Henan Province,China(35°00′N,114°24′E).The study area has a warm temperate continental monsoon climate with an annual precipitation of 615 mm and a mean annual temperature of 13.9°C.The soil was originated from alluvial sediments of the Yellow River and classified as an Aquic Inceptisol with a sandy loam texture(Caiet al.,1999).A rotation system with winter wheat and summer maize has been in place for more than 50 years.A detailed field description was provided in Menget al.(2005)and Xiaet al.(2011).Soil samples were collected after the harvest in December 2016,sieved through 2-mm mesh,and stored at 4°C until further use within 5 d.The basic soil physical and chemical properties were determined shortly after transportation to the laboratory as follows:water content,82.3 g kg−1;maximum water-holding capacity(WHC),47.1%; NH+4-N,3.85 mg kg−1;NO−3-N,32.4 mg kg−1;total nitrogen(TN),1.26 g kg−1;total carbon(TC),18.6 g kg−1;soil carbon to nitrogen(C/N)ratio,14.8;and pH(H2O),7.8.

Soil microcosms

Microcosms were constructed to investigate heterotrophy-coordinated diazotrophy in soils with and without glucose amendment in the presence and absence of urea N(Fig.S1,See supplementary Material for Fig.S1).Prior to soil microcosm construction,12 g of soil(dry equivalent)were pre-incubated in darkness at a moisture content of 40%WHC and 28°C for 24 h under ambient air conditions.The pre-incubation was to stabilize soil indigenous microbial communities and to minimize effects associated with subsequent disturbances(Blagodatskaya and Kuzyakov,2013).After pre-incubation,6 g of pre-incubated soil were collected for soil15N,inorganic N,TC,and TN measurements,and for the extraction of soil DNA.

Two oxygen conditions were also established:aerobic(20% O2, volume/volume) and anaerobic (0% O2, volume/volume)incubations(Fig.S1).Each oxygen condition included three treatments:i)H2O(control),no nutrient addition;ii)glucose,8 mg glucose-C g−1soil;and iii)glucose plus urea,8 mg glucose-C g−1soil and 1 mg urea-N g−1soil.Every treatment was conducted with three replicates of soil microcosms and statistical differences were analyzed to alleviate the limitation of biological replicates.Each of the microcosms contained 6 g of pre-incubated soil in a 120-mL serum bottle.Sterile water with or without substrates was applied to each bottle to reach the targeted 60%WHC.Each bottle was tightly closed with a butyl rubber stopper.Afterpre-incubation, the headspace gas in the bottles was evacuated for 3 min by a vacuum pump,and chemically inert argon gas(Ar)was injected into the bottle to remove as much N2as possible.Three-time evacuations were conducted,and the headspace was injected with15N2-O2-Ar(30:20:50)for the aerobic incubation.As for anaerobic incubation,15N-N2(atom percent 99%)and the inert gas Ar were injected into the bottles with a mixing ratio of 3:7 and a final volume of 120 mL.The atom percent of15N2was 99%for all incubations.Soil microcosms were incubated in the darkness at 28°C for 20 d.Gas samples(15 mL)were taken from the headspace of each bottle on days 0, 4, 7, 14, and 20.The headspace mix gases were renewed on a weekly basis.At day 20,the CO2emissions reached a plateau and destructive soil sampling was performed.Approximately 0.5 g of soil were used for15N determination;1.2 g of soil were analyzed for TC and TN;and 3.5 g of soil were mixed with 2 mol L−1KCl for measurement of inorganic N.The rest of the approximately 0.8 g of soil were immediately frozen in−20°C until the extraction of genomic DNA.

Soil C and N turnover

Soil respiration activity was assessed during incubation of soil microcosms as changes in the headspace concentrations of CO2emissions by gas chromatograph measurement(Agilent 7890, USA).Soil15N atom abundance was analyzed to determine soil N2-fixing activity using an elemental analysis-isotope mass spectrometry analyzer (Flash-2000 Delta V Advantage,Germany).Soil C and N transformation was analyzed through changes in soil TN,TC,and inorganic N.The contents of soil TC and TN were determined using a CN element analyzer(Vario Max CN,Germany).The rest of each soil sample was homogenized with 2 mol L−1KCl by shaking at 200 r min−1for 30 min,and then passing through filter paper for the determination of NH+4-N and NO−3-N using a Skalar SAN Plus segmented flow analyzer(Skalar Inc.,Breda,The Netherlands).

Soil nucleic acid extraction and real-time quantitative polymerase chain reaction(qPCR)

Soil microbial DNA was extracted using a FastDNA spin kit for soil(MP Bio,USA),according to the manufacturer’s instructions.The quality and purity of DNA were determined by a micro-UV spectrophotometer(NanoDrop ND-1000, USA).To quantify the soil bacterial and diazotrophic community, real-time qPCR was performed by analyzing 16S rRNA gene andnifHgene copy numbers in all soil treatments at days 0 and 20.The universal primer pair was 515F:GTGCCAGCMGCCGCGG/907R:CCGTCAATTCMTTTRAGTTT(Stubner,2002).The qPCR conditions of the 16S rRNA gene were as follows:initial pre-denaturing at 95°C for 3 min;followed by 35 cycles each comprising 95°C for 30 s of denaturing,55°C for 30 s of annealing,and 72°C for 30 s of extension;the generated amplicon was melted from 65 to 95°C with an increment of 0.5°C per 5 s.The diazotroph primer pair was polF:TGCGAYCCSAARGCBGACTC/polR:ATSGCCATCATYTCRCCGGA(Polyet al.,2001).The Fe protein subunit of nitrogenase reductase(nifHgene)has been most widely used as a biomarker for molecular survey of diazotrophic diversity due to the phylogenetic congruence betweennifHgene and 16S rRNA gene in diazotrophs.The qPCR conditions of thenifHgene were as follows:initial pre-denaturing at 95°C for 10 min;followed by 35 cycles each comprising 95°C for 1 min of denaturing, 55°C for 27 s of annealing, and 72°C for 1 min of extension; the generated amplicon was melted from 65 to 95°C with an increment of 0.5°C per 5 s.The qPCR reaction mixtures and the standards were performed as described previously(Xiaet al.,2011).The amplification efficiencies of the 16S rRNA gene andnifHgene were 98.1%and 100.3%,respectively.

MiSeqsequencing of the 16S rRNA gene and nifH gene

The diazotrophs in all soil treatments were identified by the 16S rRNA gene andnifHgene sequencing.The 515F/907R primer pair was used for the amplification of V4-V5 region of the 16S rRNA gene,and the end of 515F was linked with a 12-bp sample-specific adaptor sequence to distinguish different samples.The PCR conditions were as follows:initial pre-denaturing at 95°C for 3 min;followed by 30 cycles each comprising 95°C for 30 s of denaturing,55°C for 30 s of annealing,and 72°C for 45 s of extension;final extension was performed at 72°C for 10 min.The polF/polR primer pair was used for the amplification of thenifHgene,and the end of polFwas linked with a 6-bp sample-specific adaptor sequence to distinguish different samples.The PCR conditions were as follows:initial pre-denaturing at 95°C for 3 min;followed by 32 cycles each comprising 95°C for 30 s of denaturing,55°C for 30 s of annealing,and 72°C for 45 s of extension;final extension was performed at 72°C for 10 min.All PCR reactions,amplicons purification,and library construction were performed as described previously(Xiaet al.,2011).After library construction,the sequencing of paired ends (300 bp) was conducted on the Illumina MiSeq platform (Illumina, Inc., San Diego, USA).Highthroughput raw sequences that supported our findings were deposited at Sequence Read Archive (SRA) in the NCBI database(https://www.ncbi.nlm.nih.gov)under the accession numbers SRP151506 for the 16S rRNA gene and SRP267904 for thenifHgene.

Bioinformatics analyses

High-throughput raw sequences were analyzed using the Quantitative Insights into Microbial Ecology (QIIME version:1.9.1)pipeline(http://qiime.org/(Kuczynskiet al.,2011).Briefly,paired end reads were merged by the command“join_paired_ends.py”,and barcodes were extracted by“extract_barcodes.py”.Low quality sequences,which had lengths of<200 bp,an average quality score of<25,and primer mismatches,were trimmed.Chimera were also removed by“identify_chimeric_seqs.py”using USEARCH61.For the 16S rRNA gene,389 902 high quality sequences were included and further clustered into operational taxonomic units(OTUs)at 97%sequence identity with the UCLUST algorithm.A representative sequence from each OTU was picked for taxonomic identification using a Ribosomal Database Project(RDP)classifier(http://rdp.cme.msu.edu/).For thenifHgene, 798 851 high quality sequences were included and further clustered into OTUs at 95%sequence identity with the UCLUST algorithm.The representative sequences were aligned against thenifHgene database using the FunGene Pipeline(Wanget al.,2013).

Soil microbial community diversity was examined by rarefaction curves(α-diversity index)and non-metric multidimensional scaling(NMDS;β-diversity index).Rarefaction calculation was executed using the QIIME pipeline(multiple rarefactions,diversity and collate),and Bray-Curtis distances were visualized by NMDS plots in the R environment using the“metaMDS”function of the vegan package.

Identification of responsive phylotypes

We used high-throughput sequencing combined with soil microcosms to identify the responsive phylotypes significantly associated with soil heterotrophy and diazotrophy.The relative abundance of each genus was compared between different treatments by a two-tailedt-test.It should be noted that we focused on bacterial taxa at genus level rather than species and strain levels in this study because higher taxonomic resolutions often lead to greater inaccuracy of phylogenetic classification.Therefore, responsive phylotypes were defined as genera showing statistically significant increases in relative abundance in soil microcosms amended with glucose or glucose plus urea when compared to these phylotypes in the control (background soil) (Fig.S1).Responsive rare phylotypes were then defined as the significantly increased phylotypes with less than 0.5%relative abundance at genus level in the control.

We speculated that some heterotrophic non-diazotrophs may also have rapid growth, although soil bioavailable N was extremely low.Therefore,responsive phylotypes were further classified into diazotrophs and non-diazotrophs by the curated N2-fixing microorganism database embedded within the RDP pipeline.Briefly,all N2-fixing strains and potential diazotrophic species with fully sequenced genomes from the FunGene Pipeline(http://fungene.cme.msu.edu/)(Fishet al.,2013)were constructed as the reference database.The representative sequences of the 16S rRNA gene were further assigned against the reference database(Wanget al.,2007).Moreover,thenifHgene sequencing was also performed to identify the potential diazotrophs using the FunGene pipeline as described(Wanget al.,2013).In addition,the cultivation of the responsive phylotypes was also recognized using the RDP classifier.Briefly,probably cultivable phylotype indicated the taxon of the 16S rRNA gene sequence showing the highest relatedness to a cultured microorganism,while uncultivated phylotype meant the highly similar microorganism that was uncultured,although it could be confidently assigned to a certain genus.

Statistical analysis

A phylogenetic tree was constructed using the neighbor joining method in MEGA 4.0 with bootstrapping of 1 000 replicates (Tamuraet al., 2007).SPSS 22.0 (SPSS Inc.,Cary,USA)and R software packages were used for statistical analysis of the MiSeq sequencing data.Heml was used for illustrating heatmaps(Denget al.,2014),A genus tree was constructed using the Galaxy Version 1.0.0(graphical phylogenetic analysis)(Asnicaret al.,2015).

RESULTS

Soil microbial respiration and N2 fixation

Over a 20-d incubation period, the cumulative CO2emissions reached to 4.33%(volume/volume)in the control microcosms under aerobic conditions.Glucose addition significantly stimulated CO2emissions to 56.3%,while up to 60.9%CO2emissions were observed in soil microcosms amended with glucose plus urea(P <0.01).Similar trends were obtained under anaerobic conditions.Amendments of glucose and glucose plus urea significantly stimulated anaerobic CO2emissions by 5.21-and 5.18-fold relative to the control treatment,respectively(P <0.01).However,urea addition resulted in no statistically significant increase of CO2emission in glucose-amended soil microcosms(P >0.05)(Fig.1a).The control showed no evidence of15N2-fixing activity,whereas glucose addition resulted in significant15N enrichment of soil organic N, especially under anaerobic conditions(P <0.01).Soil organic15N increased from 0.37 atom%in the control to 4.23 atom%under anaerobic glucoseamended conditions.Adding urea completely eliminated the stimulatory effect of glucose on15N2-fixing activity(Fig.1b).

Glucose,regardless of urea addition,significantly altered soil bacterial community(P <0.05)(Fig.1c).The Shannon’s diversity index decreased significantly,from 8.76 and8.69 in control soils,to 7.31 and 6.03 in glucose-amended soil microcosms under aerobic and anaerobic conditions,respectively (Fig.1d).This was further evidenced by the abundance of the 16S rRNA gene, showing increases of 1.48-fold under aerobic conditions and of 2.52-fold under anaerobic conditions in the presence of glucose(Fig.1e).Similar trends were obtained in soil microcosms amended with glucose plus urea.Glucose addition also stimulatednifHgene copy numbers,from 8.34×106copies g−1soil at day 0 to 3.39×108copies g−1soil under aerobic conditions and to 3.24×107copies g−1soil under anaerobic conditions at day 20(Fig.1f).

Fig.1 Dynamic changes in soil heterotrophic and diazotrophic activities and in microbial community structures under aerobic and anaerobic conditions over a 20-d incubation period:changes of cumulative CO2 emissions(a)during soil heterotrophy in microcosms and soil 15N atom enrichment(b),β-diversity by non-metric multidimensional scaling(NMDS,c),α-diversity by Shannon’s diversity index(d),16S rRNA gene abundance(e),and nifH gene abundance(f)in microcosms.H2O,C,and CN represent soil microcosms amended with water(control),glucose,and glucose plus urea,respectively.Error bars represent the standard errors of the mean of triplicate microcosms,with some error bars smaller than the symbol size.Different letters above the bars represent significant differences(P <0.05).

Glucose,regardless of urea addition,increased the bacterial respiration activity in the soil,especially under aerobic conditions(Fig.1a).This activity was partly due to an increased abundance of the bacterial community,which was higher under anaerobic conditions(Fig.1e).Shannon’s diversity index reduction upon glucose addition indicated that some taxa predominated while others were limited(Fig.1d).Nitrogenase activity increased upon glucose addition,especially under aerobic conditions but the activity was contrasted by urea(Fig.1b).Finally,glucose induced N2fixation,especially under anaerobic conditions;this is correlated with the bacterial abundance and indicates a higher efficiency of anaerobic bacteria in fixing N2.It should be noted that N2-fixing activities were in contrast with the abundance ofnifHgenes, and this may be explained by the absence of transcriptional activities of nitrogenase innifH-carrying diazotrophs.Meanwhile, TC was significantly higher in anaerobic than aerobic soils when glucose was amended,and the relationship of C budget to N2fixation warrants further studies using C and N dual isotope tracing techniques.

Responsive taxa driving soil heterotrophyand diazotrophy

High-throughput sequencing of the 16S rRNA andnifHgenes in microcosms with glucose and/or urea addition was performed to identify responsive taxa responsible for soil heterotrophy and diazotrophy.The significant CO2emissions and15N enrichment in soils implied the significant growth of responsive phylotypes which could be evaluated by dynamic changes of the 16S rRNA gene andnifHgene biomarkers(Blazewiczet al.,2013).Responsive taxon was defined as the phylotype showing statistically significant increase of its relative abundance in glucose-or glucose plus urea-amended soil microcosms,when compared to that in control soils.Atotal of 26 phyla and 421 genera were observed among all soil treatments(Fig.S2).

A small portion of soil microbiome responded positively to glucose or glucose plus urea additions (Table I).For instance,among 421 phylotypes detected in soils,there were only 64 phylotypes that could be considered as responsive taxa in glucose-amended soil microcosms under aerobic conditions.It was noteworthy that up to 57 out of the 64 responsive phylotypes could be considered as rare taxa with less than 0.5% relative abundance in control soils.Similarly,there were 54 rare phylotypes out of 61 responsive phylotypes in soil microcosms amended with glucose plus urea under aerobic conditions.Similar trends were observed under anaerobic conditions.There were 24 and 29 responsive phylotypes in soil microcosms amended with glucose and glucose plus urea, respectively.More than 95% of these phylotypes could be assigned as responsive rare phylotypes in anaerobic soil microcosms.

TABLE INumbers of total and responsive phylotypes at different taxonomic levels in soil microcosms amended with glucose(C)and glucose plus urea(CN)under aerobic and anaerobic conditions over a 20-d incubation period

The responsive taxa showed significant increase in relative abundance upon additions of glucose or glucose plus urea(Fig.2a).Over a 20-d period of aerobic incubation,the relative abundance of 64 responsive phylotypes increased from 11.7% in the control to 66.2% in glucose-amended soil microcosms.Among them, 57 responsive phylotypes could be considered as rare taxa accounting for 80.4% of the total increases under aerobic conditions.Similar results were observed in aerobic microcosms amended with glucose plus urea.As for anaerobic microcosms,the increase of responsive phylotypes was even higher (Fig.2a).For instance, the relative abundance of 24 responsive phylotypes was stimulated from 2.07%in the control to 73.4%in glucose-amended soil under anaerobic conditions,representing a 35.5-fold increase,and 23 responsive rare phylotypes accounted for 67.7% of the total increases.Similarly, 29 responsive phylotypes represented a 37.8-fold increase in soil microcosms amended with glucose plus urea, and 28 responsive rare phylotypes accounted for 83.8%of the total increases.

The rare phylotypes represented a high proportion in both cultivated and uncultivated soil microbiomes.For instance,in glucose-amended soil microbiomes under aerobic conditions,a total of 64 responsive phylotypes showed significant increases including 54 probably cultivable(24 diazotrophs and 30 non-diazotrophs)and 10 uncultivated phylotypes(5 diazotrophs and 5 non-diazotrophs)(Fig.2b).The 24 probably cultivable diazotrophs increased in relative abundance from 6.70%in the control to 33.9%in glucose-amended soil microbiomes under aerobic conditions,while 20 out of these diazotrophs were rare phylotypes accounting for 38.0%of the total increases.Similarly,the 30 probably cultivable nondiazotrophs increased in relative abundance from 3.82%in control to 27.3%in soil microcosms amend with glucose under aerobic conditions,while 28 out of these non-diazotrophs were rare phylotypes contributing 36.1% of the total increases.Similar trends were observed in soil microcosms amended with glucose plus urea under anaerobic conditions.Under anaerobic conditions,the proportion of probably cultivable responsive phylotypes was even higher(Fig.2b).For instance,there were 11 probably cultivable diazotrophs and 9 probably cultivable non-diazotrophs with 57.6%and 11.4%relative abundance increases in glucose-amended soil microcosms under anaerobic conditions,respectively.The probably cultivable rare diazotrophs and non-diazotrophs contributed up to 48.5%and 16.0%of the total increases in glucose-amended soil microbiomes,respectively.Similarly,in anaerobic soil microcosms with glucose plus urea, the probably cultivable rare diazotrophs and non-diazotrophs accounted for 55.2%and 25.6%of the total increases,respectively.

Fig.2 Changes in responsive taxa associated with soil heterotrophy and diazotrophy under aerobic and anaerobic conditions over a 20-d incubation period:changes in relative abundance of responsive phylotypes(left ring graphs)and the proportion of rare phylotypes in responsive phylotypes(right column graphs)(a)and changes in relative abundance of responsive diazotrophs and non-diazotrophs(b).H2O,C,and CN represent soil microcosms amended with water(control),glucose,and glucose plus urea.For a,in the ring graphs,the percentages in the center represent the relative abundance of responsive phylotypes.The inset figures in b represent the propotion of rare phylotypes in the responsive diazotrophs and non-diazotrophs;the values above the column indicate the number of responsive phylotypes and error bars represent the standard errors of the mean of the triplicate microcosms.All analyses were conducted based on the 16S rRNA gene sequence.

Taxonomic identities of responsive diazotrophs and heterotrophs in soil

Among 107 responsive phylotypes detected from all soil treatments (Table SI), 49 key phylotypes contributed the majority of the total increases in relative abundance(Fig.3).Over a 20-d period of aerobic incubation,responsive phylotypes showed significant differentiation in soil microcosms with glucose and glucose plus urea.For instance,Azotobacter,Pseudoxanthomonas, andBacillus-like cultivable diazotrophs increased in relative abundance from 0.08%,0.10%, and 1.00% in the control to 7.98%, 3.48%, and 3.44%in glucose-amended soil microcosms under aerobic conditions,respectively.These three dominant phylotypes accounted for 54.8%of the total increases(Fig.3).In contrast,the cultivable diazotrophs in aerobic soil microcosms with glucose plus urea were dominated by different responsivetaxa includingChitinophaga, Enterobacter,andLysobacterwith relative abundance of 5.20%,4.48%,and 5.44%,respectively(Fig.3).Similarly,uncultivated diazotrophs in glucose-amended aerobic soil microcosms were dominated byRhizobiaceaeaccounting for 67.3%of the total increases,whileAlcaligenaceae-like probably cultivable diazotrophs showed the highest increase in relative abundance upon glucose plus urea additions under aerobic conditions(Fig.3).Similar trends were observed for responsive non-diazotrophs.For instance,Sporosarcina-andAgromyces-like dominant cultivable non-diazotrophs contributed 47.8%of the total increases in aerobic soil microcosms amended with glucose,while in microcosms amended with glucose plus urea,the probably cultivable non-diazotrophs were dominated byLuteimonasandAgromyces(Fig.3).Similarly,Intrasporangiaceae-andChitinophagaceae-like uncultivated nondiazotrophs dominated aerobic soil microcosms amended with glucose and glucose plus urea,representing 82.6%and 40.4%of the total increases,respectively(Fig.3).

Under anaerobic conditions,Bacillus-andClostridiumlike probably cultivable diazotrophs dominated the communities of responsive phylotypes in both soil microcosms amended with glucose and glucose plus urea.The relative abundance of the two phylotypes increased by 23.0%and 19.0%,respectively,contributing more than 70%of the total increases under anaerobic conditions(Fig.3).Uncultivated diazotrophs such asRuminococcaceae,Veillonellaceae,andCostridiaceae 1dominated the communities of responsive phylotypes in anaerobic soil microcosms amended with glucose and glucose plus urea (Fig.3).As for probably cultivable non-diazotrophs,Sedimentibacteralso dominated in anaerobic soil microcosms amended with glucose and glucose plus urea(Fig.3).This single dominant phylotype contributed up to 78.9%and 77.6%of the total increases in glucose-and glucose plus urea-amended soil microcosms,respectively.

Fig.3 Taxonomic identities(genus level)of responsive diazotrophs and non-diazotrophs associated with soil heterotrophy and diazotrophy under aerobic and anaerobic conditions over a 20-d incubation period.All responsive phylotypes were sorted by increase in relative abundance from most to least abundant(Table SI).Heatmaps showed the relative abundance of responsive phylotypes in soil microcosms amended with water(H2O,control),glucose(C),and glucose plus urea(CN).The orange columns represent the increases in relative abundance in C vs.H2O,and the solid green circles indicate the increases in relative abundance in CN vs.H2O.The responsive phylotypes in blue text showed significant increases in relative abundance under both aerobic and anaerobic conditions.Error bars represent standard errors of the mean of the triplicate microcosms.All analyses were conducted based on the 16S rRNA gene sequence.Clostridium S S=Clostridium Sensu Stricto.

DISCUSSION

Soil N2 fixation occurred in associat ion with C-rich glucose respiration

Our study provides compelling evidence that soil N2fixation was activated by glucose-induced heterotrophy(Fig.1).This implies that C-rich straw incorporations could have resulted in the occurrence of N2fixation as the potential N source to overcome N stoichiometric imbalance in the field.Addition of glucose significantly stimulated soil heterotrophic respiration and we speculate that the rapid growth of microorganisms resulted in the deficiency of soil bioavailable N,eventually triggering N2fixation(Fig.1 and Fig.S3).In fact, C-rich matter consisting of low-molecular-weight compound is ubiquitous,and it could likely be a potential source of C capable of meeting diazotroph energy demands(van Heeset al.,2005;Drakeet al.,2015).This is particularly evident in agriculture ecosystems as different management regimes such as the incorporation of C-rich straw into soils could result in severe imbalance of soil C/N,and N2fixation would most likely have occurred especially during the stage of the decomposition as the inorganic N is progressively limited(Trinsoutrotet al.,2000;Liuet al.,2008).It should be noted that our study cannot reflect field conditions because the release of C from crop residue decomposition is a slow process,and the availability of easily accessible C may be relatively low thus limiting the growth of microcosms in complex soil environments.Based on soil15N atom enrichment and TN,the net amount of N2fixed was estimated to be 314.7 and 560.5 kg N ha−1year−1under aerobic and anaerobic conditions,assuming a soil bulk density of 1.3 g cm−3.This estimate is 10 times higher than that of natural ecosystems(Clevelandet al.,1999).It should be emphasized that in natural environments soil physical structures allow the coexistence of aerobic and anaerobic zones in various ways,and soil microbial processes can be compartmentalized into distinct niches.Therefore,microcosm study cannot represent what is naturally occurring under field conditions,and field assessment of N2fixation budget in association with straw incorporation warrants further study.

A high proportion of rare taxa responded positivelyto soil heterotrophyand diazotrophy

High-throughput sequencing of the 16S rRNA gene further revealed that only 15.2%of soil communities responded positively to glucose additions,implying that a small portion of soil microbiome is associated with respiration-coordinated N2fixation.Based on microbial biomass analysis,it has been proposed that only a narrow subset of microorganisms is functionally active,while the taxonomic identities of these putatively active microorganisms have remained largely elusive (Blagodatskaya and Kuzyakov, 2013).Our study expanded this concept by high-throughput sequencing of“responsive taxa” as putatively active microbiome in soil microcosms.It should be emphasized that the population size rather than relative abundance of some active phylotypes might have been significantly stimulated,thereby evading detection based on the statistically analytical strategy used in this study.Moreover,up to 95%of the responses were originally rare with<0.5%relative abundance in background soils(Fig.2).This implies that soil nutrient limitation occurred for the majority of soil microorganismsin situ,and these rare phylotypes could be readily re-activated when there are favorable environmental conditions for growth.The results also demonstrated that low abundance taxa likely have a disproportionately functional effect on soil C and N turnovers.It is noteworthy that rarity is not an absolute concept but relative to soil types and their nutrients loads.In addition,the probably cultivable phylotypes contributed>80%of the total responsive communities(Fig.2b),implying their faster metabolic rates than uncultivated phylotypes.This could be explained by the ecological attributes ofrandk-selected strategists(Fiereret al.,2007).Future studies are warranted to accurately link the key rare phylotypes with heterotrophy and diazotrophy in complex soils.

Distinct nitrogenase activities under oxygen stress in soil

High-throughput sequencing of the 16S rRNA andnifHgenes(Figs.S4 and S5,See supplementary Material for Figs.S4 and S5)consistently revealed that well-known N2-fixers such asAzotobacter,Bacillus,andClostridiumlikely played important roles in catalyzing N2fixation under aerobic and anaerobic conditions(Dixon and Kahn,2004).Bacillus-like phylotypes might assimilate N in presence or absence of oxygen(Hino and Wilson,1958),as indicated by their significant increase in both aerobic and anaerobic microcosms(Fig.3).Azotobacter chroococcum,a well-known aerobic diazotroph,can utilize a wide array of diverse physiological strategies to attenuate oxygen stresses,including consumption of excess oxygen by respiration and oxygen diffusion barriers(Robsonet al.,1986;Poole and Hill,1997;Dixon and Kahn,2004).This was also supported by the fact thatAzotobacter-like diazotrophs had only 0.1%relative abundance in anaerobic microcosms amended with glucose,but up to 8.06%relative abundance under aerobic conditions(Fig.S4).Conversely,Clostridium-like anaerobic diazotrophs often persisted under anaerobic or micro-anaerobic environments (Riceet al.,1967; Kundiyanaet al., 2011), andClostridium-like diazotrophs increased in relative abundance by only 0.56%in soil microcosms amended with glucose under aerobic conditions,but up to 19.0%under anaerobic conditions(Table SI).

Strong N2fixation occurred in aerobic microcosms upon glucose addition(Fig.1b), although nitrogenase was sensitive to oxygen(Robson and Postgate,1980;Marchal and Vanderleyden, 2000).This might be explained by rapid depletion of headspace oxygen associated with microbial respiration of glucose,despite no measurement of oxygen concentrations.In the absence of oxygen,the hemoglobin of diazotrophs can provide more effective oxygen transport for energy metabolism,thus facilitating nitrogenase activities under anaerobic conditions(Dixon and Kahn,2004).In the presence of oxygen,soil diazotrophic communities would support strong respiratory activity to protect nitrogenase and obtain as much as energy to fuel nitrogenase(Marchal and Vanderleyden,2000).Although the increase in respiration of diazotrophs was suggested to decrease the oxygen concentration around nitrogenase to a tolerable level,but the N2fixation rates would likely be significantly inhibited to some extent under high concentrations of oxygen(Fig.1a,b).The higher diversity of diazotrophs was also found in aerobic rather than anaerobic microcosms(Figs.S2 and S5).This was supported by the highernifHgene abundance in aerobic soils as well(Fig.1f).It seems plausible that upland soilsin situwere constantly exposed to 21%oxygen in the atmosphere,and aerobic microbes could not be readily adapted to the strictly anaerobic experimental microcosms.The evolutionary trajectory of anaerobic diazotrophs in aerobic upland soils warrants further studies with respect to their ecological and agricultural significances.

Multiple N-acquisition strategies by soil diazotrophs and non-diazotrophs

Addition of urea completely eliminated the15N enrichment in soil organic N (Fig.1b), and indicated that soil diazotrophs preferentially assimilated soil inorganic N over atmospheric N2(Norman and Friesen, 2017).This N-acquisition strategy was supported by the limitation of soil bioavailable N in response to glucose addition including soil C/N imbalance and limited inorganic N(Fig.S3).However,some diazotrophs remained dominant in N-replete soil microcosms(Robsonet al.,1986;Dixon and Kahn,2004).For instance,high-throughput sequencing of the 16S rRNA gene showed thatClostridium-like anaerobic diazotrophsincreased significantly in microcosms amended with glucose and glucose plus urea;this is consistent with previous results in batch cultures and soil environments(Riceet al.,1967;Norman and Friesen,2017).A significant increase was also observed in relative abundances ofEnterobacter-like diazotrophs in N-rich soil microcosms(Fig.3),suggesting the inactivation of nitrogenase during strong respiration of glucose as indicated by unchanged15N atom percent in soil organic N over the incubation period of 20 d.These key diazotrophs likely acquired inorganic N through direct uptake when soil bioavailable N were replete(Norman and Friesen,2017).Conversely,some diazotrophs were sensitive to bioavailable N.For instance,Azotobactershowed 8.1%relative abundance in aerobic glucose-amended soil microcosms, but only 0.1% in soil microcosms amended with glucose plus urea(Fig.3).These diazotrophs are likely to be responsive to increases in bioavailable N, shutting down nitrogenase activity upon the supply of easily accessible inorganic N.These results consistently demonstrated that diverse diazotrophs participated not only in N2fixation but also in heterotrophic respiration processes in soils,by employing different N-usage strategies in response to soil nutrient heterogeneity and consequently shaping distinct niches within dynamic microbial communities (Hsu and Buckley,2009).

It was noteworthy thatSporosarcina-,Agromyces-,andSedimentibacter-like non-diazotrophs showed significant increases in glucose-amended soil microcosms although the soil bioavailable N was extremely low (Figs.3 and S3).These can be explained by the metabolic strategies used bySporosarcina-andAgromyces-like non-diazotrophs;i.e.,they are fast-growth phylotypes and possess extremely high nitrogen-use efficiency (Gledhill and Casida, 1969;Dworkinet al.,2006).Sedimentibacter-like non-diazotrophs could also quickly adapt to changing environments as suggested in previous studies(Obstet al.,2005;van Doesburget al., 2005).This likely suggested that oligotrophic nondiazotrophs seemed to be favored under resource-limited conditions caused by high substrate affinities.It also implies their strong physiologically metabolism potential under constantly changing environments.Cultivation of fast-growth non-diazotrophs would be crucial toward better understanding of the microbially mediated processes of soil organic matter.

It is interesting to note that there was no statistically significant difference of TC in microcosms with glucose and glucose plus urea addition under both aerobic and anaerobic conditions.It thus implies that N acquisition in microbiomes did not severely constrain degradation of glucose,although the exact mechanism of N metabolisms remained unknown.This was also supported by similar CO2emissions in glucoseor glucose plus urea-amended microcosms(Fig.1a).Soil transcriptomics can be a key technique for future study to identify the differential regulation of N metabolism genes under various C and N conditions.

CONCLUSIONS

In summary,our study suggests that 15.2%and 5.70%of a total of 421 phylotypes were significantly associated with soil heterotrophic diazotrophy under aerobic and anaerobic conditions,respectively.A high proportion of responsive rare taxa might have played important roles in catalyzing soil heterotrophy and diazotrophy includingClostridium-andAzotobacter-like diazotrophs andSporosarcina-,Agromyces-,andSedimentibacter-like non-diazotrophs.This study highlights the possible origin of bioavailable N source in support of C-rich organic matter decomposition in a fluvo-aquic soil,although heterotrophy-coordinated diazotrophy may depend on soil types and their nutrient loads.Future studies are warranted to elucidate the relative contributions of phylogenetically distinct taxa to soil C and N turnover in different soils with a variety of C sources.

ACKNOWLEDGEMENTS

This work was financially supported by the National Science Foundation of China(Nos.91751204,41530857,and 41471205), the National Basic Research Program of China(No.2015CB150501),and the Strategic Priority Research Program of Chinese Academy of Sciences (CAS)(No.XDB15040000).We are grateful to Prof.Bruce A.HUNGATE from Northern Arizona University, USA for constructive comments on the previous draft of manuscript.We thank Mr.Zhiying GUO and Ms.Huimin ZHANG for technical support and the members of our laboratory for helpful discussions.We also thank the staffof the Analysis Center at the Institute Soil Science,CAS for technical assistance,including Ms.Rong HUANG and Mr.Zuohao MA for Illumina MiSeq sequencing,Ms.Deling SUN for15N-atom abundance assay,Ms.Yufang SUN for soil C and N content assay,and Mr.Ruhai WANG for ammonia and nitrate-based N content assay.

SUPPLEMENTARY MATERIAL

Supplementary material for this article can be found in the online version.