Enhanced N2O emission rate in field soil undergoing conventional intensive fertilization is attributed to the shifts of denitrifying guilds
2021-03-17MengmengJIHaoTIANXiaogangWUJiLIYingZHUGuojunWUTingXUJingguoWANGandXiaojunZHANG
Mengmeng JI,Hao TIAN,Xiaogang WU,Ji LI,Ying ZHU,Guojun WU,Ting XU,Jingguo WANG and Xiaojun ZHANG
1State Key Laboratory of Microbial Metabolism and School of Life Sciences and Biotechnology,Shanghai Jiao Tong University,Shanghai 200240(China)
2College of Resources and Environmental Sciences,China Agricultural University,Beijing 100193(China)
(Received August 18,2019;revised September 10,2020)
ABSTRACT The emission of greenhouse gas N2O in agricultural soil is modulated by N fertilization that could be converted to N2O by denitrifiers under anaerobic condition.Nevertheless,the effect of denitrifiers on N2O emission has not been thoroughly elucidated.In this study,we explored the denitrifying gas kinetics,nitrate content,transcribed denitrifying functional genes(narG,nirS,nirK,and nosZ),and the active bacteria during anaerobic incubation of soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS),both sampled from a vegetable greenhouse experimental site.The CNS sample showed significantly higher N2O emission rates relative to the RNS sample.However,the difference in N2O emission between the soils was neither because of the cumulative nitrate content nor the quantity of denitrifying gene transcripts.The distinct fertilization regimes shaped the significantly different bacterial communities in these soils.The absolute abundance of bacteria that produce N2O but lack the nosZ gene for N2O reduction(for example,the dominant Kaistobacter)was higher in CNS than in RNS.Meanwhile,the abundance of two operational taxonomy units(OTUs),namely Rhodanobacter,belonging to the most abundant genus in denitrifying guilds,was strongly enriched in CNS and showed significant positive correlation with N2O/(N2O+N2).The predominance of these bacterial OTUs in the CNS denitrifying guild strongly suggested that high N2O emission from the soil with long-term conventional intensive fertilization might be primarily attributed to the reshaping of distinct denitrifying guilds in their bacterial communities.
Key Words: denitrification,denitrifier community,denitrifying bacteria,denitrifying functional gene,greenhouse gas,N fertilization
INTRODUCTION
N2O is a powerful greenhouse gas with a 296-fold stronger warming potential than CO2(IPCC,2001).In the last century,the concentration of atmospheric N2O has increased rapidly,rising from 270µg kg−1in 1900 to 319µg kg−1in 2005,having remained relatively stable over the previous two thousand years(Solomonet al.,2007).Global soil ecosystems are the largest source of N2O,with approximately 65% of global N2O levels originating from soil emissions(IPCC,2001).The overuse of N fertilizer is the cause of the exponential increase in N2O emission(Shcherbaket al.,2014).
The difference in N2O emission between conventional over-fertilized soil and reduced N-fertilized soil was reported in studies based on the long-term fertilization field experiments(Bremer,2006;Zhanget al.,2017).It gives rise to the question of why conventional fertilization management results in higher N2O emission.Long-term fertilization in agricultural practices could change soil physicochemical properties as well as the associated microbial community structure(Marschneret al.,2003).A few intensive studies have previously focused on the impact ofand pH on denitrification rates in soil(Firestoneet al.,1979;Thomaset al.,1994).Generally,content was considered to be a controller of denitrification.However,the critical evaluation of the importance ofcontent or the composition and activity of functional genes in the denitrifying community in affecting N2O emission has not been systematically investigated yet.
Denitrification is the main source of N2O under the anaerobic state(Zhuet al.,2013),which is the stepwise reduction ofto N2or N2O carried out by diverse denitrifiers.Investigation of functional denitrifiers is essential for understanding the relationship between the microbial community and N2O emission.Some studies have emphasized the linkage between the denitrifier community and N2O emission at the molecular level(Uchidaet al.,2014;Lennon and Houlton,2016),including the role of denitrifying functional gene abundance(Dandieet al.,2008;Yanget al.,2017),transcript numbers(Uchidaet al.,2014),and posttranscription levels(Liuet al.,2010).Studies also reported that denitrifier community size(Hallinet al.,2009)or structure and activity of denitrifying community(Dambrevilleet al.,2006)were important for N2O emission.However,it still remains unknown which of these factors are more dominant for N2O emission from soils with different N fertilization regimes and what is the ecological mechanism for denitrifier guilds influencing the emission.
In this study,we measured the N2O emission capacity of the soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS)under the samebackground to identify the effect of soilcontent.In addition,we quantified the transcript numbers of denitrifying functional genes(narG,nirS,nirK,andnosZ)and examined the composition differences of active denitrifying bacteria between the conventional and reduced N fertilization regimes.The results of these experiments will help us elucidate the microbial effect and its mechanism on elevation of N2O emission in over-fertilized crop soil.
MATERIALS AND METHODS
Soil incubation and gas kinetics
Soil samples were collected from a long-term experimental field site established in 2004 and located at Luojia Village(36°55′N,118°45′E)in Shouguang,Shandong Province of China.The plots of soil were treated with two fertilization regimes,conventional intensive N fertilization and reduced N fertilization,and each treatment was replicated thrice in each randomly distributed plot with the size of 7.8 m×5.6 m.The information for the two soils,CNS and RNS,is reported in Table I.Soil was collected from five different points in each treatment plot at top 20 cm depth and mixed to be a final sample.

TABLE ICharacteristics of the site and the soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS)in field experiment
Moist soil samples(top 0-20 cm)were sieved(<2 mm)and stored at 4°C for further use.Before incubation,soil was pre-incubated at 20°C in the dark for 7 d.Thereafter 16 g of moist soil was transferred into a 120-mL serum vial that was sealed with airtight butyl-rubber and aluminum crimp cap.To monitor denitrifying gas kinetics,vials containing CNS and RNS samples were placed on a robotized incubation system(ROBOT)(Molstadet al.,2007),a system to analyze denitrifier community using N cycling gas(NO,N2O,and N2),and were incubated at 20°C under anaerobic conditions(filled with He in the headspace).Triplicate vials were destructively sampled at 0,0.3,0.6,1.2,and 11 h for analyzing nitrogenous compound and microbiota.
Kinetics of denitrifying gaseous N(NO,N2O,and N2)were measured using ROBOT.The ROBOT system comprises a vacuum cleaning system(Jincheng Company,China),a gas chromatograph(7890 A,Agilent,USA),and an NO analyzer(NOA 280i;GE Analytical Instruments,USA).In general,the headspace gas in the sample bottle was pumped into the gas chromatograph by a peristaltic pump,and N2O and N2concentrations were determined using an electron capture detector and thermal conductivity detector,respectively.NO in the gas flow piped from the gas chromatograph was measured using the NO analyzer.The lower detection limit for the thermal conductivity detector and the NO analyzer were 5 mg kg−1and 0.5µg kg−1,respectively,and the upper detection limit for the electron capture detector was 150 mg kg−1.
Besides NO,N2O,and N2accumulation,we also calculated the N2O index(IN2O)to characterize the denitrification product ratio N2O/(N2O+N2)during the anaerobic incubation(Liuet al.,2010).The N2O index formula is:

where N2O(t)represents the accumulation of N2O at any timet,N2(t)represents the accumulation of N2at any timet,andtendis the end time of the experiment.
Measurement of N content
The inorganic N ion in the soil samples were extracted by the KCl extraction method(Kachurinaet al.,2000).The determination of-N was followed as per the method described by Bao(2000),and the-N and-N concentrations were determined using the phenoldisulphonic acid method(Eastoe and Pollard,1950).
Ribonucleic acid extraction and quantity of functional genes
Ribonucleic acids(RNA)were extracted from 1.5 g of soil,after 0,0.3,0.6,1.2,and 11 h of incubation in anaerobic vials,using the RNA PowerSoil total RNA isolation kit(Mo Bio,USA).DNA was digested and removed using the Invitrogen DNase I kit(Invitrogen,USA)until no DNA was visible on an electrophoresis gel of extracted RNA samples.RNA samples were further purified using the PowerClean®Pro RNA clean-up kit(Mo Bio,USA)and the quantification and quality were measured at 260 nm with a BioDrop Duo(BioDrop,England).Finally,extracted RNA samples were reverse transcribed to single-stranded cDNA using the SuperScript®III first-strand synthesis system(Invitrogen,USA)for reverse transcription-polymerase chain reaction(PCR).
The quantification ofnarG,nirS,nirK,nosZ,and 16S rRNA gene transcript numbers were performed with quantitative PCR(qPCR),and detailed experimental operations are presented in the Supplementary Material.
16S rRNA gene sequencing and bacterial community analysis
The 16S rRNA gene sequencing library of the V3-V4 region of cDNA from the CNS and RNS samples was conducted as previously described(Zhanget al.,2016)and sequenced with Illumina MiSeq(Illumina,USA).High-quality sequences were subsequently clustered into operational taxonomy units(OTUs)using USEARCH(version v8.0.1623)with a 97% cutoff.The alpha diversity of each sample was analyzed using the Shannon diversity index and the rarefaction curves.In addition,the representative sequences for each OTU were inserted into a phylogenetic tree using FastTree and submitted to the RDP(Ribosomal Database Project,version 2.10)classifier to identify the phylogeny with a confidence threshold of 80%.
Principal coordinate analysis(PCoA)was performed using phylogenic analysis and the relative abundance of the representative OTU sequences based on weighted UniFrac distance(Lozupone and Knight,2005).All analyses were performed using QIIME(version 1.8.0)(Caporasoet al.,2010).
Prediction of functional genes in the denitrifying community
Phylogenetic Investigation of Communities by Reconstruction of Unobserved States(PICRUSt)is an approach used to infer the functional content of the metagenome from 16S rRNA information(Langilleet al.,2013).PICRUSt was used to predict denitrifying bacterial composition.PICRUSt-compatible 16S rRNA OTU tables were developed using a closed-reference OTU-picking method in QIIME against the Greengenes database,and the resulting OTU table was later imported into an online version of Galaxy(http://huttenhower.sph.harvard.edu/galaxy/root?tool_id=PICRUSt_normalize).All KEGG(Kyoto Encyclopedia of Genes and Genomes)orthologies(KOs)were submitted to the KEGG Mapper Search Pathway(http://www.genome.jp/kegg/tool/map_pathway1.html)to identify KOs associated with denitrification(Table SI,see Supplementary Material for Table SI).Thereafter,the OTUs that contributed to denitrification were identified using QIIME(metagenome_contributions.py)to form new denitrification-OTU tables.
Random forest models(Breiman,2001)were introduced to select key OTUs that contributed to the segregation of thenosZ-containing denitrifiers in the CNS and RNS samples.The importance of the OTUs was determined by the mean decrease in the accuracy of discrimination,with a value greater than 0.000 7 considered to be indicative of the key OTUs(Table SII,Fig.S1,see Supplementary Material for Table SII and Fig.S1).
Statistical analysis
The Mann-Whitney and two-way analysis of variance(ANOVA)tests were performed using Prism version 6.01(GraphPad,USA)to examine the variations in alpha-diversity,soil nitrogenous compounds,and denitrifying gas kinetics.To explore the correlation between the relative abundance of thenosZ-containing key OTUs and N2O/(N2O+N2),Spearman’s rank correlation test was performed using MATLAB 2013a(The MathWorks Inc.,USA).Significant differences between the groups were calculated using a multivariate analysis of variance(MANOVA)in MATLAB 2013a(The MathWorks Inc.,USA).
Accession numbers
Nucleotide sequence data were deposited at the National Center for Biotechnology Information(NCBI)under the accession number SRP111441.
RESULTS
Gas kinetics and soil physiochemical parameters in anaerobic incubation
The gas kinetics of NO,N2O,and N2accumulation for soils CNS and RNS are shown in Fig.1a,b.NO and N2O showed linear increases in both CNS and RNS during anaerobic incubation,while the N2O accumulation rates in CNS were significantly higher(Ftest,P<0.000 1)than in RNS(Table SIII,see Supplementary Material for Table SIII).The amount of N2O accumulation in CNS was not significantly different from that in RNS after 11 h of incubation(P=0.34;Fig.1c).In addition,the N2O/(N2O+N2)ratio was significantly(P<0.000 1)higher in CNS than in RNS at most sampling time points,especially after 4 h of incubation(Fig.1d).Although thecontent was adjusted to the same level in both soils,N2O accumulation and its accumulation rate in CNS remained higher than that in RNS(Fig.S2a,see Supplementary Material for Fig.S2a).N2emission in the two soils increased at a linear rate over time throughout the incubation period(Fig.S2b,see Supplementary Material for Fig.S2b).

Fig.1 Comparison of gas emissions from the soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS)during anaerobic incubation:gas kinetics of NO,N2O,and N2 of CNS(a)and RNS(b),the accumulative production of N2O using the Mann-Whitney test to test the difference(c),and denitrification product ratio N2O/(N2O+N2)using a two-way analysis of variance to test the difference between CNS and RNS(d).Vertical bars indicate standard deviations of the means(n=4).The asterisks*and****indicate significant difference at P<0.05 and P<0.000 1,respectively.ns=not significant.
Quantity of denitrifying and 16S rRNA genes
The expression of the 16S rRNA gene in CNS was significantly higher than in RNS under anaerobic incubation,indicating that the microorganisms in CNS were more active than in RNS(Fig.3).During the anaerobic incubation,the transcript numbers of thenirS,nirK,andnosZgenes,except fornarG,showed no significant difference between CNS and RNS(Fig.3).
Bacterial community diversity and composition
Approximately 686 890 high-quality reads were obtained for the reverse transcribed 16S rRNA gene sequencing from 30 samplesviaquality filtering.The sequences all achieved the requirement for further analysis(Fig.S3,see Supplementary Material for Fig.S3).All high-quality reads were grouped into 3 552 OTUs,representing 27 phyla and 334 genera.The relative abundance of the Proteobacteria in the RNS sample was higher than in the CNS sample,whereas Actinobacteria and Firmicutes showed the opposite trend(Fig.4c).At the genus level,reduced N fertilization led to the higher abundance ofPseudomonas(Fig.S4,see Supplementary Material for Fig.S4).
At the OTU level,the beta-diversity,based on the weighted UniFrac distance,showed that the active members in RNS were different from those in CNS,and the MANOVA test for the two communities showed significantly different results(Fig.4a,b).Simultaneously,the Shannon index of RNS was significantly higher(Mann-Whitney test,P<0.05)than that of CNS(Fig.4d).
Comparison of predicted denitrifying bacteria
PICRUSt was used to predict the functional gene profiles of CNS and RNS by using reverse transcribed 16S rRNA gene sequencing data.The predicted result did not show anynirS-containing bacteria,but 1 112,1 515,and 767 OTUs were associated withnirK,norB,andnosZ,respectively.

Fig.2 Concentrations of soil nitrogenous compounds ,and -N during anaerobic incubation of the soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS).Vertical bars indicate standard deviations of the means(n=3).Differences between RNS and CNS at different time points were calculated using the two-way analysis of variance test.The asterisks**,***,and****indicate significant difference at P<0.01,P<0.001,and P<0.000 1,respectively.

Fig.3 Quantity of soil bacteria gene transcripts by quantitative polymerase chain reaction measurement for narG,nirK,nirS,nosZ,and 16S rRNA genes during anaerobic incubation of the soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS).Vertical bars indicate standard deviations of the means(n=3).Differences between RNS and CNS at different time points were calculated using the Mann-Whitney test.The asterisk*indicates significant difference at P<0.05.ns=not significant.
The Bray-Curtis distance showed that the structures of the predicted active denitrifying bacterial guilds were significantly different(MANOVA test,P<0.000 1)(Figs.5b,c,S5a,b,and S6a,b,see Supplementary Material for Figs.S5a,b and S6a,b).The predominant(relative abundance higher than 1%)nirK-containing bacteria were from the generaRhodanobacter,Luteimonas,Aeromicrobium,Thermomonas,Devosia,andParvibaculum(Fig.S5c,see Supplementary Material for Fig.S5c).ThenorB-containing bacteria were fromRhodanobacter,Kaistobacter,Luteimonas,Devosia,Thermomonas,Parvibaculum,andDyella(Fig.5a).ThenosZ-containing bacteria were fromRhodanobacter,Dyella,Devosia,andOpitutus(Fig.S6c,see Supplementary Material for Fig.S6c).

Fig.4 Bacterial communities of the incubated soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS):weighted UniFrac principal coordinate(PC)analysis of soil microbiota based on the operational taxonomy unit data(a),clustering of the microbiota based on weighted UniFrac distances calculated with the multivariate analysis of variance test(b),the relative abundance of the bacterial phyla(relative abundances are shown as the percentage of the total sequences to be classified in Ribosomal Database Project classifier and the others include the phyla with a relative abundance less than 1%)(c),and the Shannon diversity index using the Mann-Whitney test to test the difference between RNS and CNS(d).The asterisks*and***indicate significant difference at P<0.05 and P<0.001,respectively.
Among all sequenced reads,285 281 reads clustered into 1 515 OTUs that were predicted asnorB-containing denitrifiers.ThenorB-containing bacteria occupied 46.1%and 48.7% of the total bacteria in CNS and RNS,respectively.Quantification of the 16S rRNA copies indicated that the CNS sample contained more active bacteria(Fig.3).There were morenorB-containing denitrifiers in CNS(6.4×1011copies g−1soil)than in RNS(5.0×1010copies g−1soil).Fortyfour OTUs were selected as key OTUs for distinguishing twonorB-containing guilds using the LEfSe method.Twentyseven of the selected OTUs were significantly more abundant in CNS(average 16.4%)than in RNS(average 9.5%)(nonparametric factorial Kruskal-Wallis sum-rank test).Most CNS enriched OTUs lacked thenosZgene and were affiliated withKaistobacter(n=5),Devosia(n=3),Rhodanobacter(n=2),Bosea(n=1),Candidatus Solibacter(n=1),Aminobacter(n=1),Parvibaculum(n=1),Denitrobacter(n=1),Dyella(n=1),andLuteimonas(n=1)(Fig.5e).In addition,among thesenosZ-lacking key OTUs,OTU592473,OTU509480,OTU1787355(Kaistobacter),OTU878059,OTU81089,OTU1139973,OTU146917(Luteimonas),and OTU741918 showed significant positive correlation with N2O/(N2O+N2).In contrast,17 othernorB-containing key OTUs were significantly more abundant in RNS(the ratios of reads for enrichednorB-containing OTUs to all bacteria reads were 10.5% in RNSvs.5.7% in CNS),of which,14 OTUs lacked thenosZgene(Table SIV,see Supplementary Material for Table SIV).These OTUs were affiliated withArenimonas(n=1),Sorangium(n=1),Pseudomonas(n=1),andThermomonas(n=1).
Approximately 153 000 reads clustering into 767 OTUs were predicted asnosZ-containing denitrifiers.Rhodanobacterwas the most dominant genus among thenosZ-containing bacteria,constituting up to 60.5% and 62.1% of this community in RNS and CNS,respectively(Fig.S6c).Using Random forest models,we identified 62 key OTUs that contributed to the discrepancy of the activenosZ-containing bacterial guilds between the CNS and RNS samples.Conventional fertilization enriched 24 OTUs(Fig.6),11 of which belonged toDevosia(n=4),Rhodanobacter(n=4),Dyella(n=1),Nitrolancea(n=1),andLuteibacter(n=1).InRhodanobacter,the relative abundance of OTU714231 and OTU67341 were significantly higher in CNS(12.8% and 9.7%)than in RNS(9.5% and 6.4%),and were positively correlated with N2O/(N2O+N2)(Figs.S7 and 6,see Supplementary Material for Fig.S7).AnotherRhodanobacterOTU(OTU1136234)was more abundant in RNS(5.9%)than in CNS(5.2%)(Fig.S7).Additional N fertilizer addition reduced the abundance of,for example,Thiobacter,Opitutus,andThauera.
DISCUSSION
Effect of content on denitrification

Fig.5 The norB-containing denitrifying bacterial guilds in the soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS):norB-containing genus composition(only that of relative abundance greater than 1% is shown)(a),Bray-Curtis principal coordinate(PC)analysis of the soil norB-containing bacterial guilds based on operational taxonomy unit(OTU)data(b),clustering of norB-containing bacterial guilds based on distances calculated using the multivariate analysis of variance test(c),key OTUs selected by LEfSe with a linear discriminant analysis(LDA)score(log10 transformation)>3.0(d),and heat map of the key OTUs,the color of the spots in the left panel representing the relative abundance(normalized and log-transformed)of the OTUs in each sample,while those in the right panel denoting the R value of the Spearman’s correlation between the OTU and denitrification product ratio N2O/(N2O+N2)(e).The asterisks*and***indicate significant difference at P<0.05 and P<0.001,respectively.“+” means P value<0.05.

Fig.6 Heat map of nosZ-containing key operational taxonomy units(OTUs)identified by random forest in the soils with conventional intensive N fertilization(CNS)and reduced N fertilization(RNS).The OTUs are clustered according to their phylogenetic positions.The colors of the spots in the left panel represent the relative abundance of nosZ-containing OTUs(normalized and log-transformed)in each sample,while those in the right panel denote the R value of the Spearman’s correlation between the OTU and denitrification product ratio N2O/(N2O+N2).The genus and family affiliations of the OTUs are shown to the right.“+”means P value<0.05.
Numerous studies have deemed that N deposition in soil,caused by overuse of N fertilizers,significantly increased N2O emission(Aerts,1997;Mohnet al.,2000).Multiple studies have proven that fertilization can change the physicochemical factors of soil(e.g.,,and,the important parameters influencing denitrifying gas emissions),especiallywhich is a key regulator in controlling denitrification rates(Davidson and Swank,1986;Thomaset al.,1994).In this study,using two soils from a long-term experimental field,we observed that conventional intense fertilization significantly increased thecontent and N2O emission rates compared with those in the reduced fertilization soil.In a previous study,soil showed higher denitrification rates after receivingwithout change in the denitrifying community(Carrino-Kykeret al.,2012).This phenomenon suggested that excess fertilizer-promoted N2O emission rates in the conventional intensive fertilized soil might relate with the supplement of morein the soil.However,our results indicated that solely increasingcontent did not directly affect N2O emission.Thepool has previously been suggested to affect instantaneous N2O emission rates through the biotic community(Wallensteinet al.,2006).Altering the composition of denitrifying bacteria may change the denitrification function.Therefore,the role ofin regulating N2O emission appears to depend on the microbial community.
Other physicochemical properties of soil have direct or indirect effects on the microbial community likewise.In this study,long-term management with increased N fertilizer significantly decreased pH of CNS.The shift of bacterial community,resulting from fertilization,in CNS may be caused by reduced soil pH.However,there is the other possibility of lower activity of nitrous oxide reductase owing to hinderance of the assembly of holoenzyme in the low pH environment(Liuet al.,2010;Van Den Heuvelet al.,2011).
Molecular investigation of denitrifying functional genes
Molecular investigation in this study showed that the transcript numbers of denitrifying functional genes(nirK,nirS,andnosZ)showed no significant difference between the CNS and RNS soils(Fig.3),implying that the size of the active denitrifying guilds had little effect on N2O emission.Numerous studies have demonstrated that the flux of N2O is correlated with the quantification of denitrifying functional genes;however,these experiments were confined to the DNA level(Wang Cet al.,2013;Liet al.,2016,2017).Few studies have focused on the relationship between N2O emission and the transcript numbers of the functional genes;whereas the transcript of N metabolism genes could be more directly connected to the function of microbiomein situ.Therefore,the investigation of denitrifying community at the RNA level as in this study is apt to reach a more reliable conclusion.
Divergent bacterial communities caused by differentiated fertilization
The 16S rRNA gene transcript data illustrated that conventional intensive fertilization significantly increased the total number of active bacteria(Fig.3),but reduced microbial diversity and significantly changed the composition of bacterial community.Approximately two-fold more urea was added to CNS than to RNS by combining with basic organic fertilizer in the field experiments.The higher N fertilizer dose may accelerate the growth rates of some specific bacteria(Wang Wet al.,2013);for example,soil CNS contained more Actinobacteria.Many Actinobacterial groups have previously been identified as copiotrophic taxa,preferring more nutrition-enrichment conditions(Iizukaet al.,1998),which consequently reduces bacterial diversity in the soil and changes the community composition.Microbial diversity may also play an important role in regulating the ecosystem process(Tilmanet al.,2014).In contrast,a diverse community is likely to contain more functional species to maintain function(e.g.,denitrification)(Braker and Conrad,2011).In our study,N2O reduction may act as the rate-limiting step for denitrification.Soil RNS may thus possess a higher efficiency for consuming N2O due to having more diverse microorganisms including some highly efficient bacterial phylogenies.
Linkage between denitrifier guilds and N2O emission
It is well known that a diverse number of soil bacteria,containing denitrifying genes,compose functional denitrifying groups,so called denitrifying guilds.Many studies have revealed that regulation of the ecosystem functions is mediated by relevant functional microbial groups(Fanget al.,2015;Liet al.,2016).In terms of the denitrification process,the flux of N2O is essentially the balance of production and consumption regulated by denitrifying guilds.Our study demonstrated a somewhat different process of denitrification between CNS and RNS,which implied a difference in activity of denitrifying genes in the two soil denitrifying guilds.Richet al.(2003)reported a strong relationship between denitrifying guilds and enzymes potentially involved in denitrification.Microorganisms account for the enzyme activity in soils(Tabatabai and Dick,2002),with the difference in microbial composition potentially altering these activities.Considering the results in this study,we may conclude that the different denitrifier composition in CNS and RNS alters the activities of denitrifying enzymes and results in different patterns of N2O emission.
Denitrifiers containingnirKandnorBare the major N2Ogenerating groups in the studied soils,on account of having nonirS-containing bacteria based on PICRUSt prediction.In this study,we observed that conventional intensive fertilization significantly changed the composition ofnirK-andnorB-containing bacteria.The relative abundance of several dominant N2O-generating bacterial genera in CNS was higher than that in RNS.Among these bacteria,many lacked thenosZgene in their genomes.For example,bacteria belonging toLuteimonas(Finkmannet al.,2000)andParvibaculumlike strains only reducedto N2O owing to a lack of nitrous oxide reductase activity(Takeuchiet al.,2015).In particular,Kaistobactermay be a key N2O-producing genus in CNS.As such,five key OTUs in this genus contribute to the structural divergence ofnorB-containing bacterial guilds that were predicted to lack thenosZgene.These bacteria might be crucial contributors to the increased production of N2O in CNS owing to their higher abundance compared with that in RNS.
Previous studies have also found that bacteria containing thenosZgene showed lower abundance than those containing other denitrifying functional genes(Hallinet al.,2009;Bruet al.,2011).Based on the results of the PICRUSt prediction,approximately half of thenorB-containing bacteria identified in this study were found to lack thenosZgene in both CNS and RNS.However,the absolute abundance of the activenosZ-lacking denitrifying bacteria in CNS was higher than in RNS(Table SV,see Supplementary Material for Table SV).Philippotet al.(2011)previously emphasized the importance ofnosZ-lacking denitrifiers in regulating N2O emission.In this study,the higher number of N2O-producing bacteria in the conventional fertilized soil shown by higher N2Oproducing gene copy numbers,but lacking thenosZgene,would in turn result in the higher net production of N2O and promote its release to the surrounding environment.
As an alternative,another bacterial guild that contains thenosZ gene could act as N2O consumers in the environment.Adjusting the bacterial community by addingnosZ-containing bacteria would be beneficial for eliminating N2O emissions.There also exists an intersection betweennorB/nirK-andnosZ-containing bacterial guilds.The members at this intersection are complete denitrification bacteria that contain all genes for producing and consuming N2O.However,many studies have reported that these bacteria could accumulate N2O under certain circumstances(Sørensenet al.,1980;Miyaharaet al.,2010).In our study,one group of complete denitrifying bacteria belonging toRhodanobacter,accounted for approximately 60% ofnosZcontaining bacteria and 35% ofnorB-containing bacteria in both the CNS and RNS samples.This implies that bacteria of this genus may play a central role in denitrification of the studied soils.Greenet al.(2012)have reported thatRhodanobactershowed extremely high relative abundance in acidicrich environments and isolated strains of this genus possess strong denitrifying activity.Six isolates from a Norwegian soil sample demonstrated that four of them possessed thenosZgene but failed to reduce N2O(Lycuset al.,2017).We found that two of theRhodanobacterOTUs were more intensively enriched under conventional fertilization and their abundance correlated positively with N2O/(N2O+N2).Different strains may also show different capacities for producing and consuming N2O(Nadeemet al.,2013;Liuet al.,2016).The differentRhodanobacterstrains displayed genetic disparity in denitrification(Hemmeet al.,2016),implying that CNS-enrichedRhodanobacterOTUs may produce N2O more efficiently than they consume it.Such an explanation provides a compelling explanation for higher N2O emission in the soil with conventional fertilization(CNS)than that with the optimized fertilization(RNS).Nevertheless,it requires additional experimental validation using isolated cultures in future studies.
Yoonet al.(2016)demonstrated that the bacteria containingnosZI possessed more efficient nitrous oxide reductase than those containingnosZII,implying that denitrifying enzymes expressed by different types of functional genes may possess distinct catalytic abilities.In our study,different fertilization regimes caused distinct genetic characteristics in the microbiome,thus creating different denitrifier guilds in the two soils.The different denitrifying gene-containing bacteria existing in CNS and RNS led to distinct denitrification processes.
Although previous studies have reported the positive response of denitrifying gene transcription to N fertilizer treatment,we observed that functional denitrifying gene transcription,in parallel with our statements,showed no significant difference between the CNS and RNS samples.Liuet al.(2010)also found that the denitrifying genes and their transcription could not explain the effect of soil pH on N2O production.In the present study,soil CNS is suggested to maintain much more efficient denitrifying enzyme species than soil RNS,owing to the long-term stimulation of high dosagein CNS.This may explain why it was unnecessary to up-regulate the transcription of denitrifying genes during short-term anaerobic incubation.In this study,we analyzed denitrifying guilds based on the PICRUSt predictions and found some interesting results,although there are some known restrictions in this method.For example,we detected thenirSgene transcription using qPCR,while the results of the functional prediction showed that nonirS-containing bacteria existed in all samples.This could be because the prediction is highly dependent on the reference genome database.
CONCLUSIONS
Overuse of N fertilizer significantly enhanced N2O emission rates and increasedcontent in soil.However,increasedcontent did not make any direct contribution to the difference in N2O kinetics between CNS and RNS during instantaneous anaerobic incubation.The analysis of active denitrifying bacteria based on the 16S rRNA gene data revealed that conventional fertilization altered the composition of bacteria containing thenorB,nirK,andnosZgenes in the community.As a result,CNS possessed several abundant N2O generating bacteria.For example,Kaistobacterenriched under conventional fertilization was predicted to containnorBgene but lacknosZgene.Being representative of thenosZ-containing group,two particular OTUs fromRhodanobacterthat also contain thenorBandnirKgenes might be involved in the increased N2O emission from CNS,probably because of the high ratio of N2O production to consumption.In summary,conventional fertilization changed the composition of active denitrifying guilds,with more N2O-producing bacteria lacking N2O-consuming genes or activity.Thus,the denitrifying function associated with N2O production may be more efficient and/or the activity of nitrous oxide reductase may be insufficient in conventionally fertilized soil.All these factors are suggested to account for the increased N2O emission observed in CNS.
ACKNOWLEDGEMENTS
We would like to thank Dr.Wenchao CAO from China Agricultural University for his kind assistance in soil sample collection,and Dr.Lars BAKKEN from Norwegian University of Life Sciences for his assistance in gas dynamics analysis.This study was supported by the National Key Research and Development Program of China(No.2017YFD0200102)and the National Natural Science Foundation of China(Nos.31670105 and 41230856).
SUPPLEMENTARY MATERIAL
Supplementary material for this article can be found in the online version.
杂志排行
Pedosphere的其它文章
- Notes to Authors
- Agricultural and environmental challenges in agroecosystems
- In Commemoration of Professor Tianren YU’s 100th Anniversary
- Microbial diversity assembled from series-diluted suspensions of diseasesuppressive soil determines pathogen invasion resistance
- Global patterns of phosphorus transformation in relation to latitude,temperature and precipitation
- Chemical and spectroscopic characteristics of humic acid from a clay loam soil in Ontario after 52 years of consistent fertilization and crop rotation
