Elevated carbon dioxide stimulates nitrous oxide emission in agricultural soils:A global meta-analysis

Yilin DUXinyu GUOJinxing LIYuankun LIUJipeng LUOYongchao LIANG and Tingqiang LI∗

1Ministryof Education KeyLaboratoryof Environmental Remediation and Ecological Health,College of Environmental and Resource Sciences,Zhejiang University,Hangzhou 310058(China)

2Zhejiang Provincial KeyLaboratoryof Agricultural Resources and Environment,Hangzhou 310058(China)

3National Demonstration Center for Experimental Environment and Resources Education,Zhejiang University,Hangzhou 310058(China)

ABSTRACT Elevated carbon dioxide(CO2)(eCO2)has been shown to affect the nitrous oxide(N2O)emission from terrestrial ecosystems by altering the interaction of plants,soils,and microorganisms.However,the impact of eCO2 on the N2O emission from agricultural soils remains poorly understood.This meta-analysis summarizes the effect of eCO2 on N2O emission in agricultural ecosystems and soil physiochemical and biological characteristics using 50 publications selected.The eCO2 effect values,which equal to the percentage changes of N2O emission under eCO2,were calculated based on the natural logarithm of the response ratio to eCO2.We found that eCO2 significantly increased N2O emission(by 44%),which varied depending on experimental conditions,agricultural practices,and soil properties.In addition,eCO2 significantly increased soil water-filled pore space(by 6%),dissolved organic carbon content(by 11%),and nitrate nitrogen content(by 13%),but significantly reduced soil pH(by 1%).Moreover,eCO2 significantly increased soil microbial biomass carbon(by 28%)and soil microbial biomass nitrogen(by 7%)contents.Additionally,eCO2 significantly increased the abundances of ammonia-oxidizing bacteria(AOB)amoA(by 21%),nirK(by 15%),and nirS(by 15%),but did not affect the abundances of ammonia-oxidizing archaea(AOA)amoA and nosZ.Our findings indicate that eCO2 substantially stimulates N2O emission in agroecosystems and highlight that optimization of nitrogen management and agronomic options might suppress this stimulation and aid in reducing greenhouse effect.

KeyWords:agricultural practices,agroecosystems,climate change,experimental conditions,greenhouse gas,soil properties

INTRODUCTION

Nitrous oxide(N2O)is a powerful greenhouse gas with 265-fold greater global warming potential than carbon dioxide(CO2)(Liet al.,2020).After undergoing photolysis and oxidation to nitric oxide(NO),N2O can lead to ozone depletion in the stratosphere(McTaggartet al.,1997;Ravishankaraet al.,2009),thereby advancing global climate change,with various ramifications on human health(Ollivieret al.,2011).Agricultural soils are estimated to account for more than 50% of total atmospheric N2O emission(Sunet al.,2018)and result in significant losses of nitrogen from agricultural systems(Coskunet al.,2017).

The global climate has undergone profound changes due to anthropogenic disturbances.Since the 1750s,CO2levels in the atmosphere have gradually increased by 40%(Lamet al.,2012).Scientists have estimated that CO2levels will reach 550 mg L−1by 2050(Lobell and Asner,2003).In terrestrial ecosystems,elevated CO2(eCO2)can promote plant productivity(Leakeyet al.,2009),enhance plant tolerance to environmental stresses(Huang and Xu,2015),change soil microbial ecology(Drigoet al.,2009),and even reshape the global climate(Liuet al.,2018).Importantly,eCO2has been shown to affect N2O emission from agricultural soils by changing the interaction between plants,soils,and microorganisms(Liuet al.,2018;Jinet al.,2020).Therefore,a full understanding of N2O emission from agricultural soils under eCO2would provide a theoretical basis for predicting future climate change.

Previous studies have reported positive(Qiuet al.,2019),negative(Kettunenet al.,2007),and no changes at all(Usyskin-Tonneet al.,2020)in agricultural soil N2O emission,in response to eCO2.Several factors account for these differences.For instance,N2O emission may be affected by changes in experimental types(Chenget al.,2006;Bhattacharyyaet al.,2016),eCO2concentration(Reganet al.,2011;Cantarelet al.,2012),and experimental duration(Kammannet al.,2008;Abbasi and Müller,2011).Moreover,plant type,cropping system,nitrogen fertilization rate,and additive application may affect N2O emission.In particular,the incorporation of external additives such as biochar,nitrification inhibitor(NI),and straw often plays a critical role in mitigating the effect of eCO2on N2O emission(Donget al.,2009;Liet al.,2016;Sunet al.,2018).Importantly,different soil texture(Kettunenet al.,2006;Kanervaet al.,2007)and soil pH(Baggset al.,2003;Baggs and Blum,2004)often result in distinct N2O emission responses to eCO2.

Notably,N2O emission has been shown to be highly related to the microbial activity of the soil nitrogen turnover process.Functional gene markers for nitrifiers and denitrifiers are widely used to determine the changes in the microbial populations responsible for N2O emission.For example,theamoAgene,which encodes ammonia monooxygenase,mediates the process of ammonia oxidation(Leiningeret al.,2006;Ouyanget al.,2018).ThenirKandnirSgenes,which encode nitrite reductase,enforce the reduction of nitrite to NO(Henryet al.,2004;Kuyperset al.,2018).ThenosZgene,which encodes N2O reductase,catalyzes N2O reduction to N2(Philippot and Hallin,2005;Henryet al.,2006).However,the effect of eCO2on functional genes that regulate N2O emission in soil have not yet been quantitatively synthesized,inhibiting the elucidation of the microbial mechanisms that underlie the effect of eCO2on N2O emission.

To date,several meta-analyses have evaluated N2O emission from forests,wetlands,and upland ecosystems under eCO2conditions(Barnardet al.,2005;van Groenigenet al.,2011;Lamet al.,2012;Lianget al.,2016;Liuet al.,2018).However,the underlying mechanisms for N2O emission from agricultural soils remain unclear.Agricultural activities,such as fertilization,irrigation,and tillage,significantly alter the characteristics of agricultural soils from those of other soils.Therefore,it is necessary to compile all studies on the impact of eCO2on N2O emission from agricultural soils.

To fill the paucity in knowledge on how eCO2affects N2O emission in agricultural soil,we compiled observational data points of N2O emission rates and soil physiochemical and biological properties to investigate the responses of N2O emission to eCO2in agroecosystems.We aimed to test the following hypotheses:i)eCO2can stimulate N2O emission from agricultural soils,ii)the effect of eCO2on N2O emission can be influenced by experimental conditions,agricultural practices,and soil properties,and iii)eCO2will affect soil physiochemical and biological characteristics,thus affecting N2O emission.

DATA SELECTION AND ANALYSES

Peer-reviewed literature that reported the responses of N2O emission to eCO2and was published before May 2020 was collected from the Web of Science and China National Knowledge Infrastructure(CNKI).The key search terms were“elevated CO2”,“increased CO2”,and“N2O”.A total of 422 publications were initially collected and further refined according to the following criteria:i)the experiment was conducted in agricultural soil,ii)the effect of N2O emission was measured,iii)the climate,experimental conditions,agricultural practices,and soil were the same between the eCO2and ambient CO2(aCO2,control)treatments to avoid confounding factors,iv)eCO2concentration and experimental duration were clearly reported,and v)the means,standard deviations,and sample sizes of the selected variables were reported.Based on the above criteria,102 observational data points for N2O emission from 50 publications were included(Table SI,see Supplementary Material for Table SI).

Subsequently,we also compiled the responses to eCO2in these 50 publications of soil physiochemical properties(soil water-filled pore space(WFPS),pH,organic carbon(SOC),dissolved organic carbon(DOC),ammonium nitrogen(),nitrate nitrogen()and soil biological properties(soil microbial biomass carbon(MBC),soil microbial biomass nitrogen(MBN),and abundances of the ammonia-oxidizing archaea(AOA)amoA,ammoniaoxidizing bacteria(AOB)amoA,nirK,nirS,andnosZgenes.In total,782 observational data points(N2O emission rates and soil physiochemical and biological properties)from 50 published papers were included in our meta-analysis(Table SI).We extracted the eCO2and control treatment means,standard deviations,and sample sizes of each data point from the included studies.GetData Graph Digitizer(v2.0)was used to extract data presented in figures.

From each selected study,we collected:i)experimental conditions,including location(longitude and latitude),mean annual temperature(MAT),mean annual precipitation(MAP),experimental type,eCO2concentration,and experimental duration,ii)agricultural practices,including land use type,cropping system,nitrogen fertilization rate,and additive application,and iii)soil properties,including soil texture,bulk density,pH,SOC,and total nitrogen(TN).The geographical locations of the studies involving responses of N2O emission rates and soil properties to eCO2spanned between 40°14′S and 60°49′N,with the MAT ranging from 9.3 to 28.0°C and the MAP from 302 to 948 mm(Table SI).Categorical variables were further sorted as follows.The experimental types were grouped into field and pot.The eCO2concentrations were grouped into＜550,550—650,and＞650 mg L−1.The experimental duration was grouped into＜1,1—5,and＞5 years.Land use types were grouped into paddy field,dry land,and grassland.The cropping systems were grouped into monoculture and rotation.Nitrogen fertilization rates were grouped into＜100,100—200,and＞200 kg N ha−1year−1.Additive application was grouped into no additive,NI(nitrapyrin),straw,and biochar.Soil texture was grouped into clay,clay loam,loam,and sand.Soil bulk density was grouped into＜1.4 and＞1.4 g cm−3.Soil pH was grouped into＜6.5,6.5—7.5,and＞7.5.Soil organic carbon was grouped into＜10,10—20,and＞20 g kg−1.Soil TN rates were grouped into＜1,1—2,and＞2 g kg−1.In addition,according to the Köppen-Geiger climate classification system(Kotteket al.,2006),climate zones in these studies were extracted and classified into subtropical,temperate,and boreal.

Using a box-plot,we compared the annual N2O emission rates across different climate zones,under aCO2(control)and eCO2.Subsequently,the natural logarithm of the response ratio(lnRR)was used as the effect size,to assess the responses of annual N2O emission rates and soil properties to eCO2:

wheresTandsCare the standard deviations of variables for the eCO2and control treatments,respectively,andnTandnCare the sample sizes of variables for the eCO2and control treatments,respectively.The weighted effect sizes(lnRR++)and their 95% confidence intervals(CIs)were calculated using Hedges-Olkin random model(with the inverse variance of each effect size as relative weights)in OpenMEE(Wallaceet al.,2017).The significance of the effect size was assessed with the 95% CIs.If the 95% CIs did not overlap zero,the effect of eCO2was considered significant;otherwise,the effect of eCO2was considered to be non-significant.For a better explanation,lnRR++was transformed into a percentage(eCO2effect,%),to represent the effect of eCO2,using the following formula:

The eCO2effect values equal to the percentage changes of N2O emission under eCO2;the positive values indicate the increases in N2O emission under eCO2relative to aCO2,and the negative values indicate the decreases.

The categorical random effect model was used to examine the effect of eCO2on the variables of different climate zones,experimental conditions,agricultural practices,and soil properties.Between-group heterogeneity(Qb)was calculated to determine the effects of categorical variables.A significantQbvalue suggests that the effect of the categorical variable was significant.In addition,linear regression was used to explore the relationships between effect size on N2O emission and MAT,MAP,and changes in corresponding soil properties.All the figures were obtained using the ggplot2 package(Wickham,2016)in R 3.6.1(R Development Core Team,2016).

RESULTS

Responses of N2O emission to eCO2 in different climate zones

The overall N2O emission rates from agricultural soils were 3.20 and 4.67 kg N ha−1year−1under aCO2and eCO2,respectively.Under both CO2conditions,the N2O emission rates were significantly lower(P＜0.05)in the subtropical zone than in the boreal zone(Fig.1a).Moreover,eCO2significantly increased N2O emission(eCO2effect=44%with 95%CI of 24%—66%)(Fig.1b).There were significant differences(P＜0.01)in N2O emission in response to eCO2among the different climate zones(Table I).Under eCO2,the highest increase in N2O emission was found in the temperate zone(eCO2effect=60% with 95% CI of 29%—98%),followed by the boreal zone(eCO2effect=25%with 95%CI of 14%—37%),but there were no significant changes in N2O emission in the subtropical zone(Fig.1b).

Fig.1 Rates of N2O emission from agricultural soils under ambient CO2(aCO2)and elevated CO2(eCO2)in different climate zones(a)and eCO2 effect values(i.e.,the percentage changes of N2O emission under eCO2),calculated using Eq.3 based on the natural logarithm of the response ratio(Eq.2),on N2O emission from agricultural soils in different climate zones(b).In a,the dots indicate the outliers,the horizontal lines within box indicate the medians,the boxes and whiskers represent the ranges and 95%confidence intervals(CIs),respectively,and different letters indicate significant differences at P＜0.05.In b,the error bars represent the 95%CIs.

TABLE I Between-group heterogeneity(Qb)of elevated CO2(eCO2)effect on N2O emission from agricultural soils across all variables

Changes in N2O emission under eCO2 as affected by experimental conditions

The effect of eCO2on N2O emission decreased significantly(P＜0.000 1)with MAT(R2=0.31)across all studies(Fig.2a).In addition,the responses of N2O emission showed a significant downward-convex relationship(P=0.03)with MAP and reached the highest value at MAP of approximately 600 mm(R2=0.13)(Fig.2b).

The N2O emission significantly increased(P＜0.01)under eCO2compared with that under aCO2(Table I).The average eCO2effect values were 117% with 95% CI of 44%—226%and 42%with 95%CI of 23%—63%at eCO2of＜550 and 550—650 mg L−1,respectively.Meanwhile,no significant changes appeared in N2O emission at eCO2of＞650 mg L−1.Additionally,the N2O emission responses to eCO2depended significantly(P＜0.01)on experimental duration(Table I).The N2O emission increased consistently for the experimental duration of 1 year(eCO2effect=47%with 95%CI of 16%—85%)and more than 5 years(eCO2effect=71%with 95%CI of 32%—123%),but not for the 1—5 year experimental duration with 95%CI of−6%to 37%(Fig.2c).The responses of N2O emission to eCO2change significantly(P＜0.01)with the experimental duration,but not(P=0.76)with experimental type(Table I).However,the field eCO2treatments showed a significant positive effect(eCO2effect=47%)(Fig.2c).

Fig.2 Relationships of the effect sizes(i.e.,the natural logarithm of the response ratio(lnRR)calculated using Eq.2)of elevated CO2(eCO2)on N2O emission from agricultural soils with mean annual temperature(MAT)(a)and mean annual precipitation(MAP)(b)and eCO2 effect values(i.e.,the percentage changes of N2O emission under eCO2),calculated using Eq.3 based on the lnRR,on N2O emission under different experimental conditions.In a and b,the shaded areas represents the 95%confidence intervals(CIs),and the solid bubbles indicate the relative weights based on linear regression analysis(the larger the solid bubble,the greater the weight).In c,the error bars represent the 95%CIs.

Changes in N2O emission under eCO2 as affected by agricultural practices

Different additives resulted in significant differences(P＜0.001)in the response of N2O emission to eCO2(Table I).The N2O emission increased under eCO2with no additive and NI,with eCO2effect values of 51%(95%CI of 30%—75%)and 57%(95%CI of 22%—102%),respectively(Fig.3).Meanwhile,straw incorporation did not significantly alter N2O emission,while biochar application consistently suppressed N2O emission(eCO2effect=65%with 95%CI of−83%to−31%).The response of N2O emission to eCO2did not change significantly(P=0.10)with land use type(Table I),but the eCO2treatments had consistently significantly positive effects on N2O emission for dry land(eCO2effect=46% with 95% CI of 26%—70%)and grassland(eCO2effect=58%with 95%CI of 25%—100%)(Fig.3).Similarly,N2O emission did not change significantly(P=0.20)with nitrogen fertilization rate(Table I),but low(＜100 kg N ha−1year−1)and moderate(100—200 kg N ha−1year−1)nitrogen fertilization rates significantly stimulated N2O emission,with eCO2effect values of 82%(95%CI of 35%—144%)and 34%(95%CI of 23%—46%),respectively(Fig.3).Additionally,the N2O emission responses were consistently positive under all cropping systems(Fig.3),and no significant effect of cropping system(P=0.41)was found(Table I).

Fig.3 Elevated CO2(eCO2)effect values(i.e.,the percentage changes of N2O emission under eCO2),calculated using Eq.3 based on the natural logarithm of the response ratio(Eq.2),on N2O emission from agricultural soils under different agricultural practices.Error bars represent the 95%confidence intervals.NI=nitrification inhibitor(nitrapyrin).

Changes in N2O emission under eCO2 as affected by soil physical properties

The N2O emission showed a significant downwardconvex relationship(P＜0.01)with soil WFPS and reached the highest value when the effect size on soil WFPS(R2=0.37)was approximately 0.2(Fig.4a).Furthermore,changes in soil texture resulted in significant differences(P＜0.001)in N2O emission response to eCO2(Table I).The N2O emission from the clay,clay loam,and loam soils increased under eCO2,with the eCO2effect values being 73%(95%CI of 44%—108%),57%(95%CI of 22%—102%),and 62%(95% CI of 19%—121%),respectively.In contrast,eCO2did not significantly alter the N2O emission from the sandy soils with eCO2effect 95%CI of−24%to 0.6%(Fig.4b).Likewise,changes in soil bulk density also produced significant differences(P＜0.001)in N2O emission(Table I).In brief,soils with bulk density lower than 1.4 g cm−3(eCO2effect=72%with 95%CI of 39%—111%)showed greater increases in N2O emission in response to eCO2than those with bulk density greater than 1.4 g cm−3(eCO2effect=21%with 95%CI of 17%—25%)(Fig.4b).

Fig.4 Relationships between the effect sizes(i.e.,the natural logarithm of the response ratio(lnRR)calculated using Eq.2)of elevated CO2(eCO2)on N2O emission from agricultural soils and those on soil water filled pore space(WFPS)(a)and the eCO2 effect values(i.e.,the percentage changes of N2O emission under eCO2),calculated using Eq.3 based on the lnRR,on N2O emission under different soil physical properties(b).In a,the shaded areas represent the 95%confidence intervals(CIs),and the solid bubbles indicate the relative weights based on linear regression analysis(the larger the solid bubble,the greater the weight).In b,the error bars represent the 95%CIs.

Changes in N2O emission under eCO2 as affected by soil chemical properties

The effect sizes of eCO2on N2O emission decreased significantly(P＜0.01)with the effect sizes of eCO2on soil NH+4-N content(R2=0.14)(Fig.5a).Additionally,the effect sizes of eCO2on N2O emission showed a significant downward-convex relationship(P＜0.001)with the effect sizes of eCO2on soilcontent and reached the highest value when the effect size oncontent(R2=0.27)was about 0.25(Fig.5b).

Different soil TN contents produced significant differences(P＜0.000 1)in the response of N2O emission to eCO2(Table I).The greatest increase in N2O emission was detected in the soils with low TN content(＜1 g kg−1)(eCO2effect=138%with 95%CI of 58%—259%),followed by the soils with moderate TN content(1—2 g kg−1)(eCO2effect=37%with 95%CI of 14%—66%),but eCO2did not significantly alter N2O emission from the soils with high TN content(＞2 g kg−1)(eCO2effect 95%CI of−12%to 92%)(Fig.5c).Soil pH did not significantly affect(P=0.31)N2O emission under eCO2(Table I).The eCO2treatments in the acid soils(pH＜6.5)(eCO2effect=70%with 95%CI of 30%—124%)and the alkaline soils(pH＞7.5)(eCO2effect=48%with 95%CI of 18%—86%)consistently produced significantly positive effects(Fig.5c).Moreover,eCO2also significantly increased N2O emission under all SOC contents(Fig.5c),and the SOC content had no significant effect(P=0.41)on the response of N2O emission to eCO2(Table I).

Fig.5 Relationships of the effect sizes(i.e.,the natural logarithm of the response ratio(lnRR)calculated using Eq.2)of elevated CO2(eCO2)on N2O emission from agricultural soils with those on soil content(a)and content(b)and the eCO2 effect values(i.e.,the percentage changes of N2O emission under eCO2),calculated using Eq.3 based on the lnRR,on N2O emission under different soil chemical properties(c).In a and b,the shaded areas represent the 95%confidence intervals(CIs),and the solid bubbles indicate the relative weights based on linear regression analysis(the larger the solid bubble,the greater the weight).In c,the error bars represent the 95%CIs.SOC=soil organic C;TN=total N.

Changes in soil properties under eCO2

Across all studies,eCO2did not significantly affect SOC and soil,but significantly increased soil WFPS,DOC,and,with the eCO2effect values being 6%(95%CI of 3%—9%),11%(95%CI of 1%—21%),and 13%(95% CI of 4%—22%),respectively(Fig.6).Meanwhile,eCO2significantly reduced soil pH,with the eCO2effect value being−1%(95%CI of−2%to−0.1%).In addition,eCO2significantly increased soil MBC and MBN,with the eCO2effect values being 28%(95%CI of 16%—40%)and 7%(95%CI of 0.1%—13%),respectively.Under eCO2,the abundances of AOBamoA,nirK,andnirSalso significantly increased with the eCO2effect values being 21%(95%CI of 9%—34%),15%(95%CI of 2%—30%),and 15%(95%CI of 2%—29%),respectively,while the abundances of AOAamoAandnosZdid not change significantly.

Fig.6 Elevated CO2(eCO2)effect values(i.e.,the percentage changes of N2O emission under eCO2),calculated using Eq.3 based on the natural logarithm of the response ratio(Eq.2),on soil physiochemical(a)and biological(b)properties of agricultural soils.WFPS=water-filled pore space;SOC=soil organic C;DOC=dissolved organic C;MBC=microbial biomass C;MBN=microbial biomass N;AOA and AOB AamoA=abundances of the ammonia-oxidizing archaea and bacteria amoA gene,respectively;AnirK,AnirS,and AnosZ=abundances of nirK,nirS,and nosZ genes,respectively.The error bars represent the 95%confidence intervals.

DISCUSSION

N2O emission is stimulated byeCO2

On average,eCO2significantly increased(Fig.1b)N2O emission by 44%(Fig.7),much higher than the increase(27%)found in a previous meta-analysis based on grain crop and legume pasture systems(Lamet al.,2012).This confirmed that eCO2could increase N2O emission from agricultural soils,consistent with other terrestrial ecosystems(Lianget al.,2016).

Fig.7 Schematic diagram illustrating the elevated CO2(eCO2)effects(i.e.,the percentage changes of N2O emission under eCO2),calculated using Eq.3 based on the natural logarithm of the response ratio(Eq.2),on N2O emission from agricultural soils.The upward and downward arrows represent that the eCO2 effects(values listed in parentheses)are stimulatory and inhibitory,respectively.Blue and purple lines with arrows represent the feedback mechanisms of eCO2 and nitrogen fertilization,respectively.Yellow lines with arrows represent soil processes.Red lines with arrows represent emission pathways of N2O.The asterisks∗,∗∗,and∗∗∗indicate significant differences at P＜0.05,P＜0.01,and P＜0.001,respectively.NS=non-significant;SOM=soil organic matter;WUE=water use efficiency;WFPS=water-filled pore space;DOC=dissolved organic C;MBN=microbial biomass C;MBN=microbial biomass N;AOA and AOB AamoA=abundances of the ammonia-oxidizing archaea and bacteria amoA gene,respectively;AnirK,AnirS,and AnosZ=abundances of nirK,nirS,and nosZ genes,respectively.

Regardless of the eCO2concentration,N2O emission rates were higher in the cold(boreal and temperate)zones than in the warm(subtropical)zones(Fig.1a),in agreement with an earlier meta-analysis(Wanget al.,2018)which reported that denitrification is primarily limited by the extremely high air temperature and low precipitation in warm zones.Air temperature and moisture are known to be the dominant controllers of nitrification and denitrification at a global scale(Saad and Conrad,1993;Bouwmanet al.,2013).Therefore,the N2O emission rates in the subtropical zone declined because of high air temperature and low precipitation.Nevertheless,the difference between the temperate and subtropical zones was not significant,and more field studies under different climate zones are needed to verify if the N2O emission rates decline under high air temperature in the subtropical zone.

There are several possible mechanisms for the positive eCO2effect on N2O emission from agricultural soils,according to our results(Fig.7).First,soil WFPS significantly increased under eCO2(Fig.6a),which is consistent with the results of a previous meta-analysis(van Groenigenet al.,2011).In addition,soil WFPS also showed a significant positive relationship with N2O emission(Fig.4a).Under eCO2,the water use efficiencyviaphotosynthesis in plant improves and thus plant transpiration rates reduce,thereby increasing soil moisture,which in turn stimulates denitrification and associated N2O emission(Longet al.,2004;Butterbach-Bahl and Dannenmann,2011).These findings imply that soil water content increases under eCO2,thereby stimulating N2O emission(Barnardet al.,2005;Liuet al.,2018).

The second mechanism for the positive eCO2effect on N2O emission from agricultural soils was that soil DOC,MBC,MBN,andsignificantly increased(Fig.6)by 11%,28%,7%,and 13%,respectively,under eCO2(Fig.7).Elevated CO2promotes plant photosynthesis and,in turn,increases rhizodeposition(Kimet al.,2001;Royet al.,2012).Consequently,soil labile DOC pool and MBC increased significantly because of the accumulation of rhizodeposits(Pendallet al.,2004;Bhattacharyyaet al.,2013).Meanwhile,rhizodeposition can also increase soil MBN by strengthening microbial metabolisms(Niklaus,1998;Rakshitet al.,2012).Furthermore,it has been proven that eCO2stimulates nitrogen mineralization(Mülleret al.,2009;Ollivieret al.,2011),consistent with the increase in soilfound in the present meta-analysis.All labile carbon and nitrogen pools play an important role in nitrogen cycling in soil.Specifically,increasing labile carbon and nitrogen pools may leads to increased soil microbial activity,which ultimately stimulates nitrification,denitrification,and N2O emission(Singhet al.,2010;Andersonet al.,2011).

Third,soil pH significantly decreased(Fig.6a)under eCO2by 1%(Fig.7);in other words,eCO2led to soil acidification.This can be explained by the fact that eCO2significantly promotes plant growth(Yanget al.,2008)and the secretion of root exudates,such as citric acid,lactic acid,and formic acid(Chenget al.,2010).On the other hand,increasing labile carbon substances would improve the microbial activity under eCO2,which would then facilitate the decomposition of organic matter to acidic substances(Phillipset al.,2011;Xuet al.,2019).Both the root exudates and decomposition products of organic matter contribute to soil acidification.At lower soil pH,the production of N2O reductase that reduces N2O to N2may be prevented(Liuet al.,2010;Bakkenet al.,2012),which is in line with the reduction,although not significant,ofnosZgene abundance in our results(Fig.6b).One implication of our findings is that eCO2may decrease N2O reductase activity by decreasing soil pH,which could be a mechanism involved in the eCO2-associated stimulation of N2O emission.

Fourth,eCO2remarkably increased the abundances of AOBamoA,nirK,andnirS(Fig.6b)by 21%,15%,and 15%,respectively(Fig.7).As mentioned above,eCO2increases labile carbon and nitrogen pools,which in turn stimulates the growth of the microorganisms involved in soil nitrogen cycling.Thus,the abundances of the nitrifiers and denitrifiers that mediate nitrification and denitrification processes are expected to increase.In addition,the effect size of eCO2on N2O emission rate was negatively correlated with the effect size of eCO2on soil,but positively correlated with the effect size of eCO2on soil(Fig.5).These phenomena provide more evidence for the stimulation of nitrifiers and denitrifiers in soil by eCO2.Interestingly,our findings showed that the abundance of AOBamoAsignificantly increased in response to eCO2,though not significant(Fig.6b),supporting the theory of niche differentiation between these two communities in soil(Erguderet al.,2009;Prosser and Nicol,2012).Accordingly,higher N2O emission from agricultural soils was attributed to the enhanced abundances of AOBamoA,nirK,andnirS,induced by eCO2(Cantarelet al.,2012;Qiuet al.,2019).

Effect of eCO2 on N2O emission varies across experimental conditions

Overall,the effect of eCO2on N2O emission differed significantly among climate zones(Table I).The climate conditions in the boreal and temperate zones may be beneficial for N2O emission,compared to those in the subtropical zone,demonstrating that the microbes involved in N2O emission are limited by the high temperature of the subtropical zone.For example,N2O emission decreased in the eCO2treatments with MAT higher than 20°C(Fig.2a).A similar trend in the N2O emission response to nitrogen fertilizer was reported by Wanget al.(2018),with the largest increase in soil N2O emission in temperate regions,followed by arid regions.In contrast,eCO2always increased the N2O emission where MAP ranged from 302 to 948 mm(Fig.2b).Therefore,MAT may be more crucial to N2O emission response to eCO2than MAP.In addition,such informational syntheses for the tropical zone are lacking,owing to the small amount of data.

The eCO2concentration and experimental duration had significant effects on N2O emission in response to eCO2(Table I).At concentrations less than 650 mg L−1,eCO2significantly stimulated N2O emission,but did not alter N2O emission when the concentration was＞650 mg L−1(Fig.2c).The inconspicuous effect of eCO2at concentrations＞650 mg L−1may be attributed to the fact that higher eCO2could down-regulate plant photosynthesis,reduce the enhancement of rhizodeposition and microbial activity involved in the nitrogen cycle,and thus reduce N2O emission compared to lower eCO2(Longet al.,2004;Tenget al.,2006).Interestingly,N2O emission under eCO2consistently increased when the experimental duration was either less than 1 year or more than 5 years,but not for the experimental duration of 1—5 years(Fig.2c).Nitrogen fertilization is inevitable in agricultural soils.A previous meta-analysis summarized that nitrogen cycling functional microorganisms have different sensitivities to experimental duration under nitrogen fertilization,as indicated by statistics showing discrepancies in the responses of the abundances ofnifH,nosZ,and AOBamoAto experimental duration under fertilization(Ouyanget al.,2018).Consequently,N2O emission may be influenced by different related microbial activities during the experimental period.This mechanism could underlie the observed null effect of eCO2on N2O emission when the experimental duration was 1—5 years.Moreover,eCO2significantly increased N2O emission in field studies,but did not change in pot studies(Fig.2c).However,pot experiments usually involved higher eCO2levels(＞650 mg L−1)than field experiments(Table SI).As illustrated above,N2O emission could be suppressed by higher eCO2by affecting microbial nitrification and denitrification,which are the primary sources of N2O emission from soil(Joneset al.,2014;Donget al.,2020).This concept may explain the various responses of N2O emission to eCO2in pot studies.

Effect of eCO2 on N2O emission varies across different agricultural practices

Our analysis showed that eCO2significantly increased N2O emission from the dry land and grassland,but had no effect for the paddy fields(Fig.3).Anoxic soil conditions in paddy fields always favor denitrifiers,resulting in higher N2O emission from the paddy fields than from dry land and grassland(Longet al.,2004;Shanget al.,2020).Thus,in this case,even though eCO2had a positive effect on N2O emission from the paddy fields,the change was not significant.In addition,eCO2significantly increased N2O emission both under crop monoculture and rotation(Fig.3),but the difference between them was not significant(Table I).This agrees with the results of Omonodeet al.(2011),who reported that crop rotation had no effect on N2O emission in general.Unfortunately,there is no consensus on how cropping systems affect N2O emission from soil.Empirical evidence has shown that crop rotation has the potential to increase N2O emission by multiplying the nitrogen amount needed to sustain crop yield,which intensifies nitrogen cycling(Behnkeet al.,2020).However,the meta-analysis of Meiet al.(2018)has shown that crop rotation systems significantly reduce N2O emission compared to no-rotation systems.Altogether,the effect of crop rotation on N2O emission varies depending on irrigation,management of crop residues,and soil mineral nitrogen content(Malhi and Lemke,2007;Tellez-Rioet al.,2017).Therefore,further field experiments are necessary to understand how N2O emission is influenced by crop rotation.

Unexpectedly,eCO2generally stimulated N2O emission under the nitrogen fertilization rates of＜200 kg N ha−1year−1,but N2O emission showed a non-significant increase under the nitrogen fertilization rates of＞200 kg N ha−1year−1(Fig.3).The results implied that global warming caused by eCO2did not always increase with the nitrogen fertilization rate.Similarly,two meta-analyses examined the effect of nitrogen fertilization on N2O emission and found a significant decrease in N2O emission at high nitrogen fertilization rates(Shcherbaket al.,2014;Wanget al.,2018).One possible interpretation is that saturation ininduces nitrogen leaching and runoff,thereby limiting N2O emission(Gallowayet al.,2003;Krameret al.,2006).In addition,the Michaelis-Menten kinetics can also be used to characterize the hyperbolic responses of N2O emission to nitrogen fertilization(Kimet al.,2013;Zhu and Riley,2015).

Additive applications showed obviously different effects on N2O emission response to eCO2(Table I).First,the NI application did not mitigate the eCO2-stimulated N2O emission response in the collected studies(Fig.3).Nevertheless,recently published meta-analyses have reported that the NIs dicyandiamide(DCD)and 3,4-dimethylepyrazole phosphate(DMPP)largely reduce N2O emission(Yanget al.,2016;Liuet al.,2017;Roseet al.,2018).The contradictory results may be due to the different concentrations of the available substrates.The relatively highcontent in soil could boost plant-soil interaction and create a hotspot for nitrification potential,thereby reducing the efficiency of the inhibitors(Gerardset al.,1998;Lópezet al.,2003).Another possible factor is the NI variety.Nitrapyrin,selected as the NI in this meta-analysis,is a low water-soluble and weakly polar substance and has a great affinity for humic acid in soil(Jacinthe and Pichtel,1992).Thus,nitrapyrin had no obvious effect on N2O emission.Second,straw incorporation inhibited the stimulation effect of eCO2on N2O emission(Fig.3),congruent with the results of Xiaet al.(2018),who reported a significant reduction in N2O emission with straw incorporation in paddy fields.It is likely that straw decomposition in paddy fields enhances the immobilization of MBN,the consumption of O2,and the availability of DOC in soil,which favors the further reduction of N2O to N2(Aulakhet al.,2001).In contrast,straw incorporation significantly increased N2O emission from the upland soils,as reported by Aulakhet al.(2001)and Liuet al.(2014).This is attributed to the additional nitrogen substrate provided by straw input for nitrification and denitrification,which stimulate N2O emission from the upland soils(Davidsonet al.,2000;Chenet al.,2013).Third,our results suggested that biochar application significantly decreased eCO2-stimulated N2O emission(Fig.3).This result corresponds with the conclusions of previous meta-analyses(Hagemannet al.,2017;Liuet al.,2019;Xiaoet al.,2019),which suggested that biochar plays a crucial role in reducing the N2O emission from agricultural soils.The main potential mechanisms are the increase in soil pH,the adsorption of mineral nitrogen,the release of toxic compounds into soil,and the changes in nitrifiers or denitrifiers(Cayuelaet al.,2014;Harteret al.,2014;Sunet al.,2018).Nevertheless,owing to the limited observational data for the mitigation potential of agronomic additives on eCO2-stimulated N2O emission,we suggest that more relevant studies are warranted to verify the combined effect of eCO2and agronomic additives on N2O emission.

Effect of eCO2 on N2O emission varies across different soil properties

Soil texture and bulk density induced significant differences in the response of N2O emission to eCO2(Table I).Elevated CO2always stimulated N2O emission from the clay,clay loam,and loam soils,but had no significant effect for the sandy soils(Fig.4b).This discrepancy may be attributed to nitrogen limitations.Extremely low nitrogen was added in the studies that examined sandy soils.Thus,although carbon availability increased,eCO2did not lead to greater N2O emission.In addition,plant biomass was also unaffected by eCO2,and this may be,in part,conducive to explaining why these studies concluded that eCO2did not affect N2O emission(Rämöet al.,2006;Kanervaet al.,2007).In addition,under eCO2,low soil bulk density was shown to induce a significantly greater increase in N2O emission than high soil bulk density(Fig.4b).This was in line with the results of a recently published meta-analysis(Wanget al.,2018),which indicated a negative correlation between the effect of nitrogen fertilization on N2O emission and soil bulk density in agricultural soils.Soil O2availability is negatively correlated with microbial denitrification,and N2O emission from soils with lower bulk density appeared to be more responsive(Tiedjeet al.,1989).

The responses of N2O emission to eCO2were marginally affected by soil pH and SOC(Table I,Fig.5c).Similarly,soil pH had a marginal effect on the response of N2O emission to nitrogen fertilization,as also reported by Wanget al.(2018).This suggests that different nitrifiers and denitrifiers may affect N2O emission at different soil pH values(Ouyanget al.,2018).Furthermore,contrary to our results for SOC,Meiet al.(2018)discovered a larger increase in N2O emission from soils with high SOC than with low SOC.This difference may be the result of rhizodeposition.As illustrated above,eCO2can stimulate rhizodeposition(Kimet al.,2001;Royet al.,2012)and thus increase soil labile carbon pool,which is crucial for the microbial activity involved in soil nitrogen cycling and N2O emission.Therefore,the initial SOC might be a predictor for the N2O emission induced by eCO2.Interestingly,soil TN significantly influenced N2O emission under eCO2.With the increase in soil TN content,the stimulation of N2O emission induced by eCO2gradually weakened(Table I,Fig.5c).High soil TN provides more nitrogen substrate for nitrogen cycling(Khalsaet al.,2020),and the responses of N2O emission may be marginal even though eCO2has the potential to accelerate nitrogen transformation and stimulate N2O emission.

CONCLUSIONS

Our global meta-analysis evaluated the effect of eCO2on N2O emission from agricultural soils and soil physicochemical and biological characteristics.We found that eCO2significantly stimulated N2O emission,indicating that eCO2might be an essential indicator for predicting global climate change.Specifically,eCO2improved the water use efficiency of plant photosynthesis and thus increased soil moisture,promoting denitrification and associated N2O emission.Moreover,the rhizodeposition of labile carbon pool was likely to increase,which may result in the stimulation of microbial nitrogen cycling under eCO2.On the other hand,the increased root exudates such as citric acid and formic acid also increased the likelihood of soil acidification.Ultimately,because of all these responses of soil physiochemical properties,the activities and abundances of nitrifiers and denitrifiers could be indirectly increased,inducing greater N2O release.More importantly,experimental conditions,agricultural practices,and soil properties were important factors regulating N2O emission responses to eCO2.Collectively,our findings indicate that intensive nitrogen fertilization alleviates progressive nitrogen limitation due to eCO2and as a result eCO2has a substantially positive effect on N2O emission in agroecosystems.Nevertheless,the stimulation of N2O emission under eCO2may be suppressed by optimizing nitrogen management and agronomic options,such as straw and biochar incorporation,thereby reducing greenhouse effect.

ACKNOWLEDGEMENT

This work was jointly supported by grants from the National Key Research and Development Program of China(No.2017YFD0200102)and the Fundamental Research Fund for the Central Universities,China(No.2020FZZX001-06).

SUPPLEMENTARY MATERIAL

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

CONTRIBUTION OFAUTHORS

Yilin DU and Xinyu GUO contributed equally to this work.