Synthesizing the evidence of nitrous oxide mitigation practices in agroecosystems

Nitrous oxide (N 2 O) emissions from agricultural soils are the main source of atmospheric N 2 O, a potent greenhouse gas and key ozone-depleting substance. Several agricultural practices with potential to mitigate N 2 O emissions have been tested worldwide. However, to guide policymaking for reducing N 2 O emissions from agricultural soils, it is necessary to better understand the overall performance and variability of mitigation practices and identify those requiring further investigation. We performed a systematic review and a second-order meta-analysis to assess the abatement efficiency of N 2 O mitigation practices from agricultural soils. We used 27 meta-analyses including 41 effect sizes based on 1119 primary studies. Technology-driven solutions (e


Introduction
Increasing atmospheric concentrations of nitrous oxide (N 2 O) is a major driver of climate change and stratospheric ozone depletion (Ravishankara et al 2009, IPCC 2021).Agriculture is the primary anthropogenic source of N 2 O, contributing globally about 3.8 (2.5-5.8)Tg N yr −1 or 22% to the atmospheric N 2 O budget (Tian et al 2020).Direct soil N 2 O emissions from agroecosystems account for 61% of anthropogenic emissions (Tian et al 2020).Due to the projected increases in food demand (van Dijk et al 2021), agricultural land area, and fertilizer use, the associated N 2 O emissions are expected to rise over the coming decades (Reay et al 2012, Davidson and Kanter 2014).Climate change may further exacerbate N 2 O emissions from agricultural soils (Montzka et al 2011, Bowles et al 2018).This daunting picture challenges stated policy goals to curb greenhouse gas (GHG) emissions from agriculture, such as the Paris Agreement (United Nations Framework Convention on Climate Change 2015, Clark et al 2020).However, active management of agroecosystems offers opportunities for N 2 O mitigation without jeopardizing (perhaps even increasing) food production (Smith et al 2008, Snyder et al 2009).
N 2 O emission reductions are best achieved by altering the environmental factors that promote N 2 O production and release (most prominently soil aeration, soil N content, C availability, and soil pH), by inhibiting biochemical pathways during which N 2 O is produced using soil additives, or by precise N management to minimize excess soil N inputs (Paustian et al 2016).A wide range of mitigation practices for N 2 O abatement has been proposed and tested at fieldscale (Snyder et al 2014).Biochar (Dawar et al 2021), organic amendments (Li et al 2021, Ruangcharus et al 2021), conservation tillage (Arango and Rice 2021, Pelster et al 2021), and enhanced-efficiency fertilizers (Friedl et al 2020, Nishimura et al 2021), among others, have been assessed worldwide, with contrasting results.For instance, while biochar amendments frequently reduce N 2 O emissions (Schmidt et al 2021), the effects of organic amendments are equivocal (Lazcano et al 2021).
As a result of the burgeoning of laboratory and field-scale experiments, several meta-analyses synthesizing empirical results from individual mitigation practices have been published in recent years (e.g.Akiyama et al 2010, Cayuela et al 2014, Zhang et al 2020a).While meta-analyses may differ in scope, methods, and geographic scale, they are comparable when they assess the same response (i.e.soil N 2 O emission) for an intervention type (i.e.mitigation practice).However, results from meta-analyses regarding the same mitigation practice do not always confirm each other due to the variability of the experimental results and data collection procedures (e.g.databases, sample size, language, and quality criteria).In this context, combining a systematic review (i.e.high-level review using rigorous and explicit accountable research methods; Gough et al 2017) and second-order meta-analysis (i.e.metaanalysis of meta-analyses; Hunter and Schmidt 2004) provides a robust framework to integrate and synthesize cumulative evidence-based knowledge of N 2 O mitigation practices.Specifically, second-order metaanalysis allows to consolidate research quantitatively by assimilating and gauging meta-analyses grounded on diverse designs, samples, and variables for highlevel generalization (Castellanos and Verdú 2012) and to ascertain true effects (i.e.non-artifactual, by pooling the effects sizes to obtain a larger sample size that reduces uncertainties) for multiple meta-analyses, even when they report contrasting results (Schmidt andOh 2013, Tamburini et al 2020).Therefore, this tool can be used to advance our understanding of the variability and performance of N 2 O mitigation practices.The fast-growing number of studies assessing N 2 O mitigation practices and synthesis-based research allows to simultaneously compare the N 2 O mitigation potential of prominent mitigation practices, thus enabling a robust ranking of their mitigation efficiency.This information is crucial to guide policymaking to reduce GHG emissions from agricultural soils and identify mitigation measures requiring further research before implementation.
The objective of this study was to understand the effect of various mitigation practices on N 2 O emissions.Therefore, we conducted a narrative and quantitative synthesis of published meta-analyses focused on N 2 O emissions from agricultural soils.Specifically, we conducted a systematic review combined with a second-order meta-analysis of 27 meta-analyses, including 41 effect sizes (obtained from 1119 primary studies and 3700 pairwise comparisons).

Protocol and reporting
The initial protocol of the study was pre-registered on the Open Science Framework before data collection and analysis (https://doi.org/10.17605/osf.io/2fjhw).The reporting guidelines suggested in ROSES (Reporting standards for Systematic Evidence Synthesis; Haddaway et al 2018) were followed where applicable (figure S1).Data processing, analysis, and figure generation were executed using R v.4.0.3 (R Core Team 2020).

Systematic review 2.2.1. Literature search strategy
A systematic search combined with a semi-automatic co-occurrence network (Grames et al 2019) was performed to identify published meta-analyses.We used ISI Web of Science and SCOPUS databases and the search engine Google Scholar.Our search covered all articles having titles, abstracts, or keywords in English, with no restrictions on publication type.The cutoff date was May 2021.Further details of the literature search (e.g.keywords) can be found in Supplementary information S1.

Eligibility criteria
After removing duplications (figure S1), the articles were screened by title and abstract to meet the eligibility criteria.The fundamental eligibility criterion was that the study should be a meta-analysis focused on at least one mitigation practice for N 2 O emission from agroecosystems (tables S1 and S2).This step resulted in 142 articles.Additional eligibility criteria were considered during full-text screening: (a) Meta-analyses needed to assess the impact of mitigation practices on agricultural soils in peer-reviewed publications.For instance, if a meta-analysis reported effect sizes for different land uses or ecosystem types, only those referring to agricultural soils were retained (e.g.Wang et al 2021).(b) Systems comparisons were excluded if pairwise comparisons of the mitigation practice were not performed.(c) Regional and national level meta-analyses were included (e.g.Aguilera et al 2013, Gao et al 2021) due to their extensive geographical scope.However, meta-analyses with small regional scope (i.e.region within a country) were not considered (e.g.Xu et al 2017) due to limited sample size.(d) The studies had to provide clear information regarding what was considered treatment and control.(e) The effect size of the mitigation practice was calculated using formal meta-analytic methodologies.(f) The precision of the effect size had to be indicated (i.e. standard error or confidence interval).(g) Meta-analyses in which primary studies were fully considered in more recent meta-analyses were excluded (e.g.Akiyama et al 2010).(h) The list of primary studies had to be provided.Authors were contacted when the complete list of references was unavailable from the publication (e.g.Huang et al 2018).Removal of duplicates and screening of articles were performed using CADIMA (Kohl et al 2018).After this step, the database consisted of 57 articles (figure S1).

Considerations of statistical independence
The use of the same primary studies in different meta-analyses focused on the same N 2 O mitigation practice can be a source of non-independence among meta-analyses.To overcome this type of pseudoreplication, we quantified the overlap percentage between primary studies (figure S2).This resulted in one additional eligibility criterion: (i) only meta-analyses with less than 30% overlap of primary studies were retained in the database (Tamim et al 2011, Tamburini et al 2020).When the 30% percentage was exceeded, the rationale for deciding what study should be retained was that meta-analyses with a higher number of primary studies and rigorous methodology were preferred (i.e.high-quality studies according to table S2; see section 2.2.5).A total of 28 articles were considered statistically independent (27 in English and 1 in Chinese).The visualization for assessing statistical independence (figure S2) was implemented using the R-package tidygraph 1.2.0 (Pedersen 2020).

Data extraction and effect sizes
Study identification characteristics (e.g.title, author, and publication year), contextual information (e.g.mitigation practice and geographical range), methodological procedures (e.g.literature search strategy and experimental scale), and data analysis features (e.g.statistical model, effect size type, and precision indicator) were extracted from the meta-analyses.Data from text and tables were directly obtained from the study, while data from figures were extracted using Engauge Digitizer v12.1 (Mitchell et al 2019).If a study reported sub-group analysis but not the overall effect size of the N 2 O mitigation practice, we calculated the overall effect size with a meta-analytic fixedeffect model, which pools the partial effect sizes of the sub-group analyses (e.g.Xia et al 2017, Wang et al 2021).Sampling error variance was estimated using the commonly reported 95% confidence intervals (CIs) assuming a normal distribution (Castellanos and Verdú 2012).We transformed all the effect sizes (e.g.response ratio and percentage of change) to a standard metric (i.e.log response ratio-lnRR).However, we removed one study because we were unable to transform the mean difference to lnRR without the original data (Aliyu et al 2021).The sign of the effect sizes was reversed when the contrast was against the mitigation practice instead of the control treatment.The final dataset included 27 metaanalysis studies (n) encompassing 41 effect sizes (k) based on 1119 primary studies (m) with 3700 pairwise comparisons (l) (Supplementary data).
The N 2 O mitigation practices were (table S1): biochar, crop residue removal instead of crop residue retention, use of organic fertilizer instead of synthetic fertilizer, use of cover crops instead of fallow soil between cash crops, diversified crop rotation instead of crop monoculture, optimization of fertilizer rate (i.e. according to total N crop requirement) compared to conventional application rates, increased fertilizer application frequency compared to single fertilizer application, deep fertilization instead of superficial fertilizer application, slow-or controlled release fertilizer, nitrification inhibitor, urease inhibitor, combination of nitrification and urease inhibitors, lime amendment, no-tillage or reduced tillage compared to conventional tillage, and drip irrigation compared to traditional irrigation techniques such as sprinkler or furrow irrigation.The proposed mechanisms by which these practices may mitigate N 2 O emissions are summarized based on the systematic review in Supplementary information S2.

Methodologic quality of the meta-analyses
The quality of meta-analyses can differ widely due to methodological aspects by which primary studies are selected, the meta-analytic model used, and how results are reported (Nakagawa et al 2017, Gurevitch et al 2018, Pigott and Polanin 2020).To account for this potential bias, we calculated a quality index (see Tamim et al 2011, Beillouin et al 2021) based on 11 methodological criteria: (a) definition of the experimental (i.e.mitigation practice) and control treatment, (b) literature search strategy, (c) number of databases and search engines used, (d) number of original studies included, (e) eligibility criteria to select primary studies, (f) statistical model, (g) average effect size and precision indicator, (h) weighting procedure, (i) publication bias assessment, (j) availability of heterogeneity indicators, and (k) sensitivity assessment.Further description for each category is provided in Supplementary information table S2.Meta-analyses were ranked for each criterion with a high score (2) or low score (1).The theoretical maximum value is 22.The sum of the scores defined the general quality, which was used in further analysis (see section 2.3.2).

Second-order meta-analysis 2.3.1. Statistical model
The impact of the N 2 O mitigation practices was assessed with a multi-level mixed meta-regression model with a categorical predictor (i.e.moderator) using the extracted overall effect sizes and corresponding variance from every meta-analysis (i.e.not the effect sizes derived from individual pairwise observations included in each meta-analysis).The form of the statistical model was: with where θik is the estimate of the true effect size θ (intercept) based on the kth effect size of the ith metaanalysis, β is the regression coefficient (representing the effect size difference) of the categorical predictor D g (mitigation practice), w i is the random effect accounting for the variance between meta-analyses (σ 2 B ), u ik is the random effect accounting for the variance within meta-analysis (σ 2 W ) and e ik is the sampling error with variance v ik .The model was fitted without intercept to obtain the parameter estimates ( θik ) for each level of the categorical predictor.
The meta-analysis was weighted by the inverse of the sampling variance of the effect sizes.The following nested random effect was assumed: metaanalysis ID encompassing the effect sizes extracted from the same meta-analysis, and effect size ID representing the residual/within-meta-analysis variance.Model parameters were calculated using the restricted maximum likelihood estimator (Viechtbauer 2005).Estimates were transformed to percentages of change to ease interpretation.Estimates are presented with their 95% CIs in square brackets throughout, and statistical significance was assumed when CIs did not span zero.Moreover, 95% prediction intervals (PIs) were reported.CIs represent the range of the average true effect to be found, and PIs the range in which 95% of effects are expected to occur in similar future (or unknown) studies (IntHout et al 2016, Kim et al 2021).Significance tests of the estimates and the CIs were computed assuming a z-distribution.The omnibus test of moderators (Q M ) was reported.The percentage of heterogeneity explained by the moderator was estimated using R 2 marginal (Nakagawa and Schielzeth 2013).The meta-analysis was fitted using the R-package metafor v.3.0-2(Viechtbauer 2010).Results of the main effect model were graphically represented as lookalike forest graphics (i.e.orchard plots) using the R-package orchaRd v.0.0.0.9000 (Nakagawa et al 2020).

Publication bias and sensitivity analysis
Publication bias was explored with Funnel plots and Egger's Regression tests (Sterne et al 2006, Sterne andEgger 2006).We fitted multi-level mixed meta-regression models, including the standard error and sample size as moderators separately.The potential presence of bias was identified based on the significant deviation of the model intercept from zero.Similarly, influential studies were identified based on the leverage (i.e.hat values) extracted from the hat matrix and potential outliers based on the standardized residuals (Viechtbauer 2020).Studies with large influence were those with two times the average leverage (Habeck and Schultz 2015), whereas possible outliers were those studies with high standardized residuals.Meta-regressions including continuous moderators were plotted with the R-package ggplot2 v.3.3.5 (Wickham 2016).
We used two different approaches as a sensitivity analysis to test the robustness of the multi-level mixed meta-regression model.We used a conservative estimation with the Knapp-Hartung adjustment (Knapp andHartung 2003, van Aert andJackson 2019) based on a t-distribution.This adjustment controls for the uncertainty in the estimate of betweenstudy heterogeneity affecting the calculation of the standard error, the hypothesis tests, and CI using the Satterthwaite adjustment.To incorporate the quality of the meta-analyses, a quality effects model was fitted (Doi et al 2015).Consequently, the weights of the effect sizes were compensated by the rescaled quality index (based on the highest index) obtained for each meta-analysis, thereby reducing the weight of lowquality studies.

Systematic review
We found that 46% of the effect sizes showed significant N 2 O reductions across the mitigation practices, while 34% showed neutral responses and 20% significant increases in N 2 O emissions (figure 1).Among all the mitigation practices, the use of organic fertilizer (k = 6 effect sizes), nitrification inhibitor (k = 5), biochar amendment (k = 4), and crop residue removal (k = 4) were the most frequently assessed.Nitrification inhibitors (k = 5), biochar amendment (k = 3), and slow-or controlled release fertilizer (k = 3) reported a high number of effect sizes with significant N 2 O reductions compared to other potential mitigation practices.However, the use of nitrification inhibitors was the only practice consistently showing N 2 O reductions across all metaanalyses (figure 1).Less explored (e.g.optimization of fertilizer rate or high fertilizer application frequency) and more recent meta-analyzed N 2 O mitigation practices (e.g.lime amendment and drip irrigation) resulted in single effect sizes (figure 1).Most of the metaanalyses had a global scope (70%), followed by the country scale (China with 16%) (figure S4(A)).The overall quality score of the meta-analyses was relatively high, ranging from 14 to 22 (figure S4(B)), with a median of 18 (theoretical maximum value = 22).
There was no clear evidence of publication bias for the second-order meta-analysis.Based on the standard error fit (intercept = −0.207,p = 0.001; R 2 marginal = 0.088), effect sizes tended to become marginally smaller as standard error increased (figure S5(A)).Nevertheless, there was no evidence of publication bias derived from the sample size fit (intercept = −0.061,p = 0.386; R 2 marginal = 0.046); effect sizes tended to become marginally smaller as sample size (i.e.number of pairwise comparisons) increased (figure S5(B)).Influential studies were only identified for mitigation practices based on a single metaanalysis (figure S6); an effect size with high standardized residual (belonging to diversified crop rotation under paddy soil conditions) was identified (figure S6).The two approaches of the sensitivity analysis showed that our results are robust, as all the different models yielded similar effect sizes to our primary model (table S3).

Discussion
Several technology-driven solutions showed substantial N 2 O mitigation potentials across meta-analyses (figure 2).These included biochar, slow-or controlled release fertilizers, nitrification inhibitors, and urease inhibitors.Drip irrigation and the combination of both inhibitor types, also reduced N 2 O emissions, based on one effect size each.The main goal behind the development of these technologies was originally not to mitigate N 2 O emissions.Biochar was valued for its capacity to retain C in the long term, thereby potentially increasing soil C sequestration (Wu et al 2019).Enhanced-efficiency fertilizers were developed to achieve better synchronicity between N release and crop uptake, thereby increasing uptake efficiency while simultaneously reducing nitrate leaching (slow-or controlled release fertilizer and nitrification inhibitors) or ammonia (NH 3 ) volatilization (urease inhibitors) (Akiyama et al 2010, Timilsena et al 2015, Li et al 2018) As opposed to the technology-driven options, agroecological practices tended to increase N 2 O emissions (figure 2).Examples of these practices are the use of organic fertilizer, diversifying crop rotations, reduced/no-tillage, and the use of cover crops.Despite the potential adverse impact on N 2 O, these practices (not primarily conceived to abate N 2 O emissions) are linked to a wide range of beneficial effects, including enhanced soil biodiversity (Liu et  Therefore, a unidimensional view focused on reducing soil N 2 O emissions does not capture the multifunctional benefits of agroecological interventions (Guenet et al 2021).
Most agroecological practices had a highly variable effect on N 2 O emissions, highlighting the need to better understand under which pedoclimatic and management conditions such practices may lead to N 2 O mitigation instead of stimulation.For instance, the N 2 O mitigation of organic amendments is primarily determined by their physicochemical characteristics and N-fertilizer substitution rate (Ren et al 2017, Zhang et al 2020a).Most of the available meta-analyses focused on solid manure, whereas only one evaluated the effect of replacing synthetic fertilizers with livestock slurry with regard to soil N 2 O emissions (Aguilera et al 2013).Crop diversification may reduce N 2 O emissions depending on the specific crops within the rotation and fertilization schemes (Ijaz et al 2019).However, understanding crop diversification effects is challenging due to crop-specific confounding variables (e.g.timing and rate of fertilizer application, different rooting depths) and methodological constraints (e.g.continuous N 2 O measurements over multiple years are required).These reasons may explain the limited number of effect sizes and corresponding primary sources for this management option (figure 1).Even though conservation tillage practices (i.e.reduced/no-tillage) may increase N 2 O emissions, they could potentially decrease N 2 O emissions in dry areas over time (van Kessel et al 2013, Mei et al 2018).The effect of cover crops on N 2 O emissions varies with cover crop species (e.g.legume and non-legume), termination date, and soil incorporation (Basche et al 2014).It is necessary to update meta-analytic results with the burgeoning primary studies considering the temporal (including non-growing season) and geographical effects on N 2 O release, especially for diversified crop rotation and cover crops.
While optimized fertilizer rate according to crop needs showed important N 2 O reductions, increasing the frequency of fertilizer applications, crop residue removal, and lime application reduced N 2 O emissions only marginally.This may be because the N 2 O mitigation potential of these practices is highly context-dependent.Crop residue removal may only reduce N 2 O emissions when the residues are immature and have a low C/N ratio (Chen et (Davidson and Kanter 2014), and it can be combined with technology-driven solutions and agroecological practices.However, predicting crop N need is difficult due to variable soil and weather conditions, with increasing variability of environmental conditions induced by climate change (Hénault et al 2012, Reay et al 2012, Kanter et al 2016).Therefore, our results imply that these practices must be assessed on a case-by-case basis and at farm or regional scale before recommending their adoption for N 2 O mitigation.
Several Although agriculture is a crucial sector for the reduction of anthropogenic GHG emissions (Wollenberg et al 2016, Frank et al 2018, Tian et al 2019, 2020, IPCC 2021), only 131 countries (covering 72% of global GHG emissions) are discussing, have announced, or have adopted net-zero targets (Höhne et al 2021).One of the reasons limiting international commitments may be the perceived potential tradeoffs between GHG mitigation and food production (Frank et al 2017).However, several N 2 O mitigation practices do not compromise biomass or food production.Many of them tend to enhance crop yields (see table S4 for references), including biochar (9%-28% increase in yield) in tropical regions predominantly, deep fertilization (4%-11%), drip irrigation (12%), optimization of fertilizer rate (1%), increased frequency of fertilizer application (6%), lime amendment (36%), nitrification inhibitors (4%-10%), urease inhibitors (5%-10%), and the combined use of inhibitors (1%-9%).The use of slow-or controlled released fertilizers has uncertain effects on crop production, whereas yield decline can sometimes be observed with high substitution rate of synthetic fertilizer by organic sources (no effect to 14% reduction), reduced/no-tillage (no effect to 6% reduction), and crop residue removal (reduction 5%-8%).There is no consensus on the impact of cover crops on yield, while diversified crop rotation may have a positive effect (20%) (table S4).Efforts to reduce N 2 O emissions from agricultural soils could simultaneously improve food security.Therefore, they should represent a priority in policy agendas, providing a tool to overcome barriers to implementation (Snyder et al 2009, Kanter et al 2020).
A systematic review allows to identify knowledge gaps.In addition, a second-order meta-analysis can advance our understanding of differences and similarities among N 2 O mitigation practices by aggregating results across numerous meta-analyses grounded on hundreds of studies with thousands of pairwise comparisons.This approach can gauge the true between-meta-analyses variability of mean effect size values, and use this information to improve estimation accuracy for each first-order meta-analytic mean estimate (Schmidt and Oh 2013).For certain management practices, our second order meta-analysis synthesizes a relatively low number of original metaanalyses, which in some cases are based on relatively small datasets.However, we contend that also in these cases, a second-order meta-analysis provides important advantages over first-order meta-analyses.First, by combining data from more than one metaanalysis, we increase statistical power.This is especially important in those cases when the number of available studies is relatively low.Second, our standardized approach ensures that overall treatment effects are directly comparable and are not affected by artifacts related to differences in methodological approaches between individual first-order meta-analyses.Yet, certain shortcomings can arise from this generalization.Due to the nature of the method and the specific factors controlling each N 2 O mitigation strategy, further moderator analyses (e.g.temporal dynamics, experimental scale, management practices, pedoclimatic features, and land use) are unfeasible.This limits the capacity of this methodology to provide region-specific recommendations as effects may depend on pedoclimatic conditions or field management, which can be better identified in first-order meta-analyses.To complement our results and guide the implementation of the N 2 O mitigation practices, the assessment of yield-scaled emissions (e.g.van Groenigen et al 2010) would be valuable.
Our synthesis exposes critical research gaps to be filled in future studies.Due to the nature of meta-analyses, the N 2 O reductions are mainly shown for individual mitigation practices, but many of the examined practices may be combined at the field scale, leading to synergistic or antagonistic effects on N 2 O emissions (e.g.

Conclusions
We present a second-order meta-analysis of the effects of management practices on N 2 O emissions from agricultural soils.Despite the intrinsic variability of N 2 O mitigation practices, technology-driven solutions (enhanced-efficiency fertilizers, biochar, and drip irrigation) and fertilizer rate optimization may substantially reduce emissions in agroecosystems.These practices may favorably be implemented because they often increase crop production.On the contrary, certain agroecological practices (e.g.use of organic fertilizer) may exacerbate N 2 O release if they are not carefully managed.Our exhaustive evidence synthesis provides a state-of-the-art overview of the potential for N 2 O abatement of the main available mitigation strategies.

Figure 1 .
Figure 1.The number and direction of effect sizes for every N2O mitigation practice included in our synthesis.kindicates the number of effect sizes, m the number of primary studies, and l the number of pairwise al 2016, Venter et al 2016, Chen et al 2020, Kim et al 2020), lower weed infestation (Osipitan et al 2019), increased nutrient retention (McDaniel et al 2014, Chen et al 2018, Wei et al 2021), reduced water pollution (Thapa et al 2018), reduced soil erosion (Sun et al 2015), and other ecosystem services (Iverson et al 2014, Lichtenberg et al 2017).From a GHG balance perspective, reduced/no-tillage may increase soil organic carbon (SOC), although this topic remains widely debated (e.g.Baker et al 2007, Powlson et al 2014, Bai et al 2019).Cover crops may increase soil C storage by 6%-16% (Bai et al 2019, Jian et al 2020).

Figure 2 .
Figure 2.Effect of mitigation practices on N2O emissions.Orchard plot of the meta-analytic model showing mean estimates with black circles, 95% confidence interval with thick whisker, 95% prediction interval with thin whisker, and individual effect sizes scaled by their precision (i.e.inverse of the standard error) with grey circles.kindicates the number of effect sizes.
Fuertes-Mendizábal et al 2019, Pokharel and Chang 2021).There is a need to explore and document such interactions, including the tradeoffs and long-term effects of potential mitigation practices for designing cropping systems to minimize N emissions while enhancing production without compromising soil functions or services (Power 2010, Bommarco et al 2013, Greiner et al 2017, Bünemann et al 2018).Novel strategies with possible N 2 O mitigation capacity (and their interactions with more assessed mitigation practices) have not been sufficiently studied to be meta-analytically studied.Some examples are intercropping (Pappa et al 2011, Huang et al 2019), biological nitrification and denitrification inhibitors (Subbarao et al 2009, Bardon et al 2014), silicate additions (Vicca et al 2021), and inoculants for legumes (Bakken and Frostegård 2020).
. Drip irrigation is widely used as an irrigation practice for optimizing water supply to high-value crops (e.g.vegetables, grain crops; Vallejo et al 2014, Zhang et al 2020b, Qasim et al 2021), thus increasing crop water use efficiency as compared to flood or sprinkler irrigation (van der Kooij et al 2013); 20% of cropland worldwide is irrigated, contributing to 40% of the world food production (The United Nations World Water Development 2014).Although there are potential tradeoffs linked to certain technologydriven options that need to be considered (e.g.nitrification inhibitors may increase NH 3 volatilization; Li et al 2005, Pan et al 2016, Wu et al 2021), our results suggest that these management practices can achieve substantial N 2 O reductions from agricultural soils.
policies and regulations aimed at curbing agricultural N 2 O and other GHG emissions have been launched recently, most of which require drastic emission reductions in the near future (Rogelj et al 2016, Clark et al 2020).For instance, Ireland O mitigation at the field level may strongly contribute to achieve these mitigation targets despite the variability.Furthermore, since N 2 O emissions can dominate the GHG balance of agricultural soils (Li et al 2005, Lugato et al 2018, Autret et al 2019), failing to incorporate N 2 O mitigation practices into environmental initiatives (e.g.carbon farming practices; Tang et al 2016, Oldfield et al 2022) may hinder efforts to obtain GHG emissions reductions.Policy efforts should address the economic and social constraints limiting the adoption of mitigation practices (Smith et al 2007).