Tillage effects on growing season nitrous oxide emissions in Canadian cropland soils

Abstract Minimizing tillage has been promoted as an agricultural practice that may mitigate greenhouse gas emissions through carbon sequestration. However, there is some ambiguity regarding the effect of minimum tillage (MT) on emissions of other greenhouse gases, in particular soil nitrous oxide (N2O) emissions. To determine how effective MT could be in helping Canada mitigate greenhouse gas emissions, we used a meta-analysis to compare growing season N2O emissions from MT versus conventional tillage (CT). Overall, MT had 12% lower N2O emissions compared to CT (P = 0.03). However, there was high variability due to soil texture and growing season precipitation (GSP), with MT tending to emit more N2O than CT in climates where GSP exceeded 600 mm, particularly for soils with sand content less than 60%. Therefore, unless long-term tillage trials, which are urgently needed in eastern Canada, show a reduction in N2O emissions over time, MT should be used as a greenhouse gas mitigation measure only in dry climates or on sandy soils.


Introduction
In Canada, agricultural soils emit approximately 25 Mt CO 2 equivalent of greenhouse gases (GHG), or about 3.4% of Canada's total GHG emissions (Environment and Climate Change Canada 2020).However, agricultural soils can also help mitigate climate change by sequestering carbon (C), with CO 2 removals by cropland soils estimated to be 6.2 Mt CO 2 year −1 (Environment and Climate Change Canada 2020).One method to increase soil C sequestration in agricultural soils is to change from conventional tillage (CT) to minimum tillage (MT) techniques, including zone tillage, reduced tillage, or notillage systems (West and Post 2002).However, the effects of MT practices on soil nitrous oxide (N 2 O) emissions are inconsistent, with some meta-analyses indicating decreased N 2 O emissions with MT (Li et al. 2023;Yue et al. 2023), while others indicate increased emissions (Mei et al. 2018;Shakoor et al. 2021;Yangjin et al. 2021).
In Canada, there are large differences in the effect of MT on N 2 O emissions between eastern and western Canada (VandenBygaart et al. 2003).Across the Prairie provinces, MT has been practiced for over 40 years because it reduces soil erosion, improves water retention, and increases profitability (Zentner et al. 1992).No-till on the prairies was also found to help mitigate climate change as its use with continuous cropping was found to increase soil C by approximately 0.27 ± 0.19 Mg C ha −1 year −1 (Liebig et al. 2005).As a result, more than 65% of total land prepared for seeding on the Canadian Prairies uses MT techniques (Statistics Canada 2019).There are few recent studies from the Canadian Prairies that compare N 2 O emissions from MT with CT; however, earlier studies do suggest that MT in this region does not increase N 2 O emissions (Lemke et al. 1999;Malhi et al. 2006;Malhi and Lemke 2007).
In eastern Canada, a larger proportion of farmers continue to use CT as there are still questions regarding the potential of MT for improved profitability and improved climate mitigation.Likewise, questions remain regarding the ability of MT to sequester C (VandenBygaart et al. 2003;Powlson et al. 2014;Ogle et al. 2019), while several studies have noted increased N 2 O emissions from MT in the region (Gregorich et al. 2008;Mkhabela et al. 2008), although soil texture and drainage do seem to play an important role (Rochette et al. 2008;Pelster et al. 2021).Indeed, Rochette (2008) proposed that MT increased N 2 O emissions only in poorly drained soils.In addition, there are some indications that N 2 O emissions in MT decrease over time, often becoming equal to emissions from CT systems after about 10 years of the practice (Six et al. 2004).
To address the uncertainty of how MT affects soil N 2 O emissions in Canadian agricultural systems, we compiled a database of existing Canadian studies that compared growing season N 2 O emissions from CT versus MT.Our objective was to determine how a change to MT would affect soil N 2 O Fig. 1.Flowchart of study selection process.emissions in Canadian croplands and how the effect may vary under different soil and meteorological conditions and through time.Our hypotheses were that MT would have no effect on emissions on the Canadian Prairies, and that shortterm (<10 years) implementation of MT would increase N 2 O emissions on fine-textured soils in eastern Canada.We also hypothesized that over time, the N 2 O emissions in MT would approach emissions in CT, even in poorly drained soils.

Database generation
The data used in this study were collected from peerreviewed papers measuring the N 2 O emissions under MT compared to CT.The following tillage practices were included under MT: no-till, reduced tillage, and zone tillage.The search engines SCOPUS, Agricola, Biological Abstracts, and CAB Abstract were used with a cut-off date of April 2023.The search keywords ("soil" OR "soils") AND (tillage OR "no tillage" OR "no-till" OR "zero tillage" OR "zero till" OR "reduced tillage" OR "zone tillage" OR "conservation tillage") AND ("nitrous oxide" OR "N 2 O" OR "nitrogen oxide") AND (emission OR release) AND (Canada OR AFFILCOUNTRY(Canada)) AND NOT(Model * ) were used to search the databases for peerreviewed articles.The identified records were first screened using the title and abstract and then assessed for eligibility using the full text.A flow chart of the study selection process is shown in Fig. 1.
The inclusion criteria were (a) studies conducted in Canada; (b) research studies conducted in the field; (c) studies measuring growing season cumulative emissions (kg ha −1 ) of N 2 O under both MT and CT at the same study site; (d) means, standard deviations (or standard errors), and the number of replications that were either reported and/or could be easily calculated; and (e) experimental duration and soil texture were clearly reported.There was only one study that mea-sured emissions year-round (Wagner-Riddle et al. 2007), but we used only the growing season emissions from this study so that it was consistent with the rest of the studies.Reviews, meta-analyses, commentaries, editorials, and modelling studies were excluded.Following the above criteria, a total of 14 studies with a total of 81 paired observations (i.e., comparisons of MT to CT), reflecting the multiple treatments and years included within individual peer-reviewed publications, were selected (Table 1).The locations of the study sites are shown in Fig. 2. Eight of the studies were on the mixedwood plains in eastern Canada, where precipitation rates can be as high as 1200 mm year −1 .Four studies were on the Prairies in western Canada where precipitation can be as low as 250 mm year −1 (see http://www.ecozones.ca/english/zone/index.html for more details).Of the other two studies, one was on the Boreal Plains (precipitation of about 450 mm year −1 ) and the other on the Boreal Shield (up to 1000 mm year −1 ; Fig. 2).Data were mostly extracted from text and tables.Otherwise, the plotdigitizer web platform (https://plotdigitizer.com, 2023) was used to extract data from figures.
The following data were extracted for each study: country, province, latitude, longitude, tillage history, year of the trial, growing season precipitation, soil texture, sand content, silt content, clay content, texture class, pH, soil organic carbon, crop species, synthetic nitrogen fertilizer rate, organic nitrogen fertilizer rate, and total nitrogen fertilizer rate.Our preliminary analysis identified tillage history, growing season precipitation, sand content, and crop as the most interesting data to use in the meta-analysis as the others did not show any significant effect or contained too much missing data.

Analysis
The effect of MT on growing season soil N 2 O emissions was compared with CT by comparing the natural logarithmic response ratio (lnRR) as the effect size (Shakoor et al. 2021; where X m is the mean cumulative N 2 O emissions from an MT treatment and X c is the mean cumulative N 2 O emission from a CT treatment.The variance of lnRR was calculated using eq. 2.
where SD m and SD c are the standard deviations and Nm and Nc are the number of replicates of the MT and the CT treatments, respectively.In the case where the standard deviations were not included in the article, various methods were used to calculate the SD depending on the data provided (Eqs.

3-6).
Where a yearly coefficient of variation was provided, the standard deviation was calculated using eq .3: where N 2 O is the cumulative soil-derived N 2 O emission measured in the study for a given year, N is the number of cumulative N 2 O emissions, and CV is the coefficient of variation in percent.When the mean of a treatment and a signifi-cance probability level was provided, the standard deviation was calculated using eq.4: where X m and X c represent the cumulative soil-derived N 2 O emissions of the MT and CT treatments, respectively.τ is the bilateral inverse student law, α is the significance probability level, ν is the number of degrees of freedom, and N m and N c are the number of replicates of the MT and CT treatments, respectively.When a significant letter only was provided, a significance probability level of 0.05 was used in eq. 4. If a standard error was available for each year, the standard deviation was calculated using eq.5: where SE is the standard error given in the study and N m and N c are the number of replicates of the MT and CT treatments, respectively.When studies contained the standard error for each measurement, the standard deviation was calculated using eq.6: where SE is the standard error and N is the number of replicates of the measurement.
Missing sand content was calculated using either silt and clay contents and/or the soil texture provided, and the average sand content of the texture class was used for the metaanalysis.Missing growing season precipitations were completed using data from an environment Canada weather station located near the experiment site.
To interpret and visualize the data, the effect size as the percentage change was calculated using eq.7: Equal-effects models were used in the paper, and the weights given to the estimates are equal to eq. 8: where vi is the sampling variance of the ith study.This is called "inverse-variance weighting" and is an efficient way of weighting the estimates.The Wagner-Riddle et al. ( 2007) study used a micrometeorological measurement device, preventing spatial replicates.To assess this issue, we considered that this study contained a single replicate, which increased its sampling variance and consequently decreased its weight in the meta-analysis as a result.
All data were analyzed using the R software (R Core Team 2021).A mixed-effects model meta-analysis using the "rma.mv"function from R package "metafor" was used (Viechtbauer 2010).Since some studies produced several observations, the studies and observations were considered as random factors in the mixed-effects models.The results were considered significant if the 95% confidence interval did not overlap with zero and if the test for overall effect resulted in P < 0.05.Multiple linear regression was also used to investigate how environmental and management variables affected the lnRR using SigmaStat v.4.0.

Overall effect of minimum tillage on N 2 O emissions
The overall effect of MT on N 2 O emissions is shown in Fig. 3 along with the effect of each study used in the meta-analysis.Although only one study showed a significant effect of MT (i.e., Drury et al. 2012), the overall effect revealed a weak but statistically significant reduction (P = 0.05) in N 2 O emissions with the implementation of MT in Canadian soils.

Reduces emission
Favours emission

Environmental effects
Soil pH and organic carbon content had no detectable effect on the growing season N 2 O emission response ratio comparing CT to MT.However, the effect of MT on N 2 O emissions was influenced by the combination of soil texture, particularly the sand content, and growing season precipitation (Fig. 4).MT lowered N 2 O emissions on study sites with growing season precipitation less than 600 mm.Conversely, at sites with high precipitation (>600 mm) and in particular those sites with sand content <60%, the conversion to MT seems to lead to increased emissions (Fig. 4).However, due to the low number of studies, the confidence intervals for these effects are quite large.The lack of studies is particularly noticeable for sites where MT has been implemented on fine-textured soils for more than 10 years and for which there are only two data points for all of Canada (Fig. 5).
A parametric multivariate regression analysis indicated a significant relation with the linear combination of sand content (%) and growing season precipitation (mm) as predictors explaining 35.1% of the variation (adjusted R 2 ) (eq. 8) as follows: lnRR = − 0.538 + (0.00138 • GSP) The interaction term was not significant, each term retained in this model was significant at P < 0.05, and the overall model had P < 0.001, n = 46.The effects of sand content and growing season precipitation are also evident in Fig. 6, where all comparisons with growing season precipitation that exceeded 600 mm and where sand content was lower than 30% had an effect size (lnRR-%) greater than 0, in-dicating that MT increased N 2 O emissions under these conditions.

Management effects
Approximately 85% of the studies noted across Canada were trials conducted for less than 10 years, with most of the trials having more than 10 years of tillage history located on moderately textured soils (i.e., sand content between 30% and 60%--Fig.5).Due to the lack of long-term trials, particularly on fine-and coarse-textured soils, we were unable to evaluate the effect of duration of MT on the N 2 O response ratio.Total, synthetic and organic N application rates also had no detectable effect on the response ratio.Likewise, crop type had no significant effect on the response ratio between MT and CT (Fig. 7), though wheat tended to yield reduced emissions when used with MT compared to CT.Interestingly, soybean, a crop known to receive no nitrogen fertilization, had an effect of zero and almost symmetrical confidence intervals ranging from −0.32 to 0.33.This could suggest the absence of any consistent impact of tillage management options on N 2 O emissions under soybean.

Discussion
There have been a few previous global meta-analyses that have looked at how tillage affects N 2 O emissions in agriculture (Huang et al. 2018;Mei et al. 2018;Shakoor et al. 2021).However, their conclusions are inconsistent, with indications that regional climate may influence the results.Our study, which is the first meta-analysis to evaluate CT versus MT specifically in Canada, showed an overall slight reduction in growing season N 2 O emissions.Similar to the global analyses though, we too found that there are differences  Minimum-tillage systems are known to contribute to increased aggregate stability and mean weight diameter of aggregates (Mondal and Chakraborty 2022), while they can also reduce macroporosity and saturated conductivity, which in turn improves water retention (Singh et al. 1998;Mondal and Chakraborty 2022).Denitrification, which tends to be the dominant source of N 2 O in managed soils, requires anaero- ).Consequently, denitrification is strongly correlated with gas diffusivity (Balaine et al. 2013;Chamindu Deepagoda et al. 2019) and therefore soil water-filled pore space.Thus, under a humid climate and on fine-textured soils, the increased water retention with MT may lead to more anaerobic microsites, and in turn, increased N 2 O emissions.This is consistent with Rochette (2008), who also noted that no-till only increased N 2 O emissions when it was done on poorly drained sites.However, the mean response ratio in our current study was approximately 1.21, as opposed to the 1.50 noted in Rochette (2008), who did not constrain their study to just Canadian sites.While this could indicate differences in the response ratios between the two Fig. 7. Forest plot of the association between crops and nitrogen oxide emission response ratio between reduced or no-till and conventional tillage.regions covered in the different studies (i.e., North America vs. Canada only), there are large, overlapping confidence intervals around these means (Fig. 4), suggesting that these differences may not be significant.
Our results are also similar to a prior study included in the database used for the present meta-analysis, and that compared MT with CT in a humid climate on two contrasting soil textures and which found that N 2 O emissions were greater from the MT than the CT on the fine-textured soil, but were similar on the coarse-textured soil (Pelster et al. 2021).In the previous study, the higher emissions from the MT on the finetextured soils were thought to be due to slightly higher water content under MT compared with CT--causing higher denitrification rates during the rains that occurred shortly after N fertilization (Pelster et al. 2021).
Many of the beneficial effects of MT, however, do not appear immediately, with one study stating that newly converted MT systems require more than 10 years before seeing reductions in global warming potential (Six et al. 2004).This reduction in emissions over time is likely due to an increase in macroporosity after more than 10 years (Mondal and Chakraborty 2022), which may be related to increased earthworm populations that develop in MT systems (Clapperton et al. 1997;Tebrügge and Düring 1999) and can increase the connectivity of the macropore structure.Unfortunately, there are too few long-term tillage trials in Canada that measure N 2 O, resulting in a major data gap for Canada's mitigation options.
The lack of any effect on the response ratio between MT and CT for the other management variables (i.e., crop type and N application rate) was unexpected.We had expected that higher N application rates, would result in a larger response ratio because the higher moisture in the MT would lead to more anaerobic microsites, which combined with the increased supply of N due to higher application rates, would increase denitrification.However, the application rate is dependent on the crop type, and cumulative N 2 O emissions tend to be more strongly correlated to the N surplus (i.e., N applied − N removed in the crop) than the applied N (Eagle et al. 2020).So, higher N application rates may not create higher emissions if they were applied to crops with a higher N demand.Also, even though the absolute difference in emissions between practices may be greater with high N fertilizer rates, the response ratio of MT to CT may remain similar to those studies with low application rates.
Increased protection of microaggregates in MT compared to CT leads to greater C sequestration rates in many soils (Six et al. 1999(Six et al. , 2000)), which in fine-textured soils under a humid climate may compensate for the higher N 2 O emissions noted in the current study.However, a previous study looking at no-till on agricultural soils in Canada found that although no-till tended to increase soil C on the Canadian Prairies, it often resulted in a loss of soil C in eastern Canada, particularly on fine-textured soils (Liang et al. 2020).However, they also found that soil C increased with MT in the drier regions of Canada (e.g., the prairie region) and on medium-textured soils under the humid climate of eastern Canada (Liang et al. 2020).
The focus of this study was on how changing tillage type affects growing season N 2 O emissions.However, this focus ignores emissions that occur during the non-growing season, which in Canada average about 36% of total annual emissions with no strong differences between treatments (Pelster et al. 2022).Previous studies have noted that MT in combination with other best management practices such as optimizing N fertilizer rate or leaving crop residues on-site can reduce non-growing season emissions (Wagner-Riddle et al. 2007;Congreves et al. 2017).The large reductions in N 2 O emissions with MT during the non-growing season were related to the stubble capturing greater amounts of snow that, together with the crop residues on the soil surface, better insulated the soil causing fewer and less intense freeze-thaw events (Wagner-Riddle et al. 2007).However, both of these studies occurred at the same location in Southern Ontario, where soil freeze-thaw events can have a large effect on annual emissions (Abalos et al. 2016).Whether these effects occur in other regions of Canada, for example in eastern Canada with a much greater snowpack (van Bochove et al. 2001) or on the Canadian Prairies with lesser snowfall, is unclear, and further studies that measure emissions throughout the entire year at other sites encompassing a wider range of pedo-climatic conditions are sorely needed.

Conclusions
This study shows that while MT tends to reduce emissions overall, it should not be promoted as a climate mitigation strategy for fine-textured soils under a humid climate (e.g., eastern Canada) in the short term as it seems to increase N 2 O emissions, while having no beneficial effect on C sequestration.The long-term benefits, however, are unclear and need to be investigated before discarding this practice completely for the region.Unfortunately, the only studies on finetextured soils with high precipitation were geographically concentrated in Quebec, thus making additional studies from Ontario and the Maritime provinces a priority.In western Canada, and on well-drained soils in eastern Canada, MT does not appear to increase soil N 2 O emissions from annual croplands, which combined with the increased soil C measured in previous studies should help mitigate climate change.In sum, MT should not be promoted on fine-textured soils in eastern Canada as a potential mitigation measure; however, MT can be used in western Canada and on coarse-textured soils in eastern Canada.

Fig. 2 .
Fig.2.Location of the 14 study sites.Individual sites are indicated with a single red marker, while red markers with a number indicate more than one study (the number within the dot equals the number of studies) at that site.From: Ecozones of Canada.© Darrel Cerkowniak; adapted fromSmith et al. (2011) and licensed under a CC BY (Attribution) license.

Fig. 3 .
Fig. 3. Forest plot of the overall effects of minimum tillage compared with conventional tillage across the studies used in this meta-analysis.

Fig. 4 .
Fig. 4. Forest plot of the association between growing season precipitation and sand content versus nitrous oxide emission response ratio between conventional and minimum tillage.

Fig. 5 .
Fig. 5. Box plot showing the mean and variation of the effect size comparing nitrous oxide emissions from conventional and minimum tillage along with the sample sizes (n) in studies shorter (left) or longer (right) than 10 years from treatment establishment in Canadian soils.lnRR: logarithmic response ratio.

Fig. 6 .
Fig. 6.Effect size natural (lnRR-%) between minimum and conventional tillage in relation with sand content and growing season precipitation.lnRR: logarithmic response ratio.

Table 1 .
Detailed information of the studies included in the analysis.