Effect of soil mulching on agricultural greenhouse gas emissions in China: A meta-analysis

Human demand for food has been increasing as population grows around the world. Meanwhile, global temperature has been rising with the increase of greenhouse gas (GHG) emissions. Although soil mulching (SM) is an effective method to increase crop yield because it could conserve soil moisture and temperature, it is also an important factor affecting GHG productions and emissions. At present, research results in terms of the impact of SM on agricultural GHG emissions are still inconsistent. Therefore, a meta-analysis was used to quantitatively analyze the impact of SM on crop yield and GHG emissions in China. Overall, SM significantly enhanced not only crop yield, but also GHG emissions. Compared with no soil mulching (NSM), SM improved crop yield by 21.84%, while increased global warming potential (GWP) by 11.38%. To minimize the negative impact of SM on GHG, for maize and wheat in arid, semi-arid and semi-humid zones, it is recommended to use flat full mulching with grave or straw plus drip irrigation under neutral or weakly alkaline soil with bulk density <1.3g cm-3. For rice in humid regions, it is advisable to apply SM to minimize GHG emissions by significantly decreasing CH4 emissions.


Introduction
Food security and global warming are two critical issues in the 21st century [1]. According to Food Security and Nutrition in the World by Food and Agriculture Organization of United Nations in 2020, nearly 690 million people around the world have been starving in 2019, accounting for 8.9 percent of the global population. Human demand for food has been increasing with the increase of global population. In addition, global temperature has been rising as the atmospheric concentration of greenhouse gas (GHG) increases, especially carbon dioxide (CO 2 ), methane (CH 4 ) and nitrous oxide (N 2 O), leading to a series of problems, such as severe natural disasters, the acceleration of the extinction rate and crop yield reduction, etc. According to the assessment of Intergovernmental Panel on Climate Change (IPCC), agriculture is the second largest source of GHG emissions, accounting for about 13.5% of global anthropogenic emissions [2]. Farming and field management indirectly affect productions and emissions of GHG by changing the soil environment. Thus, it is essential to find ways to effectively reduce GHG emissions while improving crop productivity. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 SM is an effective method to increase crop yield because it could conserve soil moisture and temperature, hence it is widely practiced around the world, and China has the largest planting area with SM. Although SM has a significantly positive effect on crop yield, it is also an important factor affecting GHG productions and emissions. SM can directly impact soil respiration and alter CO 2 mainly by changing the soil surface properties, affecting the soil microenvironment, such as soil temperature, moisture, water-filled porosity, and aeration. Tyagi et al. [3] found that although SM can effectively reduce surface soil evaporation and increase soil water content in the root zone, it promotes CH 4 productions and emissions. However, Nan et al. [4] argued that the concentration of CH 4 under SM was lower than that under NSM for maize. Mancinelli et al. [5] indicated that with the increase of soil moisture due to SM, N 2 O emissions also increased, whereas Liu et al. [6] reported that SM promoted the absorption of nitrogen by plants, thus reducing the content of inorganic nitrogen and inhibiting N 2 O emissions. Furthermore, some studies found that SM increased CO 2 relative to NSM, while others showed SM reduced CO 2 by hindering the exchange between the soil and the atmosphere [7]. In general, many field experiments have been conducted to explore the effect of SM on GHG emissions, but the reported GHG emissions responding to SM varied greatly due to different climate, soil factors, crop types, irrigation methods, and mulching management practices. Therefore, a meta-analysis of SM was employed in this study to examine the variations and draw general quantitative conclusions.
Meta-analysis is the scientific synthesis of research results. It can be used to evaluate the consistency of the results of independent experiments involving the same subjects [8]. In recent years, meta-analysis has been widely used in analyzing GHG emissions. For example, employing a meta-analysis method, He et al. [9] analyzed the effects of plastic mulching on crop yield and GHG emissions, and Shan and Yan [10] explored the effects of straw mulching on crop yield and GHG emissions.
The main objective of this study was to quantitatively analyze the impact of SM on crop yield and GHG emissions in China, explore the response of GHG emissions to several critical influencing factors, such as mulching methods, climate conditions, soil properties, irrigation methods, and finally find a mulching method that is most conducive to GHG emission reductions.

Data collection and categorization
Peer-reviewed studies exploring the impact of SM on GHG emissions in China and published between 2009 and 2021 were collected. Several online databases, including Elsevier (Science Direct), Web of Science, and China National Knowledge Infrastructure, were used as search engines in this study. During the search, keywords were focused on: 1) soil mulching or 2) mulching and greenhouse gas or 3) GHG or 4) greenhouse gas emission or 5) methane or 6) nitrous oxide or 7) carbon dioxide. Using the above keywords, totally 45 publications were collected, including 183 pairs of observations of crop yield, 97 pairs of observations of global warming potential (GWP), 116 pairs of observations of CH 4 emissions, 177 pairs of observations of N 2 O emissions, and 106 pairs of observations of CO 2 emissions that satisfied the criteria for meta-analysis.
To be selected for meta-analysis, publications had to meet the following criteria: 1) only publications describing experiments conducted in the field with side-by-side comparisons of SM and NSM were included and pot studies were excluded; 2) crop yield and GHG emissions data in SM and NSM were reported. It should be noted that, in this meta-analysis, GWP, CH 4 , N 2 O and CO 2 emissions could be represented by GHG emissions; 3) the primary data of GHG emissions under SM and NSM must be comparable; and 4) the mean, sample size and a measure of dispersion (SE or SD) as numerical or graphical data were available, or SD of GHG emissions data could be calculated from the reported data for SM and NSM. If data were presented graphically, figures were digitized to extract the numerical values by the Get-Data Graph Digitizer (ver. 2.22, Russian Federation).
In addition to SM, several other variables (moderating factors) could also influence GHG emissions. Before meta-analysis, moderating factors were categorized as: 1) factors of SM presented in different ways, including mulching materials, patterns and area; 2) factors of soil, including bulk density (g cm -3 ) and pH; 3) factors of irrigation methods presented by rainfed, drip irrigation and surface irrigation; 4) factors of crop types, including maize, wheat and rice; and 5) factors of climate presented by average annual temperature (˚C) and average annual precipitation (mm).

Data analysis
The standard deviation (SD) is one of the required parameters in meta-analysis. If value of SD could not be obtained in publications, it can be calculated as: ðSE can be found in publicationsÞ ( where SE is the standard error of mean GHG emissions and crop yield data, n is the sample size and it is also the number of repetitions in publications, X is the mean GHG emissions and crop yield data of the treatment and control group, and CV is the average coefficient of variation.
In addition, the effect size was measured by the natural logarithm of the response ratio: where X t and X c are the mean values of crop yield and GHG emissions for the treatment and control group, respectively, and lnR is a unit-free index. Furthermore, the variance of lnR (V i ) was calculated as [11]: where SD t and SD c are the standard deviations for the treatment and control group, respectively, and N t and N c are the sample sizes of the treatment and control group, respectively. Usually, the weighted mean was used to produce the greatest precision since statistical precision of each experiment was different. The weighted mean response ratio (lnR ++ ) and weight (W i ) were calculated as follows [11]:

PLOS ONE
Effect of mulching on greenhouse gas emissions: A meta-analysis where i and k are the number of comparisons and the cumulative groups, respectively, and W i is the weight of the effect size of the treatment group. The 95% confidence interval (95% CI) to the lnR ++ was calculated using the following formula: SE lnR þþ ¼ ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where SE lnR þþ is the standard deviation of lnR ++ .
A random-effect model was adopted in this meta-analysis to examine the performance of crop yield and GHG emissions data under SM and NSM, respectively. To simplify the understanding, the lnR ++ results were reported as the percentage change on the basis of the comparison between the treatment and control group (ðe lnR þþ À 1Þ � 100%).

Publication bias
The funnel plot can be used to intuitively assess whether there is a publication bias. If there were no publication bias, the funnel plot would be similar to an inverted symmetric funnel. Otherwise, the plot would be asymmetric [12][13][14]. Moreover, the potential impact of SM on the overall effect size of crop yield and GHG emissions needs to be evaluated through the "trim and fill" method. According to the results of the funnel plot (Fig 1), the crop yield and GHG emissions data were near-symmetrical, and the trim and fill analysis indicated there was no missing study. Therefore, the publication bias was not a big problem for this meta-analysis.

Overview of the dataset
The database was obtained from 45 articles regarding the impact of SM on GHG in China. The effect size of crop yield, GWP, CH 4 , N 2 O and CO 2 emissions were presented in Fig 2. Overall, compared to NSM, SM significantly improved crop yield by 21.84%. However, it also increased GWP by 11.38%. Specifically, SM increased CO 2 and N 2 O by 21.62% and 1.73%, respectively, but significantly decreased CH 4 by 43.08%. Therefore, it is essential to find potential strategies that reduce GWP while maintaining or improving crop yield.

Crop types and irrigation methods
Fig 3A showed the impact of crop types on yield and GHG emissions. As indicated by Fig 3A, the yield of maize and wheat under SM were remarkably increased by 18.61% and 36.36%, respectively. However, GWP was also increased by 24.68% and 22.54% due to an increase in CO 2 emissions from maize and wheat fields by 26.07% and 25.95%, respectively. Moreover, the effect of SM on rice yield was insignificant, but GWP was drastically reduced by 34.71%. In rice fields, N 2 O was increased by 50.39%, while CH 4 was decreased by 55.63%, resulting in the general reduction of GWP.  30.60% and 21.35%, respectively. Nevertheless, surface irrigation and rainfed increased GWP by 18.70% and 32.97%, respectively, as using the above two irrigation methods CO 2 were increased by 16.51% and 19.66%, and the increase of N 2 O was 10.77% and 10.45%, respectively. Besides, drip irrigation slightly reduced GWP by 2.59% due to the decrease of N 2 O, although CO 2 was increased by 26.38%.
Therefore, it is recommended that SM be used to decrease GHG emissions through effectively reducing CH 4 emissions for rice. Regarding maize and wheat, it is advisable to use SM plus drip irrigation to minimize GHG emissions by remarkably decreasing N 2 O emissions.

Soil mulching factors
Fig 4A showed the influence of mulching materials on crop yield and GHG emissions. Compared with NSM, grave mulching remarkably increased crop yield by 39.18%while decreasing GWP by 10.86% due to the 12.83% reduction of CO 2 . Straw mulching only slightly increased crop yield by 10.93%, but the increase of GWP was also small (5.61%). Under plastic mulching, crop yield was increased by 24.23%, but GWP was also increased by 14.17% due to the increase of CO 2 .

PLOS ONE
Effect of mulching on greenhouse gas emissions: A meta-analysis  Consequently, it is advisable to use full flat mulching with grave or straw to improve crop yield while minimizing GHG emissions.

Soil conditions
Fig 5A presented the influence of soil bulk density on crop yield and GHG emissions. Compared with NSM, crop yield was significantly increased under SM regardless of soil bulk density. However, the lower the bulk density was, the greater the increase of crop yield and GWP was. When bulk density was <1.3g cm -3 and �1.3g cm -3 , crop yield was improved by 23.28% and 13.80%, respectively, and GWP was increased by 11.51% and 5.16%, respectively. Both CH 4 and CO 2 emissions were reduced regardless of the soil bulk density. However, the larger soil bulk density was, the greater the CH 4 decrease was. CH 4 was decreased by 75.19% and 42.21% with soil bulk density of <1.3g cm -3 and �1.3g cm -3 , respectively. However, the difference between the effect of SM on CO 2 emissions under soil bulk density of <1.3g cm -3 and �1.3g cm -3 was insignificant (15.12% vs. 13.78%). Additionally, there was no significant difference in the effects of SM on GWP with varied soil bulk density because there was no significant difference in CH 4 , N 2 O, and CO 2 emissions among different bulk densities. Regarding N 2 O emissions, it was reduced when soil bulk density was large and vise versa. SM decreased N 2 O by 10.05% with soil bulk density �1.3g cm -3 , and N 2 O was increased by 14.96% with soil bulk density <1.3g cm -3 . Fig 5B showed the impact of soil pH on crop yield and GHG emissions. SM increased crop yield regardless of soil pH values, and the increases were 4.80%,38.90% and 23.11% when the soil was acidic (Ph<7), neutral or weakly alkaline (7�pH�8) and alkaline (pH>8),

PLOS ONE
Effect of mulching on greenhouse gas emissions: A meta-analysis respectively. Although the impact of SM on crop yield was the smallest under acidic soil, GWP was significantly reduced by 20.04% due to the great reduction of CH 4 (52.36%). Under neutral or weakly alkaline soil CH 4 and N 2 O emissions were decreased by 12.93% and 4.71%, respectively, but CO 2 was increased by 12.39%. Under alkaline soil, N 2 O and CO 2 emissions were increased by 18.13 and 29.42%, respectively.
In conclusion, considering the small risk of increasing GHG emissions due to SM under neutral or weakly acidic soil with bulk density < 1.3g cm -3 , it was advisable to increase crop yield by SM in practice. Fig 6A presented the impact of the annual average temperature on crop yield and GHG emissions. SM increased crop yield regardless of the temperature. However, the lower the temperature was, the greater the increase was, as SM increased soil moisture, which was beneficial for crop growth. When the temperature was <13˚C and �13˚C, crop yield was increased by 26.04% and 18.49%, respectively. Low-temperature areas usually belong to arid or semi-arid zones, where maize and wheat are the major plants, and GHG emitted the most is CO 2 , and N 2 O followed, and CH 4 is the last. Therefore, when the temperature was low (<13˚C), GWP was significantly increased by 42.79% due to the 14.17% reduction of CO 2 , although both CH 4 and N 2 O are decreased by 12.71% and 9.62%. High-temperature areas tend to plant rice, and the main GHG emitted is CH 4 . Hence, when the temperature was high (�13˚C), SM decreased GWP by 15.64% due to the greater reduction in CH 4 (51.41%). Fig 6B presented the impact of annual average precipitation on crop yield and GHG emissions. According to rainfall, the region with precipitation of <400mm is generally considered arid and semi-arid zones, the region with precipitation of 400mm-800mm is sub-humid areas, and the one with precipitation of >800mm is considered humid regions. In arid, semi-arid and semi-humid areas, SM significantly increased crop yield by 20.42 and 28.98%, respectively, but GWP was also greatly increased by 15.2% and 21.19% largely due to CO 2 emitted by maize

PLOS ONE
Effect of mulching on greenhouse gas emissions: A meta-analysis and wheat. In humid regions, the effect of SM on crop yield was insignificant, but GWP was remarkably reduced mainly due to the significant decrease of CH 4 by rice. Furthermore, in humid areas, SM had no significant influence on crop yield, but it remarkably decreased GWP by 30.04% mainly due to the significant decrease of CH 4 emissions from rice. Furthermore, N 2 O emissions increased as precipitation enhanced. In arid and semi-arid areas, SM markedly decreased N 2 O emissions by 45.10% relative to NSM, but SM significantly increased N 2 O emissions by 5.09% and 15.92% in semi-humid and humid areas, respectively.
Hence, in humid regions where the annual average temperature was higher than 13˚C and the annual average precipitation was higher than 800mm, the risk of increasing GHG emissions by mulching was small, and it was advisable to increase crop yield by mulching in practice.

Soil mulching management strategy
The above analysis indicated SM significantly improved crop yield, but it also remarkably enhanced GHG emissions. To minimize the negative impact of SM on GHG, for maize and wheat in arid, semi-arid and semi-humid zones, it is advisable to use full flat mulching with grave or straw plus drip irrigation under neutral or weakly alkaline soil with bulk density <1.3g cm -3 . For rice in humid regions, it is advisable to apply mulching to minimize GHG emissions by significantly decreasing CH 4 emissions.

Crop types and irrigation methods
Compared with NSM, SM significantly reduced CH 4 emissions, but greatly increased N 2 O emissions in rice fields. CH 4 emissions are mainly caused by the anaerobic fermentation of soil organic matter resulting from the decrease of soil permeability, soil respiration and redox potential due to rice field flooding. Non-flooded mulching for rice completely changed the aquatic environment of rice. Since there was no water layer on the soil surface, soil water content decreased, soil permeability improved, and oxygen content in soil increased, and then an aerobic environment for soil was created [15], which inhibited productions of CH 4 by methanogens and stimulated the oxidation of CH 4 by methanotrophic agents, thus resulting in a great reduction of CH 4 emissions in rice fields. However, the aerobic and humid environment of the soil in rice fields is conducive to the increase of soil microbial activity and quantity, the improvement of soil enzyme activity, and the promotion of soil nitrification and denitrification, and thus leads to the increase of N 2 O productions and emissions. China is a big riceplanting country, and its planting area and yield respectively accounted for 22% and 34% around the world. Rice fields are an important source of CH 4 emissions, accounting for about 9%~19% of the total CH 4 emissions [16]. To reduce global GHG emissions, under the condition of maintaining rice yield, it is advisable to combine plastic mulching and optimize nitrogen management (e.g., split application, deep placement, use of controlled-release fertilizer, nitrification and urease inhibitors) can be considered. SM increased CO 2 emissions from maize and wheat fields compared to NSM. This was mainly because soil temperature was the main factor affecting CO 2 emissions, and SM increased soil temperature and moisture, which is conducive to the respiration of crop roots and soil microorganisms to produce CO 2 [17,18].
Compared with NSM, surface irrigation under SM can significantly reduce CH 4 emissions, and drip irrigation obviously decreased N 2 O emissions. This was probably because SM reduced surface soil evaporation in rice fields, and thus irrigation amount was remarkably reduced, soil aeration was greatly improved, resulting in the large decrease of CH 4 emissions. Denitrification, which is the main way to produce N 2 O, generated far more N 2 O than nitrification. Water and oxygen (O 2 ) indirectly affected denitrification by influencing soil redox potential. Previous research suggested denitrification is positively correlated with soil moisture and negatively related to O 2 content. Drip irrigation was an efficient water-saving irrigation method, which only wetted soil in the root zone of crops. Relative to other irrigation methods such as surface irrigation, drip irrigation decreased soil moisture and increased O 2 content in the soil, and thus soil denitrification was effectively inhibited and N 2 O emissions were significantly reduced [19]. Rainfed, drip irrigation and surface irrigation under SM all promoted CO 2 emissions, and this is probably because SM increased soil temperature, accelerated the decomposition of organic matter and improved microbial activity, thus increasing CO 2 emissions.
Relative to flat mulching, ridge mulching significantly reduced CO 2 emissions, but increased N 2 O emissions. It was because relative to flat mulching, ridge mulching can effectively retain soil moisture in ridges, thus promoting soil denitrification and increasing N 2 O emissions [16,20]. However, because ridge mulching effectively kept soil moisture inside ridges, soil temperature may drop, resulting in reduced microbial activity in the soil and slower soil respiration, and so CO 2 emissions were significantly reduced. Flat mulching significantly reduced CH 4 emissions mainly due to the great decrease of CH 4 in rice fields, while there is no significant impact of ridge mulching on CH 4 emissions.
Compared to NSM, full mulching decreased N 2 O emissions, but partial mulching increased N 2 O. This is mainly because the absorption of nitrogen increased with the increase of mulching area, and so soil nitrogen content is enhanced, resulting in the reduction of N 2 O emissions. Both full and partial mulching reduced CH 4 emissions as compared with NSM and this was also attributed to the reduction of CH 4 emissions by mulching in rice fields. However, relative to partial mulching, full mulching reduced CH 4 emissions to a large extent, as the increase of mulching area further blocked the channels of CH 4 emissions. Relative to NSM, both full and partial mulching promoted CO 2 emissions, and it is primarily because SM increased soil temperature, and so soil microbial activity was enhanced and soil respiration was accelerated.

Soil conditions
The larger the soil bulk density was, the greater the reduction in CH 4 and N 2 O emissions would be. This was possible because the larger the soil bulk density was, the denser the soil was, the worse the aeration was, and the more difficult the exchange of GHG with the atmosphere would be, thus leading to less CH 4 and N 2 O emissions. Compared with NSM, SM greatly increased CO 2 emissions regardless of soil bulk density, which may be because SM largely increased soil temperature. Neutral and weakly alkaline soil reduced N 2 O emissions. This was probably because soil with too much acid or alkali produced toxicity to crop roots [30], which was not conducive to the absorption of water and nutrients by crops, while neutral or weakly alkaline soil was conducive to the absorption of nitrogen by crops, thus reducing N 2 O emissions. As SM increased soil temperature, it promoted CO 2 emissions regardless of soil pH values. However, the effect of neutral or weakly alkaline soil on CO 2 emissions was smaller than that of alkaline soil, as neutral or weak alkaline soil was beneficial for the improvement of the activity of autotrophic microbiology, and thus the ability of carbon sequestration of autotrophic microbiology was enhanced, and finally CO 2 emissions were suppressed.

Climate conditions
South China, where lots of rice is planted, had a higher average annual temperature. The higher the temperature was, the more significant the reduction of CH 4 emissions by SM was, which might be attributed to the remarkable decrease of CH 4 emissions in rice fields. Regarding N 2 O, emissions increased when the annual average temperature was low, and vice versa. This is probably because areas with lower average annual temperature are commonly arid or semi-arid regions, where SM effectively improved soil temperature, which was conducive to crop growth, promotion of nitrogen uptake by crops, and thus N 2 O emissions were reduced. Furthermore, areas with higher average annual temperature are generally semi-humid or humid zones in China, where there are more rainfall and higher soil water content, which promoted denitrification and thus increased N 2 O emissions. The higher the average annual temperature was, the more obvious the effects of SM on promoting CO 2 emissions. It was because in regions with higher temperature, soil temperature was even higher due to SM, which was more conducive to CO 2 generations.
In humid areas, SM significantly reduced CH 4 emissions compared to NSM, and this was also attributable to the remarkable decrease of CH 4 emissions from rice fields, which were generally planted in humid areas. N 2 O emissions increased with the increase of precipitation, and this is mainly because soil moisture increased with the increase of rainfall, which promoted denitrification and thus improved N 2 O emissions. In arid, semi-arid and sub-humid regions, SM promoted CO 2 emissions, which was attributed to the increase of soil temperature by SM.

Conclusions
Meta-analysis was used to quantitatively analyze the impact of SM on crop yield and GHG emissions from farmland under different crop types, soil characteristics, climate conditions and irrigation methods. In general, compared with NSM, SM significantly increased crop yield by 21.84%, while it increased GWP by 11.38%. Specifically, SM significantly reduced CH 4 emissions by 43.08%, increased CO 2 emissions by 21.62%, but it had no significant impact on N 2 O emissions.
To minimize the negative impact of mulching on GHG, for maize and wheat in arid, semiarid and semi-humid zones, it is advisable to use full flat mulching with grave or straw plus drip irrigation under neutral or weakly alkaline soil with bulk density <1.3g cm -3 . For rice in humid regions, it is advisable to apply mulching to minimize GHG emissions by significantly decreasing CH 4 emissions.
Agricultural GHG emissions are also significantly affected by fertilization, especially nitrogen fertilizer. Therefore, the influence of fertilization application on GHG production and emissions from farmland should be considered in further research.
Supporting information S1 Table. Raw data as basis for a meta-analysis. (XLSX)