Carbon Dioxide Fluxes and Carbon Stocks under Conservation Agricultural Practices in South Africa

: Understanding the impacts of agricultural practices on carbon stocks and CO 2 emission is imperative in order to recommend low emission strategies. The objective of this study was to investigate the e ﬀ ects of tillage, crop rotation, and residue management on soil CO 2 ﬂuxes, carbon stock, soil temperature, and moisture in the semi-arid conditions in the Eastern Cape of South Africa. The ﬁeld trial was laid out as a split-split-plot design replicated three times. The main plots were tillage viz conventional tillage (CT) and no-till (NT). The sub-plots were allocated to crop rotations viz maize–fallow–maize (MFM), maize–oat–maize (MOM), and maize–vetch–maize (MVM). Crop residue management was in the sub-sub plots, viz retention (R + ), removal (R − ), and biochar (B). There were no signiﬁcant interactions ( p > 0.05) with respect to the cumulative CO 2 ﬂuxes, soil moisture, and soil temperature. Crop residue retention signiﬁcantly increased the soil moisture content relative to residue removal, but was not di ﬀ erent to biochar application. Soil tilling increased the CO 2 ﬂuxes by approximately 26.3% relative to the NT. The carbon dioxide ﬂuxes were signiﬁcantly lower in R − (2.04 µ moL m − 2 s − 1 ) relative to the R + (2.32 µ moL m − 2 s − 1) and B treatments (2.36 µ moL m − 2 s − 1 ). The carbon dioxide ﬂuxes were higher in the summer (October–February) months compared to the winter period (May–July), irrespective of treatment factors. No tillage had a signiﬁcantly higher carbon stock at the 0-5 cm depth relative to CT. Amending the soils with biochar resulted in signiﬁcantly lower total carbon stock relative to both R + and R − . The results of the study show that NT can potentially reduce CO2 ﬂuxes. In the short term, amending soils with biochar did not reduce the CO 2 ﬂuxes compared to R + , however the soil moisture increases were comparable.


Introduction
Twenty percent of the global atmospheric CO 2 concentration is emitted from agricultural lands on approximately 37% of the earth land surface [1]. In Africa, South Africa contributes the most to greenhouse gas (GHG) emissions [2], with agriculture alone contributing 9.3% of the emissions [3]. Conventional farming practices such as residue removal and/or burning, continuous tillage with heavy use of machinery, and too much fertilizer usage are responsible for the loss of half of the soil organic carbon (SOC) and contribute about 25% of the total anthropogenic CO 2 emissions [4,5]. The continued increase in the atmospheric CO 2 concentration is responsible for unpredictable and extreme rainfall events and temperatures [6] and, subsequently, crop and soil productivity loss and food insecurity. Therefore, it is important to identify suitable agricultural practices that help to effectively adapt and mitigate these detrimental global environmental effects [7].
Conservation agriculture (CA) is increasingly being promoted as a climate-smart strategy that can help arrest or reverse the processes of soil degradation, promote water conservation, and mitigate climate change [8]. Implementing CA practices can potentially increase biomass production [9]. CO 2 fluxes, carbon stock, soil temperature, and moisture in the semi-arid conditions of the Eastern Cape Province.

Site Selection
The study was conducted at the University of Fort Hare research farm (latitude 32 • 46 S and longitude 26 • 50 E) which is in a warm temperate climate. The site receives an average annual rainfall of 575 mm and a mean annual temperature of 18.1 • C. According to the International Union of Soil Sciences working group [43], the soils are classified as Haplic Cambisol.

Daily Rainfall and Air Temperature
The daily air temperature and rainfall data were collected using an automatic weather station (Weather Link, Davis Instruments Corp, California, USA) located at the University of Fort Hare research farm. The daily average air temperature and daily total rainfall for the duration of the study are presented in Figures 1 and 2, respectively. The highest daily mean air temperature was in the summer season in January (29.8 • C), and the lowest was observed in the winter season (6.5 • C) in July. Rainfall was received throughout the year, but it fluctuated. The highest daily average rainfall was recorded in March (45.4 mm).

Treatments and Experimental Design
The experiment was initiated in the 2015 winter season and continued for six seasons until the 2018/2019 summer season, when it was terminated. It investigated three factors-tillage, crop rotation, and residue management in factorial combination-and was laid out as a split-split plot design, replicated three times. Tillage with two levels (no-tillage (NT) and conventional tillage (CT)) were allocated to the main plots. The subplots were allocated to rotations that were at three levels (maize-fallow-maize (MFM), maize-vetch-maize (MVM), and maize-oat-maize (MOM)). Sub-subplots were allocated to the residue management strategies at three levels (residue removal (R−)9, crop residue retention (R+), and biochar (B)). The main plots measured 17.75 × 15 m 2 , the subplots were 5.25 × 15 m 2 , and the sub-subplots were 5.25 × 5 m 2 . The total area of the trial was 36.5 × 55 m 2 . The gross plot measured 5.25 × 5 m 2 , and the net plot was 2.25 × 4 m 2 . The CT plots were ploughed to a depth of 30 cm once at the beginning of each season-April for winter planting and October for summer planting-using a tractor-drawn disk plough, and then harrowed to make a fine tilth. Crop residue management treatments were conducted at the end of each season-i.e., residues were retained, removed, or converted into biochar. In plots with the biochar treatment, all the crop residues obtained in both the summer and winter seasons were pyrolyzed, as described by Nyambo et al. [36]. After cooling, the biochar was crushed and passed through a 2 mm sieve before being applied to the respective plots.

Management of Non-Experimental Variables
Oat (Avena sativa cv. Sederbrg) and grazing vetch (Vicia dasycarpa cv. Max) cover crops were planted in the winter season at the recommended seed rates of 90 and 35 kg ha -l , respectively. Planting in both the NT and CT plots was done by making small furrows opened using hoes. For both crops, a compound fertilizer (6.7% N; 10% P; 13.3% K) was applied at planting as a basal fertilizer at 10 kg P ha −1 . Grazing vetch was inoculated using the Rhizobium leguminosarium biovar viciae at planting. The cover crops were terminated just before flowering using Glyphosate (N-[phosphono-methyl] glycine, 360 g L −1 ), applied at 5 L ha −1 . When the crops dried, the crop residues were retained, removed, or converted into biochar.
During the summer period, a yellow open pollinated maize variety (Okavango) popular with smallholder farmers in the central Eastern Cape was planted. The maize rows were spaced at a distance of 0.75 m and 0.33 m to give a planting density of 44,444 plants ha −1 . The maize rows were made by making small furrows opened using hoes. Fertilizer was applied in all the plots at a rate of 90 kg N ha −1 , 45 kg P ha −1 , and 60 kg K ha −1 . A compound fertilizer (6.7 % N; 10 % P; 13.3 % K, +0.5 % Zn) was used to supply a third of the N and all the P and K needed by the maize crop [9]. The remaining nitrogen was side dressed at six weeks after planting as Lime ammonium nitrate (LAN). Pests were controlled in the maize and oat crops using Cylam 50EC (Lambda-cyhalothrin (pyrethroid), 50 g L −1 ).

Soil Sampling and Analysis
Soils samples were obtained from 0-5 cm and 5-10 cm depths. Visible organic material was removed before taking three random samples per plot. These samples were used to measure the SOC and total organic carbon.

Carbon Stocks
Soil C stocks (Mg C ha −1 ) were determined for the 0-5 cm and 5-10 cm soil depths using the following equation [44]: where BD = bulk density, and C (%) is the percentage of carbon. Carbon stocks from the two depths were added for each treatment to determine the total carbon stock in the 0-10 cm soil depth. Bulk density was determined at the end each season using the core method outlined in Okalebo et al. [45], while the modified Walkley-Black method was used to determine the SOC following AgriLASA (2004) [46].

Carbon Dioxide Fluxes, Soil Temperature and Moisture
Carbon dioxide measurements were taken in the field on the 10th of each month in every plot in the year in 2017. The 8100A, an automated soil CO 2 flux system from LI-COR, was used to take measurements (LI-COR, Lincoln, Nebraska, USA). The LI-8100A is a fully automated chamber system including a multiplexer, analyzer Control Unit equipped with the infrared gas analyzer (IRGA) to measure the change in CO 2 , and H 2 O vapor concentration and flux calculation software. The Li-COR uses the rate of CO 2 increase in the flow through system with the chamber and connecting housing to measure the flux of CO 2 from the soil surface into the atmosphere. The CO 2 flux (µmoL m −2 ·s −1 ) and soil temperature ( • C) data collected by the LI-COR 8100A were stored in the instrument's flash memory and transferred to a computer. Each of the 16 chambers allowed for the connection of probes for soil temperature determination to a maximum 5 cm depth. Chamber anchors were installed to more than 10 cm into the ground and extended no more than 5 cm above the surface. The chambers were permanently installed in the field.
The soil water content from a 20 cm depth was measured using a Hydro Sense II moisture probe (Campbell Scientific, Inc. Logan, Utah, UT 84321-1784, USA), with a total of three measurements across each plot at each time of sampling. Carbon dioxide flux, soil moisture, and temperature measurements were taken between 9 am and 12 noon to minimize the diurnal variation in gas measurement and to reflect the mean daily temperature [47].

Statistical Analysis
An analysis of variance (ANOVA) was performed on all the variables and statistical calculations were performed using the JMP statistical package version 14 (SAS Institute Inc., Cary, NC, USA). Mean CO 2 fluxes, soil temperature, and soil moisture were calculated by dividing the cumulative value of the parameters (collected over the experimental period) by the number of sampling times. Means were separated using the least significance difference (LSD) at the 5 % probability level.

Soil Temperature and Moisture Content
No significant tillage × crop rotation × crop residue management effects (p > 0.05) were observed with respect to the soil temperature and moisture content. Two-way interactions of tillage x crop rotation, tillage x crop residue management, and crop rotation x crop residue management did not affect either the soil temperature and moisture content. The main effect of residue management (p < 0.05) had a significant effect only on the soil moisture content. The soil moisture content was 8.2% and 4.2% higher in R+ and B, respectively, compared to the R− treatments (Figure 3).

Carbon Dioxide Fluxes
The three-way interaction of tillage × crop rotation × crop residue management did not influence (p > 0.05) the mean CO 2 fluxes. Similarly, all the two-way interactions were not significant (p > 0.05) for the mean CO 2 fluxes in the same period. Tillage (p < 0.001) and crop residue management (p < 0.01) significantly influenced the mean CO 2 fluxes. The average CO 2 fluxes were 26.3% higher in CT relative to NT in 2017 (Figure 4a  Generally, the CO 2 fluxes were higher in the summer (October-February) compared to the winter months (May-July), irrespective of the treatment factors ( Figure 5). The carbon dioxide fluxes were significantly higher in CT relative to NT in the month of October (Figure 5a). In contrast, significantly higher fluxes were observed under NT in June and January. In February, the MFM treatments had 30% and 23% higher CO 2 fluxes relative to the MOM and MVM treatments, respectively, (Figure 5b), while in November the CO 2 fluxes recorded in the MFM were 35% and 31% lower relative to the MOM and MVM treatments. The crop residue removals (3.13 µmoL m −2 s −1 ) resulted in significantly higher CO 2 fluxes relative to both the B (2.57 µmoL m −2 s −1 ) and R+ (2.62 µmoL m −2 s −1 ) treatments in January (Figure 5c). In June, R+ (2.58 µmoL m −2 s −1 ) had significantly higher CO 2 fluxes compared to both the B (1.78 µmoL m −2 s −1 ) and R− (1.93 µmoL m −2 s −1 ) treatments (Figure 5c). Significantly low CO 2 flux emissions were recorded during the winter season compared to the summer period (Figure 5d).

Carbon Stocks
The fouR−way interaction of depth × tillage × crop rotation × residue management was not significant (p > 0.05) with respect to the carbon stock. Similarly, the three-way interactions were not observed with respect to the carbon stock. Only the two-way interaction of depth x tillage significantly (p < 0.001) influenced the carbon stock. The main effect of depth (p < 0.001), tillage (p < 0.001), and residue management (p < 0.05) had a significant effect on (p < 0.001) the carbon stock. No tillage had a significantly higher carbon stock at the 0-5 cm depth relative to CT (Figure 6a). Retaining crop residues significantly increased the carbon stock compared to both B and R− (Figure 6b). Amending the soils with biochar resulted in a significantly lower total carbon stock in the 0-10 cm depth relative to both R+ and R− (Figure 7a). The total carbon stocks were 23 % higher under NT compared to CT (Figure 7b).

Soil Temperature and Moisture Content
The significantly high moisture content in the R+ treatment can be attributed to the mulching effect of the crop residues. The retained crop residues act as mulch, which absorbs the power of raindrops, allows water to infiltrate, and reduces evaporation by stopping the direct impact of the sunlight on soil [48]. Biochar, on the other hand, can retain soil moisture by increasing the soil porosity and hydraulic conductivity [49]. Biochar's high specific surface area and its internal porosity are also responsible for increasing the water holding capacity, hence the improved soil moisture content [36].

Carbon Dioxide Fluxes
The higher CO 2 fluxes in CT were possibly due to the increased decomposition of crop residues and aggregate protected OM when the soil was turned during tillage operations. Tillage stimulates soil microbial activity due to increased soil aeration [50]. According to Mangalassery et al. [51], tilling the soil affects the soil properties. which in turn influence the emission of GHGs. The same authors further state that NT reduces the soil porosity by up to 33%, while CT increases the porosity, which favors the respiration of aerobic organisms by improving water and air movement through the soil. The results of this study agree with those of Dendooven et al. [52] and Angers et al. [50], who reported significantly high CO 2 fluxes under CT compared to NT. The range of CO 2 fluxes (1.4 to 4.1 µmoL m −2 s −1 ) falls within reported ranges under similar semi-arid and/or sub-humid temperate climates [53].
Higher CO 2 fluxes under R+ relative to R− treatments can be attributed to the availability of easily decomposable organic matter under R+, which tends to stimulate microbial activity and CO 2 production [54]. Several authors have reported that crop residues stimulated and increased CO 2 emissions [55,56]. Even though fluxes were higher under R+ relative to R−, they did not result in a net loss in soil C sequestration. This is shown by the significantly higher SOC stock in R+ compared to both the B and R− treatments (Figures 6b and 7a). In this study, the increased CO 2 emission in plots with biochar can be possibly due to the increase in the decomposition of SOM fractions. Biochar addition to the soil causes fluctuations in the SOC, microbial activity, and dissolved organic C, which can impact the mineralization of soil C [36,57,58]. The results agree with Sagrilo et al. [59] and Cross and Sohi [60], who reported that short-term surges in CO 2 emissions are common after biochar amendment, despite its inherently low biodegradability. Furthermore, the decomposition of the labile fractions of the biochar could possibly be the reason for the initial increase in the CO 2 fluxes. In a study on a silty sand soil formed from river sand deposits, Haider et al. [61] reported an initial increase in CO 2 fluxes under biochar treatments. The authors attributed it to the responses of biochar to labile organic carbon fractions from volatiles adsorbed on biochar surfaces during a condensation period after pyrolysis. In contrast, in a long-term field study on silty loam soils in the semi-arid conditions in China, Shen et al. [62] reported that the cumulative CO 2 emissions decreased substantially after the field application of biochar because it suppressed soil respiration during the maize growing season, while in the short-term incubation study the biochar addition had no effect on CO 2 emissions without additional nitrogen amendment. The significantly higher fluxes in the B and R+ treatments relative to R− could be as a result of the high soil moisture content in the B and R+ treatment.
The lower fluxes in the winter season are possibly due to the low temperatures in winter. Temperature directly increases metabolic rates and biochemical processes [63], and thus it increases soil respiration and CO 2 emission [64]. According to Almagro et al. [64], a reduction in both heterotrophic and autotrophic respiration due to low winter temperatures causes low CO 2 emission. Additionally, several studies have come up with the similar conclusion that soil temperature is a major variable affecting soil CO 2 emissions, while soil moisture has little to no effect [65,66]. According to Dilekoglu and Sakin [67], CO 2 emission is positively correlated to soil temperature; the same authors reported that 0.091 g C is emitted for each 1 • C increase in soil temperature.

Carbon Stocks
The results of this study show that switching from CT to CA in the Eastern Cape Province changes the vertical distribution of SOC in the soil profile, resulting in a decrease in the SOC with depth. Less or minimum soil disturbances under NT reduces SOC decomposition, leading to SOC storage in the top soil layer in this study (Figure 6a). Tilling the soil increases the aeration and exposure of the SOM to microbial attack, which results in reduced C storage in the lower soil depth [19]. Similar observations under the same soil and conditions were reported by Dube et al. [30]. In their study, the authors reported increases in the total SOM from about 10 g kg −1 to above 20 g kg −1 after four years of CA. In a study in a subtropical monsoonal humid climate in China, Xu et al. [68] also reported higher carbon stocks in NT relative to CT. The continued input of biomass to the soil surface in the current study can be attributed to the increased SOC stock under R+ treatments. According to Nyambo et al. [9], the presence of crop residues on the soil surface increases the SOC, which also controls erosion, water infiltration, and conservation of nutrients, and is related with the soil quality.
In this study, the low SOC stocks under B treatments relative to R+ can be attributed to the quantities applied. The total quantity of biochar applied at the current stage of the study range from 9.4 to 16.7 Mg ha −1 under NT+MFM and NT+MOM treatments respectively [9]. Biochar application rates that are less than 20 t ha −1 do not results in any significant soil improvements [69]. The low biochar quantities applied in this study and, as previously discussed, the decomposition of the labile C fractions of biochar quickly could also explain the low SOC stocks under B treatments. However, it is important to note that the remaining C is more stable compared to the stocks in R+ and R− treatments and will persist in the soil much longer. Furthermore, the low SOC content in B treatments can possibly be explained by the lateral and vertical migration of biochar along the soil profile [70]. The increase in the SOC storage under NT and R+ in the study is a key indicator of increased sustainability of the system.

Conclusions
This study provided insights into the impact of CA on CO 2 fluxes from semiarid climatic conditions in South Africa. Biochar application resulted in a high soil moisture content comparable to R+ treatments. The fluxes were significantly higher in CT compared to NT. Additionally, the biochar and R+ treatments had significantly higher CO 2 fluxes compared to the R− treatments. The results of this study show that switching from CT to CA in the Eastern Cape Province changes the vertical distribution of SOC in the soil profile. This paper therefore contributes to the evidence that, in the short term, R+ and B increase CO 2 fluxes while improving the soil moisture content. Further studies that focus on CO 2 measurements after every rainfall and tillage events are needed in order to recommend full-scale successive CA application. Long-term studies are also required for better recommendations.
Author Contributions: P.N. designed, set up, collected data, analyzed data, and wrote the manuscript. C.C. designed, set up, collected data, analyzed data, and wrote the manuscript. T.A. designed, set up, collected data, analyzed data, and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.