Carbon Input and Maize Productivity as Inﬂuenced by Tillage, Crop Rotation, Residue Management and Biochar in a Semiarid Region in South Africa

: Conservation agriculture (CA) as a system is still evolving on many of the smallholder farms in sub-Saharan Africa (SSA) and questions on the impact of individual components and pathways toward adoption still require answers. A short-term study was conducted to investigate the e ﬀ ect of tillage, crop rotation, and crop residue management, including maize residue biochar on above ground biomass, cumulative carbon (C) input, soil organic carbon (SOC), and maize grain yield. A split–split plot design was used to evaluate two tillage operations (conventional tillage (CT) and no-till (NT)), three crop rotations (maize–fallow–maize (MFM), maize–oat–maize (MOM), and maize–vetch–maize (MVM)), and three-crop residue management (retention (R + ), removal (R − ), and biochar (B)). The cumulative above ground biomass produced in the MOM rotation was signiﬁcantly higher by 78.9% and 88.7% relative to MVM and MFM rotations, respectively. The cumulative C input under residue management treatments ranged from 10.65 to 12.16 Mg ha − 1 . The highest SOC was observed under R + (1.10%) followed by B (1.0%) and the lowest was in R − (0.96%). Crop residue management signiﬁcantly a ﬀ ected grain yields in 2015 / 2016 ( p < 0.05) and 2016 / 2017 ( p < 0.01) summer seasons. Biochar did not result in an obvious improvement in both C input and crop yield. Smallholder farmers can potentially switch from CT to NT without any signiﬁcant yield penalty, as well as adopt MOM and R + practices for increased biomass and C input.


Introduction
The poor adoption of conservation agriculture (CA) is a result of several socioeconomic and biophysical factors like availability, cost and knowledge associated with use of herbicides for purposes of weed control, lack of adapted implements for direct sowing and residue management tradeoffs in crop-livestock farming systems [1,2]. Crop rotation, minimal soil disturbance, and crop residue soil cover are all key elements for the success of CA [3]. Description of smallholder farmer practice of CA by many authors (e.g., [1,4,5]) shows that many adopt one or two of the principles but rarely all three [6]. This shows that Farmers are on a pathway towards CA and will undergo intermediary steps. Evaluation of interaction of factors involved in CA may reveal the best practices to promote a combination or steps that will assist to enhance improvement in early phases when there might be little visible evidence of change at field level. In this regard, a component omission experiment could be a useful tool. This approach can lead to the identification of informed and sustainable key entry points for potential adoption by the farmers [7,8].
Management practices have an important effect on the soil fertility and crop productivity [9]. Conservation agriculture is a sustainable agricultural practice because of its contribution to soil

Site Selection
Field experiments were conducted at the University of Fort Hare research farm, which is located at latitude 32 • 46 S and longitude 26 • 50 E, at an altitude of 535 m above sea level (masl). The soils are deep and of alluvial origin, classified as Haplic Cambisol [40] and dominated by mica in the clay fraction [41]. The site has a warm temperate climate with a mean annual temperature of 18.1 • C and an average annual rainfall of 575 mm received mainly during the summer months from November to March [42]. The soil type and climate at the research farm closely resemble those of smallholder farms in the vicinity of Alice, in the central EC. Prior to the establishment of the field trial, the field was planted to maize (Zea mays L.).

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 (Table 1). It investigated three factors: tillage, crop rotation, and residue management in factorial combination, and was laid out as a split-split plot design. Tillage with two levels (no-tillage (NT) and conventional tillage (CT)) was 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−), 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 (Table 1), using a tractor-drawn disk plough and then harrowed to make a fine tilth.

Management of Nonexperimental Variables
The winter cover crops were oat (Avena sativa cv. Sederbrg) and grazing vetch (Vicia dasycarpa cv. Max), planted at recommended seed rates of 90 and 35 kg ha −l , respectively. Planting in both NT and CT plots was done making small furrows opened using hoes. Basal fertiliser for both cover crops was Agronomy 2020, 10, 705 4 of 16 applied as a compound at planting at 10 kg P ha −1 , as a compound (6.7% N, 10% P, 13.3% K). Only the oat cover crop was top dressed at six weeks after planting using lime-ammonium (LAN) (28% N) at a rate of 138 kg ha −1 to make a total of 45 kg N ha −1 . The Rhizobium leguminosarium biovar viciae was used to inoculate the grazing vetch at planting. Glyphosate (N-(phosphonomethyl) glycine, 360 g L −1 ) was applied (5 L ha −1 ) to terminate the cover crops. At the end of each season, crop residues were retained, removed, or converted into biochar. In plots with the B treatment, all crop residues obtained in both the summer and winter seasons were pyrolysed using 200 L oil drum kilns 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, before planting of the winter or summer crops. The total quantity of biochar applied in the study is shown in Table 2. Okavango, a yellow open pollinated maize variety (OPV) popular with smallholder farmers in the central Eastern Cape, was planted. The maize rows were spaced at a distance of 0.75 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 [43]. The remaining nitrogen was side dressed at six weeks after planting as LAN. Pests were controlled in the maize and oat crops by using Cylam 50EC (lambda-cyhalothrin (pyrethroid), 50 g L −1 ). Supplementary irrigation was applied to both winter and summer crops when necessary (Table 3).

Measurements
Soil sampling was done soon after harvesting of each rotational crop. Before sampling, all crop residues and visible organic matter were removed. Three random composite samples were collected using a spade from the 0-5 cm depth and a 7 cm diameter auger was used for the 5-10 cm depth. The samples were taken to the laboratory in rigid containers to avoid breaking the soil aggregates. They were then air-dried and visible organic debris removed before laboratory work. These samples were used to measure SOC and total organic C.

Aboveground Biomass Yields
In each winter season, after termination, cover crop residues and weed samples were obtained from two quadrats measuring 35 × 35 cm for estimation of aboveground biomass yield. The biomass was harvested by cutting at ground level. Samples were oven-dried at 60 • C to a constant weight for determination of dry weight and further chemical analysis [44]. The seasonal aboveground biomass of the maize crop is reported in this study. Cumulative aboveground biomass was determined by summing all the biomass yield obtained in the six cropping seasons.

Soil Organic Carbon and Carbon Input
Soil organic carbon was determined following the modified Walkley-Black method as outlined in Agri Laboratory Association of Southern Africa [45]. Total soil C was measured using dried and ground plant and soil samples following the dry combustion method using a LECO TruSpec C/N auto analyzer [46]. This was used to calculate the C input by finding the product of the C concentrations (%) and the dry weights of the crop biomass in the various treatments. Cumulative C input was determined by summing C input obtained in the six cropping seasons.

Maize Yield and Yield Components
Plant height was measured on five random plants per plot. Thousand grain weight was determined by counting a thousand maize grains and weighing [44]. Grain yield (Mg ha −1 ) was corrected to 12.5% moisture level.

Data Analysis
For all the parameters measured, data were subjected to an analysis of variance (ANOVA) using the JMP statistical package version 15.0 (SAS Institute, Inc., Cary, NC, USA). The seasons were tested for homogeneity using the Bartlett's test [47]. Non-significance indicated that the variances were homogenous; hence, combined analysis of variance was done. In cases where variance was heterogeneous, seasonal analysis was done. Where significant differences occurred, separation of means was done using the least significant difference (LSD) at the 0.05 level of significance.

Aboveground Biomass Production and Carbon Input
Three-way interaction of tillage × crop rotation × residue management was not significant (p > 0.05) in relation to the maize crop aboveground biomass and cumulative aboveground biomass. Similarly, two-way interactions of tillage × crop rotation, tillage × crop residue management and crop rotation × crop residue management were not significant (p > 0.05) with respect to maize crop aboveground biomass and cumulative aboveground biomass. Only the main effects of crop rotation significantly (p < 0.001) affected cumulative biomass production. The cumulative biomass ranged from 22.83 to 40.08 Mg ha −1 and followed decreasing order of MOM > MVM > MFM ( Figure 1).
There were no significant (p > 0.05) three-way and two-way interaction effects with respect to cumulative C input. Two-way interactions of tillage × crop rotation, tillage × crop residue management, Agronomy 2020, 10, 705 6 of 16 and crop rotation x crop residue management were not significant (p > 0.05) with respect to cumulative C input. The main effects of crop rotation and crop residue management significantly affected cumulative C input at p < 0.001 and p < 0.05, respectively, whereas tillage had no effect (p > 0.05) on cumulative C input in the study. The cumulative C in the MOM rotation was 78.1% and 14.5% higher when compared to MFM and MVM treatment, respectively ( Figure 2a). Retaining crop residues (12.2 Mg ha −1 ) accounted for more cumulative C input relative to biochar (110.1 Mgha −1 ) and residue removals (10.65 Mgha −1 ) (Figure 2b). There were no significant (p > 0.05) three-way and two-way interaction effects with respect to cumulative C input. Two-way interactions of tillage × crop rotation, tillage × crop residue management, and crop rotation x crop residue management were not significant (p > 0.05) with respect to cumulative C input. The main effects of crop rotation and crop residue management significantly affected cumulative C input at p < 0.001 and p < 0.05, respectively, whereas tillage had no effect (p > 0.05) on cumulative C input in the study. The cumulative C in the MOM rotation was 78.1% and 14.5% higher when compared to MFM and MVM treatment, respectively (  There were no significant (p > 0.05) three-way and two-way interaction effects with respect to cumulative C input. Two-way interactions of tillage × crop rotation, tillage × crop residue management, and crop rotation x crop residue management were not significant (p > 0.05) with respect to cumulative C input. The main effects of crop rotation and crop residue management significantly affected cumulative C input at p < 0.001 and p < 0.05, respectively, whereas tillage had no effect (p > 0.05) on cumulative C input in the study. The cumulative C in the MOM rotation was 78.1% and 14.5% higher when compared to MFM and MVM treatment, respectively (

Soil Organic Carbon
Four-way interaction of season × tillage × crop rotation × residue management was not significant with respect to SOC at both the 0-5 and 5-10 cm soil depth. Similarly, three-way interaction of tillage × crop rotation × crop residue management was not significant (p > 0.05) at both soil depths. The interaction of season × tillage was significant (p < 0.01) only at the 0-5 cm depth, while season × crop rotation was significant (p < 0.05) only at the 5-10 cm depth. The main effects of season (p < 0.01), tillage (p < 0.01), and crop residue management (p > 0.05) were significant with respect to SOC only at the 0-5 cm depth. However, only the main effects of crop rotation significantly affected (p < 0.01) SOC at the 5-10 cm depth.
At the 0-5 cm depth, NT had higher SOC compared to CT across all seasons ( Table 4). The highest and the least SOC were both observed in season 3 under NT (1.22%) and CT (0.86%), respectively. With respect to residue management, the highest SOC was observed under R+ (1.

Soil Organic Carbon
Four-way interaction of season × tillage × crop rotation x residue management was not significant with respect to SOC at both the 0-5 and 5-10 cm soil depth. Similarly, three-way interaction of tillage × crop rotation × crop residue management was not significant (p > 0.05) at both soil depths. The interaction of season × tillage was significant (p < 0.01) only at the 0-5 cm depth, while season x crop rotation was significant (p < 0.05) only at the 5-10 cm depth. The main effects of season (p < 0.01), tillage (p < 0.01), and crop residue management (p > 0.05) were significant with respect to SOC only at the 0-5 cm depth. However, only the main effects of crop rotation significantly affected (p < 0.01) SOC at the 5-10 cm depth.
At the 0-5 cm depth, NT had higher SOC compared to CT across all seasons ( Table 4). The highest and the least SOC were both observed in season 3 under NT (1.22%) and CT (0.86%), respectively. With respect to residue management, the highest SOC was observed under R+ (1.10%), followed by B (1.0%) and the least was in R-(0.96%) (Figure 3a). The highest SOC at the 5-10 cm depth was observed in season 4 under the MOM (1.05%) rotation, while season 3 under the MFM (0.75%) rotation had the least SOC (Figure 4). At the 5-10 cm depth, SOC was in the following order: MOM > MVM > MFM (Figure 3b).      In both the 2015/2016 and 2017/2018 summer seasons, the final height of maize was higher in CT relative to NT treatments ( Figure 6). The final plant height was 5.1% and 5% higher in CT relative to NT in 2015/2016 and 2017/2018, respectively, (Figure 6a,b).

Maize Yield Components and Grain Yield
In the 2015/2016 summer season, maize grain yield was significantly higher in plots with R− (5.18 Mg ha −1 ) compared to plots with R+ (4.26 Mg ha −1 ) ( Table 5). In the same season, B (4.42 Mg ha −1 ) application was not statistically different to both R− and R+. On the other hand, application of B (4.81 Mg ha −1 ) significantly increased maize grain yield compared to both R+ (3.38 Mg ha −1 ) and R− (3.08 Mg ha −1 ) in the 2016/2017 season, while the yield in the R+ and R− was not significantly different in the same season (Table 5).  In the 2015/2016 summer season, maize grain yield was significantly higher in plots with R-(5.18 Mg ha −1 ) compared to plots with R+ (4.26 Mg ha −1 ) ( Table 5). In the same season, B (4.42 Mg ha −1 ) application was not statistically different to both R-and R+. On the other hand, application of B (4.81 Mg ha −1 ) significantly increased maize grain yield compared to both R+ (3.38 Mg ha −1 ) and R-(3.08 Mg ha −1 ) in the 2016/2017 season, while the yield in the R+ and R-was not significantly different in the same season (Table 5). Table 5. Season, tillage, crop rotation, and residue management effects on the maize grain yield

Aboveground Biomass Production and Carbon Input from Plant Residues
The high cumulative biomass yields in the study highlight the importance of winter cover crops in the farming systems of the Eastern Cape Province. The increased biomass production in MOM rotation is possibly due to the fast-growing and tillering ability of white oats as compared to grazing vetch [48]. The study concurs with results obtained by Murungu et al. [19] and Muzangwa et al. [49], who reported that inclusion of a cereal cover crop produced more cumulative dry biomass compared to grazing vetch and fallow treatments. The high crop biomass achieved in this study translates to high C input. Garcia-Gonzalez et al. [50] also observed a high cumulative C input in a cereal cover crop compared to grazing vetch. In this study, the crop biomass yield of grazing vetch was low possibly due to the effect of planting time. The crop was planted at the end of May when the temperatures were low for the optimum germination and growth of the crop (Figure 7a). In the 2017 winter season, the crop was grazed by cows; this especially affected the grazing vetch biomass yield, the oat cover crop was able to regrow ( Figure 7b). However, the high crop biomass yield in plots with cover crops relative to fallow plots can also be the best approach to address the needs of the crop and livestock enterprises in mixed cropping systems in many smallholder farms in the central Eastern Cape Province. The crop biomass yield was more than the level regarded as necessary for the success of CA, which is 6-10 t ha −1 dry matter per year [51]. According to Ferreira et al. [52], the sustained input of the large quantities of biomass to the soil surface creates a positive impact on agricultural and environmental sustainability.

Soil Organic Carbon
The lack of various and significant interactions in this study suggest that SOC is not affected by CA in the short term. This was also suggested by Blanco-Canqui et al. [53] and Poeplau and Don [54]. However, tilling the soil in the Province hastens the loss of SOC through excessive removal of crop biomass after harvest and higher decomposition rate due to increased microbial activity at the soil surface. This is evidenced by the observed higher SOC under NT relative to CT at the 0-5 cm soil depth (Table 4) in all six cropping seasons. The gradual decrease SOC in the CT plots between the winter 2015 season and the summer 2017/2018 season is possibly because of the disruption in C cycling caused by tillage. These results concur with studies by Govaerts et al. [55] and Ella et al. [56]. The higher amount of SOC in the top 0-5 cm compared to lower depths was due to the presence of crop residues on the soil surface. Slight increase of SOC at the soil surface is necessary to control erosion, water infiltration, and conservation of nutrients, and is related with the soil quality [57].
In this study, R+ increased SOC at the 0-5 cm depth relative to both B and R-treatments. Higher SOC in the R+ treatment can be attributed to the presence of residues which are broken down and decomposed by soil fauna [58]. Similar results were observed by Kassam et al. [12], Dalal et al. [59], Li et al. [60], Rusinamhodzi et al. [61], and Turmel et al. [31]. The low SOC in B treatments can be attributed to the low quantities applied in the study ( Table 2). The response of soil to biochar amended is dependent on the amount and quality of biochar used [36]. According to Monnie [62], corn cob biochar applied at rates that are 20 t ha −1 did not show any significant soil improvements. Furthermore, the low SOC content in B treatments can possibly be explained by preferential lateral movement of biochar due to low mass per volume and vertical migration of biochar due to solute transport and bioturbation along the soil profile [63].

Maize Grain Yield and Yield Components
The lack of significant differences between NT and CT in terms of maize yield shows that the smallholder farmers in the EC can potentially switch from soil tilling without any significant yield penalty. Paudel et al. [28] reported similar results under rain-fed conditions in Nepal. According to Giller et al. [64] and Thierfelder et al. [65], benefits of CA are not always immediate. For example, in arid and semiarid conditions, CA may increase crop yield through improving soil fertility through C sequestration and water conservation [66]. On the other hand, it can have negative impacts on crop yield by altering soil physiochemical and biological conditions, such as decreasing soil temperatures in temperate areas and seasons with low temperature, thus aggravating weeds [23].

Soil Organic Carbon
The lack of various and significant interactions in this study suggest that SOC is not affected by CA in the short term. This was also suggested by Blanco-Canqui et al. [53] and Poeplau and Don [54]. However, tilling the soil in the Province hastens the loss of SOC through excessive removal of crop biomass after harvest and higher decomposition rate due to increased microbial activity at the soil surface. This is evidenced by the observed higher SOC under NT relative to CT at the 0-5 cm soil depth (Table 4) in all six cropping seasons. The gradual decrease SOC in the CT plots between the winter 2015 season and the summer 2017/2018 season is possibly because of the disruption in C cycling caused by tillage. These results concur with studies by Govaerts et al. [55] and Ella et al. [56]. The higher amount of SOC in the top 0-5 cm compared to lower depths was due to the presence of crop residues on the soil surface. Slight increase of SOC at the soil surface is necessary to control erosion, water infiltration, and conservation of nutrients, and is related with the soil quality [57].
In this study, R+ increased SOC at the 0-5 cm depth relative to both B and R− treatments. Higher SOC in the R+ treatment can be attributed to the presence of residues which are broken down and decomposed by soil fauna [58]. Similar results were observed by Kassam et al. [12], Dalal et al. [59], Li et al. [60], Rusinamhodzi et al. [61], and Turmel et al. [31]. The low SOC in B treatments can be attributed to the low quantities applied in the study ( Table 2). The response of soil to biochar amended is dependent on the amount and quality of biochar used [36]. According to Monnie [62], corn cob biochar applied at rates that are 20 t ha −1 did not show any significant soil improvements. Furthermore, the low SOC content in B treatments can possibly be explained by preferential lateral movement of biochar due to low mass per volume and vertical migration of biochar due to solute transport and bioturbation along the soil profile [63].

Maize Grain Yield and Yield Components
The lack of significant differences between NT and CT in terms of maize yield shows that the smallholder farmers in the EC can potentially switch from soil tilling without any significant yield penalty. Paudel et al. [28] reported similar results under rain-fed conditions in Nepal. According to Giller et al. [64] and Thierfelder et al. [65], benefits of CA are not always immediate. For example, in arid and semiarid conditions, CA may increase crop yield through improving soil fertility through C sequestration and water conservation [66]. On the other hand, it can have negative impacts on crop yield by altering soil physiochemical and biological conditions, such as decreasing soil temperatures in temperate areas and seasons with low temperature, thus aggravating weeds [23].
The low maize grain yield in the plots with residues relative to R− and B treatments in the 2015/2016 summer season can possibly be due to the high pest and disease problem that was encountered in these treatments. Cut worms, rodents, and maize stock borer attacks were high, especially in plots with a combination of NT and R+. Wolfe and Eckert [67] and Al-Kaisi and Kwaw-Mensah [68] also reported that excessive crop residues left on the soil surface, such as in NT, can potentially reduce plant emergence, N mineralization, and crop growth, resulting in lower crop yields. In the 2016/2017 summer season, the significantly higher maize grain yields in the B treatments can be explained by the ability of biochar to adsorb nutrients and release them into the soil for uptake by plants [69]. The maize yields in the same year were low relative to the other seasons because of a combination of the low rainfall received during the crop growing season and the effect of pest incidences on crop stand. Birds and rodents damaged the crops, especially in the plots with R+. The residues provided cover for the pests and maintained a moist soil surface that made it easy for the crows (Corvus) to dig out the germinating maize seeds and crops. The attacks were so severe that much of the crop had to be replanted well after the maize planting dates (end of January). The inconsistent maize yields across seasons in this study can be attributed to the variability of rainfall. This is also corroborated by Cairns et al. [70], who reported that rainfall erraticism causes yields in a CA system not to stabilise as expected.

Conclusions
This study showed that the adoption of crop rotations and residue management practices results in an increase in crop biomass, which is also positively correlated C input. Furthermore, changes in SOC due to management practices occurred too slowly to reflect in maize grain yields in the short-term. Generally, the variability of maize grain yield in the three summer seasons makes the results inconclusive in the short term. However, the lack of significant differences between CA and CT means that adoption of CA in the Eastern Cape Province neither leads to immediate benefits nor losses in maize yield. Therefore, smallholder farmers in the Eastern Cape Province can potentially switch from CT to NT without any significant yield penalty. Addition of biochar did not result in an obvious improvement in both soil C input and crop yield in the current study. However, the fact that it performed more or less than retaining crop residues relative to crop productivity warrants further studies, especially when significantly large quantities are accrued. Therefore, continued evaluation of treatments for a medium-to long-term period is highly recommended.
Author Contributions: P.N. designed, setup, collected data, analyzed data, and wrote the manuscript. C.C. designed, setup, collected data, analyzed data, and wrote the manuscript. T.A. wrote the manuscript. All authors have read and agreed to the published version of the manuscript.