Effects of Tillage Systems and Cropping Patterns on Soil Physical Properties in Mozambique

Conservation agriculture (CA) practices are advocated to reduce soil degradation, resulting in more sustainable food production as compared to conventional tillage (CT). In this study, the short-term effects of two tillage systems in combination with cropping patterns on selected soil physical parameters on four experimental sites in Mozambique were studied. The study sites differ according to their climatic conditions, soil types, and crop adaptation. Tillage systems evaluated were CA and CT, while the cropping pattern had four levels of sole cropping and three levels of intercropping. In general, soil physical properties showed significant changes due to the tillage systems, but the cropping pattern and their interaction with tillage systems did not yield significant impacts on the soil physical properties. CA increased bulk density, penetration resistance, and saturated hydraulic conductivity as compared to CT. A significant difference due to the tillage system was observed across the four sites, and in general, evaporation was higher in CT compared to CA. The presence of crop residues in CA contributed to lower evaporation. Thus, in the short term, CA practices could be a sustainable option to conserve soil water through higher infiltration and less evaporation.


Introduction
In sub-Saharan Africa (SSA), unsustainable farming practices, such as excessive soil tillage with no soil cover, have led to a decrease in soil fertility, accelerated erosion, and degradation of arable land. In Mozambique, around 68% of the population live in rural areas, and up to 90% are smallholder farmers [1]. The most common tillage system practiced by smallholder farmers in Mozambique is where hand hoes are used to till the soil, with crop residue collection and burning during land preparation. This is referred to as conventional tillage (CT) throughout this study. The use of inappropriate tillage systems (i.e., CT) has led to the deterioration of soil quality through decay in organic matter, disruption of soil structure, and decrease in aggregate stability. This resulted in lower macroporosity, thus promoting soil erosion, organic matter loss, and fertility depletion [2][3][4][5]. The development of sustainable management practices to reduce the physical degradation of soil should, therefore, be a high priority. Conservation agriculture (CA) practices are proposed to minimize soil degradation and improve soil quality. CA is defined as a tillage system where at least 30% of crop residues are left on the field surface, and it is a conservation practice to reduce soil erosion, improve water use efficiency, and regulate upper soil temperature [6,7]. CA integrates three principles, namely minimal soil disturbance, crop residue retention, and crop diversification through rotation or intercropping [6,8,9]. These CA practices are advocated to increase the microporosity in relation to macroporosity, consequently    [22][23][24][25] where the experiments were conducted.  [26]). ** Based on USDA textural soil classification (USDA, [27]).

Experimental Design and Layout
The trials were conducted over two cropping seasons, i.e., 2016-2017 to 2017-2018, to evaluate the response on soil physical properties (Table 2). At each location, a field experiment was set up as a two factor split-plot experiment in a randomized complete block design and replicated four times. The first order factor was tillage systems, i.e., two levels-conventional tillage (CT) and conservation agriculture (CA), with 26 m × 37.5 m (975 m, including 2 m buffer zones between blocks and 1 m buffer zones between plots) main plots, and the second order factor was cropping patterns (seven levels; see Table 2) with 5.0 m × 4.5 m (22.5 m 2 , excluding any buffer zones) subplots (see Figure 2). The four sites had the same experimental design, and the treatments were applied to the same plots during the two years of study. For the CA system, glyphosate was applied at a rate of 2.5 L ha −1 (480 g L −1 of isopropylamine salt active ingredient) two weeks before planting. Under the CA system, the soil was disturbed only with a dibble stick planter when the seeds were planted, while under the CT system, soil was disturbed with a hand hoe to a depth of approximately 15 cm for seedbed preparation and planting. For the CT system, crop residues were removed, and for the CA system, the surface area was covered with at least 30% of the previous season harvested crop residues (2-3 t ha −1 ). No fertilizers were applied, since this is not common practice amongst the smallholders.
The sowing and harvesting dates were respectively in November and May every year for all crops, except for pigeon pea, due to a longer growing period, which was harvested in July of each year. Maize and pigeon pea seed were sown in rows at the recommended spacing of 75 cm between rows and 30 cm within rows, while the other legumes varied from 50 cm × 30 cm for common beans, cowpea, and groundnut, and 50 cm × 10 cm for soybean. The arrangement of row intercropping was 1:1 (one row of maize alternated by one row of legume). Maize and legumes were sown simultaneously as a sole crop or intercropped in alternative rows. Two seeds were sown per hill and later thinned to one per hill; gapping was done five days after emergence. Postemergence weeds were controlled with hand hoeing.

Measurement of Soil Properties
Soil physical properties were measured at the end of the second growing season (July 2018). Undisturbed core samples were collected from randomly located points in all plots for bulk density determination. Disturbed samples were collected at the same locations to measure particle size distribution. Saturated hydraulic conductivity, penetration resistance, and evaporation were determined in situ using methods discussed below.

Particle Size Distribution and Gravel
Disturbed composite soil samples of 50 g were from topsoil horizon of selected plots, oven-dried at 105 • C, and passed through a 2 mm sieve. The hydrometer method was used for the determination of particle size distribution [27]: sand (2.0-0.05 mm); silt (0.05-0.002 mm), and clay (<0.002 mm). This method is based on the contrasting settling velocities of the particle sizes within a water column. The textural class was deducted using the United States Department of Agriculture (USDA) textural triangle [27]. The gravel retained on the 2 mm sieve was used to estimate the percentage gravel in a composite soil sample.

Bulk Density
Undisturbed core samples were collected at the end of the second growing season (July 2018) from four randomly located points in all subplots. Cores with a 2.5 cm radius and 5 cm in length were used for sampling at Lichinga, Mutuali, and Gurúè, while slightly larger cores (3.5 cm radius and 12 cm length) were used in Nhacoongo. Bulk density was determined using the undisturbed core sampling method of Okalebo et al. [28]. Bulk density, Db (g·cm −3 ), was calculated as the relationship between dry soil mass and total ring volume using the equation below: where: Db = Bulk density (g·cm −3 ); M = Dry weight of the core sample (g); V = Volume of the soil core (cm 3 ).

Penetration Resistance
At the end of the second cropping season (July 2018), penetration resistance of the soil of each plot at all four sites was measured using a manual cone penetrometer (hand penetrometer Eijkelkamp 06.01.14). This instrument measures the soil resistance to penetration in Newton (N). Measurements were taken at 10 randomly selected points per plot within each plot to a depth of 15 cm, and the average value was calculated. The penetration resistance measurements were made by the same person in all four sites to ensure that the pressures were applied consistently on the penetrometer from one place to another using the equation below.
where: PR = penetration resistance (N·cm −2 then converted to kPa); F = normal force (N); A = base area of the cone (cm 2 ).

Saturated Hydraulic Conductivity
Saturated hydraulic conductivity was determined using the double-ring falling head infiltrometer method. The double ring infiltrometer method consisted of a 10 cm inner ring diameter and a 30 cm outer ring diameter inserted approximately 10 cm into the soil at two randomly selected points in selected plots at the experimental sites. Both the outer and inner rings were then saturated with water. The water level in the outer ring was kept above that of the inner ring to avoid lateral drainage from the inner ring. The time for water to infiltrate 2 cm was then recorded (from 10 to 8 cm above the soil surface). The infiltration test was repeated until the infiltration time became nearly constant (or less than a 5% decline from the previous infiltration time). The saturated hydraulic conductivity (K s ) was then calculated using the modified Bouwer and Rice [29] equation below.
where: K s = saturated hydraulic conductivity; L = thickness of horizon (L); t = time until constant infiltration rate was obtained (T); h 0 and h 1 = head of water above the surface before and at the start of the test and after t, respectively.

Evaporation
Evaporation (E) was measured using microlysimeters (MLs) following the second cropping season (August, September, October, and November of 2018) according to the method proposed by Flumignam et al. [30]. The MLs ( Figure 3B) consist of rigid polyvinyl chloride (PVC) pipes that were 15 cm long with 10 cm inner diameter. The wall thickness of the pipes was 0.25 cm. One of the sides was sharpened to ease penetration into the soil. A cap was inserted at the bottom to prevent deep drainage ( Figure 3A). A slightly larger envelope was then inserted into the soil used for hosting the MLs ( Figure 3C). The bottom of the MLs was sealed and raised 0.5 cm to eliminate the effects of drainage and runoff.
Agriculture 2020, 10, x FOR PEER REVIEW  8 of 17 and a 30 cm outer ring diameter inserted approximately 10 cm into the soil at two randomly selected points in selected plots at the experimental sites. Both the outer and inner rings were then saturated with water. The water level in the outer ring was kept above that of the inner ring to avoid lateral drainage from the inner ring. The time for water to infiltrate 2 cm was then recorded (from 10 to 8 cm above the soil surface). The infiltration test was repeated until the infiltration time became nearly constant (or less than a 5% decline from the previous infiltration time). The saturated hydraulic conductivity (Ks) was then calculated using the modified Bouwer and Rice [29] equation below.
where: Ks = saturated hydraulic conductivity; L = thickness of horizon (L); t = time until constant infiltration rate was obtained (T); ℎ and ℎ = head of water above the surface before and at the start of the test and after , respectively.

Evaporation
Evaporation (E) was measured using microlysimeters (MLs) following the second cropping season (August, September, October, and November of 2018) according to the method proposed by Flumignam et al. [30]. The MLs (Figures 3B) consist of rigid polyvinyl chloride (PVC) pipes that were 15 cm long with 10 cm inner diameter. The wall thickness of the pipes was 0.25 cm. One of the sides was sharpened to ease penetration into the soil. A cap was inserted at the bottom to prevent deep drainage ( Figure 3A). A slightly larger envelope was then inserted into the soil used for hosting the MLs ( Figure 3C). The bottom of the MLs was sealed and raised 0.5 cm to eliminate the effects of drainage and runoff. The MLs were saturated with water (approximately 100 mL), and the wet soil was weighed on a 2 g resolution electronic scale. Water evaporation was measured as the reduction in weight from the wet soil in the MLs on a daily basis for a week. This was repeated four times aligned to the site visits. The difference in weight between measurements was attributed to evaporation and was calculated using the equation below.  The MLs were saturated with water (approximately 100 mL), and the wet soil was weighed on a 2 g resolution electronic scale. Water evaporation was measured as the reduction in weight from the wet soil in the MLs on a daily basis for a week. This was repeated four times aligned to the site visits. The difference in weight between measurements was attributed to evaporation and was calculated using the equation below.
where: E = water evaporation from the MLs (mm); ∆M MLs = MLs variation in mass (kg); A MLs = MLs surface area with a fixed value of 0.0785 m 2 .

Data Analysis
Data collected were tested for normality with the Shapiro-Wilk test and for homogeneity with the Levene test before analysis of variance (ANOVA) with XLSTAT v 17.3 software for Excel [31]. Treatment means were separated by Tukey's multicomparison test, and differences were reported at the 5% probability level. The interaction between the tillage system and cropping pattern was given separately for those attributes that had a significant influence, and these were described accordingly.

Particle Size Distribution and Gravel Content
The particle size distribution and resulting textural classes of the soil samples from the Nhacoongo, Mutuali, Lichinga, and Gurúè experimental sites are presented in Table 3. Soil texture is unlikely to be influenced by the treatments. It is, however, important to consider the texture when the impact on other physical properties are assessed. It is also important to consider the texture if recommendations and extrapolations are made to other sites. Soils from Nhacoongo are loamy sand, Mutuali and Gurúè are sandy loam, and Lichinga is sandy clay loam. In general, the highest gravel content was observed in Gurúè (5.00%), followed by Lichinga (3.28%), Mutuali (1.20%), and Nhacoongo (0.50%).

Bulk Density
The Db was influenced significantly by the tillage system at Nhacoongo, Mutuali, and Lichinga but not at Gurúè (Table 4). However, at all four sites, CA was associated with increases in Db of 1.29% for Nhacoongo, 2.63% for Mutuali, 7.50% for Lichinga, and 1.92% for Gurúè in comparison to under CT. On the other hand, the cropping pattern had no significant influence on Db at all four sites varying from 1.03 to 1.57 g·cm −3 . The interaction between the tillage system and cropping pattern significantly influenced Db only in Nhacoongo and Lichinga but not in Mutuali and Gurúè.

Penetration Resistance
The effect of tillage systems on PR was significant only at Nhacoongo and Gurúè (Table 4). However, PR values across the four study sites were consistently higher under the CA system (Nhacoongo = 90.4%, Mutuali = 4.5%, Lichinga = 3.1%, and Gurúè = 12.2%) compared to the CT system. Cropping patterns, however, had a significant influence on PR only in Mutuali (Figure 4). The interaction effects between tillage systems and the cropping patterns had no significant influence on PR in Mutuali, Lichinga, and Gurúè. Overall, the interaction produced a higher PR under CA than under CT.

Saturated Hydraulic Conductivity
The Ks was significantly affected (p ≤ 0.05) by the tillage system in Nhacoongo, Lichinga, and Gurúè but not in Mutuali (Table 4). Generally, average Ks values were higher under the CA system (Nhacoongo = 14.7%, Mutuali = 13.9%, Lichinga = 65.5% and Gurúè = 74.2%) than under the CT system. The current study did not show any significant cropping pattern on Ks across the four sites ranging from 335 to 773 mm h −1 . This was also the case with the interaction of tillage systems by cropping patterns.

Saturated Hydraulic Conductivity
The Ks was significantly affected (p ≤ 0.05) by the tillage system in Nhacoongo, Lichinga, and Gurúè but not in Mutuali (Table 4). Generally, average Ks values were higher under the CA system (Nhacoongo = 14.7%, Mutuali = 13.9%, Lichinga = 65.5% and Gurúè = 74.2%) than under the CT system. The current study did not show any significant cropping pattern on Ks across the four sites ranging from 335 to 773 mm h −1 . This was also the case with the interaction of tillage systems by cropping patterns.

Evaporation
A significant effect of tillage system on evaporation was obtained in some months at Nhacoongo, Mutuali, Lichinga, and Gurúè (Figure 5a-d). Significant differences in evaporation were at Nhacoongo and Gurúè, recorded in September, October, and November; at Lichinga in August, September, and November; and at Mutuali only in November. In most cases, evaporation was higher under CT than CA. Across the four sites, lower evaporation was recorded in the sandy loam soils of Gurúè (CA = 3.42 and CT = 3.61 mm) and Mutuali (CA = 6.22 and CT = 6.57 mm) than in the loamy sand soil of Nhacoongo (CA = 8.60 and CT = 11.27 mm) and the sandy clay loam soil of Lichinga (CA = 11.10 and CT = 10.98 mm). Unexpectedly, in August, higher evaporation values were obtained under CA than CT in Nhacoongo, Mutuali, and Gurúè.

Bulk Density
Measurements made in the four experimental sites showed significant differences in Db as a result of tillage system. The CA system increased Db, resulting in increased compaction and less aeration compared to the CT system. Several studies (Fabrizzi et al. [2], Rashidi and Keshavarzpour [12], Malecka et al. [13], and Dembele [32]) reported greater Db values under CA than CT. Management practices adopted for CT (ploughing) tend to disturb the soil to a 30 cm depth, creating more pore space. Furthermore, the absence of tillage (CA system) increases Db because the soil becomes more compacted after only two cropping seasons. In the long term, decomposition of crop residues and organic matter accumulation might result in a decrease in Db [33,34]. It is clear that the expected long-term benefit associated with the higher organic matter has not yet been obtained. Across the four sites, the highest Db was recorded in the loamy sand soil of Nhacoongo (CA = 1.57 and CT = 1.55 g·cm −3 ), followed by sandy clay loam soil of Lichinga (CA = 1.29 and CT = 1.20 g·cm −3 ), sandy loam soil of Mutuali (CA = 1.17 and CT=1.14 g·cm −3 ), and the sandy loam soil of Gurúè (CA = 1.06 and CT = 1.04 g·cm −3 ). This was not unexpected and can be attributed to the fact that soil texture (loamy sand with >80% sand) differed from that in Mutuali and Gurúè (sandy loam with less than 75% sand). Sandy loam soils tend to have higher Db than more clayey variants [35]. At Lichinga, the soil was sandy clay loam with an average of 45% clay. Lower clay contents contribute to higher compaction, which explains the high Db in Nhacoongo (clay content < 9%) compared to other sites (clay content > 14%). Increased Db in sandy soil as compared to clayey soils was also recorded [36]. This can be ascribed firstly to the relatively high clay contents, which are not prone to compaction, and secondly to the short study period.
Cropping pattern (CP) had no significant influence on Db at all four sites. It was expected that plots with legumes and with more crop residues accumulation would have a positive effect on soil physical quality by improving the structure and decreased Db. However, no clear trends were observed between different cropping patterns. Similarly, Oliveira et al. [37] found no significant difference between beans and millet intercropping effects on Db.
A significant interaction effect between the tillage systems and CP on Db was recorded in Nhacoongo and Lichinga but not in Mutuali and Gurúè. The significant differences could be attributed to single outliers, and no clear trend was observed. Several other researchers have reported similar contradictory results. Nabanita [38] and Alam et al. [39] found insignificant interaction effects on Db between tillage systems and cropping patterns, whereas Rusu et al. [40] obtained significant interaction effects. The overall trend of tillage systems and cropping pattern interaction effects on Db under CA was higher than that under CT.

Penetration Resistance
PR values across the four study sites were consistently higher under CA systems compared to the CT system. Therefore, the CA system offered more resistance to root development than CT. This concurs with the results of Rashidi and Keshavarzpour [12] and Çelik [41]. The higher PR under CA can be due to almost no soil disturbance, which causes compaction and a decrease of macropore volume on the surface. Ploughing loosens the soil, leading to a lower PR. Across the four sites, the highest PR was recorded in Gurúè (CA = 2692 and CT = The cropping pattern influenced PR significantly in Mutuali only. The interaction effects between tillage systems and CP had no significant influence on PR in Mutuali, Lichinga, and Gurúè. Across the four sites, higher PR values were observed under the CA system with CP interaction than under the CT system with CP interaction. According to some literature, the interaction between tillage systems and CP significantly influences PR [42]. At the same time, other researchers reported insignificant interaction effects, especially with short term studies [36].

Saturated Hydraulic Conductivity
In general, the hydraulic conductivity on all the sites was high (>300 mm h −1 ), which exceeds typical rainfall intensity. The Ks values were higher under CA systems than CT. Across the four sites, higher Ks values were recorded in the loamy sand soil of Nhacoongo (CA = 743 and CT = 648 mm h −1 ), while the lowest Ks values were recorded in the sandy loam soil of Mutuali (CA = 517 and CT = 454 mm h −1 ). Singh et al. [43] emphasized that the lower Ks values recorded under CT could be ascribed to the destruction of aggregates and reduction of macroporosity in ploughed soils. The positive impact of CA on Ks was somewhat surprising, as the general theory is that an increase in compaction and bulk density is typically associated with lower conductivity. The accumulation of organic matter associated with reduced physical aeration and accumulation of crop residues could be attributed to the increased conductivity. Organic matter and continuous root channels will increase the macroporosity under the CA, thereby increasing Ks. These results are consistent with those reported by Mcgarry et al. [44] and Kumar et al. [45], who found an increase in Ks under CA, probably due to the retention of crop residues that modified soil structure, hence aggregate stability. Although our study sites had high inherent Ks, the short term impact of CA in increasing Ks can be beneficial in other areas with low infiltration rates.

Evaporation
The tillage system had a significant influence on evaporation in some months at Nhacoongo, Mutuali, Lichinga, and Gurúè: Significant differences in evaporation were recorded at Nhacoongo and Gurúè in September, October, and November; at Lichinga in August, September, and November; and at Mutuali only in November. In most instances, evaporation was higher under the CT system than CA. These results are in agreement with those of Dembele [32], Singh et al. [46], and Thierfelder et al. [47], who reported improved water use efficiency due to the application of CA instead of CT. The higher evaporation under the CT system reflects the disadvantages of ploughing, which increases soil aeration and decay of soil aggregates, leading to enhanced water loss by evaporation. Conservation agriculture systems are effective in reducing evaporation and increase water holding capacity due to the presence of crop residues, which act as a protective skin for the soil by intercepting a great part of the incident solar radiation, creating thus a wetter microclimate at the soil surface, which enhances crop water availability. These findings are in agreement with those of Hobbs et al. [8], Alam et al. [39], and van Donk et al. [48], who reported that soils dry slowly on the surface due to the presence of crop residues under the CA system. The reduced evaporation is ascribed to shading, which moderates surface soil temperature and wind effects. Crop residues in CA systems also prevent the impact of direct rainwater on the soil, thus avoiding the formation of surface sealing, in turn maintaining water conductivity. Crop residues can reduce evaporation by 10-20% [49]. However, contradictory results were noted in Nhacoongo and Gurúè in August, Mutuali in August and October, and Lichinga in August, with higher evaporation recorded under CA than CT. The inconsistent trends recorded in August can be attributed to the fact that crop residues delay heat loss from the surface during winter, and this could promote warming during summer, leading to increased evaporation [50,51]. The inconsistency in the results could also be attributed to the use of microlysimeters, since they might not contain sufficient soil water to maintain realistic evaporation.

Conclusions
This study revealed some significant differences related to the tillage system across the four experimental sites in Mozambique after two years of employing different cultivation practices. Higher values of Db, PR, and Ks were recorded under the CA system when compared to the CT system. A change in soil physical properties usually requires a significant accumulation of organic matter, which requires a long time to achieve. The increased compaction, as indicated by Db and PR, is ascribed to the absence of tillage of the soil, whereas the increase in conductivity related to increased macroporosity is associated with the accumulation of organic matter through reduced tillage and crop residues. Evaporation at the soil surface is higher in the CT system than in the CA system. The decline in evaporation under CA, when compared to CT, is a short term benefit of CA that should be explored in the future to promote CA adoption under smallholder farmers. In the present study, tillage methods that significantly affected the soil physical properties due to decreased evaporation by reducing water loss were noted. Therefore, smallholder farmers should be encouraged to adopt CA practices to increase water availability in the soil during the crop growth period. Two years of experimentations were not enough to develop clear trends of sole cropping compared to intercropping. Thus, there is a need to conduct long-term studies to quantify cropping pattern effects on soil physical properties.