Effects of tillage on runoff from a bare clayey soil on a semi-arid ecotope in the Limpopo Province of South Africa

Runoff constitutes one of the major water losses from agricultural fields in semi-arid areas. However, by adopting appropriate soil management practices, the runoff can be harnessed for improving crop yields. The main objective of this study was to quantify rainfall-runoff relationships under in-field rainwater harvesting (IRWH) using simulated rainfall, and to compare these results to those obtained with annually tilled conventional tillage (CON) (control). IRWH is a special type of no-till (NT) crop production practice that promotes runoff from a crusted runoff strip into basins where the water infiltrates beyond evaporation but is available for crop use. Runoff was related to time to runoff, total runoff, final runoff rate and runoff coefficient. This experiment demonstrated that by adopting IRWH production technique smallholder farmers could harness an additional 45.54 m3∙ha-1 of water compared to the CON system. The extra water harvested could meet about 1% of maize water requirements.


Introduction
Water scarcity is a major constraint in semi-arid areas, leading to a natural focus on in-field rainwater conservation (Jensen et al., 2003).Runoff constitutes one of the major water losses in semi-arid areas, causing the loss of valuable water, soil and nutrients (Schiettecatte et al., 2005;Vahabi and Mahdian, 2008).Research results from semi-arid regions have shown that runoff losses can be as high as 50% of the rainfall on bare untilled lands (Stroosnijder, 2003).Excessive runoff not only limits the water available for crop production, but also constitutes an erosion hazard (Rao et al., 1998).However, the goal of water harvesting (WH) is to convert runoff water 'loss' into productive use by storing it in basins where it can infiltrate and become available for crop use (Hensley et al., 2000).Water harvesting, based on the collection of runoff from a prepared catchment surface and its storage in the adjacent crop area, has been used successfully for crop and tree improvement in other parts of the world (Li et al., 2000;Li and Gong, 2002;Schiettecatte et al., 2005).Among the various water harvesting technologies available, in-field rainwater harvesting (IRWH) has been shown to be an efficient water conservation crop production technique especially appropriate for rural poor households (Botha et al., 2003).It was therefore selected for this study.
In-field rainwater harvesting (IRWH) is a relatively new in-situ water harvesting technique developed in South Africa (Hensley et al., 2000).The IRWH system is regarded as a special form of water harvesting.It is also known as minicatchment runoff farming (Oweis et al., 1999).IRWH has been tested in South Africa on clay and duplex soils in semi-arid areas where it has given maize yield increases of between 25% and 50% compared to conventional tillage practices (Hensley et al., 2000;Botha et al., 2003).The technique is illustrated in Fig. 2. The technique combines the advantages of water harvesting from the no-till, flat, crusted runoff strip, and decreased evaporation from the deeply infiltrating runoff water which accumulates in the basin (Hensley et al., 2000).Thus the IRWH partitions rainfall into runoff (on the no-till runoff strip) and run-on (in the basin).
Runoff is an important water balance component in semiarid environments (Bennie and Hensley, 2001).Zere et al. (2005) used runoff data to simulate the long-term yields for crops planted on conventional tilled soil and under in-field rainwater harvesting on Glen/Tukulu ecotope.They concluded that the PutuRun Model can be used with reasonable confidence, after calibration, to simulate long-term runoff on conventionally tilled and bare untilled plots on the Glen/Tukulu ecotopes, using daily rainfall data.In Ethiopia, Welderufael et al. (2008) used 2-year runoff data to predict a maize yield increase of between 25% and 35% under IRWH compared with conventional tillage.Some rainfall-runoff relationships from semiarid ecotopes are summarised in Table 1.The research results presented in Table 1 clearly indicate that the IRWH technique is a promising soil management technology under certain soil conditions, and that it needs to be explored further to promote sustainable crop production in marginal areas.
Results obtained with water harvesting techniques are not always transferable from one set of conditions (i.e. from a particular ecotope) to another, because of the differences in local characteristics (Ojasvi et al., 1999).An ecotope has been described as an area of land on which the climate, topography and soil are reasonably homogenous (McVicar et al., 1974).Considerable IRWH research has been done on specific ecotopes in the Free State Province of South Africa.It is, however, uncertain how the technique will perform on Ferralsols in the Limpopo province.
The agricultural industry in the Limpopo Province is made up of 2 sectors, namely, the large scale commercial and the smallholder farming systems.There were 5 000 commercial farming units and 273 000 small-scale farmers operating in  (Statistics South Africa, 2000).It was estimated that agriculture contributed 4% to the gross geographical product (GGP) of Limpopo Province in 2002 and the smallholder sector provided about 43% of total agriculture income in the province.Smallholder farmers operate from the former homelands.However, it could not be established how many of these farmers cultivate on the Ferralsols.Nevertheless, previous studies indicate that the majority of small-scale farmers in the northern Vhembe District (study area) depend on the Ferralsols for crop and fruit farming (Simalenga and Mantsha, 2003).Vhembe District is largely rural and is one of the 5 districts in the Limpopo Province where a large village population relies on agriculture for their livelihood.The area is marginal for crop production because of relatively low and erratic rainfall.In Vhembe District research results have shown that the average yield of maize was 12 bags per ha, in a good year, and about 5 bags per ha in a bad year.Poverty and food insecurity is therefore a major challenge facing smallholder farmers in the province.One of the recommendations of Simalenga and Manstha's (2003) study was that the farmers need to adopt water management strategies to mitigate the effects of bad weather.
The hypothesis is that the IRWH technique will increase crop yield on the Limpopo Ferralsols, compared to conventional tillage, due to the fact that enhanced runoff on the flat, crusted no-till runoff strip shown in Fig. 2 will result in a large fraction of the rainfall being stored in the basins.The stored water will result in a higher rainfall-use efficiency than with annually tilled conventional tillage (CON), which will have ex-field runoff losses as well as higher water losses due to evaporation.
The main objective of this research was to quantify rainfall-runoff relationships on a clayey soil under IRWH using simulated rainfall of various intensities, and to compare these results to those obtained with CON.The study will improve our understanding of rainfall-runoff processes under 2 tillage practices and thereby provide insight into the constraints and potential of IRWH production technique on this ecotope.

Site description
The study was conducted at the University of Venda experimental farm (approximate latitude 22 o 58'S and longitude 30 o 26' E; altitude 596 m a m.s.l.) located at about 2 km west of Thohoyandou town in the Limpopo Province of South Africa.Thohoyandou falls under Thulamela Municipality in Vhembe region.A map showing the location of the study area is presented in Fig. 1.The study area falls within the eastern part The highest evaporative demand also occurs from October to March.The mean annual aridity index (AI) is 0.52, causing the area to fall on the borderline between semi-arid and sub-humid according to the UNESCO classification criteria.Average rainfall is about 780 mm (range: 281 -1 239 mm).Three soil profiles were dug to depth of 1 500 mm on a research block measuring 100 m x 50 m.Soil profiles were described and classified locally as Hutton form, and belong to Land Type Ab179 (Soil Classification Working Group, 1991).
The approximate equivalent World Reference Base is Rhodic Ferralsol (WRB, 2006).Soil samples were taken from each demarcated horizon and analysed.Average results are reported.The soil profile is uniform and deeply weathered to a depth of >1 500 mm with no depth limiting material.The average clay content in the profile is 60% and is dominated by kaolinite (99%) clay minerals.The soil is also characterised by a relatively high silt content (>30%).Organic carbon is higher (1.7%) in the surface horizons than in the subsoil horizons (0.5%) (Table 3).Exchangeable cations are low and are dominated by Ca followed by Mg, Na and K, in that order.Despite their high clay content, these soils have low bulk density, ranging between 1.1 and 1.2 Mg•m -3 (Table 3).The low bulk density could partly be explained by high organic matter in the surface horizons and by the microstructure typical of Ferralsols (Medina et al., 2002).Water holding capacity is considered high due to the high clay content.<----------(cmol c (+) kg 1 )----------

Description of IRWH system
Two distinct areas can be seen from the diagram (Fig. 2): a 2-m catchment area or runoff strip and a 1-m basin or collection area.The runoff area is sloped towards the basins to direct the surface water into them.Runoff created in this way is called in-field, which differs significantly from ex-field runoff that occurs with CON; in-field runoff can be harnessed positively and used to enhance water conservation and therefore sustainability.Raindrop impact on the runoff strip causes surface compaction and therefore contributes to the formation of soil crusts, which stimulates in-field runoff.No-till is practised on the runoff strip to maintain a smooth surface (Hensley et al., 2000).The use of a 2:1 surface area ratio between the runoff and basin area is based on field experience with crops in a semi-arid environment.Crops are planted in tramlines (1 m x 2 m wide) along the basins or collection area.Tramline planting is also based on standard maize practices in the eastern Free State region of South Africa.
The role of the basin area is to: stop ex-field runoff; maximize infiltration; and store the harvested water in the soil surface beneath the basin.The basin acts as surface storage until the infiltration process is completed.The infiltrated water is stored in the rooting zone where it remains available for crop uptake but loss by evaporation is minimised.

Soil surface state characterisation
Surface (0-100 mm) soil water content was determined by gravimetric methods prior to rainfall simulation.The surface roughness index was determined in the 1 m x 1 m runoff plot with a 100 peg-board method (Zobeck and Onstad, 1987).Pegs of length 100 mm (pre-calibrated) with a diameter of 25 mm were evenly spaced on the 800 mm x 800 mm x 20 mm board, i.e. 10 x 10 holes with 60 mm intervals between rows and pegs.The holes in the board are made slightly wider than the pegs so that they can move freely.During measurements the board was placed at randomly selected sites on the soil surface and the vertical distance to the surface was then recorded for each peg.Three measurements were recorded for each plot.The mean value was taken as the roughness index of the surface.
The slope in the selected plots was determined following a simple method described in Bothma (2010).Two broomsticks, each 500 mm long were used for this purpose.One broomstick (A) was placed at the bottom slope with a cord attached 500 mm from the base of the broomstick.Another broomstick (B) was mounted at the top of the 2 m runoff area (Fig. 2).The cord from broomstick (A) was then stretched level to broomstick (B) using a spirit level.The difference in height from the soil surface at A and B was then used to calculate the slope.

Tillage treatments and historical background
The study focused on plots subjected to artificial rainfall.Treatments were: • In-field rainwater harvesting (IRWH) (No-till) • Annually tilled conventional tillage (CON) The experimental site was established in 2005 in order to evaluate the effects of IRWH on sunflower-cowpea cropping systems.The experimental design was split-plot with IRWH and CON as main treatments and cropping systems as subplots.Three cropping systems were tested: sole sunflower (Helianthus annuus L.); sole cowpeas (Vigna unguiculata L.); sunflower-cowpea intercropping.In order to establish IRWH plots, the land was ploughed initially and then disced to obtain a fairly level surface.Spades and rakes were used to create the basins and runoff area.The runoff area was levelled so as to form a slope of nearly 8% by raking and then left undisturbed so as to develop a surface crust.Weeds were controlled by spraying with glysphosphate at a rate of 360 g•ℓ -1 .This was done both during the growing period and off season.The surface soon developed a crust which enhanced runoff into the basin.
On the CON plots conventional ploughing (including mouldboard plough and harrow disc) was done at the beginning of the experiment and was later hand dug to a depth of about 200 mm before planting every season.This was necessary so that the access tubes that were installed to monitor soil water content were not damaged.Weeding during the growing season was done using hand hoes.

Rainfall simulation experiment
Simulated rainfall allows for complete control of experimental conditions (Truman et al., 2007).
Rainstorms with specified intensities were simulated with the Hofrey rainfall simulator (Fig. 3.1).The design of the simulator was based on the oscillating overhead sprinkler type  A, B, C, D) for measuring soil water.Row spacing and access tube distribution was similar in CON treatments (after Hensley et al., 2000).
Available on website http://www.wrc.org.zaISSN 0378-4738 (Print) = Water SA Vol.37 No. 2 April 2011 ISSN 1816-7950 (On-line) = Water SA Vol.37 No. 2 April 2011 170 period of 60 min; final runoff rates (mm•h -1 ).The latter was estimated as the average of nearly constant (steady) 5-min readings with the smallest difference between them.Runoff coefficients were calculated as the ratio (%) of total runoff (mm) to total rainfall (mm) applied during the simulation period.

Statistical analyses
An analysis of variance (ANOVA) was conducted using SPSS version 17.0.(SPSS Inc., 2008).Mean separations were achieved by using Tukey's honestly significant difference (HSD).A probability level of less than 0.05 was designated as significant.If there was a statistically significant interaction between main effects of tillage and rain intensity, then the interaction was presented.Otherwise only the main effects of tillage were reported.

Tillage effects
Time to runoff was significantly influenced by tillage practice (Table 4).Mean time to runoff in IRWH treatment was less than CON treatment by about 6 min or 48%.Total runoff was 1.7-fold higher in IRWH plots compared to CON plots, but the differences were not significant (Table 4).Short runoff time is beneficial for water harvesting especially on the study site where most of the rainfall comes in light showers (Mzezewa et al., 2010).A shorter hydrological response time from IRWH compared to CON strips was expected.This was attributed to the tillage effects.The formation of surface crust on the runoff strip of IRWH plots would result in early runoff compared to the CON treatment.Surface crust is a major factor in runoff generation (Philippe et al., 2001).No-till promotes surface soil sealing.Jin et al. (2008) reported that long-term application of no-till might lead to soil compaction, thereby increasing the runoff and decreasing infiltration.Higher total runoff under IRWH treatment was also attributed to low surface roughness (10 mm) on IRWH plots compared to CON (29 mm).The generation of runoff has been linked to soil-surface roughness (Carmi and Berliner, 2008).Guzha (2004) attributed higher runoff rates in no-till (NT) compared with other tillage systems due to lack of surface depressional storage.Lack of significant difference in total runoff between the 2 treatments in the current experiment may be attributed to the higher simulated rainfall intensities (23 -71 mm•h -1 ) compared to the natural rainfall normally received on the study site.High intensity rainfall could have caused slacking and collapse of clods in CON plots leading to the formation of a surface seal similar to that in IRWH plots, akin to the observation of Welderufael et al. (2008).Our results are in agreement with the findings of Rao et al. (1998) who reported no significant difference in runoff between no-till with crusted surface and conventional tillage plots on an Alfisol.Lack of difference in runoff from the 2 tillage systems was attributed to structurally unstable crusting soils.In Ethiopia, Welderufael et al. ( 2008) measured runoff, resulting from natural rain, from flat, crusted no-till plots (similar to IRWH) and conventional tillage plots on a Fluvic Regosol during the 2003 and 2004 seasons.They found no statistical difference between the runoff for the 2 treatments.They attributed this to high rainfall intensity that probably caused the clods on the CON plots to disperse and slake quickly, promoting faster crust formation, and resulting in the soil surface having similar properties to that on the IRWH plots.
Final runoff rate was significantly affected by tillage system (Table 4).By the end of the simulation period IRWH plots were discharging significantly higher runoff (26.2 mm•h -1 ) than CON plots (20.5 mm•h -1 ) (Table 4).Runoff from IRWH plots is stored in the basins where it infiltrates for use by crops whilst runoff from CON plots is ex-field and ends up in river systems.Our results indicated that the IRWH technique offers advantages of water harvesting over CON systems, as previously reported (Hensley et al., 2000;Botha et al., 2003).
Tillage system had a significant effect on runoff coefficients (Table 4).The runoff coefficient in IRWH plots was 2.1fold higher compared to CON treatment, as expected.Runoff coefficients obtained in this study are higher compared to those reported for similar production practices on some South African ecotopes (Table 1).Surface seal is the dominant factor in reducing infiltration in most African soils.The work of Stern et al. (1991) in South African soils demonstrated that kaolinitedominated soils have the most stable aggregates, but the presence of smectites in small quantities may dramatically affect the degree of dispersion.This observation was corroborated by other researchers (Lado and Ben-Hur, 2004).Contrary to findings reported in the literature, soils at this study site which contain 99% kaolinite clay minerals produced higher runoff than those at Glen (38% and 45% clay) which contain a high proportion of smectite clay mineralogy (Hensley, 2000).The dispersion of soil colloids is also controlled by the nature and distribution of the exchangeable cations, of which sodium is the most dispersive cation.However, exchangeable sodium is low at this site (< 0.12 cmol + •kg -1 ) and therefore cannot be blamed for exacerbating the collapse of soil aggregates.The differences between runoff generated at this site and Glen could largely be attributed to the poor topsoil structure of Ferralsols at this site.The differences were also largely attributed to steeper slopes on the current experimental site (8%) compared to the flat slopes used in studies such as those reported in Table 1.High rainfall intensities simulated could have resulted in heavy raindrop impact that led to the collapse of soil aggregates and therefore higher runoff than expected.However, our results are Figure 1Location of the study area Figure 2 Diagrammatic representation of the infield rainwater harvesting technique, showing distribution of access tubes (A, B, C, D) for measuring soil water.Row spacing and access tubedistribution was similar in CON treatments (afterHensley et al., 2000).

TABLE 1 Summarised measured rainfall-runoff relationships from annually tilled and bare crusted soils on semi-arid ecotopes
Available on website http://www.wrc.org.zaISSN0378-4738(Print)= Water SA Vol.37 No. 2 April 2011 ISSN 1816-7950 (On-line) = Water SA Vol.37 No. 2 April 2011 167of the lowveld which forms part of the greater Limpopo River basin.The experimental farm is 14 ha (arable) in extent on 8% gently undulating slopes running in a north-south direction.Rainfall is highly seasonal with 85% occurring between October and March (summer) (Table2).The mean maximum temperature (T max ) is around 30 o C while the mean minimum temperature (T min ) is about 20 o C during the growing period.

TABLE 4 Mean runoff, time to runoff and runoff coefficients as affected by tillage Tillage treatment Parameters Time to runoff (min) Total simulated rain (mm) † Total runoff (mm) Final runoff rate (mm•h 1 ) Runoff coefficient (%)
SEM =standard error of mean; †Rainfall averaged across all RI's over the simulation period.a Means not followed by the same letter in the same column are statistically different based upon Tukey's HSD means separation test at P < 0.05.*significant at P < 0.05; n.s.not significant.