制定和评估最佳水和盐管理做法：从非盐碱地和积水灌溉田中吸取教训

A R T I C L E I N F O

Handling Editor - Dr. Z. Xiying

Keywords:

Crop production Irrigation farming Soil salinity Water logging Water quality

A B S T R A C T

To address the on-site and off-site salt load impact associated with non-saline and water-logged irrigated fields a well-considered strategy is required. This implies minimal salt mobilisation and additions through irrigation, no crop yield losses due to excessive salt in the root zone, and minimal irrigation-induced drainage and leaching. The aim was to formulate best water and salt management practices to achieve this strategy and conduct an extensive assessment on whether they are implemented before problems appear. Weekly and seasonal data (2- years) from 19 fields (28 measuring sites) were used. Crops included barley, wheat, groundnuts, maize and lucerne grown in a semi-arid climate. Dominant soils were sandy loam, loamy sand and sandy with lateral

moving shallow groundwater tables (depth > 1.2 m and electrical conductivity < 250 mS m—1), while elec-

trical conductivity of primary water sources is < 100 mS m—1. Good decisions included the use of centre pivot compared to flood irrigation, irrigation schedules that ensured soil matric potential for maximum crop tran- spiration and limited yield losses due to water logging and soil salinity. Poor decisions were incorrect sprinkler design and excessive pumping pressure and limited use of rainfall and capillary rise to irrigate less. Growing season rainfall-plus-irrigation were a mean 55% more than the quota (annual water allocated to farmers), and 35%, 39%, 60%, 106% and 67% more than a “conservative” estimate of barley, wheat, groundnuts, maize and lucerne transpiration, respectively. The magnitude of variation in this oversupply were similar than the mean values. Approximately all salts applied through irrigation were leached from the root zone, while minimal re-use of drainage water was made (only 3 fields). In terms of resource use efficiency and preventing environmental degradation the evidence is overwhelming for improved decisions at field level. However, given sufficient supply of water, continuation of the status quo, might be sustainable in the long term.

1. Introduction

The salt load associated with irrigation remains arguable the most studied problem in arid and semi-arid crop production regions of the world. In most cases soil salinity coincides with water logging because of the presence of shallow groundwater tables. A vast amount of knowl- edge is available regarding the causes and management (methods and investments required, Wichelns and Qadir, 2015) of salt affected soils and shallow groundwater tables, as are estimates regarding the magni- tude of the problem (United States Salinity Laboratory Staff, 1954; Ayers and Westcot, 1985; Van Schilfgaarde, 1990; Rhoades, 1997; Hillel,

2000; Hillel and Vlek, 2005; Letey et al., 2011; Singh, 2014; Wichelns

and Qadir, 2015; Ritzema, 2016; Singh, 2018).

For irrigated fields with no apparent problems of salinity and water

logging a strategy that proactively not only address on-site problems but also off-site issues of water conservation and degradation of ground- water and river water sources are also required. From current knowl- edge such a well-considered strategy is to minimise salt mobilisation and additions through irrigation water, prevent decreases in crop yield due to excessive salt in the root zone and minimise irrigation-induced drainage and leaching. In fact, according to Wichelns and Qadir (2015), Prof Hilgard as early as 1893 passionately urged farmers and public officials to use irrigation water sparingly and build regional drainage systems. The authors emphasising “that his prescription for achieving sustainable irrigation is as valid in the 21st century as it was in the 19th”.

The question however, is whether the 21st century farmers are implementing such a well-considered strategy for proactive water and

* Corresponding author.

E-mail address: barnardjh@ufs.ac.za (J.H. Barnard).

https://doi.org/10.1016/j.agwat.2020.106706

Received 6 May 2020; Received in revised form 10 December 2020; Accepted 14 December 2020

Available online 5 January 2021

0378-3774/© 2020 Elsevier B.V. All rights reserved.

salt management or are they only responding once problems appear. Especially in semi-arid developing regions with generally good quality irrigation water, like for example in central South Africa, where there is no clear shift from ensuring production through irrigation to preventing off-site environmental degradation. The aim of this paper was first to formulate from literature best on-farm water and salt management practices, i.e. methods or techniques to achieve this well-considered strategy. Secondly, to assess over four growing seasons (2 years), in a semi-arid region, the extend that these formulated best on-farm water and salt management practices are implemented by farmers on 19 irri- gated fields in two irrigation schemes with no apparent salt related problems. Specifically, we want to establish whether crop yield varia- tion between fields could be ascribed possibly to mean seasonal water logging, matric stress and osmotic stress. The efficiencies of the centre pivots on the fields were also evaluated since they are integral to irri- gation farming. Furthermore, a comparison was done between the measured and allocated water for irrigation, and based on seasonal transpiration whether rainfall and shallow groundwater tables contribute to water use of the crops or not. An estimation was done to determine whether salts added through irrigation are leached from the root zone and do not accumulate in the groundwater table. We argue that such an extensive assessment will provide valuable insight into current practices and possible benchmarks after decades of researching the old enemy.

2. Formulating best on-farm water and salt management practices

In the design of an irrigation project many regional salt sources and control factors should be considered, inter alia (i) suitability of the soils for irrigation, (ii) residual salt content of the soils, (iii) irrigation water quality, (iv) topography and its effect on subsurface drainage, (v) type of system and amount of irrigation, (vi) climatic conditions and reliability of the irrigation water supply, (vii) environmental impact of drainage water and (viii) interception and possible re-use of drainage water. Many of the past irrigation developments however took place without proper consideration of these factors. Du Preez and Van Huyssteen (2020) argue that in countries like South Africa where commercial and communal farming are practiced this is worse under the latter. Hence, farmers across the globe are exposed to various levels of design and operation of regional irrigation and drainage infrastructure. Despite these constraints’ farmers should still aim to implement best on-farm water and salt management practices.

The choice of an efficient irrigation system can be regarded as the first important decision by farmers. This is crucial in reducing the amount of applied salts and decreasing the mobilisation of salts by reducing water loss from drainage and surface runoff (Abrol et al., 1988; Minhas, 1996; Kruse et al., 1996; Hillel, 2000; Oron et al., 2002; Hanson and May, 2004). According to Reinders (2011), the system should apply water at the desired amount, at an accurate application rate and uni- formly over the entire field, at the precise time, with the smallest amount of non-beneficial water consumption, and should operate as economically as possible.

Likewise, with continuous good decisions on when and how much to irrigate, salt additions through irrigation, excessive drainage and leaching from the root zone and mobilisation of salts through excessive drainage and runoff can be reduced. It is well documented that these decisions should be based on scientific based measurements (Quin˜ones et al., 1999; Leib et al., 2002; Annandale et al., 2011; Barnard et al., 2017) and/or transient soil-crop-water mathematical model-based evaluation of alternative irrigation practices. Tenreiro et al. (2020) provides an extensive review of popular crop growth (for example DSSAT, DAISY, APSIM, STICS, AquaCrop and MONICA) and

hydrology-based (for example HYDRUS1D, SWAP and SWIM) models that can be used in irrigation decision making. Atmospheric-based quantification of evapotranspiration, soil water content measurements,

crop-based monitoring and an integrated soil water balance approach, which encompasses real time and pre-programmed techniques, are some of the methods that can be used to quantify crop water requirements. Where possible rainfall and capillary rise from shallow groundwater tables should be used as water sources for crop water requirements (Ayars et al., 2006; Jhorar et al., 2009; Annandale et al., 2011; Isidoro and Grattan, 2011; Singh, 2013). When this is done then monitoring of salts in the root zone will be necessary, especially in soils with restricted natural and/or artificial subsurface drainage.

As a best practice, continuous monitoring of root zone salinity is important to determine the possibility for a reduction in crop growth and yield due to the presence of excessive salt. Irrigation-induced leaching is recommended when a reduction in crop yield is expected and not during every irrigation event. Research have shown that the efficiency of salt leaching will increase at higher soil salinities, i.e. the percentage of excess salt removed per unit drainage from the root zone (Monteleone et al., 2004; Barnard et al. 2010). Letey and Feng (2007) and Letey et al. (2011) showed that ideally a transient-state compared to a steady-state approach must be used to quantifying the leaching requirement; a state-state quantification overestimate the leaching requirement. The worldwide quest for developing relatively salt tolerant varieties (Qadir and Oster, 2004; Rozema and Flowers, 2008; Rozema et al., 2015) signify the best practice of applying irrigation-induced leaching only when necessary.

Farmers also have the opportunity to select crops that will produce satisfactorily in expected higher root zone salt conditions during the growing season. This is primarily because crops differ in their tolerance to high salt concentrations in the root zone (Maas and Hoffman, 1977; Maas, 1990; Rozema and Flowers, 2008; Rozema et al., 2015; van Straten et al. 2019). The socio-economic aspect of selecting crops and the fact that crop salt tolerance can be modified by different fertiliser applications, irrigation methods and frequencies, and a combination of soil, water and environmental factors (Meiri and Plaut, 1985) can however complicate this practice. Lastly, it has also been proposed and confirmed that even saline water (re-use of drainage water, treated waste water, treated water from unconventional gas developments etc.) can be safely used to irrigate certain crop varieties under specific soil and climatic conditions using specific water and salt management practices (Rhoades, 1992; Minhas, 1996; Sheng and Xiuling, 1997; Singh, 2004; Malash et al., 2005; Sharma and Minhas, 2005). Phogat et al. (2020), however, emphasised with a long-term modelling study that adequate management is needed when recycled water is used in order to minimise the associated adverse impacts on crop yield and soil chemical properties.

When the above-mentioned practices fail to proactively manage water and salt successfully, because of poor implementation or due to uncontrolled highly site-specific factors, productive soils become un- productive. Reclamation of saline and/or sodic soils is possible through controlled strategic leaching as well as soil and water amendments and bioremediation (calcareous soils are reclaimed without the application of amendments through the cultivation of certain salt-tolerant crops; Qadir and Oster, 2002). Fig. 1 shows the formulated best on-farm water and salt management practices.

3. Assessing best on-farm water and salt management practices

3.1. Methodology

3.1.1. Study fields

The research was conducted on 19 fields (10–60 ha) located within the Orange-Riet ( 17 000 ha) and Vaalharts ( 40 000 ha) Irrigation Schemes; the largest two schemes in the Lower Vaal River Basin, central South Africa (Fig. 2). Orange-Riet receives its water from the Vander- kloof Dam (situated in the Orange River), from where it is conveyed and distributed along the Orange-Riet canal section ( 120 km) of the scheme to Jacobsdal. Tail-end and drainage water from the Settlement

Fig. 1. Formulated best on-farm practices aimed at addressing an ideal strategy of water and salt management.

section of the scheme at Jacobsdal is transferred into the Riet River, which is conveyed downstream in an easterly direction to the Lower Riet River Section of the Scheme and the Vaal River. Vaalharts Weir on the Vaal River, just upstream of Warrenton, diverts water into the Vaalharts main canal, which supplies Vaalharts Irrigation Scheme located at Jan Kempdorp and Hartswater ( 1176 km of concrete-line canals). In addition, 314 km of concrete-line drainage canals were built to convey both storm-water and subsurface drainage water out of the irrigation scheme via the Harts River in a south westerly direction towards Spit- skop Dam.

Orange-Riet and Vaalharts are located in a semi-arid zone; long-term mean rainfall is 397 and 427 mm per year for Orange-Riet and Vaal- harts, respectively, with aridity indexes of 0.23 (ETo 1740 mm) and

0.26 (ETo 1647 mm), respectively (Table 1). Rainfall mainly occurs in the form of thunder showers during the summer months (October to April) at both schemes. The warmest months at both schemes are November, December and January with a long-term mean monthly maximum temperature of around 30 ◦C. The coldest months are June and July with a long-term mean minimum temperature of around 0 ◦C. The fields were selected to include a variety of bio-physical condi- tions, as well as agronomic and water and salt management practices (Table 2). The irrigation systems at the fields were primarily over-head sprinklers (centre pivot or linear). Flood irrigation was practiced at two fields during the first year of measurements and then replaced with centre pivot irrigation during the second year. Only selected areas within thirteen of the fields were artificially drained. The winter crops grown during the study period of 2 years (July 2007 to July 2009) at the fields were wheat and barley, while maize and groundnuts were planted in the summer (Table 2). One season of peas, cotton and seed sunflower were also grown at three different fields. Crop rotation systems employed by farmers consisted mainly of double and fallow cropping. With double cropping a wheat-maize rotation was planted alternately for two years, or only for one year where after either wheat or maize were replaced by barley and groundnuts, respectively in the second year. At some fields lucerne was also part of the rotation system. Fallow crop rotation systems consisted mainly of producing three crops com-

bined with one fallow period over two years.

The planting date, seeding densities together with nitrogen, phos- phorous and potassium applications compared well to recommended guidelines for the area (Table 3) (FERTASA, 2016). Popular soil culti- vations during a total 46 plantings of either wheat, barley, maize or groundnuts included burning the residue before planting (63%), disking (54%), wonder till (48%, shallow tined-roller seedbed preparation), ploughing (38%), bailing the residue of the previous crop before planting (28%) and ripping the soil (22%) (Table 3).

A total of 28 sites (16 m2) for taking measurements were set up in the

fields. Ten of the fields had two measuring sites. This was done as some farmers indicated that specific areas of their field are “problem” areas for example not being artificially drained, poor natural drainage, different soil type, shallow root zone etc. It was decided to include only one extra measuring site under these circumstances. The authors recognise that the soil will vary within fields. The focus of the research was, however, not on the spatial quantification of soil water and salt balances, but rather on how water and salt management practices affected the processes involved.

The measuring sites are dominated (75%) by relatively homogenous deep aeolian sandy to loamy sand soils (Table 4); the mean measured clay percentage (pipette method of Day, 1965) over a depth of 1.8 m at these sites were equal to or less than 10 and a standard deviation (stdev) equal to or less than 2. At 5 of the sites the clay percentage was more than 30, while two sites had a mean clay percentage of 17. At ten of the measuring sites the bulk density (η) over the root zone was determined; mean η for the sandy to loamy sand and clayey soils were 1.65 and

1.62 g cm—3, respectively. The measured sand, silt and clay percentage

together with η (where measurements were available) were used to es- timate the volumetric soil water content (θv) and hydraulic conductivity

(K) at a matric potential (ψm) of 6 and 30 kPa with the RETC software package (version 6.02, van Genuchten, et al., 1991). The default van Genuchten, m 1–1/n, retention curve model and Mualem conductivity model of RETC were used, i.e. the neural network predic- tion from Schaap et al. (2001) was used to generate hydraulic properties; ψm of 6 and 30 kPa or pressure heads of 60 and 300 cm rep- resents in general the range in which field capacity will vary for sandy to clayey soils (Ehlers and Goss, 2003). The pH (H2O) at all the sites except

Fig. 2. Orange-Riet and Vaalharts Irrigation Schemes located in the Lower Vaal River Basin, central South Africa (black circles indicate the sites for measurements).

Table 1

Long-term mean maximum (max) and minimum (min) temperature (temp), reference evaporative demand (expressed as the evapotranspiration of a well-watered clipped cool-season grass, ETo) and rainfall per month at Orange-Riet and Vaalharts Irrigation Schemes (raw data courtesy of ARC-ISCW, Pretoria).

Rainfall (mm) 8 9 11 33 40 42 60 64 64 43 15 8 397

ETo (mm day—1) 2.6 3.2 4.7 5.3 6.1 7.0 7.2 6.4 5.3 4.1 3.1 2.5 1740 (4.79)

Vaalharts Max temp (◦ C) 20 22 26 28 31 32 32 31 30 27 22 19 (27)

Min temp (◦ C) 1 3 7 11 14 16 17 16 14 10 5 1 (10)

Rainfall (mm) 3 4 9 34 49 48 71 83 63 37 21 5 427

ETo (mm day—1) 2.4 3.2 4.6 5.6 6.5 6.8 6.5 5.4 4.5 3.9 2.8 2.3 1647 (4.54)

Field area, type of irrigation system, number of artificial drainage laterals and crops planted at the various fields located in the Orange-Riet (or) and Vaalharts (v) Irrigation Schemes over a two-year period (July 2007 to July 2009), namely two winter and two summer growing seasons.

Field ha Measuring site Irrigation system Drainage laterals Crops

Winter-2007 Sumer-2008 Winter-2008 Summer-2009

1 43 or18, or20 Pivot – – Maize Wheat Sunflowers

2 52 or1, or2a Pivot 4 – Maize Wheat Maize

3 66 v10 Pivot – Lucerne Wheat Maize

4 32 v1a, v2 Flood-Pivot 9 Lucerne Wheat Maize

5 30 or14a, or15 Pivot 7 Barley Maize Lucerne

6 42 or12a, or13 Pivot 3 Wheat Maize – Maize

7 6 v5a Hand-shift 1 Wheat Maize Barley Groundnuts

8 20 or6a, or7 Hand-shift 1 Wheat Maize Barley –

9 17 v4a Linear 1 Lucerne Lucerne

10 10 or9a, or11 Hand-shift 1 Lucerne Lucerne

11 30 or4a, or5 Pivot 1 Wheat Maize Wheat Maize

12 52 v11a, v12a Pivot 11 Wheat Maize Barley Maize

13 18 v3 Linear – Lucerne Lucerne

14 20 v9a Flood-Pivot 5 Wheat Maize Wheat Groundnuts

15 20 v8a Flood-Pivot 5 Wheat Maize Wheat Groundnuts

16 30 v6a Pivot 4 Peas Groundnuts Wheat Cotton

17 30 v7 Pivot – Wheat Maize – Groundnuts

18 32 or17 Pivot – – Groundnuts Barley –

19
20 or19 Pivot – Wheat – Oats Groundnuts

a Measuring site located above an artificial drainage system.

Table 3

Summary of planting date, mean planting density and mean nutrient application used by farmers during the period July 2007 to July 2009 at the various fields (standard deviation is included in brackets).

Season Winter/Summer

Crop Wheat Barley Maize Groundnuts

Planting date June, July July October, December November, December

Number of plantings 17 5 17 7

Soil cultivation Burn residue 6or ; 4v 1or ; 2v 6or ; 7v 0or ; 3v Disk 4or ; 3v 3 or ; 0v              9 or ; 2v              1 or ; 3v

Plough 3 or ; 5v – 1or ; 3v 1or ; 4v

Rotavate 2or ; 0v – – –

Rip 2or ; 1v 1or ; 1v 1or ; 2v 1or ; 1v

Wonder till 1or ; 9v 0or ; 2v 0or ; 5v 1or ; 4v

Power harrow 2or ; 0v 1or ; 0v – 1or ; 0v

Bale residue 0or ; 1v – 1or ; 8v 0or ; 3v

or = Orange-Riet Irrigation Scheme; v = Vaalharts Irrigation Scheme.

one were above 7, while the cation exchange capacity (CEC) of the sandy to sandy loam soils were around 5 cmolc kg—1 and the clayey soils more

—1

determined with Eq. (1), where c1 is a parameter (0.075) that converts EC to kg salt ha—1 mm—1 water (Barnard et al., 2015)

than 10 cmolc kg (data not shown).

( ) ( ∑n )

3.1.2. Data acquisition

Two neutron access tubes (2 m) were installed 1 m apart in the

Si =

ECi(w)

(c1)

w=1

I(w)

(1)

centre of the 16 m2 measuring site. One piezometer (perforated 63 mm PVC tubes that was 3 m deep with the bottom end perforated) was

installed 2 m away, while 10 m from the 16 m2 area a square area of

6 m2 was cleared and a rain gauge was installed level to the soil surface. Weekly (w) measurements at the sites consisted of rainfall (R), irri-

gation (I), θv, groundwater table depth (ZGWT) and drainage from the

Soil samples at the measuring sites were taken every 0.3 m to a depth of 1.8 m, where possible, at the beginning and end of each growing season (5 times over 2 years), using standard auguring procedures. The soils were dried at 40 ◦C, crushed to pass through a 2 mm sieve and analysed using standard methods (The Non-Affiliated Soil Analysis Work Committee, 1990). The analysis included a saturation extract to deter-

artificial drainage system (D

AD,

see Table 2 for the number of drainage

mine electrical conductivity (ECe, mS m—1), and Na, K, Ca and Mg

(mg L—1). The concentrations of cations were determined through

laterals per field because no field were completely artificially drained), together with electrical conductivity (EC) of irrigation water (ECi), groundwater table (ECGWT) and drainage from the artificial drainage system (ECAD). Rainfall and irrigation were measured with rain gauges

(mm), θv with a calibrated neutron probe (mm mm—1), ZGWT manually

with a measuring tape (m) and DAD with a bucket and stop watch (L min—1). The EC (mS m—1) was measured with a calibrated CON 6/

TDS 6 Hand-held Conductivity/TDS Metre (Oakton instruments, Vernon Hills, USA). Seasonal salt added through irrigation (Si, kg ha—1) was

atomic absorption spectrometry. Representative irrigation, water table and drainage water samples were taken during the same time periods as those mentioned above and also analysed for dissolved cations (Ca, Mg, Na and K) using standard procedures (The Non-Affiliated Soil Analysis

Work Committee, 1990). The different crops within the site (16 m2)

were harvested at maturity, dried, weighed and threshed to determine the grain mass and total above-ground biomass.

Each centre pivot irrigation system was evaluated by placing 30 or 45 rain gauges evenly apart over the length of the system. The amount of

Table 4

Mean clay, silt and bulk density (η) over a depth of 1.8 m measured at the various sites located in the irrigated fields, together with the estimated (RETC v 6.02) volumetric soil water content (θv) and hydraulic conductivity (K) at a matric potential (ψm) of — 6 and — 30 kPa.             

Field MS Texture Clay (%) Silt (%) η (g cm—3) ψm = — 6 kPa (— 60 cm) ψm = — 30 kPa (— 300 cm)

θv (%) K (mm day—1) θv (%) K (mm day—1)

1 or18, or20 Clay 45 (4) 23 (5) 1.59 (0.06) 35.97 0.7

2 or1, or2 Sandy clay loam 34 (4) 20 (4) 1.53 (0.06) 35.38 1.3

3 v10 Sandy clay loam 31 (1) 9 (2) 1.68 (0.02) 31.46 0.4

4 v1, v2 Sandy loam 17 (3) 9 (4) 1.67 (0.03) 26.53 1.6

5 or14, or15 Loamy sand 10 (2) 5 (2) 1.62 (0.03) 20.57 9.3

6 or12, or13 Loamy sand 10 (2) 4 (1) 1.65 (0.05) 19.99 9.7

7 v5 Loamy sand 9 (1) 7 (1) – 21.83 10.2

8 or6, or7 Loamy sand 9 (1) 4 (1) – 20.49 19.1

9 v4 Loamy sand 8 (2) 6 (2) – 20.25 13.5

10 or9, or11 Loamy sand 8 (2) 3 (1) – 19.76 24.4

11 or4, or5 Loamy sand 8 (2) 4 (1) 1.63 (0.03) 17.81 12.4

12 v11, v12 Loamy sand 8 (2) 4 (1) 1.64 (0.02) 17.73 12.0

13 v3 Loamy sand 8 (1) 6 (1) – 19.76 13.5

14 v9 Loamy sand 8 (3) 5 (1) – 19.91 16.4

15 v8 Loamy sand 7 (1) 5 (1) – 19.01 17.1

16 v6 Loamy sand 7 (1) 5 (2) – 18.55 17.2

17 v7 Sand 6 (2) 4 (2) 1.77 (0.01) 14.97 6.9

18 or17 Sand 6 (1) 4 (1) – 16.28 20.3

19 or19 Sand 6 (1) 3 (1) 1.61 (0.02) 15.29 17.3

28.91

26.85

25.53

17.81

9.42

9.13

10.12

8.62

8.67

8.03

7.43

7.43

8.33

8.33

7.74

7.48

6.57

6.47

6.15

0.019

0.033

0.011

0.021

0.017

0.016

0.023

0.016

0.015

0.013

0.007

0.007

0.013

0.014

0.011

0.008

0.004

0.002

0.002

irrigation in the rain gauges were determined at a low (20%) and high (100%) speed. The Heermann Hein uniformity coefficient (CUH) was calculated with Eq. (2) and the distribution uniformity (DUIg) with Eq. (3). Where Ri is the distance (m) of the rain gauge at point i from the centre, yi the application depth (mm) at point i as collected in the rain gauge, yg the weighted average application of the total system (mm), and A the weighted average application of the lowest 25%. The amount of water pumped into the system compared to the amount of water measured in the rain gauges (application efficiency, AE) and system efficiency (SE), which is the combination of the application efficiency and the distribution uniformity were calculated with Eqs. (4) and (5), respectively. Where GA is the gross application (mm), Q the centre pivot

flow rate (m3 ha—1), t the rotation time (hours) and A the total wetted

area of the centre pivot (ha).

3.2.
Results

3.2.1. Water logging, matric and osmotic stress effects

The possibility that excessive sodium in the root zone impacted crop yield and soil permeability at the various fields were low (SAR < 6, data not shown, Le Roux et al., 2007) and therefore not included in the assessment. Fig. 3a shows the grain yield at each measuring site relative to the mean of each crop (1 representing the mean). The mean grain

yield for wheat and barley were similar (about 6-ton ha—1), while the

mean grain yield for maize, groundnuts and lucerne amounted to approximately 13, 3 and 24 ton ha—1, respectively. Barley had the lowest variation in grain yield (coefficient of variation, cv = 19%),

followed by maize (cv = 23%), groundnuts (cv = 25%), wheat (cv = 25%) and lucerne (cv = 30%). The mean measured θv(0—1.8m)