developing reservoir tillage technology for semi-arid environments

Developing reservoir tillage technology for semi-aridenvironments

C. Patr ick1

, C. Kechavarz i2

, I . T. James2

, M. O’Dogherty2

& R. J . Godwin2

1Botswana College of Agriculture, P/Bag 0027, Gaborone, Botswana, and 2National Soil Resources Institute, Cranfield University,

MK430AL Cranfield, UK

Abstract

Arable farming on suitable land in semi-arid environments is hampered by low and erratic rainfall and

droughts. To use this land effectively, techniques, such as water harvesting may improve soil water

storage and increase agricultural productivity. Laboratory experiments were conducted to assess two

reservoir tillage methods under different slopes, rainfall intensities and soil densities. A commercial

trailed tillage tool (Aqueel) was used to form discrete soil depressions by compression of the soil and

a soil scooping device with similar dimensions was used to make depressions in shear by lifting the soil

out. The results show that, for the sandy loam used in this study, reservoir tillage is an effective

method of harvesting water under high-intensity rainfall of short duration common in semi-arid areas.

It reduced surface run-off by 95% on 10� slopes when depressions were staggered and positioned with

their long axis across slope. However, high initial soil bulk densities lead to a significant reduction in

the volume of the depressions formed in compression and to internal compaction. Increasing vertical

load on the Aqueel resulted in an increase in depression volume without an increase in internal comp-

action but at high bulk densities the depression volumes remained small and high implement load

damaged the depression function and stability. This suggests a need for a pre-loosening tillage opera-

tion for compacted soils and the need to design new implements to form depressions in shear.

Keywords: Reservoir tillage, water harvesting, semi-arid, arable farming, Aqueel

Introduction

Low and erratic rainfall and droughts severely hamper arable

farming in semi-arid environments, where there is still more

than 600 million hectares of unused potentially arable land

worldwide (Alexandratos, 1988). Rainfall is irregular, falling

in infrequent spells of high intensity and short duration caus-

ing more than 90% of run-off (FAO, 1993). Water harvest-

ing has the potential to improve the productivity of these

areas.

Water harvesting is defined by the FAO (1993) as the col-

lection of run-off for productive use. There are two types of

agricultural water harvesting: (1) storage of supplemental

water off-site for application to the crop at a later stage

(Fox & Rockstrom, 2000; Barron & Okwach, 2005; Oweis &

Hachum, 2006) and (2) storage of water in the soil profile

for immediate use by the crop (direct water) (Ojasvi et al.,

1999; Fleskens et al., 2005; Brhane et al., 2006).

Reservoir tillage is an alternative method defined by

Hackwell et al. (1991) and Rochester et al. (1994) as a sys-

tem in which numerous small surface depressions are formed

to collect and hold water during rainfall or irrigation to pre-

vent surface run-off. However, currently, reservoir tillage is

used predominantly for soil erosion control in environments

with higher annual but lower intensity rainfall than semi-arid

environments. Typically, depressions are formed in compres-

sion by the use of a number of weighted, toothed discs that

are towed behind a tractor in recently tilled soils to form iso-

lated, approximately 0.5-L capacity, trapezoidal-shaped

impressions in the surface. This method has potential to

benefit semi-arid environments (Mrabet, 2002) because the

large infiltration surface area created by the numerous

depressions and the small depth of ponded water in the shal-

low depressions compared with other direct water harvesting

methods, such as ridge and furrow (Spoor & Berry, 1990;

Correspondence: C. Kechavarzi, National Soil Resources Institute,

Cranfield University, MK430AL Cranfield, UK.

E-mail: [email protected]

Received April 2006; accepted after revision October 2006

Soil Use and Management, June 2007, 23, 185–191 doi: 10.1111/j.1475-2743.2006.00069.x

ª 2007 The Authors. Journal compilation ª 2007 British Society of Soil Science 185

Kronen, 1994) are likely to result in higher infiltration rates

and therefore less evaporative losses and surface run-off.

However, under conditions of high evaporation rates and

high-intensity rainfall, reservoir tillage should be adapted to

maximize the volume of run-off collected without impeding

water infiltration through compaction during the creation of

the depressions, either by wheel traffic or by the implement

itself. For given soil conditions, the volume of water harves-

ted by a depression will depend on the volume of the depres-

sion and its depth which determines the maximum head of

water in the depression, but the volume of water will also

depend on the slope, which will influence the reservoir capa-

city. At the field scale, the number of depressions per unit

area and their spatial configuration will play a major role in

determining the volume of harvested water. In addition when

the formation of depressions under compression is not poss-

ible without excessive load because of soil hardening com-

mon in semi-arid environments, alternative methods should

be used. Forming depressions by shearing the soil using pad-

dles, spikes or scoops to lift out a small volume of soil is an

attractive alternative especially because hand tools can be

used when mechanization is not possible. Finally, the effect

of the position of planting with regard to the depressions on

plant establishment and performance needs to be investi-

gated.

The influence of these factors on the effectiveness and the

applicability of reservoir tillage in semi-arid environments

has not been fully addressed. The aim of our study was: (i)

to assess the water harvesting effectiveness of depressions

under different rainfall intensities, surface slopes and spatial

depression configuration using two contrasting depression

forming tools; (ii) to study the effect of initial soil bulk den-

sity and implement load on the internal compaction and the

volume of the depressions; (iii) to determine, for the given

geometry of the depressions, a relationship between water

harvesting reservoir capacity and surface slope; and (iv) to

determine the optimum planting position in relation to the

geometry of the depression.

Materials and methods

The soil was a sandy loam topsoil (Cottenham series). Its

texture was 72% sand, 16% silt and 12% clay with 3.7%

organic matter. The volumetric moisture content during

depression formation varied between 18 and 20% for all

experiments. All experiments took place in the laboratory or

the glasshouse in soil trays and in a soil tank.

Selected depression forming implements

Two different forms of depression formation (compression

and shear) were used. The Aqueel (Simba International Ltd)

is a commercially available, trailed tillage tool designed to

form discrete depressions for water storage by compression

of the soil surface under a normal load (Figure 1). The

implement has toothed wheels or sprockets carried on a rol-

ler shaft up to a maximum length of 4 m. The sprockets can

be spaced along the shaft; however, when placed adjacent to

each other (as shown in Figure 1), depressions occupy 70%

of the soil surface and can be produced in either a parallel

or a staggered configuration. Each sprocket is 115 mm

wide with an outer diameter of 480 mm. There are six teeth

equidistantly spaced around the perimeter of each sprocket.

Each tooth has the approximate shape of a triangular prism

(Figure 2) with a width (w) of 0.09 m, a height (y) of 0.07 m,

a top length (b) of 0.15 m and a bottom length (t) of 0.1 m.

The second tool tested was a hand-operated device which

scooped out holes of similar dimensions and shapes to those

produced by the Aqueel. Soil scooping is a realistic alternat-

ive when soil conditions are dry and dense.

Determination of harvested volume of water in reservoir

tillage

A needle-type rainfall simulator placed 9 m above the test

soil was used to determine the effect of rainfall intensity and

surface slope on the water harvesting capacity of single

depressions orientated with their long axis downslope. The

soil was packed into trays of 0.11 m depth, 0.25 m width

and 0.50 m length. These trays were designed to collect sur-

face run-off at the bottom of the slope and have a free drain-

ing lower boundary (seepage face). The three soil surface

treatments were: (i) surface with no depressions (control); (ii)

single Aqueel depression (compression soil failure); and (iii)

single scooped depression (shear soil failure). Three slope

angles were used: 0�, 5� and 10�. For each factorial combina-

tion of soil treatment and slope, simulated rainfall intensities

of 40, 60 and 130 mm h)1 were applied for 15 min and sur-

face run-off volume was recorded. This resulted in 27 rand-

omized treatments, each replicated three times.

Figure 1 Six Aqueel sprockets on a loaded tool frame towed through

prepared sandy loam soil in a soil tank.

186 C. Patrick et al.

ª 2007 The Authors. Journal compilation ª 2007 British Society of Soil Science, Soil Use and Management, 23, 185–191

Arrangement of depression configuration on surface slopes

The objective of this experiment was to study the effect of

the spatial configuration of multiple depressions formed by

compression on run-off. Soils were prepared as in the previ-

ous section, but in trays of 0.11 m depth, 0.60 m width and

0.50 m length, for depressions to be formed with their long

axis perpendicular to the slope. The soil treatments used

were: (i) surface with no depressions (control); (ii) depres-

sions not staggered, where depressions downslope were

aligned; and (iii) depressions staggered, where depressions

downslope were offset. Three slope angles were used: 0�, 5�and 10�. For each soil treatment–slope combination, a simu-

lated rainfall intensity of 50 mm h)1 was applied for 20 min

and surface run-off volume recorded. This resulted in nine

randomized treatments, each replicated three times. Because

of the difference in the ratio of depression surface area to

total catchment surface area between the single-depression

experiment (previous section) and this multiple-depression

experiment, a correction was made to scale down the volume

of run-off recorded in the single-depression experiment. This

made data from both experiments directly comparable.

Determination of internal compaction and volume of

depressions

The effect of initial dry bulk density and implement ballast

on the volume and the internal compaction of the depres-

sions was investigated in a soil tank, 20-m long, 1.7-m

wide and 1-m deep (Figure 1). Prior to depression forma-

tion by compression and by shear, the soil was compacted

in layers of 50 mm to a depth of 400 mm at four different

initial soil bulk densities: 1.09, 1.12, 1.18 and 1.24 g cm)3.

To form depressions in compression six Aqueel sprockets

were mounted on a 225-mm diameter axle and a steel

frame, and the whole unit was trailed behind a soil tank

tool carriage (Figure 1). Implement weight was varied by

the addition of ballast. The following loads were used: ini-

tial un-laden weight 2.51, followed by 4.64, 6.91 and

8.90 kN m)1. For each initial soil bulk density, the soil

tank was divided into four blocks of 2.5 m length and the

weight treatments were randomly allocated to each. A fifth

block was allocated to depressions formed in shear using

the hand-held scooping device. For each block, penetration

resistance measurements were made at three locations in

the bottom of five depressions selected randomly using a

Geopocket (Geotester) CAT T161 Penetrometer (15 mm

diameter) and compared with the initial penetration resist-

ance of the soil to assess the level of compaction inside

the depressions. In addition, for each block, the volume of

24 depressions selected randomly was determined by lining

the depressions with thin plastic film and filling them with

water to the surface.

Development of a model to describe reservoir capacity

A geometric model was developed to estimate depression

volume change with slope. The model assumed that the

depression geometry approximates that of a triangular

prism (Figure 2). In the case where the depression was

orientated with its long-axis downslope, the volume (VR)

was calculated as the depression was rotated about its

short-axis AC through the angle h (longitudinal tilt) using

equations (1) and (2). In the case where the depression

was orientated with its long axis across slope and rotated

about its long-axis AB through the angle / (lateral tilt),

equation (3) was used.

VR¼w

6ytþb2 2�sinhsinðaþbÞ

sinbsinðaþhÞ

� �sin2asinðb�hÞ

sinðaþhÞsinðaþbÞ

� �� forh�b ð1Þ

VR ¼wy2 sinða� hÞ6 sin a sin h

for h � b ð2Þ

VR ¼wy

6

cosð/þ cÞcosð/� cÞ

� �3tþ 2ðb� tÞ 1� sin/ sin c

cosð/� cÞ

� �� ð3Þ

Figure 2 Scale drawing of a model Aqueel depression. Dimensions are labelled as per the model developed in equations (1)–(3); h is the angle of

rotation about the short-axis BD and c the angle of rotation about the long-axis AB.

Reservoir tillage technology 187


Determination of the optimum planting position

This experiment was conducted in soil trays 0.13 m deep,

0.37 m wide and 0.74 m long under controlled glasshouse

conditions. For both compression and shear/scooping meth-

ods, wheat seeds were planted at the bottom of the depres-

sions where the highest moisture content is likely to be

found (treatments ABP and SBP, respectively); and in

between depressions where seeds may be planted for minimal

depression disturbance (treatments AIP and SIP, respect-

ively). In addition, for compression-formed depressions only,

seeds were planted underneath depressions before the depres-

sions were formed (treatment APP). This option is not prac-

tical for depressions formed in shear because seeds may be

scooped out. All the treatments with depressions were

packed at 1.04 g cm)3 dry bulk density. To address the effect

of increased soil bulk density, two controls without depres-

sion at two different bulk densities of 1.04 (treatment C1P)

and 1.20 g cm)3 (treatment C2P) were included. This resulted

in seven randomized treatments each replicated three times.

For each treatment, the emergence rate index (ERI) was

determined by taking daily counts of the number of seedlings

emerging until it reached a constant value. The ERI was cal-

culated using the formula given by Erbach (1982). In addi-

tion, the height of seedlings and the biomass yield of wheat

were measured.

For all the experiments, ANOVA (GenStat) was used to

test for significant differences between the treatments.

Results

Effect of rainfall intensity, slope and depression configur-

ation on surface run-off

The volume of harvested precipitation decreased with rainfall

intensity but increased with surface slope. This is illustrated

in Figure 3a,b which shows the amount of run-off obtained

as a function of slope for the rainfall intensities of 40 and

130 mm h)1, respectively. The difference between the volume

of water harvested with compression-formed depressions and

shear-formed depressions was not significant. This suggests

that for the low initial bulk density obtained when packing

the run-off trays, the increase in bulk density at the bottom

of the Aqueel depression did not significantly reduce water

infiltration.

As shown in Figure 3c, depressions which were staggered

downslope reduced run-off to a greater extent than depres-

sions that were aligned downslope (e.g. 95 against 78%

reduction for a 10� slope). This was because staggered

depressions intercepted run-off whereas it could flow in

between the rows of aligned depressions.

Figure 3 also demonstrates the effectiveness of orientating

depressions across the slope. For all tested slope angles, the

volume of run-off per unit surface area obtained with a rain-

fall intensity of 40 mm h)1 for depressions orientated

0

5

10

15

20

25

30

35

0

5

10

15

20

25

30

35

Surface slope (degrees)

Sur

face

run

-off

(mm

)S

urfa

ce r

un-o

ff (m

m)

Sur

face

run

-off

(mm

)

No depression Aqueel depression Scoop depression

LSD = 2.3 mm (P = 0.05)


No depression Aqueel depression Scoop depression

LSD = 2.3 mm ( = 0.05)P

0

2

4

6

8

10

12


No depression Parallel config. Staggered config.

LSD = 0.25 mm ( = 0.05)P

1086420

1086420

1086420

(a)

(b)

(c)

Figure 3 (a) Surface run-off from single depressions at low rainfall

intensity (40 mm h)1) and (b) at high rainfall intensity (130 mm h)1)

and (c) surface run-off from multiple depressions configured differ-

ently.



downslope was greater than that obtained for depressions

orientated across the slope despite using a larger rainfall

intensity of 50 mm h)1 across the slope.

Internal compaction and volume of depressions

There was a significant increase in penetration resistance

when initial bulk density was increased from 1.18 to

1.24 g cm)3, for both the depressions scooped out, and the

depressions formed in compression with implement weights

of 2.50, 4.64 and 6.90 kN m)1 (Figure 4a). However, at a

weight of 8.90 kN m)1 the penetration resistance decreased

from 1.18 to 1.24 g cm)3 initial bulk density and it was signi-

ficantly less than for the other implement weights because

the implement caused failure of the depression walls.

There was no compaction in depressions created by shear

with the scooping device and no difference between penetra-

tion resistance measured inside the depressions and that meas-

ured initially (Figure 4a). There was no significant difference

in penetration resistance in depressions formed in compression

for the implement weights of 2.50, 4.64 and 6.90 kN m)1 for

all initial bulk densities. However, penetration resistance

was significantly greater inside the depressions than that

measured initially for all initial bulk densities, except at the

high implement load of 8.9 kN m)1 (Figure 4a). These results

show that the wheel sprocket (tooth) compacted the soil.

Initial soil bulk density had no significant effect on the

measured volume of depressions formed in shear, except at

high initial bulk density where it was only possible to achieve

80% of the maximum volume using the hand-held scooping

device (Figure 4b). The volume of the depressions formed in

compression was significantly reduced by increasing the ini-

tial bulk density. The volume also increased significantly with

the addition of ballast (Figure 4b) although it was difficult

to form depressions of a reasonable storage capacity at high

initial bulk densities where even with the addition of substan-

tial ballast (8.9 kN m)1) depression volume was only 30% of

the maximum possible.

Modelled depression volume

Determination of VR at zero slope using the geometric model

resulted in a maximum storage volume, VRmax, of 0.47 L,

but when the depression was rotated longitudinally about the

short-axis AC (Figure 2) VR reduced to zero at h ¼ 69.7�(equal to the angle a). In this orientation, larger reduction in

volume occurs at small values of h (Figure 5). At 15�, a

maximum slope for mechanized cultivation, VR ¼ 0.25 L (a

reduction of 47%). When rotated laterally about the longi-

tudinal axis AB (Figure 2), the rate of volume change was

smaller than when rotated longitudinally in the 0–15� range.

VR reduced to zero at / ¼ 59.0� (equal to the angle c). At

15�, VR ¼ 0.33 L, a reduction of only 30% (Figure 5).

0

200

400

600

800

1000

1200

1.05 1.1 1.15 1.2 1.25

Initial soil dry bulk density (g cm–3)

Initial penetration res. Scoop depression 2.51 kN/m

4.64 kN/m 6.91 kN/m 8.90 kN/m

600

800

1000

1200

Pen

etra

tion

resi

stan

ce (

kN m

–2)

LSD = 88.0 kN.m–2 (P = 0.05)

0

0.1

0.2

0.3

0.4

0.5

0.6

1.05 1.1 1.15 1.2 1.25

Initial soil dry bulk density (g cm–3)

Scoop depression 2.51 kN/m 4.64 kN/m 6.91 kN/m 8.90 kN/m

VR max = 0.47 L

Vol

ume

of d

epre

ssio

n (L

)

VR max = 0.47 L

LSD = 0.023 L (P = 0.05)

(a)

(b)

Figure 4 (a) Penetration resistance and (b) volume of depressions

formed by shear and by compression with implement ballast of 2.51,

4.64, 6.90 and 8.90 kN m)1.

0.00

0.10

0.20

0.30

0.40

0.50

Angle of slope (degrees)

Ret

aine

d vo

lum

e V

R (

L)

Long axis down slope Long axis across slope

VR max = 0.47 L

706050403020100

Figure 5 Volume of depression when depressions are orientated with

the long-axis downslope or across slope.



These results corroborate the run-off data (Figure 3) and

confirm that on slopes, the depressions should be orientated

with their longitudinal axis perpendicular to the line of slope

for maximum volume.

Planting position

Figure 6 shows that all the treatments significantly reduced

plant performance compared with the control with the

greater soil bulk density (C2P). There were no significant dif-

ferences between any of the other treatments and the control

with the smaller bulk density (C1P). Most treatments

achieved 90% rate of emergence by day 6 except for seeds

which were planted prior to the formation of depressions

with the Aqueel (treatment APP) which gave the lowest ERI

(Figure 6a). This delay was significantly greater than for the

other Aqueel planting positions (treatments ABP and AIP).

The height of seedlings measured 27 days after planting

(Figure 6b) showed that seedlings planted in between Aqueel

depressions was significantly shorter than that of those planted

at the bottom of the depressions (treatment ABP). Seedlings

planted between scooped depressions (treatment SIP) were

also shorter than those planted at the bottom of Aqueel

depressions (treatment ABP). Generally, seedlings that were

planted in the controls grew faster than those that were

planted within reservoir treatments, except those planted at

the bottom of the Aqueel depressions (treatment ABP). The

better performance of seedlings at the bottom of depressions

formed by compression suggests that the increase in water

retention capacity because of compression had a positive

effect. The results for the higher initial soil bulk density also

indicate that a denser soil around the seeds, in addition to

enhancing emergence, improved seedling growth rate as well.

At day 27, seedlings planted on the low-density control (C1P)

were shorter than those on the high-density control (C2P).

C1P gave the greatest biomass yield and the pre-planted

treatment (APP) the least biomass yield (45% of C1P)

(Figure 6c).

Discussion

For the given experimental conditions, which included simula-

ted high rainfall intensity storms of short duration, this study

suggests that reservoir tillage is an effective method of harvest-

ing water and thus reducing erosion in semi-arid areas on

light-textured soils, such as that used in this study. Use of marg-

inal areas in semi-arid environments for agricultural produc-

tion commonly includes light-textured soils on slopes that are

prone to erosion. The results show that depressions were able

to harvest up to 95% of surface run-off for slopes of up to 10�.For the given geometry of the depressions used, this level of

water harvesting can be achieved if depressions are orientated

with the longitudinal axis across the slope and arranged in a

staggered configuration. These results suggest that, for opti-

0

5

10

15

20

25

30

0

10

20

30

40

50

C1P C2P APP ABP AIP SBP SIP


Treatment

Treatment


Treatment

Em

erge

nce

rate

inde

xH

eigh

t of s

eedl

ings

(m

m)

100

50

0

150

200

250

300

350

Bio

mas

s yi

eld

(kg.

ha–1

)

LSD = 2.5 (P = 0.05)

LSD = 6.0 mm (P = 0.05)

LSD = 72.8 kg ha–1 (P = 0.05)

(a)

(b)

(c)

Figure 6 Performance of wheat seedlings: (a) emergence rate index,

(b) height of seedlings 27 days after planting and (c) biomass

recorded 32 days after planting (C1P and C2P: controls without

depression with bulk densities of 1.09 and 1.24 g cm)3, respect-

ively; APP: seeds planted underneath Aqueel depressions before

the depressions were formed; ABP and SBP: seeds planted at the

bottom of Aqueel and scoop depressions, respectively; AIP and

SIP: seeds planted in between Aqueel and scoop depressions,

respectively).



mized water harvesting, the Aqueel should be operated across

slope as is common with other tillage practices on moderate

slopes.

Depressions formed in compression were found to be com-

pacted internally and their storage volume decreased signifi-

cantly with increasing initial dry bulk density. Increasing

implement ballast resulted in an increase in depression volume

without an increase in penetration resistance but at high initial

bulk densities it was difficult to form depressions of a reason-

able storage capacity and high-implement ballast produced

depressions with damaged walls. This has a direct implication

for tillage practices in semi-arid, high-intensity rainfall envi-

ronments where soil loosening operations, such as inversion

ploughing and disc cultivation increase soil erosion risk, water

loss through evaporation and operational costs. Where a mini-

mum tillage approach is adopted, the Aqueel is inappropriate

as depressions are difficult to form in compacted soils and the

use of large implement ballast will lead to greater draught for-

ces and further compaction. In this case, the most appropriate

technology is the scooping device. The depressions formed in

shear did not show a decrease in depression volume with

increasing initial soil bulk density and they were found to be

less compact internally. Therefore, forming depressions in

shear manually or with a single-pass operation may be a more

adapted and more cost-effective solution on compacted soils.

This solution, however, calls for the design of mechanized

tools which can form depressions in shear, such as rotating

paddles, rotating scoops or ‘L’-shaped blades similar to those

used in rotary cultivation.

On the other hand, where soil conditions are favourable or

where soil loosening operations are used, the Aqueel is of

direct benefit, firming the soil and reducing erosion risk by

intercepting water. However, combining the use of the Aqu-

eel with crop establishment is a challenge, particularly with

narrow row crops. The use of the Aqueel with wide row

crops, with depressions formed inter-row, has been demon-

strated in the UK. Although field trials are needed to assess

the effect of the Aqueel on a range of cropping systems; the

results obtained in the glasshouse in this study showed that

in semi-arid environments, this approach might be less suit-

able because moisture was concentrated at the bottom of

depressions and seeds planted nearer the surface and inter-

row performed poorly because of moisture deficits. With nar-

row row crops, the simplest approach would be to plant seed

before depression formation. However, in this study, planting

seeds before depression formation was found to cause imped-

ance to wheat emergence and lead to reduced biomass. Seed-

lings performed better when planted at the bottom of the

depressions after depression formation. However, to prevent

seeding coulters cutting through depression walls, an injector

planter attached behind the Aqueel and synchronized to

place seed in the bottom of depressions would be necessary.

Field trials to evaluate the use of a modified rotary cultiva-

tor and the Aqueel system on the production of Sorghum

and maize in semi-arid conditions in Botswana are being

planned by the Botswana College of Agriculture. These reser-

voir tillage methods will be compared with other tillage

methods, such as mouldboard or chisel ploughing and mini-

mum tillage which are practised in Botswana.

Acknowledgements

The authors would like to thank the Botswana College of

Agriculture and Cranfield University for funding this work

and the additional support provided by Simba International

Ltd.

References

Alexandratos, N.(ed.) 1988. World agriculture: towards 2000. An

FAO study. Belhaven Press, London.

Barron, J. & Okwach, G. 2005. Run-off water harvesting for dry

spell mitigation in maize (Zea mays L.): results from on-farm

research in semi-arid Kenya. Agricultural Water Management, 74,

1–21.

Brhane, G., Wortmann, C.S., Mamo, M., Gebrekidan, H. & Belay,

A.. 2006. Micro-basin tillage for grain sorghum production in semi-

arid areas of northern Ethiopia. Agronomy Journal, 98, 124–128.

Erbach, D. C. 1982. Tillage for continuous corn and soybean rota-

tion. Transactions of the American Society of Agricultural Engin-

eers, 25, 906–911.

FAO. 1993. Soil tillage in Africa: needs and challenges. FAO Soils

Bulletin 69, Rome.

Fleskens, L., Stroosnijder, L., Ouessar, M. & De Graaff, J. 2005.

Evaluation of the on-site impact of water harvesting in southern

Tunisia. Journal of Arid Environments, 62, 613–630.

Fox, P. & Rockstrom, J. 2000. Water-harvesting for supplementary

irrigation of cereal crops to overcome intra-seasonal dry-spells in

the Sahel. Physics and Chemistry of the Earth (B), 25, 289–296.

Hackwell, S.G., Rochester, E.W., Yoo, K.H., Burt, E.C. & Monroe,

G.E. 1991. Impact of reservoir tillage on water intake and soil ero-

sion. Transactions of the American Society of Agricultural Engi-

neers, 34, 436–442.

Kronen, M. 1994. Water harvesting and conservation techniques for

small-holder crop production. Soil and Tillage Research, 32, 71–86.

Mrabet, R. 2002. Stratification of soil and organic matter under till-

age systems in Africa. Soil and Tillage Research, 66, 119–128.

Ojasvi, P.L., Gayal, R.K. & Gupta, J.P. 1999. The micro-catchment

water harvesting technique for the plantation of jujube (Zizyphus

mauritiana) in an agroforestry system under arid conditions. Agri-

cultural Water Management, 41, 139–147.

Oweis, T. & Hachum, A. 2006. Water harvesting and supplemental

irrigation for improved water productivity of dry farming systems in

West Asia and North Africa. Agricultural Water Management, 80,

57–73.

Rochester, E.W., Hill, D.T. & Yoo, K.H. 1994. Impact of reservoir

tillage on run-off quality and quantity. Transactions of the Ameri-

can Society of Agricultural Engineers, 37(4), 1183–1186.

Spoor, G. & Berry, R. H. 1990. Dry-land farming tillage and water

harvesting guidelines for the Yemen Arab Republic. Soil and Til-

lage Research, 16, 233–244.



developing reservoir tillage technology for semi-arid environments

Documents