Msh

TKa

b

c

d

e

f

g

a

ARR1A

KCWGCPV

tf

0d

Field Crops Research 132 (2012) 53–62

Contents lists available at SciVerse ScienceDirect

Field Crops Research

jou rn al h om epage: www.elsev ier .com/ locate / fc r

edium-term effects of conservation agriculture based cropping systems forustainable soil and water management and crop productivity in the Ethiopianighlands

esfay Arayaa,b,∗, Wim M. Cornelisb, Jan Nyssenc, Bram Govaertsd, Fekadu Getnete, Hans Bauer f,assa Amareg, Dirk Raes f, Mitiku Hailee, Jozef Deckers f

Mekelle University, Department of Crop and Horticultural Science, P.O. Box 231, Mekelle, EthiopiaGhent University, Department of Soil Management, Coupure Links 653, B-9000 Gent, BelgiumGhent University, Department of Geography, Krijgslaan 281 (S8), B-9000 Gent, BelgiumInternational Maize and Wheat Improvement Centre (CIMMYT), A.P. 6-641, Mexico D.F. 06600, MexicoMekelle University, Department of Land Resources Management and Environmental Protection, P.O. Box 231, Mekelle, EthiopiaK.U.Leuven, Department of Earth and Environmental Sciences, Celestijnenlaan 200E, B-3001 Heverlee, BelgiumMekelle University, Department of Earth Science, P.O. Box 231, Mekelle, Ethiopia

r t i c l e i n f o

rticle history:eceived 12 May 2011eceived in revised form8 November 2011ccepted 7 December 2011

eywords:onservation agricultureheat

rass pearop residueermanent raised bedsertisol

a b s t r a c t

In the northern Ethiopian highlands, croplands yield extremely high volumes of storm runoff and arethe major contributor to sediment load in the rivers. A medium-term tillage experiment was carried out(2005–2010) on a Vertisol to quantify changes in runoff, soil loss and crop yield due to Conservationagriculture (CA) in the sub-humid May Zegzeg catchment. A randomized complete block design with 3replications on permanent plots of 5 m by 14 m was used for three tillage treatments, (i) derdero+ (DER+),permanent raised beds with 30% standing crop residue retention and no-tillage on the top of the bed, (ii)terwah+ (TER+), ploughed once at sowing with 30% standing crop residue retention and furrows made at1.5 m interval, and (iii) conventional tillage (CT) with a minimum of three tillage operations and removalof crop residues. Tillage operations in the three treatments were done using the local ard plough mahresha.Local crop rotation practices followed during the six years sequentially from the first to the sixth yearincluded wheat-grass pea-wheat-hanfets (wheat and barley sown together)-grass pea-wheat. Glyphosatewas sprayed starting from the third year (2007) at 2 L/ha before planting to control pre-emergent weedin DER+ and TER+. Runoff and soil loss were measured in collector trenches at the lower end of each plot.Soil organic matter was determined at two depths (0–15 cm) and (15–30 cm). Local farmers evaluatedcrop stands. Significantly different (p < 0.05) 4-yr mean soil losses of 14, 17 and 26 t/ha, 5-yr mean runoffdepth of 76, 95 and 118 mm, and 5-yr runoff coefficient of 19, 24 and 30% were recorded for DER+, TER+and CT, respectively. Soil organic matter was significantly higher in DER+ and TER+ compared to CT. Themean farmers’ evaluation of crop performance in the last three years (2008–2010) showed a significanthigher score for DER+ (6/8) followed by TER+ (5.6) and least for CT (4.8/8), and improvements in crop yield

were observed; however, a period of at least five years of cropping was required before the differencebecame significant. In addition to the positive effects on runoff, soil loss and crop yield, we argue thatavoiding repeated tillage which is 10–11 oxen-span days per ha and the faster ploughing pace at sowingin DER+ will enable a reduction in oxen density with further natural resource benefits. DER + and TER+ areimprovements to good local practices that qualify them as CA: we recommend large scale disseminationand implementation on Vertisols.

© 2011 Elsevier B.V. All rights reserved.

∗ Corresponding author at: Mekelle University, Department of Crop and Horticul-ural Science, P.O. Box 231, Mekelle, Ethiopia. Tel.: +251 914720426;ax: +251 0344409304.

E-mail address: tesfayw2002@yahoo.com (T. Araya).

378-4290/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.fcr.2011.12.009

1. Introduction

As a result of the human population growth and climate change,global agriculture will have to rely more on rainfed farmingin the future. In Ethiopia, agricultural productivity is low and

5 ps Res

teccsrbPapgdFhsragiociE

cifibao(sbrt(vy

cbopcactbmpfautatZkC

atmtry

4 T. Araya et al. / Field Cro

he sustainability of traditional agricultural systems is threat-ned by degradation of cropland due to complete removal ofrop residues at harvest, aftermath grazing and frequent tillage inropland (Girma, 2001; Bezuayehu et al., 2002). The main powerources in agriculture are humans and animals. Repeated tillageeduces soil organic matter and thus increases soil erosion ratesy water (Angers and Mehuys, 1989; Kay, 1990; Papendick andarr, 1997). Rainfed farming agriculture is dominant in Ethiopiand annual food production shortages are commonly linked toeriodic drought and insufficient rainfall, periodical water log-ing and high runoff rates under wet conditions in Vertisoluring the growing season (Mati, 2005; McHugh et al., 2007;reebairn et al., 1996; Deckers et al., 2001). These problems andence relatively low yields are associated with an imbalancedoil hydrology. The only input is rainfall (often erratic), but thisainwater is lost too much as blue water, i.e., as direct runoff,nd consequently, less water is available for crops, the so-calledreen water (Rockström, 1997). This imbalanced soil hydrologys due to physical deterioration of the soil quality and absencef effective in situ soil and water conservation measures in theropland itself. Therefore, in order to increase crop productiv-ty, soil and water management practices need improvement inthiopia.

Some farmers in the Tigray highlands use a conventional in situonservation tillage practice with contour furrows at 2–4 m widentervals, locally called terwah, usually only on tef (Eragrostis tef)elds. The elongated terwah furrows trap and store rain water toe used later by the tef crop during dry spells instead of being losts runoff (Gebreegziabher et al., 2009). In the Lasta highlands, southf Tigray, farmers use the derdero system, especially for fenugreekTrigonella foenum-graecum), wheat (Triticum sp.) and tef on Verti-ol. Beds and furrows are prepared along the contour after havingroadcasted the seeds over the surface. Plants are grown on theidges where they are protected from water logging, while draininghe excess water towards the furrow where it ponds and infiltratesNyssen et al., 2011). In both systems, however, all straw is har-ested, the stubble grazed and the furrows and beds destroyedearly by tillage.

Although conservation agriculture (CA) is important to reduceropland degradation and increase land productivity, it has noteen implemented in northern Ethiopia beyond the achievementsf the aforementioned traditional practices. Using CA can be aossible solution to lessen crop land degradation and increaserop productivity (1) by reducing tillage, (2) by retaining rationalmounts of crop residue in the field and (3) by using profitablerop diversification. Results from comparison of CA and conven-ional agricultural practices over different time periods have noteen consistent between socioeconomic setup, crops, tillage imple-ents and systems, soils, climate, and experiments in different

arts of the world (Ahuja et al., 2006; Giller et al., 2009). There-ore, this experiment was carried out in an area characterized by

subsistence farming system in a sub humid climatic conditionsing the local ard plough mahresha and local crop rotation prac-ices in Vertisols. The modified versions of derdero (“derdero + ”)nd terwah (“terwah + ”) local tillage systems using the tradi-ional mahresha ard plough on Vertisol were introduced in Mayegzeg in northern Ethiopia in 2005 aiming at linking indigenousnowledge with the wide international body of knowledge onA.

Our objective was to evaluate the effects on runoff, soil lossnd crop yield of these newly developed CA versions of tradi-ional tillage using local crop rotation systems during the six years

edium term study period in sloping fields. We hypothesized thathe CA based practices derdero+ and terwah+ using the local cropotation result in reduced runoff and soil loss, and increased cropield.

earch 132 (2012) 53–62

2. Materials and methods

2.1. The study area

The experiment was conducted under rainfed conditions start-ing from 2005 in May Zegzeg (13◦39′N, 39◦10′E) at an altitude of2550 m a.s.l. in northern Ethiopia (Fig. 1). Mean annual rainfall of26 years in the nearby town of Hagere Selam was 741 mm withmore than 80% from mid June to mid September (Fig. 2), character-ized by high rainfall erosivity due to large drop size (Nyssen et al.,2005). Mean monthly minimum and maximum temperatures are4–6 ◦C and 20–22 ◦C (Nyssen et al., 2007). The average length of thegrowing period is 162 days (Goebel and Odenyo, 1984).

Three to four tillage operations are conventionally done withan oxen-drawn ard to control weeds, improve infiltration and pre-pare a fine seedbed, particularly for tef. The temporal pattern ofploughing depends on the availability of oxen, type of crop andrainfall. The most cultivated crops include tef, barley (Hordeum vul-gare), wheat, hanfets (barley and wheat sown together), grass pea(Lathyrus sativus) and lentil (Lens culinaris).

2.2. Rainfall characteristics during the study period

The rainfall exceedance probability (%) and return periods forthe experimental years were calculated using the RAINBOW soft-ware (Raes et al., 2006), which resulted in normal distribution ofrainfall after applying a log10 transformation. The RAINBOW soft-ware was used to analyse 26 yr of rainfall data. The rainfall returnperiods were significantly highest (p < 0.05) in 2005 and 2010 ascompared to the other experimental years (Table 1). To the reverse,the rainfall exceedance probability was significantly highest in2007 (Table 1). There were longer than normal rainy seasons in2005 and 2010, i.e., from March to September (Fig. 2). According tothe local practices in the study area, grass pea planting is delayedby more than one month in the growing season as compared to theother crops to avoid excess moisture to the crop.

2.3. The field experiment

The experimental layout was a randomized complete blockdesign with three replications (Fig. 3). The plot sizes were5 m × 14 m and the slope was 6.5%. The soil under the experi-mental trial was a Vertisol with a high stone cover. Three tillagepractices were applied: conventional tillage (CT), terwah+ (TER + )and derdero+ (DER + ) all using the local ard. Following the localcrop rotation practice, crops grown, from the first to the sixth yearsequentially were wheat, grass pea, wheat, hanfets (wheat and bar-ley sown together), grass pea and wheat. Wheat and barley weretreated with the same seed and fertilizer (di-ammonium phosphateand urea) application rates at 100 kg/ha and the method of plant-ing was by broadcasting for all treatments. Urea fertilizer was notapplied to grass pea. The same plots were kept fixed during the sixyears of study. Weed control was done by hand weeding in the firsttwo years in TER+ and DER+, whereas from 2007 on non-selectiveherbicide glyphosate (N-(phosphonomethyl) glycine) was sprayedat 2 L/ha three to four days before planting to control pre-emergentweeds. However, weed control in CT was done by combination offrequent tillage and hand weeding.

Terwah+ (TER + ) is essentially a new tillage system developedfrom the traditional in situ water conservation method (terwah)especially used in tef where broad seedbeds are created usingthe mahresha ard plough by making furrows on the contour at

regular intervals of ca. 1.5 m (Nyssen et al., 2011), but which isin the context of this study also tested for crops other than tef,using only one ploughing operation and combined with retentionof 30% standing stubble (Fig. 3). Derdero+ (DER + ) is also a newly

T. Araya et al. / Field Crops Research 132 (2012) 53–62 55

Fig. 1. Location map of the study area at May Zegzeg catchment, Tigray, Ethiopia.

Fig. 2. Monthly rainfall (mm) distribution at May Zegzeg during the study period (own measurements), and mean monthly rainfall distribution (26 yrs) for nearby HagereSelam town weather station (Ethiopian National Meteorological Services Agency).

Table 1Rainfall exceedance probability and return periods (based on the analysis of 26 yrs of data) for the particular study periods 2005–2010 in May Zegzeg catchment, Ethiopia(p < 0.05). The 26 yrs rainfall data was from Hagere Selam (metrological weather station), 2 km to the West of the study site.

Year Rainfall (mm) Return periods (yrs) Exceedance probability (%)

2005 851 6 172006 720 2 502007 629 1 1002008 695 2 502009 710 2 502010 840 6 17Mean (6 yrs) 741 3 47

56 T. Araya et al. / Field Crops Research 132 (2012) 53–62

F its laD

ditaoufwamtffadi

2

(lpachvtte4iWiwAt(

R

2

t

ig. 3. The experimental plot at May Zegzeg as seen from a nearby viewpoint, withER+ is derdero+, TER+ is terwah+, CT is conventional tillage practice.

eveloped tillage system which is based on another traditionaln situ water conservation technique (derdero), where at the lastillage operation, the farmers broadcast the seeds over the surfacend then prepare beds and furrows along the contour at intervalsf ca. 0.6 m using the mahresha, moving the soil and seeds to anpper position on the beds (Nyssen et al., 2011). It protects the cropsrom water logging, while excess water drains towards the furrowshere it can slowly infiltrate. The ‘plus’ in derdero+ stands for the

daptations made, i.e., 30% standing stubble (Fig. 3) was left andost importantly, the beds were never ploughed, with the only

illage operation (with the mahresha) being the refreshing of theurrows at sowing. For this research, data collected on the plotsrom 2005 to 2010 were analysed. The adjacent farmer’s crop fieldt the upper side of the experimental plot was used (Fig. 3) toemonstrate the techniques and benefits of DER+ and to facilitate

ts promotion and adoption by local farmers.

.4. Runoff and soil loss data

Runoff and soil loss were measured in 4.5 m long, 1.5 m wideat the top) and 1 m deep collector trenches (Fig. 3), which wereocated at the down slope end of each plot and lined with thicklastic sheets. The plots were separated by 0.50 m wide ditches tovoid surface or subsurface flow between plots. Runoff data wereollected at 8:00 AM after each day with rain by measuring theeight of the water at three sample locations in the trenches. Theolumes of the trenches were annually calibrated at the middle ofhe growing season by relating known amounts of water to depth athree sample locations in the trench following the method of Tesfayt al. (2011). The collected runoff water was stirred thoroughly and

L was taken from each trench to determine accumulated sedimentn the trenches of each plot. These were filtered using a funnel and

hatman 42 filter paper having a pore size of 2.5 �m. The sed-ment in the filter paper was oven dried for 24 h at 105 ◦C and

eighed to quantify soil loss. Rainfall was recorded daily at 8:00M by rain gauge. A runoff coefficient (RC in %) was calculated as

he percentage of daily runoff depth (RD in mm) to daily rainfallRF in mm):

C = RDRF

100 (1)

.5. Soil organic matter

In 2010, composite soil samples with three subsamples wereaken at two depths (0–15 cm and 15–30 cm) using an auger during

y-out and view on standing stubble left on derdero+ and terwah+ planting systems.

the growing period. Organic carbon (%) was determined using theWalkley–Black method (Walkley and Black, 1934).

2.6. Crop data

Grain and straw yield were determined yearly at harvestfrom areas of 1 m × 1 m in three replicates per plot. Plant heightmeasurement was started one month after planting in 2010 whenwheat was grown and it was measured by weekly interval.

Rain water use efficiency (RWUE) was calculated as the percent-age of total grain yield (GY in t ha−1) to the amount of rain fall (RFin mm) received from planting to harvest:

RWUE = GYRF

100 (2)

Percent deviation (D in %) from the conventional tillage (CT) wascalculated as:

D = CA − CTCA

100 (3)

where CA and CT represent the measured data (grain yield, strawyield, runoff or soil loss) obtained in the CA (DER+ or TER+) and itscorresponding value in the CT treatments, respectively. This typeof equations was used in maize yield loss assessment by Odendoet al. (2003).

2.7. Local farmers’ insights

The cumulative crop performance in each of the nine plots wasevaluated with 8–10 local farmers from 2008 to 2010. The farmersfirst prepared nine sheets with a number from 1 to 9. They thencarried out a thorough group discussion during their visit to eachplot to evaluate the crop performance and placed each sheet con-taining a number from 1 to 9 on a plot, with the sheet having 9corresponding to the best crop performance while with 1 to thelowest one. Each plot thus received a different number within therange of 1–9. The maximum possible score for a given treatmentwas 8 (average of 7, 8, 9), the minimum 2 (average of 1, 2, 3).

2.8. Statistical analysis

ANOVA was used to test for statistical differences in runoff, soilloss and crop parameters between the management treatments.Data were analysed using the SAS statistical software (JMP ver-sion 5.0), and the standard error of treatment means was used for

T. Araya et al. / Field Crops Research 132 (2012) 53–62 57

F ghout at run

so

3

3

wwctwTsi(sp

ig. 4. Mean yearly runoff (a) and runoff coefficient (b) from each treatment throuillage practice. The bars shown represent standard error of mean (p < 0.05). Note th

eparation of means (SAS, 2002). Comparison of means was carriedut by Student t-tests at ̨ = 0.05.

. Results

.1. Runoff and soil loss

Runoff depth was lowest in DER+ for the complete study period,hile the largest record was from CT (Fig. 4a). Runoff depthas also lower in TER+ as compared to CT, but with signifi-

ant differences starting from 2008. Mean runoff depth duringhe rainy seasons (June–September) of the five study periodsas 76, 96 and 118 mm from DER+, TER+ and CT, respectively.

he runoff depth in 2009 in the grass pea growing season waslightly lower compared to all other drier and wetter years dur-

ng the study period for all treatments except in DER+ in 2010Fig. 4a). Five years mean runoff coefficient during the growingeason was significantly lower in DER+ and TER+ treatments com-ared to CT, with 19, 24 and 30%, respectively (Fig. 4b). The mean

t the growing period (n = 3). DER+ is derdero+, TER+ is terwah+, CT is conventionaloff for 2006 was not measured.

5-yrs percent deviation of runoff was −55 and −24% in DER+and TER+, respectively. The runoff depth percent deviation waslowest in DER+ followed by TER+ compared to CT in all years(Fig. 6).

Soil loss was significantly different (p < 0.05) between treat-ments in all years except in 2005. Four year mean soil loss of 14,17 and 26 t/ha was recorded from DER+, TER+ and CT, respectively(Fig. 5). Differently from runoff depth, highest soil loss for all treat-ments was observed with grass pea cropping in 2009, even thoughit was a relatively drier year compared to 2005 and 2010 whenwheat was grown and slightly wetter year compared to 2008 whenhanfets was grown (Table 1). The mean 4-yrs soil loss percent devia-tion was −75% in DER+ and −42% TER+. The lowest soil loss percentdeviation was observed in DER+ in 2008 and 2010 during hanfetsand wheat cropping (Fig. 6). The percent deviation of soil loss washigher as compared to runoff in 2008 and 2009 in DER+ and TER+

while only in TER+ in 2010. The percent deviations of runoff depthand soil loss in DER+ decreased with time, with the highest devia-tions in 2005 (in the first year) and the lowest percent deviation in2010.

58 T. Araya et al. / Field Crops Research 132 (2012) 53–62

F iod (ns 2007 w

3

Ddiums

3

afpfisDyiy

ig. 5. Mean yearly soil loss (t/ha) from each treatment throughout the growing perhown represent standard error of mean (p < 0.05). Note that soil loss for 2006 and

.2. Soil organic matter

Soil organic matter (%) was shown to increase significantly inER+ and TER+ as compared to CT for 0–15 and 15–30 cm soilepths, while the former two CA treatments did not show a signif-

cant difference (Fig. 7). The soil organic matter was higher in thepper 0–15 cm as compared to 15–30 cm soil depth while its incre-ent due to effects of CA treatments continued in the 15–30 cm

oil depth with similar trend to the upper soil.

.3. Crop performance

Significant improvements of grain yield have become consistentfter a period of five years of cropping in CA type treatments, i.e.,rom 2009 (Table 2). Earlier, in 2006, grain and straw yield of grassea in CT were found to be significantly higher, but in 2009, i.e.,ve years after the start of the CA experiments, grass pea grain andtraw yield were found highest in TER+ followed by DER+, whereas

ER+ followed by TER+ resulted in highest wheat grain and strawield in 2010. The mean of two years grass pea did not show signif-cant yield differences, whereas the mean of three years of wheatield was significantly highest in DER+ (p < 0.05) (Table 2). Percent

Fig. 6. Runoff and soil loss percent deviation between 2005 and 2010 (n = 3). Not

= 3). DER+ is derdero+, TER+ is terwah+, CT is conventional tillage practice. The barsere not measured.

deviation of grain yield in TER+ was negative showing lower yieldin CA treatments as compared to CT until year 2007 while onlyin 2006 in DER+. The highest percent deviation of grain yield wasrecorded in TER+ (28.6%) in 2009 followed by DER+ (23.1%) in 2010when grass pea and wheat were grown, respectively (Fig. 9). Thelast 3-yrs mean percent deviation of straw yield was 14.5 and 10.5%in DER+ and TER, respectively. Plant height was found to be signif-icantly higher in DER+ followed by TER+ compared to CT in 2010throughout the growing season (Fig. 8).

The mean farmers’ evaluation for crop performance was highestin DER+ (average of 7.6 on maximum of 8) in 2010 when wheat wasgrown, with a positive percent deviation of 43% (Table 2). The meanfarmers’ evaluation for 3-yrs was 6 for DER+, 5.6 for TER+ and 4.8for CT.

3.4. Rain water use efficiency

The mean rain water use efficiency (RWUE) of wheat cropping

was found significantly higher in DER+ in 2010, compared to TER+and CT. The 3-yrs average for wheat was significantly higher inDER+ compared to CT. Grass pea showed significantly lower val-ues of RWUE in 2006 for the CA treatments, but higher values in

e that runoff for 2006 and soil loss for 2006 and 2007 were not measured.

T. Araya et al. / Field Crops Research 132 (2012) 53–62 59

cm s

2eC

4

4

rpcwstw

TG(at

Fig. 7. Mean soil organic matter (%) for the three treatments for 0–15 and 15–30

009. The total productivity of the cropland during the six yearsxperimental period was slightly higher in DER+ as compared toT (Table 2).

. Discussion

.1. Effects on soil loss and runoff

Over the five monitored years, DER+ significantly (p < 0.05)educed runoff depth and runoff coefficient, followed by TER+ com-ared to CT (Fig. 4a and b). This consistent reduction in runoffoefficients in DER+ was likely due to the capacity of the 25–35 cm

ide furrows to retain a large proportion of runoff as depression

torage (Sayre, 1998). Reduced runoff means an improvement inhe soil water status in the root zone and a reduction in soil loss,hich in turn leads to reduced land degradation and reduced crop

able 2rain and straw yield, rain water use efficiency and farmers’ evaluation from 2005 to 201

p < 0.05). DER+ is derdero+, TER+ is terwah+, CT is conventional tillage practice, SEM is stapplicable. Means not with same letter are significantly different. Note that straw yield wreatment where the maximum for a treatment is 8, the minimum is 2.

Year Crop type Treatment Grain yield (t/ha)mean ± SEM

2005 Wheat DER+ 3.1 ± 0.2a

TER+ 2.7 ± 0.2a

CT 2.8 ± 0.1a

2006 Grass pea DER+ 2.1 ± 0.1b

TER+ 2.1 ± 0.1b

CT 2.9 ± 0.2a

2007 Wheat DER+ 3.0 ± 0.1a

TER+ 2.9 ± 0.2a

CT 2.9 ± 0.3a

2008 Hanfets DER+ 2.0 ± 0.1a

TER+ 1.9 ± 0.1a

CT 1.9 ± 0.1a

2009 Grass pea DER+ 2.2 ± 0.2ab

TER+ 2.8 ± 0.1a

CT 2.0 ± 0.2b

2010 Wheat DER+ 5.2 ± 0.1a

TER+ 4.5 ± 0.15b

CT 4.0 ± 0.2c

2006, 2009 Mean grass pea (2yrs) DER+ 2.2 ± 0.2a

TER+ 2.4 ± 0.1a

CT 2.5 ± 0.2a

2005, 2007, 2010 Mean wheat (3yrs) DER+ 3.7 ± 0.1a

TER+ 3.4 ± 0.2ab

CT 3.2 ± 0.2b

2005–2010 Mean (6yrs) DER+ 2.9 ± 0.1a

TER+ 2.8 ± 0.1a

CT 2.8 ± 0.2a

oil depths. DER+ is derdero+, TER+ is terwah+, CT is conventional tillage practice.

water stress (Stroosnijder, 2009; Rockström, 1997). When wheat,barley and hanfets was harvested (October–November), ample cropstubble (30% soil cover) remained on DER+ and TER+ plots (Fig. 3)that induced a vertical mulching effect, resulting in higher waterinfiltration than in fields without residue (Mostaghimi et al., 1988).Grass pea has pods containing grain on the lower parts of the stemmaking difficult to leave 30% soil cover of crop residue at harvest.We tried to bring part of the grass pea straw back to the plotsafter threshing, but it was taken away by wind. In addition to thecontribution of leaving crop residue and reduction in tillage everyyear, variations across years may be related to crop or variety type(canopy cover and root structure difference between crops) and

rainfall amount (wetting degree of Vertisols), intensity and distri-bution (length of dry spell). Also the Vertisol soil was often dryand deeply cracked. Thus, in the presence of stubble cover andcracks, early summer rainfall infiltration was higher in DER+ and

0 (n = 3) for the experimental site on a Vertisol in May Zegzeg catchment, Ethiopiandard error of mean (p < 0.05), RWUE is rainwater use efficiency; (−) indicates notas not measured for 2005. Farmers’ evaluation values shown are mean scores per

Straw yield (t/ha)mean ± SEM

RWUE (%) mean ± SEM Farmers evaluationmean

– 6.8 ± 0.2a –– 6.0 ± 0.2a –– 6.0 ± 0.1a –2.8 ± 0.1b 5.2 ± 0.1b –2.9 ± 0.1b 5.0 ± 0.1b –4.1 ± 0.2a 7.0 ± 0.2a –6.8 ± 0.3a 9.2 ± 0.1a –6.7 ± 0.6a 9.0 ± 0.2a –6.9 ± 0.8a 9.0 ± 0.3a –3.1 ± 0.1a 4.1 ± 0.0a 4.83.0 ± 0.2a 4.0 ± 0.1a 5.03.0 ± 0.2a 4.0 ± 0.1a 4.93.9 ± 0.4ab 6.5 ± 0.2ab 5.54.4 ± 0.2a 8.0 ± 0.1a 6.93.5 ± 0.3b 6.0 ± 0.1b 5.36.7 ± 0.7a 12.9 ± 0.1a 7.65.3 ± 0.6b 11.0 ± 0.1b 5.34.7 ± 0.1c 10.0 ± 0.1c 4.33.4 ± 0.3a 5.8 ± 0.2b –3.6 ± 0.2a 6.0 ± 0.1ab –3.8 ± 0.2a 7.0 ± 0.2a –6.7 ± 0.5a 10.8 ± 0.1a –6.0 ± 0.6ab 10.0 ± 0.2ab –5.8 ± 0.5b 9.0 ± 0.2b –4.2 ± 0.2a 7.3 ± 0.1a –4.0 ± 0.2a 7.0 ± 0.1a –4.0 ± 0.2a 7.0 ± 0.2a

60 T. Araya et al. / Field Crops Res

Fir

TfiAbma

phictpbiAsttstsfbt

ig. 8. Plant height of wheat during the growing season of 2010 (n = 10). DER+s derdero+, TER+ is terwah+, CT is conventional tillage practice. The bars shownepresent standard error of mean (p < 0.05).

ER+ compared to CT treatments, where cracks, moreover, werelled with soil due to the frequent tillage at and before sowing.fter swelling and closing of cracks, infiltration rate is determinedy the amount of stubble protecting soil, crop cover and organicatter accumulated in the soil (Fig. 7) from direct rainfall impact

nd inducing preferential flow along the stubble.CA treatments retained the greatest stubble cover (Fig. 3) and

roduced the least average runoff during the study (Fig. 4a). Theighest soil loss was produced during the grass pea cropping season

n 2009, which is related to late sowing date and low vegetationover of this plant due to wider spacing at sowing (Fig. 5). Accordingo the crop requirements and local practice in the study area, grassea is usually sown in the third week of August while wheat andarley are sown from mid June to early July. The higher soil loss

n grass pea was because the fresh tillage carried out at sowing inugust made the soil susceptible to erosion resulting in increasedediment concentration in each runoff (data not shown). In additiono the previous tillage in May and June, the CT system was tilledwice to control weeds before sowing, i.e., in August 8, 2009 and atowing in August 18, 2009 while once at sowing in the CA systemso bury sown seeds. These tillage operations in August aggravated

oil erosion rates (Fig. 5). In 2009, August received highest rainfallollowing the month of July (Fig. 2). The general differential trendetween CA and CT treatments in runoff and soil loss from the firsto the sixth year was increasing especially after four years in 2008.

Fig. 9. Grain yield percent deviation from

earch 132 (2012) 53–62

This may be attributed to the conversion of crop residue retainedin CA treatments and accumulation of organic matter over time inDER+ and TER+ plots that probably caused a significant differencein soil quality (Fig. 7; Oicha et al., 2010). The increment in organicmatter in the CA treatments may contribute to the reduction ofdirect runoff and increase infiltration rate.

The types of CA practices evaluated in our research, i.e., DER+(permanent raised beds with retained crop residues) and TER+ (fur-rows at wider intervals plus retained crop residue) are newly testedin Ethiopia. Our research shows that, in addition to its positive effecton soil loss and runoff reduction, DER+ drains excess rain waterfrom the raised beds to the furrow (where it is kept for later use bythe crops during dry spells), enhancing the crop grown on top ofthe raised bed, thus indicating its usefulness also in the high rainfallareas of central highlands of Ethiopia. Erkossa et al. (2005) reportedhigher runoff and soil losses in central highlands of Ethiopia frombroad bed and furrows compared to conventional tillage. This isbecause they constructed graded furrows to drain excess waterfrom the fields. However, such kind of graded furrows may aggra-vate the cropland degradation and lead to lower crop yields in thelong term. Our study also demonstrated that CA based on DER+planting with oxen ploughing system keeps no-till condition andstanding stubble on top of the permanent raised bed. The top soilwhere farmers also apply fertilizer every year is the most fertilesoil while this part of soil is exposed to water erosion. Therefore,the reduction of fertile top soil loss and addition of crop residueevery year that increased organic matter in DER+ and TER+ plantingsystem could bring a sustainable crop yield improvement over time.

4.2. Effects on crop productivity over time

This study showed that CA treatments are promising for thefarmers on Vertisol with equal or higher crop yield in CA systemsthan CT during the 6 yrs study period except for grass pea in thesecond year (2006) (Table 2). Improvements in crop yield in CAtreatments required a period of at least five years of cropping beforethey became significant, which coincides with findings by Govaertset al. (2005). In contrast to our findings, Tesfay et al. (2011) obtaineda significantly higher crop yield in CA system in a shorter period,i.e., after three years which may be due to lower rainfall amount,

more level land and longer dry spells at the Adigudem study sitethat can result in lower runoff and soil loss in the CA system whichcontributed to obtaining sustainable higher yields after three yearsof implementation.

CT between 2005 and 2010 (n = 3).

ps Res

TsfiuotcaStssTm

ilburrowiCittrps

adiiyma

ueicr

slgddpdtostodhCrf2d

T. Araya et al. / Field Cro

Unlike 2006, the yield of grass pea was significantly higher inER+ followed by DER+ compared to CT in 2009. The main rea-ons for lower grass pea yield in CA treatments in 2006 could berstly the absence of chemical weed control which resulted innsuppressed weed growth and yield losses. A similar effect wasbserved in tef in a nearby area (Oicha et al., 2010). In this study,he use of glyphosate herbicide as of 2007 and mulching effects ofrop residue in CA plots greatly declined weed population in everydditional study year by reducing the weed seed bank in the soil.econdly, it could have resulted from excess wetness. Farmers inhe area plant grass pea late in the cropping season to avoid exces-ive soil moisture conditions which may explain the presence oflightly higher percentage crop yield in TER+ as compared to DER+.o further increase grain yield in DER+ adjustments in planting timeay be required.In 2010, the shortest duration of crop emergence was observed

n the DER+ planting system. Plant height in 2010 was significantlyarger in DER+ throughout the growing season but especially in theeginning of the rainy season, when, unlike in CT, growth remainsnaffected by short dry spells (Fig. 8). Seleshi and Camberlin (2006)eported that in northern Ethiopia (Mekelle) the length of dry spellseaches 20 days on average, with a concentration at the beginningf the main rainy season. Our results demonstrated that the rainater use efficiency was higher in DER+ and TER+ systems indicat-

ng their importance in reducing dry spell effects (Table 2), as theA treatments capture rain water, improve infiltration, and hence

mprove crop yield. Farmers in northern Ethiopia widely practicehe derdero planting system for fenugreek, wheat and tef and theerwah system for tef (Nyssen et al., 2011). However, our researchesults support the use of a modified version of these traditionalractices, i.e., DER+ and TER+ system, for all types of crops to ensureoil and water conservation.

In their controversial paper, Giller et al. (2009) doubt thepplicability of CA on smallholder farms in Sub-Saharan Africaue to yield losses or no yield benefits up to nearly 10 yrs of

mplementation. Our results demonstrate however that consistentmprovement of grain and straw yield are obtained as of the fifthear of CA implementation. This implies that long term involve-ent is required to convince farmers to adopt CA in similar target

reas.Our results also show that local farmers’ evaluation of crop stand

nder DER+ and TER+ treatments is increasing each year while theirvaluation of crops under CT declines (Table 2). The percentagencrement of wheat straw in 2010 in DER+ was higher by 43% asompared to CT, which is 13% higher than the 30% crop residueetained.

However, CA can also be promoted by several other benefitstarting from the first year. Farmers who do not have oxen often sowate what results in lower yields, or they need to pay 50% of theirrain and straw yield to get their land ploughed. Here, the intro-uction of CA will significantly minimize the cost of hiring oxenue to the avoidance of repeated ploughing and the fast ploughingace. An average of three to four tillage practices using the oxenrawn mahresha ard plough is common for most crops grown inhe area. The first and the subsequent tillage requires an averagef 10–11 oxen-span days per ha, depending on the strength of theoil. However, all these tillage are avoided in the CA system. In addi-ion, the tillage at sowing was 4 times faster in DER+ system (1xen-span day per ha) compared to the CT system (4 oxen-spanays per ha). It will also be a cheaper technology for resource poorouseholds–women are traditionally discouraged from ploughing.onservation agriculture can also be beneficial as the lower traction

equirement allows replacing oxen with milk cows which enablesarmers to decrease expenses of oxen maintenance (Tulema et al.,008). A large number of livestock is required to meet the poweremand of the current conventional tillage system which also

earch 132 (2012) 53–62 61

contributes considerably to land degradation by extreme grazing(Gebregziabher et al., 2006). Targeting poor resource farmers suchas those without oxen in CA extension can fasten adoption in areaswhere there is a zero grazing system implemented. Furthermore,CA and the concomitant spraying of glyphosate strongly decreasethe need for women’s and children’s labour in weeding. Therefore,promoting conservation agriculture is an important interventiontowards achieving food security through sustainable farming.

5. Conclusion

The derdero+ tillage system with a permanent raised bed plant-ing system and retention of crop residues, and to a lesser extentthe terwah+ tillage system with retention of crop residue werefound to be beneficial for increasing yields of wheat and hanfets,and reducing runoff and soil loss in northern Ethiopia. Soil organicmatter was significantly increased in the CA treatments leadingto the reduction of runoff and increase infiltration rate. Overall,the DER+ planting system can be an efficient in situ soil and waterconservation strategy through reducing the runoff coefficient anddraining excess water from the beds through furrow storage. Ourresults demonstrate the importance of DER+ in reducing soil lossand runoff, thus increasing crop productivity and avoiding fur-ther land degradation. However, the improvement in soil physical,chemical and biological properties is not immediate and the fullbenefit of permanent raised beds plus retention of crop residues canonly be expected after several years. Planting time adjustments inDER+ treatments are suggested to further increase grass pea yield,which in our experiments was higher in TER+ than in DER+ afterfive years of CA.

Notwithstanding the overall better results of DER+ compared toTER+, the latter tillage system can be recommended as a first stepfor reducing runoff and soil loss, while increasing crop yield. Thelong-term goal should be to achieve a permanent raised bed plant-ing system along with the use of crop residues (DER+). In additionto the positive effects on runoff, soil loss and crop yield, we arguethat avoiding repeated tillage which is 10–11 oxen-span days perha and the faster ploughing pace in DER+ will enable a reduction inoxen density with further natural resource benefits. DER+ and TER+are improvements to good local practices that qualify them as CA;we recommend large scale dissemination and implementation onVertisols and possibly on other soils in northern Ethiopia.

Acknowledgments

This research was funded by the Institutional University Cooper-ation (IUC) Programme in Mekelle University through the FlemishInteruniversity Council (VLIR, Belgium). The authors are sincerelygrateful to Romha Assefa and Alemu Gebremedhin for theirunreserved assistance during the field work and managing theexperimental plots. Many local farmers provided useful insightsand assistance to the researchers. We are grateful to Relief Societyof Tigray office, Cultural and Information office, and Agricultureand Rural Development office in Dogua Tembien district for theirgenerous hospitality and support.

References

Ahuja, L.R., Ma, L., Timlin, D.J., 2006. Trans-disciplinary soil physics research crit-ical to synthesis and modeling of agricultural systems. Soil Sci. Soc. Am. J. 70,311–326.

Angers, D.A., Mehuys, G.R., 1989. Effects of cropping on carbohydrate content and

water-stable aggregation of a clay soil. Can. J. Soil Sci. 69, 373–380.

Bezuayehu, T., Gezahegn, A., Yigezu, A., Jabbar, M.A., Paulos, D., 2002. Nature andcauses of land degradation in the Oromiya Region: a review. Socio-economicsand Policy Research Working Paper 36. International Livestock Research Insti-tute (ILRI), Nairobi, Kenya, p. 34.

6 ps Res

D

E

F

G

G

G

G

G

G

K

M

M

M

N

N

2 T. Araya et al. / Field Cro

eckers, J., Spaargaren, O., Nachtergaele, F., 2001. Vertisol: genesis, properties andsoilscape management. In: Syers, K., Penning de Vries, F., Nymundeza, P. (Eds.),The Sustainable Management of Vertisols. CABI Publishing, New York, pp. 3–20.

rkossa, T., Stahr, K., Thomas, G., 2005. Soil tillage and crop productivity on a Vertisolin Ethiopian highlands. Soil Till. Res. 85, 200–211.

reebairn, D.M., Loch, R.J., Silburn, D.M., 1996. Management of Vertisols in rainfedconditions. In: Ahmad, N., Mermut, A. (Eds.), Vertisols and Technologies for theirManagement. Elsevier, Amsterdam, pp. 303–362.

ebregziabher, S., Mouazen, A.M., Van Brussel, H., Ramon, H., Nyssen, J., Verplancke,H., Behailu, M., Deckers, J., De Baerdemaeker, J., 2006. Animal drawn tillage, theEthiopian ard plough, mahresha: a review. Soil Till. Res. 89, 129–143.

ebreegziabher, T., Nyssen, J., Govaerts, B., Getnet, F., Behailu, M., Haile, M., Deckers,J., 2009. Contour furrows for in situ soil and water conservation, Tigray, northernEthiopia. Soil Till. Res. 103, 257–264.

iller, K.E., Witter, E., Corbeels, M., Tittonell, P., 2009. Conservation agriculture andsmallholder farming in Africa: the heretics’ view. Field Crops Res. 114, 23–34.

irma, T., 2001. Land degradation: a challenge to Ethiopia. Environ. Manage. 27,815–824.

oebel, W., Odenyo, V., 1984. Ethiopia. Agroclimatic resources inventory for land-use planning. Ministry of Agriculture, Land Use Planning and RegulatoryDepartment, UNDP, FAO. Technical report DP/ETH/78/003. Vol. I, p. 208, vol.II, pp. 95.

ovaerts, B., Sayre, K.D., Deckers, J., 2005. Stable high yields with zero tillage andpermanent bed planting? Field Crops Res. 94, 33–42.

ay, B.D., 1990. Rates of change of soil structure under different cropping systems.Adv. Soil Sci. 12, 1–52.

ati, B.M., 2005. Overview of water and soil nutrient management under small-holder rain-fed agriculture in east Africa. International Water ManagementInstitute. Working Paper 105.

cHugh, O.V., Steenhuis, T.S., Abebe, B., Fernandes, E.C.M., 2007. Performance ofin situ rainwater conservation tillage techniques on dry spell mitigation and ero-sion control in the drought-prone north Wello zone of the Ethiopian highlands.Soil Till. Res. 97, 19–36.

ostaghimi, S., Dillaha, T.A., Shanholtz, V.O., 1988. Influence of tillage systems andresidue levels on runoff, sediment, and phosphorus losses. Am. Soc. Agric. Eng.31, 128–132.

yssen, J., Vandenreyken, H., Poesen, J., Moeyersons, J., Deckers, J., Mitiku Haile,Salles, C., Govers, G., 2005. Rainfall erosivity and variability in the northernEthiopian Highlands. J. Hydrol. 311, 172–187.

yssen, J., Poesen, J., Gebremichael, D., Vancampenhout, K., D’aes, M., Yihdego,G., Govers, G., Leirs, H., Moeyersons, J., Naudts, J., Haregeweyn, N., Haile,

earch 132 (2012) 53–62

M., Deckers, J., 2007. Interdisciplinary on-site evaluation of stone bundsto control soil erosion on cropland in northern Ethiopia. Soil Till. Res. 94,151–163.

Nyssen, J., Govaerts, B., Tesfay, A., Cornelis, W.M., Bauer, H., Haile, M., Sayre, K.,Deckers, J., 2011. The use of the marasha ard plough for conservation agriculturein northern Ethiopia. Agron. Sustain. Dev. 31, 287–297.

Odendo, M., Ouma, J., Wachira, S., Wanyama, J., 2003. Economic assessment of maizeyield loss due to stem borer in major maize agro-ecological zones of Kenya. In:African Crop Science Conference Proceedings, vol. 6, pp. 683–687.

Oicha, T., Cornelis, W.M., Verplancke, H., Nyssen, J., Deckers, J., Behailu, M., Haile,M., 2010. Short-term effects of conservation agriculture on Vertisols under tef(Eragrostis tef (Zucc.) Trotter) in the northern Ethiopian highlands. Soil Till. Res.106, 294–302.

Papendick, R.I., Parr, J.F., 1997. No-till farming: the way of the future for a sustainabledryland agriculture. Ann. Arid Zone 36, 193–208.

Raes, D., Willems, P., GBaguidi, F., 2006. RAINBOW – A software package for hydrom-eteorological frequency analysis and testing the homogeneity of historicaldata sets. In: Proceedings of the 4th International Workshop on Sustain-able management of marginal drylands. Islamabad, Pakistan, 27–31 January,pp. 41–55.

Rockström, J., 1997. On-farm agrohydrological analysis of the Sahelian yield cri-sis: Rainfall partitioning, soil nutrients and water use efficiency of pearl millet.Dissertation. Stockholm University. Akademitryck AB, Edsbruk, Sweden.

SAS, 2002. JMP Software Version 5. SAS Institute Inc., Cary, NC, USA.Sayre, K.D., 1998. Ensuring the use of sustainable crop management strategies by

small wheat farmers in the 21st century. Wheat special report no. 48. Mexico,D.F.: CIMMYT.

Seleshi, Y., Camberlin, P., 2006. Recent changes in dry spell and extreme rainfallevents in Ethiopia. Theor. Appl. Climatol. 83, 181–191.

Stroosnijder, L., 2009. Modifying land management in order to improve effi-ciency of rainwater use in the African highlands. Soil Till. Res. 103,247–256.

Tesfay, A., Cornelis, W.M., Nyssen, J., Govaerts, B., Bauer, H., Gebregziabher, T., Oicha,T., Raes, D., Sayre, K.D., Haile, M., Deckers, J., 2011. Effects of conservation agri-culture on runoff, soil loss and crop yield under rainfed conditions in Tigray,northern Ethiopia. Soil Use Manage. 27, 404–414.

Tulema, B., Aune, J.B., Johnsen, F.H., Vanlauwe, B., 2008. The prospects of reducedtillage in tef (Eragrostis tef Zucca) in Gare Arera, West Shawa Zone of Oromiya,Ethiopia. Soil Till. Res. 99, 58–65.

Walkley, A., Black, C.A., 1934. Estimationof organic carbon by chromic acid titrationmethod. Soil Sci. 37, 29–38.