African Crop Science Journal African Crop Science Society ISSN: 1021-9730 EISSN: 2072-6589 Vol. 8, Num. 2, 2000, pp. 159-170

African Crop Science Journal, Vol. 8. No. 2, pp. 159-170

The effect of plant population and water conservation methods on grain yield of Early-maturing maize varieties in the Moisture-Stress areas of Southern Ethiopia

M.A. Hussein, Asmund Bjornstad¹, Are Halvor Aastveit², Trygve Berg³ and Haile Geremew⁴
Awassa College of Agriculture, P. O. Box 5, Awassa, Ethiopia
^1,2,3Agricultural University of Norway, Department of Horticulture and Crop Sciences, P. O. Box 5022,
Department of Mathematics, P. O. Box 5035, NORAGRIC , P. O. Box 5001, N-1432, Ås, Norway
⁴645 Marshall Avenue , Lincoln , NE 68510 , U.S.A.

(Received 24 February, 1999; accepted 14 February, 2000)

Code Number: CS00017

INTRODUCTION

It can generally be said that the moisture stress zones devoted to rainfed agriculture are those that have 3-5 humid months. Humid months are those when precipitation exceeds evapotranspiration (Kumar et al., 1993; Chapman and Barreto, 1996). The annual rainfall in these regions is usually from 200-400 mm, but sometimes the sub-humid areas with annual rainfall of 600-1000 mm can be included in the moisture stress areas.

Moisture-stress zones are characterised by high interannual rainfall variability ranging from 25 to 50% and an erratic and unpredictable rainfall coming in few heavy showers within a short rainy season that may contain prolonged dry spells of 2-5 weeks (Edmeades et al., 1987; Rowland and Whiteman, 1993). A large amount of this precipitation, therefore, does not infiltrate into the soil, but causes runoff and erosion. The annual mean potential evapotranspiration is much higher than annual precipitation, and this causes a moisture deficit which can be as high as 1000 mm annually (Educational Materials Production and Distribution Agency (EMPDA) of Ethiopia, 1984).

Among the cereals, sorghum and pearl millet have been the main crops grown in semi-arid regions due to their ability to efficiently extract and use water and due to their heat tolerance (Kumar, 1993). Maize is becoming one of the major cereal crops due to its potential to give higher yields in good years, although it is much inferior to sorghum and pearl millet under low rainfall conditions (Ohiagu, 1987; Rodriguez, 1987). It is sensitive to drought during establishment and 1-2 weeks before and after flowering. Maize is unusual in that the stigma and pollen tubes must extend to a length of about 30 cm within a very short span of time. Therefore drought at silking, which can cause delayed silk growth and embryosac abortion can lead to disastrous yield losses. Except for these critical periods, maize is relatively drought tolerant (Edmeades et al., 1987). As such, it is favoured by farmers in the moisture stress areas and is now moving into more arid areas replacing pearl millet and sorghum. This success of maize is in part due to its low labour demand, tolerance to bird attacks, ability to fill the hunger period before sorghum and pearl millet harvests and diverse use (Ohiagu, 1987; Asfaw et al., 1992; Lemma et al., 1992 ).

Fascinating traditional water harvesting techniques have been documented (Reij et al., 1988), but simple in-situ water conservation methods requiring minimum labour and capital have been found to increase yields dramatically in the moisture stress areas. These are ridge and furrow systems where the ridges serve as the catchment area and the furrows as storage place. They minimise runoff and maximise infiltration. Maize grain yield increase due to tied ridges (closed furrows) can be as high as 2000 kg ha-1 when soil fertility is not a yield limiting factor, and yield increase of 1 ton ha-1 are common (Rodriguez, 1987). Summarising trials conducted in Tigray (Ethiopia), Alem (1987) reported that tied ridges increased yields by 50 to 80% as compared to flat planting. Open furrows (ridges) were better than flat planting but gave yields inferior to those of tied ridges. Thomas and Kaspar (1995) in a greenhouse experiment reported that unridged plants developed an average of 38.4 functional nodal roots, while ridged plants developed 50.0. Increasing ridge height increased the number of nodal roots. In the field, plants from no-till developed 53.1 and 46.0 functional nodal roots in 1992 and 1993, respectively, while ridged plants developed 58.7 and 49.1 nodal roots.

Mulching is another water conservation system that is widely used in the drought-prone areas. Mulching protects the soil surface from direct hit by rain and reduces surface capping thereby reducing runoff and erosion and increasing infiltration. It reduces soil temperature thereby reducing evaporation from the soil surface and conserving moisture. Adeoye (1984, 1990) quoted by Kumar (1993) reported that application of grass mulch at a rate of 5 t ha-1 at Zaria (Nigeria) reduced soil temperature by 4-10 0C at 5-10 cm soil depth, increased maize grain yields by 10% and maintained soil moisture in the 0-20 cm soil horizon at a favourable level. Mando and Stroosnijder (1999) also showed that the biological activity of termites on mulched plots significantly increased water infiltration and dry matter yield. Numerous termite voids were opened on the sealed surface of the crusted soil. Runoff was significantly reduced, thereby reducing erosion.

Alem (1987) reported that tillage methods and mulching did not show any significant advantage in the black clay soil at Mekele, whereas seed bed preparation methods increased the yield of wheat, sorghum, cowpea and mungbean in the drought affected areas of Mekele, Quihea and Kobbo. Jones et al. (1997) found out that no-tillage residue management increased average soil water content at planting of the next crop by 22 mm with wheat-fallow, and 29 mm with continuous wheat systems. Shock et al. (1997) studied the effect of mechanical straw mulching of irrigation furrows on soil erosion and nutrient losses. They reported that 900 kg ha^-1 wheat straw mechanically applied to irrigation furrows reduced runoff by 43%. Cumulative sediment loss after 17 irrigations was 17 megagrams ha^-1 for unmulched furrows. Straw reduced NO₃, NH₄⁺ and PO₄ losses inrunoff solution and sediment.

Investigating the effect of plant population on two maize varieties, Dowker (1963) concluded that above a seasonal rainfall of 200 mm 36000 plants ha-1 gave the best yield for a late variety that matured in 5 and half months. For an early variety maturing in 4 months 72000 plants ha-1 gave the highest yields. Below this amount of seasonal rainfall, 12000 to 18000 plants ha-1 gave superior yields for the late variety whereas the early variety had its maximum yield at a population of 36000 plants ha-1. Rodriguez (1987) recommended a plant density of 50000 to 65000 plants ha-1 for intermediate maturity maize varieties and a population density of 65000 to 90000 plants ha-1 for early varieties in the semi-arid areas of West Africa under optimum fertility and rainfall conditions. He recommended a density of about 25000 plants ha-1 under very dry or poor soil fertility for intermediate varieties. Tollenaar et al. (1994) found out that increasing maize plant density from 4 to 10 plants m-2 reduced weed biomass by up to 50%. The grain yield reductions attributed to high weed pressure were 26, 17, and 13% for maize densities of 4, 7, and 10 plants m-2, respectively. The authors concluded that the competitiveness of maize with weeds can be enhanced by increasing plant density.

The moisture stress zones of Ethiopia correspond to an annual rainfall of 700 mm in the North, 775 mm in the Northeast and East and 900 mm in the South, where the rainfall pattern is bimodal. These areas cover 55% of the whole area and 38-42% of the total maize growing areas (Haile, unpublished manuscript,1988). The dryland areas of Ethiopia account for 46% of the total arable land, and the semi-arid areas cover 40% whereas the arid areas cover 60% of the dryland areas (Alem, 1987; Mengistu and Ermias, 1988).

Because maize is one of the most important crops in these areas, an intensive breeding programme ("Maize improvement for the moisture stress zones of Ethiopia") was initiated at the Awassa College of Agriculture in 1976, with a mandate to develop varieties that could mature in less than 120 days and give reasonable yields (up to 7-8 t ha-1 under favourable conditions and above 1 t ha-1 under relatively unfavourable conditions). A synthetic variety, Alamura Yellow, was released in 1984 and two other varieties, Fetene and Tesfa, were released in 1996. Agronomic studies on maize dryland farming techniques, such as the various water conservation methods, effects of sowing dates, time of fertiliser application, plant densities and intercropping were conducted during 1989-1992 in cooperation with NORAGRIC (Centre for International Environment and Development Studies). The purpose of these studies was to investigate the success of the breeding programme by comparing the developed varieties with the check, Katumani, and find out the best combinations of agronomic practices that maximise yield, making a package of information ready to be released with the varieties. In this paper the results of the multilocation experiment on the effects of planting densities and the various water conservation methods on maize grain yields are reported.

MATERIALS AND METHODS

Four early-maturing maize varieties (about 120 days to maturity), Katumani, Birkata , Mirtchaye and Alamura Yellow were investigated under four plant population levels and four water conservation methods at five locations in Southern Ethiopia, Awassa (380 29’E , 70 9’N), Zuwai ( 380 7’ E, 70 9’N) , Yabello ( 380 7’E, 40 9’N), Mega (380 22’ E, 40 N) and Bidre ( 390 27’E , 50 53’N) during the 1989 cropping season. During the 1990-92 cropping seasons two of the above varieties, Katumani and Birkata, were further investigated under four water conservation methods and five plant population levels.

The soil of Awassa site is Fluvisol (inceptisol) and appears to be suitable for mechanical cultivation and has good moisture and nutrient availability. Soils at Zuwai are also Fluvisols, shallow, and the availability of phosphorous and some micronutrients is hampered by the high pH (Sheleme, 1994). Chlorosis of maize leaves is frequently observed at this site. Moreover, there is a sand layer cemented by calcium carbonate followed by an impervious layer of platty structure at the depth of about 1 metre that prevents root penetration. Water holding capacity of these soils is poor. Abundant termite mounds on which crop growth is very poor contribute to the worsening of water holding capacity of the soil at this site. Crop growth on termite mounds at Yabello and Bidre is better than on uninfested soil. The presence of humus and clay illuviation in the soil profile might explain this phenomenon (Sheleme, 1991).

The varieties studied. Katumani is a widely used early-maturing variety in moisture-stress areas of East Africa where the rainy season is short. It is a synthetic variety developed in Kenya in 1967. The other three varieties were developed by the maize improvement programme of the Awassa College of Agriculture, Ethiopia.

Birkata is an improved population which was developed from Katumani after two cycles of mass-selection alternated with one cycle of ear-to-row selection. Further improvement of this population was conducted by the half-sib selection method. Mirtchaye is a top-cross between Alamura white (a synthetic variety developed in 1984 by the maize improvement programme of the Awassa College) and an inbred line developed from Katumani (Kat-1-1-80). Alamura Yellow is a synthetic variety released in 1984 and was developed from 12 inbred lines.

Water conservation methods, plant population levels and experimental design. Three water conservation methods: mulching, closed or bunded furrows and open furrows were used. These were compared with the conventional planting which is conducted on flat seedbeds. Mulching material available at each station was utilised for the first method and the ground was covered immediately after planting. For bunded furrows, seeds were planted in furrows and the furrows were closed from both ends of the row. In open furrows they were left open.

During the 1989 cropping season, four plant population levels, 80 x 15 cm (83333 plants ha-1), 80 x 20 cm (62500 plants ha-1), 80 x 25 cm (50000 plants ha-1), and 80 x 30 cm (41666 plants ha-1) were studied. The seeding rate was increased by 50% at planting to ensure the required plant population levels, and plants were later thinned if more than the number required had established.

The treatments were arranged in a split-split-plot design and replicated three times, with the varieties as the main plots, population densities as the sub-plots and water conservation methods on the sub-sub-plots. The sub-sub-plots contained 5 rows each 3 m long, with a total plot area of 12 m2, except in Bidre during the 1989 cropping season where the rows were only 2 m long and the plot area 8 m2 due to shortage of land. A fertiliser rate of 46 kg P2O5 ha-1 and 64 kg N ha-1 was applied. The experimental field was kept free of weeds.

Because the 1989 trial indicated that increasing plant population increased yield, a higher plant population, 80 x 10 cm (125000 plants ha-1) was added and the same experiment conducted during the 1990-1992 cropping seasons in four replications but with only the two highest yielding varieties, Katumani and Birkata. Although the main analysis focused on the 1989 trials where the three elite varieties produced by the Awassa College of Agriculture have been compared with Katumani, conclusions on the effects of plant densities and water conservation methods are based on data of the four years (1989-1992).

Statistical analysis. Square root transformation of grain yield was used for analysis to overcome problem of heterogeneity of errors across locations (Box and Cox, 1964). The split-split-plot model (Montgomery, 1984) was adopted for the statistical analysis, where locations and replications within locations were treated as random effects and all the remaining factors were considered fixed. The effect of plant population on yield at different water conservation methods was further investigated using regression analysis. Quadratic regression was fitted where it gave a significant improvement over the linear one.

RESULTS AND DISCUSSION

Results of the analysis of variance for each year are summarised in Table 1. In the combined analysis of 1989 the location effect was dominant (57% of the total variance). Contribution of the replications, the sub-sub-plot error, and the sub-plot error were also high in the combined analysis of this year. Similar trends were observed in the remaining three years, except that the variety contribution to the total variance was reduced due to the fact that only Birkata and Katumani were included in these experiments. Despite the large errors that concealed differences between the factors under study, significant differences could be observed for the factors studied at individual locations and in the overall analysis.

Difference between locations. Difference between the five locations was highly significant in all years (P < 0.0001). Over the four years, yields were correlated with the amount of rainfall during the growing season (r=0.7, P = 0.01) , with each mm of rainfall within the range of 392 to 764 mm increasing grain yield by 9.47 kg ha-1. The Awassa site was the highest yielding in most years, with almost a two-fold yield of the other sites when averaged over the four years of the study (4.2 t ha-1 vs 2.7 t ha-1 or less). But, even at this relatively favourable site, dramatic yield reductions occurred in some years. For example, Awassa was one of the lowest yielding sites in 1990 and ranked 4th, while in 1992 the difference between Awassa and Zuwai, one of the low-yielding sites, was only 0.5 t ha-1.

Varietal differences. The varieties differed significantly (P < 0.05) from each other in grain yield in 1989 (Table 1). The difference between Katumani and Birkata on one side and Alamura Yellow and Mirtchaye on the other, was statistically significant. The difference between Katumani and Birkata was not statistically significant. The same was true for Alamura Yellow and Mirtchaye. None of the cultivars outyielded Katumani at any location during the 1989 cropping season, although its superiority diminished as environments became more unfavourable (Fig. 1). The other three interchanged ranks over locations, and the variety-by-location interaction was significant (P = 0.001). Katumani was superior to Birkata during the 1989-1991 cropping seasons, with Birkata outyielding it only during the 1992 cropping season. This might be the result of further selection by the half-sib selection method continued during this period, or it may have been due to year x variety interaction effects.

Effects of plant populations on grain yield. The results of regression analysis of the effects of plant populations on grain yield are given in Figure 2. To highlight the results, we have included only three of the five locations during the growing seasons of 1989 and 1990. We have regressed yield on the number of plants per plot at harvest as converted to thousand plants per hectare. Where quadratic fit brought about a significant improvement over the linear one, we fitted the former (Zuwai, 1990; Bidre, 1990) (Fig. 2). We have also included the actual data points for only one water conservation method (Flat with mulching) as dots on the graph. The annual rainfall and the trial mean yields in t ha-1 are shown on the title of the graphs.

Out of 16 trials conducted during the 1989-92 cropping seasons, the effect of plant populations was statistically significant (P< 0.05) in only 6 cases (Awassa 89, 91, 92, Zuwai 90, Mega 90, and Bidre 91). These were mostly seasons with high rainfall, with seasonal rainfall above 600 mm. Under such favourable conditions higher plant densities gave higher grain yields. The highest yield was obtained around a population density of 83000 plants ha-1 and its average yield advantage over the conventional plant density of 53000 plants ha-1 was 13.3% (a range of 8 to 17%), an average increase of 0.32 t ha-1. Even under such favourable conditions, plant densities above 83000 plans ha-1 decreased yields in almost all cases. In very dry seasons higher plant populations decreased yields. During the 1990 cropping season at Awassa yield was significantly decreased from a maximum of 1.93 t ha-1 at about 53000 plants ha-1 to 1.61 t ha-1 at about 80000 plats ha-1, a 16.6% decrease. At Yabello during the same season (not shown) this decrease was from 2.48 t ha-1 to 2.08 t ha-1, a 16% decrease which was statistically significant.

Although total biomass always increased with plant population (analysis not shown), grain yield did so only up to about 83000 plants ha-1.

Effects of water conservation methods. The effect of water conservation methods was significant (P<0 .05) in 13 of the 16 trials. In the combined analysis, however, these effects were significant (P< 0.05) in only two years out of 4, mainly because the location-by-water conservation methods interaction was highly significant. The efficiency of the water conservation methods was pronounced most in dry years such as at Awassa in 1990, where the advantage of mulching, bunded (closed) furrows, and open furrows over flat without mulching was 266, 63, and 48%, respectively. The average effects of the water conservation methods over the four growing seasons is depicted in Figure 3. At Yabello there was almost no yield advantage from the three water conservation methods. The highest gain was obtained in the potentially high-yielding site of Awassa, where mulching, bunded furrows and open furrows gave 52.3, 30.8 and 28.5% more yield relative to flat planting. The average gain relative to flat planting over the five locations during the four seasons from mulching, bunded furrows and open furrows were 23.5, 12.9, and 11.4%, with respective average yields of 3.2, 2.9 and 2.8 t ha-1. Mulching material could be applied abundantly only at the main station of Awassa and the sub-station Zuwai where sedge could be cut for mulching from the nearby Zuwai lake. At the other three sites, especially at Yabello, mulching material was very scarce. The effect of mulching was, therefore, most obvious at Awassa and Zuwai. The necessity of applying high doses of mulching material (3-4 t ha-1) for a significant increase in maize grain yields has been emphasised by Rodriguez (1987).

During years when heavy rains fall around planting, stand establishment might be poor when planting is in the furrows. In our experiments the final stands at harvest under the water conservation treatments were reduced in 8 of the 16 experiments. The mean reduction in plant stand was 3, 6.4, and 7.2% under mulching, bunded furrows and open furrows, respectively. Grain yields, however, were less than those under flat planting only for two trials, Yabello 89 (not presented) and Bidre 90. Under all other cases the water conservation treatments gave higher yields than flat planting despite reductions in plant stands.

One of the advantages of the water conservation methods during years when there is no water logging during planting is that more plants survive the drought stress and reach harvest. In our trials 7.8, 18.2 and 15% more plants survived until harvest under flat with mulching, bunded furrows, and open furrows, respectively, as compared with flat without mulching. The plants on flat plots without mulch wilted before those under water conservation methods and were stunted in growth whereas plants on plots with the water conservation methods were vigorous and maintained turgor longer than those on flat without mulching. This was very apparent during many growing seasons and hence the yield advantages mentioned above were obtained.

DISCUSSION

In the moisture stress areas of Ethiopia, some years are so favourable that even these early varieties can give grain yields as high as 7-8 t ha-1 while other years are so dry that yields are lower than 1 t ha-1 or total crop failure occurs. In some years early onset of rains might tempt farmers to plant even mid-late (150-day) maize varieties, but there is no possibility of predicting the expected type of season in any one year. The agronomic practices recommended to maximise yield under the most favourable years might cause yield losses during very dry years and vice versa. In our experiments we obtained contradictory results with planting densities, and, to a lesser extent, with water conservation methods. Many authors have reported similar results.

In very dry years, under the moisture-stress conditions of Botswana, Rowland and Whiteman (1993) recommended a plant density as low as 10-15 thousand plants ha-1, while Tilahun (1992) reported that a plant density as high as 133000 plants ha-1 maximises yield under favourable conditions of Ethiopia. Tilahun (1995) investigated early and mid-late maturing maize varieties and their mixtures at Awassa and Zuwai during the 1991-92 cropping seasons. He obtained results similar to ours, where higher plant populations maximised yields under favourable conditions and lower plant populations maximised yields during drier seasons.

Due to poor stand establishment and factors such as low soil moisture at planting and surface capping, it is common for farmers in the moisture stress zones of the country to plant 2-3 times (40 to 80 kg ha-1) the normally recommended rate of maize (25-30 kg ha-1) (Kidane, 1992) . Some research stations resort to drilling maize seeds as in small grain cereals and then thin to the required plant densities after emergence. Peasant farmers in these zones follow the same principle and do the thinning using the same implement used for ploughing, an operation locally known as ‘Shilshalo’ (Rezene et al., 1992). In the Ethiopian rift valley one can usually observe peasants leaving a very high plant stand until harvest. Nadar (1984) has shown that if in dry years plant densities are reduced to 20000 plants ha-1, an increase as high as 30% (450 kg ha-1) was possible as compared to the conventional 70000 plants ha-1, but if this reduction was done on the wrong assessment of the season (i.e., the rain was more favourable than anticipated) a loss of up to 4500 kg ha-1 was possible.

Our results have indicated that simple in-situ water conservation methods, especially mulching, can sometimes give 2-3 fold yields as compared to flat plating. Many authors have reported similar results. A mulch of 3 t ha-1 tef straw gave 3.6 t ha-1 while the control gave 2.5 t ha-1 and 13.7 t ha-1 (3 cm depth) of scoria (red ash) gave 4.8 t ha-1 (Kidane, 1992). Som et al. (1987), quoted by Kumar (1993), reported grain yields of 2.0, 2.6 and 3.1 t ha -1 for flat, open furrows and bunded furrows, respectively. At an annual rainfall of 1050 mm. This gave, a yield advantage of 30 and 51% for open and closed furrows, respectively. During a three year study at Kobbo and Malkasa in eastern Ethiopia, Kidane (1992) reported an average increase of 45 and 64% from open and bunded furrows, respectively, as compared to flat planting at Kobbo and 73 and 128% more yield for the two water conservation methods, respectively, at Malkasa. These water conservation systems are superior to flat planting in many of the cropping seasons. In our experiments they were superior in 11 of the 16 trials. It is only in years with water logging that the ridge and furrow systems might give less yield than flat planting. Some researchers, therefore, recommend planting either on the ridges or on the side of the ridges (Macartney et al., 1971; Kidane, 1992; Rowland and Whiteman, 1993). This, they suggest, reduces surface capping as compared to flat planting or planting in the bottom of the furrows and avoids suffocation of seeds by water logging. Kidane (1992), however, has indicated that planting in the furrows gave higher yields than planting on the ridges.

Despite competition for the stalks as construction material and livestock feed and fuel, farmers in the Omo valley leave harvested stalks of sorghum and maize in the fields. The stalks and constructions similar to bunded furrows (soil bunded in rectangular structures of about 3 x 2 m, and about 10 cm high), are used on many hand-cultivated fields. Hand cultivation is used despite the availability of animal draft, for the mere reason of not destroying these structures.

CONCLUSION

The three varieties tested in our experiments were not superior to the check, Katumani; not even the top-cross Mirtchaye. Half-sib family selection was continued in Birkata and a variety was released from the recombined elite half-sibs in 1996. This variety was named ‘Tesfa’. The inferiority of our varieties to Katumani points to the necessity of modifying the selection strategies and widening the genetic base.

Maize populations as high as 80000 plants ha-1 gave the best yields in favourable years, whereas increasing populations beyond 50000 plants ha-1 decreased grain yields in dry years. Best use of these results could be made if the season could be predicted with greater accuracy. Persuad et al. (1987), analysing long term rainfall data of many stations in Niger, suggested that the midpoint of the season can be used as a good predictor. The mid season is expressed in number of weeks after January 1 up to which the cumulative rainfall constitutes half of the seasonal rainfall. For three stations they found a midpoint of 30.3, 30.4, and 30.4 weeks with standard deviations of only 1.3, 1.3 and 1.5 weeks. They concluded that at any date within the interval of one standard deviation from the mean, there is a 60% confidence that the rainfall for the remainder of the season will be the same as that obtained prior to this date. They suggested that this would help make calculated decisions on management options for the remaining period to avert the risk of crop failure. Such studies should be conducted for the moisture stress areas of Ethiopia to develop predictive capability. The "response farming" concept proposed by Stewart and Faught (1984) and its use in parts of Eastern Kenya for maize and beans is based on this same principle. According to this approach, early onset of rainfall justifies higher seed rates and the application of initial fertiliser. Late onset of rains is an indication of poor rainfall expectation, and seeding rate is reduced and no fertiliser applied. Rigorous analysis of rains 50 and 75 days after onset of rains would enable one to make the correct decisions such as adjusting plant stands and fertiliser side dressing in the remaining part of the season.

Despite the significant grain yield advantage of mulching, there is competition for the harvested stalk. It is used as feed for cattle, as firewood and as construction material. Other sources of mulching should be sought. Agroforestry might provide fast-growing tree or shrub species whose branches could be cut and used for mulching.

Although the ridge-furrow water conservation methods are superior to flat planting, they require extra labour for construction of the structures. Ridge-makers that are affordable to peasant farmers should be made available to exploit the expected yield increase from these methods.

Acknowledgement

This research was supported by the Centre for International Environment and Development Studies (NORAGRIC). We are grateful for this financial support. We thank the Norwegian Church Aid (NCA) for the cooperation in conducting the trials at Bidre. We are also indebted to the reviewers for their valuable suggestions.

References

1. Alem, G. 1987. Soil and water management research in the semi-arid regions of Ethiopia. In: Food Grain Production in Semi-Arid Africa. Menyonga, J.M., Taye, B. and Youdeowei, A. (Eds.), pp. 465-474. OAU/STRC-SAFGRAD, Ouagadougou.
2. Asfaw, N., Yeshi, C., Kassahun, S. and Aleligne, K. 1992. Importance, production practices, constraints and research needs of maize production under smallholder farmers in Ethiopia: Review of farming system research. In: Proceedings of the First National Maize Workshop of Ethiopia. Benti, T. and Ransom, J.K. (Eds.), pp. 105-112. IAR/CIMMYT, Addis Ababa.
3. Box, G.E.P. and Cox, D.R. 1964. An analysis of transformation. Journal of the Royal Statistical Society B 26: 211-243.
4. Chapman, S.C. and Barreto, H.J. 1996. Using simulation models and spatial databases to improve the efficiency of plant breeding programs. In: Plant Adaptation and Crop Improvement. Cooper, M. and Hammer, G.L. (Eds.), pp. 563-587. CAB International, IRRI, ICRISAT.
5. Dowker, B.D. 1963. Rainfall reliability and maize yields in Machakos district. East African Agricultural and Forestry Journal 28:134-138.
6. Edmeades, G.O., Fischer, K.S. and Islam, T.M.T. 1987. Improvement of maize yield under drought sterss. In: Food Grain Production in Semi-Arid Africa. Menyonga, J.M., Taye, B. and Youdeowei, A. (Eds.), pp. 191-210. OAU/STRC-SAFGRAD, Ouagadougou.
7. EMPDA 1984. Atlas for Secondary Schools of Ethiopia. Macmillan Publishers Ltd., Addis Ababa.
8. Haile, G. 1988. Maize Improvement Program for Low- and Mid-Altitude Moisture Stress Zones. Awassa College of Agriculture, Awassa, Ethiopia.
9. Jones, O.R. and Popham, T.W. 1997. Cropping and tillage systems for dryland grain production in the Southern High Plains. Agronomy Journal 89:222-232.
10. Kidane, G. 1992. Agronomic practices sustaining crop production in semi-arid areas of Ethiopia. In: Problems and Prospects of Dryland Farming in Southern Ethiopia. Norwegian Church Aid (NCA) and Awassa College of Agriculture (ACA) (Eds.), pp. 1-12. NCA-ACA, Awassa.
11. Kumar, V. 1993. Crop production in the West African drylands. In: Dryland Farming in Africa. Rowland, J.R. (Ed.), pp. 109-141. CTA, Macmillan. London.
12. Kumar, V., Rowland, J.R. and Smith, P. 1993. The dryland farming environment. In: Dryland farming in Africa. Rowland, J.R. (Ed.), pp. 19-45. CTA, Macmillan, London
13. Lemma, G., Alemu, T. and Abubaker, H. 1992. Maize utilization in Ethiopia. In: Proceedings of the First National Maize Workshop of Ethiopia. Benti, T. and Ransom, J.K. (Eds.), pp. 154-159. IAR/CIMMYT, Addis Ababa.
14. Macartney, J.C., Northwood, P.J., Dagg, M. and Dawson, R. 1971. The effect of different cultivation techniques on soil moisture conservation and the establishment and yield of maize at Kongwa, Central Tanzania. Tropical Agriculture (Trinidad) 43:9-23.
15. Mando, A. and Stroosnijder, L. 1999. The biological and physical role of mulch in the rehabilitation of crusted soil in the Sahel. Soil Use and Management 15:123-127.
16. Mengistu, N. and Ermias, B. 1988. Towards a comprehensive agro-ecological zonation of Ethiopia. In: IAR Proceedings of the 20th NCIC. Mengistu, N. and Ermias, B. (Eds.), pp. 150-157. IAR, Addis Ababa.
17. Montgomery, D.C. 1984. Design and Analysis of Experiments, Second Edition. John Wiley & Sons, New York.
18. Nadar, H.M. 1984. Maize yield response to row spacing and population densities under different environmental conditions. East African Agricultural and Forestry Journal 44:157-165.
19. Ohiagu, C.E. 1987. Storage of food grains in the Savanna zones of Northern Nigeria. In: Food Grain Production in Semi-Arid Africa. Menyonga, J.M., Taye, B. and Youdeowei, A. (Eds.), pp. 361-370. OAU/STRC-SAFGRAD, Ouagadougou.
20. Persaud, N., Ouattara, M. and Alari, I. 1987. Analysis of rainfall records and its implications for improving rain-use efficiency for cereal production in Niger. In: Food Grain Production in Semi-Arid Africa. Menyonga, J.M., Taye, B. and Youdeowei, A. (Eds.), pp. 371-380.OAU/STRC-SAFGRAD, Ouaga-dougou.
21. Reij, C., Mulder, P. and Begemann, L. 1988. Water Harvesting for Plant Production. World Bank Technical Paper No. 91, The World Bank, Washington, D.C.
22. Rezene, F., Woldeyesus, S., Aliye, H. and Asfaw, N. 1992. Weed control research on maize in Ethiopia: A review. In: Proceedings of the First National Maize Workshop of Ethiopia. Benti, T. and Ransom, J.K. (Eds.), pp. 62-73. IAR/CIMMYT, Addis Ababa.
23. Rodriguez, M.S. 1987. Agronomic practices for reducing drought stress and improving maize grain yield in the semi-arid tropics of West Africa. In: Food Grain Production in Semi-Arid Africa. Menyonga, J.M., Taye, B. and Youdeowei, A. (Eds.), pp. 493-510. OAU/STRC-SAFGRAD, Ouagadougou.
24. Rowland, J. and Whiteman, P. 1993. Principles of dryland farming. In: Dryland Farming in Africa. Rowland, J.R. (Ed.), pp. 68-94. CTA, Macmillan, London.
25. Sheleme, B. 1991. Characterization of land qualities for farming systems. In: Annual Research Report of the Awassa College of Agriculture. Awassa College of Agriculture, Awassa, Ethiopia.
26. Sheleme, B. 1994. Characterization of land qualities for farming systems. In: Farming Systems in Southern Ethiopia. Awassa College of Agriculture/NORAGRIC, Awassa.
27. Shock, C.C., Hobson, J.H., Seddigh, M., Shock, B.M., Stieber, T.D. and Saunders, L.D. 1997. Mechanical straw mulching of irrigation furrows: Soil erosion and nutrient losses. Agronomy Journal 89:887-893
28. Stewart, J. I. and Faught, W. A. 1984. Response farming of maize and beans at Katumani, Machakos District, Kenya: Recomme-ndations, yield expectations and economic benefits. East African Agricultural and Forestry Journal 44:29-51.
29. Thomas, A.L. and Kaspar, T.C. 1995. Maize nodal root response to soil ridging under three tillage systems. Agronomy Journal 87:853-858.
30. Tilahun, A. 1992. Quantitative and physiological traits in maize (Zea mays L.) associated with different levels of moisture, plant density and leaf defoliation in Ethiopia. In: Proceedings of the First National Maize Workshop of Ethiopia. Benti, T. and Ransom, J.K. (Eds.), pp. 74-80. IAR/CIMMYT, Addis Ababa.
31. Tilahun, A. 995. Yield gain and risk minimization in maize (Zea mays) through cultivar mixtures in semi-arid zones of the rift valley in Ethiopia. Experimental Agriculture 31:161-168.
32. Tollenaar, M., Dibo, A.A., Aguilera, A., Weise, S.F. and Swanton, C.J. 1994. Effect of crop density on weed interference in maize. Agronomy Journal 86:591-595.

The following images related to this document are available:

Line drawing images