The experimental soil was sandy loam comprising of 60% sand, 29.5% silt and 10.5% clay in block C of the model watershed at Hamelmalo Agricultural College (15˚52'20.6"N and 38˚27'57.6"E at 1280 msl), in the semiarid region of Eritrea. Annual rainfall in the past seven years ranged from 370 - 663.1 mm with a mean of 488 mm and average annual pan evaporation of 1931 mm. Highest mean monthly temperature occurred in May (35.7˚C) and lowest in January (11.1˚C). Total rainfall was 388 mm in 2013 and 429 mm in 2014 (Figure 1).

The soil was non-saline (EC 0.18 dS∙m⁻¹) with pH of 8.2 and average bulk density of 1.5 Mg∙m⁻³ (Table 1). Field capacity and wilting points of soil were 0.175 and 0.053 m³∙m⁻³, respectively. The organic matter content was 0.27% and available N was 0.05%. The available P and K were 0.88 mg∙kg⁻¹ 0.25 cmol∙kg⁻¹, respectively.

The experiment was conducted in split plot design with three tillage treatments viz., conventional tillage (CT), reduced tillage (RT) and zero tillage (ZT) in main plots and four irrigation treatments viz., rainfed (I₀), 50% of full irrigation (I₁), 75% of full irrigation (I₂) and full irrigation (I₃) in subplots in three replications. CT refers to three passes of traditional bullock-drawn plough followed by row planting and RT refers to one pass of bullock-drawn plough 4 days after heavy rainfall followed by row planting. Zero tillage (ZT) was direct planting in rows. Full irrigation was 50% depletion of available soil water in 1 m profile. All irrigations were stopped 15 days before sorghum maturity. Each subplot was 4.0 m × 4.5 m, separated by 2 m passage. Bunds of width 0.4 m and height 0.3 m were formed around each plot to avoid any runoff or run-on.

Sorghum variety ICSV 210 (BUSHUKA) and pigeonpea variety ICEAP 00040 were planted at a seed rate of 12 and 10 kg∙ha⁻¹, respectively, on July 7 in 2013 and July 14 in 2014 in alternate rows, 0.375 m apart. The

sorghum was planted at a distance of 0.2 m and pigeonpea at 0.4 m within rows. Fertilizers applied were DAP at 100 kg∙ha⁻¹ before sowing and urea at 50 kg∙ha⁻¹ at 25 and 45 days from planting recommended for sorghum by National Agricultural Research Institute, Halhale, Eritrea. Hand weeding was done before topdressing urea. Soil moisture was determined gravimetrically by sampling 0.05 m length and diameter soil core from the midpoint of 0.25 m depth increment down to 1 m at sowing and 20 days interval until first week of September and at 10 days interval thereafter to determine irrigation date. Irrigation day was at 50% depletion of available soil moisture in 1 m profile. Net irrigation was 60 mm in I₃ (full irrigation), 45 mm in I₂ (75% of full irrigation) and 30 mm in I₁ (50% of full irrigation) applied on September 11 and 21 and October 2 in 2013 and September 24 and October 4 in 2014. Hand weeding was done twice and CARBARAYL insecticide was applied two times @ 2 g per litre of water to control pod borer. Grain and stover yields were determined by harvesting central 3 m × 3 m area and threshing manually. Pigeonpea was harvested when over 80% of the pods become brown. The grain was dried to 14% moisture.

Crop water use (ET, mm) was determined using water balance equation as

where RF is rainfall, mm, SI is supplemental irrigation, mm, ΔS is change in soil moisture storage, mm, DP is deep percolation, mm, and RO is runoff, mm. Both DP and RO were zero because all plots were well bunded and storm rainfall never exceeded available water storage capacity of the root zone.

Root length density (RLD) was determined at harvesting of the two crops by line intersection method of Tennant [43] . Root samples were drawn from each treatment in 0.15 m soil depth increments down to 1.2 m by placing 0.1 m diameter root sampler on the harvested hill. Root samples were collected in plastic bags for saturation overnight followed by washing in soil-root wash basin. Soil-root mixture in the wash basin was stirred to disperse roots and water was supplied continuously to allow suspended roots to pass through the drain pipe into the sieves arranged in the order of 2, 0.650, and 0.355 mm. Roots were randomly spread by tweezers in a dish containg a film of water and number of roots with vertical and horizontal grid lines of 10 mm were counted and RLD (cm∙cm⁻³) was calculated as

where R is root length, cm, expressed as

where N is number of intersections and V is soil core volume, cm³. Percent root distribution was calculated as

The sorghum equivalent grain yield (SEY) was calculated to express total yield in terms of sorghum for sorghum-pigeonpea intercrop. The SEY was based on per kg market price of the two crops using the relation:

where Ys is sorghum yield (kg∙ha⁻¹), MRp is market rate of pigeonpea (Eritrean NKF kg⁻¹), MRs is market rate of sorghum (Eritrean NKF kg⁻¹), and Yp is pigeonpea yield (kg∙ha⁻¹). Prevailing market price of the crops kg⁻¹ was collected from the open market in ERN (Eritrean Nakfa). Pigeonpea is not common in Eritrea but it is one of the costliest pulses in the international market and, therefore, Its market price was considered twice that of the sorghum.

Sorghum vegetative growth was relatively better in CT and RT plots than in ZT, perhaps because ZT was more affected by weeds in the initial establishment period. Irrigations were applied from reproductive phase. Sorghum growth was faster than pigeonpea during the initial 85 days from sowing although both were planted on the same date (Figure 2). Pigeonpea grew faster only after sorghum harvesting in 115 - 120 days and thus did not compete with sorghum crop in the initial stages. At sorghum harvesting, only 3% of pigeonpea plants were approaching flowering. Pigeonpea is known to lack synchronous flowering and maturity. However, variety planted in the experiment appears to be wilder type because flowering and maturity continued for several months after sorghum harvesting. Pigeonpea plants survived green even beyond March but net plot was harvested by end of January. Shorter duration improved pigeonpea varieties are now available and would be better option.

About 80% sorghum roots in sorghum-pigeonpea intercrop were within 0.6 m soil profile, of which more than 60% were in the top 0 - 0.20 m layer (Figure 3). On the contrary, more than 75% pigeonpea roots were below 0.60 m depth in the soil. Rooting patterns of sorghum-pigeonpea intercrop thus have a weak competition between them for water and nutrient extraction zone in the soil profile.

Mean sorghum yields due to tillage were not significant in both the years (Table 2). But mean yields due to supplementary irrigations were significantly greater in I₂ than in I₁ both in 2013 and 2014. Yields were at par in

I₂ and I₃. Interaction of tillage and irrigations showed that grain yields in RT + I₂ and RT + I₃ were at par in both the years and significantly greater than in RT + I₁. However, despite more and better distributed rainfall in 2014 yields were lower than in 2013 except in I₀. Decreasing trends in yield indicate growing deficiency of nutrients other than applied N and P. Potassium deficiency symptoms were common.

Mean stover yields were independent of tillage but were significantly higher in I₂ and I₃ than in I₁ (Table 3). Interaction effects showed significantly higher yields in RT + I₃ in 2013 and in RT + I₂ in 2014. Halving et al. [20] also observed greater stover production in sorghum due to improved nutrient uptake.

Two-year average grain and stalk yields of pigeonpea are shown in Table 4. Mean grain yields of pigeonpea due to tillage were significantly higher in CT (1098 kg∙ha⁻¹) and that due to irrigations were higher in I₃ (1271 kg∙ha⁻¹). Increases in yields were 215% in I₁ from I₀ and 153% in I₂ from I₁. Interaction effects showed that pigeonpea yields increased significantly with irrigations for sorghum in ZT and RT. Yields were at par in CT + I₂ and RT + I₃. Since all inputs such as tillage, fertilizers and irrigations were applied based on requirements for sorghum, possibility of raising pigeonpea on residual soil moisture in sorghum-pigeonpea intercrop are promising.

Mean stalk yields of pigeonpea due to tillage were significantly higher under CT than ZT and RT (Table 4) and that due to irrigations were significantly higher under I₂. Interaction effects showed significantly higher stalk yield in RT + I₃. Since sorghum is main crop and water is a most limiting factor of crop production in Eritrea, RT + I₂ would be better choice.

Mean sorghum equivalent grain yields due to tillage were at par in RT and CT but significantly greater than in ZT (Table 5). However, mean sorghum equivalent yields (SEY) due to irrigations increased significantly from I₀ to I₃. Interaction effects showed that yields were at par in RT + I₃, CT + I₂ and CT + I₃. Since sorghum was the main crop for which RT was most optimum, the choice would go for RT + I₂, which was the next best tillage + irrigation combination. Choice for I₃ should depend on availability of irrigation resources with the farmer.

Mean residual soil moisture at sorghum harvesting increased from 60 mm∙m⁻¹ under rainfed to 80 mm∙m⁻¹ under 75% of full irrigation (Table 6). Results thus show that supplementary irrigations to optimize sorghum yields leave considerable quantity of residual moisture, which can be used by the pigeonpea crop. About 85% - 90% of residual moisture was consumed by pigeonpea.

Water use by sorghum-pigeonpea intercrop was almost independent of tillage but increased with irrigations (Table 7). Highest water use was recorded in RT + I₃ (522 mm) and lowest in ZT + I₀ (276 mm). Average water use by sorghum-pigeonpea was lowest (366 mm) under RT and highest (440 mm) under CT.

Production function showed that sorghum equivalent yield increased with water use rapidly to 9363 kg∙ha⁻¹ for which water use was 478 mm (Figure 4). At harvesting of pigeonpea mean water use increased to 438 mm from 374 mm at sorghum harvesting. As shown in Figure 5, sorghum yield of 6600 kg∙ha⁻¹ and pigeonpea yield of 1422 kg∙ha⁻¹ could be obtained by water use of 478 mm. The water use efficiency was maximum (19.6 kg∙ha⁻¹∙mm⁻¹) in CT+I₂ and minimum (6.8 kg∙ha⁻¹∙mm⁻¹) in ZT + I₀.

Benefit-cost ratio (B:C) of sorghum and sorghum equivalent yields (SEY) show that sorghum-pigeonpea intercrop was beneficial at all tillage and irrigation levels (Figure 5). Maximum benefit was 7.1 from SEY as against 5.2 from sorghum alone in RTI3 followed by that in RTI2 and CTI2. Residual soil moisture after sorghum harvesting was more important for pigeonpea than tillage. Benefit was doubled by even 50% of full irrigation and was 4 times in RTI2 (75% of full irrigation) compared to ZTI0.