WATER PRODUCTIVITY ANALYSIS FOR SMALLHOLDER RAINFED SYSTEMS: A CASE STUDY OF MAKANYA CATCHMENT, TANZANIA J. MUTIRO1*, H. MAKURIRA 2, A. SENZANJE 3, M.L. MUL2,4 1 WREM programme, Dept of Civil Engineering, University of Zimbabwe. P.O. Box MP 167, Mt Pleasant, Harare, Zimbabwe 2 Dept of Civil Engineering, University of Zimbabwe. P.O. Box MP 167, Mt Pleasant, Harare, Zimbabwe 3 Dept of Soil Science and Agricultural Engineering, University of Zimbabwe, P.O. Box MP 167, Mt Pleasant, Harare, Zimbabwe 4 UNESCO-IHE, Institute for Water Education, PO Box 3015, 2601 DA Delft, the Netherlands ABSTRACT Increasing food insecurity as a result of ever-increasing population, less water availability and soil degradation is common in countries in sub-Saharan Africa. While most of the developed fresh water resources are heavily committed to irrigation, about 90% of subSaharan populations rely solely on rain fed agriculture for their livelihoods (Rockstrom. et. al., 2003). The majority of the population is therefore not directly benefiting from developed water resources but are, in fact, subsistence rain fed farmers. Thus, in subSaharan Africa, techniques which help to improve water productivity (WP) can assist in alleviating the impacts of water scarcity especially for crop production purposes. The Makanya catchment in northern Tanzania is one typical semi arid area where farmers depend on rain fed subsistence farming for their livelihoods. The area receives an average annual rainfall of about 500 mm/yr over the two rainfall seasons they experience in a year. An assessment of the current WP in rain fed and partially supplementary irrigated agriculture was made in Makanya catchment. The study assessed the effect on WP (of maize) and infiltration capacity of several rainwater harvesting techniques. The crop water requirement for maize in the study area was found to be 508 mm/season by using the CROPWAT model compared to total received rainfall of between 276.06 mm and 383.86 mm during the same period. An attempt was made to separate transpiration from evapotranspiration using a transpiration meter. Results indicate that currently WP for maize in the catchment is low (0.1 to 0.6 kgm-3). Introduction of improved techniques increased WP by between 90% and 110%. Infiltration rates also increased from 6 cm/hr to 26 cm/hr. The conclusion from the research is that there is room to significantly improve the water use techniques being applied for crop productivity in Makanya catchment. A clear understanding and quantification of the water partitioning processes is required to maximise water used for transpiration by the plant as transpiration is directly related to biomass production. However, scientific approaches need to be complemented by efficient farm management practices for any technologies to be successful. Key words: supplementary irrigation, water productivity, smallholder systems innovations, deep ripping, rain fed. * Corresponding author: mutirojt@yahoo.co.uk 1 INTRODUCTION Increasing food insecurity as a result of ever-increasing population, less water availability, low crop yields and soil degradation is common in countries in sub-Saharan Africa. Whilst most of the developed fresh water resources are heavily committed to irrigation, about 90% of sub-Saharan populations rely solely on rain fed agriculture for their livelihoods (Rockstrom et. al., 2003). In sub-Saharan Africa, techniques which help to improve water productivity (WP) can assist in alleviating the impacts of water scarcity for crop production purposes. There is presently an increase in water scarcity due to increased competition for the same water from non-agricultural sectors and, also from deteriorating standards of usable water due to pollution. Improved water productivity, which is producing more crops per drop (van Dam. et. al, 2003), can be achieved through the introduction of improved soil, water and crop management. Carrying out a water balance analysis reveals the gaps in knowledge regarding best ways to increase water productivity. Most of these gaps relate to the inability to fully quantify all flow components, their interactions with crops, agricultural inputs and the environment in the process of producing substantive yields. According to the Tanzania Agricultural and Livestock Policy of 1997, one of the specific objectives is ‘to promote integrated and sustainable use and management of natural resources such as land, water, soil and vegetation’ (United Republic of Tanzania, 1997b). This can only be met by targeting rain fed agriculture, which accounts for about 63% of the agricultural Gross Domestic Product (GDP) (World Bank, 1994). Low and erratic rainfall have contributed to the low yields that threaten the contribution of rain fed agriculture to the GDP. Therefore, water saving technologies such as conservation tillage practices is required. Farmers have to be encouraged to rely as much as possible on rainwater and in rain fed systems, to introduce supplementary irrigation as per the crop needs and rainfall pattern. This seems practical in the semi arid areas of Tanzania such as Makanya catchment, which receives rainfall below 600 mm annually (Slaymaker et. al., 2002). However, relying solely on rainfall for crop growth in Makanya does not achieve the best results as witnessed by very little to zero yields currently achieved hence the need to introduce techniques that improve water productivity. Most catchments in sub-Saharan Africa are ungauged hence it is difficult to establish water availability in the systems. Low rainfall usually results in very low crop yields annually, for example, maize yields are around 1 ton/ha (Rockstrom, 2001) compared to the expected 5 tons/ha. The water productivity under the current management level is not known hence the need to evaluate the current water productivity levels being achieved. The potential of water system innovations to improve livelihoods while also improving water productivity remain largely unquantified in many arid to semi arid places. Quantifying field water productivity helps to reveal gaps in knowledge regarding the best ways to increase water productivity. This paper reports on work that was done whose main objective was to assess the crop and water productivity of small scale farming systems in semi-arid Makanya catchment in Tanzania. 2 The specific objectives were to: (i) (ii) Assess how much water is received at farm level. Assess crop yield and water productivity levels being achieved by small holder farmers using current water use technologies (iii) Analyse the effect of new techniques on crop and water productivity. (iv) Quantify transpiration as a proportion of total water available at field scale. The work was premised on the hypotheses that: (i) Water supplied to the field through rainfall and supplementary irrigation is insufficient for effective yields to be achieved under current water management practices. (ii) More efficient smallholder systems innovations including on-farm soil and moisture conservation can improve crop water productivity and increase crop yields in water scarce areas. METHODOLOGY Study Area: Makanya Catchment, Tanzania. The Makanya catchment has been selected as a case study and is located in Pangani river basin in Tanzania. It lies between latitudes 4o15' to 4o21' S and longitudes 37o48' to 37o53' E. It lies at an altitude ranging from 500 to 2000 m above sea level. The catchment has low and unreliable rainfall ranging from 400-600 mm. The rainfall distribution in the catchment is bimodal with the rainfall occurring in November - January (locally called 'Vuli') and March May (locally called 'Masika'). Vuli rains are highly unreliable and the seasonal amount exceeded 70 percent of the time is only 178 mm. The mean minimum temperature ranges from 16o C to 18o C and mean maximum temperature ranges from 26o C to 32o C. The main crops cultivated are maize, lablab beans, vegetables and beans. The farmers get water for supplementary irrigation from the storage reservoir locally known as Manoo Ndiva with a capacity of 1650 m3. Experimental Treatments The field experiments were conducted with a total of four smallholder farmers namely Omari, Wilson, Walter and Elizabeth. Soil samples were collected from each experimental plot and sent to the laboratory for the analysis of the following soil properties: soil texture, soil bulk density, porosity, pH; cation exchange capacity (CEC), organic carbon (OC), mineral nitrogen (N), available phosphorus (P) and potash (K). Infiltration tests were done using the double ring infiltrometer between the rip lines and in the rip lines. A flow meter was installed to measure the amount of irrigation water entering each plot. Another flow metre was installed to measure water going out of each field as run-off. Rain gauges were installed to measure the amount of rainfall received in each plot except in Omari plot. There were two sources of water inflow, supplementary irrigation and rainfall. The soil management practices analysed were deep ripping, conventional tillage, and fanya juu. Maize (Zea mays) was the study crop as it is widely grown in the study area. The maize variety was TMV1 that takes about 110 to 115 days to mature. Ripping using a Magoe ripper and sowing of seeds in the Vuli season was done on 04/10/04 (dry planting) and the recommended inter row spacing was 75cm and the in row spacing was 30 cm. In all treatments, farmers applied manure only (before planting) since it was their usual practice. They do not apply inorganic fertilizers because they believe that it makes the soils barren such that if you stop using inorganic fertilizer afterwards the yield will be low. Again organic fertilizers are very expensive 3 Weeding was done using a hand hoe. Gap filling was done from day 10 –12 after planting. Supplementary irrigation was done on day 65. Dursban was used to control ball worm. Supplementary irrigation was applied in Elizabeth’s plot (once) and in Wilson’s homestead plot (twice) during the growing season. Tasseling commenced on 30/11/2004. In Elizabeth tasseling commenced on 16/12/2004. Harvesting in all the plots was done on 06/02/2005. At harvest a sample plot area of 30 m2 was selected and the grain yield as well as the number of plants in that sample plot area were measured. This was then converted to per hectare basis. In the Masika season the same treatments that were applied in the Vuli season were repeated. The only difference was that there was also run off diversion on the rain fed plots (Walter and Iddi). In Elizabeth plot planting was done on 20/03/05, Iddi plot – 08/03/05, Walter plot – 08/03/05, Wilson chini – 09/03/05 and in Wilson home plot planting was done on 06/03/05. Experimental design Below is the table (Table 1) that shows the different experimental treatments in each of the plots. Table 1: Treatments for the Vuli season PLOT TREATMENT Supplementary irrigation Ripping Fanya juu Elizabeth x x x Wilson Home x x x Wilson chini x x x Walter x x Omari x x x Transpiration Measurement Measurement of transpiration was done using the Sap Flow Meter that was installed in Elizabeth’s plot 64 days after planting to measure sap flow. Water lost as direct evaporation during this period was calculated by subtracting the transpiration measured from evapotranspiration measured over the same period. The water lost as deep percolation was also monitored using the TDR tubes installed in the experimental plots. The water applied to the field as run-on, rainfall and supplementary irrigation minus the water lost from the field as run-off and deep percolation is equal to the water evapotranspired. To calculate the WP following equation was used; Yield (kgha1 ) WP WaterEvapoTranspired (m 3 ha 1 ) Equation 1 The paired samples t-test was carried out using SPSS Statistical program for Windows to find whether there was any significant difference between the means of the different parameters among different treatments. RESULTS Soil physical and chemical analysis The results of soil chemical and physical properties before the commencement of the Vuli season are shown in Table 2 below. Most of the soils in Makanya catchment are alkaline in nature with pH ranging from 7.4 to 8.8. Total nitrogen ranged from 0.14 to 0.26 with the lowest being recorded in Wilson chini’s plot. The available phosphorus ranged from 12.8 mg/kg to 47.2 mg/kg with the highest being recorded in Elizabeth’s plot. Soils with available phosphorus greater than 4 6 mg/kg are considered to be good soils. Organic matter and available potassium ranged from 2.93% to 4.91% and 1.74 to 8.9 Cmols/kg, respectively. Results show that the manure applied is rich in available phosphorus, %N, K and organic matter. Porosity and bulk density ranged from 42 to 45% and 1.456 to 1.534 g/cm3, respectively. Soil depths were as follows; Elizabeth plot (50 cm), Wilson home (56 cm), Wilson chini (72 cm), Omari (130 cm), Walter (20 cm). Table 2: Soil analysis results. Farmer/plot Depth pH %N Available (cm) P- mg/kg Elizabeth 0-20 8.8 0.19 47.2 Manure 8.5 11.82 712.0 Wilson 0-20 8.2 0.14 30.4 Manure 8.7 12.73 524.0 Iddi 0-20 7.4 0.15 21.8 Manure 9.3 16.13 754.0 K Cmol/kg 6.46 26.58 4.1 28.68 4.12 36.47 %OC %OM 2.07 130.0 1.52 140.0 1.64 177.4 3.57 13.79 2.62 13.67 2.83 13.72 BD (g/cc) 1.534 % Porosity 42 1.456 45 1.504 43 Crop yield results: The tables below Tables 3, 4, 5, 6 and 7) show the maize grain yield from each experimental plot during the Vuli season. The grain yield results for the Masika season could not be collected because of time constraint. Table 3: Maize yield in Elizabeth’s plot (received only one water allocation) Number of Planting Sample plot plants in density 2 PLOT number & treatment area (m ) sample plot (plants/ha) A (Ripping & irrigation) 30 116 38667 B (Irrigation & no ripping) 30 108 36000 39667 C (No irrigation & no ripping) 30 119 Yield (t/ha) 1.67 0.80 1.00 Table 4: Maize yield in Wilson Homestead’s plots (received two water allocations) Sample plot No. of plants Plant density Yield PLOT area (m2) in sample plot (plants/ha) (t/ha) A (Irrigation & ripping) 30 75 25000 1.83 14460 B (with irrigation & no ripping) 69.85 101 0.21 44333 C (No irrigation & no ripping) 30 133 2.08 Table 5: Maize yield in Wilson Shamba chini plots (received no water allocation) Sample plot No. of plants in Plant density Yield PLOT area (m2) sample plot (plants/ha) (t/ha) 27667 A(Ripping & no irrigation) 30 83 0.67 34940 B (No irrigation & no ripping 66.4 232 0.15 C (No irrigation & no 30000 ripping) 30 90 0.75 Table 6: Maize yield in Omari plot - 04/02/05 (RECEIVED TWO ALLOCATIONS) Sample plot Number of plants Planting density Yield (plants/ha) PLOT number & treatment area (m2) in sample plot (t/ha) A (Ripping & irrigation) 30 80 26667 2.50 B (Irrigation & no ripping) 30 93 31000 2.42 5 C (No irrigation & no ripping) 30 75 25000 2.83 Table 7: Maize yield in Walter’s plot (rain fed only) - WALTER MJEMA - 07/02/05 Sample plot area No. of plants in Plant density PLOT (m2) sample plot (plants/ha) Yield (t/ha) 30000 A (ripping) 30 90 0.50 B (no ripping) 30 100 33333 0.25 The yields in the different treatments were comparable as shown in tables 3 to 7 above. In Elizabeth’s plot, ripping, fanya juu and one allocation of water produced the highest yield of 1.67 t/ha. Yield of plots that received supplementary irrigation coupled with ripping was higher than of supplementary irrigation with no ripping but statistically the increase was not significant (t = 1.930 and p = 0.193). The difference in yield between the treatment with no irrigation and no ripping and the treatment with irrigation and no ripping was not significant (p = 0.940). So basing on the statistical analysis the hypothesis that systems innovations (ripping) improve yields can be rejected. Water application and water productivity Each farmer only got one water allocation per season as supplementary irrigation except Omari plot and Wilson’s homestead plot, which received two allocations. In the Vuli season studied, out of 150 families that are supplied by Manoo Ndiva, only 86 farmers got an allocation. Table 8 below shows the amount of water received in each experimental plot throughout the growing season. The rainfalls received in the experimental plots were comparable. Walter plot received the highest rainfall and Wilson home plot received the lowest rainfall. Table 8: Water received as rainfall and supplementary irrigation in the four experimental plots for the period from planting to harvesting. water Total rainfall Water applied as Water out of the Total system (mm) applied (mm) recorded supplementary Plot name (mm) irrigation (mm) 0 318.45 Elizabeth (ripping+suppl irrig) 277.14 41.31 0 270.12 Wilson Chini 270.12 0 Walter 323.07 0 0 323.07 Table 9 below shows how water productivity was changing in response to different treatments. The water lost through evaporation and deep percolation (TDR tubes) could not be measured because of equipment failure so the water balance equation could not be closed. Instead the water productivity was then calculated using the total amount of water applied to the field. The field water productivity was calculated as follows: WP Yield (kgha1 ) TotalWaterApplied (m 3 ha 1 ) Table11: Yield and Water Productivity responses to different treatments Effective rainfall + Yield Water productivity Volume applied Plot name & treatment suppl. Irrigation (mm) (kg/ha) (WP) (kg/m3) (m3) 6 Elizabeth A 263.0 1665 0.63 2630.2 Elizabeth B 263.0 799 0.30 2630.2 Elizabeth C 221.7 1000 0.45 2217.1 Wilson Chini Ripping 216.1 667 0.31 2161.0 Wilson Chini No ripping 216.1 151 0.07 2161.0 Walter Ripping 258.5 500 0.19 2584.6 Walter No ripping 258.5 250 0.10 2584.6 NB- Elizabeth A –Fanya juu + irrigation + ripping, Elizabeth B –Funya juu + irrigation + no ripping and Elizabeth C – Fanya juu + no irrigation + no ripping. Crop water requirements and maize yields The crop water requirements (CWR), reference crop evapotranspiration and effective rainfall was calculated using the CropWat 4 Windows 4.3 version The crop water requirement for maize was far below the total rainfall received during the growing season (figure 2). The cumulative rainfall received (total 209 mm and effective rainfall 192.8 mm) in Elizabeth plot was far less than the crop water requirement for maize (508 mm) in the Vuli season, as shown also in figure 2. This shows that maize cannot produce significant yield if rain-fed only especially in the Vuli season. Cummulative rainfall & CWR for maize in Elizabeth plot CWR and rainfall distributionover the growing period 80.00 cum m ulative CW R & r ainfall ( m m ) 600 70.00 water (mm) 60.00 50.00 40.00 30.00 20.00 10.00 10 /6 / 10 200 4 /1 6/ 10 200 4 /2 6/ 2 11 004 /5 /2 0 11 /1 04 5/ 2 0 11 /2 04 5/ 20 0 12 /5 4 / 12 200 4 /1 5/ 12 200 4 /2 5/ 20 04 1/ 4/ 20 05 1/ 14 /2 00 1/ 24 5 /2 00 5 0.00 400 300 200 100 0 0 20 40 60 80 100 120 140 -100 rainfall CWR time (10 days sum) 500 CW R Time (days after planting) rainfall (a) (b) Figure 2: Comparison of crop water requirement (CWR) and rainfall (a) and cumulative CWR and rainfall received in Elizabeth plot (b). The rainfall recorded by rain gauges in the experimental plots and that, which was recorded by the automatic met station installed in Bangalala village during the Vuli season was compared. There was almost the same trend in rainfall events and total rainfall for the period under observation (see Figure 3). Automated met station data Bangalala sec school W alter plot Rainfall distribution 200.00 180.00 W ilson home plot 160.00 W ilson shamba chini plot 120.00 100.00 80.00 60.00 40.00 20.00 Ap ril (F irs t 2w ks ) Ma rch y Fe br ua ry Ja nu ar er er em b De c em b No v Oc tob er 0.00 Se p te mb er rainfall (mm) 140.00 month 7 Figure 3: Graphs showing rainfall recorded in the experimental plots (Walter, Bangalala secondary, Wilson home and Wilson shamba chini). Sap flow4 measurements The graphs below (figure 4) shows how the daily sap flow measured were behaving in Elizabeth’s plot. The sap flow rate, which equates to transpiration rate showed a great variation on different days and time of the day. The transpiration flux was varying from as low as 18 g/hr/ plant to as high as 139 g/hr/plant (1876.5 ml per day per plant) in Elizabeth’s plot and in Wilson’s plot it was varying from as low as 15 to 139 g/hr/plant. The sap content constitutes about 99% water so to find the volume of water transpired it was assumed that 100% of the sap content was water. This means 1 gram/hour sap flow equates to 1 ml/hour. Daily sap flow measurements in Elizabeth plot daily sap flow measurements in w ilson plot 120 160 sap flow (grams/hr) sap flow rate (grams/hr) 140 100 80 60 40 120 100 80 60 40 20 20 0 9:00 0 9:00 10:12 11:24 time of the day (hours) 12:36 13:48 sap flow sap flow sap flow sap flow 10:12 11:24 12:36 13:48 time of the day (hours) 15:00 on 21/12/04 on 26/12/04 on 27/12/04 on 12/1/05 sap flow on 22/12/04 sap flow on 23/12/05 sap flow on 1/1/05 sap flow on 31/12/04 sap flow on 9/1/05 (a) (b) Figure 4: Daily sap flow measurements in Wilson’s plot (a) and in Elizabeth’s plot (b). The graphs below (Figure 5) show the maize sap flow measurements done in Wilson Home experimental plot and in Elizabeth plot. sap flow and rainfall for Elizabeth plot 1400 70.00 12.00 1200 60.00 10.00 1000 8.00 800 6.00 600 4.00 400 2.00 200 0.00 0 1 6 11 16 21 Time (days) 26 31 900.00 800.00 rainfall (mm) 700.00 50.00 600.00 40.00 500.00 30.00 400.00 300.00 20.00 200.00 10.00 100.00 0.00 36 0.00 1 6 11 16 21 26 31 days rainfall sap flow sap flow (grams/day) 14.00 sap flow (grams/day) rainfall (mm) sap flow and rainfall data for Wilson plot 36 41 46 51 56 rainfall sap flow (a) (b) Figure 5: The average sap flow and rainfall recorded from day 62 to day 93 after planting in Wilson’s plot (a) and in Elizabeth’s plot (b). A total of 31.808 litres of water was transpired by each plant in Elizabeth’s plot from day 64 to day 123 after planting. In Wilson’s plot 32.797 litres of water was transpired by each maize plant 8 from day 62 to day 93 after planting (31 days). In Wilson’s plot there was a mechanical fault on the sap flow recorder on 12/01/05 such that it had to be dismantled before the maize crop was fully mature. In Elizabeth’s plot the plant population was 116 plants in a 30 m2 plot (Table 3) which gives a plant population of 38 667 plants per hectare. This means that the total water transpired per hectare (from 07/12/04 to 28/01/05) was 1229.93 m3. It was assumed that the transpiration period was from 0530hrs to 1900hrs because transpiration only occurs during the day when there is light and during the night plants hardly transpire. It was also assumed that the water consumed by the plant in the process of photosynthesis and accumulation in the plant cells to maintain the osmotic gradients was negligible because about 99% of the water taken up by plants is lost as water vapour (transpiration). The rainfall received in Elizabeth’s plot during sap flow measurement was 106 mm. The effective rainfall calculated using the Cropwat was 97.88 mm. The crop water requirement during the same period was 342.36 mm. Contribution of water from rainfall only was 978.8 m3/ha and the crop water requirement over the same period was 3423.6 m3/ha (for the maximum potential yield to be achieved). Water applied as supplementary irrigation amounted to 413.1 m3. In total, the water applied to the experimental plot (Elizabeth plot) was 1391.9 m3. Infiltration The infiltration results are shown in the graphs (Figure 6) below. The basic infiltration rates on the rip lines were greater than the infiltration rates between the rip lines that are on the ridge. For example in Elizabeth plot on the ridge it was 6 cm/hr whereas the rate on the rip line was around 26 cm/hr (figure 6 (a) and (b)). From the statistical analysis ripping significantly increased infiltration rate (p=0.007). Elizabeth plot - infiltration rates Infiltration rate on hand hoe plot Elizabeth plot infiltration-ridge 140 90 infiltration-rip infiltration rate (cm/hr) infiltration rate (cm/hr) 160 120 100 80 60 40 20 80 70 60 50 40 30 20 10 0 0 0 50 100 0 150 10 20 30 40 50 time (mins) time (mins) (a) (b) Figure 6: Infiltration rates for three different sites in Elizabeth plot on the ridge (between two rip lines) and on the rip line (a) and in hand-hoe ploughed plot (b). infiltration rate (cm/hr) Walter plot - infiltration rates 120 infiltrationridge 100 infiltration - rip 80 60 40 20 0 0 50 100 150 200 time (mins) Figure 7: Infiltration rates for Walter plot on the ridge (between two rip lines) and on rip line. 9 DISCUSSION Soil physical and chemical analysis Soils in the Makanya watershed are alkaline in nature with pH ranging from 7.9 to 8.8 because the mean annual rainfall is low. Also the use of manure with a high pH of around 8.5 contributes to the increase in the overall soil pH. High evaporation in the area is also contributing to accumulation of salts in the soils. High soil pH affects the availability of micronutrients needed by maize such as boron and zinc in the soil. In saline soils more water is applied to the field to balance the osmotic pressure between the soil and plant roots and to drain excess salts. Maize is relatively sensitive to salinity and does well in well-drained soils with a pH range of 5.57 (Landon, 1991) but can tolerate pH up to 8. Beans do well in pH ranges of 5.5-6.5 such that the above pHs recorded might be affecting bean crop yields. It might not be beneficial to the farmers to switch to the bean crop that is regarded as a high value crop because of the pH levels of most of the soils as its yield is likely to be very low. Soils in the study area are rich in the major nutrients required for crop growth such as N, P and K as well as the percentage organic matter. This can help explain that the addition of fertilizers might not significantly increase crop yield and water productivity in the area. The porosity of the soils in the study area is on the average to slightly above average side and from that it shows that the soils are medium textured (sandy loam) soils (Hillel, 1982). Porosity generally ranges from 30% to 60% with coarse-textured soils being less porous than fine textured soils. On bulk density, usually sandy soils have high bulk density of around 1.5-1.6 g/cc whereas in aggregated loams and in clay soils it can be as low as 1.1 g/cc. The bulk density is affected by soil structure and degree of compaction. Poor structure and high degree of compaction results in low infiltration and poor root penetration. Ripping destroys the plough pan that normally reduces infiltration rate. Low infiltration rate affect water productivity as it limits root growth and infiltration of water into the soil. Even though ripping increased infiltration, high concentration of salts in the soils makes it difficult for plant roots to absorb as much water as possible. Crop yields Deep ripping only and deep ripping coupled with supplementary irrigation increased yield. However, statistically, the increase was not significant because some parameters like rate of manure application to the field and the residual effect of manure from the previous seasons were not monitored in the experimental plots. Therefore though ripping increased infiltration capacity, this did not result in the increase in yield, which means that it is not only infiltration rates, which affect yields. Again presence of salts creates an osmotic gradient that make water absorption by roots difficult. Due to limited time and experimental plots there were very few replicates for the different treatments so the real effect of ripping on yield was not significant. Rate of manure application and other management practices such as weeding also have an effect on yield. In the experimental plots the farmers did not timely weed because of the labour problems. Infiltration rate increased as a result of ripping because ripping improves the soil structure and increase the rate of basic infiltration thereby reducing run-off. This will then increase soil water available to the crops. Ripping is considered an in-situ water harvesting technique. The manure being applied to the fields are rich in nutrients as shown in the analysis results. Therefore the combined use of deep ripping and manure contributed much to the increased yields though the increase was not significant. Ripping increased infiltration rate significantly. The rate of infiltration relative to the rate of water supply determines how much water enter the root zone and how much will run off (Hillel, 1982). The knowledge of infiltration process helps in the efficient 10 management of soil and water. Supplementary irrigation did not coincide with the period when the maize plant required a lot of water resulting in low yields being achieved. The period from flowering up to grain filling requires a lot of water. When the maize plant is stressed at this stage yields are greatly reduced as this period is very sensitive to water stress. However in Makanya especially during the dry spell the water is usually unavailable in the storage structures (Ndivas) to irrigate. This results in yield reduction. Stewart (1988) based on research in East Africa, indicated that severe yield reductions due to dry spells occur once or twice in five years. The crop water requirement was far below the total rainfall received during the growing season in Makanya catchment. That is the crop water requirement for the maize crop was 508 mm (from CropWat4 calculations) whilst the rainfall for the corresponding period was only 209 mm and of this, effective rainfall was 192.8 mm. This shows that in Makanya, maize cannot produce significant yield if rain-fed only. For maximum yields to be realized an additional 315 mm needs to be applied either as supplementary irrigation or in-situ rainwater harvesting as it makes more water available to the plant roots. Makanya catchment receives up to 600 mm annually but the major problem is the spatial and temporal distribution of the rainfall. The rainfall distribution in the catchment is bimodal such that almost half of the rainfall is received in one season whilst the other half is received in the other season. Therefore even though maize can do well in areas with seasonal rainfall of about 600 mm the rainfall received in one of the seasons cannot sustain good yields as it falls far below the 600 mm required by the maize crop. In Makanya catchment farmers receives supplementary irrigation before the dry spell or well after the dry spell so instead of bridging the dry spell period, the water is allocated too early or too late. For example Elizabeth received water on the 8 th of December 2004 just before the dry spell commenced. So during the dry spell period from day 67 up to day 88 after planting there was no water allocation to Elizabeth’s plot. Water productivity Water productivities in the experimental plots responded well to the introduction of deep ripping coupled with supplementary irrigation as ripping tends to increase infiltration rate hence making more water available to the plant roots. For example the plots that were only rain fed with ripping the water productivity was relatively low compared with the plots with ripping coupled with supplementary irrigation. Conversion of more evaporation into transpiration boost water productivity as water evaporated from the soil is considered as lost. Therefore ripping improves water productivity though statistically there was no significant difference. In areas that received partial supplementary irrigation, such as Elizabeth’s plot, the water productivity recorded was 0.6 kg/m3, which was relatively low, compared to international standards. For example water productivity for irrigated cereals (excluding rice) ranges from 1.0 to 1.7 kg/m3 in China, the United States and Brazil whilst in western European countries it ranges from 1.7 to 2.4 kg/m3 (Rosegrant et. al., 2002). However the water productivity results from this study compares well with reported results for sub-Saharan Africa, which ranges from 0.1 to 0.6 kg/m3 with an average of 0.3 kg/m3 for cereals (excluding rice) (Rosegrant et al., 2002). Sap flow4 measurements From the sap flow measurements results maize plants absorb more water after a rainfall event resulting in a high sap flow up the plant. As the soil dries up the average sap flow decrease as it becomes more difficult for the plant roots to absorb water. This usually results in reduction in yield. For example from 12 up to 30 December 2004 there was no rainfall yet it’s the critical period when the maize plant required a lot of water (that is during tussling and grain filling). This can be identified as the two weeks dry spell and it explains why the average yields from Wilson 11 homestead plot was only 1.833 t/ha instead of the potential 5 t/ha which can be achieved by communal farmers with TMV1 maize variety. Supplementary irrigation was supposed to bridge dry spell. High sap flow mostly occurred when the plot received water allocation on the same period and also when it rained. As expected when it rains more water is available in the root zone such that the plants easily absorb water through the roots resulting in high sap flow being realized. The sap flow varied with time due to variations of weather conditions. However a sudden change in one climatic factor changes the sap flow pattern. Transpiration rate reduces to almost zero when the plant dries up, as transpiration would have almost ceased. Nearly all water taken up by the plant is lost by transpiration and only a tiny fraction is used within the plant (Allen, et al, 1998). This explains why there was high sap flow especially after a rainfall event. Also the amount of water in the soil, the growth stage and the ability of the soil to conduct water to the roots determine the transpiration rate and the amount of nutrients taken up by the plants. Reduction in transpiration rate greatly affects yield. Water, which is the main component in the uptake of nutrients and in the formation of carbohydrates during photosynthesis, determines the crop yield apart from fertilizers and other factors so if transpiration rate is reduced then yields are greatly affected. Amount of water lost as evaporation from the experimental plot (Elizabeth) was around 11%, (assuming that all water added to the field as rainfall and supplementary irrigation was evapotranspired) which compares well with what is in the literature. The period measured coincides with the mid to late season stage when there is full crop cover. For example according to Allen, et. al, (1998) at sowing nearly 100% of evapotranspiration comes from evaporation while at full crop cover more than 90% of evapotranspiration comes from transpiration. The water lost through evaporation is relatively higher than the expected value of less than 10%. This can be explained by the low plant density of 38 667 plants per hectare. Although the plant population is within the standard range of 36 000 to 60 000 plants per hectare its on the lower side of the range. Low plant population reduces the canopy cover thereby allowing more water to be lost through direct evaporation instead of transpiration. Water Productivity and Transpiration Water actually transpired was much less than the water applied to the field and later “lost” as evaporation and transpiration. The transpired water always makes the benchmark of water actually used productively. Previous research (Papendick et al., 1988) has shown that there is a strong relationship between crop transpiration and dry matter production. Papendick et al. (1988) showed that for every one-millimeter of water transpired by crops about 0.032 t/ha of wheat dry matter could be produced. Also Rockstrom et al. (2003) found out that rainfall water productivity of maize in Kenya ranged from 2.2 to 3.1 kg/ha/mm, which is typical of rain fed agriculture in sub-Saharan Africa. In this study the rainfall water productivity ranged from 0.7 to 3.1 kg/ha/mm, which is quite comparable with results found by Rockstrom et al. (2003). Assessment of water received at farm level could not be adequately accomplished as the water balance could not be closed because of the failure of the equipment used to measure deep percolation (the TDR tube). This calls for further studies in order to close the water balance equation. Other methods such as digging a monitoring well in the experimental plots will show the movement of ground water through capillary action and water loss through deep percolation. 12 CONCLUSION The conclusion from the research is that there is room to significantly improve the water use techniques being applied for crop productivity in Makanya catchment. A clear understanding and quantification of the water partitioning processes is required to maximise water used for transpiration by the plant as transpiration is directly related to biomass production. However, scientific approaches need to be complemented by efficient farm management practices for any technologies to be successful. Maize grain yield was not significantly increased by supplementary irrigation coupled with deep ripping. Ripping significantly increased infiltration rate of the soil in the study area from 6cm/hr to 26cm/hr (Elizabeth’s plot). It can be concluded that supplementary irrigation together with rainfall contribution cannot meet the overall crop water requirement for maize in the study area. Sap flow4 recorder can be used in the separation of transpiration from evapotranspiration. RECOMMENDATIONS Recommend an integrated approach in rainwater management if higher water productivity is to be realised. Ripping coupled with supplementary irrigation should be further analysed over a long time and more replicates in order to see if ripping does not actually increase yields. These introduced innovations are recommended for further use especially ripping. However these innovations should be further studied to find their response to extreme weather conditions like prolonged dry spells and meteorological droughts. Growing of crops with lower crop water requirement like beans is recommended. Use of the Sap flow4 recorder on the measurement of transpiration rate have proved useful and is recommended for further use as it can better quantify transpiration as a proportion of what comes into the field. ACKNOWLEDGEMENTS This study was conducted as part of a collaborative research program between the Smallholder Systems Innovations in Watershed Management (SSI) and the Soil Water Management Research Group (SWMRG) of Sokoine University of Agriculture (SUA) and was funded by Waternet and SSI. LITERATURE CITED Allen, R. G., Pereira, L. S., Raes, D. and Smith, M., 1998. Crop evapotranspiration – Guidelines for computing crop water requirements. FAO Irrigation & Drainage Paper 56. Rome, Italy. Brady, N. C. and Weil, R. R. (1996). The nature and properties of soil. Eleventh edition. Dam, J. C. van, and Malik, R. S. (eds), 2003. Water productivity of irrigated crops in Sirsa district, India. Integration of remote sensing, crop and soil models and geographical information systems. WATPRO final report. 173pp Hillel, D. 1982. Introduction to soil physics. Academic Press, Inc. NY, USA Landon, J. R. 1991 (ed). Booker tropical soil manual: A handbook for soil survey and agricultural land evaluation in the tropics and sub-tropics. Owesis, T., Hachum, A. and Kijne, J. 1999. Water harvesting and supplementary irrigation for improved water use efficiency in dry areas. SWIM Paper 7. International Water Management Institute. Colombo, Sri Lanka Papendick, R. I. and Campbell, G. S., 1988. Concept and management strategies for water conservation in dry land farming. Proceedings of International Conferences on Dry land Farming. Texas August 15-19, 1988. 13 Rockstrom, J., Barron, J., Fox, P., 2003. Water productivity in rain fed agriculture: Challenges and opportunities for smallholder farmers in drought prone tropical agro-ecosystems. CAB International. Rockstrom J. 2001. Green water security for food makers of tomorrow. Windows of opportunity in drought-prone savannahs. Water Science and Technology 43 (4): 71-78) Rosegrant, M. W., Cai, X. and Cline, A. S. 2002. World water and food to 2025: Dealing with scarcity. International Food Policy Research Institute. Slaymaker, T. and Blench, R., 2002 (eds). Rethinking natural resource degradation in sub-saharan Africa: Policies to support sustainable soil fertility management, soil and water conservation among resource-poor farmers in semi-arid areas. Volume II – Country case studies. Stewart, J. I. 1988. Response Farming in Rain-fed Agriculture. Wharf Foundation Press, Davis, California. United Republic of Tanzania (URT) 1997b. Tanzania Agriculture and Livestock Policy. Ministry of Agriculture and Co-operatives. Dar es Salaam. 14