Tree species and pruning regime affect crop yield on bench terraces in SW Uganda

Integration of trees on farms may exert complementary or competitive effects on crop yield. This 4 year study examined novel systems in which Alnus acuminata (alnus), Calliandra calothyrsus (calliandra), Sesbania sesban (sesbania) or a mixture of all three were grown on the degraded upper part of bench terraces in Uganda; beans or maize were grown on the more fertile lower terrace during the short and long rains. Three pruning treatments (shoot, root or shoot + root pruning) were applied to the tree rows adjacent to the crops; shoot prunings were applied as green manure to the woodlot from which they came. Pruning increased survival in calliandra and reduced survival in sesbania; alnus was unaffected. Pruning reduced tree height and stem diameter in alnus, but did not affect calliandra or sesbania. Maize yield adjacent to unpruned calliandra, alnus and sesbania or a mixture of all three was reduced by 48, 17, 6 and 24% relative to sole maize. Shoot pruning initially sustained crop performance but shoot + root pruning became necessary when tree age exceeded 2 years; shoot + root pruning increased maize yield by 88, 40, 11 and 31% in the calliandra, alnus, sesbania and tree mixture systems relative to unpruned trees. Bean yield adjacent to unpruned calliandra, alnus, sesbania and the tree mixture was 44, 31, 33 and 22% lower than in sole crops and pruning had no significant effect on crop yield. The results suggest that sesbania fallows may be used on the upper terrace without reducing crop yield on the lower terrace, whereas pruning of alnus is needed to sustain yield. Calliandra woodlots appear to be unsuitable as crop yield was reduced even after pruning.     

Alley cropping of maize with calliandra and leucaena in the subhumid highlands of Kenya: Part 2. Soil-fertility changes and maize yield

Although N-rich leaf biomass of multipurpose trees is known to be a good source of N to crops, integrating such trees into crop production systems is a major challenge in the development of viable agroforestry systems. An approach to integrating calliandra (Calliandra calothyrsus Meissner) and leucaena (Leucaena leucocephala (Lam.) de Wit), two promising agroforestry tree species, into maize (Zea mays L.) production system was investigated in the subhumid highlands of central Kenya during four maize-growing seasons from 1994 to 1996. The experiment consisted of maize plots to which tree prunings obtained from hedgerows grown either in situ (alley cropping) or ex situ (biomass transfer from outside) were applied. When alley-cropped with leucaena, maize produced significantly higher yields compared to maize monoculture (both non-fertilized and fertilized) treatments, but when alley-cropped with calliandra, the yield of maize was less than that of the monocropped unfertilized control. Application of ex situ grown calliandra and leucaena prunings with or without fertilizer resulted in higher maize grain yield than in the nonfertilized and fertilized treatments. Yields of calliandra alley- cropped maize were 11% to 51% lower than those of nonalley-cropped treatments receiving calliandra prunings from ex situ grown trees; the decrease was 2% to 17% with leucaena, indicating that calliandra hedges were more competitive than leucaena hedges. The alley-cropped prunings-removed treatments produced the lowest maize yields. The study showed that, in the subhumid tropical highlands of Kenya, inclusion of calliandra hedges on cropland adversely affected maize yields. On the other hand, alley cropping with leucaena was advantageous. This revised version was published online in June 2006 with corrections to the Cover Date.     

Introduction of Raikhanim (Ficus semicordata) in a Maize and Finger-Millet Cropping System: An Agroforestry Intervention in Mid-Hill Environment of Nepal

Fodder trees are integral part of farming system in the hills of Nepal, but designed agroforestry interventions targeted to particular trees and crops are not widely available. This paper examines the joint productivity of an agroforestry practice in which Raikhanim (Ficus semicordata) is planted in a maize (Zea mays) and finger-millet (Eleusine coracana) cropping system at Keware Bhanjyang of the western mid-hills of Nepal. Raikhanim seedlings were planted in a row on terrace risers 2, 4 and 6 m apart in ordinary farming conditions, in a randomized block design with three replications. Maize and finger-millet were grown on the terraces as intercrops with a control plot without trees on risers in each replicate. Growth parameters of Raikhanim—height, diameter at 30 cm above ground (D₃₀) and survival rate—were recorded annually in December until trees were lopped for fodder biomass, and crop yields were measured to determine tree-crop interaction effects. Tree height and D₃₀ differed significantly between spacings until trees reached the lopping stage 3½ years after planting, with the highest growth in 4 m spacing. Tree lopping checked the height growth but the diameter growth continued to increase and differed among spacings after lopping. Fodder biomass increased with tree age and was highest under 4 m spacing (7.294 t/ha) followed by 6 m (5.256 t/ha) and 2 m (3.84 t/ha). Finger-millet yield in the experimental plots decreased with tree age due to shading effects, while maize yield was not substantially affected. Among spacings, control plots produced the highest finger-millet yield (1,624 kg/ha) while the 6 m spacing produced the highest maize yield (2,463 kg/ha). It is concluded that planting Raikhanim at 6 m intervals will produce additional fodder without significant effect on maize yield and only a moderate effect on finger-millet yield. The agroforestry practice of planting fodder trees on under-utilised terrace risers is a viable option for mid-hill farmers for simultaneous production of fodder and cereal crops while sustaining the hill farming system.     

Tree growth and management in Ugandan agroforestry systems: effects of root pruning on tree growth and crop yield

Tree root pruning is a potential tool for managing belowground competition when trees and crops are grown together in agroforestry systems. We investigated the effects of tree root pruning on shoot growth and root distribution of Alnus acuminata (H.B. & K.), Casuarina equisetifolia L., Grevillea robusta A. Cunn. ex R. Br., Maesopsis eminii Engl. and Markhamia lutea (Benth.) K. Schum. and on yield of adjacent crops in sub-humid Uganda. The trees were 3 years old at the commencement of the study, and most species were competing strongly with crops. Tree roots were pruned 41 months after planting by cutting and back-filling a trench to a depth of 0.3 m, at a distance of 0.3 m from the trees, on one side of the tree row. The trench was reopened and roots recut at 50 and 62 months after planting. We assessed the effects on tree growth and root distribution over a 3 year period, and crop yield after the third root pruning at 62 months. Overall, root pruning had only a slight effect on aboveground tree growth: height growth was unaffected and diameter growth was reduced by only 4%. A substantial amount of root regrowth was observed by 11 months after pruning. Tree species varied in the number and distribution of roots, and C. equisetifolia and M. lutea had considerably more roots per unit of trunk volume than the other species, especially in the surface soil layers. Casuarina equisetifolia and M. eminii were the tree species most competitive with crops and G. robusta and M. lutea the least competitive. Crop yield data provided strong evidence of the redistribution of root activity following root pruning, with competition increasing on the unpruned side of tree rows. Thus, one-sided root pruning will be useful in only a few circumstances.     

Agroforestry for soil and water conservation in the western Himalayan Valley Region of India 2. Crop and tree production

A ten-year-study (1983 to 1992) conducted on nine 15 × 90 m runoff plots at 4% slope compared production efficiency of Leucaena leucocephala and Eucalyptus hybrid based agroforestry as well as monocropping landuse systems in the warm, subhumid climate of the western Himalayan region of India. Treatments for the first sequence were: monocropping systems of leucaena, eucalyptus, Chrysopogon fulvus grass and maize – wheat rotation, and alley cropping systems of grass and crops at 4.5 and 10.5 m alley widths with paired contour tree rows of leucaena and eucalyptus. In the second sequence, alley width increased to 22.5 m in 1989, grass was replaced by turmeric Curcuma longa and paired contour rows of leucaena hedges were introduced in monocropping systems of grain crops and turmeric. Integration of leucaena and eucalyptus trees with crops caused severe reduction of crop yields ranging from 21 to 92% for wheat grain, 59 to 69% for maize grain, 60 to 67% for dry grass and about 50% for turmeric rhizome depending upon the age of trees and alley width. The grain yield of crops stabilized at about 50% reduction with 22.5 m alley width. Total crop biomass (grain + straw) also revealed a similar trend; however, its magnitude of reduction was less severe than for grain. Production of biomass was much lower near the tree rows than in mid alleys. Managing leucaena as contour hedgerows eliminated crop yield reduction in alleys. Performance of grass and turmeric in alleys was not found to be satisfactory. Biomass produced from trees adequately compensated the crop yield reduction. Land equivalent ratios of agroforestry landuses were comparable or even better than monocropping systems indicating suitability of these systems for the western Himalayan valley region. This revised version was published online in June 2006 with corrections to the Cover Date.     

Trees on farms in Bangladesh: 5. Growth of top- and root-pruned trees in wetland rice fields and yields of understory crops

Growth of trees and seasonal yields of understory crops were measured over a 5-year period for 4 crops grown under 17 tree species at 8 × 8 m spacing in wetland rice fields. All tree species grew well in rice fields, at rates comparable to their growth in forest plantations. Top and root pruning reduced average tree girths by up to 19% and average tree volumes by up to 41%, depending on intensity of pruning. The crops monitored were Oryza sativa, Triticum aestivum, Corchorus olotorius, and Lens culinaris. Crop yields under the trees averaged 93% of the corresponding yields outside the tree canopy. The most important factor affecting the yields of undercrops was tree size (height and/or girth). Differences among tree species and the interaction with species of undercrops were not significant after controlling for tree size. Pruning of roots and branches significantly improved crop yields under trees by amounts proportional to the intensity of root or top pruning. This revised version was published online in June 2006 with corrections to the Cover Date.     

Early growth of Eucalyptus camaldulensis under agroforestry conditions at Mafiga,Morogoro, Tanzania

The growth of Eucalyptus camaldulensis clean weeded, spot weeded and intercropped with maize and beans was studied. At 4 m × 4 m and 5 m × 5 m spacings trees were significantly shorter after 15 months under a conventional spot weeding regime than with clean weeding or intercropping with beans. A satisfactory maize yield (683 kg ha^?1) was recorded from plots with trees spaced at 5 m × 5 m. Plots where trees were spaced at 4 m × 4 m and 3 m × 3 m gave significantly lower yields (444 kg ha^?1 and 283 kg ha^?1, respectively).     

Intercropping ginger and turmeric with poplar (Populus deltoides ‘G-3’ Marsh.)

Performance of two rhizomatous crops i.e. ginger (Zingiber officinale Rosc.) and turmeric (Curcuma longa L.) was investigated under rainfed conditions in pure stands and as intercrops with 5-year-old poplars planted at three spacings viz., 5×5 m, 5×4 m and 5×3 m. The average illumination below the canopies was 53, 46 and 38% of incident radiation, respectively. Both crops performed better as intercrops than as pure stands. Survival was inversely correlated to light intensity. Plant height, tillers per plant and leaves per plant in ginger and leaf length and leaf breadth besides plant height in turmeric were significantly enhanced when intercropped. The rhizome length, rhizome breadth, yield per plant and yield per ha in ginger exceeded under poplars but showed a drastic reduction under the closest poplar spacing. In turmeric, the trend for the first two characters was the same, whereas yield per plant as well as yeild per ha were slightly greater in the open than under 5×3 m spacing. Dry matter content varied significantly with spacing. For quality parameters, only oil content in ginger and oleoresin in turmeric showed significant differences. The cultivation of turmeric proved more remunerative than ginger.Among the poplar spacings, 5×4 m for ginger and 5×5 m for turmeric were delineated as the best spacings.     

Interactive effects of nitrogen and water availabilities on gas exchange and whole-plant carbon allocation in poplar

Cuttings of balsam spire hybrid poplar (Populus trichocarpa var. Hastata Henry x Populus balsamifera var. Michauxii (Dode) Farwell) were grown in sand culture and irrigated every 2 (W) or 10 (w) days with a solution containing either 3.0 (N) or 0.5 (n) mol nitrogen m(-3) for 90 days. Trees in the WN (control) and wn treatments had stable leaf nitrogen concentrations averaging 19.4 and 8.4 mg g(-1), respectively, over the course of the experiment. Trees in the Wn and wN treatments had a similar leaf nitrogen concentration, which increased from 12.0 to 15.8 mg g(-1) during the experiment. By the final harvest, mean stomatal conductances of trees in the wN and wn treatments were less than those of trees in the Wn and WN treatments (1.8 versus 4.6 mm s(-1)). Compared to the WN treatment, biomass at the final harvest was reduced by 61, 72 and 75% in the Wn, wN and wn treatments, respectively. At the final harvest, WN trees had a mean total leaf area of 4750 +/- 380 cm(2) tree(-1) and carried 164 +/- 8 leaves tree(-1) with a specific leaf area of 181 +/- 16 cm(2) g(-1), whereas Wn trees had a smaller mean total leaf area (1310 +/- 30 cm(2) tree(-1)), because of the production of fewer leaves (41 +/- 6) with a smaller specific leaf area (154 +/- 2 cm(2) g(-1)). A greater proportion of biomass was allocated to roots in Wn trees than in WN trees, but component nitrogen concentrations adjusted such that there was no Wn treatment effect on nitrogen allocation. Compared with WN trees, rates of photosynthesis and respiration per unit weight of tissue of Wn trees decreased by 28 and 31%, respectively, but the rate of photosynthesis per unit leaf nitrogen remained unaltered. The wN and Wn trees had similar leaf nitrogen concentrations; however, compared with the Wn treatment, the wN treatment decreased mean total leaf area (750 +/- 50 cm(2) tree(-1)), number of leaves per tree (29 +/- 2) and specific leaf area (140 +/- 6 cm(2) g(-1)), but increased the allocation of biomass and nitrogen to roots. Net photosynthetic rate per unit leaf nitrogen was 45% lower in the wN treatment than in the other treatments. Rates of net photosynthesis and respiration per unit weight of tissue were 48 and 33% less, respectively, in wN trees than in Wn trees.     

Net photosynthetic response to light intensity of shoots from different crown positions and age in picea abies (L.) karst

Carbon dioxide uptake response to light intensity of detached shoots of open‐ and forest‐grown Norway spruces was investigated in controlled standard conditions. The initial slope of the CO₂ uptake light response curve, calculated on leaf area basis and the internal conductance, calculated on leaf weight basis were almost constant for shoots from different positions in the forest canopy. In open‐grown trees the maximum photosynthesis and internal conductance (on leaf weight basis) decreased in relation to the tree age, so that in the 8‐year‐old tree it was about two times of that in the 66‐year‐old tree. These results suggest that 1) the enhanced efficiency of “shade” shoots is caused by morphologic adaptation, i.e. the more sparsely packed photosynthetic apparatus in needles and shoots, and 2) adaptational possibilities of Norway spruce photosynthetic apparatus, its ecological plasticity diminish during tree ontogenesis.     

Performance of the West African vegetable cropTelfairia occidentalis as a function of distance from a row of mango trees

Vegetable yield ofTelfairia occidentalis, grown at various distances (3, 4, 5 and 6 m) from a row of mango trees, remained unaffected by proximity to the tree line, until about 170 days after planting (DAP). The onset of the dry season, when water usually becomes limiting, occurred at about 170 DAP. Results support the suggestion that crops whose harvestable parts are vegetative tend to be less affected when grown in proximity to trees, but provided water is not limited.     

Growth and yield of sorghum or cowpea in an agrisilviculture system in semiarid India

This study was conducted near Hyderabad, India during 1991–1994 to quantify the effects of shoot pruning, fertilization, and root barriers around Leucaena leucocephala trees on intercropped sorghum(Sorghum bicolor) or cowpea (Vigna unguiculata) crop production under rainfed conditions. Crop plants grown with pruned trees attained higher dry matter and leaf area index than did those with unpruned trees. Two-year mean grain yields of sorghum with no root barriers were76% and 39% of pure crop yield (1553 kg ha–1)for pruned and unpruned trees, respectively. Corresponding values for cowpea were 49% and 26% of pure crop yield (1075 kgha–1). Sorghum or cowpea intercropped with trees responded to fertilizer application more strongly than did their respective pure crops, suggesting an increased need for fertilizer application in this agrisilviculture system over that currently used for pure crops. Impact of root barriers was small on either crop. Irrespective of root barriers, a high response to tree pruning suggested above ground competition for light dominated tree/crop interactions in this agrisilviculture system. This revised version was published online in June 2006 with corrections to the Cover Date.     

Improving the Traditional Acacia Senegal-Crop System in Sudan: The Effect of Tree Density on Water Use, Gum Production and Crop Yields

The traditional Acacia senegal bush-fallow in North Kordofan, Sudan, was disrupted and the traditional rotational fallow cultivation cycle has been shortened or completely abandoned, causing decline in soil fertility and crop and gum yields. An agroforestry system may give reasonable crop and gum yields, and be more appealing to farmers. We studied the effect of tree density (266 or 433 trees ha⁻¹) on two traditional crops; sorghum (Sorghum bicolor) early maturing variety and karkadeh (Hibiscus sabdariffa), with regard to physiological interactions, yields and soil water depletion. There was little evidence of complementarity of resource sharing between trees and crops, since both trees and field crops competed for soil water from the same depth. Intercropping significantly affected the soil water status, photosynthesis and stomatal conductance in trees and crops. Gum production per unit area increased when sorghum was intercropped with trees in low or high density. However, karkadeh reduced the gum yield significantly at high tree density. Yields of sorghum and karkadeh planted within trees of high density diminished by 44 and 55% compared to sole crops, respectively. Intercropping increased the rain use efficiency significantly compared to trees and field crops grown solely. Karkadeh appears to be more appropriate for intercropping with A. senegal than sorghum and particularly recommendable in combination with low tree density. Modification of tree density can be used as a management tool to mitigate competitive interaction in the intercropping system.     

Effect of crop load on fruiting and leaf photosynthesis of 'Braeburn'/M.26 apple trees

Four-year-old apple (Malus x domestica Borkh.) trees cv. 'Braeburn' on M.26 rootstock were thinned at full bloom to establish six crop loads ranging from a heavy crop to a deflowered treatment. At harvest, mean yield per tree varied from 0 to 38 kg and mean fruit weight ranged from 225 g in the heaviest cropping treatment to 385 g in the lightest cropping treatment. Light cropping resulted in a significant advance in fruit maturity as indicated by background color, starch/iodine score and soluble solids. There were small differences in leaf photosynthetic rate among the treatments when shoot growth was active. However, in early January, coincident with cessation of shoot growth and maximum rate of accumulation of fruit weight, leaf assimilation rate was reduced by as much as 65% on the deflowered trees compared to the trees carrying the heaviest crop. Leaf assimilation rate showed a curvilinear response to crop load at this time, with little increase in leaf assimilation when crop load exceeded 12 fruit m(-2) leaf area.     

Root development in a Sesbania sesban fallow-maize system in Eastern Zambia

Roots of trees (Sesbania sesban) and crops (Zea mays) were quantified during two tree/crop cycles in a sequential tree — crop system at Chipata, Eastern Zambia. The experiment included one- and two-year fallows as well as fertilized and unfertilized controls. The roots of S. sesban represent a standing biomass in the soil of 3 Mg ha^t-1 in the top 1.5 m after two years, with 45–60% and 70–75% being in the top 25 and 50 cm respectively. S. sesban fallow improved early rooting and growth of the following maize crop. Increased soil infiltration was also observed in the two-year fallow treatment, as well as decreased bulk density and resistance to penetration in the soil. No differences between maize root parameters could be detected at tasselling, nor differences between nutrient status of the different treatments. Study results indicate that under the drought-prone conditions of Eastern Zambia, where improved soil physical conditions are important for early deep rooting of crops and access to water and nutrients, tree roots could play an important role in the fallow effect. Further studies are required to assess the relative importance of the improvement of soil chemical and physical properties.Submitted as ICRAF Journal Article # 95/28.     

Soil water competition in a temperate hedgerow agroforestry system in South Africa

An on-farm trial was conducted to determine dry matter production of four fodder tree species and their effect on soil water and maize production. The trees were planted in rows intercropped with maize. The four tree species selected were Acacia karroo Hayne (indigenous fodder tree), Leucaena leucocephala (Lam.) De Wit (nitrogen fixing), Morus alba L. (fodder and fruit), and Gleditsia triacanthos L. (fodder and fuel). Volumetric soil water was measured in the upper 0.3 m of soil in each row of the trial using the time domain reflectometry technique. The neutron probe technique was used for monitoring the water content deeper in the soil. Geostatistical methods were used to analyse treatment differences in the upper 0.3 m of soil. The soil water content did not differ significantly between the maize and tree rows indicating that competition for water in the upper horizon was not the reason for lower maize yields. However, at greater soil depths (75–125 cm) trees in the wide spacing used less water than those in the narrow spacing. Light interception was an important factor in reducing maize yields in the row nearest to the trees. High soil water values recorded during summer indicated that in the current cycle of good rainfall the plants in the agroforestry trial were not stressed. Thus the trees do not compete with the crops for soil moisture in good rainfall seasons. However, this study would need further evaluation for the competition for water for the low rainfall years. Since the trees have access to water at greater depths, they are likely to be more productive into the dry season than shallow rooted crops.     

Trade-offs analysis for possible timber-based agroforestry scenarios using native trees in the Philippines

To assess possible new agroforestry scenarios the tree–soil–crop interaction model in agroforestry systems (WaNuLCAS 3.01) was used based on-site specific data collected from Tabango (Central Philippines). Three native timber trees (Shorea contorta Vid., Pterocarpus indicus Juss., and Vitex parviflora Willd.) and one widely spread exotic specie (Swietenia macrophylla King.) were simulated under different intercrop scenarios with maize (Zea mays L.) and subsequently compared. Model simulation results quantified and explained trade-off between tree and crop. For example, higher tree densities will lead to a loss of crop yield that is approximately proportional to the gain in wood volume. However, beside this trade-off effect, there is considerable scope for tree intercropping advantage under a fertilization scenario, with systems that yield about 50% of the maximum tree biomass still allowing 70% of monoculture maize yield. Maximum tree yield can still be obtained at about 20% of the potential crop yield but intermediate tree population densities (400 trees ha⁻¹) and the resulting larger stem diameters may be preferable over the larger total tree biomass obtained at higher tree densities. Another advantage from intercropping systems is that trees directly benefit from the inputs (i.e., fertilizer) that are applied to the crops. The three native trees species studied have different performance in relation to productivity but are similar to (or even better than) S. macrophylla.     

Effects of tree shading on maize crop within a Poplar-maize compound system in Hexi Corridor oasis,northwestern China

Effects of tree shading on crop within agroforestry system in the oasis regions or under irrigation agriculture are poorly understood. Objective was to determine maize yield variation and its reasons when tall Poplar trees (Populus gansuensis) were grown in western and eastern margins of the crop field in Hexi Corridor desert oasis in northwestern China. Maize growing among the 50 m alley space were divided into five regions, including shading region in western side (W-S), non-shading region in western side (W-NS), middle region (M), shading region in eastern side (E-S) and non-shading region in eastern side (E-NS), for all measurements. Measurements of leaf area expansion were carried from seeding to flowering stage, photosynthetic physiological parameters and microclimate characteristics were measured at booting stage, grain yield and other harvest parameters were obtained at harvest time. We found that tree shading reduced the crop yield by 27 and 22% in western and eastern regions, respectively, and also, mean crop yield for western side was 23% lower than eastern side. The direct reason of yield variation was transpiration rate (E) variation at booting stage, that is, maize which had higher daily mean E would obtained higher yield. Moreover, changes of incident photosynthetically active radiation (PARi), air temperature (T _a) and CO₂ concentration (C _a) were the basic reasons of yield variation among different regions. Because higher PARi, higher T _a and lower C _a, which caused by the tree shading, would all led to higher E and finally higher crop yield.     

Defining competition vectors in a temperate alley cropping system in the midwestern USA: 1. Production physiology

With renewed interest in the use of ecologically-designed, sustainable agricultural systems for temperate regions of the world, agroforestry is being proposed as an alternative to intensive production of crops in monocultures. However, the knowledge-base for understanding and managing complex, multi-strata systems worldwide is limited, particularly so for temperate regions. We examined an alley cropping system in the midwestern US where maize (Zea mays L.) is grown in alleys between tree rows of either black walnut (Juglans nigra L.) or red oak (Quercus rubra L.). During a course of ten years, crop yields in rows adjacent to tree rows declined by 50% or more. With the experimental introduction of barriers to separate tree and crop root systems, yields in the rows near trees were equal to those of the center row (and monoculture). Irrespective of a high correlation between photosynthetically active radiation and net photosynthesis, shading did not have a major influence on crop yield. At this stage of system development (11 year old trees), influence of incident PAR on crop yield seems to be minimal. Subsequent papers in this series examine the sharing of belowground resources between trees and crops to quantify the competitive interactions that impact crop yields and their implications for economic return to the farmer.This revised version was published online in November 2005 with corrections to the Cover Date.     

Water use in a Grevillea robusta–maize overstorey agroforestry system in semi-arid Kenya

Novel approaches involving a combination of sap flow measurements of transpiration and allometric estimates of biomass production were used to determine seasonal water use by trees and crops in agroforestry systems. The results were used to test the hypothesis that agroforestry may improve productivity by capturing a greater proportion of annual rainfall than annual crops. Grevillea robusta A. Cunn., which is reputed to have a deep rooting habit, was grown in semi-arid Kenya either as sole stands or in combination with maize (Zea mays L.). Water use by individual trees and maize plants was determined using constant temperature heat balance gauges and scaled to provide stand-level estimates of transpiration based on linear relationships (r²>0.70) between sap flow and leaf area across a range of tree ages and environmental conditions. Maximum stand-level transpiration rates for grevillea ranged from 2.6 to 4.0 mm per day, consistent with previous studies in similar environments. Biomass production by grevillea was closely correlated with stand-level transpiration (r²>0.69–0.74), suggesting that non-destructive estimates of biomass increments can be used to provide reliable estimates of seasonal transpiration. Cumulative water use by grevillea over the 4.5-year observation period was comparable in the sole tree and agroforestry treatments, reaching a maximum utilisation of annual rainfall of 64–68% 3–4 years after planting. Approximately 25% of the water transpired by the trees was used during the dry season, indicating that they were able to utilise off-season rainfall, comprising 16% of the total annual rainfall, and residual water remaining in the soil profile after the cropping period. During the 1995 long rains, when 221 mm of rain was received, transpiration by sole maize was <50% of precipitation, compared to ca. 85% by the trees in the sole grevillea and agroforestry treatments. These results confirm that agroforestry systems may greatly increase rainfall utilisation compared to annual cropping systems. However, careful consideration of the tradeoffs between the loss of crop production and the additional value provided by tree products is essential.