The performance of newly developed extra-short-duration pigeonpea (Cajanus cajan) genotypes and traditional short-duration pigeonpea cultivars was compared in rotation with wheat in on-farm trials conducted in 1996–97 and 1997–98 in Sonepat (28° N) district in Haryana, and in 1996–97 at Ludhiana (30° N) district in Punjab, India. At both locations, a wheat crop (Triticum aestivum cv. HD 2329) followed pigeonpea. At Sonepat, an indeterminate extra-short-duration genotype ICPL 88039 matured up to three weeks earlier, yet gave 12% higher yield (1.57 t ha−1) and showed less susceptibility to borer damage than did the short-duration cv. Manak. At Ludhiana, extra-short-duration pigeonpea genotypes, ICPL 88039, ICPL 85010 and AL 201 gave similar grain yields to the short-duration T 21 in spite of maturing three to four weeks earlier. Yields of wheat crops following extra-short-duration genotypes were up to 0.75 t ha−1 greater at Sonepat and up to 1.0 t ha−1 greater at Ludhiana. The results of the study provide empirical evidence that extra-short-duration pigeonpea genotypes could contribute to higher productivity of pigeonpea–wheat rotation systems. Most of the farmers who grew on-farm trials in Sonepat preferred extra-short-duration to short-duration pigeonpea types for their early maturity, bold seed size, and the greater yield of the following wheat crop.

A study was conducted in the humid-forest, forest-savanna, and Guinea savanna zones of West Africa from 1993 to 1999 to examine the effect of managing crop residues from cereal–legume cropping systems for mulch and fodder for sheep. Increasing the proportion of total crop residues produced from a unit area of land and used as mulch increased maize grain yield, soil organic carbon, nitrogen and available phosphorus. The extra increases obtained when more than half the crop residues were applied as mulch were relatively small, however, suggesting that 25–50% of the crop residues could be removed as feed without any detrimental effect. When any crop residues rejected by sheep were mixed with livestock urine and faeces and returned to the respective fields from where the crop residues had been removed, subsequent grain yield and soil organic carbon, nitrogen, and available phosphorus increased. The study demonstrated the possibility of managing crop residues for increased productivity in smallholder mixed crop–livestock systems.

The last 50 years have seen a rapid expansion of cultivated area in semi-arid areas of West Africa. This has precipitated a change from traditional fallowing to more pro-active soil fertility management techniques. Smallholder farmers employ a range of technologies to enhance soil fertility and manure is a cornerstone of many of the soil fertility management strategies they use. This paper reviews manure management by smallholder farmers. It considers factors that affect the quality of the manure used, including methods for keeping livestock and storing manure. The paper reviews the strategies, such as night parking and crop-livestock integration, which farmers employ to ensure that manure reaches their fields. The nutrient balances of two farming systems are presented as evidence for the importance of manure as a nutrient source. Rangeland-to-cropland nutrient transfers are contrasted with nutrient recycling through crop-livestock integration. The paper concludes that within the constraints in which smallholder farmers operate in semi-arid West Africa, manure will remain an important component of soil fertility management strategies for the foreseeable future. Integrated nutrient management strategies that take into consideration the circumstances of farmers, and the resources available to them, are the best way forward. Appropriate interventions need to focus on improving manure management to ensure that the material which farmers so laboriously prepare and transport is of the best possible quality.

The on-farm evaluation of three feeding strategies for smallholder campesino dairy herds was undertaken in the highlands of Central Mexico with cooperating farmers: traditional, alternative and intermediate feeding strategies. All three incorporated grazing of cultivated pastures, but concentrate use was 4.0 and 7.0 kg per cow per day in the rainy and dry seasons respectively for the traditional feeding strategy, 3.3 for the alternative feeding strategy in both rainy and dry seasons and 5.0 and 9.0 kg per cow per day in the rainy and dry seasons respectively for the intermediate feeding strategy. Feeding during the dry season was based on maize straw in the traditional strategy, on pasture complemented with maize silage in the alternative feeding strategy, and with some maize silage but mostly maize straw in the intermediate feeding strategy. The trial ran between 25 September 1996 and 24 September 1997, divided by season into four, thirteen-week periods. Milk yields were recorded once per week, and live weight and body-condition score, every 28 days. Three cows per strategy that completed each period were blocked according to stage of lactation and used for the statistical analysis of a split-plot design with feeding strategies as main plots and weeks as split-plots. Overall mean milk yields were 15.5±2.05, 13.4±2.43 and 12.4±2.37 kg milk per cow per day for alternative, intermediate and traditional feeding strategies respectively, with significant interactions (P<0.05) for autumn, winter and summer. The alternative feeding strategy (with over 50% less concentrate) produced milk yields 0.24 and 0.38 higher than the intermediate and traditional feeding strategies in the dry season. In the rainy season, milk yields with the alternative feeding strategy (30% less concentrates) were 0.09 and 0.16 higher than the other two strategies. Margins per day of family labour were: alternative feeding strategy US$36.48 per day, intermediate feeding strategy US$9.22 per day and the traditional feeding strategy US$9.11 per day, although in the case of the last two there were two family members in charge of the dairy herds. Results demonstrate the successful integration of grazed pasture and maize silage for the efficient production of milk, and provide evidence on the productive use of limited campesino land resources in the face of unviable economic conditions for maize grain production.

Donors have invested more than US$45 million in seed production projects in Niger during the past two decades. These investments have largely failed. Public seed systems consistently supplied less than 2% of the total national seed planted by farmers. Through subsidies, seed prices represent less than one-third of the average cost of seed production. In contrast, at the village level, most farmers consistently obtain pearl millet (Pennisetum glaucum) seed from their own harvests, from neighbours or from village markets. Seed is of acceptable quality and a range of varieties is available. Village seed systems offer a cheaper and more efficient means of delivering seed to farmers. Future investment in seed systems development should target improvements in the capacity of village seed systems to maintain and distribute seed security stocks in drought years. Efficient seed producers or groups of farmers in each community should be identified and encouraged to become entrepreneurs tasked with the multiplication and distribution of new pearl millet varieties.

Cotton (Gossypium hirsutum) is the main agricultural export commodity from Tanzania. The most significant disease of the crop is fusarium wilt caused by Fusarium oxysporum f.sp. vasinfectum. Phytosanitary measures instituted at the cotton ginneries to prevent the distribution, for planting, of seed infected with the wilt fungus have become difficult to apply since economic liberalization and the entry of the private sector into cotton ginning and lint marketing. Surveys of cotton fields, ginneries and cotton buying posts were conducted in order to determine the factors affecting disease incidence and spread. In affected fields, disease incidence was generally less than 5%. Where it was greater than this, wilt symptoms were associated with root damage caused by the root-knot nematode (Meloidogyne incognita). At a number of ginneries, herdsmen were allowed to remove seed husks that accumulate at the ginneries as a byproduct of oil extraction. The husks are used as cattle feed and this was identified as a potential source of disease spread. At the buying posts visited, there was no system for separating cotton varieties or for identifying seed cotton purchased from villages infected with fusarium wilt. As a result, seed subsequently distributed for planting is likely to be a source of infection for the spread of this disease. The implications of economic liberalization in the cotton sector are discussed with respect to seed distribution and management of fusarium wilt.

The effect of the root hemiparasitic angiosperm Striga hermonthica on the gowth, photosynthesis and partitioning of biomass in maize was studied in glasshouse experiments in two seasons. In both seasons, nitrogen was applied to the plants at rates equivalent to 20, 60 and 120 kg ha−1. There was no significant Striga×nitrogen interaction on the responses measured. Averaged across all nitrogen treatments, maize plants infected with S. hermonthica had smaller leaf areas and accumulated less biomass, than did uninfected plants. The leaf area of infected and uninfected plants increased asymptotically from emergence to the final harvest (about the 18-leaf stage) at which time the leaf area of infected plants was 63% that of uninfected plants. The rates of photosynthesis of the youngest, fully expanded leaves of infected plants, averaged across N treatments, were significantly lower than for uninfected plants. However, stomatal conductance and the sub-stomatal CO2 concentration were unaffected by Striga infection. Although infection with Striga significantly reduced shoot biomass (dry weight at final harvest was 37% that of uninfected plants in 1995 and 63% in 1996), there were no significant effects of Striga infection on root biomass so that the infected plants partitioned a significantly greater proportion of their total biomass to roots compared with the uninfected plants. The allometric coefficients, though, were similar for both infected and uninfected plants. Averaged across infected and uninfected plants, application of nitrogen increased total leaf area per plant and root and shoot biomass but did not change the proportion of total biomass partitioned to roots. These results did not show a major effect of nitrogen on the relative growth response of maize to infection with Striga.

A three-year field experiment was undertaken to evaluate the effects of tillage methods on grain yield and yield components of wheat (Triticum aestivum) in continuous cropping. The experiment was conducted on a Ramjerd, fine, mixed, mesic, typic Calcixerepts soil. Wheat was sown: (1) after burning residues followed by conventional tillage, (2) after complete residue removal followed by conventional tillage, (3) after soil incorporation of residues followed by conventional tillage, (4) into untilled residues, (5) using chisel seeder after field irrigation, (6) using chisel seeder plus herbicide application, and (7) after disking. Residue burning and removal increased spikes per square metre, grain per spike, 1000-grain weight, grain yield and harvest index compared with other treatments. This was due primarily to weed interference and lack of uniform crop establishment in the presence of residues. Reduced tillage methods retained more residues on the soil surface, which provided unsuitable conditions for crop emergence and growth. The incorporation of residues led to a build up of carbon in the soil, with lower grain yields compared with residue burning and removal, but these yields were higher than those of chisel-seeded plots.

Permanent tree crop agriculture and multi-strata agroforestry are among the most promising options for the agricultural use of the mostly nutrient-poor and physically fragile soils of central Amazonia. In general, though, information on the optimum management of local tree crops under these conditions is inadequate. Annatto (Bixa orellana) is a small tree of tropical American origin whose seeds contain a non-toxic, carotenoid dye that is used widely in food and cosmetic products. The authors studied seed yields at ages four to seven years, biomass at seven years, nutrient accumulation and mineral nutrition of annatto trees growing in multi-strata agroforestry with different inputs of fertilizer and lime on a xanthic Ferralsol in central Amazonia. Leaf samples of three age classes were collected four times during one year, and nutrient concentrations were related to soil nutrient status, growth and yield of the trees in order to develop an optimum sampling scheme for foliar analysis. Growth and yields of the trees showed a pronounced response to increased fertilizer and lime input, which seemed to be due mainly to improved availability of phosphorus. Nitrogen fertilizer had no effect and may not be necessary for well-established trees on this soil. High litter quality and substantial nutrient recycling with annual pruning make annatto a valuable component for agroforestry systems. In this experiment, however, yields were low and decreased after the fifth year, presumably as an effect of infertile soil, shading by larger trees and, possibly, a negative effect of the drastic annual pruning with removal of the entire leaf and small-branch biomass on the vitality of the trees. Annatto is probably best suited for associations with small tree crops. Less drastic pruning treatments than those practiced in this experiment may be preferable.

New varieties are often developed on research stations with limited farmer involvement until the final stages of testing. Recently, farmer participatory approaches have been employed to include farmers' input into breeding objectives at much earlier stages of varietal development. This paper reports on a farmer survey in Uganda to record farmer selection criteria for banana cultivars as a pre-breeding activity. From ranked data, bunch size and crop maturation time were the key selection criteria in the county's most important commercial production zones. Stand longevity, taste and crop maturation time were key factors where banana production is in decline and sale of bananas less important. Principal component analysis revealed that, nationally, stand longevity and tolerance of marginal soils were the most important criteria, probably because the larger part of overall production is for subsistence.

Soil fertility constraints to crop production have been recognized widely as a major obstacle to food security and agro-ecosystem sustainability in sub-Saharan West Africa. As such, they have led to a multitude of research projects and policy debates on how best they should be overcome. Conclusions, based on long-term multi-site experiments, are lacking with respect to a regional assessment of phosphorus and nitrogen fertilizer effects, surface mulched crop residues, and legume rotations on total dry matter of cereals in this region. A mixed model time-trend analysis was used to investigate the effects of four nitrogen and phosphorus rates, annually applied crop residue dry matter at 500 and 2000 kg ha−1, and cereal-legume rotation versus continuous cereal cropping on the total dry matter of cereals and legumes. The multi-factorial experiment was conducted over four years at eight locations, with annual rainfall ranging from 510 to 1300 mm, in Niger, Burkina Faso, and Togo. With the exception of phosphorus, treatment effects on legume growth were marginal. At most locations, except for typical Sudanian sites with very low base saturation and high rainfall, phosphorus effects on cereal total dry matter were much lower with rock phosphate than with soluble phosphorus, unless the rock phosphate was combined with an annual seed-placement of 4 kg ha−1 phosphorus. Across all other treatments, nitrogen effects were negligible at 500 mm annual rainfall but at 900 mm, the highest nitrogen rate led to total dry matter increases of up to 77% and, at 1300 mm, to 183%. Mulch-induced increases in cereal total dry matter were larger with lower base saturation, reaching 45% on typical acid sandy Sahelian soils. Legume rotation effects tended to increase over time but were strongly species-dependent.

Tea (Camellia sinensis) clones (PC113 and SFS204) sensitive to very dry air and clones (PC114 and SFS150) that are tolerant, were studied at two tea estates (Tshivhase and Grenshoek) in the Northern Province of the Republic of South Africa. Among the morphological leaf traits studied, stomatal density, pore diameter and pore depth were not linked consistently to stress tolerance. Cuticle thickness was not a good indicator of stress tolerance because genetic differences between clones were confounded by the clonal response of wax production to stress. In contrast, measured leaf conductance to water vapour transport was larger and leaf water potential was lower in sensitive clones, but only with more severe atmospheric stress (Grenshoek). Also the ratio of the calculated maximum stomatal conductance in old and young leaves was higher in sensitive clones, suggesting that the loss of a larger fraction of the total stem flow by old leaves enhanced the stress experienced by the young leaves. However, this indicator was valid only under the more stressful microclimate of Grenshoek. The results indicate that even promising criteria for stress tolerance should be tested by exposure to stress during selection.

The East African highlands, home to more than 80 cultivated varieties of locally evolved bananas, constitute a secondary centre of banana diversity. Uganda is the leading producer and consumer of banana in the region and also enjoys the highest diversity of a group of bananas uniquely adapted to this region. These East African highland bananas comprise cooking and brewing types. The former is a staple for more than 7 million people and thus important for food security. Little is known about the distribution of the vast germplasm and this study was set up to help determine a distribution pattern and to understand the dynamics of cultivar change using farmers participatory appraisal methods. The study involved a guided interview with 120 farmers, at 24 sites throughout the banana-growing region of Uganda, to reveal cultivar diversity, proportions, distribution and preferences. Cultivar diversity ranged from 18 to 34 (mean = 26) cultivars per site, and from 4 to 22 (mean = 12.3), cultivars per individual farm. Such high diversity was attributed to a variety of end uses, better food security and the perception that each cultivar had a unique range of strengths and weaknesses. Highland banana (AAA-EA) represented 76% of total production while Kayinja (‘Pisang Awak’ subgroup) (ABB) contributed 8%; Ndiizi (’Ney Poovan’ subgroup) (AB) 7%; Kisubi (‘Ney Poovan’ subgroup) (AB) 5%; Gros Michel (‘Bogoya’) (AAA) 2%; and plantain (AAB) 2%. Although 130 highland cultivars were recorded, only 10 constituted 50% of highland banana production while 45 cultivars were found at only 1 or 2 sites. A few cultivars showed more universal distribution and it is proposed that these may be the oldest and best performing local landraces.

Land degradation is one of the major problems related to low productivity in Ethiopia. Vertisols are potential agricultural lands but are underutilized due to waterlogging during the rainy season and are prone to soil erosion due to the traditional practices of planting crops at the end of the rainy season exposing bare land to heavy rains. An animal-drawn implement, the broadbed maker (BBM), was tested to facilitate drainage of the Vertisols. Improved drainage enables earlier planting of crops to use the resultant prolonged growing period for attaining higher yields. In order to further improve the BBM package, the frame of the BBM was used for attachment of other implements to minimize tillage practices and reduce seed and fertilizer rates, labour and animal power requirements and soil erosion. The attachments were tested both on-station and on-farm with positive results.

Enset (Ensete ventricosum) production is declining, and it faces genetic erosion due to drought, diseases and population pressure. Participatory Rural Appraisal (PRA) and additional formal survey studies on 315 households were conducted over three consecutive years (1998–2000) in the Sidama, Wolaita and Hadiya ethnic regions of southern Ethiopia to assess traditional cultivation methods, analyse the production systems, and evaluate farm-based enset biodiversity. The regions differ in terms of cultural background, resources, farming systems, population density, and agro-ecology. Furthermore, the methods for initiating suckers and the frequency of transplanting vary among the three regions.

Diverse enset landraces were identified in the Sidama (52), Wolaita (55) and Hadiya (59) regions. Sidama farmers had the highest number of landraces per farm, 57% and 21% more than found on Wolaita and Hadiya farms respectively. In all three regions, landrace diversity was influenced by household resources, cultural background, population pressure, and agro-ecology. There were significant differences in the average number of enset landraces and livestock between rich and poor households in the three regions. Rich farmers had more land and manure-producing livestock, and they planted more enset landraces than did poor farmers. In all three regions, women proved to be more experienced than men in identifying enset landraces.

The number of enset landraces per farm was significantly correlated with other household characteristics for resource-rich Sidama farmers and with the number of livestock and area of farmland for resource-rich Hadiya farmers. This suggests that middle-income or poor farmers concentrate on annual crops, rather than on growing the perennial enset plant. More research is needed to identify, characterize and conserve genetic diversity, and to improve the cultivation practices for enset. The cultural, socio-economic, and gender-associated aspects of enset cultivation need to be assessed to understand the dynamics of enset biodiversity.

In Malawi, intensive production of maize (Zea mays) is pursued on over 60% of smallholder land, yet application of nutrients is almost nil. To improve adoption of soil productivity-enhancing technologies, two participatory methods were pursued: (i) a novel ‘mother-and-baby’ trial design and (ii) participatory action research with communities in a southern Malawi watershed. The central ‘mother trial’ was managed by researchers (replicated within a site) and systematically linked to farmer-managed ‘baby’ trials to cross-check biological performance with farmer assessment. The watershed approach involved a partnership of researchers and farmers addressing soil management. Technologies tested in both approaches integrated legumes into existing maize-based systems, sometimes in combination with inorganic fertilizers. Across methods, legume intensification increased yields by approximately 40% (net benefit increase of approximately US$50 ha−1) and fertilizer increased yields by approximately 70% compared with continuous maize grain yields of about 1100 kg ha−1. Farmer assessment prioritized technologies that included secondary benefits, such as weed suppression, grain legume yields, and low-labour-demanding fertilizer. A survey indicated that participating researchers and extension staff had reservations about the amount of time required to interact with farmers, and no clear consensus emerged regarding the best approach. There has been wider adoption of the mother-and-baby trial method by scientists in neighbouring countries, indicating the value of systematically incorporating farmers' input.

Foliar sprays of 25–100 mg L−1 dinitrophenol, sodium azide (NaN3) and thiourea solution applied to Azolla microphylla (strains 202 and 203) at seven days after inoculation significantly increased the sporulation frequency and number of micro- and mega-sporocarps per plant. Azolla biomass production was decreased by the application of NaN3 at 50–100 mg L−1 and dinitrophenol or thiourea at 75 and 100 mg L−1. Both the sporulation frequency and sporocarp number were maximal with dinitrophenol and NaN3 at 100 mg L−1 and thiourea at 75 mg L−1. At this concentration, NaN3 was more effective than dinitrophenol and thiourea in enhancing sporulation in A. microphylla (strain 203). The sporulation frequency and micro-sporocarp number were significantly higher with gibberellic acid (2.5 mg L−1) than with indole acetic acid (6 mg L−1) application, while the effects of NaN3 were comparable with those of gibberellic acid. The mega-sporocarp numbers in these treatments were comparable. The combined application of any two of the metabolic inhibitors had antagonistic effects and decreased the sporulation frequency and sporocarp number over their individual applications. Combined applications also substantially reduced the biomass production. The application of gibberellic acid or indole acetic acid along with one of the metabolic inhibitors had synergistic effects on the formation of sporocarps and significantly increased the biomass production, irrespective of the metabolic inhibitor treatment. The number of both micro- and mega-sporocarps in the gibberellic acid + NaN3 and indole acetic acid + NaN3 treatments, and micro-sporocarps in the gibberellic acid + dinitrophenol, indole acetic acid + dinitrophenol and indole acetic acid + thiourea treatments was significantly higher than when each of these chemicals was applied alone. Of the different combinations of gibberellic acid, indole acetic acid and NaN3, gibberellic acid + NaN3 recorded the highest sporulation frequency and micro-sporocarp number, and indole acetic acid + NaN3 the highest mega-sporocarp number. Application of all three chemicals together had no extra advantage.

The effects of four calcium-containing materials on soil pH, nutrient availability and productivity of groundnut (Arachis hypogaea) on acid soils were studied in field experiments conducted at the Horticulture Research Centre and the Makoholi Experiment Station in Zimbabwe. The materials were calcitic limestone, dolomitic limestone, gypsum and single super phosphate.

Application of dolomitic or calcitic limestone significantly increased soil pH and exchangeable calcium and magnesium levels, whereas gypsum and single super phosphate did not. Application of 4000 kg ha−1 calcitic limestone increased calcium levels by up to 125% at the Horticulture Research Centre and 872% at Makoholi Experiment Station, while dolomitic limestone increased magnesium levels by 183% and 473% at the Horticulture Research Centre and Makoholi Experiment Station respectively. Combining gypsum and single super phosphate with 2000 kg ha−1 calcitic or dolomitic limestone did not affect soil pH differently than the limes alone. Furthermore, gypsum and single super phosphate applications on their own had no effect on pH, calcium and magnesium levels. The phosphorus, potassium and nitrogen levels were not significantly affected by the calcium-containing materials.

Application of either dolomitic or calcitic limestone significantly improved productivity of groundnut, with 4000 kg ha−1 calcitic limestone producing the largest increases in kernel yields at both sites. Yield responses to gypsum were not significantly different from the control treatment. Despite the notable changes in soil chemical properties observed at both sites, deficiencies of nitrogen, phosphorus, potassium, calcium and magnesium, coupled with low pH per se, remained the primary reasons for poor growth on the acid sandy soils at both sites.

Combined socio-economic and agronomic surveys were conducted during two growing seasons with 20 rice (Orgyza sativa) farmers in the irrigation scheme of Guédé, northern Senegal. Farmers' knowledge of recommended cropping practices was relatively poor. The surveys identified sub-optimal timing of N fertilizer application and late harvesting as major constraints. Before the start of the second season, farmers were given field-specific recommendations on timing of key crop-management interventions. Average yield compared with the first season increased by more than 1 t ha−1, which was attributed to a slightly higher recovery efficiency of fertilizer N (REN) (average: 0.47 kg crop N kg−1 fertilizer N), a change in cultivars used (resulting in a better uptake of indigenous soil N) and more favourable weather conditions in general. It is concluded that if farmers are given better access to information, improved rice technologies, inputs and decision making, rice production on irrigated land in West Africa may leap forward rapidly, as potential production gains are still large.

Population pressure has diminished the role of the traditional bush fallow system in restoring the fertility of depleted soils following several years of cultivation in the derived savanna of southwest Nigeria. The search for alternative fallow systems led to the development of alley cropping using Leucaena leucocephala and cover cropping using Pueraria phaseoloides. These soil-conserving systems also embody the principles of sustainability by ensuring stable crop yields in spite of an increase in land-use intensity. Simulating smallholder farm conditions and based on a split-plot experiment with the three fallow systems as the main plots and four different land-use intensities (LUI) as the subplots, simple direct observation and record keeping were used to collect data on labour use in all farm operations including fuelwood cutting and stacking in alley cropping systems with fallow phases. Data collection covered a period of four years (1993–96), corresponding to one out of the three phases (or complete cycles) of the trial. Partial budget analysis was used to estimate profitability of cassava (Manihot utilissima)–maize (Zea mays) under each LUI. Results based on crop production indicate that each of the cassava-maize enterprises was profitable but the level of profit varied from 10 819 to 50 289 Naira ha−1. Production under cover cropping has a net benefit advantage of 4–25% (over that under bush fallow) and 25–60% (over alley cropping). Profitability increased as the LUI decreased, depicting the overall importance of fallow periods on soil fertility restoration. Results based on crop production plus fuelwood operations indicate that production with cover cropping has a net benefit advantage over that of both bush fallow and alley cropping under both zero-fallow and one-year fallow LUIs. However, results from two- and three-year fallows indicate that fuelwood yield generated sufficient revenue to put alley cropping at a net benefit advantage of 29–42% over bush fallow and 26–37% over cover cropping. From these results, cover cropping is recommended, especially to farmers in areas where, although land shortage is a problem, fuelwood and the other sources of domestic energy are still cheap and easily accessible. However, in areas with abundant land and where farmers can still wait for fallow periods of at least two years, alley cropping is recommended since, in addition to soil fertility improvement, substantial revenue is obtained. However, considerations of yield variability, labour scarcity, as well as risk-aversion behaviour among farmers may alter their technology choice.