The CGIAR System conducts research to produce international public goods (IPG) that are of wide applicability, creating a scientific base which speeds and broadens local adaptive development. Integrated natural resources management (INRM) research is sometimes seen to be very location specific and consequently does not lend itself readily to the production of IPGs. In this paper we analyse ways in which strategic approaches to INRM research can have broad international applicability and serve as useful foundations for the development of locally adapted technologies. The paper describes the evolution of the IPG concept within the CGIAR and elaborates on five major types of IPGs that have been generated from a varied set of recent INRM research efforts. CGIAR networks have both strengths and weaknesses in INRM research and application, with enormous differences in relative research and development capacities, responsibilities and data access of its partners, making programme process evolution critical to acceptance and participation. Many of the lessons learnt regarding challenges and corresponding IPG research approaches are relevant to designing and managing future multi-scale, multi-locational, co-ordinated INRM programmes involving broad-based partnerships to address complex environmental and livelihood problems for development.

Three different saffron strains (Crocus sativus, C. sativus var. ‘cashmerianus’ and C. cartwrightianus) were tested for two years in field experiments to study their red stigma yield and quality. The experiments were performed at a high-elevation (rainfed or irrigated) site and a coastal site in Lebanon. On average, the two C. sativus strains gave more flowers and stigma yield than C. cartwrightianus. The former was more productive in the more favourable environments, especially at the warmer coastal site, than the latter. C. sativus ‘cashmerianus’, which yielded better at the coastal site in 2004 and gave a stronger colour strength and aroma, appeared to be the better choice for the area. Mean yield at the coastal site was twice that at the non-irrigated high-elevation site. On average, saffron produced at the coastal site had more colouring strength and bitterness than that produced at the high-elevation site. Nevertheless, viable commercial saffron production may still be possible in the semi-arid, high-elevation Bekaa Valley if irrigation can be provided.

Horsegram (Macrotyloma uniflorum) is sown late in the rainy season by resource-poor farmers in marginal, drought-prone areas of India. Sowing and early crop growth coincide with declining rainfall so crop establishment is often poor and yields are low. Horsegram is a ‘neglected’ crop and farmers' choice of varieties to grow is limited to poor-yielding landraces or to modern varieties for which access to seed is limited. In on-station trials, we compared the yield and relative merits over three years of seven varieties potentially suited to conditions in Jharkhand State and sown with or without priming by soaking seeds in water prior to sowing. Yield stability of the varieties was tested using data from these and five additional trials. Farmers' opinions were recorded during six participatory exercises based on farmers' trials, and their rankings were compared with those from the on-station evaluations. Significant variety × year interactions were observed and yield was generally inversely proportional to drought. Both linear and non-linear genotype × environment interactions were significant for all varieties. Variety BK 1 was the most desirable with stable yields across environments and a higher overall mean grain yield. Variety VLG 1 was unresponsive to more favourable environments but yielded more in the low yielding environments. However, it was the earliest to mature and was identified as a promising variety, which farmers preferred for its grain and brown seeds. This could be used to diversify farming systems with additional options for farmers. Although soaking seeds for 12 h was most effective in increasing germination in an in vitro study, soaking for 8 h before sowing increased emergence and final stand by 11 % and grain yield by 10 % averaged over all varieties and years. Soaking advanced flowering by 1.3 d and maturity by 1.9 d. All varieties responded positively to soaking and did so in all three years, although the response was stronger in drier years.

Arbuscular mycorrhizal fungi (AMF) may have a major role in phosphorus nutrition of crops in Lombok, where fertilizer use is low. As a start to understanding this role, AMF dynamics were monitored from the 1999 non-rice season to the end of the 1999/2000 rice season at 32 sites including dryland systems with no rice, upland rice and flooded systems with one or two rice crops per year in the rotation. Over all four systems, root colonization was greater in vertisol (22.3 % of roots) than in regosol (9.5 %) soil, possibly due to lower Bray-1 P content of the vertisol (6.2 v. 13.7 mg kg−1). Colonization was poor in flooded rice (3.1–5.1 %); at the same sampling times it was better in upland rice (10.6–13.4 %) and in non-rice crops growing in dryland systems (13.8–17.0 %). Therefore, the low colonization in flooded rice appeared to be the result of flooding, rather than the rice itself. Flooding also reduced transparent spore numbers, but sufficient inoculum appeared to survive flooding for plants in the following non-rice season to be well colonized (19–33 %) regardless of system. These non-flooded crops appear to replenish depleted AMF propagules.

Maize production in Malawi is limited by high costs and sub-optimal use of chemical fertilizers under continuous cultivation. A long-term gliricidia/maize trial was undertaken on a Ferric Lixisol from 1991/92 to 2001/02. The purpose of the study was to assess the performance of a gliricidia/maize intercropping system as a low-input soil fertility replenishment option in southern Malawi. The experiment was a 2 × 3 × 3 factorial design with three replications. Treatments included two maize cropping systems (with and without gliricidia trees), and three rates of inorganic N fertilizer (0, 24 and 48 N kg ha−1 representing 0, 25 and 50% of the national recommended N rate), and three rates of P fertilizer application (0, 20 and 40 P ha−1 representing 0, 50 and 100% of the recommended rate). No effect of P was detected on yield early in the trial, and this treatment was discontinued. The gliricidia pruning biomass did not decline after 10 years of intensive pruning, with strong correlation between tree biomass production and years after establishment (r = 0.91, p < 0.001). Application of gliricidia prunings increased maize yields by three times compared to the yield of unfertilized sole maize. Maize yield from the unfertilized gliricidia pruning treatment was superior to the yield from sole maize supplemented with a quarter or half the recommended N rate. The study confirmed that a gliricidia/maize intercropping system is a promising soil fertility replenishment option in southern Malawi and elsewhere in southern Africa.

Over 90 % of the farmers around Marsabit Mountain, a pocket of arable land amidst a predominantly arid area in northern Kenya, are former pastoralists, unfamiliar with productive and sustainable farming technologies. This paper describes how encouraging farmer innovativeness enhanced adoption and adaptation of water-saving technologies to suit individual farmers' needs. When farmers understood the principles behind a technology they were more likely to adapt it. Both men and women farmers were most interested in increasing their available income through improved technologies, but more women than men ranked labour saving as an important attribute of a new technology. More women adopted the new technologies, but men used them on a larger area than women. The paper also shows that agricultural productivity can substantially increase even if the theoretical full potential of the recommended practices might not be achieved. It is concluded that farmers need to be exposed to new ideas in exciting and thought provoking ways to tap their innovation potential.

A field study of cowpea, soybean and maize was made at Ibadan, Nigeria, to determine the uptake of mineral nutrients and quantities removed in crop residues and seed. Concentrations of N, the major cations (K, Ca, Mg and Na) and major anions (P, S and Cl), and the micronutrients (Fe, Mn, Zn and Cu) were determined in shoots harvested at flowering and maturity. Concentrations of excess cations in shoots at flowering were 135 cmolc kg−1 for soybean, 158 cmolc kg−1 for cowpea and 58 cmolc kg−1 for maize. In crop residues and seed at maturity, the values decreased to 75 and 56 cmolc kg−1 respectively for soybean, 68 and 27 cmolc kg−1 respectively for cowpea, and 52 and 10 cmolc kg−1 respectively for maize. Nutrient and excess cation concentrations were also measured in 13 soybean and eight cowpea lines grown in replicated field trials at Fashola in the derived savanna zone and Shika in the northern Guinea savanna zone of Nigeria to assess the variation within these species. Maize, mucuna and lablab were included at these sites. The results are discussed in the context of soil nutrient depletion due to removal of nutrient elements in harvested seed and stover. The contribution of seed and stover removal to the proton (H+) budget and acidification of the soil is also discussed. It is estimated that the quantity of lime required to neutralize the acidity resulting from the removal of 1 t seed is 28 kg for soybean, 14 kg for cowpea and 5 kg for maize. The corresponding values for stover were soybean 35, cowpea 24 and maize 26 kg lime t−1 stover. Given the soil pH buffer capacity at the Ibadan site for the surface 16 cm, the yields obtained and assuming that seed and crop residues are removed, carbon cycle acidification would lead to a fall of 1 pH unit after 75 crops for soybean, 114 crops for cowpea and 68 crops for maize. The inclusion of legumes in cropping systems to increase N supply, and the use of P fertilizer to maximize N-fixation by legumes will raise production, but will inevitably accelerate removal of alkalinity and nutrients in harvested products, and the consequent decline in soil fertility. The data presented in this paper can be used in estimating the lime and fertilizer nutrients needed to maintain soil fertility and sustain yields.

Mixed cropping of teff (Eragrostis tef) with faba bean (Vicia faba) was compared with sole cropping in the 2002 and 2003 growing seasons at Holetta Research Centre, in the central highlands of Ethiopia. The treatments were sole teff (25 kg seed ha−1), sole faba bean (200 kg seed ha−1) and 12.5, 25, 37.5, 50 and 62.5 % of the sole seed rate of faba bean mixed with a full teff seed rate. A randomized complete block design replicated four times was used. Treatment effects were significant for seed and biomass yields of each crop species. Increasing the seed rate of faba bean in teff/faba bean mixture increased faba bean seed yield but decreased teff grain yield. Nonetheless, mixed cropping of faba bean with teff increased land use efficiency and gave higher total yields compared to growing either species in sole culture. Teff yield equivalent, land equivalent ratios (LERs) and system productivity index (SPI) of the mixtures exceeded those of sole crops especially when the seed rate of faba bean in the mixture was increased to 50 kg ha−1 (25 %) or more. The relatively high crowding coefficient values indicated yield advantages from mixed cropping of the two species. The highest values of teff yield equivalent, LER and SPI were obtained when faba bean was mixed at a rate of 62.5 % with the full seed rate of teff. We suggest that, at the current prices of the respective crops, up to 62.5 % of faba bean can be mixed in normal teff to get better total yield and income than sole culture of either species.