Farming Systems Research (FSR) has emerged as a major theme in international agricultural research in the past decade. However, despite widespread use of the term, its meaning and the types of research objectives, approaches, activities and methods to which it applies remain ambiguous. The lack of precision in concepts and terms will discredit FSR and jeopardize donor support. This paper seeks to clarify and define the major types of research approaches and activities encompassed by the general term FSR. Six types of research are distinguished and a new, more precise terminology is proposed.

A late-maturing, high-yielding maize released for the Burundi highlands was rejected by subsistence farmers. Analysis of this variety in terms of the local cropping system shows that it offers no advantage over earlier maturing varieties when whole-system biological and economic yield and Land Equivalent Ratios are examined, while farmer risk is apparently substantially increased. The value of incorporating a ‘farming systems’ approach into national commodity research programmes in developing countries is demonstrated.

Maize grain yields were significantly higher following one to three years of Stylosanthes than after natural fallow or repeated cropping. This demonstrates an important benefit of the forage legume in addition to its value as a ruminant feed supplement. The implications for the development of traditional agro-pastoral production systems are discussed.

Farming systems research (FSR) has come to prominence as a part of tropical agricultural research in the past decade or so. The main stimulus was the growing realization that small farmers are ‘economically rational’ and agricultural economists have been the leading agents in developing FSR ideas. Three broad aspects are distinguished: (1) FSR sensu stricto, the deep analysis of farming systems as they exist, which is essentially an academic activity, (2) OFR/FSP (‘on farm research with farming systems perspective’), a practical adjunct to research which seeks to test the socio-economic suitability of research ideas on-farm before recommending extension and (3) NFSD (‘new farming systems development’) which seeks to develop complex, radical change rather than the stepwise change characteristic of OFR/FSP. The technical components of NFSD are probably mostly available but the necessary transdisciplinary, synthetic thinking is not. Research on perennial plants in tropical agriculture has been rather neglected but there are signs of growing concern; they must be an important feature of any NFSD for the wetter areas.

In monsoon climates, the mean daytime saturation vapour pressure deficit (D), measured at screen height, decreases from a maximum of 3–4 kPa attained several months before rain arrives to a minimum of 0.5–1 kPa in the month of highest rainfall. Climatic records from India and West Africa were analysed to give the relation between D and precipitation (P mm month−1) as:

where n ranged from 0.5 to 0.9 between stations.

The dependence of dry matter production on D and therefore on P is evaluated when growth is restricted by rain and when this restriction is removed by irrigation. In the first case, the decrease in demand for water associated with a decrease in D is comparable in importance with the increase of water supply (P). The analysis has implications for the marginal response of crops to supplemental irrigation and for the interpretation of experiments with a line-source or rain-shelter.

The traditional management of cattle manure in a cropping system in the savanna zone of central Nigeria is discussed in conjunction with the agronomic benefits and problems of manure use. Nitrogen and phosphorus contents of the manure varied seasonally. On-farm trials showed that maize grain yields were about 1 t ha−1 more and weed growth 90% greater in manured than in non-manured areas. It is important to reduce the competition between crops and weeds if grain crops are to obtain the full benefit of manure in such cropping systems.

A computer technique for predicting the yields of the component crops in intercropping systems is described. Predictions using this procedure showed reasonable agreement with results obtained in an experiment using tomatoes and dry beans. Reasons for the discrepancies observed and limitations to the applications of the technique are discussed.

The study was carried out in an unheated glasshouse under natural conditions of daylength and temperature. Rates of pod and seed growth and development, and of desiccation at maturity, are reported in single genotypes of Lupinus albus, Lupinus mutabilis, Vicia faba, Pisum sativum and Lathyrus latifolius. The histology of tissue layers in pods are also described. The culti-var of Pisum sativum completed dry weight accumulation in pods in 20 days from anthesis; the equivalent time in Lupinus albus (Kievskij mutant) was 60–70 days and in genotypes of Lupinus mutabilis and Vicia faba about 50 days. The heavier pods and seeds of the L. albus and V. faba cultivars were due to a prolonged rapid growth phase: rates of weight accumulation in seeds during the rapid growth phase were similar for all cultivars. The histology of the pod wall varied between species but the rate of dehydration of pods was not a factor in determining date of maturity. The results are discussed in the light of reservations based on known intra-specific variation in crop species.

A simple yet effective field screening technique for estimating drought tolerance is described. It involves growing plants in sloping plots which are opposite each other and connected to sub-channels lined with polyethylene sheet. The slopes are designed to achieve instant surface run-off and the sub-channels connected to a main channel for rapid drainage of rain-water. Stress is created at different stages of crop growth by manipulating the timing of irrigation and covering the soil surface with polyethylene sheeting. The technique proved effective even in the monsoon season which is characterized by irregular intensity and distribution of rainfall. Data on six characters for 80 genotypes of pearl millet grown in six artificially created environments representing different levels of moisture stress are discussed. The ‘slope’ technique was effective in creating different levels of moisture stress at various stages of crop growth and in distinguishing categories of genotypes by their relative drought tolerance.

Experiments comparing the effects of different legume intercrops on available soil nitrogen, bacterial activity and the yield of maize over two years showed that soyabean and blackgram were more suitable for this purpose than groundnut, substantially increasing NO3 and NH4 concentrations and populations of active bacteria in the maize rhizosphere. This increased maize yield by 15–20% and grain protein content by 20%.

Differences in the growth and development of five genotypes of chickpea (Cicer arietinum) were studied until the three branch stage (18 or 21 days after sowing). Seedlings were grown in nutrient solution in clear Perspex cells immersed in temperature controlled water baths. Daily removal and photographing allowed serial measurements of the rate of root extension to be made. An accurate, detailed description of the rooting pattern was obtained by sectioning along the primary axis after harvest. Significant differences between the genotypes in root and shoot growth were recorded. The Syrian local landrace, ILC 1929, and the Ascochyta blight resistant genotype, ILC 428, had faster rates of extentions of the primary root axis and of lateral root development than the other genotypes. The possible significance of these results to subsequent growth and yield in Mediterranean environments is discussed.

Pods of vegetable cowpea Vigna unguiculata Walp. cv. Dinner were harvested at various developmental stages and their biochemical components and other characteristics analysed in two years. Pod length, percentage dry matter and starch, fresh and dry weights and total yield of sugar and starch per unit area increased with age. The percentage total yield of sugars was greatest in pods aged between 11 and 13 days. Vitamin C, total acidity, crude fibre and crude protein percentage decreased with age. Content of minerals except iron also decreased with age of pods. Magnesium, phosphorus, potassium, manganese and copper contents were greater in 1977 than in 1979 but iron and calcium contents were smaller. The vegetable cowpea pods were a rich source of calcium, magnesium, potassium and iron. Pods harvested between 7 and 10 days after flowering were crisply tasty and high in nutrients and therefore nutritionally the most acceptable for consumption.

Improved groundnut production practices are compared with traditional methods. Improved sole crop production practices show significantly greater net returns per hectare than traditional methods but when other considerations such as the riskiness of the various enterprises are examined, the superiority of the improved sole crop technology becomes less clear-cut. A case is made for designing improved technologies involving crop mixtures.

Mixtures of legumes and cereals in various proportions (seed ratios) were evaluated in Northern Syria for forage production and quality and for yield of barley grain in the following year. Both a vetch-cereal and pea-cereal mix were tested in proportions of 0:100, 33:66, 50:50, 66:33 and 100:0. The study was carried out between 1982 and 1985, covering two seasons for forage production and two for grain production. Dry matter yields of the mixtures exceeded the yield of either one or both components of the mixture grown as pure stands, depending on the rainfall. The maximum overall yield and quality of forage was recorded for the 66:33 legume-cereal combination. The greatest benefit to subsequent barley grain yield was realized when the legume was grown in a pure stand or formed a high proportion (66:33) of the mixture. Herbage quality was better in forages containing vetch than in those containing pea.

The effect of strain of Rhizobium and of temperature on nodule formation and early growth of Trifolium semipilosum was studied in controlled environment cabinets. Time required for nodules to form was affected by strain of Rhizobium and by root temperature, acting independently. Strains CB788 and CB2116 required only 9.5 to 10.0 days, averaging over temperatures, whereas strains CB782 and CB526 required 12.8 to 13.0 days. Nodulation was earliest at whole-plant temperatures of 25/20°C and 30/25°C day/night (average 9.5 days); 15 days were required at 20/15°C. Nodule number was independent of bacterial strain but dependent on temperature; nodules were fewest in the lowest and most in the highest temperature regime. Plant yield was determined by interactions between temperature and strain and was generally greatest at whole plant temperatures of 25/20°C.

When nodulated plants growing at 25°C were transferred to a range of root temperatures, plants grew best at 20° and 25°C and with strains CB782 and CB2116.

Field trials are described in which the seed yield response of row-sown sesame to increasing plant population was examined, and in which row-sowing was compared with broadcasting at different seed rates. Maximum yields were obtained at an intended 200 to 250 × 103 plants ha−1 but yields were not greatly affected by a wider range of populations. Broadcasting, the normal local practice, gave yields slightly higher than those from rows 50 cm apart, the un-thinned rate of 800 × 103 viable seeds sown ha−1 yielded as well as treatments thinned to the optimum population, and double this seed rate, unthinned, gave the lowest yields, especially in rows.

Field trials are described in which sesame varieties maturing in 125 days at 2 m high, were grown in alternate rows with sorghum maturing in 175 days at 4 m. Compared to sowing both crops together, the yield of sesame sown 10, 14 and 28 days after sorghum fell by 53, 68 and 100%, respectively, but sorghum yield was not consistently affected. Improved sesame varieties, despite being shorter and less branched, performed better than the local variety, and nitrogen fertilizer applied to one or both components of the mixture did not appear to favour one crop at the expense of the other.

The calibrating efficiency of the pre-experimental yield of coconuts was examined using ten years data from a calibration experiment. On the basis of a fully randomized design it was found that the two-year pooled pre-experimental yield on four-tree plots produces consistent calibration and reduces the experimental error mean square by about 73%. This brings down the mean coefficient of variation to 9.7% from its pre-calibration levels of 36 on one-tree plots and 18 on four-tree plots.

Changes in plant dry weight and nitrogen content of Trifolium semipilosum cv. Safari and T. repens cv. Grasslands Huia were monitored when the root systems of effectively nodulated 28-day-old plants were exposed to a range of constant and diurnal temperatures for 21 days. Nitrogen fixation was more sensitive to high root temperatures than was dry weight accumulation, and T. semipilosum was relatively more tolerant of high root temperatures than T. repens for both dry weight and nitrogen content. The optimum temperatures for dry weight yield and nitrogen fixation (≡nitrogen content) were similar (21–23°C) for both species.

Our data suggest that growth and nitrogen fixation in T. semipilosum are more tolerant of short-term exposure to increase in root temperature than in T. repens. In addition, both species accumulated more dry matter and fixed more nitrogen when night temperatures were reduced from either 30 or 35°C to 25°C. The increase was greater with T. semipilosum and may be a major factor in its adaptation to the humid subtropics and cooler (elevated) tropics.

The future evolutionary potential of pulses is determined by the nature and extent of the genetic variability in the primary gene pool. This is extensive in the major grain legumes, notably the groundnut, soyabean, Phaseolus and faba beans, and in the cowpea. It is less extensive in the cultivated lupins and this appears to be a serious limitation to their development as useful crops. The considerable development of isolating mechanisms between related species of legumes has restricted the development of the secondary gene pool. This is most developed in the genus Arachis, where section Arachis provides an extensive secondary gene pool for the groundnut. Tertiary gene pools are potentially quite extensive for many grain legumes but since most interspecific hybrids are inviable this resource would be difficult to exploit with present techniques. The development of sophisticated gene transfer techniques for grain legumes is inhibited by their lack of amenability to in vitro culture. The suggestion is made that genetic resources profiles could be constructed on the basis of an expansion of Harlan and de Wet's gene pool system. These could serve as a guide to the present state of germplasm collections, indicating their strengths and weaknesses, which would be useful in formulating future collection and evaluation strategies.

The potential for more intensive exploitation of such legumes as the winged bean and lupins undoubtedly exists. The reasons for past under-exploitation of the winged bean need to be determined.

Grain legumes have an assured future for good economic and nutritional reasons. More attention probably should be devoted to exploiting the legume-Rhizobium symbiosis, fundamental to the development of efficient and economic farming systems in the developing world.