Farming systems in the wetter tropics may be divided into three simplified models: the irrigated paddy-rice multistorey homestead garden complex of Asia, the tree and cash crop plantations of Latin America, and the mixed root crop-bush fallow systems of Africa. In upland ecosystems, sustainable farming systems development rests on a delicate balance between conservation and utilization. To meet the basic food and nutritional needs of the indigenous population, there is great potential for the improvement and adaptation of multistorey homestead gardens and mixed systems which include trees and annual and perennial crops. The principle and practice of ‘alley cropping’ provide an ecologically sound basis for future farming systems research and development in the region.

Additivity, variance homogeneity, and normality are often regarded as prerequisites for analysis of variance. Although they are desirable, their truth is never certain; they are important to probability levels in significance tests, but not to the general informativeness of the analysis. Transformation of data may improve conformity to these desiderata at the cost of producing results on a scale ill-suited to the objectives of the experiment. In this paper the generalized power transformation and the inverse sine transformation are described, and the practical problems of choosing an optimal transformation discussed. An ideal choice can seldom be based on the evidence of one experiment, and adopting the same transformation for a series of similar experiments is usually more helpful than a search for the ‘best’. Warnings are given about inappropriate guides to the choice of transformation. In some experiments, separate analyses for two or more obviously distinct sets of treatments may be more rewarding than any attempt to find a transformation that will enable all to be handled at once. When one or two treatments (or treatment combinations in a factorial experiment) show data wildly out of line with all others, consideration of reasons should precede and possibly replace any approach through transformation. A sound general principle is that data are usually most clearly interpreted on the original scale of measurement. Therefore, transformations should be avoided unless clearly necessary; by this advice, I do not imply prohibition. Finally, attention must be given to the presentation of results; the most effective presentation does not necessarily employ a transformation or other special device adopted earlier in a statistical analysis.

Paddy rice was grown without fertilizer after jute which had been supplied with nitrogen as neem cake coated urea (NCU), wax coated urea (WCU), ammonium sulphate or standard urea at rates of 0, 40, 80 or 120 kg N ha−1. NCU or WCU applied to jute increased the dry matter and grain yield of the rice, suggesting that enough nitrogen is left behind to grow a subsequent rice crop without further fertilizer application.

Seed yam production potential of three yam species (Dioscorea rotundata, Dioscorea alata and Dioscorea cayenensis) was assessed for two systems – the traditional method using 100 to 200 g setts, and the mini-sett technique using 25, 20, 15 and 10 g setts cut from mother tubers. The mini-sett technique required only 6–33% of the number of tubers needed for the traditional planting system, and resulted in plants superior to those produced by the traditional method in all their production attributes especially in the case of D. alata and D. rotundata.

Three irrigation regimes (250, 450, 650 mm year−1) were applied to guayule (Parthenium argentatum), a rubber-producing plant, in such a way as to provide periods of good soil water status alternating with periods of water stress. The relative water content and transpiration rate of the plants decreased only when the available soil water was depleted to about 40%. Very low values (< 30%) were measured for the relative water content. Water-stressed plants recovered as soon as irrigation was resumed. Growth and biomass accumulation, as well as the yields of rubber and resins, were directly related to the irrigation rates. The water regimes applied and the use of the drip irrigation method resulted in a relatively high water use efficiency.

In a field experiment, the effects of irrigating crops alternately with sodic water (high in sodium adsorption ratio and ) and good quality canal water were investigated for six years on a well drained sandy loam (Typic Ustochrept). The irrigation treatments included: irrigation with non-sodic canal water (CW), irrigation with sodic water (SW), CW irrigation alternating with one or two SW irrigations, and two CW irrigations alternating with one SW irrigation. The results showed that the use of sodic water increased the sodium saturation of the soil and decreased rice and wheat yields. The build-up of sodium depended on the number of SW irrigations during the season. The increase in sodium saturation and decline in crop yields were progressive over the years. The improvements in yield due to alternating sodic and non-sodic irrigations compared with the use of sodic water alone increased over the years. Alternating sodic and non-sodic irrigations could therefore be considered a practical way to alleviate the problems caused by sodic water. The number of sodic irrigations during a season should, however, be kept to a minimum and the build-up of sodium in the soil over time should be monitored.

Continuous cropping of a tropical Ultisol with or without nitrogen fertilizer increased soil acidification and reduced soil organic matter and productivity. Addition of nitrogen fertilizer failed to increase yields. This was apparently due to the decline in soil pH, an increase in the amount of soluble aluminium, and consequent inefficient utilization of added nutrients. Continuous use of ammonium sulphate as a nitrogen source hastened the development of both surface and subsoil exchange acidity.

Results of farm-level experiments with pearl millet are reported from four villages in semi-arid Niger in 1982 and 1983. Crop establishment, yield loss factors (insects, diseases, striga) and economics of new practices were studied in a randomized design which included density, fertility, cultivar, and intercropping with cowpea as treatments. Improved cultivars did not establish significantly better than local ones, but fertilizer occasionally improved establishment. Incidence of yield loss factors was low, and only small effects of cultivar and fertilizer were seen. Fertilizer increased grain yields of the local cultivar by as much as 153%. The use of fertilizer and improved cultivars increased grain yields by as much as 171% over an unfertilized local millet. Even without government subsidies, fertilizer use was profitable for 56% of farmers on all cultivars. In general, fertilizers and improved cultivars had a small positive impact on profitability, with little adverse impact on grain yield variability, so that they could be recommended to farmers with some prospect of success. The intercropping treatment failed in both years, and would have to be modified before extension.

The effects of coffee berry disease (CBD) on yield are usually estimated by counting infected berries on marked branches throughout fruit development. This method is impractical when surveys are required of yield losses over large areas of extensively cultivated peasant coffee. A survey method is described for estimating yield potentials and the loss in yields due to severe CBD infection under such conditions. The method exploits the relation between numbers of flowering nodes and yield per tree, and requires a single subjective estimate of disease incidence on young expanding berries. Yield losses are estimated from the changes in this relation resulting from berry infection. Results are presented for peasant coffee at different stages of rehabilitation in Ethiopia. They indicate that the method is suitable for estimating.production in agronomic improvement projects and for identifying areas where CBD incidence might economically justify fungicide spraying.

Dry matter production in a bushy and spreading cultivar of sweet potato was studied in two trials comparing dry matter partitioning between plant components. The bushy cultivar, VSP-2, accumulated more tuber dry matter than the spreading cultivar, BNAS-51. The number and yield of tubers of VSP-2, was also greater but its foliage dry weight was less than that of BNAS-51. The vine and root dry weight of BNAS-51 increased with time but that of VSP-2 reached a peak about 10–12 weeks after planting. The tuber dry weight of BNAS-51 at harvest was similar to its foliage dry weight but that of VSP-2 was three times greater. The proportion of total dry matter partitioned to the tubers of VSP-2 was greater than that in BNAS-51 at all stages of growth.

Root growth of two drought tolerant and two drought sensitive bean (Phaseolus vulgaris L.) lines was compared at two locations in Colombia differing primarily in soil conditions. At Palmira, roots of drought tolerant lines reached a depth of 1.3 m, while drought sensitive lines only reached 0.8 m. These differences were associated with differences in seed yield, crop growth, canopy temperature and soil moisture extraction. Under acid soil conditions at Quili-chao, seed yields of supposedly drought tolerant genotypes were similar to those of the drought sensitive lines, and root growth of all four lines was restricted to less than 0.8 m. Drought avoidance through greater root growth and extraction of soil moisture appears to be an important drought tolerance mechanism in common beans, but its usefulness is limited where soil conditions restrict root growth.

In many experiments which compare a large number of different treatments or varieties, it is necessary to decide which sub-sets of treatments or varieties do not differ significantly from each other. Multiple comparison procedures provide a commonly used, but much criticized, way of tackling this problem. As an alternative, techniques known collectively as cluster analysis can be used. An advantage of using certain types of cluster analysis is that their results can be displayed as a dendrogram or tree diagram which gives a very useful picture of the overall relationship between treatments or varieties. These procedures can also be extended readily to multivariate observations from each plot.