There is a need to shorten the selection process for sugarcane (Saccharum hybrids) in Mauritius in order to improve the efficiency of the varietal improvement programme. On average six to seven ratoon crops are grown in Mauritius and selection for ratooning ability is of major importance. The current selection cycle lasts around 15 years and ratooning ability is tested on five occasions. Data (estimates of sucrose content, cane and sugar yields) from the Mauritius Sugar Industry Research Institute's selection trials planted in 1986–90, representing 85 trials (9680 genotypes) in Stage 3 (one-line stage) and 141 trials (2620 genotypes) in Stage 4 (two-line stage) were analysed. It was hypothesized that 1st ratoon (1R) data in Stage 3 and 2nd ratoon (2R) in Stage 4 were effectively redundant and that the cycle could be shortened by two years without loss of precision. Repeatability estimates, between plant cane (P) and the mean of P and 1R (P + 1R) in Stage 3, and between (P + 1R) and the mean of plant cane, 1st and 2nd ratoon (P + 1R + 2R) in Stage 4, were found to be positive and highly significant for all characters indicating that the extra ratoon data were unnecessary.

Present and proposed selection scenarios were studied further using differential-selection methodology. The coincidence indices (CI) obtained with the two scenarios were extremely high (63–82% in Stage 3 and 91–96% in Stage 4) again indicating that the additional ratoon data were not cost effective. Realized gains from selection in Stage 4 based on (P + 1R + 2R) as opposed to (P + 1R) were shown to be small or even negative. The results concur extremely well with published data from Australia and the USA. It was therefore recommended that the cycle be reduced by two years and that the resources be more usefully allocated to test genotypes over more sites and more replicates.

Two series of trials were initiated to compare different lengths of crop cycles for sugarcane harvested every 12 months under the local conditions over a six-year period. The first series involved harvesting cane every 24 months and in the second series cane was harvested after 16 or 20 months alternately. All trials included four cane varieties except one in the second series where only three varieties were planted. Cumulative results for both series of trials showed a difference in favour of the 12-month cane crop. In the first series after six years, cane and sugar yields were reduced by 19–48% and 22–60% respectively in the 24-month-old canes compared with 12-month-old crops. The level of reduction varied with cane variety. In the second series, the declines in cane and sugar yields were smaller, 5–44% and 7–49% respectively. Cane lodging, which increased with length of crop cycle, led to greater weed infestation, increased damage by rats and rendered manual harvest more difficult.

One of the major research activities of the Mauritius Sugar Industry Research Institute (MSIRI) is the breeding and selection of sugarcane varieties adapted to the different agro-climatic environments. Most of the on-farm experimentation is carried out in the fields of miller-planters (planters who own the sugar mills), where resources in terms of land, labour and other inputs are usually easily available. The non-miller-planters are operating in environments which are not always similar to those of the miller-planters. There is a need for on-farm experimentation under these conditions in order to improve productivity. Variety trials, observation plots and industrial trials to compare commercial varieties with promising and newly released ones have been established on-farm in order to improve the adoption rate of new varieties. The paper elaborates on the objectives of this major extension activity during the last decade. A major inference is that the planters obtain first-hand information on promising varieties prior to their release for commercial plantation, hence influencing their decision for adoption or rejection. Moreover, additional information on the performance and behaviour of these promising varieties is made available to research staff.

Crop residue management is posing a serious problem in the rice (Oryza sativa)–wheat (Triticum aestivum) cropping system, which is widely practised in the Indian Subcontinent and China, and covers about 22.5 × 106 ha. The problem is serious because there is very little turn-around time between rice harvest and wheat sowing. Three practices, namely, residue removal, residue burning and residue incorporation were compared in two field experiments, one with the rice residues and the other with the wheat residues. Results obtained showed that both rice and wheat residues can be safely incorporated without any detrimental effects on the crops of rice or wheat grown immediately after incorporation. Incorporation of crop residue also improved soil fertility status as judged by organic carbon and available phosphorus and potassium contents. Residue incorporation should be preferred over residue burning, which results in the loss of valuable plant nutrients and is both an environmental and a health hazard.

The main objective of this study was to determine the best soil amendment and tillage practices for sustainable millet yield and grain and stover quality. The treatments included tillage practices (immediate-, late- and no-till) and soil amendments (sheep manure plus urine, manure, millet stover (stalks, leaf blades and leaf sheaths) and millet stover ash) in factorial combinations with fertilizer nitrogen levels of 0, 15 and 30 kg ha−1 plus controls. Results showed that (i) higher yields were obtained in tilled plots than in no-till plots; (ii) tillage timing may not be a significant yield determining factor; (iii) the application of animal urine resulted in significantly higher yield and greater nitrogen, phosphorus and potassium uptake than the application either of manure alone or of millet stover. Urine application (ruminant urine contains virtually no phosphorus), which elevates soil pH especially during the first week after application, may have resulted in the dissolution of phosphorus from the aluminium–iron complexes of kaolinitic clays. This is corroborated by the significantly higher phosphorus uptake from the manure-plus-urine plots than from plots amended with either manure alone or millet stover. Long-term implications of ‘mining’ soil phosphorus with repeated applications of animal urine in these fragile ecosystems remains unclear.

Nepalese hillside terraces are classified into two broad groups, sloping or bari terraces which are only rainfed, and level or khet terraces which are commonly flood irrigated but may also be rainfed. Although such terraces may be adjacent, the process of irrigation by flooding and puddling changes the pattern of variability both within and between terraces. Data collected from Pakhribas Agricultural Centre, Dhankuta, Nepal, were used to describe this variability. The variability is manifest not only in the soil characteristics of the two terrace types, with khet terraces being more uniform than bari terraces both within and between terraces, but also in the variability of field experiments on the two terrace types. However, both terrace types exhibit similar patterns of variability which influence the design of experiments. These include the following: a gradient between the terrace riser and wall which is generally more important than gradients along the terrace, variability within terraces which can preclude the assumption that plots on the same terrace are similar, and variability between terraces which requires experimental units in the same block to be assigned to the same terrace.

Ten planting dates were selected at weekly intervals for the purpose of determining the optimum planting date for two late maturing sorghum (Sorghum bicolor) cultivars in the 1995 and 1996 cropping seasons. Both the floral characteristics and the ultimate grain yield were affected by variation in planting date. The highest grain yield for each cultivar (977 and 958 kg ha−1 for Janare and Farin Dawa respectively in 1995, and 975 and 955 kg ha−1 respectively in 1996) resulted from planting delayed for exactly 51 days after the onset of the rains. Consequently, 10–12 June appeared to be the optimum planting dates for sorghum in the area. A comparison of the result with previous work done in the same ecological unit revealed that delaying planting until 10–12 June increased grain yield by about 25–27 kg ha−1. The effects of the characteristics of the onset of the rains on the optimum planting date have been discussed and the implication of the study for traditional agricultural practice in Nigeria has also been noted.

Participatory varietal selection in a development project in western India showed that the rice (Oryza sativa) variety Kalinga III was highly preferred by farmers. The spatial diffusion of this variety from three villages (two project and one non-project) was studied. Seed of Kalinga III had spread from the three villages in 1994 to 41 villages by 1996 and is estimated to have reached over 100 widely distributed villages by 1997. Farmer-to-farmer spread was as high from the non-project case study village that received no further seed from the project, possibly because farmers are more likely to spread seed of a new variety to other farmers when they have no assured supply. Project interventions used key villages, informal-sector seed merchants, and Non-Government Organizations in the spread of seed. The project also collaborated with Rajasthan State Agricultural University and Kalinga III has been proposed for release in that state. A financial analysis revealed the very high internal rates of return that are possible from investment in participatory varietal selection.

A series of experiments on communally-owned grasslands in the barley–livestock zone of north Syria were conducted to test the hypothesis that introduction of Mediterranean annual legumes will increase productivity. The experiments were preceded by a survey to determine farmers' attitudes, describe the farming systems and to select appropriate collaborators. The first experiment examined the establishment of medics (Medicago spp.) and clovers (Trifolium spp.) distributed by hand, and monitored their effects on biomass and seedbank size. Later experiments extended these results to other villages. The principles of farmer participation in research were used to overcome the constraints imposed by the communal ownership of land. 

The survey revealed that the average size of the 20 villages was 36 families and that each village owned 887 sheep and 790 ha land. All villages had access to communally-owned grasslands, although their dependence on income from sheep varied greatly. These villages were subsequently divided into groups of high, intermediate and low potential.

Of the 11 clovers sown in the first experiment, seed numbers of T. tomentosum, T. purpureum, T. haussknechtii, T. pilulare and T. resupinatum increased over three years. By 1996, there were more than 3000 legume seeds m−2 in the seeded treatment compared with less than 2000 in the unseeded treatment (mainly the naturally-occurring Trigonella monspeliaca). The number of medic and clover seedlings also increased significantly, while the number of Trigonella seedlings decreased significantly. Biomass production increased in the final two years and there was no response to added phosphorus.

There were similar results in the later experiments. Seedbank size was greater in seeded treatments than in unseeded treatments, there were more seedlings in the seeded treatments, and the most successful species were T. campestre, T. tomentosum, T. speciosum and M. rigidula. The response in biomass was limited to the legume component, although total biomass increased in at least one of the two years. The highest biomass produced was 1112 kg ha−1 and there was no response to added phosphorus.

The results suggested that the on-station research previously conducted at ICARDA headquarters was applicable to communally-owned land, although important modifications were needed. For example, at ICARDA phosphorus was necessary to stimulate the growth of legumes; in contrast, it was necessary to sow legumes at the four villages involved in these experiments. The results also suggested that the grasslands were common property, owned and controlled by defined groups of farmers.