This thesis investigates options for sustainable rice cultivation and general agricultural development in the Mwanza and Shinyanga regions in northwestern Tanzania, often called Sukumaland due to the predominance of Wasukuma people. Generally Sukumaland has a semi-arid climate; agriculture is constrained by unreliable and low rainfall. In the past fifty years the population density has doubled in most parts. This has triggered several changes in farming systems. One important change is a reduction of grasslands in the valleys, due to increased cultivation of rice. Rice cultivation systems in Sukumaland serve here as a case study that allows the investigation of the interplay of social, economic and biophysical sustainability factors at field, farm, watershed and regional/national levels and their importance to the development of sustainable agriculture. Because soil fertility management is currently more important to sustainable rice cultivation in Sukumaland than water use efficiency or pest and disease susceptibility, it is the focus of the investigation. Economic reform programmes known as structural adjustment programmes started in Tanzania around 1986, guided by the International Monetary Fund (IMF) and the World Bank. These programmes required drastic changes in Tanzanian national economic policies and had great impact on the marketing of agricultural outputs and inputs. Liberalized markets and private traders were expected to improve the agricultural sector via a much needed intensification of agriculture, involving higher consumption of mineral fertilizers, herbicides and pesticides. However, the `liberalization' of international agricultural trade provided only a limited increase in access to developed country (DC) markets for less developed countries (LDCs) like Tanzania, and included few restrictions on the dumping of agricultural products by DCs. Farmers in LDCs cannot compete with farmers in DCs, and this lack of market opportunities, in combination with low agricultural prices and the low purchasing power of LDC consumers, pose major constraints on LDC food security. Specific data for Tanzania show that for this country the per capita food production increased in the 1970s, stabilized in the 1980s and started to decline in the 1990s. From a national point of view this is obviously not sustainable agricultural development. The gradual process of soil nutrient depletion in many parts of sub-Saharan Africa (SSA) is thought to be a reason, though rather hidden, for slow agricultural growth in the face of high increases in population. An integrated nutrient management (INM) strategy, which combines the use of locally available resources with the use of external inputs and includes both management practices to save nutrients from being lost from the system and interventions to add nutrients from outside, has been advocated to increase production and develop sustainable agriculture in SSA. INM methods are one of the important strategies of low external input and sustainable agriculture (LEISA) policies. The LEISA approach is aimed at making optimal use of local available resources, adding limited external inputs and using them in the most efficient way. In managing soil fertility, low inputs of mineral fertilizers must be combined with applications of farmyard manure and, where applicable, green manures, compost, agroforestry and erosion control. The expected result is sustainable increases in production. The objective of this thesis is to evaluate whether it is feasible for farm households in the Sukumaland rice cultivation systems to adopt INM and LEISA, given the current economic climate fostered by structural adjustment programmes. More generally, whether farm households in SSA countries can adopt INM/LEISA as a way of generating sustainable agricultural development in the context of liberalization of international agricultural trade, structural adjustment programmes, and fast population growth, is examined. It must be noted that in many locations within Sukumaland, cassava, maize or sorghum are more important food crops than rice; cotton, maize or horticultural crops can be more important cash crops than rice. Just as elsewhere, sustainable agricultural development in Sukumaland depends on the performance of all cropping systems present, and on the interactions of these cropping systems with livestock and other subsystems at the farm level. When performance in the food and cash crop sectors of Tanzania and the availability and consumption of agricultural inputs in 1986-1996 are compared with periods prior to IMF/World Bank backed reform, the positive developments of the first five years of reform appear not to be sustainable. At present, rural productivity levels per capita for important food and cash crops are declining. High increases in fertilizer prices and input availability problems in the villages are related to liberalization of agricultural input supply and pricing. The removal of subsidies on agricultural inputs from 1991 onwards is crucial in explaining the decline in production of maize, the main food crop in Tanzania. Structural adjustment programmes usually include far reaching reductions in the role of government. However, adequate government involvement may be necessary to ensure greater use of agricultural inputs and thus improved performance of the agricultural sector in Tanzania. In Sukumaland, historical material makes it possible to put current conditions in an historical context. A description of farming systems dynamics in Sukumaland over the past 100 years shows that people were attracted to the area by its low incidence of human and animal diseases. Cultivation was first restricted to the sandy upper parts of slopes because they are easy to prepare by hand hoe. Increases in population density led to enlargements in farm size directed towards the lower parts of slopes. Rice cultivation in these areas reduced grazing space for livestock. Households with large herds migrated to new, tsetse free areas; large farms were made possible by the availability of ploughs and oxen. In the 1950s and 1960s there were strong increases in household income in Sukumaland due to extensive, financially rewarding cotton growing. In the 1970s and 1980s this became much less profitable, leading to diversification in cash crops in accord with agroecological variations and distance to Mwanza town. Ongoing increases in population density caused a decrease in arable land and livestock units per capita, plus shifts in crops grown. Less demanding (cassava) and higher yield (rice, maize, cassava, sweet potatoes) crops were substituted for the traditional crops (sorghum, bulrush millet). These developments varied across the Mwanza and Shinyanga regions, due to differences in population density and agroecological conditions. At present there are signs of agricultural intensification near Mwanza town, while extensive farming dominates in the remaining parts of Sukumaland. Recent agricultural surveys conducted in Sukumaland have drawn attention to the importance of rainfed, lowland rice in the farming systems studied. More than a third of rice produced in Tanzania comes from Sukumaland. Farmers increased their rice production quickly when rice cultivation became more profitable in comparison to cotton and other crops, as well as more popular as a food crop because it can produce high amounts of calories on small pieces of land. The strong increase in rice cultivation during the last 25 years is remarkable, given the low and highly unreliable rainfall in Sukumaland. Farmers have developed highly productive rainfed, lowland rice systems solely on the basis of their knowledge of soils, rainfall patterns and topography, and on their experiments with water management systems, cultivars, and planting and land preparation methods. Rice management practices closely follow differences in ecology and household characteristics. Selection of rice cultivars is largely determined by water conditions in the field. The cultivation of rice is more intensive in Mwanza region, where transplanting takes place; on the larger rice fields in Shinyanga region, broadcasting dominates. Households grow rice for both food and cash - mainly for food in Mwanza region and mainly for cash in Shinyanga region. Water and weeds are the major production constraints, but low soil fertility is also a problem on the sandier fields of Mwanza region. Yields have declined due to continuous cultivation, with almost no application of organic or mineral fertilizers. On the more clayey rice fields in Shinyanga region yields are, however, still satisfactory at present due to the relatively short period (10-20 years) of cultivation. However, in one out of three years farmers fail to get food and cash from rice due to insufficient, too late or too unreliable rainfall. In response to farmers' complaints about declining rice yields, on-farm soil fertility research was carried out in the rice fields of Sukumaland between 1990 and 1996, using a Farming Systems Research/Extension (FSRE) methodology. The decline was thought to be related to a decrease in soil fertility. On-farm research showed that broadcasting 30 kg N ha -1in the form of urea in wet rice fields at the tillering stage increased rice grain yields by 500-900 kg ha -1in almost every type of rice field cultivated in Sukumaland. Doses higher than 30 kg N ha -1were less economical at 1996 prices for crops and fertilizers. The crop yield response to urea was better when rice plants were at the maximum tillering stage, when water depths were less than 15 cm at application, and when the sand content of fields was higher. The relatively small differences each year in response per field did not justify multiple extension messages. A single dose of 30 kg N ha -1in the form of urea to rice at tillering was thus recommended for Sukumaland as a whole. Despite the relatively high average productivity index for a low dose of urea in rice, there was almost no adoption by farmers. The main factors were problems in the availability of urea in villages and decreasing profitability of the rice-urea technology, due to IMF/World Bank instigated reform measures. Non-adoption was also due to absence of real need to use urea on the more clayey rice fields; poor involvement of the extension service; confusing research messages related to rice soil fertility management; the high degree of uncertainty in Sukumaland farming systems; and low participation of farmers during priority setting for on-farm activities. Effective adoption of agricultural technologies generated by an FSRE methodology calls for strong, institutionalized links with the extension service, commodity research and policy makers. Better coordination of activities between donors and governments is an essential precondition to make such links work. Failures in the adoption of use of urea in rice encouraged researchers and farmers in Sukumaland to look for alternative ways to improve soil fertility in rice fields. Research has been done on the use of locally available resources such as kraal manure and rice husks, and the introduction of green manure and multipurpose trees as an alternative to urea. The performance of green manures and multipurpose trees was meagre due to limited potential for biomass production in the semi-arid climate. Half of the households in Sukumaland have no easy access to cattle manure, and in any case the quality of the available manure is low, due to open air collection and very low addition of crop residues. The relatively large amount of labour involved in transporting and incorporating bulky organic materials like kraal manure, green manure, rice husks and tree leaves in the relatively far and less easily accessible rice fields is also a serious problem. The increase in labour required per hectare is difficult to realize in the thinly populated Shinyanga region, and furthermore is not seen as desirable by households anywhere in Sukumaland, due to the expected decrease in labour productivity. Apart from that, farmers with clayey rice fields see no need to invest so much in soil fertility management. A nutrient balance calculation for the rainfed lowland rice cultivation systems in Sukumaland gave no serious depletion rates for major nutrients, which seems to support the farmers' attitude. Despite massive campaigns promoting the use of organic fertilizers in Sukumaland during the colonial period and recent attempts by the Tanzanian government, the adoption rate is still very low in almost all cropping systems. Only near Mwanza town are farmers applying kraal manure to horticultural crops and, to a lesser extent, maize/cassava fields. The quantities applied are, however, not sufficient to achieve positive nutrient balances on these sandy upland soils. A review of the literature suggests that INM/LEISA successes in SSA are characterized by relatively high intensity of land use, high population density, medium to high livestock density, good market access and presence of large urban markets (in particular for horticulture products) in the vicinity. Further, the active support of SSA governments for their agricultural sectors will be needed if sustainable agriculture is to be attained through INM/LEISA approaches, as well as active intervention to protect their agricultural sectors against competition from industrial countries. INM/LEISA approaches are not appropriate for SSA farming systems that lack these characteristics. Several examples from SSA and Asia show that severe dependence on labour-intensive methods may lead to decreases in labour productivity. LEISA advocates seem to be largely unaware that farming methods based primarily on labour-intensive techniques can lead to the impoverishment of farm households. For resource-poor farmers, sustainable agriculture must first of all be socio-economically viable. Insufficient use of external inputs can turn LEISA into a non-sustainable form of agriculture; therefore LEISA advocates should take a critical look at the impact of structural adjustment programmes in SSA. The main conclusion of this thesis is that INM/LEISA approaches are currently not an appropriate way to generate sustainable soil fertility management in the rice cultivation systems of Sukumaland. Farmers with rice fields located on fertile clayey soils are still satisfied with their grain yields, and are not yet motivated to invest labour and cash in soil fertility maintenance. However, especially in rice fields located on sandier soils, in the future farmers will have to invest considerably in soil fertility maintenance to achieve sustainable rice cultivation. The current situation in Sukumaland makes such investments highly unlikely. The huge increase in rice cultivation in Sukumaland is on the other hand a good example of farmer adaptation to increasing population densities, changes in market opportunities, and soil fertility advantages in the valleys. Negative nutrient balances furthermore do not always justify recommendations to farmers that involve the immediate use of mineral fertilizers and/or organic fertilizers. More farmer participation, especially in priority setting, is necessary to prevent misunderstandings regarding farmers' objectives. Any strategy for future sustainable rice cultivation and agriculture in Sukumaland, including INM or LEISA, must be based on a thorough analysis of biophysical, socio-economic and public policy factors and their linkages. Such a strategy requires a conducive economic and policy environment. In Sukumaland this will require improvements in infrastructure, increased government support to agriculture, reduced taxation in the cotton crop sector, reduced reliance on rice imports, higher population densities, intensive livestock keeping and a greater variety of off-farm employment. A good INM strategy in Sukumaland would then be to use urea in rice, and farmyard manure in the nearby cassava/maize and cotton fields. Without a conducive economic and policy environment, population growth in Sukumaland will lead to an intensification largely based on labour inputs. Instead of agricultural evolution, agricultural involution will be the result.
