|Table of Contents | EHN Homepage | AAAS Homepage|
Basia Zaba, Centre of Population Studies, London School of Tropical Hygiene and Medicine
Ndalahwa Madulu, Institute of Resource Assessment, University of Dar es Salaam
"Nearly half of the Earth's population does not have enough water to drink. And demand is doubling every 21 years" (Vidal, 1995). This summarizes the headline in a newspaper, depicting the World Bankís views on impending water crises as barriers to development and sources of international conflict. Population growth is seen as a major factor in precipitating these problems. This chapter expands on that view and examines how migration can also be a factor. It concentrates in some detail on the situation in rural and urban parts of Mwanza region in Northern Tanzania.
The relationship between human population and water resources is viewed in this chapter as a two-way processóthat is, instead of regarding population growth as the only cause of water shortages, water availability will also be considered as a possible ìpushî or ìpullî factor in explaining migratory change and other socio-demographic outcomes. In concentrating on migration, this chapter departs from the usual Malthusian approach, which dwells on the mortality consequences of population and natural resource imbalance, in the belief that a migratory response is more likely to be detectable in the short run, and in the knowledge that migration is one of the most important determinants of population growth in villages, between villages, and in towns, the natural planning units for water supplies. Although migration will not ìsolveî problems of national or sub-national water scarcity, it does act as a mechanism to redistribute population in greater conformity with availability.
Three broad areas of water use will be covered: cultivation, other rural uses, and urban uses. This does not correspond to the usual divideóagricultural, domestic and industrialóbecause in this part of Tanzania, as in much of sub-Saharan Africa, the former breakdown achieves the clearest distinction between sources and between the technologies used to exploit them. Rural domestic use draws on the same sources (shallow aquifers and surface catchment) that are used for pastoral agriculture, whereas cultivation is directly dependent on rain; and only urban areas benefit to an appreciable extent from piped supplies. Within each of these broad areas of water use, needs, and availability, we discuss in this chapter the dynamic relationship between water and population, and adaptations which make possible the more efficient use of scarce resources.
Throughout the chapter, attempts will be made to see how the attitudes of water users and water engineers in industrialized countriesówith their emphasis on purity, private access, pumping and irrigationólead to distortions both in the analytical methods used and in the range of policy solutions adopted.
The Mwanza region was chosen for this illustrative study of interaction between demographic trends and water needs because it lies at the heart of the semi-arid area of East Africa which has been identified by eminent hydrologists (eg. Falkenmark, 1989; Cleaver and Schreiber, 1992) as facing acute water shortage problems by the end of this century. It is not at present involved in any major dam construction schemes which would require a socio-political analysis largely involving extra-regional factors, so local demographic trends might be expected to be pre-eminent determinants of local water problems.
Mwanza region, with an area of nearly 20,000 square kilometers (km2) and a population of over 1.8 million in 1988, is the most densely settled of Tanzaniaís 20 mainland administrative regionsóthe rural population density is around 70 persons per km2. It lies on the southern shore of Lake Victoria and stretches across areas with a wide range of rainfall levelsófrom over 1,400 millimeters (mm) per year on the islands on the lake, to 1,100 mm at the Western end of the shoreline, right down to under 700 mm in its most Easterly inland corner, over 80 kilometers from the lake. Eighteen percent of the population lives in urban areas, mostly in Mwanza town which is Tanzaniaís second largest and an important international ferry port and road and rail junction. Rural population growth in the last intercensal decade (1978-88) was 1.7 percent per year, the urban population grew at 8.2 percent. There are only two tribes of any numerical significance in the rural areas: the Sukuma who occupy all the mainland areas form 90 percent of the rural population, the remaining ten percent are the Kerewe island dwellers.
The region has two distinct rainy seasons each yearóthe ìsmall rainsî generally come in October and November, and signal the start of planting. The main rainy season is from February to April, and its intensity is the main determinant of crop yields. The countryside is undulating, with granite outcrops and shallow soils on top of low hills, and seasonal swamps in many of the broad, clay bottomed valleys. There are no major perennial rivers. All the wet season rivers drain into Lake Victoria and its deep inlet, Smith Sound, which nearly cuts the region in half. Ground water levels and the level of the lake show marked seasonal variation: the height of the water in shallow aquifers drops by as much as seven meters, the lake level by 0.75 meters (Birkett, 1995). The lake, with a total area of almost 70,000 km2, is the second largest freshwater lake in the world, and at its northern end is one of the sources of the Nile.
Agriculture is concentrated on the light soils of the hillsides, with maize, sorghum and bulrush millet the principal food crops, supplemented by cassava, sweet potatoes, legumes and fruit trees. Cotton and rice are the principal cash crops, and these are concentrated in the more fertile areasócotton on sandy upland soils, rice on lower lying soils where these cover an impervious ìhardpanî layer. Sisal is also widely grown for hedging and furnishing material. The hilltops provide wet season grazing, the valley bottoms are used for dry season grazing. Apart from reserves on the islands and in Geita district, there is very little primary forest left, but about seven percent of the total land area is covered by secondary ìMiomboî woodland.
Fertility and mortality rates in the region are fairly typical of the levels observed in Tanzania as a whole, with an appreciable urban-rural differential: the crude birth rate in rural areas is 40 per thousand, 33 in urban areas; the crude death rates are 12 and nine per thousand respectively. This yields a natural increase of 2.8 percent in rural areas, and 2.4 percent in urban areas. From 1978-88, rural areas in the Mwanza region experienced a net rate of out-migration of around five per thousand, whereas urban areas experienced a net in-migration of 40 per thousand. Fertility and mortality rates are fairly uniform within the rural areas: 90 percent of the birth rate estimates for villages lie between 35 and 47 per thousand, for death rates the corresponding range is nine to 16 per thousand. Migration rates vary enormouslyóover 30 percent of villages in the region experienced positive net migration in the intercensal interval: ten percent actually had net in-migration in excess of 25 per thousand; ten percent had net out-migration in excess of 60 per thousand. Migration is therefore a significant determinant of the relative growth rates of villages in this area. The population is highly mobileóin rural areas, 12 percent of adult males and 22 percent of adult females have lived less than five years in the village in which they currently reside; for urban areas, the corresponding figures for neighborhoods are 22 percent for males and 29 percent for females. The main migratory movements amongst the rural populations in recent years have been away from the lake shore towards the southwest, so that Geita district has experienced the most rapid population growth.
The results reported below are an amalgamation of findings from three distinct but related pieces of work: an analysis of census data on water supply in 1978 and 1988 (Zaba and Kiwasila, 1994); a comparative study of water supply and population growth data from all 640 villages in Mwanza region, based on the hydrological survey conducted by Brokonsult in 1978 (Zaba, 1996); and two household surveys conducted in 1995 by the author and colleagues from the Institute of Resource Assessment from Dar es Salaam University (Zaba and Madulu, 1996). The first of these surveys covered 1,239 households in 12 villages in the rural area, the other was a small scale urban survey in Mwanza town, covering 175 households; both endeavored to explore the links between water supply and migratory patterns in the region.
It is useful to review the scale of human needs for water and approximate volumes usedóthese are shown in Table 1. The ìminimum needsî row shows us that direct consumption requirements (cooking and drinking) are modest compared with what we need for hygiene (washing bodies, clothes, utensils etc.) which requires about four times as much water; and that these domestic uses pale into insignificance when compared to the water needed to grow food for an adequate diet: plant transpiration to meet basic calorie needs requires at least 200 times the minimum needed for all domestic purposes, even if all nutritional requirements are met directly from plant sources rather than by eating animal flesh. It is therefore immediately apparent that if half the worldís population were not able to get even enough water to drink, the human race would be facing imminent extinction from hunger. In pointing out such contradictions, the purpose is to suggest that the natural tendency to concentrate on drinking in the context of water needs, addresses a rather insignificant part of the problem of water resources, and biases the range of solutions which are likely to be proposed for perceived shortages.
The identification of water-scarce countries and regions referred to earlier is based primarily on agricultural rather than domestic needs. Hydrological data on rainfall and river flow are analyzed to see how much water enters an area, allowance is made for evaporation and transpiration, unmeasurable river outflow, and water lost to deep, inaccessible aquifers, and then the balanceóthe accessible run-offóis deemed to represent the usable freshwater resources of the area (Falkenmark, 1989a). In the semi-arid tropics accessible run-off would generally be of the order of two percent of the total rainfall. Allowing for the relatively small per-capita amounts needed for domestic and industrial uses, it is possible to estimate how much food could be grown using this amount of water for irrigation, and how large a population could be supported with an adequate diet. Depending how close the actual population of a region is to this theoretical maximum, it allows the analyst to classify the severity of current water resource problems, and allowing for future population growth, to assess the future position (Falkenmark, 1994).
Obviously this approach gives a much more reasonable picture of the population-resource balance than the journalistic caricature which we started with in this chapter. One readily apparent limitation of this approach, is the assumption that all the food needs of an area should be met by local agricultureóan assumption which rarely holds in MDCs, but which is more justifiable in the case of subsistence economies which have little to trade. But there is a more fundamental flaw in this approach, which emerges as soon as we look at the reality of subsistence farming in regions such as East Africaóthe negligible contribution made by irrigated agriculture to total food production. In Tanzania as a whole 96 percent of food crops come from rainfed farming (World Bank, 1994) which is completely ignored in the above calculations. In the Mwanza region there are no large scale irrigation projects at all.
Bringing rainfed agriculture into the picture does not make everything rosy. Irrigated agriculture is generally assumed to produce much higher yields per acre, or per cubic meter of water on cultivated land than rainfed cultivation, since much more intensive methods are used to compensate for the initial outlays and overheads of pumping water. But the total amount of water available to food plants directly from rainfall is very important because of the large areas round the world dedicated to this activity, and the large numbers of people dependent entirely on rainfed crops.
We can get a rough idea of the outcome for ìwater stressî calculations of neglecting rainfed agriculture by estimating the maximum proportion of the total rainfall which might be used in crop transpiration. Consider the components of the local hydrological cycle in Mwanza region: between 40 percent to 60 percent of rainfall evaporates back into the atmosphere without passing through green plants, and between five percent to ten percent ends up as run-off in aquifers or rivers draining into Lake Victoria, so that between 30 and 55 percent is transpired by plants (Brokonsult, 1978). The amount transpired in a given area depends on the type of vegetation, the soil properties and local climatic variationóbut if we assume that on average proportions transpired in cultivated and non-cultivated areas are broadly similar we see that the amount of water transpired in cultivated areas which form over one third of the total land area in this region is between ten percent and 20 percent of the total rainfall. If we were to allow for the relative concentration of cultivated areas in those parts of the region with higher than average rainfall, this estimate would be higher. By contrast, the accessible run-off in Mwanza region, which is the basis of most water stress calculations, amounts to between one percent to five percent of the total rainfall, if we follow the usual assumption that between 50 percent to 80 percent of run-off is lost in river surges and deep aquifers. In other words, the resource base is underestimated by a factor of five or so, by concentrating on the potential for irrigation.
Given this reassessment of the agricultural water resource base, can we then conclude that with the current extent of cultivation the rural population of Mwanza region could satisfy its own minimum calorie requirements on the basis of local production? The answer is a somewhat uncertain ìonly just.î Simple calculations of rainfall on cultivated area show why.
The estimate of a yearly per capita requirement of 700 cubic meters (m3) per year (Cohen, 1995) is based on a per-capita daily consumption of about one kilogram of cereals per day, an assumption that edible grain is about 50 percent of the dry matter produced by the cereal plants after allowing for chaff and losses in processing and storage, and the empirically based observation that in general a cereal plant would need to transpire approximately 1,000 times its own harvested weight of water in the course of its lifetime.
The annual volume of rainfall in Mwanza, assuming an average depth of 0.9 meters per year, and a total area of 20,000 km2, is 18 billion m3, of which one third, about six billion cubic meters, would fall on agricultural land, and around 2.4 billion m3 (using an average value of 40 percent transpiration) would be transpired. Dividing this by the rural population, estimated in 1988 as 1.5 million gives 1,600 m3 per capita per year. Allowing for 15 percent growth in the rural population since 1988 would bring this down to 1,400 m3 per capita per year by 1996.
Although this is twice the amount of 700 m3 quoted as the subsistence minimum, it is important to remember that not all the transpiration on cultivated land is due to food cropsóa small fraction of cultivated area is lost to access paths, boundary markers, shade plants, water retaining structures etc. Some land is devoted to growing non-food cropsóin Mwanza the principal cash crop used to be cotton, and in 1978 it was estimated that about seven percent of the cultivated area was devoted to this crop, though in recent years the cotton acreage has fallen due to problems of pricing and irregular payments to farmers, and rice is replacing cotton as the main cash crop. Most important of all, a large fraction of the total transpiration on cropped land will be due to weeds, to dry season cover growing on residual moisture after the harvest or thriving at the beginning of the wet season if the main crop is planted late, and to deliberate fallowing for preservation of soil fertility. Reliable data on such factors at farm level are not available, but data from experimental plots in the zonal agricultural research station at Ukiriguru (Dogget, 1988; Collinson, 1976) show that yields can be reduced to 20 percent of the optimum obtained with traditional methods if planting is delayed until after the start of the rains, or to 30 percent by inappropriate spacing, which allows weeds to flourish in between.
The proportion of total area cultivated has increased since 1978ócurrent data on cultivated area are only available on a sample basis, but the increase in cultivated area is believed to be proportionately smaller than population growth. There is some recent evidence of intensification of agricultural production, with increased use of cattle manure for maintaining soil fertility. In general, the situation for those making a living from subsistence agriculture in this region is precarious, with localized problems in years of low rainfall, and fairly widespread nutritional problems in families with less access to land.
Figure 1 shows the distribution of the population in the rural areas in 1978, according to the per capita cultivation water availability in the ward in which they live, where cultivation water availability is defined as water transpired in the cultivated area. The tremendous variability between the endowments of villages is immediately apparent from this diagram. About 150,000 peopleóover ten percent of the rural populationólive in areas in which the water transpired in cultivated areas falls around or below the ìminimumî needed for subsistence, but 25 percent live in areas where the per capita availability is three or four times this level. Clearly farming intensity or nutritional levels or non-farming income must also vary considerably between villages.
Rainfall volume and reliability have clearly been important determinants of population movements in the past: historically Ukerewe, with an annual rainfall approaching 1.4 meters per year has supported population densities three or four times those of Kwimba which has half this rainfall. The proportion of land cultivated on the islands was also almost twice that of the dryer mainland districts. As a result by 1978 there was a much more even distribution of the population per million cubic meters of crop transpired rain than per square kilometer.
There is very little mechanization anywhere in this region, and little use of chemical fertilizers or pesticides except for cotton growing. However, the two most densely settled districts: Ukerewe Island and the rural environs of Mwanza town, where around 65 percent of the land is cultivated, practice more intensive forms of traditional agriculture than the other districts, which also have lower proportions of their land (35 percent to 45 percent) under cultivation. In both these densely settled districts fallowing has been cut back, but apart from this, intensification takes on different forms: around Mwanza town, there is a lot of vegetable growing and milk production; on Ukerewe there is careful use of human and animal manure, fruit tree cultivation and some timber planting. With respect to non-farm income, fishing is very important in Ukerewe, and proximity to Mwanza town provides some opportunity for petty trading and casual labor.
In spite of the above impression of subsistence farmers approaching the limits of agricultural production for their traditional technology and natural resource base, the complaint most frequently articulated by farmers with respect to agricultural production is the shortage of labor at times of peak demand: planting and harvesting. Every survey since World War II has shown that this is a major problem (Rounce, 1951; Larsen, 1974; Collinson, 1976; Brandstrom, 1985; Meertens et al 1995), and the widespread awareness of this problem undoubtedly drives the high fertility norms prevalent in the area.
There is a connection between this cultural pressure for high fertility and the temporal variations in water availability. With hoe cultivation, the land area needed to grow enough cassava, maize and millet to support a typical family even in years of low rainfall, will absorb all the adult labor and more at critical seasons. The months of labor needed for breaking up the soil to ready it for planting when the rains come, for weeding during the growing season, and for harvesting the crop, cannot be spread over the yearóthey must all be available at once.
Families with a large proportion of young adults whose labor can be controlled by the household head are the most advantageous from the headís point of view, and according to Sukuma custom, adult sons are obliged to cultivate their fatherís land even after their marriageóonly when a young couple have shown themselves to be proficient cultivators would they be allocated their own parcel of land. High fertility norms will probably start to change once all the potentially cultivable land in the area has been claimed, and there are indications that this is beginning to happen. Already in many areas land is being traded between and inherited within families, rather than allocated by clan leaders, as was the Sukuma tradition. Fertility is positively correlated with per capita crop water transpiration, and this effect is statistically significant at the village level. Ukerewe and Mwanza, the most intensively farmed districts, have total fertility rates under six compared to over 6.5 in the other districts (Zaba and Madulu 1996). These two districts also have higher child and adult mortality than the other four, though this is not correlated with crop water availability.
Rainfall is no longer a determinant of migrationónet migration shows a very weak correlation with crop water availability. The few villages which are currently experiencing long-distance net in-migration are those with good road access and relatively large areas of pasture which can be converted for cultivationóthese happen to fall mainly in the middle of the rainfall range, in areas most recently cleared of forest. At the same time, there is strong outmigration from those rural areas close to towns, where rural-urban links have become more strongly established. As a result the distribution of population with respect to crop water availability is becoming more, rather than less, polarized. Shortage of cultivable land in all rainfall regimes is one of the prime factors governing male rural to urban migration. But at a more local level, when it comes to movements between sub-villages, or between adjacent villages, land availability is not such an important factor, as spare land can be reallocated without moving houseóvillagers often rent land from each other, or practice share cropping.
The shortage of the most desirable agricultural landóthe light, productive ìLuseniî soils on the hillsidesóhas led to expansion of agriculture in both directions: up, onto the poor shallow ìLuguruî soils of the hilltops, limiting the area previously used for grazing; and down, onto the rich mbuga clays in the valleys. These clay soils are very heavy and almost impossible to cultivate with a hand hoeóuse of ox plows and even tractors is increasing as a result. The clay-bottomed valleys are naturally swampy, and readily lend themselves to rice cultivation, which is expanding rapidly, and supplementing cotton as the main cash crop for many small-holders. Traditionally rice is grown in small paddies formed to hold back seasonal run-off for one or two months after the end of the rains. It is a high risk crop, liable to fail completely in exceptionally dry yearsóin crisis years, people harvest more cassava, which can be left standing for two or three years, so that the starchy root it produces is the growth of more than one seasonís rainfall.
Adams (1992) describes this kind of flood cropping or rainwater harvesting as ìblurring the distinction between rainfed and irrigated farmingî, and cites a 1986 survey by FAO as estimating that 82 percent of Tanzaniaís irrigated area is in fact residual soil moisture management practiced by small-holders, as opposed to large scale schemes which would be recognized as irrigation in more developed countries. In Mwanza in 1978, rice cultivation was practiced on one percent of the total land area (about three percent of the total cultivated area). Flooded grassland accounted for nine percent of the total area, so that the potential for expansion of this kind of cultivation may be substantial, though not all impervious soils are suitable for rice cultivation. This type of small scale irrigation does not compete with urban uses as it draws on different sources, so farmers would not be priced out of water access by having to compete against industrial users. However the expansion of rice farming might lead to slower recharge of the shallow aquifers which are the main sources of domestic water in rural areas, it could further curtail seasonal grazing, and may divert surface flows from ponds and water holes.
Rice is a more useful crop than cotton in many ways: it provides food as well as income, it does not require the same level of protection from insects, it needs less processing before final use so there is less dependence upon unreliable local industry, and there is a large local market for it, so that its value is much less affected by international commodity price movements. With gradual improvements in transport, internal trade networks within Tanzania are improving. Mwanza region has for some time been a net importer of maize, which comes to it principally from regions further south (World Bank, 1994). It is now becoming an exporter of rice to the rest of the countryóalthough the net flow of grain by weight is still inward, rice is a more valuable crop than the other local staples, as it has better storage properties and is more in demand by the wealthier urban elites. The focus on large scale irrigation as the main determinant of agricultural production has led to a dearth of data on the activities of small scale cultivators and their traditional water harvesting activities, and we simply do not know how much rice is currently grown by smallholders in Mwanza, how much this contributes to their diet and to their income, or how far this is likely to expand in the future.
Table 1 expressed all water needs on a yearly basis, to enable us to compare the scale of agricultural and domestic needs. But unlike agricultural needs, which can be satisfied on a seasonal basis by storing rainfed crops for use in the dry season, domestic needs have to be met more or less uniformly throughout the year. In rural parts of Less Developed Countries these needs are met by the siting of settlements close to perennial sources of water. As noted in the introduction, ground water levels fluctuate markedly between wet and dry seasons, and the population is attuned to coping with very large changes in water availability, which have nothing, or very little, to do with population pressure.
Because of these large seasonal variations, and lack of technological resources for storing water from one season to the other, in this section we shall present most of the statistics on use and availability in terms of liters per day at the height of the dry season, rather than on an annual basis. The minimum needed for domestic use on an annual basis, 3.5 m3 per year, translates to a daily requirement of ten liters per capita.
White and Bradley (1972) measured water consumption in 19 villages in rural areas of East Africa, at times of the year when water was reasonably abundant in the neighborhood of the settlement. They found the average rate of daily use for domestic purposes ranged from about four to 18 liters per head. More recently, Drangert (1993) measured it in six Sukuma villages, and reported a range of between 22 and 36 liters per head. In our own study we found quite a wide variation between individual householdsóten percent of households used less than ten liters per capita, and five percent used more than 40, but at the village level the average amount is remarkably constant, given the rainfall differences: 18 to 22 liters in mainland villages, 15 in Ukerewe. These estimates of mean levels of use are based on the amount carried, and therefore are slight underestimates of total use, as they exclude water collected in the dwelling (for example from an iron roof), and water used at sourceósuch as water used in washing clothes on the river bank. Because domestic catchment and use at the source are higher in the wet season than in the dry, the amounts carried in the wet season are typically some five liters per capita lower than the dry season amounts quoted above. A year-round availability level of 20 liters per capita per day has been adopted by the Tanzanian government (Mambali, 1992) as the standard to which all rural dwellers should be entitled within their own villageóthis corresponds to seven cubic meters per capita per year, about twice the ìbare minimumî needed for survival.
According to hydrological surveys (Brokonsult, 1978) almost a quarter of the villages in the Mwanza region had supplies which yielded in total less than five liters per capita per day during the peak of the dry seasonóthat is less than half of ìminimumî need. The counts of sources obtained in this region-wide investigation, did not tally with our own counts of water sources, conducted in 12 selected villages 17 years later, even after taking account of sources developed in the interval. The difference was mainly due to the neglect in the earlier survey of small ponds and dams which were used principally for washing or for watering cattle. Various studies (eg. White and Bradley, 1972) have shown that even when water is scarce, or has to be carried from a long distance, people use more for washing (bodies, clothes, utensils etc.) than for direct consumption (drinking and cooking). In fact, Cairncross (1990) has shown that in general, the quantity of water used by a household is a more important determinant of child health than its quality, because of the critical role of personal hygiene. But water sources which are not fit for human consumption are often overlooked when estimating total available water resources, because of the focus on the ìdrop to drinkî at the expense of the ìbucket to bathe.î
This is not to deny the real hardship experienced in some villagesóa water source ìwithin the villageî could mean a walk of four or five kilometers for some inhabitants, as village areas are much larger than is customary in Europe. In our survey villages about half of all traditional and improved sources ran dry for two or three months a year, so many households cannot use their nearest source throughout the year. At the end of the dry season, water seeps so slowly into traditional catchments that long queues of buckets can be seen whilst women patiently scoop water from shallow puddles in their deepest recesses. In many places, women no longer adhere to the traditional sentiment that water should be drawn only during daylight hours, and go to the sources before daybreak, to take advantage of the nightly recharge.
An important and controversial use of village water sources in the rural areas is for watering cattle. The Sukuma are settled agriculturists who traditionally reckon their wealth in cattleóthe possession of livestock ìÖresembles invested capital at a high rate of interestÖî (Malcolm, 1953). They are very energetic in forest clearance, because of their need for agricultural land and the desire to eliminate tsetse. The 1978 estimate of the livestock units (LSU) in the region was 1,185,000 (where four goats or one cow make up a LSU) which means that the human population of the rural area exceeded the LSU count by less than ten percent. It is hard to find a widely accepted ìminimum water needî figure for cattleóthe Brokonsult report (1978) uses a figure of 30 liters per day per LSU, exceeding the human per capita target for all domestic uses; Rounce (1951) estimates the requirement to be 25 liters; Drangert (1993) gives current use rates between 15 and 20 liters.
Reduction of stock policies designed to alleviate the soil degradation caused by cattle, particularly in the vicinity of water sources, are reported to have met with little success (Brandstrom, 1985) but we do not have a reliable count of livestock for recent years. It is generally believed that the cattle population of the region is decreasing rather than increasing, as fallows used for grazing have been shortened to meet the needs for increased food production. Anecdotal evidence suggests that Sukuma herds are grazed in various sparsely populated parts of Tanzania, and some are even reported to have reached Zambia and Zaire.
Our own survey shows that livestock ownership is very unevenly distributed: on the mainland over 55 percent of households own no livestock at all; the average number owned by households with livestock is around 13 LSU, but in most villages there are a few households with herds numbering over 100 LSU, though many of these large herds would be put in the care of others and dispersed outside the region. A very different situation pertains on the Ukerewe islands where land is in short supply: here two thirds of households own livestock, but the number owned by each household rarely exceeds two or threeóthese animals are kept tethered close to the dwelling, and grass is often cut and carried home to feed the animals.
Outsiders have believed, since colonial times, that the Mwanza region has been seriously overgrazed and persistently advocated reduction in stocks. Agricultural extension work in animal husbandry has concentrated on improving the quality of the livestock, and providing veterinary services, whilst emphasizing the need to decrease overall numbers. Because of these policies, villages received little help in developing water sources for cattle. But even today, the visitor will be struck by how much energy and ingenuity is expended on maintaining the small surface reservoirs (lambos) used for watering the cattle, with all the resources and technical know-how coming from the villagers themselves. These lambos have to be cleared of reeds and silt every few years, and the dams repaired. The work has to be done by large gangs of men at the height of the dry season, yet no outside inspectors are needed to harangue the villagers to undertake this workóin sharp contrast to the maintenance of the SGAB-HESAWA (Swedish Geological AB-Health Sanitation and Water Programme) wells. Drangert (1993) sees this as a reflection of the balance of power between the sexesóproviding water for cattle is the responsibility of men, whereas domestic water provision is womenís work. It is also a reflection of the respect accorded to wealthóthe equivalent of marble facades on our banks.
The traditional surface reservoirs also provide for uses other than watering cattleóclothes are often washed nearby, there may be a bathing place adjacent to the main pond, and fruit trees or vegetable crops may be grown in the seepage area. The water in them is not of drinking quality, but they alleviate the need to draw on high quality well water for bulk use in activities which do not need very pure water.
In the rural areas it is more difficult than in the urban areas to
assess whether domestic water provision is keeping pace with population
growth, since the census cannot tell us anything about the accessibility
and reliability of the wells used by the vast majority of the
population, and remote sensing can tell us very little about the long
term trends in the height of the water table which feeds the shallow
aquifers tapped by these wells. Against the background of wide seasonal
and inter-annual variation in local water levels, and the unexplained
jump in the height of Lake Victoria in 1961, the relative contribution
of population growth to water shortage is extremely difficult to assess.
There has been no regional hydrological survey to match the Water Master
Plan (WMP) of 1978, but we can at least compare the sources currently
used in the 12 villages surveyed in 1995, with those reported in
The population of the 12 villages grew by 40 percent in this
interval, from just over 21,000 to nearly 30,000. Excluding rivers, and
access to Lake Victoria, the WMP listed 64 sources for these villages.
In 1995 we found 258 sources in useóthat is four times as many.
Excluding all the HESAWA shallow ring wells and other improved wells
still leaves 198 traditional sources, i.e. three times as many as in
1978. We could conclude that water source development has far exceeded
population growth, but it is fairly certain that the WMP undercounted
the available sources for a variety of reasons: the survey was conducted
soon after the villagization program so it is probable that many of the
sources in the peripheral hamlets were ignored; it seems likely that in
concentrating on potable water many small ponds and swamps useful for
washing and cattle watering were discounted; finally, there may have
been deliberate omission by the villagers in their reports to the WMP
engineers, if they felt that villages with poor supplies would be given
priority assistance in the development of new wells. But on balance it
would seem probable that the number of sources per capita has increased
in this 17 year period.
Data on yields are even more problematic. We measured only current daily
usage, without monitoring the overnight re-charge, so we do not know
whether the rate of use was sustainableóthough in all likelihood it was,
since we were in the field towards the end of the dry season. The WMP
made proper estimates of total daily yield for both wet and dry season,
but only for those villages which did not have an ìinfiniteî source,
such as Lake Victoria, or a perennial riveróthis applied to eight of our
12 survey villages. The 51 sources which they identified in these
villages were capable of yielding a total of 90 m3 a day in
the dry season in 1978. In our 1995 survey, 204 sources were reported in
these eight villages, of which 100 lasted throughout the dry season, 84
of these were unimproved traditional wells or ponds, and the villagers
reported carrying away about 110 m3 a day from all of these
sourcesówe did not attempt to measure usage at source, such as watering
cattle, bathing or washing clothes, which may easily double this
estimate of the total amount taken.
These data, though somewhat unsatisfactory, certainly do not suggest any
decline in availability. We could interpret them either as evidence of a
large increase in rural domestic water consumption, or as indicating
slowly growing consumption and a large undercount of sources and yields
in 1978. It is interesting to note that in 1978, it was envisaged that
increased demand due to rural population growth would have to be met
principally through the sinking of deep boreholesóin fact, only eight
percent of the sources developed under the HESAWA program up to 1994
have been boreholes reaching deep aquifers, which can be viewed as
tapping ìnewî water resources. Ninety percent are new or improved access
points to shallow aquifersóechoing the point made by Kolsky (1996) that
water aid programs do not in general ìgiveî villagers new supplies, but
rather improve existing ones and make them more accessible. The
remaining two percent of sources developed under this program are tanks
for holding rainwater collected from iron roofs, mainly in schools and
other institutional premises. This type of rain harvesting has met with
considerable success in other parts of Tanzania (Zaba and Kiwasila,
The data collected by the 1978 hydrological survey allow us to calculate
theoretical levels of per capita availability for the rural parts of
Mwanza taken as a whole: according to the estimates of potential yield
of shallow aquifers and surface reservoirs, the mean volume available
per capita in 1978 was about 20 liters per day in the dry season.
Including the potential yield of boreholes accessing deeper aquifers
raises this by a factor of 20, to around 400 liters per capita per day.
We have already noted reservations about the reliability of the WMP
estimates of shallow aquifer yields, so it is possible that the
estimates of potential are also biased downwardsóbut it is immediately
apparent why boreholes were thought to be the only viable option in the
face of population growth which would lead to a doubling of the rural
population in about 40 years.
Data on mean levels of availability are in any case very misleading, if
we are dealing with a situation where the water sources (ponds, wells,
or boreholes) must be located within the village they are to serve, as
pumping schemes to serve the rural areas are beyond the countryís
financial horizon. Availability statistics have to be computed at the
village level, and thus the discussion below is not presented in terms
of mean liters per capita, but in terms of proportion of villages
falling into different availability categories.
About 25 percent of the regionís 640 villages lie within three
kilometers of Lake Victoria, or one of its few perennial riversóin rural
areas, the water from these sources is no worse in terms of bacterial
load than the water from traditional open wells, though there is a
higher risk of schistosomiasis in the reedy areas of the shoreline.
These villages are classified as having enough water resources to meet
their domestic needs, both now and in the future, assuming the same
technology and therefore per capita use as at present. A further 15
percent of all the regionís villages are situated above shallow aquifers
which have the potential to supply their current populations with over
40 liters per capita per day in the dry season, and so could comfortably
sustain a doubling of their populations without compromising the
government standard of 20 liters per capita, though this might entail
digging new wells or deepening existing ones. According to the WMP, the
remaining villages were evenly divided between those which even in 1978
could not obtain ten liters per capita per day from shallow wells in the
dry season, and those which were above this level but projected to fall
below sooner or later due to population growth. Most of these villages
are described as having a more than adequate potential yield from
boreholesóenough to supply hundreds of liters per capita per day at 1978
population sizes. But there are 56 villages, housing about ten percent
of the total population, which have no significant borehole capacity,
and which also had estimates of potential shallow well yield below ten
liters in the dry season.
Even if the WMP estimates of shallow well yield are underestimated,
these villages face a critical shortage of domestic water now, or in the
very near future. Unfortunately, the HESAWA program implementation
reports only provide data on numbers of sources, but not on yield rates,
or even months of water availability. There is an urgent need to monitor
yields from traditional and improved sources in water stressed villages,
if we are to obtain a truer picture of this critical population-resource
Figure 2 illustrates the relative importance of ease of access and
availability of household labor in determining household use rates of
water. The distributions of households by amount of water used per
capita in the dry season differ very little when we compare sub-villages
which have year-round working sources and those in which the supply
fails during the dry season. But there is a large decrease in the
percent who make do with less than 20 liters per capita in households in
which more than half the members are engaged in carrying water.
The population of the 12 villages grew by 40 percent in this interval, from just over 21,000 to nearly 30,000. Excluding rivers, and access to Lake Victoria, the WMP listed 64 sources for these villages. In 1995 we found 258 sources in useóthat is four times as many. Excluding all the HESAWA shallow ring wells and other improved wells still leaves 198 traditional sources, i.e. three times as many as in 1978. We could conclude that water source development has far exceeded population growth, but it is fairly certain that the WMP undercounted the available sources for a variety of reasons: the survey was conducted soon after the villagization program so it is probable that many of the sources in the peripheral hamlets were ignored; it seems likely that in concentrating on potable water many small ponds and swamps useful for washing and cattle watering were discounted; finally, there may have been deliberate omission by the villagers in their reports to the WMP engineers, if they felt that villages with poor supplies would be given priority assistance in the development of new wells. But on balance it would seem probable that the number of sources per capita has increased in this 17 year period.
Data on yields are even more problematic. We measured only current daily usage, without monitoring the overnight re-charge, so we do not know whether the rate of use was sustainableóthough in all likelihood it was, since we were in the field towards the end of the dry season. The WMP made proper estimates of total daily yield for both wet and dry season, but only for those villages which did not have an ìinfiniteî source, such as Lake Victoria, or a perennial riveróthis applied to eight of our 12 survey villages. The 51 sources which they identified in these villages were capable of yielding a total of 90 m3 a day in the dry season in 1978. In our 1995 survey, 204 sources were reported in these eight villages, of which 100 lasted throughout the dry season, 84 of these were unimproved traditional wells or ponds, and the villagers reported carrying away about 110 m3 a day from all of these sourcesówe did not attempt to measure usage at source, such as watering cattle, bathing or washing clothes, which may easily double this estimate of the total amount taken.
These data, though somewhat unsatisfactory, certainly do not suggest any decline in availability. We could interpret them either as evidence of a large increase in rural domestic water consumption, or as indicating slowly growing consumption and a large undercount of sources and yields in 1978. It is interesting to note that in 1978, it was envisaged that increased demand due to rural population growth would have to be met principally through the sinking of deep boreholesóin fact, only eight percent of the sources developed under the HESAWA program up to 1994 have been boreholes reaching deep aquifers, which can be viewed as tapping ìnewî water resources. Ninety percent are new or improved access points to shallow aquifersóechoing the point made by Kolsky (1996) that water aid programs do not in general ìgiveî villagers new supplies, but rather improve existing ones and make them more accessible. The remaining two percent of sources developed under this program are tanks for holding rainwater collected from iron roofs, mainly in schools and other institutional premises. This type of rain harvesting has met with considerable success in other parts of Tanzania (Zaba and Kiwasila, 1994).
The data collected by the 1978 hydrological survey allow us to calculate theoretical levels of per capita availability for the rural parts of Mwanza taken as a whole: according to the estimates of potential yield of shallow aquifers and surface reservoirs, the mean volume available per capita in 1978 was about 20 liters per day in the dry season. Including the potential yield of boreholes accessing deeper aquifers raises this by a factor of 20, to around 400 liters per capita per day. We have already noted reservations about the reliability of the WMP estimates of shallow aquifer yields, so it is possible that the estimates of potential are also biased downwardsóbut it is immediately apparent why boreholes were thought to be the only viable option in the face of population growth which would lead to a doubling of the rural population in about 40 years.
Data on mean levels of availability are in any case very misleading, if we are dealing with a situation where the water sources (ponds, wells, or boreholes) must be located within the village they are to serve, as pumping schemes to serve the rural areas are beyond the countryís financial horizon. Availability statistics have to be computed at the village level, and thus the discussion below is not presented in terms of mean liters per capita, but in terms of proportion of villages falling into different availability categories.
About 25 percent of the regionís 640 villages lie within three kilometers of Lake Victoria, or one of its few perennial riversóin rural areas, the water from these sources is no worse in terms of bacterial load than the water from traditional open wells, though there is a higher risk of schistosomiasis in the reedy areas of the shoreline. These villages are classified as having enough water resources to meet their domestic needs, both now and in the future, assuming the same technology and therefore per capita use as at present. A further 15 percent of all the regionís villages are situated above shallow aquifers which have the potential to supply their current populations with over 40 liters per capita per day in the dry season, and so could comfortably sustain a doubling of their populations without compromising the government standard of 20 liters per capita, though this might entail digging new wells or deepening existing ones. According to the WMP, the remaining villages were evenly divided between those which even in 1978 could not obtain ten liters per capita per day from shallow wells in the dry season, and those which were above this level but projected to fall below sooner or later due to population growth. Most of these villages are described as having a more than adequate potential yield from boreholesóenough to supply hundreds of liters per capita per day at 1978 population sizes. But there are 56 villages, housing about ten percent of the total population, which have no significant borehole capacity, and which also had estimates of potential shallow well yield below ten liters in the dry season.
Even if the WMP estimates of shallow well yield are underestimated, these villages face a critical shortage of domestic water now, or in the very near future. Unfortunately, the HESAWA program implementation reports only provide data on numbers of sources, but not on yield rates, or even months of water availability. There is an urgent need to monitor yields from traditional and improved sources in water stressed villages, if we are to obtain a truer picture of this critical population-resource relationship.
Figure 2 illustrates the relative importance of ease of access and availability of household labor in determining household use rates of water. The distributions of households by amount of water used per capita in the dry season differ very little when we compare sub-villages which have year-round working sources and those in which the supply fails during the dry season. But there is a large decrease in the percent who make do with less than 20 liters per capita in households in which more than half the members are engaged in carrying water.
One of the reasons for the Tanzanian governmentís Ujamaa villagization policy was to bring the population together in nucleated settlements, to facilitate self-help in developing clean water supplies, and to increase access to primary schools, dispensaries, and agricultural extension services. In the Mwanza region, the policy was enforced in 1974-75, and the current administrative structure of villages reflects the centers of population designated under this plan. However, settlement policy was relaxed in the 1980s, and families have gradually moved back to the smaller hamlets (subvillages), to live closer to the plots which they cultivate. Movement from the center to the periphery of a village is not circumscribed by local custom or regulations, nor is movement from village to town.
According to Sukuma tradition, women move to their husbandís household upon marriage, but it is somewhat less common for men to move from one village to another unless they are looking after cattle. For a man to settle and acquire land in another village would require the agreement of village elders, and increasingly these days would involve a transaction of money or cattle. In spite of this, migration rates between villages are high: we noted in the introductory section that about 12 percent of adult men have been resident in their current village for less than five yearsóof these only one percent are returning from urban areas, 11 percent came from other villages.
There is some evidence that domestic water availability is one of the determinants influencing short-distance inter-village migration. The 1978 to 1988 intercensal growth rates of villages are loosely correlated to their water source potential as measured in 1978, though this correlation was not apparent when the analysis was repeated at the ward scale. In our detailed survey of 12 villages (Zaba and Madulu, 1996) we found that households which reported having to go to another village or sub-village for water in the dry season lost more of their male members, and therefore gained fewer female members through marriage than households with a more convenient supply. Very few individuals actually report water availability as a reason for moving, though two thirds of recent movers reported that the domestic water facilities in the village of their current residence were an improvement over the ones available at their previous home.
In our detailed survey of 12 villages, we were able to relate child mortality to water use patterns. We discovered that certain water supply variables were more significantly associated with child survival than maternal education. The most important supply factor was whether the main source of water used was located within the subvillageówomen with a conveniently close supply reported five percent lower child mortality than those who had to go further. Also, those who reported using an improved, protected source had lower child mortality than those using a traditional, open source (Evison, 1996).
Not all villages are situated in areas close to deep aquifers which can be accessed through boreholes, so if at some future time population growth does outstrip the potential yield of local shallow aquifers, some villages may indeed find their growth constrained by seasonal domestic water shortages. But in general, shortages of cultivable land are likely to curtail village growth sooner than domestic water shortages.
The HESAWA program of water source development has brought about considerable changes in the way that water is used in the villages served by the program. The quality of the water from HESAWA wells which are lined with concrete rings, capped by a concrete dome and accessed by a hand pump, is much better than that from the traditional open sources which villagers dig for themselves. HESAWA villages now have more sources per capita than non-scheme villages, though the reliability of the HESAWA sources, in terms of mean months of water yield, does not appear to be much greater than that of traditional sourcesólocal opinion puts this down to the fact that many of the HESAWA wells were sunk at the start, rather than at the end of the dry season, when water tables were still fairly high, so that digging did not continue to the required depth.
These shallow ring wells have augmented rather than replaced traditional sources, because for many villagers the traditional sources are more conveniently located, more reliable, or have water which tastes better. About 60 percent to 70 percent of households in HESAWA villages continue to use traditional sources for drinking and cooking, compared with 90 percent in non-scheme villages.
HESAWA wells are designed to discourage at-source use: the only access is through the hand pump, spillage is channeled away from the well cover, the area around the well is supposed to be fenced and kept free of vegetation, and no provision is made nearby for laundry or bathing facilities. These are all very sensible precautions for keeping the water inside pure, but it means that households who use these wells must carry more water home for laundry and bathing, if they do not have an alternative traditional open source for these activities.
Although water carrying is mainly a female activity, over 35 percent of males in our survey also reported carrying wateróthe proportion was particularly high among teenage boys, and higher in the dry season than in the wet. We estimated that around 17 percent of all the water carried in the dry season was brought to the home by boys and men. Men are more likely to carry water in large quantities when the source is far from the homeómany use bicycles for this purpose, with plastic jerrycans strapped to the luggage rack.
Men are more actively involved in water management in HESAWA villagesóthe mechanics trained to repair the hand pumps are invariably men, and the committees formed to deal with the financial aspects of well maintenance have a preponderance of male members. In some villages, hand pumps are kept locked for most of the day, with nearby households taking it in turn to look after the key. During the designated hours of use a responsible adultóagain, usually a manóis required to sit by the well and make sure that children do not waste water, dirty the surroundings or damage the equipment. This has undoubtedly raised male awareness of water issues in these villages, though not all HESAWA wells were maintained to the agreed standard, which has led to some discord between villagers and district water engineers who inspect the wells from time to time.
Some of our informants claimed that as a result of the emphasis on the upkeep of HESAWA ring wells and other improved sources, traditional source maintenance was being neglected. This was not evident as far as the cattle watering catchments (lambos) were concerned, but it is likely that there are now fewer small ponds for laundry and bathing than there used to be. This trend could be accelerated if a decrease in the livestock population were to lower the demand for lambos, which are often used for laundry and bathing also. This could have the unfortunate consequence of increasing the demand for high quality well water for bulk use in activities which are traditionally carried out at source, and which do not need very pure water.
Villages close to larger market places are increasingly turning to vegetable gardening as an additional source of cash, though vegetables have always been grown to add flavor to staples. In our sample villages 15 percent of households reported vegetable growing as a year-round activity, with a further five percent confining themselves to wet season gardening. Even in the wet season about 25 percent of vegetable growers supplement the rain from traditional surface sources, in the dry season nearly 40 percent of them use water from traditional open wells or ponds in their gardens, three percent actually report taking this from improved wells. Similar proportions report using well water for brickmakingóa traditional dry season activity, as the bricks are baked dry in the sun. Such activities may also raise the demand for well water in the future, especially if the small surface catchments are increasingly neglected.
In urban peripheries, where there are good markets for milk, a few Sukuma cattle owners are gradually moving to the Kerewe style of cattle keeping: fewer animals of higher quality, grazed close to home or fed on cut grass collected and brought to the stall. Increased participation in the cash economy could eventually lead to less reliance on cattle for investment, and alleviate the pressure on hilltop pastures. Already some villages have declared a moratorium on tree felling above a certain height, and some are even planting trees for fuelwood, building material and watershed management.
Sukuma custom dictates that water for domestic use is freely available to anyone who requires it, though a charge may be levied for watering cattle (Drangert, 1993). In rural areas there is little sign of change in this respect, though there is anecdotal evidence that people who drive in from a town to fill containers at a particularly prized source may be asked to pay.
In urban areas, it is much more important than in rural areas for the population to have a piped water supply, as the aquifers which can be accessed by wells are liable to be contaminated by fecal matter from pit latrines and septic tanks (Feachem, 1983). In 1978, 88 percent of Tanzaniaís urban population had access to a piped supply for drinking water, 34 percent in their own dwelling or yard, the rest from public standpipes. By 1988 this proportion had fallen to 79 percent, with the fall in private sources, down to 27 percent, being most marked. The urban areas of Mwanza region have been amongst the worst affected by this general decline in standards, with an overall decline from 82 percent in 1978 to 66 percent in 1988, with private and public piped source access declining to the same extent.
In Mwanza town, according to the 1988 census, over 90 percent of households used a piped supply for drinking, but the situation in the smaller townsóthe district capitalsówas much worse: 20 to 30 percent of households living in the other two towns on the lake shore had piped drinking water, in the inland towns the proportion was only one to five percent. Although Mwanza town is the second largest in Tanzania, nearly half the urban population of the region lives in these small towns, which are growing faster than the regional capital, and are thus likely to experience far worse water supply problems in the near future.
Our survey of regional water engineers and observations in the field, showed that the main problems in providing piped supplies in urban areas are maintenance and installation of infrastructure (the main problem in regional headquarters like Mwanza town), and obtaining fuel for pumping (the overriding problem for smaller towns). Two of the smaller towns in the region have a network of pipes in place, and functioning pumping stations located by the lake, but water is rarely delivered to the network because the fuel allocation can only meet the requirements of a few weekís operation per year.
In the small towns, virtually all the households which do report using piped water supplies are dependent on public standpipes, whereas in Mwanza town around 20 percent of households have water piped into the home, or the yard in which the dwelling is located, and ten percent have full plumbing including flush toilets. Until recently, those who had water piped into the home or yard paid a flat rate standing charge, those who used standpipes obtained the water free, though more affluent households might pay a carrier to deliver the water to their home. In Mwanza town, about half of the households which have their own piped supply sell water to their neighbors. When their own piped supply fails (on average in our sample two days a week), 80 percent of these households have recourse to a free supply, of those households which do not have their own supply 45 percent use only free supplies, the rest buy some or all of their water from neighbors or from carriers. This means that although water is widely traded in urban areas, the municipal water authority receives very little revenue from the consumersócertainly not enough to cover the cost of pumping.
Recent government policy changes under the International Monetary Fund structural adjustment program have made it possible for municipal water authorities to install water meters in private households and to charge for water provided at standpipes. Meters have begun to be installed in Mwanza town, and there are plans to build kiosks at standpipes to charge for water from these. Just over ten percent of those of our survey respondents who had water piped into their homes reported having water meters, and the average monthly bill for these users was more than double the flat rate charge which was just over US$3óapproximately the price of a liter of petrol. Those households without a piped supply which paid for water also had a monthly expenditure more than double the flat rate charge, but the average extra expenditure on water from external sources incurred by households which paid for water when their piped supply failed was over four times the flat rate charge. Taking into account the proportion of households in each category, we have estimated that the water authority receives just 37 percent of the total expenditure on water incurred by the citizens of Mwanza, direct re-sale by private individuals accounts for another 20 percent, and about 43 percent goes to professional carriers, some of whom may also be buying water from private households. In fact only 15 percent of households report buying water from carriers, as opposed to 35 percent from neighbors, but a bucket of water fetched to the door by a carrier costs about US$.10ónearly seven times as much, on average, as a bucket purchased from a neighbor and carried home by the user.
Urban dwellers appear to be less discriminating than rural dwellers in their choice of a particular source for a particular end useóif a source is used at all, it tends to be used for all purposes. The exceptions to this pattern are neighborsí piped supply, which is used more for consumption than for washing, and wells, rivers and the lake, where the opposite is true. But 13 percent of households without their own piped supply drink water from the lake, and a further seven percent of households with a piped supply use this as an alternative drinking water source when their pipes fail. The townís sewage output goes into the lake, as does waste water from a fish canning factory, a brewery, several ginneries and some small tanneries. Although the lake is also the source of the piped water supply, at least the intake for this is situated a few hundred meters off-shore, and it is chemically purified to some extent.
We can get a very rough idea of the amount of water which should be available to Mwanza town on a ìsustainable useî basis, by looking at the recharge rate of the part of Lake Victoria which could be said to ìbelongî to Mwanza region. The lake has a total area of 69,000 km2, half of which is within Tanzanian territory (Okidi, 1990), bordered by the three regions of Kagera, Mwanza and Mara. On Tanzanian maps, approximately 10,000 km2 of the lake are shown as lying within the Mwanza administrative area, that is less than 30 percent of the Tanzanian total, even though Mwanza accounts for over 50 percent of the lakeís Tanzanian shore line. The urban population of the region is distributed in six towns, with Mwanza town accounting for approximately half the total; three towns (including Mwanza town itself) are situated on the lake shore, of the others one lies less than ten kilometers away, the other two over 30 kilometers. We will assume that Mwanza town is ìentitledî to half of what can be pumped sustainably from the regionís share of the lake, on the assumption that all the regionís urban areas have an equal entitlement to pumped water, but that rural areas will continue to depend mainly on the run off which passes through rivers and shallow aquifers.
Seventy-five percent of the water input to the lake comes from rainfall over the lake itself (varying between 80,000 to 90,000 million m3 per year on average), 25 percent from river discharge and seepage from underground. In the absence of interference by humans, between 60 percent and 75 percent of the annual input would be lost by direct evaporation, the remainder would flow out into the Nile at Jinja (Brokonsult, 1978). The absolute amount lost by evaporation is fairly constant, (at around 70,000 million m3 per year) but the outflow varies considerably with the water level in the lake, which dramatically increased (by 1.5 meters) in 1961, and has been slowly declining since, though it is still about 0.5 meters higher than the mean level from 1945 to 1960 (Sutcliffe and Lazenby, 1990). There is no agreed figure for what is considered a ìsustainableî rate of abstraction from the lake: in 1952 Hurst considered that a proposal to abstract 2,500 million m3 a year in Smith Sound for irrigation purposes (never implemented) ìÖcould make no appreciable difference to the NileÖîómore cautious attitudes would no doubt prevail today. For illustrative purposes, I propose to use a figure based on a maximum abstraction rate of two percent of the rainfall on the regionís share of the lake, reduced by half to allow for equal consumption rates by other urban parts of this region. Two percent of rainfall is the same proportionate amount which is reckoned as the ìusableî run-off over land.
Assuming a very conservative rainfall level of one meter per year over the 10,000 km2 which ìbelongî to Mwanza region, Mwanza town could thus ìsustainablyî use around 100 million m3 per yearóthis is more than 5 times the current rate of abstraction of 18.4 million cubic meters (Municipal Water Engineerís office, personal communication). This does not leave very much room for expansion of demand due to population growth and increased consumption, especially for industrial purposes. If the infrastructure required to pump this volume of water was available immediately, todayís population of around 300,000 would be able to use about 330 m3 per capita per yearóit is therefore obvious that even if future domestic consumption did not rise above 30 m3 per capita per year (a low rate of use if full plumbing and flush toilets become universally available), industrial use could not rise above 300 m3 per person per year even if population growth ceased altogether. Although this is well below the industrial use rates of urban populations in developed countries, it is 100 times the rate at which water is consumed in the townís industrial and commercial establishments today, so there would appear to be a substantial growth potential for industrial activity, provided that pumping capacity can be enlarged. In this respect Mwanza would appear to be one of the best endowed towns in Tanzania. Naturally, this hinges on the ìguesstimateî of two percent of rainfall over the lake as a ìsustainableî rate of use, and this may be unduly optimistic, unless the right technology were employed so that a large fraction of treated waste water could be returned safely to the lake.
The focus above has been on rates of abstraction, but an equally important determinant of sustainable use would be the capacity of the lake to absorb pollution from sewerage and industrial processes. No attempt is made here to assess these limits, as these would depend on the level of treatment of wastes (at present minimal), on data as to natural circulation of water in the lake (unavailable) and on assumptions about tolerable levels of various pollutants.
Lack of fuel and maintenance problems explain the decline in the number of households receiving internal piped water supplies in certain urban areas of Tanzania in the last intercensal decade, and these problems are not directly linked to the scale of local population growth. Although there are few large towns in Tanzania as favorably situated as Mwanza with respect to water supply, local hydrologists and water engineers do not feel that towns are threatened by an absolute shortage of water for domestic use, even if they continue to grow at current rates for the next two or three decades.
Yet there is a statistically significant relationship between the rate of urban population growth and the recent relative decline in piped water supplies (Zaba and Kiwasila, 1994), which is partly explained by the settlement patterns of urban in-migrants, and their relative lack of access to the amenities enjoyed by those who have been resident longer in urban areas.
Figure 3 shows the distribution of household water supply in urban areas of the region as reported in the 1988 census, according to the migrant status of the household. Households in which all or some of the members had been resident in the area for over ten years had a much better access to piped water than those households which consisted entirely of people who had moved into the area in the last ten years. This pattern is explained by the fact that in-migrant households dominate the smaller towns of the region, which have no piped supply. Within Mwanza town, those who want to build new houses often move to squatter settlements which are unplanned developments, relatively far from the public standpipe network. Also households composed entirely of immigrants have less well developed social networks, and do not find it so easy to obtain piped water from other private households. As a result, over 50 percent of immigrant households in Mwanza town depend partly on wells for their drinking water, compared to about ten percent of households with members who have lived for over ten years in the town (Stephenson, 1996).
Although only half of migrant households in Mwanza manage to obtain water regularly from a piped supply, this is still appreciably higher than in the rural areas as a whole where four percent get water from a piped supplyóeven in the urban fringe of Mwanza town less than 15 percent of households have piped water. However, the distance which urban residents must walk to reach their water source is very often similar to that reported by villagers. In our urban sample from Mwanza town, migrant households had to go 750 meters on average to obtain water, compared with 600 for resident households. By comparison the average distance to a water source in rural areas was about 800 meters in the wet season and 1,250 in the dry. Migrant households pay twice as much on average per bucket than resident households, because a lower proportion are able to get free supplies.
Individuals who migrate from rural areas to join households whose other members have lived in the urban area for a long time do much better in terms of water supply than rural families which set up a new urban household. The latter would hardly seem to improve their water supply situation by movingóalthough half can expect regular access to a piped supply, the other half may be worse off with respect to water quality, as urban wells are more likely to be contaminated. They may live marginally closer to their water source, but for most, domestic water will be an added expense which they did not have to bear in the village.
There is some evidence that within Mwanza town, water availability influences the choice of destination for those moving house from one neighborhood to another. Those households in which all the adults had lived in the town for over ten years, but had moved recently within the town itself had a lower mean distance to their water source (400 meters), paid slightly less per bucket for purchased supplies, and were marginally more likely to end up in a dwelling with piped water, than those urban households in which all the adults stayed put.
Water availability is not recognized as a pull factor in rural to urban migration: the main reasons for movement to town are employment, schooling, access to health care, and of course family reasons. However, water availability for industrial processes is an important determinant of economic development, and therefore may act through employment opportunity to influence the scale of future labor migration.
The proposed schemes for charging for water from standpipes will go some way to ensuring that municipal authorities recover more of the costs of pumping and maintenance from those who use the water, which should in turn enable better planning of investment for maintaining and expanding the public stand pipe network. This expansion is urgently needed in squatter settlements in Mwanza town, and in all the small towns of the region. Finding the appropriate pricing mechanism for standpipe supplies relative to those delivered to private households will not be easy: as long as flat rate charging is applied to the latter, the owners can undercut the standpipe price and still make a profit which will not benefit the water utility. Water metering is itself costly and takes time to implement, and any escalation in the general price of piped water will drive the poorest households to get more of their supplies from polluted wells and from the lake.
Vending by private households should not be bannedónot only would a ban be difficult to implement and monitor, but even if only partially successful would result in hardship for some current buyers, who would end up having to walk further and queue longer for their supplies. It should be possible to devise a system of licensing households with internal supplies to re-sell water which would ensure a reasonable level of profit for both the utility and the better endowed householders.
Much of the current use of neighborsí water supplies is quite altruisticópeople go to relativesí homes to bathe, women carry clothing to their friendsí houses where they do their weekly wash together, public spirited businessmen provide outlets on the outside of their premises for community use. Even in relatively wealthy households with full plumbing, water plays an important part in the gestures of hospitalityówhen guests sit down to a meal, the hostess will come round with a bowl, soap and a kettle of warm water so everyone can wash their hands at the table. These patterns of use point the way in which municipal authorities could act to alleviate the burdens of increased charges on the poorest members of the community, as well as reducing the amount of water carrying which falls disproportionately on women. The building of bath-houses and laundry centers, where water use on the premises could be subsidized, would ensure that a significant proportion of the water required for personal hygiene would be available at source, at a low cost to those needing it most. More research is needed however, on the current extent of at source use and water resale in urban areas, and on the pricing mechanisms which have evolved.
Sandra Postel (1992) has characterized the situation of humanity with respect to water resources as ìdrinking at the last oasisî. Her ìlast oasisî is a metaphor for the adaptive possibilities of concentrating on small scale improvements in use-efficiency, a moderation of unrealistic expectations, and a curb on uses in which one community undermines the welfare of another. Much of our work in the Mwanza region vindicates her findings.
One of the most important lessons has been about the misleading nature of ìmeanî measures of availability in circumstances where natural endowments vary tremendously even on a local level, and mechanisms for distributing surpluses to areas of deficit are not available. The scale of differences between villages and between individual households may be a more important determinant of resource consumption patterns than overall endowment.
Gloomy prognoses which show that ìidealî consumption standards will not be met if population continues to grow at a certain rate often fail to note that these standards have never been universally attained in the past, are not realized now for a large majority of people, and possibly are not even aspired to. Ideal standards often incorporate tacit assumptions about the necessity of carrying out all water consuming activities in the home, even though there may be considerable energy and infrastructure savings from using public facilities. It is easy enough to show that one family with full plumbing and flush toilets uses as much water as two families each with a single yard tap, or four families relying on stand pipes half a kilometer away. It is rather harder to discover the relative infrastructure costs of providing each kind of supply. It is even more of a challenge to find an appropriate measure for the relative health benefits which would be associated with equal investments in each type of supply, but these are the kind of considerations which need to guide water investment decisions. Research which looks at how the poorest sections of the community actually obtain and use water will suggest modest ways of improving the quantity and quality of water available to the poor which are not necessarily based on the standards and delivery systems used by industrialized or water rich nations.
There is an ultra-cautious view which implies that community self-sufficiency in food is a desirable way of balancing population and natural resources, including water. Taken to its logical conclusion this view would imply that the solution to the problems of local resource imbalance can be found only in drastically reducing human numbers, and begs the question as to how community boundaries should be drawn. The alternative is to encourage trade in resources and products. For human populations to exploit the natural resources each environment offers with the least disruption to other species, the best strategy is usually to trade in foodstuffs rather than moving water. As Allen (1995) has pointed out, food is ìvirtual water,î but the sheer bulk to be moved is just one thousandth of the ìreal waterî needed to grow it. Realistic estimates of how much of its food a community could produce should be based primarily on rainfed cultivation, rather than irrigation potential. A focus on rainfed cultivation would encourage work on development of hardy varieties of cereals and rain harvesting techniques in agriculture.
There is no doubt that rapid population growth increases the problems of water access and supply maintenance in agricultural and domestic spheres. Even for populations living in environments in which the accessible renewable water resources are very comfortably above per capita needs and usage, rapid population growth means rapid increases in the infrastructure needed to deliver the water to where it is needed. There are many compelling reasons for enabling couples to control their fertility other than resource constraint arguments, but it is important to bear in mind that halting population growth in itself will not improve access to water, nor will it alleviate environmental problems. Indeed, problems of resource access for human populations, and habitat loss and extinction for other species may continue to escalate rapidly even if human populations stop growing, locally or globally, if equal attention is not given to consumption and distribution issues.
Adams, W.M., 1992. Wasting the rain - Rivers, people, and planning in Africa. Earthscan: London
Allen, J.A., 1996. The political economy of water: reasons for optimism but long term caution, in Water, Peace and the Middle East - Negotiating Resources in the Jordan Basin. eds. J.A. Allan and J.H. Court, volume 9, Library of Modern Middle East Studies: Tauris, London
Berry, L., 1973.Tanzania in Maps. University of London Press: London.
Birkett, C.M., 1995. ìThe contribution of TOPEX/POSEIDON to the global monitoring of climatically sensitive lakes,î Journal of Geophysical Research. Volume 100, number C12, pages 25179-25204.
Brandstrom, P. 1985. The Agro-pastoral dilemma: Under-utilization or over-exploitation of land among the Sukuma of Tanzania, Working Papers in African Studies. African Studies Programme, Uppsala University: Sweden.
Brokonsult, 1978. Water Master Plan for Kagera, Mwanza and Mara Regions. Taby: Sweden.
Bureau of Statistics, 1978. Census, Volume VIII - analytical report, Volume II - Ward Characteristics. Takwimu, Bureau of Statistics: Tanzania.
Bureau of Statistics,1988. Census, Regional Profiles, Vol XIX: Mwanza. Takwimu, Bureau of Statistics: Tanzania.
Cairncross, S., 1990. Water Supply and the Urban Poor, in The Poor Die Young. eds Hardoy, Cairncross and Satterthwaite. Earthscan: London.
Cleaver, K. and Schreiber, G., 1992. The Population, Agriculture and Environment Nexus in Sub-Saharan Africa. World Bank: Washington DC.
Cohen, J., 1995. How Many People Can the Earth Support? Norton: New York.
Collinson, M.P., 1976. Guidelines for the long-term development of Agriculture in Mwanza region, in Mwanza Integrated Regional Planning Project, vol III. SIDA: Sweden.
Conyers, D., 1970. Agro-economic zones of Sukumaland, Research Report no.16. Bureau of Resource Assessment and Land Use Planning, University of Dar-es-Salaam: Tanzania.
Dogget, H., 1988. Sorghum. Longman Scientific and Technical: UK (IDRC).
Drangert, J., 1993. Who Cares About Water? Household water development in Sukumaland, Tanzania, Linkoping Studies in Arts and Science: Tanzania.
Eele, G., Duncan, A., Lawson, A. 1992. Environmental Change and the Response by Small-holder Farmers: Some Evidence from Tanzania, in Environment and Population Change, eds. Zaba, B. and Clarke, J. IUSSP Ordina Editions: Liege.
Evison, G., 1996. Child mortality and water use in Mwanza region, Tanzania. Unpublished MSc dissertation, CPS, London School of Hygiene and Tropical Medicine.
Falkenmark, M., 1989a. ìThe Massive Water Scarcity Now Threatening Africa - Why Isnít It Being Addressed?î AMBIO, vol 15, no. 4.
Falkenmark, M., 1989b. ìMacro-scale water scarcity requires micro-scale approaches,î Natural Resources Forum, November 1989.
Falkenmark, M., 1994. Water Availability as Carrying Capacity Determinant - a New Factor in Third World Demography, in Environment and Population Change. Eds. Zaba, B. and Clarke, J. IUSSP Ordina Editions: Liege.
Feachem, R., 1992. Sanitation and Diseases - Health Aspects of Excreta and Wastewater Management, in World Bank Studies in Water Supply and Sanitation, vol 5.
Gillman, C., 1945. ìPopulation Problems of Tanganyika Territory,î East Africa Agricultural Journal, vol 11.
HESAWA, 1994. Rural water supply, environmental sanitation and health education programmes in Kagera, Mwanza and Mara regions: program implementation statistics quoted from agreed minutes of extended management meeting. SIDA: Sweden.
Hurst, H.E., 1952. The Nile: a General Account of the River and the Utilization of Its Waters. Ministry of Public Works: Cairo.
Kauzeni, A.S., Mujwahuzi M., and Kiwasila, H., 1990. Study on Community Participation Approaches in Rural Water Supply and Sanitation Sector in Tanzania. Institute of Resource Assessment, University of Dar-es-Salaam: Tanzania.
Kiwasila, H., 1994. Water Supply Aspiration and Willingness to Pay: A socio-economic study of Mwanza urban area. Unpublished report, Institute of Resource Assessment, University of Dar-es-Salaam: Tanzania.
Kolsky, P., 1996. personal communication (postgraduate lectures on rural water supply, London School of Hygiene and Tropical Medicine).
Larsen, A., 1974. Agriculture in Sukumaland, Tanzania. Okonomisk Institut- Lanbrugsokonomiske studier no. 5: Denmark.
van der Linde, M., Heemskerk, W.C.S., Kaitaba, E.G., Law, L.J., 1995. Ginnery Location Study, Tanzania. Winand Staring Centre for Integrated Land, Soil and Water Research: Wagenigen, Netherlands.
Malcolm, D.W., 1953. Sukumaland: an African people and their agriculture. OUP: London.
Mambali, S.S., 1992. The National Water Policy, paper presented at seminar on re-organization of the Makonde Plateau water supply, vol. 2 of proceedings. MAJI: Tanzania
Meertens, H.C.C., Ndege, L.J., Enserink, H.J., 1995. Dynamics in farming systems: changes in time and space in Sukumaland, Tanzania. Royal Tropical Institute: The Netherlands
Okidi, C.O., 1990. History of the Nile and Lake Victoria Basins through Treaties, in The Nile: Resource Evaluation, Resource Management, Hydropolitical and Legal issues. eds. Howell and Allan. SOAS, University of London: London.
Postel, S., 1992. The Last Oasis - Facing Water Scarcity. Earthscan: London
Rounce, N.V., 1951. Technical considerations in the economic development of Sukumaland. Empire Journal of Expatriate Agriculture, vol 19, no 76, pp 253-264.
Stephenson, R., 1996. Poverty, Migration and Water Resources in Urban Mwanza. unpublished Msc dissertation, CPS, London School of Hygiene and Tropical Medicine.
Sutcliffe, J.V. and Lazenby, J.B.C, 1990. Hydrological data requirements for planning Nile Management, in The Nile: Resource Evaluation, Resource Management, Hydropolitical and Legal issues. eds. Howell and Allan, SOAS, University of London: London.
Tiffen, M., Mortimore, M. and Gichuki, F., 1994. More People, Less Erosion - Environmental Recovery in Kenya. ODA: London.
Vidal, J., 1995. No Water World. The Guardian newspaper, August 8th, vol 46,314, London.
World Bank, 1994. Tanzania Agricultural Sector Memorandum, Agriculture and Environment Operations Division. World Bank: Washington DC.
White, G., Bradley, D., and White, A., 1972. Drawers of Water: Domestic Water Use in East Africa. University of Chicago Press: Chicago.
World Resources Institute, 1993. World Resources 1992-93. Oxford University Press: New York.
Zaba, B., 1996. Village Growth and Pressure on Land and Water Resources in the Mwanza Region of Tanzania. Paper given at British Society for Population Studies conference, Brighton 1995, forthcoming in CPS research paper series.
Zaba B., Diamond I., Jones G., Kamuzora C., McMurray C., 1991. An Initial Analysis of the 1988 Census of Tanzania. Proceedings of a Workshop, unpublished report to ODA.
Zaba B., and Kiwasila, H., 1994. Domestic Water Supply and Population Growth in Tanzania: Urban and Rural Contrasts. in Environment, Population and Development, eds. Clarke, J., and Tabbah, L. CICRED: Paris.
Zaba B., and Madulu, N., 1996. Water Use and Population Mobility - An Analysis of Data Collected in a Household Survey of 12 Villages in Mwanza Region, unpublished report to ESRC and ODA.
|Table of Contents | EHN Homepage | AAAS Homepage|