Car1 Christiansson

Scandinavian Institute of African Studies, Uppsala and Department of Physical Geography, University of Stockholm Soil Erosion and Sedimentation in Semi-arid Tanzania

Erosion and Sedimentation -arid Tanzania

Studies of Environmental Change and Ecological Imbalance

Car1 Ghristiansson

Scandinavian Institute of African Studies, Uppsala and Department of Physical Geography, University of Stockholm This study also appears as Medd~lnnd~Nr A l19 from the Department of Physical Geography, Universit~of Stockholm, Box 6801, S-l 13 86 Stockholm, Sweden

Car1 Christiansson and Scandinabian Institute of African Studies 198 1 ISBN 91-7106-197-5 Printed by Borgstroms Trvckeri AB, Motala, Sweden 198 1 Abstract

Christiansson, C., 1981: Soil Erosion and down to bedrock within 50-100 years. Sedimentation in Semi-arid Tanzania. Stu- Important erosion processess in the in- dies of Environmental Change and Eco- logical Imbalance. Scandinavian Institute vestigated area are splash erosion, sheet of African Studies, Uppsala and Depart- wash, rilling and gullying. At present ment of Physical Geography, University of the first two are quantitatively domi- Stockholm. Uppsala 1981 208 p. ISBN nant. 91-7106-197-5. Due to high sediment production, two of the studied reservoirs have an ex- Research on soil erosion and related pected life of 3545 years only. The problems was carried out within the other two have an expected tota! life of scope of the DUSER project in Ugogo in 80-90 and 120-130 years, respectively. semi-arid central Tanzania in the years To assess long term environmental 1968-74. changes, land degradation is viewed in a The present thesis describes and historical perspective based mainly on analyses factors of importance for the written sources. According to these it initiation and acceleration of soil erosion seems as if the initiation and large scale and discusses the geomorphological development of soil erosion in the area effects of erosion processes in the semi- are due to: arid environment. Pressure from hostile neighbours on A twofold approach to the problems is the Wagogo, who during part of the used: 19th century crowded around the cen- 1. Recording of the present situation tral range of hills causing overpopula- 2. Reconstruction of past environmental tion and overexploitation of the land; change effects of the 19th century caravan To assess the present situation the trade, with large demands on food and types, rates and extent of soil erosion firewood; 20th century expansion of the and sedimentation were studied within livestock herd sizes well above the carry- five selected catchments, four of which ing capacity of the area; increased popu- with reservoirs. The principal methods lation density resulting in local land used are field surveys and air photo in- shortage and reduced fallow periods. terpretation. To give an idea of the present land The rates of erosion in the investigated potentials of Ugogo an attempt is made catchments are high, corresponding to to estimate the population and livestock sediment yields of 174 to 602 m3/kmzper carrying capacity of the area. Possible year as averages for the longest periods effects on the environment of the new of available records. Annual denudation settlement pattern are also considered. rates on moderately grazed upper In the concluding section suitable meas- pediments average between 1 and 2 mm ures for conservation of the marginal per year rising to around 10 mm per lands of Ugogo and similar semi-arid vear on overgrazed slopes with erodible areas in East Africa are reviewed. soils. Cultivated areas experience erosion rates in the same range. With the C. Chrzstzanrson, Department of Phjszcal present agricultural system the soil Geography, Unzverszty of Stockholm, Box cover on parts of the slopes will be lost 6801. S- 113 86 Stockholm, .S-r~leden.

Contents

1. Introduction 1 7. Ikowa dam and catchment 5 1 2. Scope of the study 4 Ikowa dam. Construction 5 1 Ikowa irrigation scheme 52 3. Methods 5 Reservoir surveys 54 General 5 Sediment yield 56 Catchment studies 5 Morphology and types of Reservoir surveys 6 deposits 57 Population and land use 7 Catchment erosion 61 Inselbergs and other hill- 4. Location of the study area 9 slopes 61 5. The environment 12 Upper pediment slopes 62 Geology 12 Lower pediment slopes 63 Topography 14 Gently undulating areas 64 Soils and vegetation 15 Stream channels and sand Vegetation types 15 fans 65 Soil zones and associated vegeta- Rainfall 66 tion 22 Infiltration and soil moisture 67 The skeletal soil zone 23 Evapotranspiration 68 The red soil zone 23 Drainage pattern and dis- The hardpan soil zone 24 charge 68 The zone of grey and black clays 26 8. Imagidam 72 Forms of erosion 27 9. Msalatu dam and catchment 74 Climate 31 Construction and operation of the Rainfall 33 dam 74 Temperature, evaporation, Drainage pattern 78 humidity 35 Erosion and conservation 79 Wind 35 Present distribution of erosion in Drainage 37 the catchment 82 6. Population and land use 38 Sediment yield and reservoir capa- Population. Immigration into city 83 Ugogo 38 Morphology and types of Spatial organization 40 sediments 89 Residential mobility 4 1 Grain size composition and concen- Land use 42 tration of suspended sediment 91 Cultivation 42 10. The proposed Mkonze dam and Crops 42 catchment 96 Cropping methods 44 Livestock and grazing 46 11. Matumbulu dam and catch- Water supply 49 ment 98 "A land condemned to perpetual Construction and potential use of drought" 49 the dam 98 Water reservoirs 50 Drainage 103 Erosion and conservation 103 Environmental degradation in the Erosion 103 precolonial and colonial period 150 Conservation 106 Adaptation to the environ- Upstream sedimentation 107 ment 150 Sediment surveys 110 The formation of "the cultivation Morphology of sediments 113 steppe" 152 Grain size composition 116 "Land development" in Ugogo during the colonial period 157 12. The methods-A critical re- Transformation of the socio- view 117 economic structure 157 Representativity of the study Anti-erosion measures 158 area 117 The caravan trade-its role in Relevance of the methods em- the development of the soil ero- ployed 117 sion complex 159 Field measurements 117 Early regional trade 159 Satellite image and air photo in- Caravan routes through terpretation 119 Ugogo 160 Reliability of basic data 12 1 19th century food re- 13. Contemporary erosion and sources 160 sedimentation in Ugogo. Discus- Food requirements of the cara- sion and comparisons with other vans 163 areas 123 15. The post-Independence Rates of runoff 123 situation 166 Sources of sediment and types of Availability of land and population erosion 127 carrying capacity 166 Erosion on inselbergs and Ikowa catchment 167 bedrock hills 127 The Dodoma catchments 17 1 Erosion on pediments-splash Livestock density and carrying and sheet wash 128 capacity 173 Overgrazed upper pediments 128 16. The future 178 Cultivated upper and lower The new settlement pattern; its im- pediments 129 pact on the semi-arid Vegetationless surfaces 132 environment 178 Cropping areas 133 17. Summary and Conclusions 182 Deciduous thicket 134 History of the soil erosion complex Grass 135 in Ugogo 182 Gully erosion 136 Processes and rates of erosion and Gullying on pediments 136 sedimentation 183 Gullies associated with fault Conservation and management 187 lines 138 Rates of catchment denudation and Appendix 190 sedimentation 139 Acknowledgements 195 14. The historical perspective 147 References 197 Early observations of soil erosion in Ugogo 147 Maps, air photos and satellite im- Erosion by water 147 ages 207 Erosion by wind 148 Abbreviations 208 List of tables

Table l Climatological statistics. Dodoma Table 18 Rainfall and runoff in Bubu and Meteorological Station. Kinyasungwe catchments. Table 2 Windspeeds in Ugogo. Table 19 Percentage runoff of total precipi- Table 3 Ikowa dam. Construction data. tation and soil lost per ha with different land use and plot treatments at Mpwapwa soil Table 4 Ikowa catchment. Sediment yield erosion experiments, 1933-38. and soil denudation rate. Table 20 Percentage runoff of total precipi- Table 5 Ikowa reservoir. Rate of sedimentation and soil lost from identical plots with tation. different land use at the Mpwapwa soil ero- Table 6 Grain size composition of sediment sion experiments, 1946-54. samples from Ikowa reservoir. Table 2 1 Percentage runoff of total precipi- Table 7 Monthly rainfall at 9 stations in tation and soil lost from plots with different Ikowa catchment and 3 stations at Dodoma. land use at Ouagadougou, Upper Volta. Table 8 Areal inventory of main landform Table 22 Reservoir data, sedimentation and units. soil denudation rate for five catchments in semi-arid Tanzania. Table 9 Msalatu dam. Construction data. Table 23 Dodoma district, Ikowa and Do- Table 10 Msalatu dam. First eight years of doma catchments. Total population and operation. population density. Table 11 Dodoma catchments. Areas of se- Table 24 Ikowa and Dodoma catchments. vere erosion, 1960. Population density of enumeration areas, Table 12 Msalatu catchment. Sediment 1967. yield and soil denudation rate. Table 25 Ikowa catchment. Population and Table 13 Msalatu reservoir. Rate of livestock numbers, 197 1/72. sedimentation. Table 26 Dodoma catchments. Areal cover- Table l4 Grain size composition of soil age of woodlandlthicket and cultivations, samples from Msalatu catchment. 1960. Table 15a Msalatu dam. Total amount of Table 27 Size of Wagogo homesteads, size suspended matter in the reservoir. of homestead herds and average hectarage cultivated. Table 15b Msalatu and Matumbulu dams. Chemical analysis of the water in the reser- Table 28 Ikowa and Dodoma catchments. voirs. February, 1970. Landform units with associated soils, vegetation and land use in relation to soil ero- Table l6 Matumbulu dam. Construction sion. data. Table 17 Transect record from Matumbulu catchment. List of figures

Fig 1 Sounding in Matumbulu reservoir. Fig 24 Gully erosion, Mkonze catchment. Fig 2 Ugogo Fig 25 Badlands, Kondoa. Fig 3 Dodoma region. Relief, drainage and Fig 26 Tanzania. Mean annual rainfall. administrative boundaries. Fig 27 Rainfall at Dodoma Meteorological Fig 4 Ikou~a and Dodoma catchments. Station, 192 1-1977. Geology. Fig 28 Bush fire near Kilimatinde Fig 5 Plateau landscape with inselbergs. Fig 29 Tanzania. Population density, 1967. View of Kongwa plains, E Ugogo. Fig6 Tor topography on the Central Fig 30 Aerial view of part of Matumbulu plateau. catchment. Fig 7 Reconnaissance view of central Fig31 Example of the distribution of Ugogo. Part of LANDSAT scene homestead fields and garden plots in a clus- 1155-07133, 25 Dec, 1972. ter of homesteads in Ugogo. Fig 8 Ikowa and Dodoma catchments. Fig 32 A Wagogo homestead in Matumbulu Landforms and zones of erosion and de- catchment. position, 1960. Fig 33 Newly cleared bush field on upper Fig 9a-d Matumbulu, Msalatu, Imagi and pediment, NW slope of Nyankali ridge. Ikowa catchments. Hypsographic curves. Fig 34 Cattle herd near Ikowa reservoir. Fig l0 Ikowa and Dodoma catchments. Fig 35 Overgrazed area suffering from Communications and settlements, 1972. sheet wash and rilling, E part of Matumbulu Fig 11 Dodoma catchments. Relief, drain- catchment. age and location of rainfall stations. Fig 36 Aerial view of Ikowa dam. Fig 12 View towards the E from the top of Fig 37a-c Ikowa dam. Annual record of wa- Imagi hill. ter levels in reservoir, 1958-1974. Fig 13 Deciduous bushland, Mpwapwa, E Fig 38 Ikowa dam. Reservoir contour maps, Ugogo. 1955, 1959 and 1969. Fig 14 Thicket in Ugogo. Fig 39 Ikowa dam. Accumulation of sedi- Fig l5 Miombo woodland in the dry season, ment, 1957-1970. Selected cross sections. Chenene hills. Fig40 Ikowa dam. Accumulation of sedi- Fig 16 Wooded grassland, E Ugogo. ment, 1957-1974. Long profiles and average Fig 17 Severe overgrazing near watering depth of cross sections. point, N of Msalatu. Fig41 Ikoua dam. Capacity curves, Fig 18 Gullies cutting through a surface soil 1957-1974. crust. Fig 42 Soil pedestal left after gullying. Fig 19 Thicket vegetation on eroded crust. Fig 43 Root exposures on heavily grazed NW slope of Nyankali ridge. pediment. Fig 20 Dry season view of mbuga area. Fig 44 Tree mounds on upper pediment Fig 21 Soil pedestals under protective grass slope. clumps. Fig45 Oblique aerial view of the gullied Fig 22 Sheet wash and rilling on upper pediment slope. N of Ngolo hill. Iko~va pediment, L$' of Imagi Hill. catchment. Fig 23 Rill formation along cattle track. Fig 46 "Termite mound terrain". E of Do- Imagi catchment. doma. Fig47 Minor ant hills on the pediment Fig 71 Matumbulu dam. Daily rainfall and above Msalatu dam. reservoir water levels, 1969, 1972 and 1973. Fig48 Bank erosion caused by undercut- Fig 72 Imagi, Msalatu and Mkonze catch- ting. ments. Air photo 49/TN/6/126,June 1960. Fig49 Water stage and approximate dis- Fig 73 Matumbulu catchment. Air photo charge, Kikombo river at Kikombo. 49/TN/6/153,June 1960. Fig 50 Imagi dam and Imagi hill from the Fig 74 Bush field on inselberg slope, E. Mkonze catchment. Fig 75 Culverts choked by sediment, NW of Fig 51 Oblique aerial view of Msalatu dam. Matumbulu dam. Fig 52 Dodoma town and surroundings. Air Fig 76 Matumbulu dam. Reservoir contour photo 49/TN/5/012,June 1960. maps, 1960, 1971 and 1974. Fig 53 Badly eroded slope with soil crust at Fig 77 Matumbulu dam. Accumulation of Mkonze, 8 km SW of Dodoma. sediment, 1960-1974. Selected cross sec- Fig 54 Dodoma catchments. Land use and tions. areas of severe erosion, 1960. Fig 78 Matumbulu dam. Accumulation of Fig 55 Msalatu catchment. Soil cover erod- sediment, 1960-1974. Long profiles and av- ed down to iron crust. erage depth of cross sections. Fig 56 Msalatu dam. Reservoir contour Fig 79 Matumbulu dam. Capacity curves, maps, 1944, 1960 and 1974. 1960, 1971 and 1974. Fi 80 Matumbulu dam. Area curves, 1960, Fig 57 Msalatu dam. Accumulation of sedi- 19%1 and 1974. ment, 1944-1974. Selected cross sections. Fig 81 Matumbulu delta, hlarch 15, 1970. Fig 58 Msalatu dam. Accumulation of sediment, 1944-1974. Long profiles and average Fig 82 Matumbulu delta, 1972. depth of cross sections. Fig 83 Grain size composition of sediment Fig 59 Msalatu dam. Reservoir area curves, samples from Matumbulu reservoir. 1944-1 974. Fig 84 Satellite view of Ugogo. Fig 60 Msalatu dam. Reservoir capacity Fig 85 Matumbulu dam. Annual record of curves, 1944-1974. water levels, 1963, 1966-1 974. Fig61 Pit for charcoal burning. Msalatu Fig 86 Trade caravan at Dodoma, 1915. catchment. Fig 87 "Erosion pavement" of laterite Fig 62 Advancing front of sand lobe on the nodules. main delta, Msalatu reservoir. Fig 88 Active gully erosion on pediment, Fig 63 Part of delta at the mouth of channel Kondoa area. 2, Msalatu reservoir. Fig 89 Example of thin soil cover on upper Fig 64 Profile of sediments at pit M17, Msa- pediment. latu reservoir. Fig 90 Sand fan developed in front of a gullied fault scarp, Singida area. Fig 65 Detail of the uppermost 7 cm of the profile shown in Fig 64. Fig 91 Relation of mean annual sediment yield to relief ratio. Fig 66a-b Grain size composition of sediment samples from Msalatu reservoir. Fig 92 Ikowa reservoir. Water levels and storage capacit\ , 1957-1 974. Fig 67 Msalatu darn. Grain size con~position of suspended sediment. Fig 93 Msalatu dam. Changes in reservoir capacity, 1944-1974. Fig 68 Grain size analysis of material from an abandoned river channel at Mkonze. Fig 94 Matumbulu dam. Changes in reservoir capacity, 1960-1974. Fig 69 Oblique aerial view of Matumbulu dam. Fig 93 Soil cover eroded to bedrock. Fig 70 Rainfall at Matumbulu dam, Fig 96 "A tree-less country with frequently 1962-1 974. occurring bare soil or sparse tufty grass". Fig 97 Early documentation of soil erosion Fig 102 Matumbulu dam and surroundings. in Ugogo (From Speke 1863). Fig 103 Caravan routes through Ugogo. Fig 98 grazing On stubble after Fig l04 Ikowa and Dodoma catchments. harvest, Matumbulu catchment. Population density, 1967. Fig 99 The cultivation steppe E of Dodoma. ~i~ 105 lkowa and Dodoma catchments. Fig 100 Vertical air photo of the area shown Areas under cultivationp in Fig 99. Fig 106 Central province. Fluctuations in Fig 101 Wooded grassland in the Kongwa livestock numbers, 1925-1964. area, early 19th century. Fig 107 Severely degraded land at Singida. l. Introduction

The semi-arid environments in East tion and an analysis of the underlying Africa face serious threats. Increased causes of the processes. As the study population,' shortage of land suitable moves away from the purely technical for cultivation, heavy grazing and in- focus which characterizes many other creasing demands for firewood and studies of soil erosion, opportunities charcoal have led to erosion, i.e., loss of have been added to emphasize new di- productive soil, increased sediment load mensions of the problem. So for in- in streams and consequent rapid silta- stance, for the first time the East African tion of water reservoirs. caravan trade is analysed from the point Agents of soil erosion are wind or wa- of view of its direct and indirect impact ter. In the East African drylands water on the natural environment and its pos- erosion is the dominant type. It occurs sible role in the initiation of soil erosion. as splash, sheet wash and channel ero- The research was carried out within sion. It is evidenced by features of the scope of the DUSER project in macro size such as gullies, obvious to Ugogo2 (Dodoma district and adjoining anybody, or features that form the areas), central Tanzania in the years micro relief such as rills, soil pedestals, 1968-74. The area was selected for tree mounds, etc. These are less obvious study in 1968 by A. Rapp, of the Uni- at first sight as are such consequences as versity of Lund, formerly of the Uni- changes in natural vegetation and de- versity of Uppsala, L. Berry and P. H. creasing yields on cultivated fields. The Temple, formerly of the University of problem of soil erosion is sometimes Dar es Salaam. The criteria for selection treated as an isolated physical phe- are presented in Chapter 3. Field work nomenon, but although many of its was performed in October-November roots are in the physical sphere the 1969 and February 1970 by A. Rapp ultimate causes are rarely physical but and D. H. Murray-Rust and by the pre- are found among socio-economic sent author and co-workers in October factors. 1970, January and November-De- The present thesis is an account of cember 1971, January and December studies of contemporary soil erosion 1972 and October-December 1974. The and sedimentation in the semi-arid savanna areas of central Tanzania, and a review of the degradation of the land in In Kiswahili, the official language in Tan- a historical perspective. The main zania, which in general in this thesis is used approach is physical geographical but it to indicate ethnic groups, individuals within is not primarily an investigation of the the groups and names of places and areas, the prefix ki- indicates a language; wra- indi- processes as such but a study of existing cates a group of people; m- indicates an in- features of soil erosion and sedimenta- dividual person and U- shows that it concerns a geographical area. In the local language Ugogo is called Wugogo. There is also a ' Toda), in Tanzania. more than 3 mill. ~rordCigogo which indicates the cultural people lite in the semi-arid areas as defined boundaries of the Wagogo, i.e., the ap- by the East African Royal Commission proximate area within which people belong (TMA 1977 p. 6). to the same cultural and social system. main report, where the present writer pastoralists, which means they are de- was one of the co-authors, included ma- pendent on agriculture for subsistence terial collected up to the season 1971/72 but maintain large herds of cattle, sheep (Rapp, Murray-Rust, Christiansson and and goats for social, economic and sec- Berry 1972). In the present volume the urity purposes. Traditionally the Wa- results then published are further de- gogo system of animal husbandry in- veloped. volved seasonal transhumant move- Ugogo is typical of large parts of in- ments between settlement areas and terior Tanzania in terms of the envi- areas of dry season grazing and water ronment. Dominating climatic features supplies. Today mobility is limited, thus are a long dry season and a short wet local overstocking occurs, in many season. The dry season lasts for about places causing overgrazing and soil ero- 7-8 months, generally begnning in sion. The present livestock carrying April and ending in November or De- capacity of the area is 2-5 halstock unit. cember. The mean annual rainfall is The traditional type of cultivation in 500-600 mm; potential evaporation Ugogo is shifting cultivation with bush ranges between 2000-2500 mmlyear. fallows of varying length. This is now- The alternating wet and dry seasons adays supplemented by manured and have considerable importance with re- more or less permanently cultivated gard to erosion development. The dry fields adjacent to homesteads. Soil ero- season allows the soil to dry out sion under shifting cultivation is not a thoroughly. In addition, the vegetation serious problem provided that clearing dies off so that bare soil is exposed to the is restricted to gentle slopes and that next rains. Soil exposure is further in- short periods of cultivation are followed tensified by overgrazing and clearing of by long periods of fallow. This system of fields for cultivation. land use can be satisfactory if the ratio The topography is characterized by of land to people is high. Due to the plains with scattered inselbergs or ridges increasing population pressure the or rows of hills. The soils occur in catena available land is now being cultivated sequences. On the upper slopes of the more intensely. The fallow periods in inselbergs thin stony soils dominate. the cultivation cycle are becoming shor- The pediment slopes are covered by red ter (TMA 1977). 'To compensate for de- or grey sandy soils, usually poor in creasing fertility and consequent low nutrients. Black or grey deposits prevail yields (there is a great lack of fertilizers on valley bottoms and flood plains. and a lack of means to collect and dis- Existing and abandoned cultivations tribute cattle manure) the cultivated take up considerable areas. Throughout area is continually being extended. New most of Ugogo the "natural" vegetation fields are being cleared on the upper is bush of secondary character. The parts of the pediments around the in- field layer is dominated by fast growing selbergs. Soils in this position are highly short-lived annual grasses and herbs. In erodible. Thus the fields loose their top places the bush appears in the form of soil and rills and gullies dissect the dense thicket or miombo woodland. slopes. Most of the inhabitants of the study Since the early 1970s a new factor has area are Wagogo. Up to the early 1970s become involved as the entire popula- the): lived in scattered homesteads locat- tion of the area has been moved into ed according to the availability of water, villages. This implies large concentra- cultivable soils and suitable grazing tions of people and cattle, which in turn areas. The Wagogo are cultivating may mean accelerated land degradation around the village^.^ Further increases land improvement projects it is impor- in population and strains on the land tant to make the local population aware may be exercised by the expansion of of the benefits of soil and water conser- Dodoma town, which since 1973 has vation and of the necessity of coopera- been undergoing development into the tion. But the benefits have to be ex- new national capital. plained and demonstrated in a way that However, a particular area can local farmers and herdsmen understand support only a limited number of peo- and appreciate. The details of this edu- ple if damage is not to be caused to the cation system have to be worked out and land. Restraints are imposed by factors transmitted by persons who are familiar such as soils, climate and type of land with the principles of both the natural use (representing the interaction be- and social environment. tween ecologcal and economic factors). Restoration of land quality is a long By the obvious fact that erosion occurs, term project, because the processes that unambiguously documented by rapid ruin the soils destroy the results of reservoir sedimentation, it is apparent centuries of geological and biological that with the present farming system the activity. Necessary initial measures are land carrying capacity is being exceed- better management of grazing lands, ed. To many planners the panacea protection of cultivated fields on gentle seems to be systems of more intensive slopes against splash and sheet wash, agriculture to increase productivity, but physical measures (terracing, cut off if they threaten the long term health of drains and contour ridgng) to prevent the environment through risk of acce- rapid runoff from fields on more steep- lerated erosion, they cannot be consid- ly sloping ground, protection of natural ered viable. woody vegetation and introduction of Land degradation is generally not village forest plantations for firewood promoted by lack of knowledge among and charcoal. For long term success all cultivators and pastoralists but rather by land use schemes must from the initial a combination of physical, social and stage include relevant basic research economic forces over which they as in- and accurate documentation so that mis- dividuals have little control. Thus, for takes can be recognized and corrected.

3" It is a mute point \+.hetherthe indigenous system of agriculture can adapt itself to static corlditions before irrepairable damage has been caused to the land around the villages" (TMA 1977). Scope of the study

The aim of the study is to describe and and sedimentation were studied within analyse factors of importance for the five selected catchments, four of which initiation and acceleration of soil erosion with reservoirs. and to document and discuss sedimen- Parallel to the account of the present tation and other geomorphological situation, i.e., the early 1970s, an effects of the erosion processes. attempt has been made to view the en- Moreover the study serves as a test of vironmental change and the develop- some selected methods for collection mknt of the soil erosion complex in a and analysis of indicators of land degra- historical perspective. This reconstruc- dation. A further aim is to make an tion is for the period from 1949 based assessment of the past and present on air photos combined with written re- pattern of land use and the population ports whereas for the period before and livestock carrying capacity. It is 1949 it is based entirely on written hoped that the material presented will sources. Particular interest has been de- form a reliable basis for future en- voted the impact of the caravan trade in vironmental monitoring within the area. Ugogo and its implications on land de- The approach to the problems is from gradation. two sides: 1. Recording of the present To give an idea of the land potentials situation. 2. Reconstruction of en- of Ugogo an attempt has been made to vironmental change during a recent his- estimate the carrying capacity of the torical period (1850s-1970s). area. Some possible effects on land de- The first approach has concentrated gradation of the new settlement pattern on documentation and evaluation, have also been considered. In the con- quantitatively and qualitatively, of soil cluding section suitable measures for erosion and sedimentation and their re- conservation of the marginal lands of lations to landforms, soils and vegeta- Ugogo and similar semi-arid areas in tion. To assess the present situation the East Africa are reviewed. types, rates and extent of soil erosion Methods

General - Good quality air photo coverage from several years exists. This section presents a summary of the methods employed in the study. For a To obtain an overall view of the present discussion of the relevance of the situation concerning degradation of methods see Chapter 12. (See also Rapp, vegetation and potential erosion in cen- Murray-Rust et al. 1972.) tral Ugogo, field reconnaissance trips The investigations are concentrated were undertaken supplemented by stud- on: 1. Measurement of sedimentation ies of early (1972) LANDSAT images rates in reservoirs and studies of types (Figs 7, 84). To relate the present situa- and morphology of sediment accu- tion to conditions in the recent past a mulated in the reservoirs and upstream thorough survey of older literature was in the catchments. 2. Recording of types made. Thus, trends of change in the of erosion and their relation to land use natural vegetation and in the intensity in the catchments and surrounding of land use could be determined. areas. 3. A survey of soil erosion in Ugogo during the last 125 years and its relation to land use, and man-land Catchment studies ratios. In the assessment of erosion the A number of methods have been ap- catchment basin as used as the primary plied to identify the spatial variation, the study unit. The selected catchments are extent and rate of erosion within the suited for investigations of this kind for selected catchments. a number of reasons: Mapping of land use and erosion features within the catchment was carried - They are situated near Dodoma town out by means of blacklwhite panchro- and are easily accessible all the year matic air photos (Figs 52, 72, 73) round. supplemented by field checks. The air - All the catchments but one have photo technique has proved to be well dams. Thus, the reservoirs act as suited for studies of the relatively open sediment traps allowing calculation of dry savanna environment of central rates of erosion in the catchment. - More background data are available than from most other reservoirs in material that throw light on some of the earlier obscurities. Thus differences exist be- the country.' tween some of the figures presented in this article and corresponding figures published ' In 1972 after the initial field season, the earlier. first results from the studies in the Dodoma All older published data are expressed in area were published (Rapp, Murray-Rust et inches, feet, etc. The data now published al. 1972). Follo~v-upand summarizing arti- have been standardized to the metric system cles, all based on material collected up to the except in a few cases where it has been con- end of 197 1, were published in the following sidered meaningful to keep the original un- years (Christiansson 1974, Rapp 1975, its of measurement. In most figures and dia- (I:hristiansson 1978). After 1971 I have grams both the old and the new units occur found, in local archives, additional valuable parallel to each other. Tanzania (Rapp, Murray-Rust et al. exist to employ special gauge readers 1972 p. 259; Christiansson 1972 p. 3 19; and full time supervisors of equipment. Murray-Rust 1972 p. 325, 329; Cook Suspended sediment samples were 1974). For the catchment surveys air collected from a small stream in Msalatu photos from the years 1949 (scale catchment. For this purpose an au- 1:33,000), 1953 (1: 10,000), 1960 tomatic sediment sampler of the type (1:40,000 and 1 :12,500) and 1970 Hayim 7 was used. (Schick 1967 p. 18 1; (1:10,000) were used. The vertical air Cf. Rapp, Murray-Rust et al. 1972 p. photos were supplemented by obliques 298.) taken during reconnaissance flights in a small aircraft in March 1970, October 197 1 and November 1974. Reservoir surveys The maps resulting from the surveys show areas of denser vegetation, culti- In Ikowa reservoir sediment surveys vated fields, erosion, sedimentation, etc were made by WD&ID in 1960, 1961 (Figs 8, 54). For calculation of areas a and 1963. Further surveying was photoelectric planimeter (Ekman & carried out by DUSER teams in 1969 Wastenson 1968) and a stereo mapping with some additions in 1970 and 197 1 instrument (Wild B8 Aviograph) com- (Table 5). Eventually the 1961 survey bined with an automatic co-ordinate re- w7as discarded when it was found that corder (EK-22) were used. Both the map was not accurate enough for methods give results of high precision. comparisons to be made. In order to get detailed views of veg- Preliminary studies of the Dodoma etation and erosion in the catchments, reservoirs and catchments took place in transects were measured from hill tops to valley bottoms. Along the transects readings of slope gradient were made every 25 or 50 m. The percentage of grass and bush cover was estimated for each 25 or 50 m section. Notes were taken on bedrock, soils, erosion and sedimentation features. Table 17 is an example of a transect record from Matumbulu catchment. For records of transects from Ikowa and Imagi catchments, see Rapp, Murray-Rust et al. (1972 p. 272 and 276) and Christiansson (1979 p. 136). Other methods employed in the catchment erosion studies include collection and analysis of soil samples and measuring of root exposures, tree mounds and soil pedestals. Originally, detailed precipitation and soil loss measurements were included in the plans. However, that part had to be left out as the equipment was removed by the local population and destroyed by Fzg 1 Sounding in Matumbulu reservoir. (Photo: grazing cattle. Economic means did not Rolf Heyman, Nov. 1974.) the period 1968-1970. They were fol- wire loop (Rapp, Murray-Rust et al. lowed by detailed surveying and map- 1972 p. 287). The profile sections and ping in 1971-1974. At suitable angles the sediment samples were analysed in with the dam walls base lines were laid the soils laboratories of the Physical out. From the base lines parallel trans- Geography Depts. of the Universities of ects were run at intervals. General Stockholm and Uppsala. methods of transect surveying have been described elsewhere (Rapp, Murray-Rust et al. 1972 p. 256). Population and land use In cases when the reservoir bottoms were covered with water the bottom The boundaries of drainage basins were topography was determined by means used as spatial limits for this study. For of sounding from a small, inflatable discussions on population and land use rubber-boat. Experiments with this it was however necessary to refer to method were initiated in October 1970. administrative boundaries. After improvements (steel wires Before 1967 the population was re- stretched across the reservoir, etc.) the corded on a divisional basis only, and method worked successfully in the 1974 before 1947 even larger enumeration surveys (Fig 1). For the soundings a units were used. For the 1967 census the simple measuring rod with a 13 cm di- divisions were, however, divided into ameter bottom plate was used. enumeration areas. These are the A sufficient number of bottom eleva- smallest units for which the population tions were thus determined and contour is stated. maps of the reservoir bottoms were The population of the study area prepared. In order to compute the vol- prior to 1967 lvas computed by use of ume of sediment in the reservoirs the topographic maps, air photos and data difference between the orignal storage from agricultural surveys. One method capacity and the present capacity of the applied in the present study was to dams was calculated. Areas enclosed by count the number of homesteads selected contours were determined by marked on the 1 :50,000 topographic planimeter and the volumes betrveen maps. In other cases the information consecutive contours were calc~lated. Tvas taken directlj. from air photos. The summation of volumes for the dif- The number of inhabitants in the ferent contour intervals plus the volume study area in 1967. the census year, was below the lowest contour gave the total calculated using information from all volume of the reservoir. (For details of enumeration areas wholly or to some the calculation method, see Rapp, part covered by the investigated catch- Murray-Rust et al. 1972 p. 257.) The ments (Fig 104). In enumeration areas morphology of the sediment accumula- divided by the catchment border the tions M.as studied on air photos and in population was assumed to be propor- the field. For studies of the characteris- tional to the area on each side of the tics of the sediment, samples were col- border. At the 1973 socio-economic lected from different parts of the reser- survey of the Ikowa area the catchment voirs (Table 6; Figs 66, 83). boundaries Tiere used as limits for the Sampling troughs were used in I ko~va census. C:alculations of the population reservoir to collect sections of the sedi- and livestock carrying capacity were ment in pits. The metal troughs, 1 m X made using methods employed h>- 5 cm X 2,5 cm Ivere hammered into a earlier tvriters (Berry 197 1b; Moore vertical pit face and then cut free with a 197 1 ; IVigg 1973). The historical part of the study is value a thorough analyses of each re- based entirely on written material, most cord has been required as well as fre- of it provided by early European quent comparisons between reports de- travellers (explorers, scientists, adminis- scribing one and the same area. It trators, etc.). The contents of those re- should also be borne in mind that the cords may sometimes be questionable observations made along the trade but the general impression is that many routes are not necessarily representative of the travellers made careful and sys- of the hinterlands. Thus conclusions tematic recordings of what they saw, drawn from the historical material only often with economic consideration in refer to areas in the vicinity of the cara- mind. However, to be of any scientific van routes. 4. Location of the study area

The studies are focused on the central doma town. The catchments are Tanzanian dryland and particularly its situated between longitudes E heartland, Ugogo (Figs 2, 26). Ugogo is 35°42'-36014' and latitudes S 6"6'-6" 19' thus understood to be an area with no and include parts of Chilonwa, Dodoma fixed boundaries, but inhabited by peo- Makulu and Mvumi divisions of Do- ple with common practices of land use doma district, i.e., the central part of and common cultural values-the Wa- Ugogo (Figs 4, 104). The largest of the gogo (Rigby 1969b p. l l). This country catchments, Ikowa, covers an area of covers some 20-25,000 km2 of the 612 km2. The main township in the plateau areas of interior Tanzania. The catchment is Kikombo (-3000 inh., core area of Ugogo is formed by Do- 1967). Dry weather roads connect Ki- doma district (16,576 km2) (Fig 3). Do- kombo with Buigri in the N and Han- doma town (23,559 inh., 1967) is the dali in the S. There are also motorable administrative centre of the area. tracks leading from Kikombo to The majority of the field data were Ihumwa and to Ikowa reservoir (Figs 4, collected from five catchments together 10). covering some 700 km" and E of Do- The four smaller catchments are all

Fzg 2 Ugogo. =ABOVE 6000' 114000'- 5000' 0BELOW 3000' 5000'- 6000' 13000'- 4000' 1SWAMP

"-"" REGIONAL BOUNDARY --- DISTRICT BOUNDARY • DISTRICT CENTRE

F~g3Dodoma regon. Relief. drainage and administrative hounciaries. Cgogo covers Dodoma district, the Lipart of hip%+-ap.r\.adistrict and the E 11a1-tof SIanyoni district (left edge of map). The frame indicates the area sho.rt.n in Fig 8. located S of Dodoma town. Matumbulu Iringa road runs W of the catchment catchment covers an area of 17.9 km2. and the road from Dodoma to Mvumi The Matumbulu dam is situated in the follows the E water divide. The Imagi SE part of the catchment some 12 km S and Msalatu catchments, situated 2-3 of the town. km S of Dodoma, cover 1.5 (4.3)' km2 Mkonze catchment, 25.3 km2, is locat- and 8.5 km2 respectively of the N slopes ed N of Matumbulu. The Dodoma- of Dodoma hills (Fig 11).

The natural catchment of Imagi covers 1.5 km2. This area has been extended by means of furrows dug along the hill sides. At present the effective extensions add 0.7 km2 to the natural catchment. The extensions have temporarily added larger areas and the max. size of the catchment amounted to 4.3 km2. 5. The environment

The natural and human environments others. James (1950 p. 1) groups all are described in some detail in the these components together under the following sections as they are both of term migmatite. immediate significance for the sediment Of the granitoid rocks, biotite granite production in the catchments studied. and schistose granite form the cores of Knowledge of the total environment is the isolated inselbergs, the hills and the necessary for understanding the ridges that characterize the landscape of agricultural pressure on the land and its the Central Plateau of Tanzania (Wade relation to the development of the ero- & Oates 1938 p. 18). The granitic rocks sion complex. have as large plutonic intrusives expanded into the very old rocks (3000 mill. years?) of the Dodoma Formation. Geology Before the intrusion of the granites, in the form of "bottomless" batholites, An outline of the geology of the investi- an ESE trending foliation occurred, gated catchments is shown in Fig 4. traces of which may be seen in the older Rocks of precambrian age, belonging to rocks which form parts of the Dodoma the Central Tanzanian Granitoid Shield, hills (Wade & Oates 1938 p. 17). form the geologcal foundation of The Dodoma Formation is character- Ugogo (Saggerson 1972 p. 70). They are ized by argillaceous and arenaceous made up of granites, granitic gneisses, metamorphosed sediments (metasedi- amphibolites, banded ironstones and ments) as gneisses, quartzites, amphi-

SUPERFICIAL DEES BEDROCK m Alluvium m Gronltes, rnlgrnotltes ond granitold gnelsses / Mlnor lntrusives (Dolerlte. Epldorltei 0Others (caliuviurn, eluviurn lmm puortz~t~cand colcoreeus rnetasedirnents ArnphjboLltec rocks ,/Fault Line Fzg4 Ikowa and Dodoma catchments. Geology. Slodified from Itade and Oates (1938) and King (1953).

12 bolites, etc. These rocks are mainly The continuous rising of the conti- found in the valleys and along the lower nent was, in Mid-Pliocene, interrupted slopes of the inselbergs. The occurrence by widespread block dislocation, as- of the metasediments, which are both sociated with the formation of the Rift foliated and more easily weathered than Valley System (Morgan 1973 p. 9) the granites, has to some extent come to Effects in Ugogo caused by these dis- determine the direction of the valleys . turbances are some major and several Where the rocks of the Dodoma For- minor faults (Figs 4, 84). The most mation are found near the surface, par- prominent of the faults form the W limit ticularly the amphibolite shows up clear- of Ugogo. The tectonics in Ugogo have ly. On weathering they generally form been studied and described by Dantz deep red earths which differ in colour (1902), Meyer (1915), Obst (1923), from surrounding soils and which also Wade & Oates (1938) and Fawley support a somewhat different vegeta- (1958a). tion. Due to the tectonic movements and to In many places smaller intrusives in the occurrence of "barrier" mountains the form of dolerite dykes and quartz in the E, Ugogo came for a period to and pegmatite veins occur. Of these les- have entirely internal drainage. This ser intrusives the dolerite is of dark grey was further emphasized through to blackish colour. Where it occurs on shallow downwarping associated with the surface it has often weathered the faulting. The rain that fell over the through exfoliation to more or less area was trapped. The consequence was rounded forms. The quartz veins gen- a high ground water table, extensive erally occur as small ridges, resistant to shallow lakes and a fairly humid climate weathering, on exposed rock surfaces or (Heidke 1923 p. 128). In the soil profiles as horizontal or vertical "stone-lines" in hydromorphic concretions (limestone the soil profiles. Within badly eroded nodules) and sometimes more extensive areas the quartz pebbles often cover the layers of CaCO, are found. This shows entire ground surface (Fig 95). that waterlogged conditions did exist in Ever since the late Precambrium ero- the past. sion seems to have dominated in Ugogo. Gradually the larger rivers, Kinya- On several occasions the land surface sungwe and Gt. Ruaha, formed passages was eroded to a pedi- or peneplain through the mountains by means of (King 1967; Morgan 1973). It is headward erosion. The lakes were assumed that during the middle part of drained completely or shrinked consid- the Tertiary the surface was eroded to erably. Many of them are today repre- near the base level (Quennel et al. 1956) sented by lacustrine deposits only. and Ugogo then formed a lowland plain The most prominent example of an with a hot and humid climate. earlier lake is Bahi swamp in W Ugogo The peneplanation was followed by a (Figs 3, 84). It is now filled with exten- rising of the continent, probably begin- sive layers of Tertiary and Quaternary ning in Miocene. The speed of the rising sediments and gives the impression of a since then has been faster than the nat- lake only in the rainy season. ural erosion. Thus, today the peneplain in Ugogo is found at an altitude of 900-1200 m (30004000 ft) a.s.1. (Fig 3). Fzg 5 Plateau landscape with inselbergs. View of Kongwa plains, E Ugogo. (Photo: C. Christiansson, Oct. 1974.).

Topography rise. The most prominent of these are the NW-SE trending Chenene hills and "Ugogo is a 1-ast plain-now flat, now heav- the WNW-ESE trending Dodoma hills ing uprvards, here level as a table. there tilted (Figs 99). up into rugged knolls bristling with scores of 3, 12, rough boulders of immense size which lie The study area comprises the central piled one above another." (Stanley 1872 p. and E part of the Dodoma hills and 174). plain and inselberg terrain to the N (Fig 7). Regarding the topography, the area The landscape shows a typical semiarid may be divided into five major parts. topography represented by pediment a) In the S a complex of mighty granite plains dotted with granite mountain blocks with steep pediments dominates. remnants and hills, a topography which b) In the E and NE the most prominent has fascinated many travellers (Figs 5, physiographic feature is isolated in- 6). selbergs, smaller and lower than the S From the even surface of the plateau massifs but still of considerable size. The also more continuous ranges of hills pediments here are more gently sloping

Fig 6 Tor topography on the Central plateau. In foreground: severely degraded land along cattle track. (Photo: C. Christiansson, Dec. 197 1.). and extend over wide areas. They are study area extend in two main direc- cultivated to a larger degree than in the tions, WNW-ESE and NE-SW. They S. merge without marked differences in c) Fault topography occurs in the W. level into corresponding valley systems The watershed of the Ikowa catchment in the adjoining catchments. is here formed by the Iyumbu-Madengi A tectonic line which has been inter- ridge which is bounded by faults with preted as a continuation of the their escarpments facing the Hombolo Iyumbu-Madengi fault (Fig 4) has been depression 75 m below the crest of the demonstrated by the present author ridge (Fig 4). The topography found from the area S of Msalatu reservoir around the upper reaches of Luaha and (Rapp, Murray-Rust et al. 1972 p. 290) Kikombo rivers has also been interpret- and from Matumbulu catchment. Parts ed as affected by tectonic movements of the fault are covered by loose deposits related to the Iyumbu-Madengi system. and are in places difficult to identify in d) The NW section of Ikowa catchment the field but it is clearly visible on air is flat or gently undulating. This type of photos. In some places along the as- terrain extends towards the central sumed fault the soil cover suddenly be- parts of the catchment where it is re- comes thicker in a way that commonly placed by the wide central valleys with reveals tectonic features (Christiansson even lower gradients than in the NW. 1972 p. 323). Where gullies cross the The very gentle slopes culminate in the fault the bottom level drops abruptly be- flat mbugal Nyika ya Itumba, 900 tween 1 and 5 m. m.a.s.l., in the eastern part of the Of the total area of the Dodoma catchment. catchments 75 % are situated at levels e) The Dodoma catchments in the W between 1200-1300 m.a.s.1. Only some exhibit a hilly landscape (Figs 72, 73). 10 % are made up of inselberg slopes Dominating topographical elements are over 1300 m. Another 10 % are situated steep-sided hills and ridges, steep to between 1170 and 1200 m (Fig 9). gently sloping pediments and wide part- In Imagi catchment, having a small ly alluviated valley bottoms. Msalatu and surface area, the steep inselberg slopes Imagi catchments occupy parts of the take up a large share, 51 %, while the inselberg and pediment slopes of pediments cover 46 %. The correspond- Chimwaga, Sagala, Imagi and Iseni hills. ing figures for Msalatu are 19 % and As the reservoirs are situated right on 81 % respectively. In Matumbulu the pediment slopes, larger flat areas catchment inselberg slopes take up are laclung nithin the catchments. 27 %, pediments some 48 % and valley Downstream of the reservoirs the gra- bottoms 25 5% of the area. dients of the pediments decrease and the slopes merge into nearly completely flat plains (Figs 11, 12). The most prominent of the inselbergs in the area are Soils and vegetation Gobi (1504 m), Ilende (1435 m), Chim- Vegetation types ttaga (1490 m) and Imagi (1395 m). The vallers within this part of the To a superficial observer much of the vegetation of Ugogo appears "natural" in that the area covered by ongoing cultivation is small. A closer study, how- ' The term "mbuga" 1s explalried In the section on "So11 zones and assoclatecl 'iegeta- ever, reveals that the influence of culti- non: The zone of gre) ancl black cla)sn. vation, grazing and fire have made great Fig 7 Reconnaissance view of ce~ltralUgogo. Part of LANDSAT scene 1135-07133, 25 Dec. 1972. Note extensive areas of sparse vegetation (~vhitetone), particularly around concentrated settlements (surroundings of Dodo~natown. Handali, Mvumi etc.). Note also thicketlwoodland caps 011 inselbergs (dark tone), and fault lines in upper left corner and centrally in picture. Water NE of Dodoma is Lake Hombolo. Hilly area in upper right corner is Chenene hills. For identification of objects within the investigated catchments. see Fig 8.

MATUMBULU CATCHMENT MSALATU CATCHMENT

Fzg 9a-d Matumbulu, Msalatu, Imag and Ikowa catchments. Hyposographic curves (based on 1:50,000 topographical ~naps).Maps of SE Ikowa catchment (-15 % of the area) published without contours. The curve is based on the contoured area only. IKOWA CATCHMENT 5280 ft Fzg 1 I ' Dodoma catchrnenrc. Relief, drainage and location of rainfall stations Position No Name of station Year opened Year closed 1. 96.35001 Dodoma Met. Station 1910 2. 96.35002 Dodoma Reservoir No 1 1931 197 1 3. 96.35003 Mvumi Mission 1926 1965? 4. 96.35005 Dodoma Geol. Office 1936 0. 96.35006 Alliance Sec. School 1939 6. 96.35008 Dodoma Reservoir No 2 1945 7. 96.35012 WD 8s ID. Dodoma 1961 8. 96.35014 Matumbulu Dam 1962 ' Due to technical r-easons Fig 11 appears on this page. Fig 10 is on page 20 - Rood or track

Fig l0 Ikowa and Dodoma catchments. Communications and settlements (Ujamaa villages), 1972 changes and vegetation of true "climax" b) Woodland (which today in Ugogo character is rare. only exists as patches on inselbergs and The vegetation of the area (where not in the Chenene hills area) consisting of a totally removed) may be divided into mantle of trees whose crowns more or four broad categories (Pratt et al. 1966). less touch to form a light but mainly a) Bwhland (which is the most common continuous canopy over a ground cover type of vegetation) is characterized by of grasses and shrubby plants (Fig. 15). small trees of bushy habit, i.e., branch- C) Wooded grassland which is an open ing or forking from the base, with only a mixture of trees and shrubs standing in thin field layer underneath (Fig 13). tall growth of grass but not forming a Where the branches interlace and im- canopy over it. (This category still exists pede passage so called thicket is formed in parts of Ugogo but the general over- (Fig 14). grazing seems to have reverted it into

Fzg 12 View towards the E from the top of Imag hill. In the foreground is Imag dam and Iseni ridge, in the distance on the right is Msalatu dam. In the far distance beyond the plains are the Chenene hills. (Photo: C. Christiansson. Dec. 1970.) Fig 13 Deciduous bushland, Mpwap~va.E Ugogo. (Photr,: C. Christiansson, Nov. 1974.)

Fzg 14 Thicket in Ugogo. (From Busse 1908a.) Fig I5 Miombo woodland in the dry season, Chenene hills. The trees have shed their leaves, the grass in the foreground has been burnt. (Photo: C. Christiansson, Oct 1974.)

Soil zones and associated vegetation bushland. Descriptions and photo- The soils within the study area occur in graphs from the German period show catena sequences. Such sequences were wooded grassland as typical for exten- mentioned from Ugogo as early as 1894 sive areas in Ugogo.) (Fig 16). by Stuhlrnann (pp. 40, 52-53). How- d) Grassland is made up of virtually ever, the catena concept and the rela- tree-less open grassy areas. In Ugogo tions between slope gradient, drainage, this category is confined to seasonally soil and vegetation were first described inundated mbugas (Fig 20). by Milne (1935, 1947).

Fig 16 LVooded grassland, E Ugogo (cf. Fig 101). Usagara mts in the distance. (Frorn \fever 1909.) In 1954 Calton introduced a soil which is hillier than Ugogo in general, classification based on genetic character- they cover an estimated 8-10 %. istics. According to that system the soils The vegetation is of miombo type (low of Ugogo in general belong to the so- deciduous trees and bush) with among called Ugogo catena, which is briefly de- others Brachystegza and Isoberlinia species scribed as a kaolinoid red earth catena, (Temperley 1938 p. l l) or dense thicket with a black calcareous bottom member with Euphorbia and species of Acacia, (Calton 1954, 1963; Moore 1971 pp. Combretum, Commiphora and Grewia 28-29). interspersed with trees like Adansonia, A more recent classification system is Fagara, and Pterocarpus (Fig 74). The used in Atlas of Tanzania (TML 1967) field layer is sparse. Where grasses based on that presented in Soil Map of occur they are generally species of Aris- Africa (D'Hoore 1964). tida, Eragrostis, Hyparrhenia, or Setaria During the course of preparation of (Vageler 1912 p. 45; Burtt 1942; Milne this thesis the FAO Soil Map of the 1947; Skerman 1968 p. 9; Mathew 1974 World, Vol. VI: Africa, was published p. 2; Lind & Morrison 1974 p. 17; (FAO-Unesco 1977). However, the Stromquist 1976 p. 24). circumstances have not permitted adaptation of this chapter to the FAO-system The red soil zone and as there is as yet no generally ac- cepted scheme of soil classification for Dark red soils occur within a narrow East Africa the presentation of soils in zone at the contact with the skeletal soils the present work is primarily based on on the uppermost part of the pedi- the simple system used by Lind and ments. This transitional area between Morrison (1974) and the more elaborate wooded areas on the hill tops and system used by Scott (1962, 1972). dominating grass and bushland on the pediments generally supports Acacza - Albizia - Combretum - Commzphora - Lan- The skeletal soil zone nea associations. Dark red soils also The exposed bedrock on the upper occur in other situations, particularly in parts of the inselbergs gradually disin- association with basic dikes and patches tegrates through weathering. The of amphibolite and meta-dolerite. How- weathering products form a narrow ever, those soils do not belong to the zone, with badly developed soil hori- normal catena. zons, immediately below the outcrop- Pale red soils are widespread on the ping parts (Scott 1972 p. 95; Lind & pediments. The pale red soil zone is Morrison 1971 p. l7 l ; I'ageler 1912 p. dominated by the coarse, easily eroded, 45). slightly transported soils which form the The most easily weathered rock is the fan slopes around the inselbergs and hill amphibolite. The gneiss, in turn, weath- ranges. These soils are generally red- ers more easily than the granite. The dish but may also be yello~lish or granite forms a coarse sandy material brownish on the surface. The colour while gneiss and particularly amphi- depends on the concentration of iron bolite disintegrate to finer fractions. oxides. The skeletal soils cover only limited In general these loose deposits have a areas-some few percent of Ugogo. thickness of 112 to a few metres but they However, they occur on the upper may in extreme cases form layers of 20 slopes of practically every inselberg or m or more (Fawley 1954 p. 2). With hill complex. Within the study area, increasing distance from the hills the Fig 17 Severe overgrazing near watering point, N of Msalatu dam. Effects of grazing as well as trampling are evident. Grass left underneath thornbushes only. (Photo: C. Christiatlssotl, Oct 1971.) clay content in the soils increases, the Comtniphora spp., Combretum spp. and soil becomes paler and may change to Acacia mellifera, Boscia grandiflora, brownish-yellow with pale red patches. Dichrostachys glomerata, and Sapium In most cases the red soils cover layers bussei (Mathew 1974 p. 4). Single stands of grey to yellow sediments which are in of Acacia tortilis and Adansonia digztata some localities exposed below the red occur. soil zone. James (1950 p. 2) terms these The field layer consists of short-lived soils "hillsands". annual grasses and herbs. Themeda spp. The soils of the red soil zone are poor and Hyparrhenia spp. are the most com- in plant nutrients (Skerman 1968 p. 1; mon. Cenchrus ciliaris, Heteropogon contor- cf. TMWD 1974, App. I). However, tus and Panicum maximum also occur. within a narrow zone on the lower parts Around settlements, induced growth of the slope brownish soils may form. of regenerating bush occurs. Here They are supplied with water from the species of Boscia, Calotropis, Combretum, upper parts of the slope. They have a Cordia, Hibiscus etc. may be found. fairly heavy texture and are relatively fertile. Rather high concentrations of The hardpan soil zone calcium may occur, excess calcium qv- ing rise to concretions in the subsoil. When material is transported down- According to Vageler (1912 p. 40) the slope through natural or accelerated red soil zone covers less than 10 % of the erosion a sorting of the particles takes total area of Ugogo. place, which on the lower part of the The vegetation within the zone has slope may give rise to a zone of undergone such strong modifications as "cemented" sandy material, generally a result of heavy grazing and cultivation grey in colour. that distinction of a "natural" vegetation Due to the specific hydro-chemical is impossible (Figs 17, 44). Common conditions in this zone the clay material species in this environment are: is dispersed and the separate particles Fig 18 Gullies cutting through a surface soil crust. (Photo: C. Christiansso~~.Dec. 1971.)

find their way down through the pro- Because of their hardness the crusts file. At various levels they get stuck, themselves are very resistant to erosion. block the pores and act as cementing However, gullies may cut through such agents. The resulting hardbedded de- crusts and they will then readily stand posits are called hardpans or crusts with vertical walls (Fig 18). (Stuhlmann 1891 p. 52; Milne 1936 p. The crusts do not crack on drying, 10; James 1950 2; Handley 1952 h. nor do they disintegrate when wetted. 2). Thus infiltration is efficiently impeded. The deposits are composed mainly of Rain rapidly runs off on the surface. grains of quartz and feldspar, clay and The moisture conditions thereby be- sometimes limonitic nodules. Conglo- come extremely unfavourable and plant meratic bands are common at their base. species found within this zone are often An analysis by Fawley (1954 p. 4) of of a more arid type than in surrounding such material showed that it consisted of areas. Wherever these crusts are present approximately 40 % fine sand and 60 % the loose deposits above are shallow and kaolin. Permeability tests of similar ma- of little value for agriculture because the terial showed a very low permeability lower layers of the profile will not allow (Fawley 1956b p. 73). They are wide- satisfactory root development. Further, spread around the foothills in the Do- the amount of nutrients is low and the doma area from where the); were first top soil is extremely liable to erosion. described by Vageler (1912 p. 49). They The most common type in Ugogo is dip off the hills towards the valley the light grey argillaceous type where bottoms. Their average thickness is 1-2 clay mineral forms the cementing agent. m, but according to James (1950 p. 2) Nevertheless, crusts with calcium, silica they may be up to 30 m in some localities and iron as cement are also common (Figs 19, 53). (Morgan 1973). When the crusts are uncovered, by The vegetation is commonly made up the removal through erosion of the up- of thicket or more widely spaced clumps per horizons, they emerge as sterile of thorn bush with bare ground or patches sometimes of considerable size. sparse grass in bet~veen(Figs 19, 53). The thicket may consist of very dense reaches the lower parts of the catena stands of shrubs through which single and which is usually rich in bases. Some Euphorbias or Baobab trees project. of the calcium which is absorbed onto Generally the vegetation of the field the humus colloids gves rise to a dark- layer survives on the moisture that may coloured complex which is responsible be present in the upper horizon. Some for the colour of the valley soils (Dudal of the shrubs, however, have roots that 1965 p. 72). Calcium in excess of that are able to penetrate the hardpans (Lind absorbed onto the clay humus complex & Morrison 1974 p. 173). usually precipitates in the soil to form Where the vegetation is not altered by concretions of calcium carbonate (Lind man Combretum-Grewza associations with & Morrison 1974 p. 173). The common common occurrences of Acacza spp. name in Tanzania for these open flat dominate. Dzchrostachys glomerata 1s verv stretches of country that make up the common. Species of Albzzza, Boscza, Cas- lowermost member of the typical Ugogo sza, Commzphora arid Strophanthu may also catena is mbuga (Fig 20). A local term in be found mat he^\ 1974 p. 3). Kigogo is "nyika" ~vhichindicates an un- Other common species encountered inhabited marshy grass-pan (Rigby in these areas are: Termznalza stuhl- 1969b p. 116). mannzz, Zzzzphus mucronata and Croton An mbuga profile may contain more SPP. or less sandy clays and marls and sometimes fairly solid limestone. Pure sandy beds occur in places along the margins The zone of grey and black clay of the depressions and along drainage The majority of the fine fractions trans- courses. ported dow nslope are eventually depo- Mbuga deposits of thicknesses of over sited on the valle) floors and in the de- 100 m have been recorded in Cgogo presslons \\here the\ come to form (Fa~vley 1959. Coster 1960, TMWD dark-coloured clal e) soils. The dark 1974). colour is not primarilj due to a high The porosity of the mbuga soils is in- content of organic matter but originates itially high because of the cracking In the runoff and seepage water that character of the material. However, Fig 20 Dry season view of mbuga area. Cattle herds are grazing on the flats. (Photo: C. Christiansson, Nov. 1974.) when the soil gets soaked, it swells, the parts of mbugas the salt-indicating Aca- porosity decreases and the drainage be- cza stuhlmannz grows (Stromquist 1976 p. comes poor. During the rains the 23). Far out on the mbuga flats only the mbugas are waterlogged or inundated specially adapted Acacia seyal thrives, for longer or shorter periods. Thus, apart from grasses and herbs. Main much water is lost by evaporation. species of grass found in mbugas in- Sometimes so called "gilgai relief' clude: Erapostis superba, Heteropogon con- may be seen on mbuga surfaces. This is tortus, Hyparrhenia spp., Pennzsetum, due to alternating shrinking and s~vell- Sporobolus spicatur and Themeda trzandra. ing of the material and with loose mate- Even Sorghum spp. may be found rial falling into the open cracks heaving (Mathew 1974 p. 1). into a pattern of low mounds when the soil swells (Dudal 1965 p. 3). The ~egetatlonon mbugas in Ugogo Forms of erosion is dominated b~.grassland with fe\\ scattered shrubs and trees. Along some of Agents of soil erosion may he wind or the streams emptving into the depres- \vater. In the semi-arid environment of sions rlrrerine kegetation including Ugogo, both are very effective because doum palms (H~phaenacorzacea) occur. of the long dry season which weakens At the fringes of the plalns Acacza spzro- the protective function of the vegetation carpa is 0bser.r ed as well as Acacza drepa- cover. nolobzum, Acacza mellfira, A~acza tan- Man-made grass and hush fires, over- gan~zkenszs and Bauhznza thonnzngzz grazing and burning of mulch and (hlathew 1974 p 2) In poor11 dralned stubble after harvest aggravate the sit- Fzg 21 Soil pedestals formed through splash erosion and sheet wash follo~ilng overgrazing. (Photo: C. Christiansson, Dec. 1971.) uation (Fig 28). They all further reduce aggregates are destroyed by splash the protective cover and they also cause effects. a reduction in the humus content and in Although wind erosion is common in the biological activity in the top soil. the area under study (Ann. Rep. Agr. D Thus the formation of aggregates (peds) 1938 p. 42; Rigby 1969b p. 158; Chris- which assist in creation of a favourable tiansson 1979 p. 138) it is not as prom- soil structure is obstructed and existing inent as the various types of erosion by

F2g 22 Sheet \?ash and rilling on upper pediment, W of Imae hill. Surface layer of quartz pebbles resistant to erosion. In the foreground are small anthills (cf. Fig 47). The far advanced degradation of the vegetation and the soil is noteworthv. (Photo: C. Christiansson, Dec. 1972.)

2 8 Fig 23 Rill formation along cattle track, Irnagi catchment. Grass and herbs have been removed through grazing and trampling by stock. Runoff \\-a- ter follows the track downslope and initiates erosion. The bush canopy prevents serious splash effect . (Photo: C. Christians- son, Dec. 197 1 .)

water. These are: splash erosion, sheet of soil; and step erosion, which may occur erosion, step erosion, rill erosion, gully where layers of different density im- erosion and tunnel erosion. pede the free infiltration of water down Splash eroszon is caused by the direct into the soil. Then subsurface water impact of rain drops on the soil (Fig 2 1). flows down the slope on top of a more The splash makes the fine soil particles or less impermeable layer towards a cut start moving. It also packs the soil in the profile. Material is removed from surface and seals the pores, thus de- the face of the cut and thus the step creasing the infiltration rate and in- retreats upslope. creasing surface runoff. Forms of run- Splash and sheet wash detaches, off erosion common in the Dodoma transports and redistributes top soil area are hee et eroszon (Fig 22), which is downslope. It generally operates selec- the uashing amay of a thin surface layer tively, carrying away first the finer Fig 24 Gully erosion, Mkonze catchment. (Photo: C. Christiansson, Dec. 197 1.) particles and lighter organic matter accumulation or soil burial generally most valuable to plants. The soil is thus occurs on lokver slope segments. impoverished and thinned on the upper The channel erosion forms of runoff segments of the slope while concomitant erosion are rill erosion, gully erosion

Flg 25 Badlands. Ko~ldoa.(Photo: C. Christiansson. Dec. 1971 .)

3 0 Fig 26 Tanzania. Mean annual rainfall

and tunnel erosion. Rill erosion occurs the soil in cracks or insect passages. It where surface runoff concentrates in then seeps downslope along under- minor natural or artificial depressions ground zones of weakness, sorting out along a slope. It creates small furrows of certain fractions which are removed. some centimetres or tens of centimetres Thus tunnels are created. When the in depth (Fig 23). Long parallel systems roofs of the tunnels collapse the system of closely spaced rill channels are re- is transformed into an open gully sys- ferred to as shoe-string rills. tem. Gully erosion is a later-developing stage of rill erosion. Gully heads often migrate up the rill channels. Gullies are Climate considered to have formed when the furrows cannot easily be obliterated by The whole of Ugogo is located within a ordinary tillage implements. They may belt cf dry areas that extends from the be several metres in depth (Fig 24). NE towards the SW through the central Areas seriously affected by gully erosion part of Tanzania (Fig 26). On general- are gradually transformed into "bad- ized climatolo~calmaps the area falls lands" (Fig 25). Tunnel erosion has been ~vithinthe sub-humid tropical zone. observed but it is not common within However, a closer look at the climate the areas under consideration. It occurs reveals that it is of semi-arid character. where surface crusts overlie more easily Morgan (1973 p. 43) classifies it as a eroded layers. The runoff water enters "Dry season savanna climate" (mean Tablr l. C:limatological statistics Dodoma Meteorological Station, no 96. 351001. Lat. 06"101S, long. 35'46'E. Altitude: 1,119 m (3,670ft) a.s.1. (Sources: Heidke 1913, EAMD 1922-29, EAMD 1966, Woodhead 1968, TMWD 1974, EAMD 1975, Rodhe & Virji 1976.)

Month Atmospheric Temperature (1932-70) Relative Wind pressure humidity speed (1932-70) Means Extremes (193670)

1200 GMT Max. Min. Range Highest Lowest 0600 GMT 1200 GMT 1200 GMT (mb.) ("C) ("c) ("c) ("C) ("C) (%) (%) (knots) January February March April May June July August September October November December Year 888.3 28.9 16.4 12.5 36.4 7.6 7 5 44 8

' Penman's method. Mean annual rainfall for the period 191 1-70 is 566 mm according to EAMD 1975. According to calculations based on Heidke 1913, EAMD 1966, TMWD 1974 and Rodhe & Virji 1976 it is 553 mm.

annual rainfall 500-1000 mm, 5-6 followed by a generally drier period months during the year with a total (Uhlig 1907 p. 497; Vageler 1912 p. 28, rainfall above 50 mm). The characteris- 71; Jager 1913 p. 180; Obst 1923 p. 123, tics are, however, closer to a drier semi- 192; Baker 1977 p. 76). Baker mentions arid type of climate, which would agree the possibility of 10- or l l-year rainfall well with Koppen's classification where- cycles in East Africa while Mnyawonga by it should be regarded a "Tropical (1973 p. 13) states that in Ugogo "there steppe climate" (BS) (Liljequist 1970 pp. is a six-year cycle from one drought pe- 164-170). It should be noted that the riod to another". However, no evidence limited rain falls during the hot part of of any general climatic trend can be ob- the year, causing high evaporation loss- served during the period covered by es. If the water balance is used as the climatic records. main criterion for classification of the Statistical data on the climate of the climate, Ugogo is pinpointed as the area are presented ill Table 1. heart of the drylands of Tanzania. As a matter of fact it should then, by defini- Rainfall tion, be classified as arid rather than semiarid (Pant & Rwandusya 197 1). Because of its margnal nature rainfall is The possibility of short term climatic the most important single climatic ele- fluctuations in Ugogo and the sur- ment in semi-arid Tanzania. It is a rounding areas has been suggested by factor of greatest interest to all cul- several authors, most of them emphasiz- tivators and herders, i.e., the vast ing a period of unknown length of wet- majority of the population in Ugogo. ter conditions prior to the early 1890s The coming or not of the rain may Y Evapor- Cloud Rainfall (l91 1-70) Rain- 1run ation' (195544) days i5-64) (1955-64) (1931 -60)

y aver. Mean Mean % of mean Highest Lowest Max. 24 hour fall Mean 1 (mm) (oktas) (mm) annual (mm) (mm) (mm) no

mean life or death to humans and conditions in the atmosphere, the varia- livestock, and the amount and intensity tions in the precipitation become signifi- of the rain is of decisive importance for cant and the patterns of rainfall from soil erosion as well as for crop yields. day to day often appear unrelated The general rainfall patterns of the (Johnson 1962). area are determined by the movements In 1971 there were 6 meteorologxal of the Inter Tropical Convergence Zone stations in Ugogo and an additional 20 (ITCZ) between the northern and the stations where only rainfall was record- southern hemisphere. Thus, the year ed. Between 1971 and 1974 the station may be divided into two distinct seasons, network was extended (TMWD, 1974 V one dry season lasting from May to p. 2). Rainfall measurements in Ugogo November and one rainy season from date back to 1898 when the recording December to April (Table 1). station at Mpwapwa was opened (Jack- During the dry season high pressure son 1968). At Dodoma rainfall has been conditions dominate over E Tanzania recorded since April 19 10 (Vageler and gve rise to dry weather in Ugogo. 1912) (Table 1 and Fig 27). A high cor- Steady south-easterly winds blow from relation (r=0.92) has been found to the subtropical belt of high pressure to- exist between Dodoma Met. Station and wards the equatorial belt of low the rainfall of the rest of the Dodoma pressure, which at this time of the year is regon (TMWD 1974). Consequently north of the equator. the climate of Dodoma may be used as a During the period December to April representative indicator of the climate unstable conditions in the upper air of the whole of Ugogo. pressure field give rise to wetter condi- The average annual rainfall in central tions in Ugogo. Due to the changeable Ugogo amounts to between 500 and laooi Rainfall year Fzg 27 Rainfall at Dodoma Meteorolog~calStation 192 1-1977. During the period of record, 9 seasons (13 calendar years) experieced a rainfall 20 % below normal or lower. During the same period 10 seasons (10 calendar years) got a rainfall 20 % above normal or higher. The diagram well shows that rainfall figures referring to calendar year may be greatly misleading. For the agriculture in the area it is the rain that falls during one growing season (one rainy season) and its distribution in space and time that is of greatest importance.

600 mm. Parts of E Ugogo, around quent crop failure (See page 46).* Of Mpwapwa, receive around 700 mm importance in this context are also the while Mtera in the extreme S receives recurring dry spells within the rainy sea- only 450 mm (Stromquist 1976 p. 16). son (Fig 7 l). These are very destructive The average rainfall at Dodoma Met. to most crops and may reduce a promis- Station is 546 mm (55 years). The total ing harvest to nothing. precipitaiton varies strongly from one The highest monthly rainfall is re- season to another (Figs 27, 71). The corded in January, which generally re- lowest rainfall recorded during one sea- ceives between 100 and 150 mm al- son at Dodoma is 221 mm, the highest is though monthly totals of over 400 mm 1083 mm. have occurred. The highest daily falls Rainfall reliability is low (Jackson are also recorded in January. Rainfall 197 1 p. 28; Nieuwolt 1973). Over large exceeding 75 mm in one day may be parts of the area an annual rainfall of expected once in two or three years less than 500 mm may be expected in at (Rapp, Murray-Rust et al. 1972 p. 263). least one year out of five (Griffiths 1972 p. 109). Rain in the district seldom falls evenly over a wide area but commonly occurs as scattered local showers or Total ralnfall in the 1972173 season recorded at Dodoma Alrport !\as 637 mm, cloudbursts of short duration. Practi- 19 R abo\ e ax erage In the same season the cally every year some part of Ugogo Hombolo area 20 km to the NE experienced experiences local drought with conse- a sex ere drought. Temperature, evaporation, humidity The relative humidity is low. Early Like precipitation, temperatures also morning values may be in the range df shov seasonal variations although the 84-93 % but in the afternoons strong transitions are not as sharply marked. winds in combination with frequent The months June-August form a bright sunshine bring the humidity somewhat cooler period compared with down to low values. The average mid- the rest of the year. The coldest month day minimum of 34 % is reached in is July with an average temperature of October (TMWD 1974 V p. 18). 19.8"C, the hottest is November with The number of sunshine hours 24.8"C. The average temperature for amounts to around 3000 per year the year is 22.7"C (EAMD 1963). (Griffiths 1972 p. 116). Kongwa in E On approximately 40 days each year Ugogo experiences an annual mean of midda) temperatures exceeding 32" are 8.4 hours of sunshine per day. A maxi- recorded (EAMD 1963 p. 7). The high- mum of almost 10 hours per day occurs est temperature recorded in the period during the months September-Novem- 1932-1970 at Dodoma is 36.4"C and the ber (EAMD 1975 p. 12). lowest 7.6"C. The daily and annual tem- During the rainy season dew occurs. perature amplitudes amount to 12.5"C The mean annual dew-point lies be- and 5.0" respectively. tween 16-17" (Griffiths 1972 p. 115; The potential evaporation is high. EAMD 1975 p. 12). At Mpwapwa hail When the driest period sets in and the has been observed on a few occasions force of the wind increases considerably, (Heidke 1913 p. 64; 1923 p. 103). evaporation also increases and reaches its highest monthly average in October. The average annual potential evapora- Wind tion at Dodoma is 2123 mm and at Kongwa 1873 mm. At Mtera the annual Observations of the directions and evaporation amounts to 3261 mm forces of the wind exist only from a few (Stromquist 1976 p. 16). stations in Ugogo (Vageler 1912;

Tabl~2. iVindspeeds in Ugogo. (Monthly averages in km/hr. Recordings 2 m above ground.) Dodorna Met. Station is in central Ugogo in flat countrj-.Kongwa is in eastern Ugogo in open flat country. Ikombo is in northern Ugogo in hilly and well vegetated country.

Station Dodoma hlet. Station' Kongwa Pasture Res. Station2 I komboJ Time 6 a.m. 12 a.m. 6 a.m. 12 a.m. -

Januarx Februar) March Aprll 51a) June July August Septemhe~ October No\ ember December

' 1936-70 Sources: EAMD 1975: TMWD 1974 I 195446 Period of measurernent unkno~cn Fig28 "Normally the visibility is good but smoke haze from bush fires and dust in the air restrict it toliards the end of the dry season. The haze reaches altitudes of about 3,000-3,500 m.a.s.1." (EAMD 1963 p. 7).

Bush fire near Kilimatinde. (Photo: C. Christiansson, Dec. 1970.)

Heidke 1923; Fawley 1958b; TMWD and often in association with the intense 1974; EAMD 1975). thunderstorms that frequently occur in In the Dodoma area the winds blow the rainy season. However, strong winds from the E or SE throughout the )ear. of longer duration are recorded mainly From April and onwards the force of during the dry season. The possible the wind increases progressively and utilization of wind-powered generation during the period August to November of electricity and wind-powered pump- when steady south-easterly winds pre- ing for rural water supply is mentioned vail the highest daily averages are meas- by Vageler (1912 p. 21, 84), Wade and ured (Table 2). At Kongwa, ~vheregen - Oates (1938 p. 8) and TMWD (1974 V erally lower windspeeds are measured p. 15). The force of the wind increases than at Dodoma, the pattern of wind gradually during tht dry season, which direction is more varied and during the is of importance when water has to be rainy season the direction is particularly pumped from successively greater variable with frequently occurring NE depths (Heidke 1923 p. 97; Fawley and NW winds (Heidke 1923, p. 97, 1958b; EAMD 1975 p. 6, 12). 109; EAMD 1975 p. 67,71). Over hot vegetationless ground so- Apart from seasonal variations the called "dust devils" often develop in the speed of the wind aaries according to dry season. These are whirlwinds of time of the day and topographic loca- limited areal extent but of sufficient tion of the observation point. The high- power to carry dust and larger particles est forces are recorded during da) time high into the air (see page 150). Drainage found it half submerged and brawling streams of yellow water rushing towards the All streams in Ugogo are ephemeral. north." (Stanley & Neame 1961 p. 34). Flow occurs only during and soon after When there is flow in the streams large heavy rainfall (Fig 49). It starts suddenly amounts of sediment are transported as and may for some few hours be very bed load and in suspension. The sedi- intense (Obst 1923 p. 129; Wade & ments are successively deposited in the Oates 1938 p. l l). stream channels and sometimes fill "About three miles to the south-tvest, we ob- them up completely (Dantz 1902 p. 57) served a thick fog, and knew that rain was or spread as fan-shaped accumulations falling, but we only received a few drops. onto adjoining land (Figs 30, 99, 100). Half an hour later, a broad and dry sandy The proper channel bottom, under- stream-bed (Kikombo river), in which we had commenced to dig for water, was trans- neath the sand, is generally formed by formed into a swift torrent 18 inches deep the bedrock surface or some imperme- and 50 yards wide, the general direction of able layer in the soil profile. Over that which was north bj east. Within two or three layer ground water seeps long after the hours, there were only a few gentle threads of water remaining; the torrent had subsid- surface flow has ceased, commonly right ed as quickly as it had risen." (Stanley 1878 through the dry season (Grove 197 1 p. p. 100). 15). These water potentials are used by the When these flash flows rush down the local population who collect the water sides of the hills they often set extensive through wells dug into the dry stream areas of the lowlying plains under water. beds. Irregularities in the substratum "From Dudoma we travelled through a may obstruct the seeping water so that dense bush clothing a slightly- elevated ridge. the water surface is raised and small from which after a couple of hours we pools may be formed on the sandy emerged amid a terrific rainstorm in view of the basin of Zingeh which as we looked at it stream bed. These pools indicate the from under our rain-dripping helmets occurrence of more water at some depth seemed a lake. Descending into the plain we below the surface (Coster 1959 p. 45). 6. Population and land use

Population. Immigration from the time before 1920 are not available. Contrary to official opinions into Ugogo provided by the colonial powers, Kjeks- hus (1977), supported by Kuczynslu The average population density in (1949), reckoned that there was a con- Ugogo is low, but there is a marked un- siderable decrease in population all over evenness in the distribution. Large East Africa in the period 1890-1920. areas, particularly in the N and in the S This is evidenced by early German are virtually uninhabited while central population statistics, the reported Ugogo is part of a zone of more densely effects of the rinderpest (Stuhlmann populated areas that extends from the 1894), severe drought years, locust in- central coast towards Lake Victoria. vasions (Deutsches Kolonialblatt 1895 p. Another densely populated zone 100) and effects of the 1914-18 war extends from Tanga in the NE to Lake (Brooke 1967). However, most probably Nyasa. The two zones of higher density this trend started much earlier with the cross each other in Ugogo (Fig 29). spread of new diseases to populations Reliable population data for Ugogo along the trading routes (Burton 1860, I

I \ l Fzg 29 Tanzania. Population densitv 1967. p. 165; Stanley 1872 p. 533; Becker newcomers (Claus 191 1; Obst 1923; 1899 p. 761; Mkunduge 1973 p. 67). Schaegelen 1938 ; Carnell 1955; Oliver In 191 1 the population of Ugogo 1967; Rigby 196913). (then 22,4 10 km2) was estimated at some Kimambo (1969) has termed those 125,000 or 5.5 inh/km2 (Vageler 1912 p. parts of W and central Tanzania in 80; DSAS 1914 pp. 4041; Obst 1923 p. which the Wagogo oripnate "the mtemi 260). By 192 1 the number of inhabitants region". Common for this area is an had decreased to between 90,000- economy based on an uncertain pro- 100,000, due to hardships during the duction of grain combined with cattle 1914-18 war and a dramatic influenza herding. Traditionally the central epidemic followed by a terrible famine power in each local society was personi- in the years 1919-20 (Brooke 1967 p. fied by a leader who was responsible for 15; CDA 1975 p. 27). By 1934 the pop- the religious rituals and for certain relation in Dodoma district (then 17,150 ligous symbols (Bull 1973 p. 88). No km2) amounted to 130,000 or 7.6 central political power existed. inh/km2, of whom some 5,000 lived in In the N, S and E the mtemi repon is the township of Dodoma. Thirty-three surrounded by high rainfall areas with a years later, in 1967, the figure had risen sedentary population of farmers. When to 32 1,000, the mean population density the population in the mtemi region in- being 19.4 inh/km2 (rural areas only: creased, expansion was possible on11 18.0 inh/km2). In 1977 the population towards the central parts of present of the district was estimated at 354,000 Tanzania. Here a number of chiefdoms or 21.5 inh/km2 (TMA 1977). came into existence, the members of During the period 1957-67 parts of which were successively regarded as one the district showed a very marked in- ethnic group, the Wagogo. For details crease in population. The average den- about the origin of the Wagogo and the sity in 1957 in Dodoma Makulu division routes of immigration into Ugogo, see was 19.5 inh/km2. By 1967 it had risen Rigby (1969b). to 35.5 inh/km2 (Berry 1971b p. 3). In- Down through Central Tanzania a creases of similar magnitude were re- wedgeshaped corridor extends with corded in the neighbouring Chilonwa very low and irregular rainfall (Fig 26). and Mvumi divisions which, in 1967, In the 18th and 19th centuries this area had average population densities of 34.4 teemed with game for the hunter and and 4 1.2 inh/km2, respectively. In the grazing was also sufficient for Ikowa catchment ~vhichforms parts of nomadic pastoralism. Tsetse was non- the above divisions the increase was existent (Burton 1860 I p. 276; DSAS 39.4 % in the years 1960-67 i.e., 5.6 % 1900 p. 95; Vageler 1912 p. 78). For the per annum. (See further Chapter l5 pure cultivators, however, it was of little and Tables 23, 24). interest due to the margnal rainfall and When the present inhabitants of infertile soils. The cultivators avoided Ugogo, the Wagogo-a bantu-speaking these parts of the country as areas suited people of mixed ethnic origin, moved for arable farming existed both E and W into the area in the second half of the of the dry corridor. 18th century it had long been populated For the Wagogo, being cultivating by small groups of people (Mankala, pastoralists, this environment was Ngulimba) who lived by hunting and acceptable. But they arrived in Ugogo gathering. Being inferior in numbers to relatively late and the Maasai were the immigrants the orignal inhabitants already moving towards the same area. were dricen away or assimilated by the This resulted in warfare that also involved the Wahehe in the S, and that did Spatial organization not end until around the turn of the 19th century (Herrman 1892 p. 200; The population of Ugogo nowadays Schele 1896 p. 76; Meyer 1909 p. 167; lives in villages. However, until the be- Carnell 1955 p. 26; Bujra 1974 p. 2; cf. ginning of the 1970s the Wagogo, like Kjekshus 1977 p. 12, 19). the majority of the rural population in These circumstances reduced the Tanzania, lived in dispersed farmsteads possibilities of further expansion for the (Fig 30). Each farmstead, or homestead, Wagogo. They were instead, at least for consisted of one or more production certain periods pushed into the hilly units (households),' (Table 27). The areas in the central parts of their pre- members of the different production sent area of distribution. This heavy units within the homestead formed a concentration of people and livestock homestead group with well defined in- had devastating effects on the environ- ternal organization (Rigby 196913). With ment. some modifications this internal organi- Periodically for some 10 years from zation as well as the physical appearance 1883 all S Ugogo as far N as the Do- of the homesteads are still the same as doma hills was occupied and devastated before the villagzation (Bujra 1974 p. by the Wahehe (Meyer 1909 p. 267). 9). Man) of the attacked Vagogo fled N to The homestead (village unit, kaya) the more powerful chiefdoms around consists of one house (itembe) or a the Dodoma hills (Fig 103). However, group of houses. The general shape is a here the numerous refugees caused square or a rectangle. The building ma- overpopulation which in turn resulted terial is a wooden framework plastered in rapid land degradation and repeated with mud. Together the buildings famines. surround, wholly or partly, a courtyard or boma where cattle are kept at night "Repeated attacks from the Wahehe to the south and the Wahumba (IvIaasai) to the (Figs 30, 32). north have reduced them in numbers, and Traditional concepts of spatial or- compelled them to seek refuge on the hill- ganization beyond the homestead are su~nmits."(Stanley 1878 p. 99). chiefdoms (ritual areas) and neighbour- It was not until the German authorities hoods. Traditionally the head of the intervened in the 1890s that the fighting chiefdom, the mtemi (sultan, jumbe, ceased (Gwassa 1969 p. 93) and the Wa- chief) led the seasonal rituals of gogo were able to move out on the rainmaking and fertility within his area. plains (Fonck 1910 p. 275; Schaegelen Moreover, he had the power to prevent 1938). A similar sequence of events is certain resources from being used for described by Owako (1971 pp. 180-18 1) certain periods and to close off areas from the Machakos area in Kenya where within his chiefdorr, from exploitation the M'akamba, also orignally cultivating of various types. But he could do so only pastoralists, were pushed into the hills for reasons of religous nature (Rigby by the Maasai, probably already in the 1969b p. 99). second half of the 18th century. It was only with the establishment of British ' hlany socio-economic surveys mix up the rule that they again moved down and term production unit ~viththe European- settled on the surrounding plains. oriented concept of core family. Families in the European sense are uncomrnon among the Wagogo. The problem of definition has been further developed by Egerii & Henin (1973 p. 163). Fzg30 Aerial view of part of Matu~nbulucatch~nent. The angular structures are homesteads. Dark boundaries of the smaller fields arc thornbush fences. Such fences are frequently used to enclose garden plots to protect them from grazing animals. The lower left area is fallow fields used for grazing. Note the occurrence of sheet wash and channel erosion features. To the right are fields under cultivation. The upper left part shuxvs a big sand fan. r\-atel-conditions are favourable on the fan. It is thus used for grorving of vegetables etc. The view is typical of thc so-called cultivation steppe areas in the period before the villagiration. (cc. Figs 54 and 100). (Photo: D. H. Murray-Rust, March 1970.)

Thus, single trees or bushes left in the vaguely defined areas with no definite terrain may have a ritual function. Only borders. Common to the inhabitants of the mtemi may decide whether these one neighbourhood is "a complex of trees may be cut down or not. When it cattle transactions which . . . link them comes to other trees or bushes the in a network of common rights and mtemi has no power. This explains why obligations in all other transactions" it was so difficult for the colonial au- (Righy 196913 p. 52). thorities to persuade the Wagogo to carry out communal ~vork,for instance Residential rnobility soil conservation work (Ann. Rep. Agr. D 1960 p. 18, cf. Krokfors 1978 p. 204). Before the forming of villages in the The single unifying authority was the early 1970s the M'agogo were residen- mtemi but he only had power over and tially mobile. The movements generally interest in the religious sector. originated in scarcity of land suitable for Each chiefdom consists of a number cultivation or decrease in yields due to of neighbourhoods (itumbi). These are exhaustion of the soils (Ngaula 1976b p. 132). In older times threats from neigh- new fields explains to Gogo the evident cycli- bouring people were reasons for the cal rise and fall of population in specific movements as well as the drying up of neighborhoods, but individual moves are almost always the result of a search for graz- wells and waterholes and search for ing, or to avoid famines or witchcraft ac- food in times of famine (Kannenberg cusations. This is primarily due to the fact 1900a, b; Fonck 1910 p. 275; Vageler that land for cultivation is hardly ever scarce 1912 p. 72). Along the caravan routes and a very low value is placed upon it. In- heriting of land appears ridiculous to Gogo." harasses and repeated thefts of food (Rigby 1969b pp. 1 18-1 19). caused large areas to be deserted (Fonck 1910 p. 281). Witchcraft accusations have also been stressed as an important Land use reason for movements (Rigby 196913 p. 117; Mnyawonga 1973 p. 23; cf. Owako Cultivation 1971 p. 188). However, the primary Generally spealung, Ugogo is unsuited reasons behind the residential mobility for growing of crops, due to the low and of the Wagogo are of ecologcal nature, unreliable rainfall and inherently in- such as irregular supply of water, un- fertile soils. Still, cultivation forms the reliable supply of grass for the livestock most important part of the Wagogo and exhaustion of agricultural land. subsistence economy. Normally each "hfajima, a homestead oltner In Temang- household (production unit) produces hu ku had come from Musalato ten all the food it consumes. The average miles to the southeast to Temanghuku ne~ghborhooddur~ng the 1939 famine In cultivated area per household is around 1962 he returned \\~thhis dependents to 2.5 ha (5 halhomestead) divided into build a nev homestead at Musalato small and scattered plots (Fig 3 1). How Maj~masa~d that when he had moved from Musa- large areas are actually cultivated by Iato tuent\-three )Par7 before the land and bush each homestead is dependent on the size there had been exhausted, and the general fam~neof 1939 had prec~pltatedh~s change of the group and its energy resources in of res~denceHe noill mazntazned that the The- the form of the labour force. manghuku nezghborhood and ~tsresources CL1ere exhausted, tth~leat Musalato the bush had regenerated and Mas su~tablefor the clear~ng of nem fields Crops Llaj~maexpla~ned People are lear~ngth~s Traditionally sorghum and bulrush area because the land has become old It IS old because there is no longer anl bush to millet are the staple crops. hlaize is get- clear for neu field ting increasing importance and is no\$- People generall~clear a field one 1 ear and gro~vn over significant areas (Rigby cultllate ~t,the next \ear the! clear a little 1969b; Ndorobo 1973 p. 33). It requires more, and again the follo~%~ngTear Then a person \\ill lea1 e the field he cleared the \ear more rain than sorghum or millet but is before last and cult~rate last \ear's field plus not as easily attacked by birds. the nen1.i cleared and burnt field If there w In the 1971172 season millet was pnough bush he ill11 clear nrir areas each liar grolvn on 37.030 ha, sorghum on 5230 leaz~zngthe old ha and maize on 4829 ha of the total il'hen bush IS exhausted In an\ ne~ghbor- hood and the cleared fields are no longer area of Dodoma district (TMWD 1974 productire In splte of manurlng the I11 p. 3). These main crops are usually homestead population begn to d~min~sh interplanted ~cith legumes and root Homestead omners then mole to nex\ areas crops. Maize is most common in the or to old ones M here the bush has regenerated Hor\e\er, soil exhaustion and scarcit) of northern part of Ugogo ichich receives cult~rahleland are seldom the onl~reasons slightly more rain than the rest. Sor- for moles The search for bush to clear for ghum dominates in the \'V where soils (-J.. . ., ...... Field with grain crops fl Track 0Field with small crops 1Footpath or cattle track m@ Homestead

Fig 31 Example of the distribution of homestead fields and garden plots in a cluster of homesteads in Ugogo. (Kwilamba cluster, Temanghuku neighbourhood 1962. Modified after Rigby 1969b.) Grain crops (millet, sorghum and maize) are sometimes grown in pure stands but they generally occur interplanted with other crops. Small crops (including vegetables, pulses, groundnuts, tobacco, etc.) are frequently grown on old abandoned homestead sites. Numbers of fields refer to residence of present user. Fields without number are cultivated by persons from other clusters. All land at present not occupied by cultivations is used for grazing. Note that bush fields cultivated by the inhabitants of the cluster are not visible here. Waterpan = mbuga. Approximate area covered by map 180 ha. are heavy. Bulrush millet prefers light Food crops include-apart from soils. It is the dominating crop in central maize, millet and sorghum-sweet Ugogo and in the S where rainfall is potatoes, cassava, groundnuts, bam- very low (Ngaula 1976b pp. 132-133). barra groundnuts, cow peas, gram, marrows, water melons, cucumbers, etation, colour and texture of the soil pumpluns and squash. Gourds and and apparent moisture conditions. calabashes are grown to serve as pots "There is a wealth of sound knowledge and and vessels after the fruits have dried power of good judgement of agricultural and the contents been removed. Castor, matters held by the native of the cultivation ste pe " (Rounce 1949 p. l ; cf. Vageler 1912 sesame, groundnuts, tomatoes and on- p. !L; Allan 1965 p. 5). ions are sometimes grown as cash crops, but production for the sole purpose of There is no central control of land use. earning cash is (still) uncommon The head of a production unit or the (Nyange 1976 p. 135; Hamad 1976 p. head of a homestead acquires the right 136). to a piece of land through clearing it Around Bahi swamp and locally near and then actively cultivating that land. other mbugas and at Ikowa irrigation According to praxis the rights to the scheme, rice also plays a certain role as land cease when an area has been left cash crop. Cotton was tried by the fallow for two years. It is generally not Germans but the low rainfall and the possible to own or inherit land. This re- frequent dust storms prevented devel- fers both to cultivated fields and grazing opment of large scale cotton growing areas (Rounce 1949; Bryceson 1967; (Vageler 1912 p. 97). Recently wine has Rigby 196913; Conyers 1971). In certain been introduced as an alternative cash areas with high population density the crop. Cultivation of wine requires right to cultivate land may be bought for irrigation. Therefore it is primarily cash, grain or livestock (Rigby 1969b p. grown on a communal basis in ujamaa 29; cf. Manshard 1974 p. 66). villages. The original cropping system in Ugogo is shifting cultivation (Gillman 1930 p. 4; Rounce 1949 p. 25) of bush Cropping methods fallow type (medium fallow periods) Whch areas will be chosen for cultiva- (Boserup 1965). This implies movement tion depend on the distance from the of both economic area and settlement at homestead and the fertility of the soils. certain intervals (Ruthenberg 197 1 ; This is judged by observations of veg- Manshard 1974). Maintenance of

Fig 32 A M'agogo homestead in Matumbulu catchment. Honiesteatl field in the foreground, grazing on the slope in the background. The wooden fence indicates the cattle pen where livestock is kept at night. (Photo: C. Christiansson, Nor. 197 1 .)

44 Fig 33 Newly cleared bush field on upper pediment, NW slope of Nyankali ridge. The picture is taken 100 m upslope of the area shown in Fig 19. (Photo: C. Christiansson, Nov. 1974.) fertility is based on regeneration of nat- (infields) lie concentrated around the ural vegetation and to some extent on homesteads (Fig 30). They are to some use of ashes as fertilizer (Hornby & extent manured with cattle-manure or Hornby 1943 p. 45). A common practice termite mound soil. With the practice of is inter-cropping whereby, commonly, a manuring, weeds become a great prob- mixture of grains and legumes is sown lem and work-intensive weeding is in the same field. Parallel with the tradi- necessary to prevent the weeds from tional form of cultivation a semi-con- suppressing the crops. In pre-colonial tinuous type of cropping occurs: rota- and early colonial times the only form of tional bush fallow, land rotation or re- manuring that seems to have been prac- current cultivation (Allan 1965; Morgan ticed was to let cattle graze over the 1969; Manshard 1974) in which system stalks in the fields after harvest (Vageler settlements are not moved; only the 1912 p. 72; Hornby & Hornby 1943 p. cultivated area is changed. This system 45; Rounce 1949 p. 27). At that time includes the use of cattle manure which permanently cultivated fields were un- makes longer cropping periods on one common. This is one of the reasons why field possible (Rigby 1969b p. 29). It ex- the Wagogo in older days were more presses a relatively stable relationship mobile than today when the more or less between land and population (cf. Mor- permanent cultivation ties the Mgogo gan 1969 and Prothero 1972). farmer to a particular piece of land. Centred on the settlements, zones of Even today it is only on a limited scale land use of different intensity may be that the Wagogo use cattle manure distinguished. These zones begin with (Conyers 1971 p. 2). Frequently, heaps intensive garden cultivation and extend of manure are burnt (Ndorobo 1973 p. outwards to the extensive cultivation of 36). the bush (Figs 52, 72, 73) (cf. Manshard Bush fields (outfields) are cultivations 1974). Three types of fields occur: a) of temporary character (Figs 33,34).Af - Garden plots, b) Homestead fields. c) ter they ha've been cleared they are Bush fields. cultivated for a few seasons and left Garden plots and homestead fields fallow when the yields decrease. The bush fields are never manured due to better. If with this preparatory activity it their location far from houses and cattle rains, good harvest may be expected and the people can engage in festivities such as bomas. When a bush field is cleared the circumcision, folk dances and songs, and larger trees are cut approximately 1 m there is a substantial amount of drinking. In above the ground. The larger trunks are view of these good times ahead, the Wagogo saved as building material or fire wood. prepare the fields with rising expectations." Brush-wood, smaller branches and twigs (Mnyawonga 1973 p. 14). are placed in heaps and left to dry However, it is not uncommon that the whereafter they are burnt. The ashes rains dwindle away after a promising are then used as fertilizers. start. Good harvests are acquired in the first "At this stage people still have hope for the and second years when the crops profit crops if the rain returns within a fortnight. from nutrients in the ashes. The returns Nevertheless all the people especially the old then become poor and the fields are ones, begin to think of living through a abandoned and recolonized by natural famine period. They think about the sale of property, and journeys they have to make to vegetation (Hornby & Hornby 1943 p. get food, and the harshness of previous 45; Hamad 1976 p. 136). famines, such as the great famine known as If fallow periods are short, coarse 'nzala ye kaputula'." (Mnyawonga 1973 p. grasses dominate. If the fields are left 22). for longer periods a mixture of grasses, Also the spatial distribution of the rains herbs, and shrubs develop. Woody may be very uneven. Within the same fallows are a more efficient way of re- rainy season one area may receive nor- storing the nutrient status than grass- mal rains and adequate harvests may be herb fallows but grass may be enough to obtained while an adjacent area may ex- restore soil structure (Moss 1969 p. perience severe drought and total crop 221). Also, the grass fallow is easier to failure. Thus, during the 197017 1 rainy clear for new cultivation. Fallow fields season Dodoma Met. Station recorded always serve as grazing areas. As they 639.1 mm, i.e., a rainfall well above av- are continuously grazed, and often erage. During the same season only overgrazed, erosion is generally intense 15 1.2 mm (30 % of "normal") was re- (Figs 30, 35, 98). corded at Matumbulu some 12 km S of The Wagogo divide the agricultural Dodoma. year into two parts, "the rains" and "the The irregularity of the rainfall has dry period", these parts in turn are made "rain-making" an important cul- divided into a number of shorter tural and religous function (Burton periods according to the dominating 1860 I p. 302; Stanley 1872 p. 637; Obst agricultural activities. 1923 p. 283; Hartnoll 1942 p. 59; Rigby The year starts with the preparation 1969b p. 34; Mnyawonga 1973; Han- of the fields before sowing. This occurs kins 1974 p. 98). just before the beginning of the rains. Such signs as the appearance of leaves on trees and migration of birds are used Livestock and gazing to tell the advent of the rainy season. Then a period of dry sowing of millet, The Wagogo, being a semi-pastoral maize, pumpkins and melons ensues. people subsist primarily on an agricultural economy but livestock play an "Drl sowng is good for tuo reasons, one is essential role in the socio-economic sys- the a! oldlng of pests. and the second IS pre- caution In case there is l~ttleram that partlcu- tem (cf. Jie and Karamojong of N lar season The earlier that people plant the Uganda and Wasukuma S of L. Vic- Fig34 "Cattle serve as the basic 'value units' in Gogo social values. They can be used to evaluate anything from wives and children to the seriousness of an offence and the number of bags of grain borrowed in a famine year. Their value in terms of modern external market prices, or grain trading, is constantly changing. But this bears no relation to the values attached to livestock in the contexts of bridewealth exchanges, compensation and trustee-ship contracts." (Rigby 1969b p. 45). Cattle herd near Ikowa reservoir. (Photo: C. Christiansson, Oct. 197 1 .) toria). Most Wagogo "households" animals to a stock trustee. Thus, of the possess cattle, sheep, goats and poultry. total number of animals herded by a Some households have donkeys for homestead owner only a part is his own transport purposes. Livestock are never property (Rigby 1969b; Nyange 1976 p. regarded as consumption goods but can 134; TMA 1977 p. 6). be exchanged for agricultural products Some important effects of this system uithin the local economic system. In are: contrast to the situation with land, 1. A network of social and economic re- livestock are heritable property. lations, involving individuals and kin The cattle produce protein in the groups, is created and maintained. form of milk and meat. Under famine 2. If fatal droughts or diseases hit a conditions Wagogo also practice blood- particular area, part of the capital of letting. Moreover, the livestock produce the cattle owners in that area will be manure and hides. Milk production saved through the trustee system. amourlts to 1-2 1 per cow per dav during 3. The system helps to relieve certain the rainy season but drops to a areas from excessive grazing minimum during the dry season. pressure. Traditionally cattle are never slaught- A general transhumant grazing cycle is ered solely for meat. However, cattle as practiced by the Wagogo herders. Dur- well as small stock may be slaughtered ing the wet season the cattle are usually for ritual purposes and consumed in grazed near the homesteads and during connection with various ceremonies the dry season when streams and ponds (Hartnoll 1942 p. 59; Rigby 1969a p. dry up the herders move great distances 43). A great number of naturally de- in search of water and grazing (Rigby ceased animals are also eaten (Herr- 1969a p. 45; Ngaula 1976a p. 130; mann 1892 p. 196; Meyer 1909 p. 267; 1976b p. 133; cf. Ann. Rep. VSD 1938 Brandstrom 1976 p. 33). p. 32). Due to many dry season grazing An important factor with ecologcal as areas now successively being trans- well as socio-economlc inlplications is formed into semi-permanent cultiva- the "cattle trusteeship system". The svs- tions, this pattern is not as evident today tem implies that a stock optner gves as it used to be. Nowadays it is not un- temporary disposition of one or more common that stock are kept near settle- Fig 35 Overgrazed area suffering from sheer ~vashand rilling, E part of Matumbulu catchment. (Photo: C. Christiansson, Nov. 1971.) ments until the grass is completely ex- all cultivated land once the crop is re- hausted (Figs 35, 107). Few areas of moved. This means there is no individ- expansion of grazing occur as Ugogo is, ual acceptance of responsibility for the and has long been, surrounded by control of grazing or for measures to tsetse-infested areas, which under nor- improve pasture (Ann. Rep. Agr. D mal conditions are avoided by the 1955 p. 24; Ann. Rep. Agr. D 1957 p. herders (Ford 197 1). However, even 19; Skerman 1968 p. 49). these areas are reported to have been However, comparatively severe pen- used for grazing purposes in extreme alties are reported to be assessed by local emergency situations (Brooke 1967). courts for burning of good quality (For earlier and present distribution of pasture land (TMWD 1974 111 p. 13; tsetse see Kjekshus 1977 pp. 164-165 Rigby 1977 p. 100). This reflects some and TML 1967.) of the ecologcal knowledge possessed The grazing resources around areas by the Wagogo. They are authorities on with permanent water are generally too grasses and are capable of assessing the limited for the great numbers of cattle feed value of different grazing lands that gather around the watering points. and their stock carrjing capacity. Within these areas problems of over- Parts of Ugogo are taken up by belts grazing are very accentuated (Fig 17) of deciduous thicket. These are only of (Ann. Rep. Agr. D 1945; Ann. Rep. limited value as grazing resources. VSD 1961 p. 10; UNDP/FAO 1968 p. 5; Grasses and herbs are sparse and during Skerman 1968 p. 47; Johansson 1976 p. the dry season, when the bush vegeta- 67). Natural bodies of water as streams, tion could serve as a supplementary water pans and ponds are open to any- source of food, most shrubs drop their one and traditionally land for grazing is leaves (Skerman 1968 p. 18). held communally (Rigby 1969b p. 57). Within more open areas the field Communal grazing is also practiced on layer is dominated by annual grasses and weeds which grow rapidly and set Accessibility or rather inaccessibility seed early. These include species of to water used to be the most important Arzstida, Chloris, Cynodon, Dactyloctenium, guiding instrument for which routes the Digztaria, Eragrostis, Heteropogon and Se- caravans chose through the area taria. Among legumes Stylosanthes is (Stuhlmann 1894 p. 42). common. This a perennial which mul- Later, water was needed for the rail- tiplies rapidly when exposed to heavy way and the history of Dodoma town is grazing. Various species of Crotolaria the history of the struggle to obtain suf- and Indigofera also occur (Skerman 1968 ficient water for the population and for p. 7). The drier mbugas support good the railway (Fawley 1956a p. 62; Chris- quality grass of high grazing value and tiansson 1979 p. 122). play an important role as dry season Burton considered the land to be grazing areas (Ann. Rep. Water Dev. condemned to perpetual drought and Dept. 1948 p. 5). The grasses of the meant that nowhere was water either mbugas are often perennials and in- good or plentiful. He vividly described clude Cynodon plectostachyus, Themeda the problems. triandra and species of Echinochloa, Pani- cum, Pennisetum, Sporobolus and Sorghum. "At 9 p.m. we sighted the much talked of The more frequently inundated mbugas Ziwa. The Ziwa (pond) which is 3100 feet above the sea, occupies the lowest western are vegetated by tall grasses such as Digz- level of Marenga Mkali and is the deepest of taria and Thelepogon but these rapidly the many inundated grounds lying to its N, run to seed in the rains and thereby lose NE and NW. The extent greatly varies: in their nutritive value (UNDPIFAO 1968b September 1857 it was a slaty sheet of water . . . about 300 yards in diameter . . . In p 3). early December 1858 nothing remained but On somewhat higher ground with col- a surface of dry, crumbling and deeply- luvial soils there is a wider variety of cracked sand and according to travellers it more palatable grasses than on the had long, in consequence of the scanty rains, mbugas which play a valuable role in wet been in that state." Caravans always camp at the Ziwa when season grazing (Vageler 1912; UNDP/ they find water there . . . When the Ziwa is FAO 1968b p. 3). desiccated, travellers usually encamp in a The significance of livestock in the thick bush, near a scanty clearing about one economy of the Wagogo is of irnpor- mile to the NW,where a few scattered vil- tance for the analyses of land degrada- lages of Wagogo have found dirt); white \vater, hard and bad in pits varying from twenty tion. However, it should be emphasized to thirty feet in depth . . . A demand is al- that the ratio of livestock to grazing re- lvays made for obtaining permission to draw sources is only one factor among many water. - A venerable custom dating from which together may be used to explain the days of Moses. Yet, thirsty, like hungry men are not to be trends in environmental change in trifled with. Fatal collisions have resulted ugogc. from this inhospitable practice. Some years ago a large caravan of Wanyamwezi was annihilated in consequence of a quarrel Water supply about water, and lately several deaths occurred in a caravan led by an Arab "A land condemned to perpetual merchant, Sallum bin Hamid, because the rvells \vere visited before the rate of payment drought" was settled. In several places we were fol- ~~~di~i~~~ll~,in the wet season, M,ater lowed upon the march lest a gourd might be furtively filled. To prevent exhaustion the was obtained from ponds and water- people throw euphorbia, asclepias, and sol- holes and in the dry season from i\-ellsin anacious plants into the ~vellafter a certain dry river beds. hour, andrvhen not wanted it is bushed over, to keep off animals, and to check evapora- Ikowa dam to be suitable for such reser- tion." (Burton 1860 I p. 296). voirs (Figs 3,4). Although of bad quality water was, dur- During the 1920s one or two sub-sur- ing the dry season, a very expensive face dams were built by the British at commodity which in some cases had to Dodoma. The dams, however, did not be collected from water holes situated at operate successfully and were soon several hours' walk from the settlement. abandoned (Fawley 1956a p. 62). Pro- Together with firewood and food it was posals to construct sub-surface dams sold by the local population to passing across infilled valleys as a solution to the travellers (Junker 1891 p. 691; Casati problems of water supply were again 1891 I1 p. 285; Fonck 1910 p. 323). heard in the 1950s (Ann. Rep. GSD, Sometimes it commanded the same 1954 p. 12). price as the local beer (Speke 1865 p. These proposals followed a number 75). Armed guards protected the water of unsuccessful boring programmes, holes and often quarrels and fights oc- begun in the mid-1920s, the construc- curred over the scanty water (Wissmann tion of the surface water reservoir at 1889 p. 280; Reichard 1892 pp. Imag in 1929 and the construction in 282233). the mid-1940s of a second, larger reser- The wells and the water holes have now been supplemented by boreholes. voir at Msalatu. In both reservoirs However, in several places even the sedimentation turned out to be a con- supply from the boreholes is irregular siderable problem. and of bad quality (TMWD 1974, I1 p. The Msalatu dam was the first in a 1, X p. 2). In the dry season most of the number of new water supplies to be boreholes are so low yielding that peo- built in Ugogo during the 1940s. The ple still to a great extent have to rely on reservoir was intended mainly for the temporary wells in the river beds urban water supply while others were to (TMWD 1974). enable the opening up of new grazing areas and to encourage people to leave overgrazed areas and to control the movements of the cattle herds (Patton 197 1). Water reservoirs In the 1950s and 1960s the Tanzanian A simple improvement of the traditional authorities showed great interest in the wells was suggested by Vageler (19 12 p. construction of surface water reservoirs 29). He showed how water seeping in and a number of dams were built. Thus, the stream bed could be collected it was assumed, the water supply prob- through constructing small sub-surface lems would be solved in combination dams across the streams. He considered with flood-control and possible irriga- the small streams at Dodoma, Kikombo tion prospects. During this period the river at Kikombo, Little Kinyasungwe at large reservoirs at Hombolo, Dabalo Chilonwa and Kiegeha near the present and Ikowa were built (Fig 7). 7. Ikowa dam and catchment

I kowa dam. Construction covering 6 12 km2 (photogrammetric delimitation of catchment-Wild B8, Construction data on Ikowa dam are photo-electric computation of catch- summarized in Table 3. ment area (Ekman & Wastenson 1968). In early 1955, after two years of se- Altitudes range from 900-1600 m.a.s.1. vere famine in Ugogo the plans of a The area of the catchment has been reservoir at Ikowa for flood control with variously stated as 466 km2 (WDD subsidiary cattle watering and irrigation 1957), 613 km2 (Berry & Kates 1970), prospects took form (Horst 1964 p. 6). 640 km2 (Rapp, Murray-Rust et al. The original plan involved an area of 1972), 635.8 and 634.6 km2 (TMWD some 65 km2 to be used for mixed culti- 1974). (Computation methods un- vation and grazing. Of a total irrigable known.) area of 600 ha, between 240400 ha The reservoir was designed to hold were initially to be set aside for irrigated 3.82 mill. m3 with a max. depth of 9 m (6 cropping (WDD 1955; WDD 1957). m if the stream channel is excluded) and A survey of the proposed dam site, a max. surface area of 182 ha (WDD situated S of the Dodoma-Mpwapwa 1955). Construction of the reservoir road approximately 55 km due E of Do- started in 1956 and in early 1957 the doma on the river Majenjeula (Kie- dam was completed and taken into use. geha), was undertaken in July 1955. The length of the embankment is 570 m The river has a highly variable sea- and the crest width is 15 m. The crest is sonal flow with a max. recorded runoff situated 12.5 m above the oripnal river of 700 m3/sec. It has a catchment area bed (12.8 m according to WDD 1957).

Fig36 Aerial vie!\ of Ikowa dam. Inflo~vand large vegetationless delta area in left part of picture. Outlet to irrigation channels in lower right corner. Note difference in vegetation between indhergs and pediments. Inselherg on the extreme right is Rfado hill. (Photo: C. Christiansson. Sept. 1971.) Table 3. Ikowa dam. Construction data.

Year of construction 1956-57 Area of catchment: 612 kmz

Altitude (asl) of river bed at dam site 890 m Assumed level of reference bench mark 134.00 ft Length of embankment 570 m Height of embankment (above river bed) 12.5 m(12.8 m) Width of crest of embankment 15 m Slope of embankment, upstream 3: 1 Slope of embankment, downstream 2: 1 Earthwork 117,715 m3 Width of spillway channel 150 m Height of spill\$ay crest above river bed = max total depth of reservoir 9 m' Depth of reservoir (river channel excluded) 6 m Surface area of reservoir 182 ha Total reservoir capacity 3,807,000 m3

' Crest of spillwa) = Full supply reference le\el (FSL) 134.00 ft

The spillway (FSL) is situated at an The same pessimistic views were ex- assumed level of 134.00 ft (= 134.0 on pressed in the annual reports of the gauge), some 10 ft (3 m) below the ref- following two years. erence bench mark level. The width of "On the grey calcareous soil very few of the the spillway channel is 150 m. crops gave profitable yield. Growing condi- The cost of construction of the dam tions in the dry season in the central pro- was 635,487 Sh. with an additional vince with the intense sun and almost con- 99,111 Sh. for channels and canals. Dur- tinuous desiccating wind are difficult for most crops." (Ann. Rep. Agr. D. 1958 p. 19.) ing the first 10-12 years maintenance "At Iko~vagame, birds and pests caused costs lvere estimated at 6-12,000 Sh. serious damage to a number of crops." (Ann. (Ann. Rep M'D&ID 1960). Rep. Agr. D. 1959 I p. 18; I1 p. 9.)

At the time of the planning of the pro- kowa irrigation scheme ject due consideration does not seem to I have been gven to the question of In 1956, 100 ha of irrigable land on the drainage. In some areas !\.aterlogging Kiegeha mbuga were cleared and by the began to occur at an early stage with end of the year some 30 ha were under consequent great reduction in crop experimental cultivation by the Agri- yields. Moreover, by 1960 increased sa- cultural Department (Figs 4, 8). linity of the soils was discovered (Ann. The experiments were successful Rep. Agr. D. I1 p. 6 1960). (Patton 197 1) and the initial yields were In 196011961 the area got very little good and prospects encouragng (Berry rain and towards the end of the dry & Kates 1970). However, the 1957 season the dam dried out and no water annual report of the Agr. Dept. notes was available for irrigation (Figs 27, 37). concerning the Ikowa irrigation scheme Under those conditions many of the that: farmers lost interest in the scheme. They seem to have felt that traditional "In general the yields obtained lvere no better than those obtained under favourable wet season farming was adequate to conditions without irrigation elsewhere in supply their own food requirements the territory." (Berry & Kates, 1970 p. 18). l r ANNUAL RECORD OF WATER LEVELS l i IN RESERVOIR 1970 1974 l l l

Fig 37a-c Ikowa dam. Annual record of water levels in reservoir, 1958-1974. Rising of water level marks inflow into reservoir follorring rainfall in the catchment. Sinking is due to evaporation and use of water for irrigation. Towards the end of the 1963 dry sea- Consequently the farming system needs son the reservoir dried up again, dis- to be adapted to cope with such situa- couragng more farmers from proceed- tions. In Ikowa a second long term prob- ing with irrigated cultivation. lem is that of the continual silting up of In 1964 a Farmers' Association was the reservoir, which threatens the fu- started. TYL l volunteers and a group of ture prospects of the project. The silting women moved in to cultivate the farm. is illustrated by the contour maps (Fig However, they had little previous farm- 381, cross profiles (Fig 39),Long profiles ing experience and after about a year (Fig 40) and capacity curves (Fig 4 1). See their numbers dropped rapidly. also Tables 4, 5. In the following year, it was noted, in In December 1959, after the reservoir a report by FAO, that only about 16 ha had been in use for 3 seasons, a survey were being irrigated. The dam mainly of the sediment volumes accumulated in served the purpose of flood control and the reservoir was carried out by domestic water supply for people (3000) WD&ID. It became apparent that the and cattle (4500-5000). capacity had decreased from 3,807,000 Much water was lost in the canals as to 3,111,000 m3 or 81.6 % of the orig- there was no co-ordination between the inal volume. The FSL area now meas- amount of water released at the outlet ured 144.2 ha against originally 182 ha. and water requirements on the farm. The average depth amounted to 2.2 1 m Moreover, the canals suffered from only. heavy silting because of the great con- A second sediment survey was carried centration of suspended material in the out in 1961 by WD&ID but unfortu- irrigation water. It was recommended nately the resulting data contain serious that investigations be started in order to errors and thus, the survey has been revive the scheme on a small scale. excluded from the present report. In 1968 about 40 ha were again irri- In 1963 the reservoir was again sur- gated. The few farmers participating in veyed by WD&ID. The capacity of the the scheme obtained food enough to reservoir had by then been reduced to subsist and a number were able to sell about 2,740,000 m3, a reduction of crops. An investigation into the use of 28.1 of the oripnal volume. the available water revealed great in- The rapid silting of the reservoir efficiencies. Of the water supplied in caused serious concern among the au- 1968169 (0.08 m3/sec. for 11 hours a thorities involved. In a FAO report in day) a considerable amount was lost to 1965 the following investigations Tvere unused land (Berry & Kates 1970 p. 22). suggested. The year 1969 mas dry and the reser- 1. To investigate the possibility of di- voir again dried up completely, causing recting the Majenjeula river into the serious problems for the farmers and Nyika ya Itumba mbuga. If easy diver- the three professional fishermen (Fig sion Jvere possible the mbuga could be 37). used as a silt-trap. Also, diversion of the river through the mbuga \vas hoped to Reservoir surveys reduce the peaks of flash floods. 2. To investigate the possibility of en- An irrigation scheme in the drier parts larging the reservoir capacity either by a of Tanzania will inevitably experience permanent increase of dam height and years in which water supply is poor. spillway level or by temporary measures ~vhich would be \cashed away by ' TANU Youth League extremely high spill~vaydischarges. IKOWA DAM Original survey 1955

'6789

E' 0 250 500 m - FULL SUPPLY LEVEL IFSL) RIYER CHANNEL l-E CROSS PROFILES

0 1000 2000 it - DEPTH BELOW F51 (FEET) = ROAD Y -Z LONG PROFlLE

Sediment survey December 1959

D-

Sedlment survey October 1969

Fzg 38 Ikol\a dam Reserxoir contour maps 1959, 1959 arid 1969 Orignal survev undertaken 1955 Reserlo~rtaken into use in earl) 1955 Depth In m (ft In brackets) on 1969 map Pos~t~onsof long profile and cross profile4 lnd~cated Note the existence of natural levees along orlgnal stream channel bet~eendepth contours 9-14

In 1969 a complete silt survey was survey the reservoir was completely dry again carried out ~vithinthe scope of the and the whole bed was accessible for DUSER project. At the time of the levelling. The total remaining volume was found to be 2,315,000 m3 which was 1957-1960 the annual sedimentation 60.8 % of the original volume (Rapp, was 23 1,680 m3. However, over a longer Murray-Rust et al. 1972 p. 28 1). period (1957 -1 969) the sedimentation In October 1970 a partial survey was was only 124,800 m3/year (Rapp, carried out by the DUSER team. As the Murray-Rust et al. 1972 p. 3 1 1). reservoir was fairly well-filled it was not possible to carry out levelling of the entire bottom. In the area adjacent to the Sediment yield embankment short sections from the full supply level towards the edge of the Data on catchment erosion and sedi- water were surveyed. In the upstream ment yields are given in Tables 4, 5. dry part of the reservoir the sections In the period 1957-74 the total loss of were extended right across the basin reservoir capacity was 47.5 % of the (Christiansson, 1970 pp. 34). origmal volume, corresponding to an A complete sediment survey was annual loss of 2.8 %. The accumulation planned for November 1971 but un- of sediment below FSL during the 1- fortunately had to be restricted to year period 1957-69 was 1,492,000 m3, soundings along a few transects near the or an average of 124,333 m3/year corre- embankment to check discrepancies be- sponding to an average annual sediment tween earlier surveys. The discrepancies yield of 203.2 m3/km2. were considered to result from disturb- In the following five years (1969-74) ance of the sediments by the fishermen the total accumulation was estimated at who drag their nets across the bottom 315,000 m3 or 63,000 m3/year indicat- whereby already deposited finer frac- ing an average sediment yield of only tions are again brought into suspension 103.0 m3/km2 per annum. It should be and then removed from the reservoir noted that the figures referring to the with the irrigation water. 1969-74 period are estimations based In the first half of 1972 Ikowa catch- on one complete long profile and a ment was selected one of three specially number of incomplete profiles sounded instrumented index areas for the Do- across the deeper parts of the reservoir. doma Water Master Plan investigations. It was assumed that the distribution of New metric gauges were installed at the the deposits was similar to that recorded reservoir (TMWD 1972b p. 7). In 1974 a in 1963-69. Under the above assump- long profile was sounded by the present tion the accumulation of sediment in the author and an estimation of the total silt 17-year period 1957-74 was estimated accumulation since 1969 was made. The at 106,294 m3/year or a total of some remaining capacity of the reservoir was 1,870,000 m3. The average annual then approximately 2,000,000 m3 or sediment yield would be 174 m3/km2. 52.5 % of the orignal capacity. Provided that rate of sedimentation In 1959, 1960 and 1967 the concen- prevails for the next 15 years or so, the trations of suspended sediment in the life of the reservoir will be some 35 Majenjeula river were measured. An av- years. The economic life, which ends be- erage figure arrived at was 3000 ppm, fore it is totally filled by sediment, is which would imply an average siltation much shorter and will probably have of 190,000 m3/year (TMWD 1974 X p. been reached by the time this thesis is 3). This seems to show fairly good con- published (198 1). cordance with the recorded average The figures for sediment volumes sedimentation in the early years of trapped in the reservoir are only existence of the reservoir. Between minimum figures of the total amounts Table 4. Ikowa catchment. Sediment yield and soil denudation rate.

Catchment Relief ratio Period of Total sedi- Annual sediment yield Soil denuda- area (mlkm) recording ment yield tion rate (km2) (m3) Volume Vol./area Weightlarea (mnllyear) (m3) unit (m3/km2) unit (t/km2)

Regarding differences from figures earlier published, see page 5. Regarding transformation of sediment yield in m2/km2 into values of tons/km2 see Table 12

Table 5. Ikowa reservoir. Rate of sedimentation.

Sediment Reservoir capacity Annual loss of capacity Expected total survey life of reser- (year) Remaining volume Period of recording Sedimentation voir (years) (m3) (7%) (years) (no. of (m3) (5%) seasons)

The survey \

LJ . SEDIMENTATION the end of the dry season the reservoir NOV 1969- NOV 1970 m JAN 1960 OCT 1963 dried up again. The follouling season, m NOV 1963-OCT I969 IAN ,957- DEC 1959 1961162, was one of the wettest on re- ---e LONG PROFILE INTERSECTlON cord. The reservoir filled up in eight

0 100 200 300 100 5M m days in mid-December, the result of six 0 500 1000 1500 feet storms above 30 mm in 17 days. The highest water level recorded at Ikowa Fig 39 Ikowa dam. Accumulation of sediment 1957-1970. Selected cross sections. has on 6 January 1962 when the height of Mater over the spillway was 0.68 m. This period of high discharge is most ment yield varies considerably betxveen like11 the cause of the large deltaic successive years due to fluctuations in accumulation. xlhich can be distin- runoff and erosion. guished in the long profile (Rapp. Heavy sedimentation has also taken Murraj-Rust et al. 1972 p. 283). Inflow place in the delta area. The deposition in the season 1962163 was again poor

IKOWA DAM, Capacity curves 1957-1974

Fig 41 Ikowa dam. Capacity curves 1957-1974. The 1974 curve is uncertain. It is based on one long profile and two cross profiles only. and the reservoir dried up before the including redeposited material from the next rains. The paucity of sediments in shore-line belt. On exposure above the the deeper areas may be due to the low water in the dry season the sediment water levels in 1961 and 1963 with cor- layers are disturbed by compaction, responding decreases in the deposition cracking, stock movement and insect of suspended material. activity (cf. Figs 64-65). The sediments of the reservoir are The dry bulk density of the sediment stratified into varves, each representing is around 1.6 g/cm3 (Rapp, Murray-Rust a major inflow of water. Each inflow is et al. 1972 p. 281). If the average bulk marked by deposition of a silty layer density of the soils in the catchment is grading upwards into finer clayey sedi- assumed to be the same as that of sediments. Sedimentation in the dry season ment the average soil denudation rate of very fine suspended material results measured over the whole catchment in accumulation of a rather thick layer would be 0.175 mmlyear. However, the

Tabl~6. Grain size composition (weight %) of sedime~ltsamples from Ikowa reservoir. The samples were collected 200 m upstream of the embankment in the N part of the reservoir. Sample 1 was taken from a depth of 1 m near the FSL. Samples 2 and 3 were taken from a depth of 0.5 and 1.0 m respectively in a pit dug into the reservoir bottom on the 2.4 m level below FSL. Analyses by G. Hellstrom, M:D&ID, Dar es Salaam.

Sample-. Grain size No. Coarse sand Medium sand Fine sand Silt and clay <0.074 mm Fzg 42 Soil pedestal left after gullying. The protective effect of the vegetation is obvious. (Photo: C. C,hristia~lsson,Dec. 197 1.)

bulk density of soils ranges widely with guish the type, and where possible, the location and depth. A common pattern intensity of the erosion processes. The is increasing bulk density and decreas- fol!owing section is based on Rapp, ing void ratio with increasing depth. Murray-Rust et al. (1972 pp. 270-278) In Ikowa catchment, soil bulk densi- and refers to the situation in 1972 and ties of between 1.3-1.7 g/cm3 have been earlier. No detailed studies of catchment recorded (TMWD, 1974 X pp. 16-19). erosion were made after that year. The During erosion studies in Kenya, Dunne map of erosion and sedimentation in the (1976) recorded top soil densities close catchment (Fig 8) is based on air photo to 1.0. interpretation and refers to the situation Locally much higher rates of denuda- in 1960. tion than 0.1-0.2 mrnlyear must be expected as practically all coarse material Inselbergs and other hillslopes is trapped upstream in the catchment and thereby "never" will be included The slopes of inselbergs and other among the reservoir sediments (Chris- bedrock hills in the area seem to be of tiansson 1979 p. 129). Moreover, a large little importance as sources of sediment. part of the catchment is effective only in They represent a small total area (some extremely \vet years as the extensive 10 %) of the catchment (Table 8) and to mbuga, Nyika ya Itumba, acts as a trap a certain extent they consist of bare rock for both water and sediments. without signs of intense weathering. They also have a fairly dense vegetation cover made up of trees and bush and Catchment erosion sparse grass. In some places the vegetation of the inselberg slopes has been The purpose of the erosion survey is to cleared (cf. Figs 74, 102), resulting in locate areas of erosion and to distin- local erosion. Fig 43 Root exposures a5 a consequence of splash erosion on heavily grazed pediment in hiatumhulu catchment. (Photo: C. Christiansson. Nov. 1974.)

Upper pediment slopes pebbles, root exposures, tree mounds The upper pediments extend from an and networks of rills). In this zone grass angle of about 10" do~vnto gradients of cover is very sparse or entirely missing about 2-3" where they gradually merge due to overgrazing (Figs 4244). into the lower pediment slopes. This Nearly all upper pediments are zone is marked by a dense netrvork of marked by dense gullying on both 1949 convergmg gullies, generally 24 m and 1960 air photographs. Comparisons wide and max. 3 m deep. There are also in 19'71 in the Ikowa catchment showed many signs of severe splash and sheet that gullied zones still covered these ~vash(splash pedestals under stones and areas ivithout an! clear evidence of

Fi,q 44 TI-ee mounds on upper pediment slope. Ikotva catchment. In hackground Dunda hill. (Photo: C. Christiansson. Nov. 15174.)

62 Fzg 45 Oblique aerial view of the gullied pediment slope N of Ngolo hill, Ikowa catchment. Upper zone of converging gullies, lower zone of parallel channels across cultivated areas. (Photo: P. H. Temple, Sept. 1971.) gullies spreading into earlier ungullied years. Hence such trees and mounds areas. show a general lowering of the The main impression is that no large- surrounding ground by about 1 cm per scale gully extensions have occurred year on the most intensely eroded parts since 1949. However wall collapse due of the upper pediments. This is a to undermining by seepage has been ob- minimum value as splash erosion also served (Fig 48). In general, gullies seem acts on the mounds, lowering their to be deeper and the pattern of channel surface somewhat. erosion denser in the western part of Ikowa catchment, a tendency that can be Lower pediment slopes followed into the Dodoma catchments. Also the intergully slopes of the upper The upper pediments grade into the pediment zone show many signs of se- lower at a slope gradient of about 2-3". vere and active soil erosion. Among the In this belt gullies tend to gradually de- most informative features in this respect crease in depth and be filled with sandy are so-called tree mounds. The tree deposits until they disappear and merge mounds are best developed on the flat- into thin sheets of sand (Fig 52). The ter interfluves of the upper gullied lower pediments are intensely cultivat- pediments (Fig 44). The margin of a ed. After the harvest the dry stalks from tree mound corresponds to the perip- the crops are burnt in the fields (Fig hery of the crown of the tree or bush. 102),if they are not grazed by livestock, The mounds are slightly crusted and leaving the ground bare. reach 1540 cm above the surrounding The main erosion processes active in ground. They are obviously erosion this zone are splash, sheet wash and ril- remnants formed during the lifetime of ling. In the more open areas sorting of a tree. Tree ring counts at Ikowa indi- surface material by wind activity is also cate that trees growing on 3040 cm common (cf. page 150). high mounds have an age of 20-30 No estimate of the amount of erosion Fig 46 "Termite mound terrain" E of Dodoma. (Photo: C. Christiansson, Sept 1971.) in this zone was made due to difficulties gradient decreases the rills end in minor with instrumentation. However, active fans of sandy or silty material. Accumu- erosion is evident from the occurrence lation also occurs in minature traps be- of networks of shallo~rrills across the hind roots and clumps of grass. On cleared fields after rainfall. Where the fields bordering mbugas "micro fans" interfinger with the black clays of the mbuga deposits. A lo~verpediment area dissected by long parallel gullies, 2-3 m deep and 4-5 m wide, is found on the slopes SE of Kikombo station (Fig 45). The gradient is below 1.5' and yet the slope is gullied. This may be due to rather abundant supply of runoff water from some of the highest and most extensive hillslopes in the catchment.

Gently undulating areas Considerable parts of Iko~vacatchment are taken up by gen~lyundulating areas characterized by so-called pock mark morphology or termite mound terrain. Fig 47 Minor anrhills on the pedirrlent above Msalatu darn. An important source of sediment in The remnants of thousands of old all parts of the catchment is earth freshly brought termite mounds appear in the terrain as to the surface by ants and termites. The material is deposited on the gl-ound in the form of srnall circular patches, devoid of vegetation, heaps or cones rangmg in height fl-on1 a few to 10 slightly domed and 10-25 m wide. The or l5 cni. The ut~consolidatedheaps are particu- patches occur at regular intervals and larly s~~sceptiblcto splash and sheet wash (cf. Young 1972 p. 67). (Photo: C. <;hristiansson. Oct. are very striking \.\-henseen from the air 1970.) or on aerial photographs (Figs 46, 52). Fig 48 Bank erosion caused by ~indercutting.Bank erosion is not a major source of sediment in the i~lvestigatedcatchments but contributes some of the coarse material foutld downstream. (Photo: C. Christiansson, Dec. 197 1.)

Lacking vegetation they are likely to be When the stream channels are filled exposed to heavy splash and high up by sand the material spreads as fans runoff. Special studies of these over adjoining areas (Figs 8, 99, 100). phenomena were not included in the The most extensive are the twin sand present project. (For details, see Johans- fans downstream of Kikombo station. son & Stromquist 1977.) The fans are not only remarkable as sites of sand accumulation but also as zones of' heavy water infiltration. Their Stream channels and sand fans location is probably controlled by a de- In most parts of the study area stream crease in channel gradient or by loss of channel bottoms are formed either by \vater through downward seepage, or by bedrock, common in the upper reaches, both factors in combination. or hardpans (cemented sediments) and The map (Fig 54) sho~vssand fans in clay in the lower valleys. On the proper typical positions where a tributary channel bottom layers of sand are gen- channeljoins a channei of higher order. erally deposited. The depth of the sand There are several examples of this type layer ranges from 112 m or less to sev- also in the upper part of Ikowa catch- eral metres. The stream banks are gen- ment. But many of the sand fans are erally vertical and may reach from 1 to 3 found along single stream courses, e.g., m above the sand bed. In times of flow the three lo~verfans of the Luaha river the banks are freq~~entlyundercut and (Fig 8). The one near Buigri is so close large clumps of soil slump into the to the drainage divide that sometimes stream (Fig 48). Thus some of the ma- the channel shifts over the fan to the terial deposited in the reservoirs origi- north spilling water over the normal di- nates from the stream banks although vide and i~ltothe Buigiri reservoir. the total volume of sediment from this Some fans are situated immediately source is insignificant. below short steep overgrazed pedi- Tablr 7. Monthly rainfall (mm) at 9 stations in Ikowa catchment (Fig 8) and 3 stations at Dodoma (Fig 11) during the 1972173 rainv season. Only stations with complete records for the whole season are included.

Station Oct. No\. Dec. Jan. Feb. Mar. Apr. May. Total

No. 1 3.3 32.6 (2)' (6) No. 3 4.3 13.0 (1) (3 No. 4

No. 6 10.2 35.5 (2) (2) No. 7 3.0 12.5 (1) (4) No. 8 36.5 (5) No. 9

No. 10

No. 11

Dodoma 31et. Stat Dodoma Geol. Office Matum- bulu Darn

' Fi<~rein hrackets = nurnber of raindays. Sources: TMMTD1974, EAMD primary records ments. They reveal that much of the set up. The locations are shown in Fig. eroded material in the catchment orig- 8. The stations are separated from each nates from such areas. other by a distance of 7-8 km, the sta- In some areas with very low gradient tion net~vorkthus having a density of 1 the streams build up "flood plains" with per 50 km2. Although each station predominantly clayey soils and sandy shows its olvn pattern of distribution levees along the banks, as indicated on and occurrence of rainfall, the total re- U the orignal map of Ikowa reservoir site corded rain at each station does not de- (Fig 38). viate more than 20 70 of the season's average total rainfall for all stations (Table 7). Rainfall The intensity of the rain falling at each station during the season was also In order to determine the characteristics recorded. The maximum intensity (1.7 of rainfall in Iko~ra catchment a mmtmin.) occurred at station No 3 on net~vorkof raingauges was installed in 30 January 19'73 providing a total the area by the Dodoma Water Master amount of 60 mm in 35 min. Such in- Plan Team in 1972 (TMWD 1974). tensities are. ho~vever,uncommon and Altogether twelve gauging stations were the general opinion is that 50 7c of the Tabl~8. Area1 inventory of main landform units (cf. Fig. 8)

Entire study Ikowa catchment area (%) 06) Inselbergs and bedrock hills 12

Upper pediments 20

Lower pediments + gently sloping upper pediments 18

Valley floors and associated landforms except mbugas 25

Mbugas 6

Gently undulating terrain (Termite mound areas) 19

showers provide less than 5 mm at in- tents were low' throughout the profiles, tensities between 0 and 0.2 mmlmin varying from 3.1-8.9 (4.3-12.5 % in (TMWD 1974). Such limited rainfall volume, assuming an average bulk den- evaporates quickl\ and generally does sity of 1.4 g/cm3). The field capacity, not initiate stream discharge. (5 mm is which is assumed to he around 20 % in approx. the average daily evaporation volume, was never reached. Such low from a free water surface in the rainy soil moisture contents prevailing even at season.) the end of the rainy season provide clear Rains of more than 20-30 mm rep- evidence of a permanent soil moisture resent almost 50 % of the annual total deficit. amount of precipitation but account for Detailed soil moisture measurements only 10-20 $% of the annual total Jvere carried out from late March to number of sho\\ers (cf. Fig 71). mid-May 1973. Samples were collected at intervals from two sites representing lower pediments and valley bottoms. Infiltration and soil mois- The vegetation at both sites consists of ture fairly tall grass and scattered Acacia trees. In 1973 infiltration experiments and The site representing valley bottoms soil moisture measurements were is situated near the confluence of the carried out in Ikowa catchment by The Kikombo and Majenjeula rivers (Fig 8). Dodoma Water Master Plan Team The site representing lower pediments (TMWD 1974). is situated below the small inselberg Ki- Very high rates of infiltration Icere donyasi, some 1.5 km N of the first site. observed (average 0.5-5 mmlmin for At the valley bottom site the soil mois- 120-180 minutes). The lowest rates ture varied from 2-22 5% in the upper (30-80 mmlhr) were recorded in areas 50-60 cm and from 7-15 % in the lower with rather compact profiles. The high- layers down to l50 cm. An average est rates (160-300 mm/hr) Icere record- around 10 S prevailed from 40 cm and . ed in areas lvith hon~ogeneousdeep do\cnlvards. The lower pediment site profiles of loamy sand. sho~vedthe same pattern. The soil mois- During the course of the recordings ture varied from 2-18 % in the upper (March) the average soil moisture con- 3040 cm and from 3-13 % in the lower part (down to 150 cm), also with an av- layers, as shown by Staples (1937b) after erage around 10 % below the 40 cm experiments at Mpwapwa. Consequent- level. Thus, variations are largest in the ly, there is virtually no recharge to upper layers, while in the deeper part of ground water in these areas except the profiles variations are rather small. where water courses cross fracture The rainfall intensity in the study area zones. seldom exceeds 60 mm/hr and the infiltration rate in the sampling area is high (see above). Under such circum- Evapotranspiration stances almost all rain can be immediately infiltrated into the soil. This is also The evapotranspiration, which includes indicated by the small runoff coefficient transpiration by plants, has a strong in- of 1.5-8.3 % for entire subcatchments fluence on the water balance in semiarid (see page 70). Under such conditions areas. The potential evapotranspiration soil erosion occurs only during excessive in the rainy season in the study area as storms. Similar effects were shown from calculated by Penman's method is as Zimbabwe by Elwell and Stocking follows: November 209 mm, December (1974). 188 mm, January 167 mm, February However, a sub-surface storage of 153 mm, March 160 mm, April 149 mm 10-20 % around the end of the rainy (Table 1). season is dried up in the succeeding dry This evapotranspiration would be season to between 0.5 and 1 % in the maintained provided the water supply uppermost 20 cm and between 2.5 and were sufficient. However, in the study 3.3 % at a depth of 80-100 cm. area the deficit in the soil moisture is a In the Dodoma area an amount of 30 limiting factor. The movement of the mm rainfalliday occurs approximately soil moisture downwards to deeper once a month. Should 30 mm fall at the layers is almost negligible (see above). begmning of the rainy season on the Therefore, the decrease in soil moisture extremely desiccated ground surface is mainly caused by evapotranspiration. and the supposed runoff coefficient is This leads to the conclusion that the ac- 5 $6, then about 28 mm will be infiltrat- tual evapotranspiration can be obtained ed. However, that amount of water is from data on the decrease in soil mois- sufficient to fill only 9.3 cm of soil to ture. The results from two sites (lower field capacity, assuming a field capacity pediment and valley bottom) were 2.9 of 20 % of the soil volume. mmldap and 2.4 mmtday, respectively. The infiltrated Ivater in the top soil Thus the actual evapotranspiration in layers is always subject to heavy evapo- the studv area is some 2-3 mm per day transpiration. Only a very limited during the rainy season in areas covered amount may be absorbed by the drier by vegetation similar to that at the sam- soil below. In general, it may be con- pling sites. cluded that the high infiltration rate favours the potential sub-surface water storage but the actual infiltration is Drainage pattern and dis- strictly limited bp the low amount of charge rainfall. Due to the general low soil moisture Three of Tanzania's major drainage and the high evapotranspiration even in areas meet within the stud\ area at the rainy season, moisture does not Chimmaga hill, SE of Dodoma. Trib- move gravitationally to the deeper utaries of the Gt. Ruaha flou tol\ards Fig49 LVatel- stage and approximate discharge, Kikombo river at Kikombo 10-1 1 Apr. and 18-20 Ilec. 1973. (Modified from TMWD 1974.)

the SE, rivers belonging to the internal amounts of sediment which are deposit- drainage system of Bahi depression flow ed at the confluence with the main river. to~vardsthe SW and W, and tributaries At Mtumba, 30 km from the source, of' the Kinyasungu-e-Wami river system the Majenjeula (Mwitikira) becomes flow towards the NE and E. Many of the narrow and winding, remaining so until smaller streams in the area never reach it reaches the small mbuga at Mkwala S the major rivers but ebb away in the of Buigri. Another 5 km downstream, plains or in local discharge-less depres- at an altitude of 985 m.a.s.1. it is joined sions (Fig 52). by the Kikombo river. The Majenjeula (Mwitikira, Luaha, The Kikombo river originates at an Kiegeha), which ori~natzsat an altitude altitude of 1500 m on the slopes of Ki- of 1400 m on the slopes of Chimwaga kanga and Itamba hills (Fig 8). Like the and Kituli hills, drains the major part of Majenjeula, when reaching the general Ikowa catchment (Fig 8). It is a tributary level of the plateau, it becomes a large of the Little Kinyasungwe which in turn sandy river 50-100 m wide. At Nkam- is one of the major tributaries of the baku and N of Kikombo station the river Kinyasungwe-Wami. E of Chimwaga, 7 has deposited impressive sand fans. km from the source, the upper Majen- The total length of the upper Majen- jeula enters flat land in the valley of jeula (upstream of the confluence with Nkongontha. It here becomes a typical Kikombo) is 45 km. The area of this "sandy river" and expands to a xcidth of subcatchment is 264.3 km2 with a drain- over 100 m. age density2 of'0.55 km/km2. The length Just N of the village of Ihumrva a small stream joins from the W. This stream drains the actively eroding NM' ' Computed front the drainage marked on corner of the catchment. It carries large the 1:.i0,000 topographical rnaps. of the Kikombo upstream of the conflu- runoff coefficient (ratio between dis- ence with the Majenjeula is 26 km. The charge and effective rainfall) of the subcatchment area is 85.1 km2 and the former area varies from 1.5 % to 7.2 % drainage density is 0.88 km/km2. The and of the latter from 2.1 % to 8.3 %, gradient downstream towards Majen- with averages of 4.0 and 5.3 % respec- jeula is 0.0026 m/m (2.6 mlkm). tively. These runoff coefficients are in From the confluence the lower Majen- the same range as those recorded at jeula flows across the plains of the E part Imag close to Dodoma (Christiansson of the catchment to the Ikowa reservoir. 1979) but lower than the runoff figures The gradient is low (0.0039 m/m). for Msalatu catchment (Fawley 1956b). To the S of the lower Majenjeula val- On 10 April 1973 runoff from the ley is found the large mbuga, Nyika ya Kikombo catchment (85.1 km2) gave a Itumba. It receives water through short discharge of 224,000 m3 at a runoff of channels orignating in the Madowa and 4.4 % and an area rainfall of 63.5 mm. Manyeche hills and through larger The figures indicate a severe "loss" of streams (the Kolongo and the Msamalo) water. Some 95 % of the rainfall or which originate in the Tasangala, Ma- more than 60 mm did not leave the lamo, Ngolo and Kisunguru hills (Fig 8). catchment as surface runoff. The daily Alluvial fans have developed at all average potential evapotranspiration in places where the channels leave the the area in the rainy season is only 5-6 pediment slopes and enter the plains. mm/day. Therefore, it must be assumed Major sand accumulations are also that most of the "loss" of water is due to found where the streams reach mbugas infiltration into the soil. and the drainage becomes undefined. If 60 mm of rain is absorbed by the Nyika ya Itumba is drained north- soil and distributed to gve a 10 % in- wards to the Majenjeula. In the E part a crease in moisture in the upper part of pattern of shallow drainage channels the profile it will be absorbed by a soil has developed on the mbuga surface. layer of 60 cm. This moisture will be Discharge from the mbuga through the subject to evapotranspiration during the channels rarely occurs, except after first dry spell following the rain. heavy rainfall near the outlet. When Accordingly, the low surface runoff is rain falls further upstream there is gen- explained by the high infiltration capa- erally no flow through the outlet city and the high rate of evapotranspi- channel. Compared with the other sub- ration (see above). catchments the discharge from the In order to determine the discharge Nyika ya Itumba is negligible (TMWD conditions, the areal rainfalls were cal- 1974 X p. 36). culated for all runoffs in the Kikombo In the upper Majenjeula river there and upper Majenjeula sub-catchments. were 16 flows and in Kikombo river 19 The total areal average rainfall for the flows in the 1972173 season (TMWD 1972173 season was 569.3 mm and 513.0 1974). The heaviest of these lasted for mm for Kikombo and upper Majen- 24 hours, but usually flow in the streams jeula, respectively. The total discharge does not last for more than 6-7 hours for the whole season from Kikombo (Fig 49). catchment was 22.9 mm, xvhich is 4.8 % The area average rainfall which of the total effective rainfall (476.4 mm) caused flow in the rivers ranged from or 4.2 'j%of the total areal rainfall (= the 7.5 mm-22.6 mm in the upper Majen- overall runoff coefficient). jeula catchment and from 12.2-63.5 Corresponding figures for upper Ma- mm in the Kikombo catchment. The jenjeula catchment were: total discharge 10.7 mni = 4.0 % of the total effective of the upper Majenjeula is almost dou- rainfall (268.1 mm) or 2.1 'j4 of the total ble that of the Kikombo by virtue of its area1 rainfall for the season. The effec- larger catchment area, the latter is the tive rainfalls in Kikombo catchment was more efficient from the point of view of 84 % of the total rainfall while in the surface drainage. The water yield of the upper Majenjeula it lvas only 52 %. Kikombo was 270,000 m3/km2and that The difference in the runoff pattern of Majenjeula was 45,000 m3/km2 clur- between the two catchments is partly ing the year. Thus it can be expected caused by differences in morphology that Kiko~nbocatchment is the area and partly by differences in rainfall dis- suffering most severely from soil ero- tribution. The Kikombo basin is steeper sion and thus yielding the greatest than the Majenjeula, especially in the S amounts of sediment per area unit. where also hardpans occur on the hill Regarding the 48 5% loss of water in slopes. These impermeable crusts most the upper Majenjeula catchment, it may certainly increase the runoff rates (cf. be partially attributed to the heavy Msalatu catchment). In the Kikombo evapotranspiration in the thicket areas, catchment valley bottoms and flat land but also to considerable sub-surface represent 32.4 %, of the entire catch- storage in alluvial fans and sandy ment. In the Majenjeula a larger share riverbeds. Some water may also have (40.1 %) is composed of flat and gently been infiltrated into deeper strata. undulating land and alluvial deposits. However, as mentioned previously, the Moreover, a considerable part of this soil moisture rarely exceeds field capa- catchment is covered by deciduous thic- city. Therefore, it is very unlikely that ket ~vhichgives extremel:. small runoff recharge of the underground water coefficients (Christiansson 1979 p. 134). occurs by means of percolation through Another reason for differences in the the soil layer. runoff is the varying rainfall pattern. The fact that ground water is pumped Rainfall at stations No. 7 and No. 9 tvas up in the catchment (7 boreholes) does low (417 mm and 4 l l mm, respectively). not necessarily indicate that recharge Rains of more than 20 mm occurred occurs at present as supposed in the only 5 times at station No. 7 which Dodoma Water Master Plan (TMNTD means that the share of effective rainfall 1974). The ground water may be fossil Ivas small. Station No. 8 in the S of Ki- (Woodhead 1968). However, in some kombo catchment recorded the season's areas recharge occurs when the surface highest rainfall (620.6 mm). Furth- flow meets with a favourable recharge ermore, showers yielding more than 20 area such as a fracture or fault zone mm occurred 1.5 times, gving more exposed in the river course. These than 50 % of the total rainfall at this places are commonly indicated by station. extensive sand accumulations which Although the annual total discharge block the channel (cf. pages 107-1 10).

Rainfall in excess of'the anlount ~leededto wtisf\ the itorage (apacit\ of [tie catchment. 8. Imagi dam

The following is a summary of an article through leakage and evaporation from about Imagi dam published elsewhere the reservoir amounted, on average, to (Christiansson 19'79). 50 % of the max. volume stored in the The present effective catchment area respective years. The small size of the of Imag covers 2.2 km2. The vegetation catchment, the low rainfall and the great is dense thicket and miombo woodland water losses all contribute to the reser- on inselberg slopes and shrub and voir being filled only in one year out of sparse grass on pediments (Fig 50). The five. soils are of low fertility. Present domi- The reservoir is gradually losing nating land use is grazing by livestock. capacity through sedimentation due to For a period of more than 20 years, soil erosion in the catchment. As part of beginning in 1929, when the reservoir the DUSER programme the volume of was built, the area was reserved as sediment was assessed through levelling catchment for the Dodoma town water and sounding of profiles across the supply. In that period hydro-meteoro- reservoir. Inventories of vegetation and logical research was carried out in the erosion features in the catchment were catchment. Much of the rainfall was made through air photo interpretation found to be lost by evaporation from and field studies. interception storage and transpiration Erosion was found to be slight on the from the catchment vegetation. Annual densely vegetated inselberg slopes while runoff from the catchment varied be- on the pediment surfaces runoff and tween 3 and 17 % of the rainfall. Losses soil loss are more pronounced. Splash,

Fig 50 Imagi dam and Imagi hill from the E. Foliage has developed on the thicket/woodland vegetation but runoff inflow into the reservoir has not started. Note bare rock with exfoliation on hill crest and many boulders on the hill slopes. (Photo: C. Christiansson, Dec. 1971.) sheet wash and rilling are the dominat- 0.78 % p.a. of its original volume and ing types of erosion. Gullying occurs the maximum depth has decreased along cattle tracks. from 10.8 to 5.7 m. Provided the aver- The observed variations in sediment age rate of sedimentation remains the yield are shown to be directly related to same in the future the total life of the changes in land use. When the earlier reservoir will be 120-130 years (Table protected catchment was opened to cat- 22). tle grazing the sediment yield rose by At the time the reservoir was built it 50 %. The average annual sediment was considered that Dodoma had got a yield for the period 1934-7 1 was 602 m3 very reliable water supply (Ann. Rep or 900 tons/km2, which is comparatively GSD 1931 pp. 32-33). However, the low considering the steep slopes and the consumption gradually increased and small catchment size (Fig 91). The rate after some years the reservoir was found of soil denudation distributed over the to be inadequate for the needs of the whole catchment varied during the town. In 1943 a severe drought hit the same period between 0.50 and 0.83 mm area. The Imag reservoir dried up and p.a. which implies a denuding effect, the town became desperately short of measured on the pediment only, of be- water. This initiated the planning of a tween 1 and 3 mm p.a. second town dam. On average the reservoir has lost 9. Msalatu dam and catchment

Construction and operation ments were removed before construc- of the dam tion (Handle? 1952 p. 3). Material for the clay core was taken partly from a Construction data on Msalatu dam are clay layer in the stream bed below the summarized in Table 9. dam and partly from the railway brick- As a consequence of the 1943 drought field at Dodoma. The remainder of the immediate action was taken and a sec- construction material, surface debris ond dam site was selected some 3.5 km (red sandy loam), was taken from the SE of the town (Fig 52). In November area adjoining the dam site. In 1943 surveys of the site were carried out September 1945 the reservoir was and by March 1944 the construction brought into use. A very valuable drawings were completed. In April account of the reservoir, operation and tenders were submitted for the con- problems during its first 74years was struction work and in October the published by Fawley (195613). embankment, an earth dam with a clay Initially the new water supply was core, stood ready. Unfortunately, renamed Town Dam No 2 or The New cords of the construction of the dam are Dam. When discussions later were not complete. It is assumed that the dam raised on a third reservoir The New rests on a layer of cemented sediments, Dam was renamed Msalatu after the avith low permeability, which are under- small stream whose water is stored in the lain by granitic bed-rock. Sandy soils reservoir. Originally it Lvas estimated originally overlying the cemented sedi- that runoff from an area of 18.1 km2 (7

Figil Oblique aet-ial view of 5Zsalatu dam. In foreground the spillwa) and channel 2 ~vithticlta. Lcft in picture. the delta in front of the main inflou. 111 background Iseni ridgc. (Photo: C:. Christiansson. Sept. 197 1 .) Table 9. Msalatu dani. Construction data and changes of capacity.

Year of construction: 1944 Area of catchment: 8.51 km2

Altitude (asl) of river bed at dam site Altitude (asl) of spillway crest Length of embankment Height of embankment (above river bed) Width of crest of embankment Slope of embankment Width of spillway channel Height of spillway crest above river bed Max. total depth of reservoir1 Height of lower sluice invert above river bedZ Height of spillway above lower sluice invert Height or spillway above lower sluice invert: gauge reading Max. effective depth of water Max. length of storage basin Max. width of storage basin (excl. spillway area) Reservoir surface area Total reservoir capacity

Losses of depth, area and volume due to sedimentation have been allowed for Lower sluice invert = reference level (0.0 ft) on gauge. sq. miles) would be collected in the stream enters the reservoir was some reservoir (Ann. Rep. PWD 1944 p. 16). 600 m and the surface area was 125,000 Later surveys, however, showed that the m'. true size of the catchment was less than The embankment has a length of 366 half that area (Ann. Rep. PWD 1945; m and a maximum width of 75 m at the James 1950 p. 4). A preliminary survey base and 4-5 m at the crest. The dam in 197 1, by the present author, based on crest is situated 12.8 m above the former 1:50,000 topographical maps gave an river bed. area of 8.67 km2 (Rapp, Murray-Rust et The spillway channel was excavated al. 1972 p. 289). Later detailed checks into the loose soil E of the river. The on large scale (1:15,000) photo- crest was situated 2.4 m below the top of grammetric maps gave a catchment size the embankment. The width of the of 8.51 km2. channel was 76 m. The original FSL of the reservoir is Records of the operation of the dam situated at an altitude of 1169 m.a.s.1. As during the first 8 years are summarized the elevation is not enough for gravity in Table 10. In the first season flow to the town water-works, the water (194411945) rainfall was near the aver- is first pumped to a high level storage age for the area but it is not known if all tank. the runoff was collected in the reservoir. According to the orignal design However, enough was collected to bring (Ann. Repts. PWD 1944 and 1945) the the reservoir into use in September initial reservoir capacity was 358,660 m3 1945. In the second season (1945/46), (79 mill. gals). A later check indicated rainfall was well below average but the 354,000 m3. The maximum depth was reservoir filled. The third season 10.4 m. The distance from the em- (1946147) experienced very heavy rains bankment to the spot where the main and there was considerable flow over Fig 32 Dodoma to.r\n and surroundings. Air photo 49/TNi5/012, June 1960. Imagi and Msalatu dams i~ilower part of picture. Eroded slopes-sheet wash and gull)-ing--clearli visible in Msalatu catchment and adjoining areas. Note also do~instreanideposi~ion on sand fans and the final disappearan~eof Xlsalatu stream on the plains NE of Dodoma. Tivo PI-ominent inselbergs are visible in the picture, one M' of Dodoma ~cithremnants of contour banks on the pediment and Xilim\\-a hill NE of the town surrounded bl "tern~itemound terrain". Cf. Figs 1% 46. 190-101. the spillway already in January. By this reduced, due to sedimentation, by as time it was noted that substantial vol- much as 9100 1n3 (2 mill. gals) per year. umes of sediment accumulated in the The follo~ving t~vorainfall seasons, storage basin. Sonie years later it Tvas 1947148 and 1948149, experienced little estimated that the reservoir capacity was rain and, although the tolvn non. relied %hi(, 10. Msalatu dam. First eight years of'operation. Rainfall, runoff' and irlflou into reservoir are measured per rainfall year. Amount of water stored, consumption and losses are measured per calendar year. It has not beet1 possible to obtain accurate runoff' fip~resfro~n the catchment as the amount of' water that overflowed into the spillway was not mcasuretl. In the table no allowance is made for the successive loss of capacity through sedimentation. Thus figures for inflow into the reservoir (recorded from lllc increase ill water stage and computed against the capacity curve) are somewhat overestimated, as are the figures for total arnou~ltof water. stored. It sllould also be noted that the original capacity was assumed to be 358,700 m3 (79 mill. gals), a figure which appears in all official records. 7'he figures in this tablc are based on that "original volurne". Later checks, however, revealed a somewhat smaller volume (SWpage 77). Figures tot- consumption were measured at the town water works. Accurate figures for losses through evaporation and leakage are not available. 'The losses I-ecol.ded are the diff'ererlccs between the total recorded amount of water stored in the reservoir during the year (generally somewhat lal-ger than the rnax. reservoir capacity) and the volume actually consumed.

A. B. C. D. E. F. G. H. I. Rainfall 'l'otal Ir~flo~in- Ovrrflobt Rurioff2 Calendal- 'Tot. amount Annual consumptiorl Natural losses veal- rainhill to rrse1-- (months) (%I) year of' water (lur~c-May) (mm) voir stored (1,000 m3) (O/c of G.) (1,000 m3) (96 of'G.) ( 1,000 m3)' (1,000 m3) March, April Jdnual-y, February, March. April Reserv. rlot filletl Fcbt-uary, March

January, February, March March, April March, April Rrserv. nor filled

' Volurnc of overflo~not recordrd. G Until 1-eservoiroverflowed. G Sources: Fabrley 1956h; PM'D Records on two major supplies, rationing of wa- rains of April and May. As a result of ter was again necessary in 1949. To this the entire town was supplied from compensate for sedimentation and as an the Makutapora scheme" (Ann. Rep. attempt to improve the storage capacity, PWD 1957 p. 9). From then onwards the spillway level of the reservoir was Makutapora served as the main water raised by 2 feet in 1950, thus increasing source for Dodoma. The two town dams the volume of the reservoir by some got the role of supplementary supplies 65,000 m3 (Fawley 1956b p. 72) (Fig 93). only. However, the increase in volume also In 1971 the catchment was selected by led to the head of water increasing and the Dodoma Water Master Plan team thereby also the seepage losses. To see if for special instrumentation (TMWD improvements could be effected, in- 1972a). The plans were never realized. vestigations of the seepage from the A modification of the reservoir capa- dam were undertaken in 1952 and city was again carried out in December 1953, but without marked success (Ann. 1972 when the spillway level was raised Repts PWD 1952 p. 9; 1953 p. 11). by 0.3 m (1 foot). The total capacity of The drop in the reservoir due to the reservoir was thereby increased by evaporation and seepage losses ranges approximately 30,000 m3 (Figs 60, 93). from 7.6-12.7 mmlday. Most of the drop is attributed to evaporation (cf. Ann. Rep. GSD 1932 p. 23; Christians- Drainage pattern son 1979 p. 137). The seepage losses were estimated at about 200 m3/day or a The Msalatu catchment is drained by daily lowering of the water surface by 3 several parallel, NW-flowing, small mm with the reservoir half full. streams which originate on the slopes of In 1953 new attempts were made to Chimwaga hill. The main stream, Msa- improve the storage capacity, and a total latu, is only some 10 km long, whereof 5 of 8000 m3 (l 314 mill. gals) of silt were km are situated upstream of the dam. excavated from the reservoir (Fig 93). Although short, all the streams draining Moreover, it was recommended that the slopes of the Dodoma hills carry about 800 m (2600 ft) of furrows should masses of sediment away from the be constructed to divert the flow of uTa- pediments. In the case of Msalatu the ter from some of the streams to the E sediments are trapped in the reservoir. into Msalatu reservoir. That extra flow In other cases the sediments are of water was expected to cause the brought down to the plains where they reservoir to fill even during dry years are deposited as large sand fans into (Fawley 195613 p. 8 1). which the runoff \vater is rapidly infil- In 1956 attempts were made, at the trated. The drainage pattern becomes start of the dry season, to conserve the sluggish and streams which a few km reservoir for anti-evaporation experi- upstream are 50 m ~videjust ebb away ments only, but these were upset owing (Figs 1 1, 52, 99-100). to the necessity of using the dam-water On the slopes above the dam the when problems arose with the supply channels are narrolver and rather shal- from the Makutapora boreholes (Ann. low. On the upper pediments channel Rep. PWD 1956 p. 10; Fawley 1958 a) bottoms are close to the bedrock. The (cf. page 97). depth seldom exceeds 1 m. On the lower In 1957 "the No. 2 Dam supply par- pediments the channel width and depth tially failed during December due to is greater. The ~vidthmay amount to heavy silting up caused by the heavy 8-10 m and the depth to 34m. Crusts

now been converted into a road; the small mulated in the stream channel down- drainage ditches which bordered it are now of the present ~~~l~~~reservoir. large enough to accomodate a motor car. stream l.he area, olltivated just The sedimentation rate in the channel befbre the war, is now in an advanced state averaged 15-20 cm annually over a pe- of 'badlands' erosion." riod of some 25 years. The first detailed accounts on erosion The present appearance of the above features and estimated rates of erosion area is shown in Fig 53. in Msalatu catchment were presented by Grantham also comments on the in- Fawley (1956b p. 74). crease in sediment transport in Msalatu stream during the same period. "Numerous 'korongos" intersect the area and have steeply banked channels from 3 to "Three kilometres east of Dodoma is a sandy more than 20 feet deep. The larger 'koron- river, crossed by a railway built just before gos' are rapidly widening their banks; small the war. At the time of building the bridge 'korongos', branching out from the larger the clearance betureen the river-bed and rails ones, are rapidly cutting their lvay head- was 14 feet; now it is only about 3 feet, owing nards. All the material eroded in the 'korongos' is to the silting up of the channel by sand. In eventually carried into the reservoir where the particularly wet season of 1929-30 there most of it is deposited. A proportion remains were no less than twenty-six wash-alvays at in suspension in the reservoir water and this one river crossing. Soil erosion thus en- must be filtered out before the water is used. gages engineering. The trouble here was An average of about 12,000 cubic yards of due almost entirely to bad methods of native sediment are deposited in the reservoir each cultivation in the hills above; soil is eroded, year, indicating that about 4 inches of soil the sand being redeposited and the clay are removed from the catchment area in ev- fraction carried away." ery 100 years. This high erosion rate could not have been maintained for more than a Also Wade and Oates (1938 p. 56) refer few hundred years, hence it indicates that a to the same stream and stress the seri- drastic change must have taken place in this area, either with regard to climate or vegeta- ousness of the heavy sediment-transport tion." In the water courses draining the Do- doma hills. It is doubtful whether the headward erosion in the 1950s was as serious as "East of Dodoma, the railway has repeatedly been threatened with obliteration by the described by Fawley. Comparisons be- shifting sands of another drainage channel tween the 1949 air photos and photos from the hills to the south. It is reported that from later years show only minor in 1916, before these hill slopes were so changes in the gully pattern. extensively cultivated, the bridge conveying Nevertheless in 1953-54 gully recla- the railway over this watercourse was ten feet above the sandy bed. At the present time the mation was carried out over the whole accumulatecl sands have reached the lo~ver catchment area. During the dry season edges of the steel joists of the bridge, and the larger gullies were filled with already the railway engineers have been brushwood to a depth of 5 to 10 feet for forced to add additional spans to it to main- a considerable part of their length ivith tain communications." the objective of causing silt and sand to At the spot where the stream is crossed be trapped in the gullies instead of be- by the railway it is some 50 m wide (Figs ing transported into the reservoir. 99-100). Initially the benefit of this work was From the above accounts one can apparent. However, in the long run the draw the conclusion that from 19 10 (the year when the railway reached Dodoma) to 1938 (or one or two years earlier) a I "Korongo" is a local term for a gully or a total of some 4.27 m (14 ft) of silt accu- cleepl? incised strearu channel. DODOMA CATCHMENTS. LAND USE. EROSION AND SEDIMENTATION 1960

Mixed thorn bush and grass

Cultivated fields

- Main rocd Homestead or house - - Catchment boundary

Fig 54 Dodonia catchments. Land use and areas of severe erosion in 1960. Dense vegetation is found on the upper parts of most inselhergs. All pediments carry secondary bush or are under cultivation. Soil erosion is intense on upper and middle pediments. The homesteads are concentrated to valley bottoms and lolver pediments. Where houses and homefields are located on the upper pediments they seem to he of recent date and are probably indications of scarcity of land suitable for cultivation. For description of transect A-B see Table 17. Table 11. Dodoma catchments. Areas of se- Present distribution of ero- 1 ere erosion. 1960. sion in the catchment Catchment Total area Sex ereh eroded area (km2) (km2) (7%) Three or four zones of erosion can be distinguished in the catchment (cf. Imag 4.3l 0.7 16.3 Msalatu 8.5 1.8 21.2 Table 28). Mkonze 25.3 5.6 22.1 On the inselberg slopes erosion is Matumbulu 17.9 3.3 18.4 generally very slight as they are occu- Total 56. l 11.4 20.3 pied by dense vegetation (Figs 7,54). In places where this vegetation has been Including artificial extensions. removed erosion proceeds up the slope. Some 3 % of the inselberg slopes are subject to gullying in this way. The upper pediments are dominated by a dense gully pattern and extensive brush was only partially successful as a bare surfaces. The vegetation is sparse silt-trap, as the torrential rains of De- thorn bush and patches of grass. The cember 1953 and January 1954 cut new soils are thin, 0.5-1.0 m in the upper channels around or under the brush, parts (cf. Fig 89). Both splash, sheet and also carried a great deal of the silt ~vash,rilling and gullying are active. ,411 and sand through the brush. Some of fallow fields in this zone are dissected by the brush itself was carried into the rills and shallow gullies, many of them reservoir. having developed down to the iveather- Still, in the 1954 dry season the work ing bedrock. Further down the slopes was continued with famine relief labour. the gullies are generally deeper, 1-5 m, Apart from packing gullies ~vithbrush - lvith vertical crusted side ~valls.Down- wood a hedge was planted around the slope below the gullied area is a zone reservoir to keep cattle out (Ann. Rep. where sheet ~vashdominates or where PWD 1934 p. 10). gullying is less evident.

Fzg 55 hlsalatu catchment. Soil col er eroded don n to iron crust (dark surface in centre of picture). The ~ihitetone is exposed suhpoil resting on the crust. Surface la~rrof laterite nodules in foreground. 7'hitkness c;f eroded laler indicated h\ the step (approx. 1 rn high) in the hackground. (Photo: C. Christiansson. Dec. 1972.) Table 12. Msalatu catchment. Sediment yield and soil denudation rate.'

Catch- Relief Period of Total sedi- Annual sediment yield2 Soil denuda- ment ratio recording ment yield tion rate area (mikm) (years) (m3) Volume Vol.iarea Weightiarea (mmlyear) (kmz) (m3) unit unit (m3/km2) (t/km2)

' Regarding differences as compared with figures earlier published, see page 5. The sediment yield in m3/km2 has been transformed into values of tons/km2 under the assumption that the dry bulk density of sediments as well as soils is 1.5 g/cm3 (cf. Christiansson 1979 p. 128 and Hughes et al. 1979).

These are the main grazing areas started growing new rills develop across within the catchment although very little the fields. grass can be found except underneath In the period 1970-74 only few cases the thorn-bushes where it is out of reach were observed where larger gullies had of the cattle (Fig 17). Nearly everywhere eaten backwards or had been exten- the top soil has been removed by erosion sively widened. The stability of the and the infertile, cemented layers are pattern is probably due to the gullies exposed (Fig 55). These bare surfaces already being cut down into the firm are particularly conspicuous along the cemented sediments. cattle tracks. Below this zone the gradi- At present no conservation measures ent steepens again and another set of are being maintained in the catchment. gullies has developed on the short slopes towards the lower course of the Msalatu Sediment yield and reservoir stream. On all newly cleared fields without capacity growing crops sheet wash and rilling are Data on sediment yield and rate of obvious. The smaller rills are later ob- sedimentation are presented in Tables literated by cultivation but when heavy 12-13. The decreasing reservoir capa- rains fall before the new crops have city is shown in Figs 60 and 93.

Tnble 13. hIsalatu reserloir. Rate of sedimentation.'

Sediment Reservoir capacity Annual loss of capacity Expected total survey life of reservoir year Remaining volume Actual rem. vol.' Period of recording Sedimentation (years) (m3) (%l (m3) (%) (years) (no of (m3) (%) seasons)

' Regarding differences as compared rvith figures earlier published, see page 5 In l952 an amount of 8,000 m3 of silt was excavated from the reservoir. MSALATU DAM tions and grazed fallow areas dominated in Msalatu catchment and its surroundings in the period before the construction of the reservoir. When the reservoir was built in 1944 restrictions on land use in the catchment were introduced. For 15 years no cultivation occurred and grazing was limited (Handley 1952 p. 4). The rate of sedimentation in the first 5 seasons of operation of the dam is not known in detail. Fawley (1956b p. 74), noted that considerable amounts of silt entered the storage basin. The original capacity of the reservoir v- y LONG PIOFIIE 3 /),- U SUPPLY LEVEL was assumed to be 358,700 m (79 mill. 0-- WATLR INLET iOWEsi \ \ 15p,llwo)I cr.$+ of 30 tt 1 gals) (Ann. Repts. PWD 1944, 1945), a O YC *BY 100 4- 500 /r/l l ' ,' '8 ' i 'X ,.' O $0 0 I50 1D"X volume that was increased considerably in 1950 by the addition of 0.61 m (2 ft) Sediment survey Nov. 1960 to the spillway crest (Ann. Rep. PWD 1950; Fawley 1956b). In 1972, a worn dye line copy of the original contour map of the reservoir site was found. It was thus possible to make a check of the 1944 reservoir volume. From the origmal map, which has 5 ft contour intervals, a new map was traced, showing the outline of the reservoir bottom topography (Fig 56). By computing the volumes of the consecutive contour intervals below FSL and then summing them the total capacity of the reservoir was assessed. The resulting figure, 354,000 m3 (77.9 mill. gals) = 1944 true max. volume, differs slightly Sediment survey Nov. 1974 from the earlier one. The difference DEPTH i~ M~TRES B A' may be due to lack of details on the origmal map or refinement of the computation methods. From the information on the map it was also possible to calculate volumes up

Fig 56 Msalatu dam. Reservoir contour maps 1944, 1960 and 1974. Figures in brackets on 1974 map show depth in feet. Numbers M1 1-20 indicate spots of sediment sampling (see Figs 66a-b). to 5 ft abo\ e the full supply level. Thus, MSALATU DAM it was noted that if a 2 ft higher spillway Accumulation of sediment 1944 - 1974 Selected cross sections

had been built already in 1944 the total m ss., reservoir volume would then have been some 420,000 m3 (92.4 mill. gals) = 1944 theoretical max. volume (cf. Ann. Rep. PWD 1950: 90 mill. gals). In 1950 "the capacity of silt was obtained" (PWD 1958) probably by means of a reservoir field survey. Thus the volume of the reservoir was determined, and in "reservoir capacity tables" (acci- dentally found among rubbish in a store room in Dodoma) referring to the year 1950 it is shown as being slightly less than 389,000 m3 (85.6 mill. gals). Theoretically some 3 1,000 m3 of silt had entered the reservoir in 6 seasons. A small portion may, however, have been trapped on the delta between the original and the new FSL. For the construction of the embankment, soil material from the areas adjoining the dam was used. It is nowhere indicated exactly from where the material was taken. If taken from inside the SEDIMENTATION E. . i5 storage basin the accuracy of the above DEC ?%?l-197' figures will be affected. m OEC 1960 - 1911 The rate of sedimentation for the first oci il'i- isso six years seems somewhat high as it re- LONG PROFILE INTERSECTION fers to years when there was no cultivation in the catchment and grazing was restricted. It should, however, be observed that the period includes the season 1946147 in which unusually heavy rainfall was experienced. To sum up, between 1944 and 1950 Fig 57 Msalatu dam. Accumulation of' sediment some 3 1,000 m3 of silt was brought into 1944-1974. Selected cross sections. the reservoir, corresponding to a total annual sediment yield of 5167 m3 or 607.1 m3/km2.The annual loss of vol- supply of 3900 m3 or 458.3 m3/kmz.By ume was 1.46 C/c. By 1950 the reservoir 1960 the reservoir volume had de- had lost 8.8 % of its true orignal volume creased from 389,000 m3 in 1950 to = 7.4 % of the theoretical max. volume. 358,000 m3. Unless de-silting had oc- The conservation measures under- curred the volume would have been taken during the 1950s (page brought 350,000 m3. The new volume was calcu- the total sediment yield for the period lated from a reservoir map produced 1950-1960 (l0 seasons) down to 39,000 from air photos taken over the area in m3, equivalent to an average annual November 1960. MSALATU DAM

Long profile v-X-y 1944-1974

t SEDIMENTATION E a DEC 1971 - NOV 1974 m DEC 1960- NOV 1971 m OCT 194A- NOV 196'; I C INCOMPLETE LONG PROFILE OCT 1967 SEDIMENT SURVEY

Average depth of cross sections 1944 - 1974 m feet Y !

AVERAGE SEDIMENTATION 6-20 m DEC 1960-NOV 1974 y!:z L25 r(lll OCT I944 - NOV l960

,600 500 LOO 300 200 100 2000 1500 ldo0 5bo 0 feet

- 1944 BOTTOM OF RESERVOIR - SURVEYED POINT A-G CROSS SECTIONS WATER INLET Towen Fzg 58 hfsaldtu darn Actuinulation of sedlment 1944-1974 Imng profiles and alerage depth of cxoss Fections

MSALATU DAM, Reservoir areo curves 1944-1974

~ -

p--- --

AWL iN :VOULA>IC SQUARE FEET Fig 59 hlsalatu darn. Rcscrvoir area cur-~cs1944-1974 MSALATU DAM, Capacity curves 1944 -1974

p ---

1 1 RESERVOIR CAPACITY iN THOUSAND CUBlt METRES 50 100 150 200 250 300 350 LOO 32 l I l I l I l l I l C 10 2(1 30 10 50 60 70 80 B0 RESERVOIR CAPACITY IN MILLION GALLONS Fzg 60 Msalatu dam. Reservoir capacity curves 1944-1974.

In late 1933 Fawley estimated the ing a marked increase in sediment yield capacity of the reservoir at 317,800 n13 in the years 1960-1971 compared with (70 mill. gals) assuming an average the foregoing period. annual rate of sedimentation since 1944 An amount of 60,000 m3 or 5454.5 m3 of 9100 m3 (2 mill. gals). As shown by per year accumulated in the reservoir, later sediment surveys the rate of corresponding to a sediment yield of sedimentation is lo~verthan anticipated 651.0 m3/km2per year. The annual loss by Farvle)., provided that large amounts rvas l. l3 % of the 1944 max. theoretical of silt have not been removed from the volume. By 197 1 altogether 3 1.0 R of reservoir without documentation. the capacity had been lost. However, as In 1967 the reservoir xvas again sur- desilting had occurred that figure is veyed. The resulting map contains a theoretical and the true remaining number of errors. It has however been capacity rias 70.95 %. utilized to construct a long profile show- The increased rate of sedimentation is ing the approximate position of the largely explained by the increase in reservoir bottom in 1967. population (Table 24) particularly in the In October 1970 a preliminary study areas N of the catchment and the conse- of the reservoir took place follo~ving quent changes in land use. Already in selection in 1968 of the reservoir and the late 1950s the earlier imposed re- catchment as one of the reference areas strictions on cultivation and grazing had of the DUSER project (Christiansson been abandoned. The 1960 air photos 1970 pp. 5-7). ~h;preliminary work show that grazing xvas again common was followed by detailed surveys and and bush fields by that time covered mapping in October and December 4.4 R of the catchment area. 197 l. The total capacity was then found By 1970 the grazing pressure had in- to be 298.000 1n3 (65.6 mill. gals) indicat- creased, and ~voodcutting for charcoal Fzg 61 Pit for charcoal burning, Rlsalatu catchment. Charcoal is probablye the nlost i~nportantsource of energy in Dodorna totvn and in settlements rvhere firewoocl is not readily available. Due to their proximity to Dodoma the investigated catchmrnts all serx-e as srippliers of charcoal to the town. Charcoal is produced through slot<-burning of trunks and bt-anches of hush and tr es. The rvood is placed in shallow pits and covered with soil to reduce the access of air during the process. In the study area the felling of tl-ees atid bush for firewood and charcoal procluctio11 means a reduction in tlle erosion protection forrnecl by the vegetation. (Photo: C. Christiansson, Dec. 1971.) was common even on the steeper in- years 1971-74, 14,000 111' of silt had selberg slopes (Fig 61). Cultivated or entered the reservoir, corresponding to recently abandoned fields now covered an annual sediment yield of 4666.7 m3 some 20 '% of the catchment. (548.4 m3/km2).This period of only 3 One of the effects of the increased years is too short horvever to be repre- pressure on the land was a 40 % in- sentative if treated separately. Occa- crease in the sediment yield in the peri- sional climatical extremes affect the re- od 1960-70 compared with the period sult greatly and minor errors in the field 1950-1960. survey arid in the nlap construction be- To counteract the effects of the loss of come overemphasized. reservoir capacity through seditnenta- The raising of the spillma~in 1972 tion the spillway rvas raised by- approx- means that a good deal of the sediment imately 0.3 tn (1 foot) in December is non trapped h~gherup on the delta, 1972. The capacity Ivas thereby in- ~naterial M hich I\ ould have been washed creased from 293.000 m3 to 330,000 1n3. further do~tnif the FSL had been lor\er It1 early November- 1974 a complete In the 1974 sune\ the L olurne account- sediment survey bvas again undertaken. ed for tlas onl\ that ~hichhad accu- The capacity of the reservoir below ~nulatedbelou the old FSL Because of the 1950-72 spilllvay lel,el was by then this it r\ould prohablt be reasonable to 284,000 n13 (62.5 tnill. gals). In the three add some 6-8 r( to the total ~olumeof Fzg 62 Advancing front of sand lobe on the main delta. Msalatu reservoir. The lobe, the result of one heavy rainstorm, extends from the main inflow (right in picture) and covers a more fine grained layer (darker tone due to grass vegetation) to the left (cf. Fig 64). The hill in the background is Chimwaga. (Photo: C. Christian5son. Oct. 1970.)

sediment recorded in 197 1-74. That Morphology and types of ~vouldimply an average annual sediment yield for the period of almost 600 sediments m3/km2. The majority of the sediment which en- The total loss of capacity recorded in ters the reservoir is carried in the main the period 1944-1974. including the channel (1) hut one of the tributaries volume of silt excavated Tvas 144,000 m3 (channel 2) also supplies considerable or 34.3 %. The average annual sedi- amounts. Thus, in front of these ment yield in the same period was 364.0 streams distinct deltas have been m3/km2. formed (Figs 5 1, 56). In November-December 1974 an While gradually advancing, the main estimated 6000-7000 m3 of the sedi- delta has cut off the entrance of the ment in the delta area lvas excavated to southernmost of the tributaries which be used for building purposes in Do- nou. forms a pond separated from the doma. It is assumed that these activities rest of the storage basin. Another of the proceeded into 1975. Vnfortunately. tributaries is-at present-being cut the exact volume excavated does not off in the same way (Fig 62). seem to have been recorded. hlean~vhile,upstream sedimentation Fig 63 Part of delta at the mouth of channel 2, Msalatu reservoir. Around 1970 channel 2(Figs 51 and 56) flowed across its own narrow delta in a channel bounded by natural levees. A heavy rainfall in the catchment in mid-December 1972 caused one of the levees to collapse and considerable sediment masses were washed out towards the spillway area. (Photo: C. C:hristiansson, Dec. 1972.)

takes place in the channels above the FSL. The lower reaches of the channels are being filled by sand and silt. Thus, cm ,- - immediately upstream of the delta the 0-105 Cloy and sllt wllh bed of the main stream has been raised ..- mi~r~-"or~e~ 10-20< Sand and grovel some 2 m indicating a decrease in 20-24 '- F~ne~nd channel gradient from 0.0 15 mlm at the 21.27)~ clay time of the construction of the reservoir to 0.0035 m/m at the time of the 1974 sediment survey. From the present upper delta-front the slope towards the embankment has increased to 0.020 m/m from an original gradient of 0.0 15 (Fig 58). The maximum average reduction in depth has occurred at cross profile E Fig 64 Profile of sedi~nentsat pit %lIf. Msalatu where the original average depth was reservoir. The location of the pit is shown in Fig 2.8 m and is now 0.75 m. The maximum .56 -. The sedi~nentsof the I-eservoir are deposited as total sedimentation, 5.3 m, has occurred distinct layers, or varves, each indicating one sepa- where profile D crosses the former rate flow into the basin or the total inflow during stream channel. Closer to the ernbank- one season. Often coarse nlaterial at the bottom of. each layer grades upwards into finer fractions. ment the total sediment accumulation is In the delta areas the layers are thick and less pronounced. However, in this area dominated by sandy and silty material (cf.. Fig 62). the max. depth of the reservoir has de- In the deeper parts of the reservoir micro varves with a high percentage of fine fractions are corn- creased from 10.4 to 6.5 m. [non. They are generally of a darker colour than The decreasing capacity of the reser- the coarse delta material. Similar occur-rences of layering have been described from Ikowa, voir is also illustrated by the capacity Kisongo and Imagi reservoirs (cf. Murray-Rust curves and the area curves (Figs 59-60). 1972 p. 339; Christiansson 19711 p. 128;). tblp 14. Grairl size composition (~veight%) of soil samples from Msalatu catchmetlt. Dodoma, December 17 1. Analysis by Leif Tjernstriirn, Dept. of' Physical Geography, Uppsala. Samples 21a and b, data fi-on1 ~rvley19.56b. mple Depth below Grain size (mm) o surfice (cm)' >2 2.0-0.2 0.2-0.02 0.02-0.002 <0.002 Notes

11.2 7.0 Abandoned field. Upper pediment 11.2 13.6 Abandoned field. Upper pediment 7.4 6.5 Gully cutting. Upper pediment 11.6 7.4 Cultivated field. middle pediment 4.2 27.7 Accumulated material. Lower end of cultivated field )a 0-60 39.2 37.0 8.Y 14.9 0.0 Exposed cemented sediment )h 60-120 4.2 48.8 32.2 11.6 7.4 Cemented sediment in channel .i\,all Grain sire (mm)

a 30-60 1.25 10.58 29.81 18.38 18.41 14.26 7.3 1 Surface material near reservoir b 60-90 0.81 13.20 32.70 14.20 16.97 17.77 8.25 Surface material near reservoir

Grain sire composition and in Yovember 197 1 in channel 2. approximately l50 m from the mouth of the concentration*. Of stream. The stream has a small catch- sedlment ment covering only l km2. The sediment deposited in the central parts of the reservoir is predominantll- clay, silt and fine sand; that deposited near the entrance is predominantly coarse sand and gravel. It should, her\-- ever, he noted that coarse layers occur among the more fine grained in the deeper parts of the reservoir and fairly thick layers of clay and silt are found among the delta sedi~nents.The grain size ~0l~p~l~itiOllOf ample^ collected F,~147 Derail the uppermost 7 (,fthe from the reservoir is shown in detail in shown in Fig 64. The upper 4 cm ctltlsi5t of undisturbed 1a)ers of fine material po,c~blyindicat - Fig 66a-h. ing deposition during one 5eason. Belot\ is a Lone To record concentrations and grain wirh inclistinct layers affected h\ insect actixitv a11ti sizes of the sediment suspended in the "a( kirlg cle\eloped in rhe prcccding dr) seaton. l'hc deeper parts of the profile are dominated h\ water flo~vinginto the reservoir. an au- sedimenr 64 (Photo: (,, tomatic sediment sampler Lvas installed (:hr-istiansson, Dec. 1971.) C LAY SILT SAND GRAVEL

CLAY SILT SAND GRAVEL

Fig 66a-b Grain size composition of sediment samples from Msalatu reservoir. Samples XI 11-14 represent accumulation on the upper parts of the delta near the main inflo~iinto the reservoir. Samples M 15-16 represent delta "foreset beds". Sample hI 19 from bed of channel 2, above FSL. Samples M 17-20 from central parts of reservoir and from spill~vayarea (Fig 56). All samples except no 17 represent a 15 cm surface layer of the deposits. \vhich in the centre of the basin is made of ~naterial accumulated during several seasons and on the deltas usually consist of material deposited during one single season. No l ia represents a laver 0-10 cm belo~vthe surface. Nos 17b and c represent depths of 30 and 50 cm respectively (Fig 64). C LAY SlLT SAND

Grain size, mm Fig 67 Msalatu dam. Grain size composition of suspended sediment. The 12 samples shown were collected in channel 2, Msalatu in February 1972, during the rising stage of a flash flood caused by a rainfall of 85 mm in one day. Sampling in automatic sediment sampler Hayim 7. Lowest intake (sam le no 1) 15 cm above channel bed, thereafter one intake every 5 cm. Highest intake (sample no 12) 70 cm above channel bed. Lowest concentration of sediment (sample no 2) 14.98 g/l, highest concentration (sample no 10) 74.72 gll. High percentage of clay and silt indicates that material originates from intergully splash and sheet wash.

When installing the sampler the loose ures are in the same range as the con- sand was removed down to the crusted centrations measured in runoff water bottom of the channel. At the bottom a 1 from erosion plots at Mpwapwa by m3 concrete block was placed to which Staples (1935 p. 139). There, the aver- the sampler Jvas attached by means of age annual concentrations during two iron bars. The sand was then replaced to years of recordings ranged from 26 to cover the concrete block. 87 g/l. The figures include suspended The first rains of the 197 1/72 wet sea- load and bed-load. son did not cause enough flow even to It was intended to continue the re- reach the lowest intake. However, heavy cordings but due to lack of field rains in January (85 mm in one day) filled the bottles of the sampler with wa- Lower Bubu, while in that same season con- ter containing sediment concentrations centrations of 5.0-2 1.0 g/l were recorded in ranging from 13 to 75 g/l.' These fig- the Upper Bubu at Kondoa and 24.0-52.8 gll in the tributaries Kelema and Mkondoa (WD8cID 1968 p. 3). Such massive suspend- 'Such high concentrations have been re- ed sediment concentrations were considcred ported even from the larger rivers in the by the author of the report to be among the region. During the 1959160 season concen- heaviest in the \vorld in a sizeable river (cf. trations of 0.6-0.8 g/l \\.ere recorded in the Stromquist 8c Johansson 1978 p. 49). Tnblp 15a. Msalatu dam. 'Total arnount of During 12 months in 195 1/52 the suspe~~ded[natter in the reser~oir. concentration of suspended sediment was recorded in Msalatu reservoir. That Date Vol. of water Total susp. matter is the only existing regular check of sus- in reservoir in reservoir (1,000 1n3) (m3) (ppnl) pended sediment in the investigated reservoirs. The average" amount recorded during the sampling period was 605.8 ppm, ranpng from 47 1.4 at high stage in May to 773.6 at low stage in December-January. The total inflow of water into the reservoir during the period of investigation kvas >550,000 m3, 210,000 m3 of which were consumed and some 250,000 m3 were lost due to evaporation and seepage. More than average 185 101 605.8 90,000 m3 with an average suspended sediment concentration of 584.4 ppm Source: Fawley (l956b) flowed out of the reservoir over the spillway in late March and April. Thus in 12 months a rnirlinlum of 128 m3 (the personnel no more recordings were highest consumption of water is record- made in the 197 1/72 wet season. In De- ed in the months with the highest sedi- cember 1972 the same rainstorm that ment concentrations-the dry season) changed the stream course across the left the reservoir with the consumption delta washed the sampler away. It was water and another minimum of 53 m3 later found some 30 m do~mstream left with the overflow water. Altogether partly buried in the sand. In January the suspended sediment that left the 1973 it Mas reinstalled but before it reservoir from June 195 1 to May 19.52 could be safely anchored it was again amounted to l81 m3 or 4.6 % of the washed away and this time it was buried 1950-60 average annual sediment yield. so deeply in the sand that it took 14 The 195 1/52 season was above the aver- people 4 hours to get it out again age for the 1945-53 period concerning (Personal communication H. Kuschel, water consumption and near average as Ministry of Water Dev. and Power). It concerns evaporation and seepage. The was then not possible to reinstall it again overflow losses, however, were more before the end of the rains as the track than 10 times as high both in 1946147 leading to the reservoir was cut off by and 1949150. two streams making it impossible to Fawley (1956b pp. 78-79) character- bring cement to the site. ized the \vater in the reservoirs as "soft

Table 15b. hfsalatu and Matumb~~ludarns. Chemical analysis of the water in the reservoirs. Fehruar).. 1970.

Keservo~r pH zz0.106 Ca Mg Sa K A(HCO,) SO, Cl Si mg hfn rng

Msalatu 7.37 53.8 0.244 0.119 0.134 0.072 0.428 0.123 0.043 4.31 0.00 Marumbulu 7.18 46.4 0.134 0.084 0.187 0.07 1 0.375 0.069 0.047 4.10 0.00

Sou~ce:Rapp, Murrat-Kust et al. (1972). but rusty ~olouredand containing a Results from chemical analysis and great deal of colloidal and suspended analysis of the concentration of su- matter . . . . Water seeping through the spended material in the reservoir water strata below the dam is far clearer than are presented in Tables 15a-b. the reservoir water, but is also harder". 10. The proposed Mkonze dam and catchment

After the 1949 water shortage, the in- to be 10.3 and 7.6 sq. miles respectively adequacy of the water supply afforded in the draft report by James (1950 p. 5). by the two existing dams became appar- The inefficiency of part of the catch- ent and investigations for a new damsite ment is due to a number of small were initiated. streams in the NE part of the area end- A site at Mkonze, 6.5 km SSW of Do- ing in a small mbuga, area1 extent less doma, near the Dodoma-Iringa road than 1 km2, that lacks visible drainage to was selected and surveyed in 1950 (Figs the rest of the catchment (Fig 54). Faw- l l, 72). However, the investigations re- ley considered that during years of low vealed that the site would probably be rainfall the runoff reaching the pro- expensive to develop, as, in order to posed reservoir would be derived almost make full use of it, a considerable struc- entirely from the reduced catchment, ture ulould be required. but during wet years the runoff would In 1952 diamond drilling disclosed be from the entire catchment. that unweathered bedrock was situated On the S boundary of the catchment at a depth of 40-50 m below the surface. some runoff flows into a swampy area This great depth of superficial sedi- lying astride the lvater divide. Part of ment, mainly claycrete, overlying the that water probably seeps out of the bedrock meant that the dam construc- Mkonze catchment into the ineffective tion had to be reconsidered, but as it was part of Matumbulu catchment to th,e S supposed that all large river valleys in (Fig 73). the Dodoma area would contain thick Fawley (1954 p. 22) estimated the layers of loose deposits, Mkonze was still runoff from Mkonze catchment, using thought to be one of the best sites. It was the runoff figures from Msalatu catch- thought that silting of the dam should ment (eight seasons, average 14.5 % not exceed the degree of silting record- runoff) which at least in some respects is ed in the Imagi and Msalatu reservoirs. similar to Mkonze. He arrived at a prob- It was estimated that the water would able minimum runoff of 222,500 m3 in contain about 600 ppm of supended dry years with runoff from the reduced matter (Ann. Repts. PWD 1950 p. 7; catchment only and a maximum runoff 1951 p. 6; 1952 p. 14; Ann. Repts. GSD of 3,518,500 m3 in wet years with runoff 1950 p. 47; Handley 1952 p. 8; Fauley from the larger catchment. The average 1954 p. l, 24). would be around 830,000 m3. Assuming The catchment area was surveyed in such high runoff figures a reservoir 1950 by T. C. James. It was found to ~vitha capacity of 1.4 mill. m3, as pro- cover 24.1 km2 (9.3 sq miles) of which posed, ~vouldhave been filled in 4 )-ears only some 70 % or 16.6 kmz (6.4 sq out of 8 in the period 1945-53 or 20 miles) were considered to be fully effec- years out of 30 in the period 1944-1973. tive. Holvever, recent detailed meas- In these calculations capacity losses urement on large scale (1: 15,000) maps through silting have not been accounted indicated a total catchmenr size of 23.56 for. km2. Note that the total and the effec- When half full the dry season daily tive catchment areas are rvrongly stated losses from the reservoir by evaporation SILT SAND GRAVEL

100

9 0

3 60 ci ,250 3 40

3 0

2 0

0 0,0062 0.0 2 0,074 0,125 Q25 0.5 1.0 2 4 8 Grain size,mm

Fig 68 Grain size analysis of material from an abandoned river channel 2t Mkonze. Samples taken from depths between 0.3 and 5.0 m. (A: 0.3-0.6 m, B:0.6-0.9 m C:1.2 -1.5 m, D:1.5 -1.8 nl, E:2.7-3.0 m, F:3.6-3.9 m, G:4.8-5.0 m). Note the gradual change ~vithdepth from a dominance of the sand fraction to sandy clay. Samples F and G represent cemented clay (claycrete). Between 3.0 and 3.5 m is the boundary between the loose (recent) sediments and the older cemented sediments. The sorting is more pronounced with the loose sediments than with the cemented sediments. (Data from Fawley 1954.)

were estimated at 0.24 % of the actual However, as an alternative to building ~rolume.It would decrease to around another dam and reservoir at Dodoma 0.16 % ~viththe reservoir full and in- the underground water supply was increase to 0.35 % with the reservoir filled vestigated. In 1948 deep trial boreholes to only 15 % of its capacity. were sunk N of the town, but the water The Mkonze (Mohansa) river on its obtained proved to be saline. No prom- way across the alluvial valley floor ising supply of suitable underground carries large amounts of sediment and it water was found until a successful bore- \$asrealized that the degree of silting of hole which yielded plenty of water of a dam at the proposed site ~vouldbe low salinity was drilled at Makutapora considerable. To prolong the effective 24 km N of Dodoma. Following the suc- life of the dam it was recommended that cessful conclusion of the investigations far-reaching preventive measures the PWD started the construction of a should be undertaken partlv on an pipe line from Makutapora to supply experimental basis under the combined part of the water required at Dodoma. viglance of the Water Development In September 1955 the new water sup- Dept. and the Geologcal Surve) Divi- ply scheme was taken into use (Handley sion (Handle\ 1952 p. 4). 1952 pp. 8-9; Ann. Repts. GSD 1953 p. l l; 1954 p. 12). l I. Matumbulu dam and catchment

then had a maximum height of 10.06 m Construction and ~otential1 use of the dam (33 ft) above the river bed and a length of 219.5 m. The width of the spillway In 1960 the Water Dept. completed 33 channel was 48.8 m (160 ft). The con- dams, more than in any previous year. struction of the embankment included One of those was Matumbulu situated some 25,000 m3 of earthwork. Apart alongside the Dodoma-Iringa road 12 from the erection of the embankment km S of Dodoma (Figs 11, 73). It was and the cutting of the spillway channel described as an "earth dam for cattle the project included realignment of a with a capacity of 310 acre ft (382,400 part of the Gt. North Road that other- m3) and with a surface area at FSL of 48 wise would have been flooded. Total acres (19.5 ha)" (Ann. Rep. WD&ID expenditures for the work amounted to 1960 p. 18). Construction data are 140,000 Sh. (Ann. Rep. WD&ID 1960 p. summarized in Table 16. 18). Surveys of the catchment (7sq miles In November the first rains fell over 2 -18 km ) and the dam site were carried the Dodoma area and the reservoir out in March 1960. By July the started filling. The average annual embankment was near completion. It rainfall for the area had, rather

Fzg 69 Oblique aerial view of Matu~nbuludam. Note eroded slopes in centre of picture and dense shore vegetation around upstream part of reservoir (cf. Figs 81-83). (Photo: C. Christiansson. Sept. 1971.)

98 Tahlc 16. hlnt~~~~~buludam. C:onstruction data.

Year of construction: 1960 Failure of embankment: 1961 Reconstruction conlpleted: 1962 .Area of iatchment: 17.93 ktnZ(effective 15.0 km2)

Original dam 1960 After reconstruction 1961-1962 .4ltitude (asl) of river bed at clam site Altitude (asl) of spill\\.a! crest Assumed level of river bed (Reference level) (169.00) Length of embankment 2 19.4.5 m (720 ft) Height of embankment (above river bed) 10.06 rn (33 ft) 10.67 m (35 ft) Width of crest of embank~nent 3.66 rn (12 ft) 3.04 m (10 ft) Slope of embankment, upstream 1.3 1:3 Slope of embankmerlt, downstream 1:2 1:3 LVidth of spillway channel 48.77 m (160 ft) Height of spill\\.ay crest above ri\-er bed = max. toral depth of resenoir 7.93 m (26 ft) Height of sluice invert above ri~er bed ' 1.83 m (6 ft) Height of spillx\ay above sluice invert = max. effective depth of reservoir2 6.09 m (20 ft) Gduge read~rlgat full suppll le~el(FSL) (200 0) hlax. length of storage basin 900 m LIax. width of storage basin 200 m Reservoir surface area 194,256 mZ (48 ai)? 141,645 m2 (35 ac)' 'I'otal reservoir capacity 382,382 rn3 (3 l0 acft) 363,880 m3 (295 a~ft)~

I Sluice invert level = reference level 176.00 (0.00 ft on gauge) ' Spill\\.a) crest lex el (FSL) = reference level 196.00 (200.0 ft on gauge) Embankment top level = rcference level 202.00 Revised figure: 135,.500 m2 ' Revised figure: 333.000 m3 optimistically, been estimated at 24 in- seem to have been documented. Holv- ches (609.6 mm). With that rainfall a ever, according to the rainfall records of runoff from the catchment of 3.55 R the area it must have occurred on would be suf'ficient to fill the reservoir. February 12 or 13, 1961. Ho~vever.in the period 1962-1974 an hose were the only days during the average annual rainfall of only 501.1 first half of the rainy season that the mm r\.as recorded (Fig 70). Dodoma area experienced particularly Shortly after the dam M-as completed a heavy rains. Unfortunately the Matum- heavy storm in the catchment area bulu Dam rain gauge was not yet in op- caused the :\.ater level to rise rjithin a eration. fe~rfeet of the spill~\.a)..At this stage a Totals for the rainfall stations in the ~veaknesson the right bank caused the Dodoma area for 12-13 February 1961 embankment to fail, fortunately ~vithout (2 days) were as follo-rvs: Dodoma Met. damage to property do~vnstrearn(Ann. Station 53.9 mm. Imag dam 58.4 mm, Rep. \1'D&ID 196 l p. 13) (Fig 76). Dodorna Ceol. Office 72.4 mm, Alliance The date ~vheilthis occurred does not Sec. School 38.1 mm. hlsalatu dam 93.0 mm 10001 Calendar year N, a total of 485.7 mm was recorded in December and January. By January 1962 a raingauge had been installed at Matumbulu dam. In the first month of operation 260.6 mm was recorded. In the 1962 dry season reconstruction 500 years average) work proceeded. When the first rains of the 1962163 rainy season fell over the catchment at the end of December 1962 the dam operated properly again (Ann. Rep. WD&ID 1962 p. 16). By then the height of the embankment had been raised by two feet to a maximum of 35 ft (10.7 m) above the orignal level of the river bed (Ann. Rep. WD&ID 1963 p. 17). The downstream slope, which oriqnally was 1:2, had been changed to 1:3, the same as the upstream slope. The capacity was now given as 295 ac. ft (363,880 m3). The work with the reinforcement of the dam proceeded into 1964. By then 445,000 Sh. had been spent on the re- 100qJ RalnfoLl year mm construction work and improvements. Fig 70 Rainfall at Matumbulu dam 1962-1974. Matumbulu dam was constructed for Data for the 1963164 and 1964165 seasons uncer- the primary purpose of supplying water tain due to missing records for certain months. to cattle. It is, however, possible that the Note the extremely dry season 197017 1. potential use of water for irrigation downstream was also considered. In mm. In each case more than 95 % of the 1970 close investigations were made by amount mas recorded on one day, either WD&ID people in order to check the the 12th or the 13th. (There are varia- possibilities for irrigation and domestic tions in the principles of recording.) water supply for the proposed Matum- In the 196 1 dry season repairs were bulu Ujamaa Village 4 km do\%nstream commenced. The whole of the river bed (Fig 11). The water requirements for was cleared down to rock under the irrigation of a vineyard of 40 ha were broken section of the embankment. A estimated at 1225 m3/day for 7 months new corewall was constructed on the or 256,000 m3/year. For domestic use right bank as well as horizontal stone some 70 m3iday would be needed all the and gravel filters at the base of the dam. year round, making a total of 25,000 m3. For the core wall, impervious black That volume tvas to be shared between cotton soil was brought to the dam site 800 persons requiring 45 l/person/day,' from an area 2 km NW of the reservoir (Ann. Rep. WD&ID 1961 p. 13). The reconstruction work was held up by heavy floods in December 196 1 and ' The present average domestic ~vaterco11- January 1962. At Mvumi mission SE of sumprion, escl~iding\\.ater for livestock. in the villages of the study area is 3040 1,'day hlatumbulu 443.5 mm was recorded in per household or 3-8 ]/capita (TXIIVD 1974 December alone. At Imagi dam, to the X p. 2). MATUMBULU DAM Dally ra~nfalland reservoir water levels 1969, 1972 and 1973 60 l '. .20 I l969 1 50 - l 18 l ~ I l 40 l 16 '1 I /I 30 -- -- 14

I 20--1 l -- 12 I Total ra~nfoll1969 271 8 mm l 10-- -

8 JAN. FEB. MAR. APR. MAY JUN. JUL. AUG. SEP. OCT. NOV. DEC.

Fig 71 hlatumbulu dam. Daily rainfall and reservoir water levels in 1969, 1972 and 1973. Lowering of \vater surface due mainly to evaporation and leakage and watering of cattle. Rapid lowering from October 1973 due to pumping for use of water in I\.latumbulu Ujamaa Village. 1969 dry, 1972 wet and 1973 normal year. Rainfall recorded once daily. Water levels are read three times daily.

1200 cattle requiring 23 l/head/day and Water Development Authorities in Do- 700 small stock requiring 5 l/head/day. doma Mere doubtful whether it was It rvould be possible to draw 25,000 ~vorth going ahead with the project m3 for the domestic icater supply but (WD&ID 1970). According to the survey there \could be far too little water left the reservoir had lost 113 of its volume for irrigation in a dry year like 1965 or in 7 years due to sedimentation. More- 1969 and there ~vouldnot be enough over, evaporation and seepage losses left even in a \vet year. In the light of the \\ere knom n to be considerable. 1969 preliminar). sediment survey, In hfarch 197 1 it Ilas decided central- carried out by the DUSER team. the l\ at WD&ID in Dar es Salaam that there Fig 72 Imagi. Msalatu and Sfkonze catchments. Air photo 49!TN!6/126. June 1960. The dense gully pattel-n in Msalatu catchment is rtriki~lg.In the lolrer right corner of picture are examples of hush fields extending up the slope of Ilende hill. Note gullies developing across the fields. Light area in centre of picture is a small mhuga (cf. Fig 54).

was no future for an irrigation scheme \ias also stated that lvith the high rate of of 40 ha and it xvould not be economical sedimentation a bad year in the near to construct a pipe line for a domestic future would not even provide enough water supply project based on hlatum- water for domestic purposes. bulu reservoir as a source of \vater. It It rvas suggested that the local \vater authorities should look for "some ber of tributaries the Matumbulu river alternative source of water" (WD&ID drains some 165 km2 of the southern 1971). slopes of Dodoma hills. The river is However, later sediment surveys have short but impressive. Where it leaves the shown that the possibilities of using the hills it has a width of between 50 and reservoir water for domestic purposes 100 m (Fig 73). It ebbs away in an and irrigation are not quite as bad as mbuga area SE of Luatu only some 30 indicated by the earliest investigations. km from its sources. The Mohansa is The rapid sedimentation gives no room still shorter, flowing westwards for a dis- for long term projects, but as a tempo- tance of less than 20 km before ending rary solution of the problems concern- in the Nzinge swamp. ing water supply for Matumbulu Uja- Those parts of the rivers which are maa Village it should be possible to use situated within the study areas comprise Matumbulu reservoir. only some 6 and 10 km respectively of As the reservoir is being filled mainly the total river courses. The width of the by sandy sediments it might be possible stream channels here range from a few to use it as a ground water reservoir m up to 10-20 m. They are usually in- after it has been filled up completely cised from 1 to 3 m into the loose de- (Fig 83). The ground water capacity posits and they follow relatively stable may be as high as 113 of the original meandering courses across their flood storage volume or the same as the total plains. The slope of the SW tributary, pore volume of the sediments Matumbulu catchment, which was (3040 %). Evaporation losses will de- measured in detail, is between 0.5 and crease markedly, making available a to- 1 from the delta and some 200 m tal volume of water of some 100,000 m3 upstream. The gradient then increases annually, provided the area is kept free to 2 %, a slope that remains unchanged from deep-rooted and \vater consuming all the way up to the valley head. weeds. In the NE end of Matumbulu catch- Since several boreholes in the area ment a small earth dam with a capacity failed in the years 1970-72, the water of a few thousand m3 has been con- situation for the village became acute structed across the stream channel. (WD&ID 197 1). Thus, \%,ater from this part of the Thus, in the dry seasons of 1973 and catchment flows past the dam only at 1974, water was taken from the reser- high water stage. However, even before voir (Fig 7 l) and transported by truck to the construction of the small dam this the village where it was used without part of the catchment must have been a filtration or other treatment. (Concern- relatively insignificant source of water ing the quality of the \$.ater see Tables and sediment for the larger hfatumbulu l 5a-b.) reservoir. Most of the sediment is trapped further upstream where the drainage passes across a grassy plain with no Drainage defined stream channel. Xfkonze and Matuinbulu catchments are drained b) the hlohansa and the Erosion and conservation Matumbulu (Mandina) rners (Fig 11) tno of the Inan\ streams in the area Erosion flow~nginto Bahi suamp or other smal- At the time of the 1960 aerial surver, ler depresiions Toyether ~titha num- vegetationless surfaces, caused primari- Fig 73 Matu~nbulucatchment. Air photo 49/TR'/6/153. June 1960. The dam, in centre of picture, was under construction when the photo was taken. Remnants of contour bank systerns are visible on both sides of the road NW of dam site. The large sand fan shown on Fig 30 is situated N of the dam site on the right side of the road. The fan is supplied by sediment originating on bush fields and grazing areas on the S slopes of Nyankali ridge (cf.Figs 11, 30 and 54). The photo gives a good idea of the inselberg terrain in the study area.

1) by overgrazing, occupied some 8.5 % :cpllying. The erosion, as reflected by of the Maturnbulu catchment. This fig- the density of the gully pattern and the use does not include cultivated areas size of the gullies, was most intense in ~vhichlack vegetation during part of the the SW and NE parts (Fig 34). The rea- year. Some 17 % was affected by serious son for this distribution does not seem 7 -mm - U Y C

.- =>. .P$ 3 C m: bl: 9, 11 v; " X

X 2. am XXX_x C C XXXX 2 E E E E 5?$z G-c Q xxxxxa 09-3 .M e maza= >,X8 g 3 8 --m- C c C C C ,lJ---- *.XXX -G-GZ-G 2 2 2 2 2 &&+-AAA CCCC wrn* ,c F ccccP2:2 mmmm m m;% m.2.2 2 m m m mmwm

.-.-.-.-.- m c XW.XXX-aC m " + X >. r. X 'Z 2 2 2 2 :%%z?-$% 000COddEdgd 03OG to be due to the length or steepness of rather intensely used for cultivations the pediment slopes as they do not dif- until some time around 1940. The fer considerably from those in other pediment vegetation on the 1949 air parts of the catchment. photos is a sparse open bush dominated In the SW, however, the inselbergs by small shrubs indicating recent above the gullied pediment are larger abandonment. Only few fields were un- and higher than in other places along der cultivation. During the 1930s and the catchment boundary. A possible rea- 1940s the Agricultural Department son for the drastic effects of erosion in carried out an intense propaganda pro- the NE is the loose and poorly consoli- gramme for soil conservation, including dated deposits associated with the contour banking, and restrictions were assumed extension of the Iyurnbu-Ma- introduced against cultivation and graz- dengi fault line (Fig 4). The gullies in ing on the upper pediments. that area have small drainage areas with At the same time the Wagogo were low gradients, but still several of them taught to use cattle manure, which they reach an impressive depth and size. had not been doing before (page 45). The distribution of erosion within the During the 1940s ~nanuringbecame catchment seems to have changed very more and more widespread and a semi- little during the period from 1949 to permanent type of cultivation could be 1974. Where it has been possible to maintained in the valleys. Many of the compare the size of single selected fields 011 the upper pediments, ~,hichat gullies (on air photos fro~nthe different that time rvere already badly eroded and years) few extensions of the systems most probably gave poor returns, rvere have been observed. abandoned. The upper pediments then However, the field studies showed served mainly as grazing areas and land that the gullies, particularly in their for fuelwood production. downstream parts, widen through bank By 1960 the bush-vegetation had got erosion. Extensions in the form of up- denser, but grass seemed to be scarce. In slope gro~ithof existing gullies also the early 1970s the bush rvas still denser occur, particularly along cattle tracks and partly impenetrable. (Fig 23). In 1974 a general cultivation cam- On the N slope of Gobi and N of' paign was launched in Tanzania. The Matumbulu reservoir areas of badland aim was to increase the production of character occur. These areas were de- food crops in order to make the country void of vegetation already in 1949 and less dependent on imports. Around Do- by then the top soil and parts of the doma, and particular1)- in hIkonre subsoil r\.ere also gone. Up to 1974 the catchment and its surroundings, many regrowth of vegetation on these spots fields were again cleared on the upper has been negligible. pediments (Figs 33, 74). The result was Many features in the catchments indi- increased erosion and in some cases new cate that at least a number of the gullies gullies rvere formed (Christiansson 1979 Tvere formed relatively long ago. This is p. 141). evidenced by the dense vegetation in ancl along gullies (Table 17). In many Conservation gullies the vegetation Lvas \yell established already in 1949. During the later half of the 1930s and in Froni 1949 and o~~wardsthe devel- the 1940s, large contour-banking prog- opment may be traced on the air photos. rammes Liere initiated in the Central It seems that the pediment slopes Tvere Province. In that penod parts of the which is typical along the sides of older inactive gullies. All this is evidence of the relatively late date of formation of these gullies. Obviously the conservation programme failed. The banks were not strong enough to keep the runoff water back (Ann. Rep. Agr. D 1955 p. 25). On occasions with intense rain the water broke through one contour bank after the other. The reason for the collapses may have been seepage underneath the ridges or defects in the design.

Upstream sedimentation Upstream sedimentation in Matumbulu catchment occurs in four different positions (Fig 54): Fzg 74 Bush field on inselberg slope, Mkonze catchment. Observr the steep slope, coarse stony r On alluvial fans. soil and typical vegetation. (Photo: C. Christians- r In a swampy area in the NE part of son. Nov. 1974.) the catchment. r In the stream channels. r In the main reservoir and particularly pediment slopes in Matumbulu and in the delta area. Mkonze catchment were "banked". (Ann. Rep. Agr. D 1940 p. 6). In positions where the velocity of the The systems of contour-banks stand water running down the gullies sudden- out clearly on the 1949 air photographs. ly decreases, i.e., where relatively steep Although gradually disappearing they pediment slopes end directly on the flat can still be distinguished on later air valley bottom, the transported sediment photos (Figs 69, 72). is deposited. It forms fan-shaped The ridges, approximately 112 m accumulations extending from the slope high, run parallel to the contours at cowards the lowest parts of the valley. spacings of 2-3 m. They were built on Such fans are typical also for parts of the least damaged parts of the slopes. Ikowa catchment. Areas already affected by gully erosion This type of accumulation is rep- do not seem to have been treated. resented by the fan in the SW part of the Today gullies may be found running catchment. It has a typical cone shape right through the system of ridges in with a convex upper surface. On the places leaving narrow slope sections 1960 air photos there are signs of culti- with remnants of banks. These gullies \-ations on the water divide above the also dissect cultivated fields which most fan. In 1949 no cultivations existed in certainly were intact at the time of the the area. Cultivations in such an ex- construction of the banks as these were posed position may have added built partly to protect. the fields below. considerable amounts of sediment to the The guilies also lack the vegetation fan but as far as can be judged from the air photos the size and the shape of the cone has not changed since 1949. In the NE part of the catchment there is a whole series of accumulations of this type deposited at the mouth of gullies where these reach the lower pediment or the valley bottom. The largest of these accumulations is exactly on the NE catchment boundary. It is situated at the mouth of an impressive gully, 8-10 m wide and 34m wide, but less than 1 km long. It drains a severely eroded area of badland character. The infiltration of water on the fan is significant as, according to the local population, a number of wells in the lower part yielded plenty of water all the year round. The fan has been built up to Fig 75 Culverts choked by sediment, NW of such dimensions that nowadays it Matumhulu dam. By 1974 the passage under- extends into Mkonze catchment. neath the road was completely cut off and the water flowed across the road. (Photo: C. Christ- Sandy accumulations are also found iansson, Jan. 197 1 .) in places where smaller stream channels or gullies reach channels of higher order. Sand may accumulate in such positions because of rapid infiltration and surface but a network of sandy rills indi- decreased flow of surface water. As part cate the path of the water at the latest of the surface flow is infiltrated into the flo\v. bottom of the channel the sediment The fan extends as far as the NE concentration in the remaining surface channel which cuts its front. Sediment flow will increase. If, in such a situation, which reaches the front is transported no tributaries add more water to the further in this channel to the reservoir. stream a successive aggradation occurs. Most of the material on the fan seems to This may start with filling of the stream oripnate from cultivated fields on steep channel whereafter sediment later slopes (>5") on the S side of the Nyank- added is spread in a fan-like shape if no ali ridge. obstacles are present (Schumm & Had- Sedimentation in the river channels is ley 1961 p. 12). also considerable. The depth of the de- An example of this type of accumula- posits varies between 0.5 and 3.0 m. To tion is the complex fan at the confluence get an idea of the volume of the sedi- between the N and NE channels in ments in the channels some wells in the Matumbulu catchment (Figs 30, 54). river bed were studied during the The h' channel above the h11is cut 1 m 197 1/72 dry season. into the surface deposits. It has vertical A large well in the main stream 100 m sides and a sandy bed. When approach- upstream of the 197 1 reservoir FSL had ing the upstream part of the fan the a total depth of slightlv less than 3 m. channel depth gradually decreases. The The ground water surface Ivas situated fan has a typical convex form both in 2.70 111 below the present river bed and cross profile and long profile. There is probablj. just ab0i.e the substratum on no defined stream-channel across the which the sand was resting. The sub- MATUMBULU DAM Approximate outline 1960

DEPTH BELOW F51 Iml m MAIN TRIBUTARY Y -X LONG PROFILE HEIGHT ABOVE FSL (rnl m SEC:ION WASHED AWAY 1161 . . . EMBANKMENT TOP LEVEL BLOCKS - _ -D WlTH BRlDGE Tt-- \NET CHAMBER

Sediment survey November 1974

Flg 76 hlatumbulu dam Reser~olr contour maps. 1960, 197 1 and 1974. stratum was of the same type of cement- main road the SW channel splits into ed sediments as found below other t\\o branches (Fig 73). At the construc- stream beds in the area. Further up- tion of the Matumbulu dam in 1960 stream all \\ells nith a depth of less than attempts were made to divert all the wa- 1 m were drv. ter through the N branch, blocking the Just before the passage under the S branch b~ a concrete wall. However, MATUMBULU DAM the water soon found its way past the Accumulation of sediment 1960 - 1974 wall and in 1970 approximately half of Selected cross sections the water flowed through the N branch A " fee. and half through the S. The culverts, placed where the road crosses the S branch, are not dimensioned for the amounts of sediment now being transported that way. Where the rvater meets the roadbank and the underdimen- sioned culverts, large amounts of sediment are being deposited (Fig 75). In 1974 the sedimentation upstream of the road-bank amounted to 1.2 m. The corresponding accumulation downstream of the road-bank amounted to 0.6 m.The total period of accumulation was 14 years. Thus, 9 and 5 cm respectively were added to the height of the riverbed annually. The most rapid sedimentation occurred in the years 1970-74, during which period the culverts choked completely. It is not known whether the culverts have been choked and successively cleared in earlier years.

Sediment surveys The results of the sediment surveys are shown on the reservoir maps (Fig 76), long profiles and cross profiles (Figs 77-78), area and capacity curves (Figs 79-80). In 1968 Matumbulu dam and catchment were selected as reference units of the DUSER project. In late October the H H' J --.pSEDIMENTATION '1 following year a preliminary sediment +S OEC IBII-NOY li71 0- 0 survey was undertaken. As the 1968169 11 m 3CT 1Pbi-NO< '37' L 5 7 - wet season experienced very little rain I I' the reservoir was at a remarkably low 4- LONG OROFlLE NTEQSELilOY -'SL .... ~~ LEVELLING Di'i L&LY!ir stage (1 1 ft) by October, enabling considerable parts of the reservoir floor to be levelled. A map of the bottom topography was constructed. Unfortu- nately it larer turned out to include Fzg 77 Matumbulu dam. Accumulation of sedi- some errors and it was apparent that ment. 1960-1974. Selected cros sections. only the long profile could be used for MATUMBULU DAM

Long profile V-X 1960 -1974 m .feet

WATER LEVEL l0 l1 71

SEDIMENTATION

0NOV 1971 -NOV. 1974 OCT 1960-OCT 1971

...- INCOMPLETE LONG PROFILE. OCT. l969 SEDIMENT SURVEY

m feet Average depth of cross sections B 275

JERAGE SFDIMENTATION 127 NOV 1971 - NOV 1974 0OCT 1960- OCT 1971 300 200 100 0 m 900 800 700 600 500 LOO l I 1 I l I 0 feet 2500 2000 1500 l000 500 - - 1960 BOTTOM OF RESERVOIR SURVEYED POINT A-J CROSS SECTIONS

Fzg 78 Matumbulu dam. Accun~ulationof sediment. 1960-1074. Long profiles and average depth of cross sections.

Fig 79 Maturnbulu darn. Capacity curves for 1960. 1971 and 1974. By l974 the reservoir had lost 3 1.4 g of its original capacity. The dotted curves a. and b. are early estimations of the reservoir capacity made b! WD & ID. MATUMBULU DAM, Reservoir area curves 1960 - 1974

Fig 80 Matumbulu dam. Area curves 1960. 1971 and 1974. The dotted 1962 curve is from secondary WD & ID data. The 1960 curve is from modified primary data. comparisons with results from later the reservoir as well as for the amount surveys. of water available. A new survey of the reservoir was In 1974 a complete reconstruction of planned for October 1970. However, the 1960 bottom topography of the re- 1970 being a relatively wet year posed servoir was undertaken. The whole new problems. This time there were storage basin above the actual water considerable amounts of water left in level and the surroundings of the re- the reservoir (15-16 ft) at the end of the servoir site were surveyed. The position dry season, restricting a detailed survey of the embankment was carefully de- to the delta area only. Some experi- termined by triangulation. The result ments with soundings from a small rub- was checked against the 1960 air photos. ber boat were initiated (Christiansson After this reconstruction a detailed 1970 pp. 8-10). sediment survey rvas undertaken includ- In November 197 1 the whole storage ing sounding, levelling and sediment basir, was surveyed. The sediment sampling. The survey indicated that of accumulation was recorded b) means of an original volume of 333,000 m3 ap- levelling and sounding. proximately 228,500 m3 remained During the course of the survey (Table 22). In the period 1962-1974 (12 errors in the original construction dra~v- seasons), i.e., from completion of the re- ings became apparent and later it was construction of the embankment in late found that also the oripnal reservoir 1962 to the reservoir surve!? in Novem- capacity figures were incorrect. ber 1974, 104,500 m3 of sediment All assessments made up to then (in- accunlulated in the reservoir. Each year cluding those of the feasibility of the an average of 8.708 m3 were transport- reservoir as a basis for an irrigation pro- ed from the catchment to the reservoir. ject) were based on data drawn from the That implies a rninimunl annual sedi- original drawings. Thus they gave in- ment yield during the period of 581 correct figures for the rate of sedimen- m3/km2.The total loss of capacity was tation and hence for the expected life of 3 1.4 % making an average loss of 2.6 % per year. During the later part of the the delta, stream channels, previously 1 period (1971-74) the rate of sedimenta- or 2 m deep, are being completely filled tion was only 2.0 % per year as com- with sandy sediments. The sand is now pared with 2.8 % per year in the period spreading into cultivated fields adjoin- 1962-7 1. The average sediment yield ing the stream and bloclung culverts per km2 decreased from 626 m3 in the under the road bank (Fig 75). early period to 445 m3 in the later peri- During 1974 sand was taken from od. The expected total life of the re- several places in Matumbulu river and servoir is 3545 years. its tributaries upstream of Matumbulu It is, however, doubtful whether the dam for building purposes. This makes rate of erosion in the catchment actually future measurements of the rate of decreased during the last three years of sedimentation imprecise as the natural recording. There is a possibility that no river bottom along which material is particularly intense rainstorm occurred transported has been disturbed. during the period. The decrease in the rate of sedimentation may also be due to the coloniza- Morphology of sediments tion of shore vegetation (Polygonurn sene- galense, Echinochloa haploclada and Cype- The decreasing capacity due to sedi- rus digztatus) on the delta (Backeus 1976 mentation and the effects of the advanc- p. 120) (Figs 8 1-82). Other reasons may ing delta on the water surface area are be sediment spilling out of the channel illustrated by the reservoir maps (Fig and temporary accumulation behind 76), the long and cross profiles (Figs barriers (see below). It is also possible 77-78), the capacity and area curves that there actually was a decrease in the (Figs 79-80) and by the oblique aerial rate of erosion in the catchment as all photograph (Fig 8 1). the inhabitants were moved from the According to the long profile, sedi- area to Matumbulu and Mkonze ujamaa ment accumulation seems to be some- villages in 197 1-72. This must have im- what irregular. This is explained by the plied a marked change in the pressure profile showing a section cut right on the land. It should, however, be through the reservoir not avoiding pro- noted that cultivation and grazing were truding topographical features. still going on but probably at a reduced The best view of the distribution of intensity. sediment in the reservoir is gven by the Eroded material is trapped upstream profile showing the average depth of of the reservoir to a much greater extent the cross sections (Fig 77). than in lCIsalatu and Imag catchments. The original channel gradient of At Msalatu and Imagi, where the whole 0,009 mlm at the reservoir site has been catchments consist of inselberg and re- markedly changed by the sedimenta- latively steep pediments only, practically tion. The gradient of the channel above all eroded material is transported di- the FSL has been reduced to 0,0025 rectly to the reservoir. m/m. This very gentle slope is inter- In h3atumbulu catchment a consid- rupted near the FSL by the steeper delta erable amount of coarse material is de- front ~itha slope of 0.02 mlm. In the posited on alluvial fans and in the central part of the basin the gradient has stream channels, as indicated for in- remained approximately the same as be- stance bj- the gradual rising of the beds fore the construction of the dam but the of the feeding streams above the re- former channel bottom has been raised servoir FSL. Several hundred m above bet\veen 1 and 2 m. Fig81 \latu~nbuludelta area on 15 hlarch. 1970. The approxi~natepositions of the long profile and cross profiles F-I haye been marked for comparison n.ith reservoir maps and profiles (Figs 76-78). The \vater surface of the reservoir is at hill supply level. One week's rain during which the rain gzauge at the darn recorded a total of 94 mm adcted large amounts of sediment to the delta. the fl-ont of which extended to a position marked by the dashed line. By 1974 the delta extentied another 60 m out into the reservoir (dotted line). (Photo: X. Rapp. hlarch 1970.)

The long profiles from the surveys in effect of shore r egetation Before 197 1 1969, 197 1 and 1974 shorv a close corre- the delta as practicall, free of ~egeta- lation. The extending delta front is well tiorl In 197 1 a ueed, Pol\gonum senrga- illustrated. There is a tendency of an ~P~ZTP,started invading the lol\er parts of increase in the accumulation of material the area (Fig 82) and bs the end of 1974 on the delta above the FSI, in the later the \\hole area betneen profiles E-H part of the period. It should be noted rjas co~eredbp a clense grou th of leg- that the extension of the 1969 profile is etation uncertain. Fig 81 shoks the delta died on 15 The rapid development of the delta hfdrch 1970 The approxinldte positions and increase in upstream sedi~nentation of the long profile dncl cross profiles F-I in 197 1-74 is explained by the ti-appi~lg h,i\ e been narked for cornpailsol1 ~ith Fig 82 Matunihulu delta, 1972. Shore vegetation is invading the earlier vegetationless surface. Note sediment trapping effect in foreground. By 1974 the density of shore vegetation had increased considerably. (Photo: C. C:hristiansson, Dec. 19'72.)

reservoir maps and profiles. The water of 94 mm had added large amounts of surface of the reservoir is at full supply sediment to the delta. The front of the level. One week's rain during which the delta had extended to a position marked rain gauge at the dam recorded a total on the photo by the dotted line. The

C LAY Sl LT SAND GRAVEL

Grain size, mm Fig 83 Grain size composition of sediment samples from hfaturnhulu reservoil-. Of the samples, no 1 represents the upstream part of the delta (abo~eFSL.). Nij~2 and 3 represent dox\.ntream delta accu~nulations.No 4 was collected in the central part of the reservoir and no .iis from the area near the embankment. deposition of sand in the reservoir re- and the storage basin. At the time of sulted in cut-off pools at the mouths of sampling the whole delta was dry. Sam- small tributaries from the E and block- ples from parts of the reservoir still con- ing of road culverts on the opposite side. taining water were scooped up from the In 197 1 exceptionally little rain was rubber boat. The grain size distribution recorded at the dam site (l80 mm). is shown in Fig 83. However, more rain must have fallen The samples represent the upper 15 over the catchment as the reservoir fill- cm of the profile at the respective sam- ed in early February and remained at pling spots. The distribution of grain FSL until mid-March. In the 1968169 sizes in a deeper profile is illustrated by season, when 296.2 mm were recorded, a 90 cm long core collected from the the reservoir did not even fill up. The delta in October 1970. (Rapp, Mur- same refers to the 1964165 season when ray-Rust et al. 1972 p. 287). It contains also some 300 mm were recorded. well stratified sediments in stormflow varves, grading from fine gravel to sand and silt and dark clay at the top. The silt Grain size composition and clay layers reflect the deposition of suspended load after the stormflow. In In November 1974 a number of repre- round figures the percentage of differ- sentative sediment samples were col- ent grain sizes in this core is: gravel 5, lected from different parts of the delta sand 40, silt 40 and clay 15 %. 12. The methods-A critical review

Representativity of the study unit when it comes to questions concerning land use and population. The water area divide sometimes acts as a "land use di- Central Ugogo is representative of vide" as well, as in the southern part of much of the interior drylands of Ikowa catchment, although this is not Tanzania. This refers both to physical the rule. In the same way, population and socio-economic factors (Rigby distribution seldom relates to catchment 1969b; Patton 1971; Rapp, Berry & boundaries. Thus the relations between, Temple 1972; TMWD 1974; TMA for instance, land use and physical en- 1977). Thus the discussions on erosion, vironmental factors are sometimes dif- sedimentation and population and live- ficult to define. This emphasizes the stock carrying capacity of the study area problem of integration of physical and are of relevance not only for Ugogo but human environmental studies, and that also for the surrounding semi-arid parts is probably one of the reasons why there of the country. These cover altogether are so few truly integrated studies of the more than 60 % of Tanzania (Peberdy erosion problem. 1969). Representative catchment basins for detailed study were selected on the basis of a number of reasons, the most important being the presence of a reservoir in the catchment. Other criteria were Relevance of the methods availability of background data, includ- employed ing maps, good quality air photos, hydrological and geological information Field measurements and accessibility. In the present study of ersoion prob- No long term records of suspended lems the catchment basin was used as sediment transport in the streams feed- the primary study unit. It is a suitable ing the reservoirs exist but by measuring unit as it is easily defined on maps and the amount of sediment accumulated in air photos. Physical parameters such as reservoirs of known age, and correlating runoff, sediment transport and deposi- this with catchment and climatic factors tion are easy to record within well-de- it is possible to determine the effects of fined catchments. Thus computation of individual factors on sediment produc- minimum rates of erosion is possible. tion (Christiansson 1979). Difficulties arise when the catchment The volume of' sediment accumulated boundary is ill defined, as in the N end in a reservoir may be determined by of hlatumbulu catchment where sedi- means of different methods. For the ment is deposited right on the catch- present study a method requiring only ment boundary, or as in Mkonze catch- simple measuring equipment, suitable ment where a section in the NE has in- for rough field conditions was em- terior drainage in dry years but spills ployed. The method is recommended over to the main catchment in wet years. by the US Soil Conservation Service for hlajor problems arise lcith this type of sediment surveys under the current conditions. A sufficient number of Measuring sediment yields is not suf- bottom elevations were determined by ficient, however, for understanding the means of sounding and levelling to re- physical significance of soil erosion. One construct the topography of areas where needs to know where the eroded soil change has occurred as compared with comes from within the landscape and the original bottom map. A new contour what the interactions are between con- map was drawn and the volume between trolling variables such as erosion inten- the consecutive contour intervals gave sity, relief, geology, soils, vegetation, the total capacity of the reservoirs. land use, etc. It should be remembered that the The relative relief of the catchment sediment trapped in the reservoir is only (Table 22) is expressed as a relief ratio a minimum volume of the amount of which is the difference in altitude be- material eroded from the slopes in the tween the spillway elevation and the catchment. Practically all dissolved ma- maximum elevation on the water divide, terial and part of the suspended leave divided by basin length measured in a the reservoir (and the catchment) straight line approximately parallel to through the sluice tower with consump- one major drainage channel. In cases tion water (domestic or irrigation) or where isolated high points on the divide over the spillway in times of overflow. tend to give undue weight to the Another portion is removed from the maximum divide elevation, an average catchment through wind erosion. Such relief was used for computing the ratio activity is particularly common on un- as recommended by Schumm & Hadley vegetated sand fans. (1961 p. 2). Recordings of suspended sediments It has been shown from experiments at river gaugng stations as well as reser- in semi-arid USA that there is a good voir sediment measurements omit all correlation between mean annual sedi- material deposited upstream. Further, ment yield and relief ratio provided the the river gauging method does not in- climate and the underlying rocks are clude bedload accounting for as much as uniform (Schumm & Hadley 1961). It 10-50 % of the total transport (Dunne was suggested that erosion classification et al. 1978 p. 133). This portion is how- of semi-arid lands can be made through ever accounted for in the reservoir a quantitative analysis of the geo- survey method. None of the methods morphic characteristics of a region. allow types and spatial variations of However, in practice, under African catchment erosion to be directly deter- conditions, the land use and the natural mined. vegetation factors out~veighmany other In small steeply sloping catchments, parameters, as shown by the present such as Imagi and Msalatu, without sig- study. Lack of meaningful correlation nificant areas of upstream sedimenta- between mean annual sediment yield tion, the amount of sediment trapped in and geomorphic characteristics was also the reservoir is closely related to erosion noted by Stall and Bartelli (1959) in the catchment. (In Imag where water (quoted by Schumm & Hadley 1961) seldom flows over the spillway it would during studies in Illionois. The reason be easy to check the amount of sus- was thought to be cultivation in the pended and even the dissolved material catchment. as all water that leaves the reservoir, Relative and absolute rates of erosion apart from seepage, goes via the and spatial variations in the rates can be Dodoma water works to the town's sup- estimated from indirect evidence of tree ply.) mounds (Fig 44). This method was used to estimate the from such studies to different localities rate of erosion on the upper pediments or to different scales. in I kowa catchment where tree mounds However for the following discussion are particularly common. The rate was on erosion on cultivated fields versus calculated by measuring the difference bare and grassed areas, I have drawn in level between the residual ground heavily on results obtained from plot surface protected underneath the studies in semi-arid environments in canopy of treeslbushes and the level of different parts of Africa. The data of the surrounding eroded unprotected greatest relevance for the present study areas. The depth of the lost layer of soil are those resulting from 13 years of re- was then related to the age of the tree search on runoff and soil loss at and the local rate of erosion could be Mpwapwa, E Ugogo (Staples 1934, calculated (Rapp, Murray-Rust et al. 1936, 1939; van Rensburg 1948, 1953, 1972 p. 277; Stromquist & Johansson 1954, 1955; Temple 1972). 1978 p. 47; Dunne et al. 1978 p. 133). To obtain a more comprehensive As splash erosion acts on the mounds by picture of changes over whole slopes lowering their surface, the relative these may be periodically resurveyed be- height as calculated above only gives a tween fixed points and the resultant minimum value of the rate of erosion. profiles compared (Menne 1959; King This method was supplemented by & Hadley 1967). In addition, changes of studies of micro-relief features such as ground surface level caused by erosion soil pedestals with or without cappings, and deposition can be measured at grass clumps and root exposures, which points identified by fixed and graduated also can be aged. rods or pins placed vertically in the Also observations along vegetation- ground (Leopold, Emmet & Myrick erosion transects supplement the other 1966; Evans 1967; Schumm 1967; methods. The percentage of grass cover Stromquist & Johansson 1978). It was gives a good indication of grazing intended to include the latter method pressure and amount of splash and among the techniques employed in this sheet wash, and local sediment accumu- study but due to problems with supervi- lations along the transects give a hint on sion that part had to be omitted positions in which sediments are (Rapp, Murray-Rust et al. 1972 p. 260; temporarily deposited within the Murray-Rust 1972 p. 337; Dunne et al. catchment. 1978 p. 133). Among techniques commonly employed in erosion studies are measurement of runoff and soil loss from ero- Satellite image and air photo interpre- sion plots (Gerlach 1967; Leopold & tation Emmet 1967; Schick 1967b). These Remote sensing techniques used in the techniques make it possible to monitor study are interpretation of satellite im- both erosion processes and rates of wa- ages (LANDSAT, blacklwhite enlarged ter and soil loss during different condi- paper copies) and air photo interpreta- tions of land use. Such experimental tion. plots need constant supervision. There- The LANDSAT images proved val- fore the method is difficult to employ uable for a reconnaissance view of the under primitive field conditions or in environment. Dominating features as remote areas. Furthermore, considera- inselbergs, sandy rivers, mbugas, etc., ble caution must be exercised in the di- could easily be interpreted as could also rect extrapolation of results obtained the large scale geomorphic process Fig 84 Satellite view of Ugogo. The dominance of light tone indicating sparse vegetation (or dry grass) and areas suffering from overgrazing and soil erosion is striking in central Ugogo (the cultivation steppe). The same refers to the densely populated and heavily grazed and cultivated areas around Singida (Figs 90 and 107). For comparison, the Kongwa ranch map sen-eas an example of a grazing area not suffering from excessive grazing pressure. The very dark tone W of Bahi swamp represents areas of dense thicket around Itigi (part of Mgunda Mkali). Earlier, similar vegetation covered large parts of E Ugog-the Kongrva area and the tracts het~veenblpwapwa and Kikombo (hfarenga Mkali) (Fig 103). Of that only patches remain today except N of Kong~varanch. The frame indicates area covered by Fig 7. 1 Singida. 2 Bahi swamp. 3 Chenene hills 4 Dodoma toTvn. 5 Dodoma hills. 6 Kikombo. 7 h'fpwapwa. 8 Kongwa ranch. he p'icture is chposed of LANDSAT images 1048-07184, 9 Sept. 1972; 1155-07133 and 1155-07140,25 Dec. 1972.

pattern including erosional and deposi- In the mid 1970s when the field work tional zones. Areas of active erosion of the present studies was being com- were identified by very light grey or pleted, only a limited number of good white tones. It should, however, be quality images over the actual area M-ere noted that more or less identical grey available. Thus problems arose when it tones on the LANDSAT images some- came to determining the significance of times represent totally different fea- different grey tones on images repre- tures (Johansson & Stromquist 1978). senting different seasons of the year. Thus very light tones may indicate With an increasing number of regstra- either a dense cover of dry grass or tions and ~vithsuccessive improvements badly eroded areas without vegetation. in image resolution this technique is, Furthermore, on the image 1155-07 133 however, extremely valuable for future Ikowa reservoir is shown \vhite although monitoring purposes (Rapp & Hellden water should be dark in band 5. The 1979). light tone may be an indication of the Studies of air photos offer opportu- high concentration of sediment in the nities to follow in detail land use and water (Fig 7). land degradation in C'gogo back to the 1940s. Repeated surveys undertaken af- these kinds of sources it is not advisable ter that period over the Dodoma area to make definite conclusions based on means that the Dodoma catchments information given by one or two authors probably possess one of the better air only, but comparisons between data photo coverages in Tanzania. Photos from all available written material sup- covering the entire study area were plemented by interviews will make oth- taken in 1949 and 1960 only, a reason erwise obscure tendencies distinguish- for referring a major part of the discus- able. sion on land use, settlement and population to the 1960 conditions. In the period 196 1-1 974 no air surveys covering the entire study area were undertaken. Reliability of basic data The existing photos (blacklwhite panchromatic), representing different The amount of available background scales, are taken in different parts of the data is directly related to the age of the year and in years with differing rainfall reservoir and the purpose for which it conditions. Thus the photos are not in was built and the thoroughness of the all respects immediately comparable. records that have been kept. Concern- However, for studies of such features as ing Imagi and Msalatu reservoirs, it is gully growth, extension of cleared areas, likely that more background data exist exposures of sub-soil and trends in than from any other reservoir in the population and livestock pressure they country. This is due to the role of these are of great value. reservoirs as the base of the Dodoma Vegetation has been studied from a urban water supply for a period of physiognomic point of view only. No nearly 30 years, starting in 1929 (Chris- attempts have been made to distinguish tiansson 1979). Further, Imagi dam and between species. This is considered to be catchment was in focus for hydrological a sufficiently detailed method for the research for a 10-year period in the purposes of the present study, but of 1930s. These conditions guarantee re- course inventories of single species or latively good accuracy of the data on studies of composition of communities erosion and sedimentation presented would have disclosed trends in the en- for these catchments. vironmental change otherwise not im- Since its construction Ikowa dam has mediately obvious (Dasmann et al. 1973 been of central importance for flood p 95). control. Further, an irrigated area below The perspective beyond the earliest the dam is dependent on the water air photos Fvas created by means of lit- stored in the reservoir. However, al- erature surveys and through interviews ready at an early stage it was evident with local farmers, officials, mission- that large quantities of sediment accu- aries, etc. Central questions in the litera- mulated in the reservoir and the Water ture surveys are: Do the texts mention Department commissioned a number of occurrence of soil erosion? In which sediment surveys and evaluations of the areas? Any indications of rates and ex- irrigation project to find out how much tent? Land use pattern? Natural vegeta- storage remained in the reservoir. Fur- tion coverage? Location of settlements? ther data on the dam and the catchment Population and livestock densities? was collected and presented by the General trends in environmental Dodoma Water Master Plan Team that change? ill 1972 selected Ikowa index catchment. It should be emphasized that ~vith IZatumbulu dam is of younger date and prior to the start of the DUSER project Photo-electric planimetering of the had not attracted much interest from catchment (boundaries from topo- either the Water Department or from graphical maps 1:50,000 and air photo researchers. interpretation) showed a total area of The accuracy of much of the basic 612 km2, a difference of some 4.5 % as data is doubtful. On the 1:50,000 to- compared with earlier estimations (cf. pographical maps many details have above: errors in the topographical been left out. In several cases 34con - maps). tours have been omitted on hilltops and A great number of obscurities have in valley bottoms resulting in errors of been found among the official rainfall more than 50 m. statistics. In some cases it is obvious that The early documentation is often in- there have been difficulties in interpret- complete and existing records are con- ing the information delivered by the tradictory in many respects. Some of the gauge reader, whereas in others, orignal maps and reservoir construc- amounts expressed in inches and mm tion drawings have turned out to be in- have been mixed up, or annual and sea- correct, necessitating a time-consuming sonal rainfall values have not been duly reconstruction work. This refers to both separated. Msalatu and Matumbulu dam. It has Population statistics from developing even been necessary to omit some of the countries are known to be particularly earlier sediment surveys (Ikowa 196 1 unreliable (Bondestam 1973). An and Msalatu 1967) due to inaccuracies example from the study area: A map of in the data, the origns of which have population density from the 1967 not been possible to identify. census, showing 26 enumeration areas E The surface area of Ikowa catchment of Dodoma wrongly classified 8 of them. was stated to be 640 km2 according to Apart from inconsistencies in the WD & ID documents. In the Dodoma background data, some errors may orig- Water Master Plan it is stated to be 634 inate from the simple field methods km2. employed in the studies. 13. Contemporary erosion and sedimentation in Ugogo. Discussion and comparisons with other areas.

Rates of runoff cemented sediments. Thus a considerable portion of the precipitation that Data on runoff and soil loss exist from falls over the upper parts of the catch- erosion plot experiments at Mpwapwa ment finds its way into the top soil layer (Staples 1939; van Rensburg 1955), to feed the roots of bush and grass, from catchment studies at Imagi, Msa- whereas the rain that falls on the pedi- latu and Ikowa and from recon- ment to a higher degree runs off on the naissance surveys in some major river surface. However, the bulk of the rain basins (Ann. Repts. GSD 1938, 1939; evaporates or is transpired by the veg- Fawley 195613; Coster 1959; TMWD etation (Staples 1934 p. 101) (cf. Fig 50). 1974): Parts of the results of the above This is in accordance with results studies have been discussed elsewhere from the Mpwapwa experiments (Stap- (Temple 1972 pp. 206-21 1; Rapp, les 1934) which show that on slopes with Murray-Rust et al. 1972 pp. 264-266; dense thicket 30 % of the rainfall is lost TMWD 1974; Christiansson 1979 pp. by evaporation from interception stor- 134-136). Further comments will be age and another 65 % is lost by transpi- made below. ration leaving only 5 % to feed water In the years 1929-1939 runoff from courses. These effects have been shown Imag catchment ranged from to be most pronounced with light 3.5-17.0 %. The highest percentage rainfall (Stromquist & Johansson 1978 runoff was recorded in the first year p. 48). Thus, when surface runoff when rainfall was much higher than in occurs in well vegetated catchments it an); of the other years covered by the generally amounts to a few percent of study. Before the start of the recording the total precipitation. However, even the area was used for shifting cultivation this limited runoff may in the absence of and grazing. From 1929 it was reserved a grasscover cause sheet wash and gully- for the Dodoma water supply, with re- ing (van Rensburg 1958 p. 190; Chris- strictions on grazing, cultivation and tiansson 1979 p. 116, 136). wood collection. Thus the vegetation In the years 1945-1953 the percent- reverted to bush, thin in the beginning age runoff from Msalatu catchment and denser towards the end of the peri- ranged from 7.3 to 25.0 % (measured od (,4nn. Rep. GSD 1937 p. 24). With from the start of the rainy season until denser vegetation the annual runoff the reservoir overflowed) (Table 10). stabilized between 7.7 and 3.5 % with Two of the years that experienced very the higher figures representing years high rainfall showed a runoff exceeding with higher total precipitation. 17.5 %. The year with the lowest runoff Imag catchment consists to more (7.3 %) was extremely dry. As also evi- than .?O % of inselberg slopes with dent from the data from Imagi there coarse soils which presumably have a exists a relationship between total high infiltration capacity (TMWD 1974. annual rainfall and percentage surface App I). The soils of the pediment in runoff (and amount of soil loss) (Christ- turn consist in man); places of layers of iansson 1979 p. 123). A similar relationship exists between single rainstorm The percentage runoff for the period amount and runoff as shown by Temple December-March, when the reservoir (1972 p. 208). overflowed, was 12.9 and 8.5 % respec- In years with high rainfall the re- tively. These figures can probably be reservoir overflowed in January indicating garded "normal" for a catchment like that a major part of the season's total fell Msalatu with some grazing, regenerat- early in the season and rapidly filled the ing bush and a large proportion of soil moisture storage whereafter con- crusted soils (Staples 1939; Handley siderable surface runoff occurred. The 1952; Fawley 1956b). important role of the vegetation in re- Imagi and Msalatu catchments, situ- ducing surface runoff is demonstrated ated next to each other, show many sim- by a comparison between December ilarities (size, gradients, soils, natural runoff and runoff measured over a vegetation and land use). Nevertheless larger part of the rainy season. For five recorded surface runoff in Msalatu is years out of nine the runoff in De- greater than in Imagi (n.b., the records cember at Msalatu was above 20 5% and cover different years). The main rea- twc? years were above 26 S.Measured sons however are probably to be found over the whole rainy season, percen- in the soil factor (larger areas of tages drop to around 10 %. cemented sediments in Msalatu catch- "It is noticeable that the runoff is much ment and a larger proportion of Imag higher ~cithheavy sho~iersduring the early catchment consisting of inselberg slope part of the season while most of the soil with permeable soils) and in the land use surface is exposed. For example during the factor (some grazing going on during first part of the 194647 season after 453.4 the recording period). rnln of rain had fallen the runoff on the cultivated land (experimental plots, At the construction of Matumbulu re- Mprvapwa) was 18.3 5% and on the grasscov- servoir a minimum runoff of 3.55 % ered plot 4.3 '% ivhile the figures for the p.a. was expected. This was considered season were 9.0 and 1.2 respectively." (Ann. sufficient to fill the reservoir even in dry Rep. VSD 1948 p. 33) "Towards the end of December 1955, years. However, the annual rainfall catastrophic storms occurred over wide areas turned out to be 50 1 mm (10 -year aver- in quick succession, causing an enormous age) instead of 6 10 mm as expected, and runoff. The top soil of an entire hill side at as the effective catchment is smaller hladotva (1kolc.a catchment) 1vas washed than originally anticipated the reservoir away together ~cithall the contour banks leaving only bare sand. Altogether in the neither filled in the dry years 1969 and Central Province some 2000 ha of African 197 l nor in the somewhat wetter 1974 fields were wrecked in this way". (Ann. Rep. (Fig 85). No recordings of the actual Agr. D. 1953, p. 25) percentage runoff Fvere ever made, but In years with lower rainfall the reservoir it seems likely that 3.5 % is a realistic did not fill at all or did not overflow figure taking into consideration the until March, indicating either a less in- gradients, the soils and the land use in tense start of the rainy season or more the catchment, all of which differ from evenly spread and gentler showers of Imagi and Msalatu but have many simi- enough dignity only to recharge soil larities with Ikowa catchment. moisture storage, or with the dominant In larger catchments like Ikowa, rainstorms TvidklF spread late in the sea- proportionally more of the precipitation son n.hen the trap effect of the vegeta- is caught in sand)- river beds, sand fans tion is rvell developed. and mbugas, from where much of it is In the 1950151 and 1951152 seasons later lost by evaporation. A considerable the rainfall Fvas normal for the area. amount is also infiltrated into loose soil 1 1 l l l JAN FEE MARCH APRIL MAY JUNE JULY AUG SEPT GCT NOV DEC Fig 85 hlatumbulu dam. Annual record of rvater levels 1963, 1966-74. The reservoir did not fill in 1963, l969 and 1974. Rapid sinking in 1973 and 1974 dry seasons due to pumping for domestic use of water (cf. Fig 71). Data from WD & ID, Dodoma. layers on pediments and plains. This Dodoma area. The Uganda study was water, even if not used up by the veg- started by EAAFRO in 1958 (Pereira etation, seldom reaches greater depths. 1962 p. 42) in the Atumatak catchment, Staples (1937b) carried out comprehen- an area of 8 km2 on a severely over- sive tests of soil moisture penetration grazed "scrub-covered undulating under different covers. Penetration was basement-complex peneplain" (op. cit. greatest with bare cultivated soil, then p. 43). The catchment is similar to those short grass, tall grass and least with de- studied by the DUSER project near ciduous thicket. Dodoma in geology, soils, gradient, A detailed study of rainfall, runoff vegetation and size. It has higher pre- and soil moisture in semi-arid ranchland cipitation (750 mm average annual in northern Uganda is of interest for rainfall 1958-61). Pratt (1962 p. 75) de- comparisons with conditions in the scribes the results of three years' de- Table 18. Rainfall and runoff in Bubu and Kinyasungwe catchments. The Bubu river catchment covers an area of 11,360 km2. The main river is 300 km long. It rises at an altitude of 2,180 m in the tracts NE of Mt Hanang and discharges into Bahi swamp at 810 m asl. Kinyasungwe is a major tributary of the Wami river. It has a catchment area of 9,900 km2. The length of the river, above the gauging station at Gulwe, is 240 km. It rises at an altitude of 1,400 m and falls to 800 m at Gulwe. It drains the central and E parts of Ugogo and the SW part of the Maasai steppe.

Bubu Annual rainfall Annual runoff (m3 X 106) (mm) (m3 X 106) (%)

196243 6,774 644 19634 9,030 850 196445 5,318 515 196546 7,605 722 196647 6,130 583 1967-68 10,923 1,039 Average 726

Kinyasungwe

196445 5,021 508 196546 6,518 660 196647 6.197 626 196748 10.204 1,033 Average 703

Source: TMIVD 1974 tailed studies of relationship of rainfall corresponds to only 2.65 mm of runoff to runoff at Atumatak and makes the from its entire catchment (less than 2 % following conclusions: of the precipitation). Coster (1959) "Stormflol(-. which constituted the whole of quotes an annual runoff figure of the streamflow was directly proportional to 1.33 % for the Ikowa catchment in the the amount of effective rainfall The runoff first year the reservoir was in operation. averaged 40 per cent of the effPctzzrr rainfall. During the 1972173 season when mea- However, the first rains of the !vet season appear to go to fill the reservoir in the coarse surements were made by TMWD the sand of the river-beds since they yield little runoff showed small variations between streamflow in spite of previous wetting of the single rainstorms. In the upper the soil surface." Majenjeula sub-catchment runoff The rainfall only rarely penetrated ranged from 1.5-7.2 '% of effective deeper than 45 cm, and water was avail- rainfall and in the Kikombo sub-catch- able in that zone for approximately six ment from 2.1-8.3 %. Naturally, the months of the year. percentage runoff calculated from As mentioned above, evapotranspira- effective rainfall is higher than if cal- tion is heavy in the Tanzanian catch- culated from total rainfall. Thus. during ments during the spells of fine weather the above season, effective rainfall in that occur throughout the rainy season. Kikombo sub-catchment was 83.7 of This efficiently reduces the percentage the total. During the 1929-1939 period runoff that reaches the reservoirs. the average effective rainfall in Imag For example, in 1960 the Ikowa re- catchment amounted to 72 7~of the to- servoir received 1.7 mill. m3 \vater which tal (Christiansson 1979 p. 123). Fig 86 Trade caravan at Dodoma 1915. Imagi hill in the background at that time practically devoid of woody vegetation (cf. Fig 50). (From CDA, 1973.)

For data on runoff from the larger bility and poor soils. However, it seems river basins of Bubu and Kinyasungwe, as if some of the hills in central Ugogo see Table 18. were being deforested during the From the above it may be concluded course of the 19th century due among that in the semi-arid East African enbi- other things to the trade caravans' need ronment, runoff from small catchments of firewood. Around the turn of the of a few km2 may reach as much as 40 % century many hills were completely de- of the ejjfectzue rainfall. However, in the void of trees, as illustrated by Imagi hill central Tanzanian catchments values S of Dodoma (Fig 86). Since then the from 20 % and downwards of total vegetation has recovered on some hills, rainfall seem more characteristic. The probably as a result of restrictions on result is highlj dependent on soil t>pe, felling of hilltop vegetation, first im- percentage vegetation cover and rainfall posed by the Germans in the early 1900s characteristics, as shown by Elwell and (Ann. Rep. VSD 1932 p. 26). Stocking (1975). In larger catchments The small caps of very dense vegeta- here much water is absorbed in sand tion on some of the higher inselbergs fans or flat areas of loose and porous are probably remnants of a continuous soils or trapped on mbugas, the rec- vegetation of this type which earlier orded runoff is considerably lower, i.e.. covered most of central and E Ugogo just a few percent of the precipitation. (Stanley 1878 p. 76; Stuhlmann 1894 p. 45). Large areas of this vegetation still col-er the tracts E of Ikowa and N of Kongwa (Figs 7,84). Sources of sediment and On the presently bare rock surfaces of types of erosion the inselbergs few weathering products are left. They are rapidly removed by Erosion on inselbergs and bedrock ~vindand \vater (James 1950 p. 3). Be- hills low the bare zone is a belt of accumu- In the earl, 19th century inselbergs and lated coarse ~veatheringmaterial. In the bedrock hill slopes were never used for Dodoma Water Master Plan it is main- agricultural purposes due to inaccessi- tained that "the potential erosion hazard of this zone is minimal" (TMWD 1974, 55). Such conditions are particularly App. I). It is true that erosion is checked obvious in the area around Dodoma and by the vegetation, the roots of which in the W parts of Ikowa catchment often penetrate to the regolith rock. where they were noted as early as in the Moreover, infiltration conditions are late 19th century. generally favourable (Christiansson "On our road hither we came across a cal- 1979 p. 137). However, the field layer is careous tufa of the colour of a dirty grey poor and litter is absent due to rapid putty in process of formation. It was about a decomposition. This makes erosion im- quarter of a mile square, and the accretions resembled very much as if a mixture of very portant as soon as the vegetation is re- fine sand, lime and alkali had been boiled moved. and poured out, and had hence become In some instances bush fields have congealed." (Stanley & Neame 1961 p. 33; been cleared on the inselbergs in the cf. Stuhlmann 1891 p. 52). study area and removal of vegetation These formations were also described for firewood and charcoal burning is from locations near Dodoma by Vageler common. This has resulted in severe (1912 p. 47). local erosion (Rapp, Murray-Rust et al. These barren areas recover extremely 1972 p. 294). With a more substantial slowly due to the unfavourable moisture removal of vegetation extensive areas of conditions and the difficulties for veg- bedrock will be laid bare and the etation to strike root (Christiansson surrounding pediments will be subject 1979 p. 134). Also ground temperature to intense gully erosion, as is already conditions become extreme and dis- common around many denuded insel- courage colonization by vegetation. bergs (Stromquist 1976 p. 39). There are indications (written docu- On the steeper bare rock surfaces, mentation, air photos and field obser- rock slides and a downward transport of vations) that they cover much larger boulders occur. Thus, on many pedi- areas today than a hundred years ago, ments a great number of large and small illustrating a clear trend in land degra- boulders are found. They are easily dis- dation. tinguishable on b/w panchromatic air Overgrazed upper pediments photos (Christiansson 1979, Fig 14). Their relative position reveal changes Types of water erosion active on the in the intensity of the slope process and pediments are splash, sheet wash, rilling the recent geomorphologcal history of and gullying. Splash and sheet wash, the area (Figs 33, 50, 52). often working in close connection, are particularly severe on the overgrazed upper pediments and in areas of so- Erosion on pediments-splash and called termite mound pattern. sheet wash The genesis of the bare patches in the Most of the pediments in the study area "termite terrain" is interpreted as are covered with a relatively thin layer follows: When abandoned by the ter- of loose sandy soil which often overlies mites, the mounds are eroded by calcium, silica, iron or clay mineral rainsplash (Hesse 1955 p. 24) and a clay crusts, "cemented sediments" (Fawley crust is formed. The crust prevents 195613 p. 72). In many areas ho~vever vegetation from recolonizing the sites, the sandy soil has been completelj re- which results in further splash and wash moved by erosion and the cemented effects and thus a certain addition to the layers are exposed at the surface, par- sediment budget of the catchments. In a ticularly along cattle tracks (Figs 19, 53, 450 km2 area NW of Dodoma the frequency of such bare patches, with an on basic schists, and an average of 14.7 average radius of some 5m, was esti- mmtyear on silty and sandy loams on mated at between 110-250 per km2 lavas. The apparent rates varied with Uohansson & Stromquist 1977 p. 106). slope gradient. Each patch represents an average bare Field measurements on erosion plots surface of 80 m2 which pves a total area in grazing areas in N and S Kenya show of vegetationless "termite terrain" of similarly high rates of erosion (Dunne 0.9-2.0 ha/km2 (Figs 43, 52). 1976). Over periods of 10-20 years soil On uncrusted areas of bare ground was lost at rates of 1-10 mmiyear, on the upper pediments a minimum equivalent to yields in the range rate of erosion of 10 mmlyear has been 1,000-10,000 m3/km2 per year. Erosion suggested, based on measurements of rates of that order of magnitude were tree mounds and datable micro relief also recorded on experimental plots at features. The upper pediments occupy Mpwapwa, central Tanzania, by Staples 20 % of the total study area (Table 28). (1939). (See below.) Of the upper pediment zone only 15 % There are indications that locally an has a relatively dense basal ground annual denudation rate of 1 cm may cover (>40 %) to judge from measure- even be low. On a small plot on a bare ments along transects. Some protection slightly crusted pediment slope (gradi- is also afforded by the bush canopy ent 4.3") near Morogoro (200 km SE of which covers some 50 % of the upper Dodoma) an average erosion rate of 2.5 pediments (based on air photo interpre- mm was recorded during 2 weeks meas- tation and estimations made during re- urements in March-April 1970 (Rapp, connaissance flights). Axelsson et al. 1972 p. 150). Assuming Thus, 67 km2 (10 % of the total study that rate as valid for the whole rainy area) seems to be a reasonable estimate season the total annual rate would be for the area of completely unprotected around 20 mm per year. It should, how- ground (no bush, no grass) with a slope ever, be noted that the average rainfall gradient of 2.5-3" or more. An annual at Morogoro (890 mm/annum) is con- denudation of 10 mm within this zone, siderably higher than in Ugogo and that equivalent to a sediment yield of 10,000 the period of measurement was ex- m3/km2 per year, as indicated by mea- tremely short. surements of tree mounds and root exposures, corresponds to an average Cultivated upper and lower pediments denudation rate over the uhole study area of 1 mm/year. Erosion of similar intensity as on the Stromquist and Johansson (1978 p. overgrazed upper pediments occurs on 47), using the tree mound method, the unprotected cultivated fields on the noted annual losses of soil of 24mm at upper and lower pediment slopes. Sheet Mtera in S Ugogo, possibly on less slop- wash and rilling are obvious. Rills are ing ground than that in Ikowa catch- often obliterated by cultivation, but ment referred to above. Measuring hill- \\.hen the cultivation ceases new rills de- side erosion in S Kenya by means of the velop across the fields (Table 17). tree mound-root exposure method, Experiments designed to obtain Dunne et al. (1978 pp. 135-137) found quantitative data on runoff and erosion an average erosion rate of 3 mmlvear on in semi-arid Tanzania were initiated in cracking clay soils on volcanic and 1933 by Staples (1934) at Rlpwapwa. metamorphic rocks and rates of The experiments were carried on until 5.3-10.5 mmlyear on sandy clay Ioams 1939. Tabl~19. Percentage runoff of total precipitation and soil lost per ha with different land use and plot t1-eatments at the Mpwap\\.a soil erosion experiments 1933-38. (Table modified after Staples 1939.) In 1933 six experimental plots were established on a uniform pediment slope of 3.5" at the foot of the Kiboriani mountain. Two plots with a slope gradient of 4.5" were added in 1934. The soil on the pediment is a red sandy friable loam. The area has an average annual rainfall of 690 mm. Plots 1-5 measured 27.7 m downslope by 1.8 m wide (area 50 m'). Plot 6 had the same width as the others but only half their length. Plots 7 and 8 had the same length as plots 1-5 hut had double the width of all other plots. The treatments of the plots were designed to simulate the major land use variations of the area. Plot 1 was designed to resemble the bare slopes and hard compacted soil common in the more densely populated areas of central Tanzania. Plot 2 represented a variant of the above with surface compaction broken bl- cultivation. Plots 3 and 4 were cropped in rough accordance with local M7agogopractice. Plot 5 was designed to test grass cover effects. Plots '7 and 8 were representative of the bush vegetation covering large areas of dry central Tanzania. Plot 1 Plot 2

Season Rainfall Bare; no Natural Bare; flat Natural mm cultivation revegetation cultivated vegetanon

Runoff Soil lost' Runoff Soil lost Runoff Soil lost Runoff Soil lor 7c m3/ha R m3/ha R m3/ha % m3/ha

Plot 4 Plot 5

Season Rainfall Maize flat Cotton flat Mixed native food Perennial tuftec m m cultivatecl cultivated crops flat culti- grass (Chenchrw vated ciliaris)

Runoff So11 lost Runoff 5011 lost Runoff So11lost Runoff So11 loq 5% m3/ha % m3/ha 7c m3/ha '3 m3/ha

' Con\ersion of original figut-es in tons to ln3 pcr area unit \\.as made by division by 1.3. This figure is the soil lost during the secontl half of the ceason on17 : during the first half, the plot !ca.; ridged on contour. Xpparentl~a mistake. -rile amount of soil lost is not reasonable in comparison x~ithother figures in the table. "This plot is only half the length of Plot 2; hence the difference in the erosion rate. At tile end of the previous dr) season. this plot had been btil-nt off. The amount of dr) grass, hov.ever. tvas only sufticient to give a ver\ light burn. BL~I-II~again before the rains rtarteit and a much ileanel- burn than in 1936-37.

After a starldstill during the war the of 8.5". Later tkvo plots on a 4.5' slope recordings Tvere recomme~lcedin 1946 \<.ere added. The treatments ivere de- and carried on until 1954 (van Rens- signed to simulate the major land use burg 1955). Initially six plots Lvere laid variations of the area. One plot was de- out on a pediment slope with a gradient signed to resemble hard compacted soil Plot 3 Plot 4

; tie-ridged Bare ridges Cotton ridges Mixed native food Bulrush millet flat Sorghum flat ~ntour downhill downhill crops on ridges cultivated cultivated downhill

,ff Soil lost Runoff Soil lost Runoff Soil lost Runoff Soil lost Runoff Soil lost Runoff Soil lost m3!ha % m3/ha % m3/ha % m3!ha '"o m3/ha '?c m3/ha

Plot 64 Plot 7 Plot 8

; flat Cotton; flat Mixed native food Deciduous Deciduous Deciduous thicket ,ated cultivated crops flat culti- thicket thicket lightly browsed ated hy goats

,ff Soil lost Runoff Soil lost Runoff Soil lost Runoff Soil lost Runoff Soil lost Runoff Soil lost m3/ha 7c m3/ha 7c m3/ha 5% m3!ha % m3/ha % m3!ha

on bare slope. Another plot represented shown in Table 20. Clear trends are dis- cultivated but not yet planted land. Two tinguishable, and being the only com- plots Mere cropped in accordance with plete series of recordings of this type Wagogo practice. One plot was designed from semi-arid Tanzania this research to test grass coxer effects and the two forms an invaluable source of data on remaining plots carried a cover of thic- rates of runoff and soil erosion in ket. a common tvpe of vegetation in the Ugogo. However, caution should be area. exercised in extrapolating the results The data summarized in Table 19 obtained from plot studies, as these may cover the first 5 seasons of the experi- not necessarily be relevant for large ments. The results of the last 8 years are slope segments (van Rensburg 1955). Table 20. Percentage runoff of total precipitation and soil lost from identical plots with different land use at the Mpwapwa soil erosion experiments 1946-54 (modified after van Rensburg 1995 and Templt 1972). For data on plot layout see Table 19).

Plots 1 and 4 Plots 2 and 5' Plots 3 and 6 Plot 5l Season Rainfall Sorghum flat cul- Perennial grass Top half sorghum Cultivated with (mm) tivated (Cynodon plecto- bottom half grass narrow grass belts stachyur, Cenchrus across slope cilians)

Runoff Soil loss Runoff Soil loss Runoff Soil loss Runoff Soil loss (%j (m31ha)(tlha) (5%) (m3iha)(tlhaj (5%) (m31ha) (tlha) (%) (m31ha) (tlha)

1946147 780 1947148 530 1948149 650 1949150 580 1950151 670 1951152 860 1952153 410 1953154 520 Average 690 (A~erage)~

I At the beginning of the 195015 1 rainy season the treatment of plot 5 was modified; previously under grass for 4 years it was now cultivated tcith 2 m \$lde grass belts left one-third and two-thirds of the distance down the plot. Including soil \cashed to the bottom of the plot but prevented by the tank lip from entering the tank.

Vegetationless surfaces The most striking tendency shown by compared with the 1 cmlyear lost from the experiments is the high rate of bare overgrazed upper pediments. runoff and the large amounts of soil lost Within the Ikowa and the Dodoma from bare ground as compared with catchments only part of the cultivations vegetated. are situated on slopes as steep as 3.5" One of the long term effects of ero- (the approximate boundary between sion on cultivated land in sloping terrain upper and lower pediments) but one is clearly demonstrated in that in a 20- can assume that most actively cultivated year period a vertical lowering of the fields on the steeper parts of the lower ground of 30 cm was recorded at the pediments experience a denudation upper end of the cultivated plots. Of the exceeding 1 cmlyear, equivalent to a eroded material some was lost from the sediment yield of 10,000 m3/km2. These plots while some accumulated near the conditions refer to some 5 $% of the total lower ends. Thus, only part of the sedi- area of the catchments. ment is included in the figures on soil The recordings of soil loss from the loss recorded in the tables. (It is not various plots in single years indicate known whether the plots were actively sediment yields of the same order of cultivated in the period 193946.) The magnitude as indicated above. The av- considerable lowering of the ground erage soil loss from a bare uncultivated surface indicates that a local denudation surface during three seasons was 86.7 rate on cultivated fields of 1.5 cmlyear is m3/ha, while the loss from a bare flat not uncommon. This figure should be cultivated area, kept ~.eedfree, was 88.5 m3/ha. On a plot planted ~vithsorghum an average annual loss of 56.7 m3/ha was recorded. The loss of water (surface runoff) during the same period averaged 48.3 % of total rainfall on bare uncultivated land and 27 % on bare flat cultivated land. Similar conditions were recorded by Neal (1937) during plot experiments in the USA. While a dry hard-baked veg- etationsless surface (slope 3.6") lost 84 % of the total rainfall (falling during 40 minutes) a surface simulating soil after tilling lost only 39 %. Runoff on the hard-baked surface occurred 4 minutes after the rain started while the roughly tilled surface did not experience runoff until after 26 minutes. Trouse and Baver (Baver, Gardner & Gardner 1972 p. 207) observed that restoration of a good tilth on fields that had experienced compaction increased the infiltration capacity in the top 15 cm by 3 to 7 times. The Mpwapwa experiments showed Fig 87 "Erosion pavement" of laterite nodules. Sheet wash can generally only remove finer that grass is the most efficient cover in particles leaving behind a residue of coarser ma- reducing runoff and soil loss. The terial on the soil surface. Unless transported by other processes (cultivation). these residual runoff from plots with an established coarser particles. which in the study area consist of grass cover averaged during two years rock fragments or laterite nodules, accumulate 1.9 % and no loss of soil Ivas recorded. until a continuous surface layer forms, which pro- tects the ground from further erosion. (Photo: C. During the same period runoff oc- Christiansson, Oct. 197 1.) curred from bare crusted soil 59 times and from a badly developed grass cover 7 times only (Temple 1972 p. 208). Ugogo (Fig 87) (cf. Murray-Rust 1972 p. Thus, grass cover assists infiltration as 332). does also cultivation, but less efficiently. Table 19 shows a decrease in soil loss Cropping areas in the third year on the bare uncultivated plot. lvhile there is a slight increase Even fields planted with grain crops on the bare cultivated plots. Staples experienced relatively high runoff and (1939 p. 117) explains this by the for- soil loss, tvith only small differences be- mation of an "erosion pavement" of tween the selected crops. According to pebbles protecting the soil on the un- the results of the plot experiments, an cultivated plot and an increase in the average annual runoff of some 20 % can erodibility of the soil in the cultivated be expected with that type of land use plots with tillage. Such pebble-strewn and a loss of soil ~'hichmay be below 10 surfaces, also described by Epstein et al. m3/ha in years with lo~,and evenly dis- (1966) are very common under natural tributed rainfall and 6-9 times that conditions ~vithinbare eroded areas in amount in years with high and intense rainfall. Narrow grass strips incorpor- years and then allowed to revegetate, ated in the cultivated plots reduced soil showed a quick decrease in both runoff loss by 53 % as compared to unprotect- and soil loss. The figures should be ed flat cultivation (van Rensburg 1955) compared with those recorded on Plot (Table 20). No. 1 and runoff and soil loss figures Surprisingly, the plots with mixed na- from Msalatu and Imag catchments. tive food crops gave the highest runoff. The considerable soil loss from Plot No. This, however, must have been due to 2 (67.7 m3/ha) in the first )-ear after special rainfall conditions, as interplant- cultivation Ivas abandoned emphasizes ing of a number of crops in each field is the importance of protecting first year a common practice in Ugogo (Rigby fallorvs from erosion until a dense veg- 1969b pp. 27-28) which is considered to etation has been established. gve better soil protection than clear From reconstruction of old plot ex- stands. Still, it is clear that the most periments at Mfumbxve, in E Ulugurus, common food crops grown flat do not Temple and Murray-Rust (1972) report offer sufficient protection for the soil, that a cultivation technique which in- and weeding operations make the situa- cludes burning of crop residues after tion worse from a conservation point of harvest, leaving the soil rvith no protec- view. After harvest, crop residues are tion, as is very common practice in Ugo- sometimes left on the fields. This re- go, resulted in a lo~veringof the ground duces splash and sheet wash as long as by 2.8 cm per year during 4 years. With the residues are not eaten by termites. 10 years of bush fallo\v the denudation However, it is a more common practice rate decreased to 1.2 cmlyear, or a re- to rake the stalks together and burn duction in soil loss of roughly 60 % dur- them and to use the ash as fertilizer. ing bush regeneration. Hokvever, it is ivhich leaves the soil without protection. obvious that bush fallow in this envi- The most efficient method of cultiva- ronment (high rainfall, steep slopes) is tion from a conservation point of view not entirely efficient in preventing soil is, according to Staples, contour ridging, loss. Due to incompleteness of the re- preferably tie ridgng. The ridges effi- cordings the authors state that the fig- ciently obstruct downslope flo~vand ures may not be entirely representative, afford a larger soil surface for moisture but the trend is clear: Grass fallow is infiltration than flat cultivation (Staples more efficient than bush fallow in re- 1934 p. 99; 1936 p. 140; 1939 p. 118; ducing soil loss by slope wash. hlore- Hanluns l974 p. 100). The method Tvas over, the depletion of soil moisture is described by Rounce (1949 p. 14) as "a much sloxcer under grass than under perfect soil' conservation measure for bush. arable land" improving the structure and the fertility of the soil as weeds and field refuse may be incorporated in the ridges. As practiced b>-the local popula- According to Staples (1939 p. 118). tion around the Veterinary Station at runoff from plots with deciduous thic- Mp~vap~vait gave good results and the ket is negligible and the protection af- yields from the ridged fields were forded to the soil is excellent. This. higher than from the flat cultivated, an hon-ever, pre-supposes that there is a observation also made b>- Rounce and dense basal ground COL-erunderneath Thornton (1939) in Sukumaland. the thicket, because other\\,ise erosion Plot No. 2 (Table 19). that was culti- may be considerable, as pointed out b!. vated and kept free of 1,egetation for 3 van Rensburg (1918 p. 190). Light ~blr21. Percentage runoff of total precipitation and soil lost from plots with different land use at nagadougou, Upper Volta. Slope gradient 0.5". (Modified after Roose and Lelong 1976.)

'ason Rainfall Cultivated Grass (burnt) Grass (unburnt) Bare soil average Ru~loff Soil loss Runoff Soil loss Runoff Soil loss Runoff Soil loss (mm) (%) (m3/ha)(tlha) (7~) (1n3/ha) ([/ha) (75) (m3!ha) (tiha) (%) (m3/ha) (tiha)

~voseasons 850 2-320.4-3.3 0.6-8 10 0.1 0.15 2.5 0.01 0.01 40-60 1.3-21.3 2-32

browsing on earlier untouched decidu- grass, particularly if unburnt, provides ous thicket immediately caused an in- the best protection for the soil and crease in runoff from 0.2 to 2.5 %. causes a minimum of water loss through According to the results from the runoff. Bare soil is shown to be particu- Mp~vapwa experiments the common larly vulnerable to losses of both soil and practice in Ugogo of clearing thicket for water. The West African example shows cultivation increases runoff some lower figures throughout as compared 50-100 times, if the bush has not been with those recorded in Ugogo, a condi- grazed previously. An annual loss of soil tion which may be due to the lower of anything from 10 to 100 m3 or more gradient of the slope (<1 %) in the first per ha, depending on rainfall character- case, or less erodible soil. However, the istics, may be experienced with cultiva- general tendency is in accordance with tion of the most common grain crops. the Tanzanian results. The relationship between slope angle and soil loss in African semi-arid envi- Grass ronment has been shown by Hudson Plot No. 5 (Table 19) ~v2splanted ~vith and Jackson (1959) from experiments in grass. In the first three years it ex- Zimbabwe. During 6 seasons the aver- perienced a slight runoff. In the fourth age soil loss from a plot on a gradient of year it was lightly burnt with no increase 3.5" was more than double that lost from in runoff. In the fifth year it lvas heavily a plot on 1.5". The plots were identically burnt and after that treatment runoff treated during the whole length of the increased to 11.6 'Z but still no soil loss experiment. was recorded. This gives a good idea of The effects of differences in soil type, the effects of burning. ~vhichis common vegetative cover, rainfall intensity and in Ugogo to improve early b.et season rainfall amount, on runoff and soil loss grazing. M;ithout an annual burn the were demonstrated by Elwell and Stock- bush will increase in many grazing areas ing (1975) in field studies on cropping and so available grazing ill be reduced and grazing lands in Zimbab~ve. (Glover 1968 p. 134). Data were collected from bare and Roose and Lelong (1976) summarize vegetated plots on two soil types. One, a results from erosion studies in seven dif- sandy clay loam formed from meta- ferent en~ironmentsin IVest Africa. sedirnents of the basement complex, is One of these environments, wooded known for its tendency to form hard grassland with 850 mm annual precipi- surface crusts ~vhen exposed to tation near Ouagadougo in Upper Volta raindrop impact. Due to this it is charac- has some characteristics in common with terized by low rates of infiltration. On Ugogo. The results of those studies are this soil all rainfall contributes to runoff presented in Table 21. As indicated by and soil loss. The other, a sandy soil of the Tanzanian figures, it is evident that granitic origin has high porosit!-. It tends, on raindrop impact, to accumu- Gully erosion late layers of coarse sand on the surface which protect the soil from compaction Gully erosion on pediments and detatchment (cf. Fig 87). The most spectacular form of soil ero- With no vegetation, precipitation sion in the study areas is gullying. Dense quantity was found to be the most impor- networks of gullies are common on most tant factor governing the quantity of upper pediments (Fig 88). In Msalatu runoff and soil loss from the sandy clay catchment, for instance, well over 20 % loam, while precipitation intensity was of the total area suffers from gully ero- considered most important for runoff sion. However, gully growth seems to be and soil loss on the sandy granitic soils. of less importance as a source of sedi- The overall most important variable was ment than splash and sheet wash. This is found to be cover density. It influences in accordance with Hudson's view that both the degree of soil protection and splash and rill erosion are the most im- the soil moisture status and thus also portant forms of erosion on arable runoff and soil loss. This tendency was lands, forming a serious threat to the also demonstrated by Dunne et al. (1978 production of food crops (Hudson p. 136) from studies in Kenya. (Erosion 197 1). rates accelerated drastically as ground Most exposed to splash and sheet cover decreased below 20-30 %.) wash are the inter-gully flats in grazing Staples (1939 p. 119) concluded from areas and unprotected cultivated fields the results of the Mpwapwa studies that on sloping ground which provide the where there is a dense grass cover there bulk of the sediments. Gullies and is no serious problem of erosion under stream banks only provide a fraction of the climatic conditions prevailing in cen- that amount. In the Kisongo catchment tral Tanzania. However, the experi- in N Tanzania it was estimated that less ments did not consider differences in than 20 % (100 m3/km2) of the sedi- rainfall intensity and they referred only ments in the reservoir orignated from to one, but a common, soil type. Still, the gullies (Murray-Rust 1972 p. 342). In length of the recording period is con- the Maralal area in Kenya, Dunne (1973 siderable and it covers both wet and dry p. 11) found that of the annual amount years and different types of land use. of soil lost from the slopes in some stu- Thus, the average figures must be con- died catchments only about 100 tons (67 sidered relatively reliable. They con- m3)/km2 originated from gullies (cf. firm, with support of results from Leopold, Emmett & Myrick 1966 p. studies in other areas, the intensity of 239). land degradation in Ugogo, also indic- In a study of 30 reservoirs and their ated by high reservoir sedimentation watersheds in the loess hills region of rates. The experiments clearly show the the Missouri River Basin, Glymph importance of splash and sheet wash on (1951) found that sheet erosion was by overgrazed and cultivated lands. They far the largest single source of sediment. also demonstrate from which environ- Sheet erosion was the only source of ments the majority of the material de- sediment in nine of the reservoir posited in the water reservoirs origi- lvatersheds. It accounted for 95-98 % of nates. the total sediment in nine other ~vatersheds.In only file of the watersheds did sediment derived from gully and channel erosion exceed 10 per cent of the total (cf. Glymph 1957). Fig 88 Active gully erosion on pediment. Example from the Kondoa area. (Photo: C. Christiansson, Dec. 1971.)

Gullies are often associated with cattle systems of gullies, even if relatively sta- tracks or other depressions or vegeta- ble, are harmful to the productivity of tionless areas on slopes. When cultiva- the land as they act as channels for rapid tions are abandoned, the fields au- transport of water and soil from the tomatically pass into grazing areas. inter-gully areas and thus speed up the Then, initially the herds and flocks eat overall erosion and desiccation of the all herbs and grasses available, laying pediment slopes. large areas bare. Gradually shrubs begn In some of the gullies bedrock is ex- to invade the land. But that takes time, posed indicating thin soils on the upper and meanwhile cattle tracks develop, pediments (Fig 89). In other cases the which canalize running water. Thus rills gullies have a hard surface bottom of are formed which may develop into "cemented sediments". Frequently the gullies (Fig 23). "If the hush does suc- water has cut through an upper, near ceed, eventually, in covering the soil it surface, crust which then stabilizes the commonly conceals gullies and a soil walls and prevents further branching. with reduced fertility status. Often the (For a detailed description of gully for- open secondary bush on seriously erod- mation in open grazing areas see Mur- ed land is a persistent stage-the stage ray-Rust 1972 pp. 33 1-333.) of man-made desert." (Hornby & Skerman (1968 p. 4) points out that Hornbv 1943 p. 46). the secondary bush on the pediments ~ullygrowth seems to be a relatively also performs an important function in slow or irregular process. The main im- holding the soil against erosion. This pression, based on air photo interpreta- effect was also shown by Staples (1939 p. tion, is that no large scale gully exten- 118). However, thicket without a basal sions have occurred in the study areas ground cover is not enough to arrest since 1949. Rapid gully cutting probably erosion as stated by Hornby and Horn- occurs at long time intervals as cata- by (see above) and van Rensburg (1958 strophic events (Ann. Rep. Agr. D 1955 p. 190). p. 25; cf. Bishop 1962 p. 163), but the Gullirs associated i~lithfault lines In parts of the study area very large gullies occur. Such features in these parts of the country are frequently associated with fault lines, as shown by the present author from the Kondoa and Singda areas (Christiansson 1972). A sequence of short, deep gullies cut into sandy colluvium in front of a fault line near the N edge of the Uluguru Mts is described by Rapp, Axelsson et al. (1972 p. 153). The general appearance of the gullies and the isolated occurrence indicate that they are directly associated with the fault and that their incision has been facilitated by the thick deposits of colluvial sandy material accumulated in front of the now buried escarpment. The S extent of the Madeng-Iyumbu fault system ~vhich forms the eastern Fig 89 Example of thin soil cover on upper pedi- escarpment of the E branch of the Great ment. The roots of the bush penetrate dorvn the Rift system has a similar appearance cracks into the weathering bedrock. The soil formation is slow and the badly protected loose ma- through the W part of the study area terial is successively \$.ashet1dolvn the slope. Thus (Msalatu and Matumbulu catchments). the depth of the soil profile remains shallo~\.. This buried scarp, which is not every- (Photo: C. Christiansson. Nov. 1974.) where obvious in the field. can be traced to its full extent on air photos. The only "Active gully erosion continues to take place locations ~vherethe fault features are under cover of indigenous deciduous thicket clearly visible in the field are where that has been given conlplete protectiorl for gullies or stream channels cross the many years, particularly along foothill areas faults and suddenly drop 1-5 m and on steep fan slopes. Although the roots of this vegetation type must inevitably afford become both wider and deeper. considerable protectio~i,erosion continues at The conditions described above, in a very marked rate. The deciduous thicket combination ~viththe way in which loose does not provide satisfactory arrest against deposits appear, indicate at least three the ravages of heavy do\+-npours,especiallv different phases in the late development during the earl!. parr of the rainy season \\-henthere is no foliage to break the force of of the present landscape around Do- the rain." doma: Similar effects 1ve1-e observed h>- the 1. ,4n earlier denudation surface has present author (Christiansson 1'379 p. been broken up bv ~videspread 137) and by Stromcluist and Johansson tectonic activity. (1978 p. 48) ~vhopoint out that high intensity rainstorms give almost the 2. Rapid deposition of large amounts of same rainfall underneath trees as in the colluvium has taken place along the open space, indicating their minor ero- fault dislocations (Milne 1936 p. 9) sive protection. Raindrops either break and soils have developed on the col- through the canopy or are rejuvenated. luvial co\ er. Fig 90 Sand fan cleveloped in front of a gullied fault scarp, Singda area. The sterile sandy sediments originating on the overgrazed slope in the background gradually cover the more fertile grazing and cultivation areas in the valley below. On the dry surface of the fan, ~$indfurther sorts the material, blowing away the remaining finer fractions thus adding to the impoverishment of the environment. (Photo: C. Christiansson. Oct. 1970.)

3. Changed climatic conditions (Milne the lower Majenjeula three such sets 1936 p. 10) from relatively wet to have been observed, indicating that at relatively dry, and the impact of man least three cycles of depositional activity has initiated a new phase resulting in occurred in the past. gullying of the residual soils and the colluvial deposits. Redeposition of sediments is now talung $ace, in part rapidly as accelerated erosion Rates. of catchment-. denuda- (~hristiansson1972 p. 320; cf. Bish- tlOn and sedimentation op 1962 p. 162, 164).*inpart slowly as soil creep (Handley 1952 p. 2). Areas Data on the rates of soil denudation, of deposition are non sand fans, sediment yield and expected length of stream channels, mhugas, lakes and reservoir life within the study areas are man-made reservoirs. Thus, the orig- summarized in Table 22. The observa- inal topograph~of Ugogo is gradu- tions are compared with data from Ki- songo catchment W of Arusha in N all1 being masked under recent deposits of great thickness, reaching Tanzania (Murray-Rust 1970, 1972). over 100 m in some mbugas (Ann. The average annual sediment yield as Repts. bVDD 1949 and 1950). computed from reservoir sedimentation over the longest available period of re- In certaln x alle~s sex era1 sets of deposi- cording in Ikowa and the Dodoma re- tion max be d~stinguished Each set servoirs is as follows: Ikoiva (612 km2), consists of bottom laJers ~ithgral el and 1957-1974, l74 m3/km2; Imagi (2.2 coarse sand and loam\ sand and silt\ km2), 1934-197 1, 602 m3/km2; Msalatu cla\ on top (ThIM'D 1974 X p '7) Along (8.5 km2), 1944-19'74, 364 m3/km2; Table 22. Reservoir data. sedimentation and soil denudation rate for five catchments in semi-arid Tanzania. Ca~chment Catchment area Relief ratio Ann. sediment yield Soil denudation rate Location of (km2) (m/kml) (rn3/kmz) (mm/yearz) reservoir

effectire 15.0 25714.4 1962-74 1962-7 1 197 1-74 hfsalatu 8.5 18314.1 1944-74 1944-50 1950-60 1960-7 1 1971-74 Imagl 2.2 13711.5 1934-7 1 1934-50 1950-60 1960-7 1

' Relief is measured from spilli\.ap lex el to highest summit on rvater divide in Ikoria and Kisongo and to average of the trvo highest summits it1 the Dodonia catchments. Concerning transformation of values of sediment yield in m3!km2 into soil denudation rate see Christiansson 1979, p. 129. Changes in capacity due to raised rpillr\-av and excaxations have not heen allorved for in this column. hlatumbul~~reservoir rras completed in 1960. In Februarl 1961 a part of the embankment rras washed arvay. The embankment r\-as repaired in 1962. "I'he rpillrval- of hlsalatu reservoir \\.as raised bp 2 feet in 1950 and bp 1 foot in 1972. The spillrvay of Imagi reservoir was raised h) 4 feet in 1932-33 and by 2 feet i11 1972. Both h1,alatu and Imagi I-eserloirshave been subject to sediment excavation,, Imap in 19.52 (9.000 m') and hrsalatu in 1953 (8,000 m'). ' Imagi reservoir isas conipleted in 1929. Hoivever. the first proper survey of the total volume of the reservoir \\-as not undertaken until 1934. X Data fi-om hlurra?--Rust(1972). ' It1 earl! l974 the cmhanknient gave I\-av.It had not vet been repaired hv the end of 1974.

hlatumbulu (l5.0 km2), 1962-1 974, 58 1 years. The catchment consists of open, m3/km2. semi-arid grazing land with some culti- From these values of annual deposi- vations. Serious soil erosion began to tion the expected total life of the re- develop soon after people settled per- servoirs was calculated (linear extrapo- manently in these surroundings in the lation). For Ikowa it leas found to be 1950s. 3040 years, for Imag and Msalatu A variety of types of soil erosion are 80-90 and 120-130 years respectively active rcithin the area. Overgrazing is and for Matumbulu '3545 years (cf. shown to be the main cause of sheet Stallings 1957 p. 47). erosion rvhich affects 14 % of the The Kisongo dam, built in 1960, has a catchment. Some areas also suffer from very short expected total life, only 25-30 gully erosion rchich is mostly formed ervoir completed Capacity .4nnual loss of capacity Expected total liment surve)) 7c of orig. through sedimentation3 life of reservoir 1') (m3) volume (75) (years)

(theor) 420,0005 389,000 358,0006 298,000 (theor) 284,0005

along stock routes. The accelerated soil In practice the useful life of Kisongo erosion within the catchment is partly reservoir ended after 14 years as in 1974 blamed on the unplanned provision of part of the embankment collapsed. As water in the area. This has caused in- the reservoir by that time was filled to creased stock numbers and increased approximately 60 % by silt it was not pressure on grazing supply, with sub- considered feasible to repair the dam- sequent destruction of ground cover. age. With a catchment area of 9.3 km2 the The rate of sedimentation in Kisongo overall sedimentation yield in the re- and the Dodoma catchments is of the servoir u.as estimated at 446 m3/km2 per same order of magnitude. However, it is year during the period 1960-1969 and not possible to distinguish a common 640 km3/km2 per year during the sub- tendency in the rate of erosion1 valid for sequent t~voyears. As the economic life-time of a re- ' The terms "rate of erosion" and "rate of' servoir is shorter than the total life the sediment production" (rate of sedimenta- Kisongo reservoir may be utilized tion). as commonly used are seldom sy- economically only for a period as short nonymous. The rate of erosion is a measure of the total amount of soil or soil material as 13 years. According to Brown (1950) moved from a fixecl position ~cithinor on "most reservoirs must be supplemented any area per unit time. The rate of sediment or replaced by the time they have lost ~~roduction,on the other hand, is the net 50 % of their capacity" while Stallings amount of solids moved out of a natural !\atershed area per unit of' time. As com- (1957) is of the opinion that "the useful monly used, sediment production excludes life of the average reservoir ends when removal of dissolved matter and remo\ral by 80 % of its capacity is gone". ~,intlerosion (Stallings 1957 p. 47). all the catchments and for the entire all material that is eroded is transported area of each catchment and, in fact, more or less directly to the reservoirs. cannot be expected as the catchments The rates of sedimentation then, in are of different size with different these cases, refer directly to denudation proportions of the main land form units on the pediments only as the uppermost and somewhat different patterns of land parts of the catchments consist of insel- use. berg slopes, in the case of Imagi, with Matumbulu and Mkonze catchments considerable areas taken up by bare have much in common with the sub- rock, which supply very little sediment. catchments of the SW part of Ikowa As in Imagi catchment, the pediment catchment. The various land form units covers only 46 % of the total area, the are represented in approximately the rate of denudation in that zone pre- same proportions. Average slope gra- sumably ranged from 1.1 to 1.8 mm/ dients and slope lengths are the same as year (= 1100-1800 m3/km2), a rate are also the proportions between which seems normal for a fairly well cultivated land and grazing areas. This vegetated pediment with reasonable in- also refers to population density. Thus, filtration capacity. Bare pediment the degree of environmental degrada- surfaces in overgrazed parts of the other tion in these areas should be comparable catchments experience local denudation although there is no reservoir to meas- rates which are 6-9 times as high or ure the rate of sedimentation in the up- around 10 mm per year. per Majenjeula (SW Ikowa) catchment. Schumm and Hadley (1961) showed klsalatu and Imagi catchments. in from studies in the semi-arid parts of turn, also have much in common. USA that sediment yield increases with Sedimentation in the reservoirs is more increasing relief ratio. Such a tendency indicative of catchment erosion than in is also distinguishable among the both Iko~vaand Matumbulu, as Msalatu Tanzanian catchments. Ikowa, Matum- and Imagi reservoirs are located closer bulu, hlsalatu and Kisongo all fit into to the "source of sediment". They can be one and the same pattern, ~vhileImagi looked upon as being representative of stands out as a catchment with low upper pediment segments of larger sediment yield in relation to its relief catchments (Fig 54). Soils and vegeta- ratio, around 40 '% of what could have tion are approximately the same and been expected (Fig 91). For probable their recent land use history shows explanations see Christiansson (1979 p. many similarities. 138). To fit into the American pattern The annual sediment yield per km2 in the sediment yield from Imagi should the two catchments is approximately the have been above 4,000 m3/km2. same if seen over the longest available However, as earlier pointed out, in period of record, 602 m3ikm2 for Imagi practice, under African conditions, land and 364 m3/km2 for Slsalatu. The corre- use and vegetation factors generally sponding rates of catchment denuda- outweigh such factors as relief and slope tion uniformly distributed over the length. catch~nentranged in the case of Imagi In a study of sediment production from 0.30-0.73 rnnl/'!,ear and in 5lsalatu from a sample of Kenyan drainage ba- catchment from 0.46-0.64 nlm/year sins (Dunne 1973 p. 7). steeply sloping (assuming the same bulk density of forested catchments sho~vedvery low sedi~nentsand soils in the catchments). annual yields of sediment (2.7-2.9 The location of both the catchments m3/km2/year).Then a successively in- on the fairly steep pediments means that creasing rate could be noted, indepen- rn3/krnz catchments are approximately the same, 10.0 L750 the expected total lives of the reservoirs differ, 80-90 years for Msalatu and 120-130 years for Imag, depending on 1900 W the relation between the size of the re- -J II 1L25 servoirs and the catchments. W IT Q 950 This effect is particularly obvious if 3 Imag is compared ~ithKisongo. The annual sediment yields per km2 in the E5 1.0 L75 a 380 t~vocatchments are ell comparable, b- 2 W 3 WLL 285 602 and 48 1 m /km respectively, but having a larger catchment area, Kisongo W [L 190 0 lost capacity five times as quickly as 6 5 Imagi due to the higher total inflow of d i 9 5 sediment. W > In comparison with Msalatu and l- = 0.1 L7, 5 Imagi catchments, Matumbulu seems to z have a higher relative rate of erosion D W V) (considering the pediment slopes only-). _I 3 2 6 The total sediment yield of 58 1 m /km 3 zZ is nearly exactly the same as in the other Q t~vobut in Matumbulu catchment with a different topographical setting and different intensity of land use a large 0.01 l amount of sediments is deposited up- 0 0.02 0.OC 0.06 0 08 RELIEF RATIO (h11) stream of the reservoir. This is particu- Fig 91 Relatio~lof mean a~lnualsediment yield to larly obvious in positions where sand relief ratio for 5 catchments in Tanzania (circles) fans have developed and near the re- and 14 small catchments in E \Vyoming. US.1 servoir where the beds of the feeding (dots). The diagram sho~vsthat in catchments of similar sire the sediment )ield per area unit is streams have been raised considerably. similar in the trco groups. Furthermore, the in- Sediments are now blocking the culverts crease of sediment yield rvith increasirlg relief ratio is evident in four '~f ihe Tanzanian catchments under the road and more material is ~vhilethe fifth (Irnagi) stands out as a catchment successively accumulating upstream of rvith lo~vsedirne~lt yield in relatiori to its relief the road bank, threatening the whole ratio. Ik=Ikon-a, Ki=Kisongo (see Table 22). hIs=hIsalatu. llt=lIatumbulu (XIodified from structure (cf. Ann. Rep. Agr. D 1954 p. Schu~nmand Hadle~1961). 32; Stromquist 1976 p. 52). Accumula- tion of sediment in this position may he clently of slope gradients, in catchments one reason w-hy Matumbulu shorved a with the following types of land use: decreasing rate of sedimentation in the Forest + cultivation, cultivation + 197 1-74 period. Another reason is the forestrl, sole culti~ation,sole grazing trapping effect of the vegetation on the and final11 cultivation + grazing. Thus delta. In 1970 there was no vegetation the largest amounts (abole 1000 on the delta, but in the following years a m3,km2)were recorded in areas ~ithvery dense growth of shore vegetation ~nixedcultivation and grazing. This is has developed. the type of land use predominant in The sediment yields of 445-626 Ugogo. 1n3/km2per year as calculated from re- Although the annual sediment yields servoir sedimentation, exclude the sub- per unit area in hlsalatu and Imagi stantial upstream deposition as indi- IKOWA RESERVOIR M;!! WATER LEVELS AND STORAGE CAPACITY 1957 - 74

Fig 92 Ikowa reservoir. Water levels and storage capacity 1957-1974. By 1969 the reservoir had lost 39.2 W of its original capacity due to sedimentation and by 1974 nearly 50 5% had been lost. The reservoir dried out completely in 1961, 1963, 1969, 1970, 1971 and 1974. The decreasing capacity between 1969 and 1974 is an approximation based on one long profile and a few cross profiles measured in November 1974. Data for 1957-1969 compiled by D. H. Murray-Rust, and for 1970-1974 by C. Christiansson.

cated above. Thus the denudation rate Matumbulu catchment. The effects of figures are not directly comparable with intensified cultivation in Msalatu catch- the figures from Imag and Msalatu that ment are described on page 173. represent inselberg and upper pedi- As proposed in the discussion above, ment denudation only, with insignifi- the denudation rate on most grazed or cant upstream losses. Thus one can cultivated pediments is probably well assume that the average rate of erosion above 1 mmlyear over large slope seg- on the slopes in Matumbulu catchment ments and may rise to 10 times that is higher, at least than in Msalatu catch- value on limited unprotected areas as ment. The reason may be more intense cultivated fields and surfaces suffering grazing and a larger percentage of the from overgrazing. pediment slopes under cultivation in Leopold, Emmett and Myrick (1966)

MSALATU RESERVOIR Decreasing capacity of reservoir

Recorded reservoir volume According to field survey - 90 or reconstruction from mops LOO o Estimated '\ E Excovotion of silt

Z > 325 - -+ U \. . S

300 - \ - 65 '.. '.. I

n5-''"""1"'11'1i1iiiii1i1111 194L -46 -48 -50 -52 -51 -56 -58 -M) -62 -64 -66 -68 -70 -72 -74 -76 YEAR Fzg 93 hlsalatu dam. Changes in reservoir capacity, 1944-1974. The decreasing capacity is illustrated by thick black line. Increases in capacit) (addition to spillway height and excavations of silt) are shown by double lines. Dashed line indicates theoretical change of capacity. Dotted line refers to estimations by Fatvley (1956b). report from studies of sediment budget MATUMBULU RESERVOIR Decreasing capacity of reservoir in small catchments in semi-arid areas in New Mexico that sheet erosion basin- - Actual change of c.poc,ty wide estimated from erosion pins gave a -..-Theoret~col change of capoc8ty value of sediment production about 4.5 times larger than could be accounted for in channel aggradation and reservoir sedimentation. The inventories of erosion and sedimentation forms in Ikowa and the Dodoma catchments very clearly show similar tendencies. Part of the eroded material is deposited in the reservoirs but particularly in the larger catchments most of it is deposited on sand fans, in channels, as thin sheets on lower pediments and valley bottoms and in mbuga areas. Wade and Oates (1938 . volume documented from field survey. p. 56) refer to a striking example of or detailed mop reconrfruct~on~ heavy channel deposition in Msalatu x volume without delatled documenfottan o or stream (cf. page 80). Volume ertrmoted extrapolated As pointed out by Langbein and 1960 -62 -6' -66 -68 -70 -72 -74 -76 Schumm (1958), Schumm and Hadley YEAR Fig 94 Matumbulu dam. Changes in capacity, (1961) and Dunne (1976), the measured 1960-1974. The decreasing capacity is illustrated sediment yield per unit area for catch- by thick black line. Dashed line indicates condi- ments usually declines as the size of the tions from the time of the failure of the embankment until the reservoir again started operation in basin increases, because then more ma- 1962. terial is trapped upstream in the catchment. This is evident in the case of creases, extended areas of cultivation Ikowa that has an average catchment and increases in the number of livestock sediment yield/km2 of some 30 'j%of that (cf. Murray-Rust 1972 p. 336). recorded in the smaller catchments in Possible explanations of the above the bodoma area. tendency are: In the case of Msalatu and Imagi, with - Few really intense rainstorms and no relatively long series of recording, cer- period with excessive total rainfall tain. trends in the rates of sedimentation during the 1963-74 period. related to changes in land use can be - An increased proportion of deposi- distinguished (cf. Christiansson 1979 p. tion of suspended load on mbugas 141). Also in Ikowa certain trends can and upstream areas. be distinguished, the most obvious being - Colonization of shore vegetation on a steady decrease in the rate of sedimen- the delta causing more effective tation (Fig 92). (In 1957-60 the annual trapping of sediments in the vegeta- loss of capacity was 6.1 X,in 1960-63 tion above FSL (cf. Matumbulu). (The the annual loss was 3.2 %, in 1963-69 it sediment surveys of 1969 and 1979 was 1.9 % and finally in the period only involved areas below FSL.) 1969-74 it had decreased to 1.7 % - Reduced trap effect of the reservoir annually.) This trend is not a result of a as concerns suspended load, because decrease in pressure on the land and of gradually reduced volume and decreased erosion. On the contrary, depth of water due to the cumulative other evidence indicates population in- sedimentation. portance in this case. Therefore the extrapolations for estimation of total reservoir life are based on the longest available period. The mean value for Ikowa of 2.8 % annual loss of reservoir capacity means a total expected life of 3040 years or complete filling up of the reservoir with sediment in the 1990s. Provided the 1963-74 average annual loss of 1.76 % 'ccill prevail for a considerable period in the future it will take until 2000-2005 before the reservoir is filled with sediment. In the dry climate of Ugogo the rate of soil formation is extremely slow, probably adding less than 0.01 mm/ year. In comparison with the present rate of erosion (

Early observations of soil the occurrence of "hills denuded of soil" (Stanley & Neame 1961 p. 33). erosion in Ugogo Particularly severe erosion seems to "I-he flatter areas are xegetated h) a tuft) have occurred between Mpwapwa and straggling grass l~kehx~ng ha), often ra~sed Chunyo where the caravan route tra- on llttle mounds, ~ithbald places betueen, versed an area badly dissected by gullies. thinl\ strel\d ~~thbits of quart7 and sandstone" (Burton 1860 I p 299) "So far the road had been tolerably good, but now gullies, ravines and broken ground made travelling difficult." (Southon 188 1 p. 548). Erosion by water This difficult passage is also men- Soil erosion is by no means a recent tioned bv other travellers (Junker 1891 phenomenon in Ugogo. As early as in p. 695; Stuhlmann 1891 p. 52; 1894 p. the 1850s Burton (1860 I p. 205, 207, 45). Kannenberg (1900 a pp. 4-5) writes 299) observed the existence of gullies as that water that drains down the hillsides well as signs of splash and sheet erosion destroys the road. In few minutes the in Ugogo. He did not, however, specu- water digs deep furrolvs across the road late over the origin of these features. making it impossible for travellers. Rilling across newly cleared fields is Where the furrows end on the plain marked on one of the illustrations in might!! sand accumulations occur. 'yournal of the discover1 of the source Werther (1898 p. 18) described the of the Nile" (Speke 1863). area bet~veenMp~apwa and Mamboya Stanlev, when travelling between Ki- as one of the least attractive in East kombo and Dodoma in the 1870s noted Africa due to the dense drv bush and

Fi,q 96 ".A tree-less country with frequentl) occurring bar-e soil or sparse tuft! grass". (Photo: C;. Christiansson. Sept. 19'71.) Fig 97 Early documentation of soil erosion in Ugogo. Rilling across newly cleared fields (From Speke 1863).

"the deep ravines cut into the soil tween the occurrence of soil erosion and cover". high population densities in semi-arid Even Herrmann (1892 p. 191) and areas. Ugogo was pointed at as an Meyer (1909 p. 264) mention gully ero- example of an area with far advanced sion in Ugogo, and Meyer also deals erosion. with the question of high sediment con- "Surrounding this 'core-land', marginal centration in the runoff water. lands . . . carry areas of soil erosion, and The earliest scientific analysis of sheet these areas seem to grow in extent in pro- wash phenomena in Ugogo and the role portion to the densit?. of' population, i.e. in of the vegetation as protection against proportion to the extent of artificially created 'culture steppe' " (Gillman 1930 p. 5). erosion was presented by Obst (1923 p. 22). Obst also mentions the existence of "It is easy to trace the progress of cultivation fiom the river banks over the river benches tree mounds. to the foot-hills. Gullies and bare patches There is documented evidence that mark the trail almost e~ery~vhere"(Ann. the pace of erosion was, alread) at an Rep. Agr. D. 1932 p. 26). early stage, greatly accelerated along the central railway line, due to locomotives and generators consuming huge Erosion by wind amounts of timber and the line itself attracting large populations (CDA 1975 Kotes on hind erosion are more com- pp. 28-30). mon than those on \\ater erosion. The! By the time the British had estab- generally refer to so-called "dust der ils" lished themselves in Tanganyika they or intense storms of short duration that soon realized the problem of soil erosion herald the rains (Fonck 1910 p. 280; in the Territory. Gillman belonged to Vageler 191 2 p. 21 ; Wade & Oates 1938 those that clearl) saw the relations be- p. 7). A comparison between the earliest descriptions and those from the late 19th century seems to indicate an in- ' All translations of' quotations fi-om original sources in German are made h\- the present crease in the frequency of sand and dust author. storms. The early reports, from the 1850- sure was moderate, erosion was no prob- 1870 period, only seldom mention the lern. However, with intense grazini- and existence of vegetationless spots or de- some cultivation the grass cover was graded vegetation. On the contrary, damaged. Large vegetationless areas Ugogo is generally described as a well developed and wind erosion was ini- wooded area with some grassy plains tiated. and small cultivated clearings. Later re- "The tountrv (N of Bah~swamp) is not par- ports, from the 1890s and the early t~cularlypleasant as it 1s completely devo~dof 1900s, often describe Ugogo as a tree- trees. At th~stlme of the year (October) a less country with frequent occurrences contlnuouy strong ~~ndr\hirls Into the air of bare soil or sparse and tufty grass (Fig enormous clouds of dust uhich are most lrrltatlng for the eyes." (Schynse 1890 p 46.) 96). The concentration of dust in the air seems to have been high and the vis- Towards the end of the 19th century the ibility often bad. natural vegetation was destroyed practi- Burton describes the conditions in the cally everywhere along the caravan 1850s. He gives a clear impression of routes through central Ugogo. Accounts central Ugogo as an area largely vege- on wind erosion occur frequently. tated by woodland and bush. However, he mentions the occurrence of "dust "Extensive areas are covered with thin devils", which require hot vegetationless steppe grass only, and some are totally de- \-clid of vegetation, so that the strong winds ground to form. He also noted the more call easily transport the sand away." (Stuhl- permanent, strong winds that during rnann 1891 p. 53.) the dry season in areas with weak vegetation moved clouds of dust into the air. "It should also be noted that in Ugogo the winds are generally blowing fiercely, whirl- "During almost the whole year a violent ing large amounts of red dust into the air, east-wind sweeps from the mountains. thus causing eye infections." (Herrmann There are great changes in the tempera- 1892 p. 193.) tures, I\-hilst the weather apparent]! remains the same, and alternate currents of hot and Fonck (l910 p. 278) writes: cold air were observed. In the long summer the climate much resembles that of Sindh; "Sand and dust storms continuously rush there are the same fiery suns playing upon across the plains of Ugogo, and practically the naked surface with a painful dazzle, cool elerything becomes covered ~itha layer of crisp nights, and clouds of dust. The succul- du5t". ent vegetation is shrivelled up and car- bonised by heat, and the crackli~lgcovering These accounts all refer to the period of clayey earth and thin sand, whose parti- around 1890, i.e., before the rinderpest cles are unbound b!, detv or rain, rises in lofty r\-hirling colun~ns like water-spouts pandemic (see page 155). The extinc- when the north u-incl from the Il'ahumba tion of the major part of the livestock Hills (Chenene hills) meets the gusts of Usa- population may have improved the gara, which are soon heatecl to a furnace- conditions for the vegetation thus prev- breath by the glol\.ing surface. These 'p'he- 110' or 'devils' scour the plain uith the rapid- enting some soil erosion. However, the ity of horsemen. and, charged ~vithcoarse 1890s experienced a row of dry years grain and small pebbles, strike ~viththe vio- thus prohibiting optimum development lence of heavy- hail." (Burton 1860 1 p. 297). of grass cover. The annual bush- and grass fires also contributed to the de- The L1' parts of Ugogo around Bahi gradation of vegetation and to the con- swamp have been treeless, hith grass as cent of' dust in the air. dominating vegetation at least since the advent in the area of the U'agogo (Vage- ". . . in a short time all the vegetation withers ler 1912). As long as the grazing pres- and grass fires transform it into carbon dust Fig 98 "Due to the existence of large rlurnbers of cattle in the area the arable soils successively deteriorate. This tnay seem paradoxical but is illustrated by the following: After the harvest the sun desiccates the top layer of the soil. The permanent easterly winds then remove the finest fractions, irlcluding organic tnatter and nutrients. If the fields were left alone after the harvest this process would soon cease. Now, however, the Wagogo let their cattle into the fields to utilize the mtarna stalks as additional feed. The trampling of the cattle loosens the top layer of the soil and successively the wind is given new points to attack. He who has once seen the enorrnous dust clouds that whirl into the air frorn these fields is riot surprised that after a few years even fairly clayey soils get a rnarked sandy charactcr." (Vagcler 1912 p 73). Cattle grazing stalks af-ter hanest, Mat~imbulucatchment (cf. Fig 6). (Photo: C. Christiansson. Ocr 1970.) which is swept across the ground by rhe Environmental degradation fierce SE trade winds." (Stuhlrnann 1894 p. 42.; in the pre-colonial and colonial period "Lztrge areas lack iegetat~onand thus the strong ~~ndshlo~l sand and ash par t~tles Adaptation to the environment2 high into the air ln the fo~n~of huge dust tle~lls" (Stuhlrnann 1894 p 47 ) In his book "Ecolog~ Control and Economic Der elopment in East African The effects of uind erosion on the soils History", Kjekshus (1977) puts for~ard of agricultural lands are described by the theor\ that the "19th centurk East Vageler and Rounce. Africans here on the offenslr e against a hostile ecological system, and until the "'The !rinds at the end of the dry season cause dust storrns when the agricultbral land end of the century the) Ivere the victors is denuded of \zegetation and the soil pulver- in that struggle" (p. 3) and "lived in an ized b!- the harvesting of the annual crops ecological system conquered b!. man" (p. and the grazing of stork." (Rounce 1949 p. 184). M~heprecolonial economies dev- 17; cf. Xnn. Rep. Agr. D. 1932 p. 26.) (Figs

For a discussion and applicntions of tile concepts "adaptation" arid "acljustrnent" in this context, see Hankins (1974 pp. i02-103). eloped in an ecological control situa- further adaptation to the situation was tion" he says (p. 181), leaving no de- to let cattle graze over the harvested tailed definition of the concept. fields to add some nutrients. When making the above statements Thus, during the first half of the 19th Kjekshus seems to consider East Africa century a certain balance seems to have as an environmentally uniform area as existed and degradation of land was he talks about "the (one) ecological con- probably never serious. However, to trol system in East Africa" (p. 49). Such maintain as Kjekshus (1977 p. 47) does, a far-reaching generalization is danger- that a more or less constant surplus and ous as will be shown below by examples material plenty existed in pre-colonial from Ugogo (cf. Mascarenhas 1974 p. East Africa is certainly an exaggeration. 22). Among the literature many reports Evidence from studies of the 19th show that some years were good, others century conditions in Ugogo indicate bad, with famines frequently occurring. that the Wagogo were well adapted to Fonck (1910 p. 278) writes that if the the harsh environment in which they rains are normal, good harvests are a lived (Bujra 1974 p. 3; cf. Rigby 1969a, rule but in other cases, and particularly b and Mkunduge 1978, p. 72). As the in years of drought or locust invasions area was not infested with tsetse, cattle the granaries are emptied before the could be kept. Cattle, apart from being end of the dry season, with famine as a the centre of social and economic life, result. formed an institute of security, a good One may assume that there existed a example of adaptation to irregular certain "ecological balance" in the early rainfall and unreliable grazing re- 19th century but it was certainly not sources, as cattle can be quickly moved man-controlled. However, when the from an area with temporary shortage balance was disrupted it was external of grazing and water to an area with human influence on the Wagogo econ- more favourable conditions (cf. the sys- omy and society that was the primarv tem of cattle trusteeship, page 47). As a reason behind the upset. This disturb- matter of fact, for cattle herders great ance led to utilization of margnal lands mobility is an optimum adaptation to and to overexploitation with land de- environmental conditions. The greater gradation as a result. the agricultural involvement the greater In the case of the Wagogo, hostile the risk of severe effects on the society neighbours prevented free movements of droughts, diseases, etc. and reduced the available grazing areas. Thus the society was organized in The caravan trade with its enormous small mobile units. The oarerall popula- food requirements increased the pres- tion density was low. Settlements, sure on the ground. During the colonial around which shifting agriculture was period a number of restrictions and practiced, Icere b>. necessity located to regulations affected the environment places with permanent water. With negativel). Among those are: adminis- plenty of land available, shifting cultiva- trative measures like forced population tion was a good adaptation to the pre- movements, creation of fixed settle- vailing conditions as the soil nutrient ments, construction of unplanned water status is restored by itself. The plots supplies and reduction of grazing areas. were scattered, mainly on flat land. Cat- That affected negatively the ratio of tle manure Tvas not collected and would livestock numbers to available grazing. have been difficult to transport to the As soon as serious damage is caused to various fields scattered in the hush but a part of the ecological system, control Fig 99 The cultivation steppe E of Dodoma. Oblique aerial view looking S towards Chimwaga hill. In foreground do~vnstreampart of hisalatu stream (note wells in dry river bed). Centrally in picture, partly cultivated alluvial fan. Beyond the fan is the Central Railway line. In the upper left corner is part of the Iyumbu-Madeng fault scarp (cf. Figs 52 and 100). (Photo: D. H. kiurray-Rust, March 1970.)

cannot be said to exist. Still, it is assumed The formation of 'lhe by Kjekshus (1977 p. 4) "that this con- steppe" trol of the ecosystems is now occurring The present worn eroded appear- in East Africa". Hoxvever, one should ance of Ugogo dates back at least to the not use the concept "control of the mid 19th century. Continuous impact by ecosystem" ~vhen,with three years of agriculture and pastoralism and, in rainfall below average in the mid 1970~, particular, excessive exploitation along the Dodoma and Singda regons to- the 19th century trade routes, has re- gether had to import 11 1,000 tons of moved much of the natural vegetation grain as famine relief food. Further- and created a so-called cultivation more, overgrazing is excessive, steep (Figs 96, 99). The heavy food inselberg slopes are being cleared and

and vegetation is removed 3 .' , , , countr)-of unpronounced topography from upper pediments which are losing heavily cultivated and grazed b!- cattle. with soil at a rate of 500-10,000 m3/km\parse areas of hush." (Rounce 1949; cf. annually. Hornhy k Hornhy 1943). The area, as de- filled by Rounce. included the lancls of ele- ven tribes, among them the Il'agogo, in Lake. Western and Central provinces (cf. Phillil;,~1929). and firewood requirements of the hundreds of thousands of travellers necessitated clearing of new areas of woodland and cultivation of marginal soils, until then uncultivated, without any conservation measures. In the dry savanna climate the result could be only one-soil erosion. The original population of Ugogo mainly subsisted on hunting and gathering (Schaegelen 1938). The invading Wagogo, bringing cattle with them found that the more open parts of the area were well suited for pastoralism. The natural grazing areas were gradually extended by means of burning. In the same way areas for cultivation were cleared by means of the axe and fire. (The early extension of cultivated areas must have been very slow as the main agricultural implement was a wooden hoe (Stuhlmann 1910 p. 58), which probably during the first half of the 19th century was exchanged for tools made of iron (Kjekshus 1977 p. 82)). Thus Ugogo in the mid 19th century was still characterised by a mosaic of deciduous thicket, open grassy areas and small cultivated clearings surrounding the scattered settlements. Extensive Fzg 100 Vertical air photo of the area shown in areas Ivere also covered by dry wood- Fig 99. In upper part of the photo is the Do- land of miombo type, "the jungles and doma-klp~capwaroad, in the middle part the Cen- forests of Ugogo" (Burton 1860), par- tral Railrtay line. N is towards the top of the picture. (TANG. 1080 No 66, June 1953.) ticularly in the E and in the N. Although gradually disappearing, large areas of ~voodlandstill remained in the second Strijmquist 1976 p. 39; Johansson & half of the century (Speke 1865 p. 74; Stromquist 1977 p. 106). Primarily it is Southon 1881 p. 550; Stuhlmann 189 1 not the common habit of' burning that p. 55; Herrmann 1892 p. 193). eradicated the miombo woodlands in The recurrent drought years xvith Ugogo. On the contrary, light (early) famine and precarious grazing continu- fires seem to be a prerequisite for the ously changed the settlement pattern continued existence of miombo in its (Kiepert 1896 p. 48; Carnell 1955 p. 26). typical form (Hornby & Hornby 1943 p. Abandoned fields were invaded by sec- 32; Skerman 1968 p. 9). If untouched ondary vegetation, first grass then thorn the miomho is eventually replaced by bush. Thus, the woodlands and the deciduous bushland (Phillips 1929) or a thicket were succeeded by open scrub dense modl land ~vitha thick under- tvhich in combination xvith grazing left storey in which grasses are uncommon large areas bare (Vageler 1912 p. 72; (Pullan 1975 p. 432). Gillman described the Makonde ments (Gillman 1936 p. 361). This effect Plateau as "scattered clearings in this was als'o emshasizedAbv differences in overall blanket of thicket". The thicket the fertility, waterholding capacity and resulted from regrowth of bush during drainage properties of the 'soils. Thus, the period of soil resting, indicated to be Meyer (1909 p. 266) wrote: at least 20 years. A few remnants of the "In the central range of hills the settlements orignal vegetation revealed that taller are located so closely to each other that they trees once had covered the area. In the form villages. In the most fertile areas, Han- early years cultivations in Ugogo were dali, Ndebwe, Mvumi and others towards the concentrated to the light soils along W, the villages line up mile after mile, each with 1000-5000 inhabitants." some river banks and on the lower pediment slopes (Ann. Rep. Agr. D. As a consequence of the expansion of 1932 p. 26). Good land for agricultural the caravan trade the pressure on cen- use may have been experienced as a tral Ugogo became excessive. The wood- scarcity at a relatively early date, particu- lands disappeared completely within larly as the paucity of permanent water large tracts (Stuhlmann 1891 p. 55). excluded huge tracts from human habi- The resulting scarcity of firewood tation. On the upper pediments and on forced both the local population and the the steeper slopes of' the inselbergs the caravans to use grass and cow dung as vegetation was dense enough to make substitute (Herrmann 1892 p. 193; access by people and cattle difficult Stuhlmann 1894 p. 50). With the (Jager 1913 p. 193). Successively, how- dwindling of the woodlands also the wa- ever, thicket areas and upper pediments ter dwindled. The total rainfall did not were cleared. decrease but due to land degradation, Along the main caravan routes more rainfall became less effective than it intensely cultivated areas emerged used to be (Busse 1908b p. 129; Vageler (Speke 1865 p. 76). 1912 p. 28; Obst 1923 p. 192; TMA 1977 p. 9). Mpwapwa, situated below "Strips of vate er less brown jungle through- Mt. Kiboriani, is an area where origi- out Ugogo divide the cultivated districts from each other and occupy about half the nally plenty of water was available all the superficies of the land". (Burton 1860 I p. year around. With a reasonably good 265). average annual rainfall (around 700 "At hTyambwa extensive fields of grain mm) and a central position along the heralded the neighbourhood of the villages". trading routes the area early attracted (Stanley 1872 p. 186). dense settlements (Stuhlmann 1894 p. 43). However, the role of Mpwapwa as Traditionally the settlements were small an important resting place for caravans and isolated. A cluster of average size led to deforestation and desiccation of consisted of some 10 households. With the area. the emerging pressure from neighbouring peoples (cf. page 40). defence "\.llth the stead! Increase in population In aspects became important and larger vil- the area the ~\oodlandssuccesslr el~had to gl'i e nal to clearings Dur~ngm! journe\ lage-like clusters were formed (Speke through Ugogo I learnt of a partlcularlr 1865 p. 76; Casati 1891 I1 p. 284; Dantz drast~cexample In hlpuapna wa here durlng 1902 p. 50). The reason for the forma- m! sta! the women each dal had to collect tion of larger clusters was also of an water for domestlc use In a drl rir er bed at a nalhng dlstance of 1 112 hours Captail1 ecologcal nature: the scarcity of water Charislus. who had got hls lnformatlon from over large stretches of land made im- an old mlsslonar!, stated that 20 \ears possible an even distribution of settle- earl~er \\hen the forest still co'i ered the rihole valley as far as Mp~vapwatownship, need of fuelwood for the railway en- sufficient water rias available close to the gines. "Over the years this was to have a to.iin at any time of the year." (Busse 1908b profoundly adverse effect on the p. 117). ecology of the district, and already in The favourable water conditions as- 1917 there were reports of Greek and sumed by Busse to have prevailed in the Indian firewood contractors denuding late 1880s were, however, more distant the country of trees" (CDA 1975 p. 29). than so. By 1890 thousands and thou- Today the only remaining woodlands of sand of caravans had already rested at any size in Ugogo occur in the Chenene Mpwapwa, filled their food require- hills area (BRALUP 197 1 p. 8). ments and used up much of the The earliest European travellers firewood supplies. noted the abundance of livestock in The area around Mpwapwa as de- Ugogo (Burton 1860 I p. 242; Stanley scribed by Stuhlmann (1891 p. 52, 1894 1872 pp. 168-169, 631-632; Dodgshun p. 4 1) gves an impression of the degra- 1878; Southon 1881 p. 548). In this pe- dation of the land: riod, however, there does not seem to "The area is covered by thin grass. In places have been any serious competition for a few acacias, euphorbias and sycamores grazing between wild and domestic occur. Even the mountain slopes show an animals as there was still plenty of game extremely sparse vegetation, similar to that in the area (Stanley 1872 p. 72). How- on the steppes further W." ever, overgrazing followed by soil erosion by wind and water occurred, par- According to Kannenberg (1900a) the area was once covered with Acacia ticularly in dry years (Burton 1860 I p. forests, a statement confirmed by 299; Stanley 1878). Stanley's description. Localized slow-downs in the land degradation processes occurred in the "Along the base and slopes of the mountains, 1890s. In 1889 a rinderpest epidemic and in its deep valleys. large trees \

Heavy rains often destroyed the banks destroying grazing and soil (Berry & or caused breaks that increased rather Townshend 1972 p. 250). than alleviated the problem (Ann. Rep. After Independence soil erosion was Agr. D. 1955 p. 25; Ndorobo 1973 p. no longer considered a problem to the 40). development of agriculture in Tanza- In addition, water supplies were con- nia. Early party policy was to denounce structed with a view toward opening up conservation measures as being part of new pasture areas, controlling stock bad colonial rule. However, as part of movements and encouragng people to the second five year plan (1969-74) soil relocate to relieve overgrazed areas conservation measures once more ap- (Ann. Rep. Agr. D. 1945 p. 2 1, 68; Ann. peared in Governrnent proposals Rep. WDD 1947 p. 5). There were rapid (Bryceson 197 1 p. 3; Berry & Town- improvements in pasture quality and shend 1972 p. 245). bush regeneration when rotational grazing was introduced into the Msomalo area of Dodoma district in the 1940s The caravan trade-its role in the de- and twothirds of the population were velopment of the soil erosion complex moved to conform with grazing and Earlj regional trade cultivation rules (TNA 3.). But when rotational grazing was introduced with- Up to the beginning of the 19th century out the construction of new tvater- the regonal trade in central Tanganyika sources, thousands of cattle converged was based mainly on iron and salt and to on one spot for water in the dry season, some extent on copper from Katanga. An internal long distance trading keep the water holes in order (cf. Bau- network based on these commodities mann 1894 p. 22). The most detailed had long been well developed when the description of the hongo system, its first Arab and Swahili traders from the causes and effects is given by Burton coast reached central Tanzania. How- (1860 I p. 252). In the early 1890s, when ever, these traders were not primarily the German colonial administration ac- interested in the traditional trading quired control over the trade routes, the goods. It was mainly in order to buy hongo system was abolished. ivory and slaves that in the 1820s they started travelling beyond the coastal hinterland. They then followed routes Caravan routes through Ugogo opened some 20 years earlier by traders The earliest caravans did not pass from Unyamwezi (Benndorf 1918 p. 63; through central Ugogo as the Wagogo Sutton 1966 p. 27; Alpers 1969 p. 35). prevented foreign travellers from Thereby a new network of regional traversing their country. At that time trade was created. the trade routes followed the Gt. Ruaha In this trade Ugogo came to play a river whereafter, near the confluence very important role. It soon rose to the with Kisigo river, they turned towards most prominent food producing area in the NW through the SW corner of central Tanganyika. In Unyamwezi and Ugogo (Fig 103). in the areas NE of L. Tanganyika (Fig 2) From around 1850, after crossing the the traders bought ivory, slaves and iron coastal hinterland, the main routes went hoes. During the transport to the coast across the E mountain areas through the some of the iron hoes and some of the Mamboya valley or through the valley of slaves were exchanged in Ugogo for Mkondoa (Stuhlmann 1892 p. 187; food, locally produced salt and more Werther 1898 p. 18; Dantz 1902 p. 50). ivory (Burton 1860 I p. 252, 308, 31 1; From Chunyo, W of Mpwapwa, one line Cameron 1877 I1 p. 238; Becker 1887 I went N to Kondoa-Irang and three p. 448; Junker 1891 p. 691; Sigl 1892; main lines went W through Ugogo: one Alpers 1969; Kjekshus 1977 p. 98). At in the N via Chilonwa, Ilindi and Bahi; times, the effects of food shortages in one central via Kikombo and Dodoma Ugogo seem to have been relieved and one S of Dodoma Hills via Handali, through trade of surplus produce from Mvumi and Nondwa (Kanyenye). The the surrounding areas (Hore 1892 p. main routes were interlinked through a 249; Kjekshus 1977 p. 36, 39). number of greater and smaller tracks The Wagogo made use of the stra- (cf. Kjekshus 1977 p. 122). tegic position of their area to get some advantages from the continuous flow of goods that they saw passing. They 19th century food resources adopted a kind of toll system that It was uneconomic for the caravans to already existed in the coastal areas. carry food supplies over anything but The system was called "hongo" or short stretches. Thus the position of "honga" from the Kiswahili verb "Ku- Ugogo was strategic, half~vaybetween honga": to pay "dash", tribute or pas- the coast and Unyanyembe and bound- sage money. Cameron (1877), Southon ed both in the E and in the W by virtu- (1881 p. 549) and Hore (1892 p. 51) ally empty wilderness (the Marenga interpreted the system primarily as a Mkali and Mgunda Mkali). It Tvas "an kind of water tax which should com- indispensable store for the commercial pensate the Wagogo for their job to trade caravans" (Junker 1891 p. 694) CARAVAN ROUTES THROUGH UGOGO

---- Early routes ( before 1870 1 - Main routes (1870 - 1910 ) Connecting routes

MAASAI STEPPE

KIBORIAHI HT5

Fig I03 Caravan routes through Ugogo. because it was one of the few areas along But most observers agree that the the caravan routes that had a population strains placed on the land and food re- large enough and sufficient arable soil sources by the caravans were enormous. to produce a surplus for the enormous To produce these huge amounts of caravan traffic. But as many of the con- grain the pressure on the land must temporary reports indicate, a surplus have been intense. Not only the best only existed in years with favourable soils were used but also the poor shallow rainfall. soils higher up the hill slopes, which had obviously not been used for cultivation "The apparently infertile land can, however, before. The fallow periods were fre- feed a fairly large population with bulrush quently reduced. There are no reports millet and water melons, and even the innumerable caravans that pass . . . unless of particular soil conservation measures there is not a famine ciue to lack of rain." from this period (Bujra 1974 p. 4). (Stuhlmann 1894 p. 50; cf. hleyer 1909 p. In May 1871 Stanley travelled through 260) Ugogo on his first journey to the interior of the continent. "Ugogo", Speke writes that the villagers of Ugogo Stanley was told, "is rich with milk and have used much land for cultivations honey, rich in flour and beans and and that the Mragogo in general cultivate almost every eatable thing" (Stanley enough, not only to satisfy their own 1872 p. 163). needs but also the thousands of people Stanley passed just after the harvest- that pass each year ~iththe caravans ing period. Thus all stores were filled (Speke 1865 p. 75; cf. Burton 1860 I p. and there was food to buy at reasonable 242). prices (cf. Southon 188 1 p. 548). However, towards the end of the dry Wagogo (Casati 1891 I1 p. 284; Stuhl- season and during the rainy season until mann 1894; Fonck 1894 p. 292; Kjeks- the first harvest, there seems to have hus 1977 p. 142). To avoid trouble and been food shortages practically every to avoid the risk of losing their only se- year, particularly along the caravan curity, which could be a few surviving routes where the most well-filled stores cows, the local population began to could be emptied in a short time by the move away and settle far from the cara- huge caravans that passed (Stanley 1878 van routes. This made the food supply p. 101; Hermann 1892 p. 197, 201; situation for the caravans precarious Fonck 1910). (Kiepert 1896 pp. 4546; Kannenberg In part the shortages were probably 1900b p. 148; Fuchs & Hauter 1907 p. due to the Wagogo selling some of the 37; Meyer 1909 p. 165,266; Fonck 1910 dry season and planting season sub- p. 218; Benndorf 1918 p. 62). sistence reserves. This occurs even to- The problem is illustrated by Fonck day: who in 1896 marched with a large cara- "In spite the fact that yearly harvests are van (1,630 persons) from Dar es Salaam small in Dodoma district, people still sell part via Mp~ap~vato Ujiji. At that time there of their food crop for cash, because there is had been consecutive famine years since no major cash crop . . . . 1890. Fonck writes: Before the next harvest, therefore, people are short of food and famine sets in." "When the caravan reached Kwa h'yangallo (Ndorobo 1973 p. 33). the shortage of food became acute. Some starving porters tried to steal sheep and During the late 1880s the food situation goats. The thieves were caught in the act and became Morse (Junker 1891 p. 692). were immediately killed by the villagers. All The main reasons for this must have Ugogo suffered from hunger and it became been a successive lowering of the pro- more and more difficult to find food. For those that could pay there were small duction potential including acceleration amounts of flour and vegetables to buy. The of soil erosion caused by deforestation porters tried to survive on grass and roots. - and overexploitation of the agricultural It was not before the caravan turned h' to- land. Probably a certain role was also wards Usandawe, off the main caravan routes, that the situation changed. Here the played by the gradual extinction of stores were filled and the caravan could re- game in large parts of central Ugogo. new its supplies." (Fonck 1910 pp. 161-163) The frequency of attacks by vermin in (cf. Fig 103). this period is not known but it is likely that it was a problem as it is today caus- Other reasons causing famine also ing considerable crop losses (Masca- occurred. During a number of years in renhas 1971 p. 259). the 1880s and 1890s many Wagogo fled The situation was drastically aggra- from their homes in S Ugogo due to vated in the 1890s through drought and aggression from the Wahehe. The re- the 1893-95 locust invasion (Mkunduge fugees crowded in the already densely 1973 p. 67). No animal products were populated areas S of Dodoma hills, re- available due to the rinderpest (Bau- sulting in over-exploitation and land rnann 1894) and crops that did not wilt degradation (Kannenberg 1900a). A \\-ere eaten by locusts. During these permanent famine situation arose, circumstances people were not willing to kvhich was successively aggravated from sell any food. This was viewed by the year to year. After the Wahehe had Europeans only as a sign of commercial been defeated by the Germans in 1898 back~cardness(Gotzen 1899 p. 33), and the majority of the refugees returned to the caravans robbed the food from the their old areas. To avoid the permanent famine along was not uncommon (Burton 1860; these routes most of the travellers now Speke 1865; Wissmann 1889; Bin Mu- followed routes N of Dodoma hills. hammed 1902; Fonck 1910). A general Thus the pressure on the N routes and estimate of the number of caravans adjoining areas became excessive travelling along the central route in the (Meyer 1909 p. 266). early 1890s is some 2000 a year. Towards the turn of the century the Reg.lstration of the caravans started in caravans again followed the S route, the 1890s. The German administration which by then had been improved, "past issued authorizations for caravan lead- the large villages (S of Dodoma hills) ers and traders. According to these where even the largest caravans can ob- documents some 200,000 people passed tain enough food" (Kannenberg 1900a I\/Ipwapwa each year in the period P 7). 1890-1910 (Stuhlmann 1894 p. 90; The most dramatic example of soil Kannenberg 1900a p. 7; Benndorf 1918 erosion in the whole of Tanzania occurs p. 63). Thus in the 1897198 season, 620 around Kondoa (Christiansson 1972). caravans travelling up-country and 880 In this area population expansion and travelling down-country halted at intensification of cultivation combined Mpwapwa. with grazing bp livestock in areas of hilly The majority of all caravans passed topography and erodible soils have during the period May-November, i.e., resulted in erosion of frightening pro- the dry season, when travelling was portions (Figs 25, 28). However, as late easiest and when there was hope of food as at the time of the 1914-18 war the in the stores (Cameron 1877). De- area does not seem to have suffered cember-April, the planting and growing from serious erosion. period, was called "the hunger period". "Subsistence depended almost entirelj- on Due to negotiations with the various the stuff the forage-parties of the companies chiefs and the often precarious water were able to bring in . . . In this extremely and food situation, the passage of rich Kondoa country the troops could obtain Ugogo required varying lengths of time a variety of food in abundance." (Lettou- (Burton 1860; Speke 1865; Stanley Vorbeck 1920 p. 139). 1872; Southon 188 1; Wissmann 1889; The parallels with the development in Schynse 1890; Casati 1891; Junker Ugogo are obvious. 1891; Reichard 1892). The average for six caravans with carefully documented time schedules was 20 days. Food requzrements of the caravans For the passage from Kilimatinde to Statistics concerning the caravan trade Itura, E of Tabora, through uninhabit- and the size of caravans from the period ed land (the Mgunda Mkali) another before 1890 are not available. Meyer 6-8 days were required (Speke 1865 p. (1909 p. 266) estimated the total num- 89; Fonck 1910 p. 277). ber of persons that passed through Thus when the hongo system was still Ugogo along the different routes in the in operation westbound caravans re- golden days of the caravan trade (1860- quired food for at least 27 days in 1885) to between 400,000-500,000 Ugogo. At the end of the 1890s the cor- persons each year. "The number of responding figure was 15 days. people who annually pass this way must Eastbound caravans loaded provisions be reckoned in hundreds of thousands," for the stretch to Kilimatinde in the (Peters 1891). The caravans were often Tabora area (Meyer 1909). From Kili- very large. A size of 1000-3000 persons rnatinde, during the time of the hongo system, it took some 20 days to Some years later when the intensity of Mpwapwa and after the abolishment of the traffic had decreased and the time thk tribute system approximately half spent in Ugogo by each caravan had that time (Kannenberg 1900a; Fonck been reduced by approximately 50 %, 1910 p. 277). the total needs of food for the caravans Using the few figures available it can were 1600 tons per year. be roughly estimated how much food In 1900, Kannenberg (1900a p. 7) the Wagogo had to produce, apart from estimated that in the Mpwapwa area their own needs, to be able to satisfy the alone some 600 tons of grain were con- caravans. sumed annually by the caravans. Of this The daily ration for a porter in Ger- volume only about 1 % was produced man government caravans was 0.640 kg locally. The remaining part was grown of grain, mainly millet or sorghum6 elsewhere in Ugogo or in neighbouring (Becker 1899; Benndorf 1918). The ra- areas. This indicates that the land had tions were probably smaller in the pre- been over-exploited for decades colonial trade caravans (Herrmann (Stanley 1878 p. 101; Stuhlmann 1891 1892 p. 201), but assuming they were p. 52; 1894 p. 4 1) and by then probably the same as in the colonial government gave very bad returns per area unit. caravans, the total annual requirements A description of the area 30 years of grain to be bought in Ugogo by west- earlier gives a hint of more productive bound caravans in the period around conditions: 1875 must have amounted to some "At first, the jungle thinned, openings in the 3,456,000 kg (200,000 X 0.640 X '27). wood appeared, then wide and naked clear- The needs for caravans travelling in the ings, then extensive fields of the hardy other direction should thus amount to holcus, Indian corn and maweri or bajri 2,560,000 kg (200,000 X 0.640 X 20). (bulrush millet) with here and there a square tembe or village." (Stanle~1872 pp. 168- Thus, the total needs for all caravans 169). amounted in this period to more than 6000 tons per year7. "At Mpwapwa food was plentiful and of sufficient variety to suit the most fastidi- ous-and cheap also". (op. cit. p. 164). 640 g of millet gives after preparation an The extensive utilization of the land for average of 2,048 kcal and 80 g of protein. cultivation meant less areas available for 640 g of sorghum gives 1920 kcal and 64 g of protein. Daily rations in famine camps dur- grazing and furthermore there Tvas ing the British time amounted to 615 g of competition of the pasture areas be- grainiday. Today, according to the Tanzania tween local herds and the sizeable herds Food and Nutrition Center the needs of an brought along by some of the caravans adult, relying on one single type of cereal (maize, millet or sorghum) is estimated at (Behr 1891; Herrmann 1893 p. 82; 700 g of flour a day Uonsson 1976 p. 2). Stuhlmann 1894 p. 47). ' 6,000 tons of millet would have been suffi- During the last decade of the 19th cient to feed some 20,000 persons, or an century the caravan traffic decreased. estimated 115 of the total population of The Arab-dominated plantations on the Ugogo at that time, during one whole year. Tanganyika coast declined, and the To produce 6,000 tons of millet an area of at least f2,000-14,000 ha of reasonably fertile trade found new ways as the routes be- soils is required. In cases of lower yields than "normal" due to drought, diseases or low ramfall the aLerage ~ieldof inillet in Cgogo fertility of the soils a larger. area is needed. !\as diound 300 kglha and the \ield of led At present the average yield of millet in East sorghum 750 kg/ha M hite sorghum ~hich Africa is 450 kgiha (Acland 197 1). Rounce is common on the l~ghtsoils of central (1949) reported that in years with adequate Ugogo, y~eldsecen less than m~llet tween Mombasa and L. Victoria in- played an important role for the food creased in significance. Nevertheless, supply. It was not until the construction due to the contemporary rise of the of the railway around 1910 that Ugogo German plantation economy the need definitely lost its role as an important of labour still existed. The old slave producer of agricultural products to be- caravans through Ugogo were replaced come an area that imported large by an enormous traffic to and from the quantities of food in the form of famine coast by employed workers (Bull 1973; relief (Mascarenhas 1967; Brooke 1967; Mascarenhas 1967). Thus Ugogo still Rigby 1969a; Patton 197 1). 15. The post-Independence situation

Availability of land and cesses of land degradation. Thus, before population carrying one can suggest efficient remedial measures against soil erosion it is capacity necessary to assess the carrying capacity Little quantitative attention has been under present and realistic, alternative paid to man-land relationships in types of land use. For this purpose, Tanzania. Imbalance in these relation- Allan (1965 p. 9) proposed the concept ships is a crucial factor in the initiation of "critical density of population" de- and acceleration of soil erosion as fined as "the maximum population indicated from, for instance, the density the system is capable of support- Kondoa area (TMA 1977 pp. 30-3 1). ing permanently in that environment The following discussion is intended to u~zthout damage to the lund". However, point at the present and potential future doubts have been expressed about the pressure on the land, the results of use of ecologcally based techniques of which may aggravate the ongoing pro- assessing critical population density (Jacks 1956, quoted by Wapakala 1974), but in the absence of a better (and rela- Table 23. Dodoma district, Ikot\a and Do- tively simple) technique the "ecological" doma catchments. Total population and parameters used by Moore (1971) will population density. be employed in the estimates below. Population density reflects the size of Dodoma diitrict (16.576 km2)' the population in a unit area. In rural 2 year inh. inh./km areas it directly measures the number of 1911 125,000 5.52 1921 90-100,000 people who attempt to gain a livelihood 1934 130,000 7.6 from the land. Somewhat less directly, 1948 218,845 12.8 1957 239,641 14.0 population density suggests such pa- 1967 320,999 19.4 rameters as the degree of land utilization, the extent to which lands unsuit- Iknu'a catchmpnt (612 km2) \ear ~nh. inh/kmz home- "house- able for cultivation are cleared and the steads holds" numbers and densities of cattle. All these factors add up to the potential degree of erosion that may be expected (Stocking & Elwell 1973 p. 101). Mknnze cJnum~ratzonarea (67 kmz) The availability of land in central Cgogo ranges from 1-5 ha/person. In 196'7 each individual in M\ u111i division disposed over an average of 2.4 ha (41.2 inh/km2). In Dodoma Makulu division the corresponding figure was 2.8 ha ' Before 1957, 17.150 kmz (35.3 inh/km2),in Chilonwa division 2.9 Refers to Ugogo, 22,410 kmz ha (34.4 inh/km2) and in Bahi division 12 1 ri-ithin the catchment boundaries 2 113 of catchment uninhabited. Population den- 4.9 ha (20.3 inh/km ). (For location of sity in remaining 2i3 = 52.7 inh.,'kmz. the divisions see Fig 104). However, these overall population less open fields around the homesteads density figures have little meaning as (Figs 52,72-73). In a given year some of only part of the land is available for ag- the fields are sown and cropped, others riculture. In practice 10-50 % is not are in fallow. On dry season air photos suited for cultivation or for grazing due grazed fallows are very difficult to sepa- to steepness of slopes, density of vegeta- rate from recently harvested fields. tion, thin soil cover, erodible soils, lia- Thus fairly large areas may have been bility of flooding, etc. (Berry 1971b p. l; interpreted as cultivated fields although TMWD 1974, I11 p. 2 and Appendix I). they were not being actively cultivated. An analysis of the area1 coverage of To get an idea of the annually the main land form units and associated cropped area within Ikowa catchment, soil types in the study area based 011 the one has to turn to district averages and 1960 air photos, shows that 33 % of the apply them to the study area. (It has catchments are directly unsuitable for earlier been shown that Ikowa catch- agriculture. Bedrock hills or ridges ment is representative of the district in cover 12 %, upper pediments with slope many respects.) gradients from 3-10" cover 20 % and An approximate maximum annually riverbeds roads, railways, built up areas, cropped area can be assessed through etc. approximately 1 %. Another 6 5% subtraction of the agriculturally un- are taken up by mbugas which with pre- suited areas from the total area and then sent methods are difficult to cultivate. leaving a relevant amount of land to They, however, serve as important dry allow for fallows. In the shifting cultiva- season grazing areas (Table 8). tion system under optimum conditions the ratio between cultivated and fallow fields should be between 1 :7-1:20. Thus Ikowa catchment Moore (1971 p. 42) estimated that in In the DUSER project air photo in- Dodoma district the area which could be terpretation was used to identify expected to carry a crop in any one year "cultivated" land and grazing areas was around 10 % of the area of the dis- (Rapp, Murray-Rust et al; 1972-p. 269, trict, less urban and "waste land" and 289). forest reserves. Of the Ikoua catchment area it was Prior to 197 1 Ikowa catchment had a estimated that, in 1960, "cultivations" population density above the district av- covered a total of 30.3 % of the catch- erage and is better supplied with water ment area. It was, however, difficult to than many other parts of the district, define the exact area under cultivation implying that the population Jvas fairly due to the very complex pattern of evenly spread over the catchment (Fig cleared and abandoned fields. Thus 104; Table 24). Further, some semi- "cultivations" in this case does not only permanent cultivation, based on the use include the area actually carrying a crop of cattle manure, was practiced in the in that particular year but also other vicinicy of all homesteads. Thus in the sparsely vegetated areas, which often 1960s the annually cropped area prob- mean intensively grazed, fallow fields. ably amounted to around 15 5% of all the Only limited areas in Ugogo are per- land in the catchment or some 9200 ha. manently cultivated, the dominant type The need of calories averages 2100 of agriculture being a combination of kcallday for each member of the com- rotational bush (or grass) fallow and munity.' To keep that standard, 288 kg more genuine shifting cultivation. The system leaves a mosaic of more or ' Protein. etc. is not considered in the following. 167 IKOWA AND DODOMA CATCHMENTS. POPULATION DENSITY 1967

inh./ kmz division boundary --- 0-30 - catchment baundory - boundary of reference area

Fzg 104 Iko~aand Docloma catchments Population denslty 1967 of millet or 306 kg of sorghum has to be the prevailing cultivation system, the produced annually per person provided upper limit of the population carrying the whole need is taken from one staple capacity was reached. An increase in crop. With an average )ield of 500 kglha population at that stage would reduce a minimum cropping area of 0.58 the cropping area per person below the halperson is required in areas where 0.58 ha limit. (The actual increase in millet is the staple crop. Maize has the population in the years 1960-1967 was same calorie content as millet but gen- 39.4 %.) erally yields better. Thus with maize (or This strongly points at the pressure sorghum) as staple crops only 0.40-0.45 on the land and the apparent risk of halperson is needed. It should, how- extension of the cultivations to the most ever, be kept in mind that millet gives marginal areas. normal yields at an average annual It is doubtful if the above conditions rainfall of 500 mm. hlaize needs were considered when, in the "Eco- 700-750 mm, which is uncommon in the nomic Report of Dodoma Region" Dodoma area, to give good yields. For (BRALUP 1971 p. 8), it was stated that instance, in the drv )ear of 1969 no "there is no land shortage and never will maize ripened on rainfed land in the be for quite some time." district. By 1967 the population of Ikowa Provided the annual cropping area in catchment amounted to 21,900 persons the earlv 1960s amounted to 15 % of the and the number of homesteads hias ap- catchment area, as suggested earlier, the proximately 2400 (cf. Rapp, Murray- cropping area per person was then 0.57 Rust et al. 1972 p. 268). If one assumes ha or 3.23 halhomestead (Fig 105). As that still around 15 % of the catchment millet strongly dominated among the was under crops, each person disposed grain crops and cash crops lvere practi- of only some 0.42 ha of cultivated land cally non-existent, one notices that with (3.854.0 halhomestead). With millet Table 24. Ikowa and Dodoma catchments. Population density of enumeration areas 1967. Number of homesteads in 1960 from topographical maps 1:50,000 and air photos. Average number of homestead members assumed to be 9.2. Number of inhabitants 1967 from census data. Note that the boundaries of enumeration areas A-Y (population reference area) do not coincide with the catchment boundaries. In the period 1960-1967 the total increase in population in the area was 39.4 % (5.6 % p.a.). In the same period the population density increased by 43.6 %. The considerable differences between different enumeration areas can be explained by overpopulation, exhausted soils and emigration in some cases and underpopulation, recovered fallows and immigration in others. Area G is Msalatu enumeration area. The Mkonze enumeration area (Dodoma catchments except Msalatu) not included in table (cf. Table 23).

Enumeration area 1960 1967 Code km2 Homesteads Inh. Inh/km2 Inh. Inh/km2

P P P -P Total 865.7 2,132 19,640 22.7 27,380 32.6 still the dominating crop2, the catch- In 197 1/72, with the villagization ment was not even theoretically self- programme (Operation Dodoma), the sufficient with food crops to maintain an pressure on the catchment eased adequate nutrition standard. temporarily. Part of the population had The decrease in available agricultural been moved to areas outside the catch- land was probably one reason for mi- ment and the remaining part, by then grations (Table 24). living concentrated in l l villages. amounted to just below 20,000 inhabitants (Manka et al. 1974) (Table 25). However, the ease in pressure on the In the 1971172 season 78.6 R of the gram crop hectarage In Dodomd d~strlct\\as tdken land \%.asprobably imagnary as some up h\ m~llet,l l 1 % by sorghum and 10 3 '2 large villages, as Buigiri, are situated b~ maize verl near the catchment boundary ancl IKOWA AND DODOMA CATCHMENTS. CULTIVATED AREAS 1960

Fig 105 Ikowa and Dodoma catchments. Areas under cultivation in 1960. The map is based on air photo interpretation. Each square represents approximately 5 ha, the average cultivated area of one homestead.

have a large part of their cultivated average need of land for cultivation in fields and grazing areas within the Dodoma district to 2 halstandard catchment although no part of the household of 5 persons4 = 0.4 population is included in the population halperson. (The figure presupposes figures for the catchment (Fig 10). maize as the main staple crop.) With an annual cropping area of 15 % of the catchment (9,200 ha) the crop- Average in the districct in 1967 was 4.3 ping area per person in 1972 was 0.46 persons, average in Chilonwa, Dodoma- Makulu and Mvumi divisions was 4.6 ha (1.93 ha/"householdn of 4.2 per- (TMMTD 1974 I11 p. 2). Average in the dis- 3 sons ). Berry (1971b p. 2) estimated the trict in 1973 was 4.12 persons per "family" (= household, production unit) (Bujra 1974 Average in 10 of the 11 villages P 8).

Tnblr. 25. Ikowa catchment. Population and livestock numbers 197 1/72

Village Population House- Cattle Sheep LU LU,' LU/ holds and goats capita household

hfulebe Nkambala hltumba Ihumrta hlgunga' Chololo Ikolra I hfnase K~kornbo' Xfakojd Kkongontha'

Total

' Figures uncertain Source: Manka et al. 1974 Table 26. Dodoma catchments. Area1 coverage of woodland/thicket and cultivations 1960.

Catchment Total area Woodland/thicket Cultivations (km2) (km2) (%) (km2) (%l Imagi 4.3' 2.3 53.5 0.0 0.0 Msalatu 8.5 1.9 22.6 0.4 4.4 Mkonse 25.3 6.5 25.8 4.8 19.1 Matumbulu 17.9 4.9 27.1 4.3 24.0 Total 56. l 15.6 27.8 9.5 16.9

Including artificial extensions.

In recent years cash crops, like wine, number of homesteads in the area only have started to compete with the sub- 121 were located within the catchment sistence crops on the best soils, leaving boundaries (approx. 11 14 inh., 23.5 per still less space for the production of kmz) at the time of the 1967 census, the calories and proteins. It increases the population within the enumeration area share of permanently cultivated fields had increased to 1945 (29.0 per km2), an and implies shorter fallow periods for increase of 18 % since 1960. the bush fallow cultivation and that in By 1960 the number of homesteads in turn requires manuring if yields per the "Msalatu enumeration area" was 77 unit area are not to decrease. Manuring and the population approximately 725 in turn requires a substantial cattle persons, a figure which in 1967 had in- population and means to transport the creased to 93 1 (38.4 inh/km2). manure to the fields. For details see For the Dodoma catchments a de- Appendix. tailed study of land use has been carried out, based on the 1960 air photos (Fig 54). Particular effort was devoted to the The Dodoma catchments problem of separating fallow fields from At the 1967 census the Dodoma catch- those that had carried a crop in that ments covered parts of two enumeration particular year. The result arrived at, areas (Fig 104). The largest of them 16.9 % of the total area of the catch- covers 67 km2 and includes Matumbulu, ments under crops in the season Mkonze and Imagi catchments ("the 1959160, is slightly higher than the Mkonze enumeration area"). The other supposed average for the district. The one covers Msalatu catchment and the relatively high figure is probably ex- areas h.' of the reservoir. It also includes plained by the fact that the catchments a small part of Ikowa catchment. are situated within an area that is in- In 1960 the number of homesteads fluenced by the high population densi- within "Mkonze enumeration area" ties around Dodoma town. It is well in amounted to 179 indicating a popula- accordance with the supposed average tion of some 1,650 inhabitants (24.61 for Ikowa catchment which is character- km2) provided the average size of the ized by similar terrain. The average area homestead groups was approximately cultivated per homestead member in the the same as in the 1940~.~Of the total Dodoma catchments, 0.85 ha, seems high, but it should then be noted that The average size of the homestead groups many of the fields in the catchments are in the entire district was then 9.57, and in cultivated by people who live in home- Ikowa catchment (Buigiri and Mulebe) 9.2 1. steads located outside the catchments. It Table 27. Size of Wagogo homesteads, size of homestead herds and average hectarage cultivated.

Area' l Population

Dodoma district Bahi neighbourhood Nagulo neighbourhood Buigiri neighbourhood Mulebe neighbourhood Mulebe ujamaa village Cilungulu "chiefdom" Temanghuku neighbourhood Matumbulu and Mkonze catchm. Iko~vacatchm Ikowa catchm.

I Bahi and Nagulo neighbourhoods are situated in W Ugogo near Bahi swamp. In this area cattle numbers were (and are) generally higher and the average cultivated area smaller than in central Ugogo. For the location of Mulebe and Buigiri neighbourhoods see Fig. 8. Temanghuku neighbourhood and Cilungulu "chiefdom" are situated some 20 km NW of Dodoma in an environment similar to that of Ikowa catchment. The figures are estimations based on counts of homesteads on air photos. Based on 10 out of 11 villages in the catchment (Nkongontha excluded). The figure is based on measurements of cultivated areas on air photos. C Figures are for maize on unirrigated land during the late part of the 1960s. The average homestead herd consisted of 11.2 head of cattle and 16.0 small stock. ' The average homestead herd consisted of 13.9 head of cattle and 13.9 small stock. Sources: Dodoma District Book; Agric. Dep. Village Economic Survey (1944); Rounce (1949); Rigby (1969b); Patton (1971); Berry & Kates (197 l); Xlanka et al. (1974).

is likely that the population that uses the (>100/km 2). This, however, is a recent Dodoma catchments for cultivation (and phenomenon as regulations during the grazing) is approximately equivalent to colonial time prevented Africans from that of "Mkonze enumeration area". settling in the outskirts of the town. Fur- The cultivated area then decreases from ther, in the 1940s and early 1950s there 0.85 to 0.58 halperson which is well in were no settlements or cultivated fields accordance with the supposed average within Msalatu catchment due to land for Ikowa during that period. use restrictions. However, in the late The population density in the areas N 1950s the regulations on migration were of Msalatu reservoir is very high no longer efficient and the restrictions Cultivated area l Number of stock

on cultivation were abandoned. Thus on Livestock density and carry- the 1960 air photos cultivations appear ing capacity in the catchment. They then covered 4.4 % of the area. The photos also show With a population of 125,000 inhabi- that grazing was common. 20 % of the tants and a total number of livestock catchment showed signs of serious soil units (LU)'j of 260,000-270,000 in the erosion. year 1911 the per capita figure of Ten years later, in 1970, cultivations around 2 LU/person has changed only covered some 20 % of the area. The little through the years (Vageler 1912; whole catchment except the inselberg Obst 1923). In 1944 the number of LU slope was badly overgrazed and the per capita in Dodoma district was 2.1 areas affected by serious soil erosion and the average homestead herd 20.1 now covered an estimated 30 % (Table 27). From areas in central Ugogo One effect of the above conditions was a 40 Q/c increase in the sediment yield in the period 1960-1970 com- 1 Livestock Unit (LU) = 1 head of cattle = pared with the period 1950-1960. 5 small stock (sheep and goats). This evaluation factor has no relation to the values the Wagogo put on their livestock. For other livestock unit conversion systems see Dahl & Hjort 1976. p. 223. Small stock /-,/------_ .-- - C ------P-- ---__I- 0.0 I 1 I I I l 1925 1930 1935 1940 1945 1950 1955 1960 1965 year

Fig 106 Central Province livestock fluctuations, 1925-1964. Arrows indicate two or more consecutive seasons with below average rainfall at Dodoma Met. Station. Stock numbers are expressed as livestock units (LU). To obtain number of sheep and goats multiply by 5. As shorin in the diagram, the total numbers of stock varied considerably over shorter periods (up to 40 % over 2 years) but the tendency seen over the whole period was astonishingly constant. Prolonged droughts have immediate effects on the livestock numbers (cf. Ndorobo 1973 p. 30; Hankins 1974 p. 98). The 1946147 season got extremely large amounts of rainfall. Considerable areas were flooded. The following season was drier than normal. The combined effect was a 21 % decrease in stock numbers. The 1952-55 period was characterised bp disastrous droughts and consequent intense overgrazing. From 1953 to 1955 there was a drop in the total stock numbers by 26.7 % while the cattlefigures dropped by 30.4 %. In Dodoma district the drop was from 450,000 to 179,000 (60.6 %) (Skerman 1968 p. 49). In the 1950s Government destocking programmes were in operation but they had only limited effects on the total stock numbers. (Ann. Rep. Agr. D. 1936 p. 33; TMA 1977 p. 13). Towards the end of the 1950s most restrictions concerning livestock were abandoned. This in combination with a sequence of years with rainfall above normal led to a 46.7 % increase in the stock numbers between 1957 and 1964. "Now (mid 1970s) numbers have increased so that even at the commencement of the dry season it is difficult to find adequate grazing." (TMA 1977 p. 29).

(Buigri and Mulebe) figures of A more detailed survey in a smaller 1.55-1.87 LU/capita and average herd area, Temanghuku, which is part of sizes of 15.0-18.1 were reported while Cilungulu chiefdom, pointed at an av- in W Ugogo homestead herds of 20-24 erage homestead herd of 16.8 LU or 1.8 LU and LUIcapita well above 2 seemed LUIcapita. However, the Wagogo do more common. In the case of Buigiri not willingly divulge information con- 60 % of the animals were cattle and cerning their cattle (UNDPIFAO 1968b 40 % were small stock. However, a more p. 5),but cases are recorded of individ- common distribution seems to be uals owning over 1000 head of cattle around 30 % each of cattle and small (Ann. Rep. VSD 1945 p. 13; TMA 1977 stock, or even a dominance of small p. 29). stock. (FAO 1967; Ann. Repts. Agr. D.). There is a good accordance between During surveys by Rigby (1969b) in the figures recorded at Mulebe and the early 1960s in Cilungulu, an area h' Buigri in 1944 and those recorded at of Dodoma, the average homestead Temanghuku in 1962 pointing at the herd was found to consist of 14.4 LU fact that both the average size of equivalent of 1.68 LUIcapita (Table 27). homestead herds and the number of The figures are based on tax material stock unitslcapita changed relatively lit- which may imply that they are too low. tle during the period. Both 1944 and 1962 were years following droughts. (calculated on the whole catchment) was Thus the figures indicate the lower level 2.7 haiLU. However, in practice, areas ~f the range within which the herd size for grazing amounted to 1.9 ha1LU. varies between good and bad years (Fig From another area in central Ugogo 106). The total number of livestock in- (Mundemu),Bujra (1973) reports an ac- creased during the 1960s, but one can tual stocking density of 1.3 haiLU. assume that the average size of the It should be noted that even in areas homestead herd remained approx- of irrigated farming, as at the Ikowa imately the same during the period. scheme, livestock play a significant part Thus it is assumed that around 1970 an in the system and in 1969 there was an average size homestead herd in central average of just under 1 head of cattle Ugogo consisted, at a minimum, of 15 per capita. LU (12-14 head of cattle and 12-14 In the Dodoma area a minimum of sheep and goats). 2.5 ha of unimproved grazing is re- If the above average figures are ap- quired for each livestock unit to avoid plied on Ikowa catchment the number overgrazing (UNDPIFAO Livestock of LU disposed of by the people living in Project Dodoma quoted by Rapp, the catchment (16,237 inh.) would have Murray-Rust et al. 1972 p. 269). A amounted to around 29,200 in 1960, maximum stocking density of 2 ha/LU is with a ratio of 1.8 LUicapita. With 29 % recommended by Berry (1971b p. 1) of the catchment not suited for grazing and TMWD (1974) and a figure of 2.8 (inselbergs, steep upper pediments, ha/LU is mentioned by Bujra (1974 p. roads, built up areas, etc.) and another 13). When assessing the rural popula- 6 % possible to utilize during part of the tion carrying capacities for the districts year only (mbugas) and 15 % of the area of Tanzania, Moore (1971 p. 27) uses a under crops (dry season supplementary figure of 5 haiLU as a minimum for feed) there was left only 1.5 ha of graz- herding on a partly commercial basis. ingiLU. Provided the number of The figure, however, refers to other LUicapita remained at 1.8 and the area than standard livestock units. Converted under crops at 15 % of the catchment to the SLU system the figure is equiva- the available grazing had by 1967 lent of 4 haiLU, a density considered (21,900 inh., 39,400 LU) been reduced suitable on mbuga areas by Skerman to 1.1 haiLU (cf. Appendix and Rapp, (1968 p. 8). UNDPIFAO (1968b) esti- Murray-Rust et al. 1972 p. 269). mated a suitable stocking density of be- The villagization programme in the tween 1 head of cattle per 4-5 ha on early 1970s changed the trend. In con- ranching schemes in Ugogo. Then some nection with the resettlement stock improvement of the natural grazing was numbers were obviously reduced as anticipated. Estimates of carrying capa- were population numbers, as parts of city in Arusha region vary between 2 the earlier inhabitants were moved out and 10 haiLU depending on water of the catchment. In 1971172 the popu- supplies. Murray-Rust (1972) estimated lation of the area amounted to 19,974 the actual density of stock in Kisongo inhabitants and the number of livestock catchment at about 1 ha/LU, suggesting units was 22,557 mahng an average of that the local carrying capacity was ex- 1.13/capita 7 (10 per homestead). The ceeded 2-5 times. The UNDPIFAO theoretically available grazing area Masailand Development Rept. (1968a p. 10) estimated the traditional stocking rate in part of Maasailand at 1 head of ' C;alculated on 10 of' the l l villages cattle to 8 ha of unimproved grazing. Results from grazing trials at Kongwa If the all-year-round grazing areas during the years 1965-68 (three rela- and the areas needed for cultivation are tively dry and one wet year) showed that put together and the temporary grazing the average carrying capacity of sown facilities on harvested fields are added perennial pastures was 1 adult beef the result will be that each homestead of steerl2.5 ha. The average carrying average size in Ugogo, with a mixed capacity of natural unimproved grass pastoral and cultivation economy of cover (annual grasses and weeds) was 1 subsistence type, needs some 40 ha of beef steer14 ha (Wigg 1973 p. 373). land for grazing and cultivation. That For techniques of assessing livestock would allow a maximum of 2.5 carrying capacity, see Heady (1960) and homesteads or 5 households/km2 = Dasmann et al. (1974). 20-25 inh./km2, if all land was available From the above it can be concluded for agriculture. that 1 LU12.5 ha seems to be the However, large parts of Ugogo maximum stocking density to allow for (10-50 %) are not suited for grazing or an economy based on cultivation for cultivation. From this follows that the subsistence with an additional small in- maximum carrying capacity of central come from sale of livestock without im- Tanzanian drylands (tbe cultivation mediate destructive effects on the envi- steppe) under the current agricultural ronments. This implies that for each system is approximately 2 homesteads homestead (8-10 inh.) with 15 LU, some (with livestock) /km2 = 4 "households" 37 ha of unimproved grazing (dry and or 17-20 inh./km2. wet season) are needed. For subsistence The Dodoma Region Water Master crops another 4-5 ha are needed. Plan (TMWD 1974) gves a more The cropping area must be protected optimistic view of the situation: against grazing animals during the peri- "Since it is known that the average family od between preparation of seedbed and in Dodoma owns about five head of cattle at harvest (4-6 months). Then it may act as a stocking rate of two hectares this suggests a supplemetary grazing area. To in- that 10 ha of land is needed per family plus possibly two ha for cultivation. From these crease yields and allow for a higher de- figures it could be concluded that 12 ha of gree of permanent cultivation increased land per family will imply a density of not use of cattle manure is necessary9. more than eight families or 33 people per km2".

R The optimum number of cattle which an area can carry (carrying capacity) is the number which can be carried comfortably The supply of manure has to come from through an average year allowing a margin the cattle bomas. An important problem is for drier conditions. that some of the dung is lost when the cattle Experiments referred to by Rounce (1949 are grazing in the bush. Thus, in practice, p. 32) indicated that under conditions prev- the above figures on possible permanently alent in Ugogo the most economic and effi- cultivated areas in relation to stock numbers cient use of cattle manure was application of have to be reduced by a certain percentage. 7.5-10 tons per ha every 24 years (cf. For permanent cultivation manure from 3 Skerman 1968 p. 36). As one cow produces LU/ha is needed. Consequently. for perma- approximately 1 ton of manure/year a nent cultivation of 12 7c (159.000 ha) of homestead herd of 15 stock units would Dodoma district, manure from 477,000 LU theoretically produce enough manure for is required. This leaves 2.45 ha of available permanent cultivation of 6 ha with an appli- grazing per LU which is the approximate cation of 10 tons every fourth year or 4 ha lower lirnit of the stock carrying capacity of with an application of 7.5 tons every second the area (Rapp. Murray-Rust et al. 1972 p. year. 269). In the above the present situation of 720,000 (= 1.84 hallivestock unit of all overstocking is not considered and in land in the district of reasonable suita- the assessment the share formed by bility for agricultural use = cultivation "small stock has been left out. Accord- and grazing) (cf. Appendix). ing to Rigby (1969b), that share often With an estimated human population forms 50 % of the herd or 10-15 sheep in 1977 of 354,000 (TMA 1977 p. 12) and goats (2-3 LU) per homestead. the ratio of humans to livestock was It is not mentioned in the Water Mas- 1-2.03, i.e., nearly exactly the same as in ter Plan whether the estimations are 191 1 (1-2.12) and in 1944 (1-2.1). Both based on the rural population only or if 1944 and 1977 were years following se- they include the total population of the vere droughts, hence the figures are district. If the urban population is in- probably low. In good years the stock cluded the figures must be slightly ad- numbers seem to build up rapidly (Fig justed as the town dwellers (10 % of dis- 106) and in periods of prolonged trict pop.) only to a limited extent are drought (approximately one in a de- cattle owners (Berry & Kates 1970). cade) the cattle numbers may be re- In 1977 it was estimated that there duced by as much as 50 %. were some 600,000 head of cattle in the This should be compared with the Dodoma district (TMA 1977 p. 12). As estimated minimum carrying capacity of during times without restrictions the the district of 2.5 ha/LU, which in 1977 ratio of cattle to small stock seems to be implied a maximum stocking rate of 1:1, one can assume that the total 1.2-1.5 LU per capita or a homestead number of small stock also amounted to herd of 10-13 LU depending on the size around 600,000, resulting in a total of the area taken up by cultivations (cf. number of livestock units of some Appendix). 16. The future

The new settlement pattern; considered by the authorities as being of its impact on the semi-arid purely technical nature. The plans and the solutions rested with the authorities. environment The major task was to supply the in- The pre-independence Tanzanian novations and to see that they were im- peasant society was characterized by plemented. The regional authorities small scattered farming units where the decided about the location of the new production was maintained by means of villages. The farmers themselves were simple hand-tools and the products seldom consulted on practical matters or were consumed within the production on the agricultural suitability of the unit (household). This economy seldom areas chosen (Rudengren 1976 p. 766). gave any large surplus. In the Dodoma area the villagization To increase production and to facili- programme (Operation Dodoma) start- tate provision of social services, Pres- ed on a large scale in 1971. By Septem- ident Nyerere (1967) in his writings on ber 1973 the whole population in the "Socialism and Agriculture" advocated district (250,000) had been moved into the idea of changing the traditional set- 142 villages, some of which were new, tlement pattern. others were traditional trading centres, The peasants were to move from their not necessarily having land resources scattered homesteads and concentrate for an expanded population. The aver- in sizeable villages located at strategc age size village in Dodoma district in points. In the villages (ideally consisting 1973 had 1,763 inhabitants making up of some 250 households each) all larger 428 households (4.12 inhlhousehold) investments (water, schools, etc.) should instead of 250 as originally planned. be on a communal basis as well as the There are examples of villages with production (crop and animal) and the more than 1,000 households (Bujra marketing. 011ly small garden plots 1974 p. 9). should be owned privately. In this way Most of the villages in the district are rural development tvas intended to be concentrated in the central belt, an area brought about quickly and efficiently which is about one-third of the total area managed. This effort is the direct of the district. Even before the move- opposite of the development program- ment this area had a much higher popu- mes of the 1950s and early 1960s'. lation density than the rest of the district The planning stage preceding the (Fig 29). Thus, at the moment 90 % of implementation of the above ideas was the population of the district is concentrated on one-third of the total area (Bujra 1974 p. 9). This has changed the l "It is hoped ~viththe development of water man-land ratio unfavourably. supplies in the Central province to spread From a soil conservation point of view the human and animal population more criticism can be raised on [he size and evenly . . . . Overcrowding naturally occurs ocation roduction in the hest jVatered parts and soils there be- l of new The p come worn out" (Ann. Rep. Agr D. 1955 p. capacity in relation to the expected 25). population was never assessed and sel- Fig 107 Severely degraded land at Singida. Orergrazing followed by wind- and water erosion has given the area a desert-like appearance. (Photo: C. Christiansson, Nov. 1974.)

dom if ever was the question considered age of 1,860 inhabitants per village, that about how to maintain a viable equals 7,986-12,276 ha or 13-20 % of agricultural production with existing the catchment (cf. page 168). Such large technique^.^ Further, I know of no case areas of agriculturally suitable soils are where a proper soil survey was.carried of course difficult to find adjacent to the out prior to the final selection of a vil- village sites. lage site (cf. Krokfors 1978 p. 205). Consequently, during the first years Still, a general aim in the district is to after moving to the villages it was com- increase the average yield per ha 3-fold mon that apart from cultivating the (Bujra 1974 p. l l). However, with the fields that were cleared in the new area present agricultural techniques the con- the farmers proceeded to cultivate land centration of population in the central in the vicinity of their old homesteads, belt makes it difficult even to maintain generally situated at a distance of 2-15 the existing production ~apacity.~ km from the site of the village. If the soil The present villages in the district reat the old homestead was considered quire 0.4-0.6 halperson or 700-1,050 better the homestead groups tended to halvillage (depending on which is the move back, at least during the planting main crop) of annually cultivated land and harvest season. just to feed the local population in years Such conditions were looked seriously of average rainfall. In an area like Ikowa upon by the authorities as they threat- catchment with 11 villages with an aver- ened the consolidation of the new villages. In some districts bye-laws were instituted to prevent cultivation of fields ' Several cases exist ~vhere$\.hole villages situated far from the villages. Con- have had ro move a second time (Rudengren sequently, large areas of marginal land 1976 p. 767). In the 55 years f'rorn 1923-1977 Ugogo closer to the villages had to be taken into experienced 18 !,ears ~\.ithconditions rang- use and in many cases fallow periods ing from "appreciable food shortage" to "se- had to be reduced. At the same time, in vere famine". Four of these occurred in the more remotely situated tracts, the bush 1970s (TMA 1977 p. 9). From 1969 until the end of April 1977 Dodorna repon received a started invading cleared areas with a total amount of f'amine relief food of 294, possible infestation of tsetse as a con- 248 tons (op. cit. p. 10). sequence. (Intense cultivation and grazing is a prerequisite of preventing is the organization of the livestock econ- spread of tsetse into critical areas.) If omy. The movement of people to the spread occurs traditional grazing areas central belt of the district has meant an are lost, reducing the total grazing po- increase in the livestock population of tential (cf. Douglas-Hamilton 1975 p. that area. Hence the risk of further 235). overgrazing is most serious in those Moreover, all households must collect parts of Ugogo which already suffer firewood and water every day. Tradi- most from soil erosion. tionally these are jobs reserved for However, "only when defined groups women. However, due to other work of people are responsible for defined (hoeing, weeding, domestic work, etc.) areas of grazing, will automatic limita- not more than a few hours a day can be tion of stock be achieved. The definition spent on wood and water collection. of village boundaries under Tanzania's Consequently, all woody material within ujamaa programme is a first step in this a radius of some 5 km mill successively direction" (TMA 1977 p. 13). be used up. And it should be noted that The concentrated settlements based this is also the zone which is most en- on communal production also have en- dangered by overcultivation and over- vironmental advantages in that the total grazing (Fig 107). labour force of a village-if organized Movement to the villages had one and motivated-can carry out such en- very important attraction to the peas- terprises as soil conservation work that ants: Clean water would be provided to in the long run may increase the returns all, according to government promises. from the agricultural sector. Thus be- However, the programme of constiuc- tween 1973 and 1978, 5,400 ha of tion of water supplies is advancing very eroded land was reclaimed in Dodoma slowly as it is not only a technical matter region as part of the undertakings of the (BRALUP 1971 p. 13). HAD0 project (The Dodoma region It is a wellknown fact that Ugogo is soil conservation project) (Daily News, devoid of permanent surface water 3 1 October 1978). A near future target except in a few reservoirs. It is also well is reclamation of 2,000 ha annually known that the ground water resources (Daily News, 30 November 1977). How- are sparse, sometimes of low quality ever, "it is doubtful whether operations (saline) and expensive to utilize.( An on the present scale can keep pace with important factor to be considered is also the accelerating rate of degradation" the estimation by Woodhead (1968) of a (TMA 1977 p. 3 1). zero net recharge of the underground Also such arrangements as village water supplies in central Tanzania. firervood plantations are environmental These conditions were clearly pointed necessities suitable to run on a com- out in the Dodoma Region Water Mas- munal basis. However, from today, it is a ter Plan (TMWD 1974) which, however, matter of decades before they are was not published until Operation established and start producing suffi- Dodoma had already been completed. cient amounts of ~vood. Another major problem of en- In the central belt of Ugogo the many vironmental significance, not yet solved, concentrated settlements, in combination with the traditional agricultural methods, are creating an ecological imbalance that within a near future may ' Of' 59 boreholes drilled in the early 1970s. 37 yielded water suitable for domestic use undermine the production potential of (Bujra 1974 p. 12) the area. The present trend is that the resting period of the fields is reduced cient to provide food for an expanding and land is cultivated before fertility is population, so the solution was adopted fully restored. This means earlier of taking up land at a distance from the abandonment and a premature move to villages, so great that it was not possible even less recovered land. And so finally to walk home every evening. "Lands" a point of no return is reached and the houses were in consequence established, community has to move to an area not where those members of the family con- yet ruined. cerned with cultivation lived for the An example of the development of whole of the cropping season. More- such a pattern is referred from Bots- over, a considerable proportion of the wana by Fosbrooke (1974). population possess cattle, but owing to The Tswana, presenting an example the absence of sufficient grazing in the of agro-urbanization in semi-arid envi- vicinity of the villages, a large propor- ronment, live in villages of considerable tion of the livestock is placed in cattle size, each with many thousand inhabi- posts far from the villages, herded by tants. Ruined villages show that this has the boys. been the dominating settlement pattern Another example of impoverishment for several hundred years. This evi- of the environment through concentra- dence also shows that the villages moved tion of people and stock was docu- from time to time. The root cause of the mented by Deshler (1960). He showed movements was probably the decreasing how, in the early 1950s, parts of Kara- fertility of the cultivated area around moja in N Uganda became wasteland the villages. About 100 years ago such due to the high densities of people and movements largely ceased. To com- livestock which the area had to support pensate for decreasing yields when the during the evacuation from tsetse inland around the permanent villages de- fested areas. teriorated until it was no longer pro- One can certainly see a possibility of fitable to cultivate it, larger areas were the repetition of the above patterns in cultivated. However. this was not suffi- semi-arid Tanzania. 17. Summary and conclusions

The purposes of the present study were Wahehe but mainly due to the expan- defined in the introductory part. They sion of the caravan trade, the pressure include description and analysis of on central Ugogo became excessive factors important for the initiation and from the mid-1800s. The woodlands acceleration of soil erosion as well as a dwindled and sloping ground was taken documentation and discussion of into use for cultivations. The regenera- sedimentation and other geomorpho- tion of trees was prevented by short logical effects of erosion in the study fallow cycles and by the high demand area. A second aim is to give an answer for firewood and charcoal, and possibly to the questions: For how long have the also by a slight shift towards drier cli- processes been going on? Are they matic conditions. With a strategic posi- accelerating at present? What are the tion along the caravan routes, Ugogo future perspectives? played an important role as a food producing area. It is thus estimated that at the peak of the East African caravan History of the soil erosion trade the soils of central Ugogo had to complex in Ugogo produce some 6,000 tons of grain annually, in addition to the consump- Reports by early European travellers tion needs of the local population. about the existence of gullies, soil Many indications show that it was only pedestals, extensive vegetationless areas, possible to produce these amounts in rapid runoff from slopes and high silt "good years". In years with low rainfall concentrations in stream water 211 indi- the pressure on the ground increased cate that soil erosion by water was com- due to scarcity of grazing and need of mon in central Ugogo already before larger areas to produce the same the colonial era. As also indicated by amount of food. As a consequence, 19th century reports, wind erosion lcas fallow periods were reduced and mar- active in the more open parts of the ginal soils were cleared, in man): cases country, frequently increasing the dust on steep slopes. Thus soil erosion that content in the air to levels troublesome earlier had been localized now spread all to humans. along the caravan routes through However, the general impression of Ugogo. The trend ~vas temporarily Ugogo as described by the early broken in the 1890s, as the grazing travellers is of a fairly well wooded pressure was reduced due to the country. The few and scattered cultivat- rinderpest. ed fields seem to have been located The processes seem to have speeded mainly on flat or gently sloping areas up again in the first decades of the 20th never affecting steeper pediment slopes. century, as noted by British colonial The area being tsetse-free and ~vithre - officers in the 1920s and 1930s. Mith latively favourable grazing conditions the establishment of a Soil Conservation had considerable herds of cattle. Cornmittee in the 1930s conservation Partly as a consequence of repeated became an integrated part of the policy threats from the Maasai and the of the Agricultural Dept. Although never particularly successful, the con- b) Effects of the 19th century caravan servation programmes and the restric- trade, with large demands on food tiona on land use in Ugogo obviously and firewood. Considerable areas slowed down the rates of erosion, at were deforested. In areas with the least locally. In many areas gully growth highest pressure on the land the was retarded as a consequence of pro- production potential was greatly re- hibitions on cultivation of upper pedi- duced due among other things to re- ments. Conservation of vegetation on duced fallow periods, as well hill tops also had positive effects. The illustrated from, for instance, most obvious results were achieved in Mpwapwa. areas which were closed to grazing c) 20th century expansion of the herd and/or cultivation. When the restric- sizes well above the carrying capacity tions were abandoned in the 1960s and of the area, particularly after se- 1970s the sediment yields rose again, quences of wet years, causing over- culminating in the 1974-75 cultivation grazing in dry years. This was made campaign when many new gully systems possible through the introduction of were formed as a result of the clearing veterinary medicine and construc- of steep upper pediments and even in- tion of new water supplies. selberg slopes (Christiansson 1979 p. d) Drastically increased population 14 1). densities in some areas particularly Thus many old, abandoned gullies after 1960 with land shortage and that were not actively eroding in the pe- reduced fallow periods as a conse- riod 1949-1960 (covered by air photos) quence. probably date back to the later half of the last century. The same refers to the many denuded vegetationless crusts Processes and rates of ero- which today show the same appearance sion and sedimentation as they did in the 1940s. It is not likely that they all lost their soil cover in this The rates of erosion are high in the century as such features were recorded catchments investigated. They corre- as early as in the 1870s. As a conse- spond to sediment yields of 174-602 quence of the above it is assumed that m3/km2 per year as averages for the many of the large sand fans that occur longest periods of available records in the study area also orignated in the (Table 22). These sediment yields are of last century although some of them have the same order of magnitude as those been extended in later periods of in- recorded in similar environments in tense erosion. other parts of East Africa. They also Thus the initiation and large scale de- correspond to erosion rates recorded in velopment of soil erosion in central catchments of similar size and relief in Ugogo can be largely attributed to: the semi-arid parts of the USA (Fig 91). The annual denudation rates range a) Pressure from hostile neighbours on from 0.45-0.75 mm in the smaller the Wagogo who during part of the catchments and from 0.10-0.38 mm in 19th centuiy crowded in and around the larger Ikowa catchment, calculated the central range of hills causing on the entire area of each catchment. overpopulation and overexploitation Considering the pediments only, the of land, best illustrated from the rate of denudation rises to between 1 Handali, Mvumi and Luatu areas in and 2 mm per pear on moderately the 1880s. grazed slopes of gradients from 2-6". Table 28. Ikowa and Dodoma catchments. Landform units with associated soils, vegetation and land use in relation to soil erosion.

Landform unrt Areal coverage in Topography (relief) Soil percent of study area

Bedrock hills 12 %. Percentage is lnselberg peaks and slopes. Steep gradients, Shallow, grey to l~ghtbrown roarse texture( or ridges. larger (about 20 A) 10-30". Rock outcrops,"tor" features and soils rnth ninny boulders. (Skeleral soils.) Por in the Dodoma catch- scattered large boulders. NE [rending fault and rvell drained Degradational phase ments and in the W structures. ' part of Ikowa catchment.

Up er 20 R Marked break of slope at inselherg contact Rather shallow, dark to pale red or brown pecLent Slope gradient 3-10" Slope profile slightl) sands loams. Depth Increases downslope. slopes concare. Around low inselbergs this zone The coarse or medium textured so~lson the may be lacking. upper part of the Lone gradually merges into finer texmred pale sandy clays. In some positions low porosity due to compact o necs, hut generally soils are reasonably well drained. Degradat~onalphase. P.0 " Lower 18 W. 10 % of rrhirh are Geomorphologically closely related to the Shallow to deep, coarse to medium texture( pediment made up of gently upper pediment slopes. Only gradual tran- pale red to yellowish s~ils.In some cases slopes sloping pediments sitions from the zones above and below. well drained hut hardpanc frequently .- and gently reaching from the Slnpe gradient 34.5". Slope profile linear prevent efficient infiltration sloping tnselberg break of or concave. a slope to the valley ;:$gents. bottom and 8 % cons~st of gently sloping pediments beth,een steeper upper pedl- ments and valley bottoms.

Valley 18 W Slightly undulating to flat. Grey to yelllow colluvial and alluvial floors Gradients less than I". soils. Great contrasts in texture. Soil and depth 8-20 m above deeply weathered plains granite. Hvdromorphic concretions and rometimes extensive hardpans. In some {laces several sets of deposiuon can e ohsened. Recent format~onof alluvial sand fans. Aggradauonal phase. / Mhugas Extremely flat. Gradients less than 0 5" Black or dark gre) allu~lalcraclung cla)s Sometimes shallow channels or more de- with concreuons of calcium carbonate. fined stream courses occur on thr mbuga Uniform soil profile. Sand, 1a)ers are surface. common in certain positions. The so11deptt is great. Thicknesses of over 50 m haie been recorded In mbugas at Ikorta. Initiall) when the rains start infiltration 1s good Into the still open cracks. Wet mbuga cla~s swell and prevent infiltrauon. The surface then becomes Inundated.

Fan shaprd arcumulatlons with slightly Grey to yellow alluvial so~ls,mainly of convex surface. Found at thr contact between sandy or sandy loamy character. The pro- pediment and valley bottom and at places files lack proper horizons and d~stingulsh- where sediment-laden small streams reach able lavers reflect current beddine onlt. channels of higher order. Gradients 0 .G>". The soil depth is considerable, g&eraIly se~eralmetres. Infiltration is significant and ground water conditions fa';,ourahlr

19 R. (22 R uf Ikowa Vrry grntly to moderately undulating. Slope Grq to !cllow, COPI.FCto rned~umrcxtuz-cd catchment) grad~entsbetween 1-0.5". soils. Soil depth 1s moderare to ~hall<~n.P<,- The mlcro relief is domtnated h\ the "pock rosm is modcrate. On the bare surfacec thr mark or termite mound attern", abandoned coil tcxturr is finer thrn in the currouncl~ng termite hllls. These are s&htlr domed, tcrraln, clar being the dominant tracr~on. nrcular >egetation-less parches of 10-25 m lnfiltrat~onon these surface3 1s neglivhle diameter. Inter\als hetxeen patches arc 10-60 m

Fault In the areal ~n- Former peneplain surface verticalli d15- Shallow to derp, sand? to loam\, gre~~sh. topograph~ ventor\ th~rcate- located through faults or flexurec. >lain hrown or red wilt. Gcnrrall~su ell dra~ned gor, 1s ~ncludcdin svstem trending NE-SW. Scarps, 50-100 ni except where hardpans i~ccur.Thrcc preie~ the figures for high. facing NIV. Sl~ghttilt of hlocks cffic~entlnfiltrat~on causlng harrcn erod- "hedrock hill?" and torrards SE. Lrsr distinct E-\V trends sug- ed inrfacrr "upprr pedimrntc". gected by certaln river gallr,$. Zones of movement frequentla h1ddr.n beneath loose deposits of great dcpth. Surface gradients of loocc depos~tsalong trc- ronlr I~nc\from a fei\ to 20". ation Land use Erosion

m~ornhowoodland and deciduous L~mitedgrazing by small stock. Scattered Some splash, sheet wash and seepage ,t. Poor field layer. cultivated plots (bush fields). Some charcoal burning. , Adanronu, Combreturn, hora, Dtchroslachq.~.Euophmbm, s bush. Thin field layer. In Grazing and limited shifting cultivation Intense rill and gully erosion. Tree denser thicket particularl) (hush fields). Increased population pressure mounds indicate heavy splash erosion ds the lnselberg slope. Vege- implies extended use of this very sensluve and sheet wash. modified by earlier culti- and marginal zone for agricultural purposes. 1. hurn~ngand grazing. , Adonronu, Carsul, Combreturn, stachyr Grasses include Chloni, .~clenrum,Dtgrfam, Erayortn. i bush, scattred trees, sparse In sparsely populated areas wet and dry On steeper sections with sparse vegetation a)er Vegetat~onmod~fied season grazing. Some cultivation. How- cover severe rill and gully erosion. Tree tiratrons, hurning and gra- ever, large parts of this zone have a dense mounds-splash erosion and sheet wash. pattern of settlements with larger areas Usuall) dense network of shallow gullies. ,s same as on upper ped~ments. under cultivation. Temporary grazing occurs Hardpans in the soil profile prevent deep- ~selypopulated areas the lower after harvest of crops. going erosion. tents are heavily cultivated On the permanently mltlvated sections of atural vegetation is largely lower slope gradient few signs of gully g- erosion. Small rills across sloping fields dary shrub on fallow fields common. Intense splash and sheet wash in le Jurttca, Solanurn and Ver- the earlv rains season. and in association with thicket Extensive cult~vation.Some dry season Erosion is limited to splash and sheet attered shrubs. The zone is grazing after harvest of crops. The zone %,ashin the early rainy season. In the y cultivated. Only remnants ot is densely settled. dry season wind erosion is significant, 11 vegetation occur often left particularly when cattle graze on har- ual reasons. vested fields and loosen the soil surface through trampling. Vegetation-less surfaces suffer from wind erosion.

~nnualinundations prevent growth Mainly dry season grazlng. Some cultivation Shallow channel flow occurs in the mar- es Vegetat~onis predornlnantl~ along the frlnges. ginal areas. Where back cutting takes and place stream channels with sandy beds r ieyal is adapted to the spe- establish themselves. Main channels may -owing conditions. Among be incised 1-3 m into the mbuga clay. .S Cynodon, Pantrum, Pennurtum 'hemrda ma, he noted.

al vegetation sparse. Complete Homesteads often lorated along the fringes Mainly deposition. Surface eroston occurs f trw vrgetatlon. Field of the fans. Significant infiltration ~m- in time of flash floods On the more of herbs and grasses. plies permanent hater in wells dug into the steeply sloping fans channels may develop. sand) sediments. Ground-water tiithln easy Eros~onalso occurs where the front of the reach pro\ides favourable conditions for fan reaches a water course. Irrigated cult~vationof vegetables on small plots. Some grazlng and cultivation of main crops on thc larger fans. itinuous hush and thicket inter- Grazing and limited rultivation. Some On vegetated surfaces generally veri I nith patches of grassland. charcoal burning and firexood collection. little erosion. Where the vegetation has ion species are Aiacu and Grruza been destroyed on gently sloping ground intense gully erosion. itarla are compietel~hare

thlrket and remnants of Same as on upper pediment slopes The loosely consolidated weathering pro- and. ~eatl\mod~fied h> ducu h~dingthe scarp faces are easily g, hu;ning and temporarr eroded. Sheet hash and gullying are severe ltlon Spetler same as In many localities. The material is often prr ped~menrslopes deposited as sand fans at the base of the scGps. Due to increased infiltrat~onsand fans also form where water courses cross fracture zones on less steep slopes. The gentlv sloping dipslopes also suffer from gull! eroslon in place5 where vegetatlon ha5 been removed. Bare pediment surfaces with sandy policy in the area during the colonial loamy soils in overgrazed parts of the period with restrictions on cultivation of catchments experience denudation rates upper pediments and the large scale of around 10 mmlyear, a conclusion contour banking programme. Examples based on geomorphic evidence from of gully extension were again observed splash pillars, root exposures and tree in the field during the survey period mounds, and on available records of (1970-74). In parts of the study area erosion plot investigations. very large gullies occur. These are asso- Accordingly, with the present agri- ciated with fault lines. Their incision has cultural system the thin soil cover on the been facilitated by the thick deposits of bare parts of the upper pediments will colluvial material in front of now com- be lost down to bedrock within the span pletely buried escarpments. of two generations. Apart from relief, soils, percentage Erosion and sedimentation in the vegetation cover and direct land use catchments occur according to a well factors, erosion is governed by rainfall defined pattern (Table 28): Some splash intensity and runoff characteristics. and rilling on inselbergs, gully erosion Concerning runoff it may be concluded sometimes combined with splash and that in the semi-arid East African envi- sheet wash on upper and middle pedi- ronment runoff from small catchments ments. Lower cultivated pediment of a few km2 may reach as much as 40 5% slopes experience less intense gullying of the effective rainfall or 50 % of the but instead widespread splash erosion, total. sheet wash and rilling. Deposition However, in the study area runoff occurs as thin sandy sheets on gently values from 20 % and downwards of to- sloping pediments and on sand fans in tal rainfall seem more characteristic. In positions where pediments meet valley large catchments like Ikowa where bottoms, in stream channels with deep much water is absorbed in sand fans or sandy accumulations, in mbugas that act flat areas of loose and porous soils or as traps of fine grained and in some trapped on mbugas, the recorded cases coarser sediments, and finally in runoff distributed over the whole the reservoirs where heavy sedimenta- catchment area is considerably lower or tion of predominantly clayey sediments just a few percent of the precipitation occurs in the larger Ikowa reservoir and (Tables 10, 18). silty/sandy material in the smaller re- The large influence of clearance, servoirs. cultivation and overgrazing on soil and Due to the rapid sedimentation, ttvo water losses from the lower pediments is of the reservoirs thus have an expected exemplified by the results from the life of only 3045 years. The other two Mpwapwa investigations. The principal have an expected life of 80-90 years and value of those results is in providing 120-130 years, respectively. data at the plot scale of the effects of Field inventories compared with various cropping and conservation studies of aerial photographs from 1949 practices. and 1960 showed that the dense The data clearly show the importance network of small gullies of the upper of some form of cover, as bare plots, pediment zones and the larger, more tvhether cultivated or uncultivated, lost widely spaced gullies of the lower pedi- extremely high amounts of soil and $\a- ment zones had the same distribution ter; nearly 90 m3 of soil per ha (9,000 pattern in 1949 as around 1970. This is m3/km2) and 50 % of the total rainfall. possibly a result of the conservation Losses were much reduced under a grain crop like bulrush millet and sor- - Identification of areas sensitive to, or ghum, and became insignificant under already suffering from, land degra- an undisturbed vegetative cover of dation. thicket or grass, which induces infiltra- - Quantitative documentation and tion and reduces soil loss. If this cover is comparison between the rates of gully damaged by (over-)grazing, water and erosion, intergully splash, sheet wash soil losses rise rapidly. and rilling. From the above observations it may be - Continuous recordings of the water concluded that both splash, sheet wash and sediment budget in existing re- and gullying form dominant processes servoirs. in the investigated area. Erosion by - Studies on losses of dissolved matter splash and sheet wash are quantitatively from the soils of the catchments. more important than gully erosion but the gullies are important features as Secondly, protection against soil erosion they act as transportation channels for for sustained yields on cropping land runoff water and soil from the in- and maintained livestock carrying tergully areas. The gullies have a dis- capacity in grazing areas should be ap- continuous growth related to the inten- plied by means of the most appropriate sity of land use and the occurrence of of the methods indicated below. maximum rain storms of long return As soil moisture is the major limiting periods. factor on crop yields under the prevail- Morphological activities of the kind ing climate there is a need to preserve described are active in all parts of the the soil structure to increase infiltration catchments, but they are greatly accel- and reduce soil erosion. Much of the erated in connection with overcropping losses can be reduced by employing re- and overgrazing on inselberg and latively simple conservation practices. pediment slopes. The details of the pre- Provided adequate fallow periods are al- sent morphology of the area can thus be lowed the losses are counteracted under understood in terms of the intensity of crops such as the traditionally grown the observable processes. These pro- bulrush millet and sorghum. cesses in turn are functions of type and If interplanted with pulses, sweet intensity of land use today and in the potatoes, gourds, etc., as is common recent past. practice in Ugogo, they afford a good The methods employed and tested in soil cover during the major part of the this investigation have provided val- rainy season. Cassava is another uable quantitative data. Sediment drought-resistant crop which gives good surveys in reservoirs, catchment erosion ~ieldson poor soils. It is one of the few surveys and air photo interpretation perennial crops possible to grow in the are, consequently, recommended for area and if planted as the last crop in a continued and extended use in crop succession it gives good protection Tanzania and elsewhere. to the ground when the field is left fallow. Conservation and manage- Over the last 20 years maize has become increasingly common in the area. ment This could have adverse effects as maize In accordance with the results of the requires good soil and adequate rainfall present study, soil and water conserva- and hill, in Ugogo, have a greater ten- tion in semi-arid Tanzania should con- dency to failure than the traditional centrate firstly on: grain crops. Its ~ultivationencourages clear stands, row cultivation and clean Thus the conservation of temporary weeding which all imply increased soil cultivations, fallow fields and grazing and water losses except on flat land. areas forms a separate complex of equal Soil and water loss from cultivated importance to that of conservation of fields can also be reduced by the leaving intensely cultivated areas close to the vil- of organic residues and the use of mulch lages. cover on the ground. However there is An integral part of the agricultural always a risk that such material is con- system and thus of the ecology of Ugogo sumed by termites before being able to is livestock. Livestock is important for fulfil its purpose. the production of manure. Livestock is Keeping fields small conserves the also a social and economic institution soil, as also does grass strips across the and a source of protein, though not the plots. Contour hedge planting by Eu- main source of subsistence. phorbia tirucallia has been shown to re- Grazing by stock should be so man- duce wash from the time of its planting. aged as to maintain a good grass cover. Being semi-poisonous this plant is never This can be achieved through carefully grazed, hence part of its value (Rounce planned provision of new water et al. 1942 p. 15). In contrast, sisal supplies, controlled grazing (fencing) hedges on the contour, practiced as a and controlled burning. This would conservation measure in parts of semi- favour grass and bush communities. arid Tanzania, only become effective as Grass, being shallower rooted than a control of soil loss when the plants are bush, loses less water by transpiration, large enough to protect the grass be- and as grass is as effective as bush in neath them from grazing. inducing infiltration such a cover con- Apart from biological measures, serves water more efficiently. physical measures are generally needed Too great a reduction in grazing to prevent excessive soil and water losses pressure will excessively favour the from cultivations on slopes steeper than establishment of shrubs (and trees) even 2". Thus, moisture absorption in the soil on the better grazing areas. This would can be much improved through the in the end reduce the grazing potential practice of growing crops on contour since many grasses rvould disappear by ridges and tie-ridges. This technique is competition. This would be unfortunate also extremely efficient in reducing soil since large parts of Ugogo are best losses. On particularly exposed hillsides, suited to production of livestock rather cut-off drains are necessary to carry off than arable agriculture (FAO 1967 p. excess water after heavy rains. 47; TMA 1977). In areas ~vherecultivation is perma- Furthermore, to keep a reasonable nent or semi-permanent, as will be the ancl sustained level of the crop yields case in the proximity of all villages in and to permit longer periods of cultiva- Ugogo, terracing seems the most tion, availability of cattle manure is adequate alternative (Wenner 1977). necessary. Thus there arises a problem The grasslbush fallow type of cultiva- of finding an optimum ratio between tion still practiced over most of Ugogo cultivated land, number of stock and makes the separation of cultivation and primary grazing areas. grazing activities difficult and poses re- Bare uncultivated land should be en- strictions on improvement measures. couraged to regenerate naturally by Over a foreseeable future that type of control of fire and grazing. In some land use will be dominarlt at least at cases total closing of areas to grazing some distance from the village sites. may be necessarj., and reseeding has to be employed to cover the eroded land. a geographical nor a biologcal problem. Emphasis should then be given to the It cannot even theoretically be solved as selection of suitable forage crops that easily as many of the other parts of the ulould be able to colonize the denuded soil erosion complex. Its solution most areas as rapidly as possible (cf. Pratt probably rests with social antropologsts 1963; 1964; Pratt & Knight 1964; Bog- rather than with political, economic and dan & Pratt 1967). agricultural experts. It may, however, Thus some of the major lines of con- be hoped that an improved and more servation should be: a) improved grass labour intensive cultivation pattern management and h) carefully planned wiould to some extent affect the livestock and distributed watering points, c) in- numbers and thus the grazing pressure creased use of manure on cultivated in certain areas. fields and d) use of suitable cover crops. An improved vegetation cover on Proper management would be assisted grazing lands combined with physical by e) the planting of sisal hedges around measures on the cultivated slopes will fields. Thorn fences as traditionally reduce the flow into the reservoirs of used should be avoided as they are in- sediment and runoff water. The former efficient in controlling soil loss and effect is obviously positive as it keeps runoff and as the material may instead more soil on the fields and increases the be used as firewood. All steep slopes and useful life of the reservoirs. The latter other areas not suited to cultivation or effect may, in some reservoirs or in grazing should be f) planted to trees and some years, reduce the water storage to serve as village firewood and timber below full supply level. It will, however, plantations. Permanently or semi-per- be compensated through an increase of manentl) cultivated land on sloping the water surplus in the catchment soils ground should be g) terraced or h) in the early dry season, and a possible cultivated by means of ridges on the increase in the subsurface discharge in contour. the sandy river beds all the year round. All large gullies should be reclaimed Desilting of reservoirs is generally by means of i) checkdams and j) gully stated to be uneconomical. However, in bottoms should be planted with grass, small reservoirs that are periodically herbs and bushy vegetation. The seed- emptied it may be a possible solution ing of Star grass (Cynodon plectostachyum) serving a triple purpose: a) to increase on gullv floors has proved efficient in the capacity of the reservoir, h) to holding moisture and trapping soil deepen the reservoir, with consequent particles, as has been demonstrated at reduction in percentage loss of water by the Veterinary Research Station at evaporation, and c) the sediments re- Mpwapua. moved can be used to raise the height of The above conservation measures the dam (Fawley 195613 p. 81). As some would all be inefficient and meaningless of the reservoirs are being filled mainly without a parallel control of stock num- by sandy sediments the); can be useful as bers. Holtever, bringing about a reduc- ground water reservoirs after filling up tion in the number of livestock is neither completely. Appendix

Dodoma district and Ikowa erage area needed for the two crops for 100 standard households is 81.5 ha of catchment carrying capacity maize and 114.1 ha of millet or together Moore (1971) estimated for Dodoma 195.6 ha (803.9 hall00 households for district the average area needed for two cash and subsistence crops). Con- main rain-fed cash crops (ground- sequently, an area of 0.39 halperson or nuts-500 kglha, 1 Shlkg and castor- between 0.8 and 1.0 haladult is needed 400 kgiha, 0.6 Shlkg) for 100 standard for subsistence crops. households of 5 persons to 608.3 ha if a Together, cash crops and subsistence gross annual monetary income of Sh crops require 1.6 1 ha/person (14-16 2500 per household would be achieved. halhomestead). Consequently an area of 1.22 halperson Applied on Ikowa catchment this or between 2.5 and 3 ha/adult is needed would mean an annual cropping area of for cash crops. 32,158 ha or 52.5 % of the total area of In Moore's example the whole calorie the catchment. As a considerable part of requirement (2000 kcal/person/day) is the catchment is unsuited to any kind of supposed to come from two major agriculture and large areas are already crops. In the case of Dodoma district seriously damaged by erosion, practi- these are suggested to be maize and cally nothing would be left for grazing, millet. They are assumed to provide nor would there be anything left for 50 % each of the calories needed. How- long term fallows. Without cattle there ever, Moore assumed the average would be no cattle manure. Thus in- yieldlha of maize to be 800 kg while it is tense fertilization would be necessary. probably 700 or less in a dry area like Moreover, without mechanization of the Ugogo. 700 kglha is the average maize agriculture it will not be possible to yield for Tanzania on unfertilized land. cultivate an area as large as 3 or 4 ha per In Ugogo the average on unfertilkzed adult. The average maximum area pos- land is 480 kg/ha and on fertilized land sible to cultivate for one adult in one 830 kg/ha (TFNC 1978). The average season with existing technology is yield of millet is, according to Moore, around 1.5 ha (Rounce 1949). 900 kglha. However, in environments like Ugogo it has been shown to be less than 500 kg/ha (Rounce 1949, Acland 19'7 l).' With the lo~veryieldslha the av-

' The figures on crop yields must be considered with great caution "since acreages are unknorzn and the yields per acre are variable" (BRALUP 197 1 p. 17). Estimated population and livestock carrying capacity, Dodoma district (Figures from Moore 197 1 ; Berry 197 1 and this study) District area: ...... 16 576 km2(100 %)2 Urban and waste land: ...... 77 km2 Forestreserves: ...... 280 (2.2 %) Theoretically available for agricultural use: ...... 16 219 km2(97.8 %) Area not suited for cultivation and grazing (inselbergs, upper pediments, urban land, reserves, etc): ...... 3 315 km2(20 %) Available for cultivation and grazing: ...... 13 261 km2(80 %) Annually cropped area (roughly 10-15 % of total district area = 12-18 % of available land for cultivation and grazing): ...... l2 % = 1 590 km2(9.6 %) 18 % = 2 390 km2(14.4 %) Area left for grazing (82-88 % of available land) a) With 12 % of available land under crops ...... l l 660 km2(70 %) b) With 18 % of available land under crops ...... 10 860 km2(66 %) Total population (1967) in the district (rural) ...... 297 440 inh Average household size ...... 4.3 pers

Yields of subsistence and cash crops and number of standard households (5 persons) to subsist on arable farming (2000 kcalipersoniday from subsistence crops and a gross annual monetary income of 25001-per household) I. Area reyuzred for subszsterzc~crops (50 % of calorze requzrements provzded by each of two maan crops) Maize: 320 kcal1lOO g =228.1 kgiperslyear Aver. )-ieldlha = 700 kg. Area requiredistand. household: 0.82 ha

Millet: 320 kcal/ l00 g = 228.1 kgiperslyear Aver. yieldlha = 500 kg. Area requiredlstand. household: 1.14 ha Total area for subsistence crops: 1.96 haistand. household (0.39 halpers) Maximu~nnumber of standard households (hh) to subsist on mixed cultivation (no cash income) and livestock husbandry. a) With 12 of available land under crops: ...... 81,288 hh (406,440 inh) b) With 18 % of available land under crops: ...... 122,188 hh (610,940 inh)

Figures 'ijithin brackets indicate percent of' total district area. II. Area required for cash crops (cash income provided by two main crops) Groundnuts: 80 % of cash income (20001-). Producers price l/-per kg Crop required: 2000 kg. Average yield 500 kglha Area requiredlstand. household: 4.0 ha Castor: 20 % of cash income (5001-). Producers price 0.61-per kg Crop required 833 kg. Average yield 400 kglha Area requiredlstand. household: 2.08 ha Total area for cash crops: 6.08 halstandard household (1.22 halpers) Total cropping area requiredlstand. household (cash and subsistence crops) 8.04 ha (1.61 halpers.) Maximum number of standard households to subsist on mixed cultivation (with cash income). a) With 12 % of available land under crops: ...... 19,776 hh (98,880 inh) b) With 18 % of available land under crops: ...... 29,726 hh (148,630 inh)

Livestock carrying capacity and number of households to subsist on cattle herding Maximum number of LU in the district at a stocking rate of 2.5 ha/LU. (All available land used for grazing. No allowance for areas under crops): ...... 530,440 LU, a) U'ith allowance for 12 % under crops (grazing reduced)...... 466,400 LU b) With allowance for 18 % under crops (grazing reduced)...... 434,400 LU alt a): 5.7 LUIstand. household= 1.14 LUipers alt b): 3.6 LUIstand. household=0.72 LUipers Maximum number of LU in the district at a stoclung rate of 4.0 haiLU. (Only "permanent" grazing areas. Fields under crops excluded): a) With 12 9% under crops ...... 29 1,500 LU b) With 18 f under crops ...... 27 1.500 LU Number of standard households to subsist and earn cash income (2,5001- per annum) from pastoral activities (sale of cattle for meat). Requires a stocking density of 4 haiLU and a minimum household herd size of 67 LU which makes a minimum grazing area of 268 haistand. household. a) ...... 4,350 hh (21,750 inh) b) ...... 4,052 hh (20.260 inh)

Population carrying capacity I. Subsistence alternative Ratio3 Subsistence from mixed cultivation and phstoralism. No cash income. alt a) 406.440 in11 1.37 alt b) 610,940 inh 2.05 ' Relation bet!\-een estimated carrving capacity and 1965 district rural population. 11. Commercial alternative Ratio 1. Subsistence from agriculture. Income from sale of cash crops. Some livestock but not alt a) 98,880 inh 0.33 for commercial purposes. alt b)148,630 inh 0.50

2. Subsistence from livestock. Income from sale of livestock. No cultivation. alt a) 21,750 inh alt b) 20,260 inh 3. Dual economy. T~vocategories. Subsistence from agriculture. Income from sale of cash crops. Permanent cultivations (require use of fertilizers). No cattle. Subsistence from livestock. Income from sale of livestock. alt la+2a 120,640 inh 0.40 Permanent grazing areas. No alt Ibt2b 168,890 inh 0.57 cultivation. Note: The need of manure for the commercial alternative of farming has not been considered

Estimated population and livestock carrying capacity, Ikowa catchment

2 Catchment area: ...... 612 km (100 %)4 Area not suited for cultivation (inselbergs, upper pediments, mbugas, roads, etc): ...... 219.9 km2 (35.8 %) Area not suited for grazing (as above, less mbugas): ...... 178.7 km2 (29.2 %) Available for cultivation: ...... 392.9 km2 (64.2 %) Area possible to use for grazing (incl. mbugas): ...... 433.3 km2 (70.8 %) Total population (1967): ...... 21,900 inh Number of homesteads ...... 2,400 Number of households ...... 4,800 Average homestead size ...... 9.2 pers Average household size ...... 4.6 pers Annuall) cropped area (roughl~15 % of catchment or 23.3 % of area a.i ailahle for culti~ations) ...... 92 km2 (15 %) L4realeft for grazing (excl. mbugas) ...... 301.3 km2 (49 %) Area required for subsistence crops.

Maize: 320 kcalIlOO g = 0.625 kglperslday = 228.9 kglperslyear Amount required to feed whole catchm. pop.: 5,109.44 tons Average yield per ha = 700 kg Area needed to feed whole catchm. pop.: 7,299 ha (11.9 %)

Millet: 320 kcal/lOO g= 0.625 kglperslday = 228.1 kglperslyear Amount required to feed whole catchm. pop.: 5,109.44 tons Average yield per ha = 500 kg Area needed to feed whole catchm. pop.: 10,219 ha (16.7 %)

Sorghum: 300 kcalIlOO g = 0.667 kg/pers/day = 243.5 kglperslyear Amount to feed whole catchm. pop.: 5,454.40 tons Average yield per ha = 750 kg Area needed to feed whole catchm. pop.: 7,273 ha (11.9 %) 2. 50 % calorie requirements provided by each of two main crops. Maize and millet: Area needed to feed catchm. pop.: 8,759 ha (14.3 5%) The carrying capacity of the catchment with 15 % of the total area under crops is as follows:

Ratio Maize and millet: 5,333 households (24,532 inh) ...... 1.10 1.80 ha/household (0.75 ha maize + 1.05 ha millet) Millet only: 4,571 households (21,027 inh) ...... 0.94 2.10 halhousehold

Livestock carrying capacity Maximum number of LU in the catchment at a stocking rate of 2.5 ha1LU. (allowance made for 15 % of catchment under crops): ....12,050 LU Allowance for dry season grazing on harvested fields and dry mbugasadds ...... 2,600LU The total carrying capacity on unimproved grazing in the catchment is 14,650 LU With a population carrying capacity of 2 1,000-24,500 this implies a maximum of 0.6-0.7 LUIpers (2.8-3.2 LU/household, 5.5-6.4 LUIhomestead). Acknowledgements

I am deeply indebted to Professor An- Hultblad for typing draft versions as ders Rapp of the Physical Geography well as the final version of the manu- Department, University of Lund, who script, to Miss Ylva Brolin and to Miss inspired me to engage in this research Annika Pettersson for secretarial work. and who actively helped me to complete Thanks are also due to Mr Are Are- the study. It is doubtful whether this schoug for assistance in photogrammet- task could have been achieved without ric work, to Mr Wolter Arnberg for his constant encouragement. He read working out computer programmes for drafts as well as the final version of the calculation of sediment and reservoir manuscript and contributed with much volumes, to Mr Karl-Axe1 Edlund and valuable advice and many important Mr Mats Akerblom for analysis of soil comments. and sediment samples. The author also I am much indebted to Professor gratefully acknowledges the assistance Gunnar Hoppe, former Head of the of Mr Johan Kleman and Mr Jan Department of Physical Geography, Onnerby for photo and printing work. University of Stockholm who encour- Further, thanks are due to Mr Samuel aged me in my research and helped in Mutiso, University of Dar es Salaam, Mr finding solutions to financial and practi- Severini Theodori, Mgeta och Mr Rolf cal problems in the preparatory stages Heyman, Stockholm, who assisted me in of this volume. I am also much indebted the field. to Professor Lars-Erik Ase, Department I much appreciate the helpful assist- of Physical Geography, University of ance of Mrs Ingrid Lindau and Miss Stockholm, who read the text and con- Birgit Lindgird, former and present tributed very valuable suggestions based Librarians of the Geographical Library, on his extensive experience of the East University of Stockholm, and the staff African environment, and to Miss Mar- of the Library of the Geological Survey gareta Elg who also read the text and Division, Dodoma, the Libraries of the made many useful comments. University of Dar es Salaam and the East Further, I particularly want to men- African Meterologcal Department, tion Mr Hammond Murray-Rust with Nairobi and Dar es Salaam. whom I have had innumerable interest- Professor Len Berry and Professor ing discussions, especially during the Adolfo Mascarenhas, former and pres- many weeks that we spent together in ent Directors of BRALUP, helped me the field in Ugogo. He collected and with transport and office facilities and analysed some of the data on which this arranged for research clearance in Tan- volume is based. zania. Mr Nigel Rollison, Uppsala, most I am ver) grateful to Mrs Birgt Hans- kindly helped me with the linguistic re- son for skilfully draxving the maps and vision. diagrams and for suggesting improve- I also {$.ant to acknowledge helpful ments to the design of a number of fig- discussions over specific points with Mr ures. I am also grateful to Miss Eivor Lars Asker, formerly WD 8s ID, Dodo- Granbom for help with the drawing. ma, Dr Hartmut Kuschel, formerly of Thanks are due to bliss Gertrud the Dodoma Region Water Master Plan Team, TMWD, Dodoma who also Fund and The Andree Fund of The helped in sediment sampling and su- Swedish Society for Anthropology and pervising of gauging equipment, Dr Lill Geography, The P. E. Lindahl Founda- Lundgren, University of Stockholm, tion, The Faculty of Mathematics and Professor Lennart Stromquist, Universi- Natural Science of The University of ty of Uppsala, and Professor Paul Stockholm, The Axe1 Lagrelius' Fund, Temple, formerly of the University of The Secretariat for International Ecolo- Dar es Salaam. gy (SIES), The Swedish International Finally I am indebted to my wife, Development Agency (SIDA), Lang- Monica, who in the early phase of the man's Cultural Fund and The C. F. studies assisted me in the field and who Liljewalch's Travel Grants. later has supported me in many ways. I am particularly grateful to the Di- Without the kind assistance of all the rector of The Scandinavian Institute of people mentioned above as well as all African Studies, Professor Car1 Gijsta the other colleagues and friends who Widstrand, for his interest and support. must remain unmentioned here, this vo- Apart from generous financial assist- lume could not have been produced. ance he kindly offered me an opportu- Financial support for this study has nity to work with the Institute for two been provided by the following funds months in early 1978, which was of and institutions: The Scandinavian In- great value for the work with the pre- stitute of African Studies, The Bank of sent volume. Thanks are also due to the Sweden Tercentenary Fund, The Vega Director for his kindness in publishing Fund, The J. A. Wahlberg Memorial the work in the Institute's series. References

Acland, J. D., 197 1: East African crops. ,4n Behr, H. F. von, 1891: Kriegesbilder aus introduction to the production of field dem Araheraufstand in Deutsch Ostaf- and plantation crops in Kenya, Tanzania rika. Brockhaus. Leipzig. 343 p. and Uganda. FAO/Longmans. London. Benndorf, J. G., 1918: Der koloniale Ver- 252 p. kehr Deutsch-Ostafrikas. Thomas & Hu- Agr. D., 1944: Village econonlic survey. bert, M'eida in Thur. 126 p. Tang. Terr., Agric. Dept. Dar es Salaam. Berry, L. (ed.), 1971a: Tanzania in maps. Allan, W., 1963: The African husbandman. University of London Press, London. 172 Oliver & Boyd, Edinburgh and New York. P. 505 p. Berry, L. 197 1 b: Dodoma population den- Alpers, E. A., 1969: The coast and the de- sity. Res. Rep. No 42. BRALUP, Univ. of velopment of the caravan trade. In: A Dar es Salaam. 4 p. history of Tanzania. U.N. Kimambo & A. Berry, L. & Kates, R. U'., 1970: Planned J. Temu, eds.) E. A. Publishing House, irrigation settlement: a study of four vil- Nairobi. 34-56. lages in Dodoma and Singida regions. Res. Annual Reports: Agric. Dept (Agr. D.) Tan- Paper No 10. BRALUP, Lniv. of Dar es ganyika Territory, Government Printer, Salaarn. 63 p. Dar es Salaam. Berry, L. & Townshend, J., 1972: Soil con- Annual Reports: Geol. Surv. Div. (GSD), servation policies in the semi-arid regions Tang. Terr., Govt. Printer, Dar es Salaam. of Tanzania, a historical perspective. A111lualReports: Public Works Dept. (PWD). Geogr. Ann., Vol 54 A 134). 241-253. Tang. Terr., Govt. Printer, Dar es Salaam. Bin hfuhammed, H. (Tippu Tip). 1902-03: Annual Reports: Veterinary Sci. and Anirn. Maisha )-a Hamed bin hluhammed. Husb. Dept. (VSD), Tang. Terr., Govt. Mitteilungen des Seminars fur Orienta- Printer, Dar es Salaam. lische Sprachen. 1902-03. Annual Reports: Water Dev. Dept. (WDD), Bishop. M'. W., 1962: Gully erosion in Tang. Terr., Govt. Printer, Dar es Salaam. Queen Elizabeth National Park. Uganda Annual Reports: Water Dev. and Irrigatio11 ,Journal, Vol 26. 161-165. Dil. (WD & ID), Tanz. Ministry of Lands. Settlement and Water, Govt. Printer. Dar Bogdan, A. V. & Pratt, D. J., 1967: Reseed- ing denuded pastoral land in Kenya. es Salaam. Kenya Ministry of Agric. and Anim. Husb. Backt.us. I.. 1976: Ecological studies of' the Nairobi. 48 p. vegetation on marshy shores. In: Ecolog- Boserup, E., 1965: The conditions of ical studies of the hltera Basin. (D. agricultural growth: the economics of Johansson, ed.) TANESCOISIZ'ECO. agrarian change under population Stockholm. pressure. Allen & Un~vin,London. 124 p. Baker, S. J. K., 1977: A background to the Bondestam, L., 1973: Some notes on African study of drought in East Africa. In: statistics-collection. reliability, and in- Drought in Africa (D. Dalby, R. J. terpretation. Res. Rep. No 18. Scandina- Harrison-Church & F. Bezzaz, eds.) Afri- vian Institute of African Studies. Uppsala. can Environment. Special Rept. 6. Inter- 50 p. national African Institute, London. 74-82. BRALUP. 1971: Economic report of Baumann, O., 1894: Durch Massailand zur Docloma region. Res. Rep. No 30. Nilquelle. Dietrich Reimer. Berlin. 383 p. BRALUP. Univ. of Dar es Salaam. 43 p. Baver. L. D., Gardner, \V. H. & Gardner, I\:. R.. 1972: Soil physics. IZ'iley, New York. Brandstrijm, P. 1976: Livestock clevelop- 498 p. ment in hl~vanza Region. Mwanra In- Becker. J.. 1887: La vie en .ifrique. (2 1-01s) tegrated Regional Planning Project, Vol 3. L. Ixbeque. Paris and Brussells. Stockholm. 1-71. Becker. hI.. i899: Ueber Bahnbau i11 Brooke, C;.. 1967: The heritage of farnine in Deursch Ostafrika. Ikutsches (;entral Tanzania. Tanz. Notes anti Rec., Jiolonialhlatt. X-01 10. 760-63. No 67. 13-22. Brown, C. B., 1950: Sediment transporta- vols). Gebr. Buchner, Bamberg. 340 + 365 tion. In: Engineering hydraulics (H. P. Rouse, ed.), Wiley, New York. 826-832. Christiansson, C., 1970: Studies of three wa- Bryceson, D. N. M,, 1967: Planning for agri- ter reservoirs in Dodoma district, central culture. Conference Papers, Conference Tanzania. (Unpubl. rep.) BRALUP, Univ. on Agricultural Planning in East Africa. E. of Dar es Salaam. 10 p. A. Publishing House, Nairobi. 1-8. Christiansson, C., 1972: Notes on morphol- Bryceson, D. N. M., 1971: Land use plan- ogy and soil erosion in Kondoa and Sin- ning in Tanzania-looking to the future. gida districts, central Tanzania. Geogr. Second Conf. of Land Use in Tanzania. Ann. Vol 54 A (34). 319-24. Faculty of Agriculture and Bureau of Re- Christiansson, C., 1974: Soil erosion and source Assessment and Land Use Plan- sedimentation in some selected catch- ning, Univ. of Dar es Salaam. 6 p. ments in the semi-arid parts of Tanzania. Bujra, A. S., 1973: A socio-economic devel- Rapp. 34. UNGI, Uppsala. 447462. opment plan for Mundemu village. (Un- Christiansson, C., 1978: Relations between publ. report) Univ. of Dar es Salaam. heavy grazing, cultivation, soil erosion and Bujra, A. S., 1974: Ujamaa rural develop- sedimentation in the semi-arid parts of ment and environment. The experience central Tanzania. Proc. First International from Dodoma, Tanzania. Pap. No 43. Rangeland Congr., Denver. 274-278. IDEP-KENDEV: Workshop on environ- Christiansson, C., 1979: Imagi dam-a study ment and rural development in East of soil erosion, reservoir sedimentation Africa, Nairobi. 16 p. and water supply at Dodoma, central Bull, E., 1973: Ostafrikansk bakgrunn. Tanzania. Geogr. Ann. Vol 61 A (34). Gyldendal Norsk Forlag AIS, Oslo. 334 p. 113-146. Burton, R. F., 1860: The Lake Regions of Claus, H., 191 1: Die Wagogo. Baessler Central Africa (2 vols). Longmans Green, Archiv, Berlin. 1-72. London. 412 + 468 p. Cliffe, L. R., 1970: Nationalism and the Burtt, B. D., 1942: Some East African veg- reaction to enforced agricultural change etation communities. J. Ecol. Vol 30. in Tanganyika during the colonial period. 65-146. Taamuli No. 1, Dept. of Political Science, Busse, W., 1908a: Deutsch Ostafrika I. Univ. of Dar es Salaam. 1-1 1. Zentrales Steppengebiet. In: Veg- Conyers, D., 1971: Agro-economic zones of etationsbilder, Bd 5 (7). Karsten & Schenk, Dodoma and Singida. Res. Rep. No 47. Jena. Tafel 4045. BRALUP, Univ. of Dar es Salaam. 75 p. Busse, W., 1908b. Die periodischen Gras- Cook, A., 1974: The use of photo-interpre- brande im tropischen Afrika, ihr Einfluss cation in the assessment of physical and auf die Vegetation und ihre Bedeutung biological resources in Tanzania. Res. Pa- fur die Landeskultur. Mitt a.d. Deutsch. per No 31. BRALUP, Univ. of Dar es Schutzgeb. Bd 2 1 (2), Berlin. 113-139. Salaam. 59 p. Calton, W. E., 1954a: An experimental soil Coster, F. M., 1959: Underground water in nlap of Tanganyika. 2nd Inter. Afr. Soils Tanganyika. (Lnpubl. rep.) Ministry of Conf., Leopoldville, 237-240. Lands and Surveys, Govt. Printer. Dar es Calton, W'. E., 1954b: The catena in relation Salaam. 97 p. to the classification of East African soils. Dahl, G. & Hjort, A., 1976: Having Herds. Proc, 5th Int. Congr. Soil Sci. Vol 4. Pastoral herd growth and household 38-62. economy. Dept. of Social Anthropology, Calton, W. E., 1963: Some data on a Tan- Univ. of Stockl-iolm. 335 p. ganyika catena. E. Afr. Agric. J. Vol 29. Daily News, 30 Nov. 1977. 173-177. Daily Nem, 3 1 Oct. 1978. Cameron, V. L.. 1877: Across Africa (2 vols). Dantz, A.. 1902: Die Reisen des Dalby, Isbister & Co, London. 389 + 366 Bergassessors Dr Dantz in Deutsch-Ostaf- P. rika in den Jahren 1898, 1899, 1900. 2. CDA, 1975: A portrait of Docloma. Capital Von Kilossa bis Mpapua und Ausfluge in Develop~nentAuthority, National Printing die Umgebung Mpapuas. 3. Von Mpapua Company. Dar es Salaam. 55 p. bis Tabora. Mitt. a. d. Deutsch. Carnell, MT. L., 1955: Sympathetic rnagic Schutzgeb., Bd 15 (2),Berlin. 49-63. among the Gogo of Mpwapwa District. Dasnlann, R. F., Milton, J. P. & Freeman, P. Tang. Notes and Rec. No 39. H., 1973: Ecological principles for Casati, C., 1891: Zehn Jahre in ~~uatoria econornic developnlent. John Wiley & und die Ruckkehr mit Emin Pascha (2 Sons, London. 252 p. Deshler, W., 1960: Livestock trypano- fotoelektrisk ytmatningsmetod. Forskn. somiasis and human settlement in north- rapport 1, Naturgeografiska inst., Stock- ern Uganda. Geographical Review, Vol holms univ. 13 p. 50. 541-554. Elwell, H. A. & Stocking, M. A., 1974: Deutsches Kolonialblatt, 1895: Kolo- Rainfall parameters and a cover model to nial-Abteilung des Auswartigen Amts. predict runoff and soil loss from grazing Mittler und Sohn, Berlin. trials in the Rhodesian sandveld. Proc. D'Hoore, J. L., 1965: Soil map of Africa. Grassl. Soc. S. Afr., Vol 9. 157-163. 1:5,000,000. Explanatory Monograph, C. Elwell, H. A. & Stocking, M. A., 1975: Para- C. T. A. Joint Project Publication No 95., meters for estimating annual runoff and Lagos. 205 p. soil loss from agricultural lands in Dodgshun, A. W., 1878: Diary notes. Rhodesia. Water Resources Research, Vol Quoted in: TMA, 1977: The threat of de- 11 (4). 601-605. sertification in central Tanzania. Tanzania Epstein, E., Grant, W. J. & Struchtmeyer, R. Ministry of Agriculture. A., 1966: Effect of stones on runoff, ero- Dodoma District Book. sion and soil moisture. Proc. Soil Sci. Soc. Douglas-Hamilton, I. & O., 1975: Among Am., Vol 30 (5). 638-640. the elephants. Wytham Publications Ltd., Evans, R., 1967: On the use of welding rods London. 260 p. for erosion and deposition pins. In: Field DSAS, 1900: Die Deutschen Schutzgebiete in methods for the study of slope and fluvial Afrika und der Sudsee. Denkschrift uber processes. Rev. Gkomorph. dyn. 42. p. die Entwicklung. Berichtsjahr 1899/1900. 165. DSAS, 1902: Die Deutschen Schutzgebiete in Afrika und der Sudsee. Denkschrift uber FAO, 1965: Soil erosion by water. Some die Entwicklung. Berichtsjahr 190 111902. measures for its control on cultivated DSAS, 1914: Die Deutschen Schutzgebiete in lands. FAO Agricultural Development Afrika und der Sudsee. Vervaltungsstatis- Paper, 8 1. Rome. 284 p. tik 191211913. Berlin. FAO, 1967: East African Livestock Survey (2 Dudal, R., 1965: Dark clay soils of tropical vols). FAO, Rome. and subtropical regions. FAO, Rome. 16 1 FAO/UNESCO, 1977: Soil map of the world. Vol 6, Africa. UNESCO, Paris. 299 P. Dunne, T., 1973: Studies of soil erosion and P. soil degradation in Kenya. (Unpubl. rep.) Fawley, A. P,, 1954: Mkonze dam site, Dept. Geol. Sci. and Quaternary Res. Dodoma. (Unpubl. rep.) File 1963, AF127, Centre, Univ. of Washington. 12 p. Geol. Surv. Div.. Dodoma. 28 p. Dunne, T., 1976: Studying patterns of soil Fawley, A. P., 1956a: Water resources of erosion in Kenya. (Unpubl. rep.) Dept. Dodoma and vicinity. Rec. Geol. Surv. Geol. Sci. and Quaternary Res. Centre, Tang., Vol 3., Govt. Printer, Dar es Univ. of Washington. 28 p. Salaam. 62-70. Dunne, T., Dietrich, W. E. & Brunengo, M. Fawley, A. P., 1956b: Msalatu reservoir, J., 1978: Recent and past erosion rates in Dodoma. Rec. Geol. Surv. Tang. Vol 3., semi-arid Kenya. Zeitschrift fur Geo- Govt. Printer, Dar es Salaam. 71-82. morph. NF. Suppl. Bd 29. 130-140. Fawley, A. P., 1958a: Geology and under- EAMD, 1922-1929: Climatological statistics. ground water resources of the Makuta- Tanzania. E. A. Met. Dept., Nairobi. pora depression. Rec. Geol. Surv. Tang., EAMD, 1963: The weather of East Africa. Vol 5., Govt. Printer, Dar es Salaam. Pamphlet No 7. E. A. Met. Dept., Nairobi. 41-52. 10 p. Fawley, A. P., l958b: Evaporation rate at EAMD, 1966: Monthly and annual rainfall Dodoma, Tanganyika. Rec. Geol. Surv. in Tanganyika and Zanzibar during the 30 Tang., Vol 5., Govt. Printer, Dar es years 1931-1960. E. A. Met. Dept., Nai- Salaam. 37-39. robi. Fawley, A. P., 1959: Diamond drilling at EAMD, 1975: Climatological statistics for Bahi swamp. Rec. Geol. Surv. Tang. Vol East Africa. Part 111. Tanzania, E. A. Met. 6., Govt. Prlnter, Dar es Salaam. 47-51, Dept., Nairobi. 92 p. Fonck, H., 1894: Bericht uber nleinen Egero, B. & Henin, R. A. (eds), 1973: The Marsch Mpwapwa-UgogcAJssandawe- population of Tanzania. Census Volume Irangi-Uasst-Ufiome-Umbugwe-Mangati- 6. BRALUP & Bureau of Statistics, Dar es Irangi-Burung-Ugogdvfp~rap~~a.Mitt. Salaam. 292 p. a.d. Deutsch. Schutzgeb., Bd 7 (4), Berlin. Eknlan, S. R. & Wastenson, L., 1968: En 291-296. Fonck, H., 1910: Deutsch-Ostafrika. Eine Handley, J. R. F., 1952: The foundation Schilderung deutscher Tropen nach 10 conditions existing at the proposed Wanderjahren. Vossische Buchhandlung, Mkonze dam site, No 3 dam, Dodonia, and Berlin. 612 p. general considerations on water supply in Ford, J., 1971: The role of the the immediate vicinity of Dodoma. (Un- trypanosomiasis in African ecology. A publ. rep.) File 2325-2326 JRFHl16. Geol. study of the tsetse-fly problem. Clarendon Surv. Div.. Dodoma. 9 p. Press, Oxford. 368 p. Hankins. T. D., 1974: Response to drought Fosbrooke, H. A., 1974: Socio-economic in Sukumaland, Tanzania. In: Natural aspects of shifting cultivation. Soils Bulle- hazards. Local, Global, National. (G. F. tin 24. SIDAIFAO, Rome. 72-77. White, ed). Oxf. Cniv. Press, New York, Fuchs, P. & Hauter, A., 1907: Wirtschaft- London and Toronto. 98-104. liche Eisenbahn-Erkundungen im mitt- Harrison, E., 1937: Soil erosion, a leren und nordlichen Deutsch-Ostafrika. memorandum. Govt. of Tanganyika Gerlach, T., 1967: Hillslope troughs for Terr., Crown Agents, London. 1-22. measuring sediment movement. Rev. Hartnoll, A. V., 1942: Praying for rain in Geomorph. dyn. Vol 17 (4), p. 173. Ugogo. Tang. Notes and Rec. No 13. Gillman, C., 1930: Notes on soil erosion in 5960. East Africa. (Unpubl. rept.) Gillman Pa- Heady, H. F., 1960: Range management in pers, Hans Cor) Collection, Univ. of Dar East Africa. Kenya Dept. of Agric. Gov- es Salaam. 4 p. ernment Printer. Nairobi. 125 p. Gillman, C., 1936: A population map of Heidke, P,, 1903-14: Meteorologische Beo- Tanganyika Territory. Geographical Re- bachtungen aus Deutsch Ost-Afrika. Mitt. view, Vol 26 (3). 353-375. a.d. Deutsch. Schutzgeb.. Bd 18-27, Glover, P. E., 1968: The role of fire and Berlin. other influences on the savanna habitat, Heidke, P.. 1923: Das Klima. In: Das nlth suggestions for future research. E. abflusslose Rumpfschollenland im nord- Afr. \Vildl. J.. Vol 6. 131-137. ostlichen Deutsch-Ostafrika. Teil 11. (E. Glymph, L. M., 1951: Relation of sedi~nen- Obst, ed), Mitt. d. Geogr. Gesellschaft in tation to accelerated erosion in the Mis- Hamburg, Bd 35, Hamburg. 90-128. souri River Basin. US Soil Conservation Herrmann, R.. 1892: Ugogo, das Land und Service. Technical Paper No 102. seine Be~vohner. Mitt. a. d. Deutsch. Glymph, L. M., 1957: Importance of sheet Schutzgeb.. Bci .5 (4). Berlin. 19 1-203. erosion as a source of sediment. Trans. Herrmann, R., 1893: Die Handelsverhalt- Am. Geophys. Un., Vol38. 903-907. nisse am Victoria-Nyansa. Deutsches KO- Gotzen, G. A., Ifon,1899: Durch Afrika von lonialblatt, Vol 4. 8 1-82. Ost nach West. Resultate und Begebenhei- Hesse, P. R., 1955: Some facts and fallacies ten einer Reise v011 der deutsch-ostafri- about termite mounds. Tang. Notes and kanischen Kiiste bis zur Kongomiindung Rec. No 39. 16-25. in den Jahren 1893194. D. Reirner, Berlin. Hore, E. C.. 1892: Tanganyika: Ele\.en years 426 p. in Central Africa. E. Stanford. 1,ondon. Grantham, D. R., 1933: Soil erosion: Discus- 306 p. sion. Geogr. J., Vol 82 (1). 146-150. Hornbv, H. E. & Hornby, R. M., 1943: .A Griffiths. J. F., 1972: Climate. In: East co11t;ibution to the study of the vegetation Africa: its peoples and resources. (\V. T. of M~M-apwa.Tang. Notes and Rec. No M:. Morgan, ed.) Oxford Univ. Press, Nai- 15. 2548. robi, London and New York. 107-1 18. Horst. L.. 1964: Kinyasungice-31kkondoa Grove, A. T., 1971: Africa, south of the Sah- river basin flood control. Report on a pre- ara. Oxford Univ. Press. London. 279 p. liminary survey. (Unpubl. rep.) IVD & ID. Gtvassa, G. C. K.. 1969: The German in- Dar- es Salaam. ter\-ention and African resistance in Hudson. N. \lr.,1971: Soil conservation. B. Tanzania. In: .A histor!- of Tanzania. T. Batsford Ltd. London. 320 pp. (Kimambo. I. N. & Temu. A. J., eds). East Hudson, N. LV. & Jackson. D. C.. 1939: Re- African Publishing House, Nairobi. sults achieved in the measurement of era- 85-122. sion and runoff in Southern Rhodesia. 3rd Inter-Afr. Soils Conf.. Ilalaba. ,575-583. Hamad, S., 1976: Izazi ward. In: Ecological Hughcs, P. J.. Hope. G.. Latharn, hl. & studies of the Mtera basin. (D. ,Johansson, Brookfielcl. M,.1979: Prehistoric man-in- ed). T'ANESCO/SLVECO, Stockholm. duced degradation of the Lakeba land- 135-137. scape: Evidence from rlvo inland sltamps. UNESCOIUNFPA. Fiji island reports No hillslope erosion. Rev. Geomorph. dyn. v01 <5.95 -1 10. 17 (4). 165-167. Jackson, I. J., 1968: Rainfall stations in Kjekshus, H., 1977: Ecology control and Tanzania. Res. Notes No 5a. BRALUP, economic development in East African Univ. of Dar es Salaam. 79 p. history. The case of Tanganyika Jackson, I. J., l971 : Atmospheric pressure 1850-1 950. Heinemann, London, Ibadan, and winds; Rainfall; E~raporation;Other Nairobi and Lusaka. 2 15 p. climatic elements. In: Tanzania in Maps. Kremer, E., 1910: Die unperiodischen (L. Berry, ed). Univ. of London Press, Schwankungen der Niederschlage und die London. 3441. Hungersnote in Deutsch-Ost-Afrika. Jager, F., 19 13: Das Hochland der Riesenk- Archiv der deutschen Seewarte, Ham- rater und die umliegenden Hochlander burg. 64 p. Deutsch-Ostafrikas. Mitt. a.d. Deutsch. Krokfors, C., 1978: Utvecklingsstrategier, Schutzgeb. Erganzungsheft Nr 8. 213 p. folkligt deltagande och rumsuppfattning: James, T. C., 1950: Geological recon- exempel frin Tanzania och Somalia. naissance of the Dodoma dam sites with Terra, Vol 90 (4). 200-207. special reference to the proposed Mkonze Kuczynski, R. R., 1949: Demographic dam. (Unpubl. rep.) File TCJI6. Geol. Survey of the British Colonial Empire (2 Surv. Div., Dodoma. 6 p. vols). Oxford Univ. Press, London. Johansson, D., 1976: Vegetation of the Mt- 821+983 p. era basin, with reference to the impound- Langbein, W. D. & Schumm, S. A., 1958: ment. In: Ecological studies of the Mtera Yield of' sediment in relation to mean basin (D. Johansson, ed), TANESCOI annual precipitation. Trans. Am. Geop- SM'ECO, Stockholm. 55-76. hys. Un., Vol 39 (6). 1076-1085. Johansson, D. 8s Stromquist. L., 1977: Leopold, L. B. & Emmett, W. W., 1967: On Termite mounds near Docloma, central the design of a Gerlach trough. Rev. Geo- Tanzania. Medd. fr. Uppsala Univ. Geogr. morph. dyn. Vol 17 (4). p. 172. Inst. Ser. A Nr 272. 102-115. Leopold, L. B., Emmett, M'. W. & Myrick, R. Johnson, D. J., 1962: Rainfall in East Africa, %I., 1966: Channel and hillslope processes Quarterly J. Roy. Met. Soc., Vol 88. 1-1 9. in a semi-arid area, New Mexico. Prof. Pa- Jonsson, U,. 1976: Planning for village food per. 352-G. Geol. Surv., Washington. production. (Unpubl. draft version). 193-253. Tanzania Food and Nutrition Center, Lettow-Vorbeck, P. von, 1920: My re- Dept. of Planning, Dar es Salaam. 7 p. miniscences of East Africa. Hurst and Junker, W.. 1891: Reisen in Afrika 1875-86. Blackett. London. Bd. 3 (1882-86). Eduarcl Hiilzel, Wien Liljequist, G. H., 1970: Klimatologi. GLA, und Olmiitz. 740 p. Stockholm. 327 p. Kannenberg, Hauptm., 1900a: Durch die Lind, E. M. 8s Morrison, M. E. S., 1974: East Marenga hlakali. hlitt. ad. Deutsch. African vegetation. Longman, London. Schutzgeb., Bd 13 (1). Berlin. 3-1 7. 257 p. Kannenberg. Hauptm., 1900b: Reise durch Manka, T. M., Mutembei, L. W. & Ndaki, A. die hamitischen Sprachgebicte um Kon- M. S.. 1974: Socio-economic studies, doa. Mitt. a.d. deutsch. Schutzgeb., Bd 13 Ikowa catchment. Tanz. Ministry of Water (2). 144-172. Del-. and Power, Research and Training Kiepert, R., 1896: Begleitworte zur Karte Div., Dodoma. 8 p. "Neue Aufnahmen deutscher Offiziere in Manshard, W., 1974: Tropical Agriculture. Lssagara. Ugogo, Uhehe und Mahenge". Longman, London. 226 p. Mitt. a.d. Deutsch. Schutzgeb., Bd 9 (1). hlascarenhas, A. C., 1967: Aspects of food Berlin. 4348. shortages in Tanganyika (1925-1945). U. Kimambo. I. N., 1969: The interior before E. ,4.. Social Sci. Conf., Nairobi. 14 p. 1800. In: a history of' Tanzania (I. N. Mascarenhas, 4. C., 197 l : Agricultural Kimambo & A. J. Temu, eds). East African vermin in Tanzania. In: Studies in East Publishing House, Nairobi. 14-33. African geography and development. (S. King. A. J.. 1953: Quarter degree sheet H. Omincle, ed), Heinemann, London, Llp~\-ap~vasouth B 37'311. Tang. Terr. Nairobi and Ibadan. 259-267. Geol. Surv. Di v.. Dodoma. Mascarenhas, A. C., 1974: Food production, King. L. C.. 1967: Llorphology of the earth. the total cnaironment and ruial develop- Oliver &. Boycl. Edinburgh and London. ment. Pap. No 12, IDEP-KENDEV: 726 p. IVorkshop on environment and rural de- King. S.,]. & Hadley. R. F., 1967: Illeasuring velopment in East Africa. Nairobi. 32 p. Mathew, A., 1974: Vegetation of' Dodoma sedimentation in Kisongo catchment, region. (Unpubl. rep.) Dept. of Biology, Arusha region, Tanzania. (Unpubl. M. A. Mazengo Sec. School, Dodoma. 16 p. Thesis), Dept. of Geography, Univ. of Dar Menne, T. C., 1959: A review of work done es Salaam. 201 p. in the Union of S. A. on the measurement Murray-Rust, D. H., 1972. Soil erosion and of runoff and erosion. Third Inter Afr. reservoir sedimentation in a grazing area Soils Conf., Dalaba. 612427. west of Arusha, northern Tanzania. Meyer, H., 1909: Das Deutsche Geogr. Ann., Vol 54 A (34). 325-343. Kolonialreich. Eine Landerkunde der Ndorobo, B., 1973: Famine problems in Deutschen Schutzgebiete. Bd I. Bibliogra- Dodoma district. In: Studies in famines phischen Institut, Leipzig und Wien. 650 and food shortages. (A. C. Mascarenhas, P. ed), Journal of The Geographical Associ- Meyer, 0. E., 1915: Die Bruche von ation of Tanzania, No 8. 2647. Deutsch-Ostafrika besonders der Land- Neal, J. H., 1937: The effect of degree of schaft Ugogo. Neues Jahrbuch fur Mine- slope and rainfall characteristics on runoff ralogie, Geologie und Palaontologie, and soil erosion. Missouri Agr. Exp. Sta. Beilagebd 38, Stuttgart. 806-881. Res. Bull. 280. Milne, G., 1935: Some suggested units of Ngaula, M,, 1976a: Mtera village. In: Ecolog- classification and mapping, particularly ical studies of the Mtera basin (D. for East African soils. Soil Research, Vol4. Johansson, ed). TANESCOISWECO, 183-198. Stockholm. 130-13 1. Milne, G., 1936: A provisional soil map of Ngaula, M., 197613: Muhigiti and Kingali vil- East Africa. Amani Memoirs, No 3 1. lages. In: Ecological studies of the Mtera Milne, G., 1947: A soil reconnaissance basin (D. Johansson, ed). TANESCOI journey through parts of Tanganyika SWECO, Stockholm. 131-132. Territory, December 1935 to February Nieuwolt, S., 1973: Rainfall and evaporation 1936. J. Ecol., Vol 35. 192-265. in Tanzania. Res. Paper No 24, BRALUP, Mkunduge, G. L., 1973: The Ukaguru envi- Univ. of Dar es Salaam. 46 p. ronment: Traditional and recent re- Nyange, A., 1976: Ugogoland. In: Ecological sponses to food shortages. In: Studies in studies of the Mtera basin (D. Johansson, famines and food shortages. (A. C. ed). TANESCOISWECO, Stockholm. Mascarenhas, ed). Journal of The 134-135. Geographical Association of Tanzania, No Nyerere, J. K., 1967: Socialism and rural 8. 63-85. development. Government Printer, Dar es Mnyawonga, M., 1973: The rainlore of the Salaam. 3 1 p. Wagogo. In: Studies in famines and food Obst, E., 1923: Das abflusslose shortages. (A. C. Mascarenhas, ed). Jour- Rumpfschollenland in nordostlichen nal of The Geographical Association of Deutsch-Ostafrika. Teil 2. Mitt. d. Geogr. Tanzania, No 8. 13-25. Gesellschaft in Hamburg, Bd 35. 330 p. Moore, J. E., 1971a: Soils. In: Tanzania in Oliver, R., 1967: Discernible developments maps. (L. Berry, ed). Univ. of London in the interior, c. 1500-1840. In: History Press, London. 28-29. of East Africa. The early period. (R. Oliver Moore, J. E., 1971b: Rural population carry- & G. Mathew, eds). Oxford Univ. Press, ing capacity for the districts of Tanzania. Nairobi. 169-2 11. Res. Paper No 18. BRALUP, Univ, of Dar Owako, F. N., 1971: Machakos land and es Salaam. 46 p. population problems. In: Studies in East Morgan, W. B., 1969: Peasant agriculture in African geography and development. (S. tropical Africa. In: Environment and land H. Ominde, ed). Heinemann, London, use in Africa (M. F. Thomas & G. W. Nairobi, Ibadan. 177-192. Whittington, eds). Methuen, London. Panburst, R., 1961: The Great Ethiopian 241-72. Famine. Univ. of London Review. Morgan, W. T. W., 1973: East Africa. Pant, P. S. & Rwandusya, E. M,, 1971: Cli- Longmans, London. 410 p. mates of East Africa. Techn. Mem. No 18. Moss, R. P,, 1969: The ecological E. A. Met. Dept., Nairobi. 13 p. background to land use studies in tropical Patton, M., 1971: Dodoma region, Africa, with special reference to the West. 1929-1959: A history of famine. Res. Rep. In: Environment and land use in Africa. No 44, BRALUP, Univ. of Dar es Salaam. (M. F. Thomas & G. W. Whittington, eds). 62 p. Methuen, London. 193-238. Peberdy, J. R., 1972: Rangelands. In: East Murray-Rust, D. H., 1970: Soil erosion and Africa: its peoples and resources (M:. T. W. Morgan, ed). Oxford Univ. Press, Nai- and sediment transport in the Morogoro robi, London and New York 153-176. river catchment, Tanzania. Geogr. Ann., Pereira, H. C., 1962: Grazing control in V0154 A (34). 125-155. semi-arid ranchland. The project. E. Afr. Rapp, A., Murray-Rust, D. H., Christ- Agric. J. Vol 27. Special Issue. 4246. lansson, C. & Berry, L., 1972: Soil erosion Peters, C. 1891: New Light in Dark Africa. and sedimentation in four catchments Ward, Lock and Co. London, New York near Dodoma, Tanzania. Geogr. Ann., and Melbourne. 597 p. Vol 54 A (34). 255-3 18. Phillips, J. F. V., 1929: Some important Rapp, A. & Hellden, U., 1979: Research on vegetation communities in the central environmental monitoring methods for province of Tanganyika Territory land use planning in African drylands. (Formerly German East Africa). South Rapp. och Notiser, No 42. Dept of Phys. African Journal of Science, Vol 26. Geography, Univ. of Lund. 116 p. 332-372. Reichard, P,, 1892: Deutsch Ost-Afrika, das Pratt, D. J. 1963: Reseeding denuded land in Land und seine Bewohner. Otto Spanner, Baringo district, Kenya. 1. E. Afr. Agric. J. Leipzig. 524 p. Vol 29 (1). 78-91. Rensburg, H. J. van, 1948: Run-off and soil Pratt, D. J., 1964: Reseeding denuded land erosion experiments. Ann. Rep. Dept. in Baringo district, Kenya. 2. Techniques Vet. Sci. Anim. Husb., Dar es Salaam. for dry alluvial sites. E. Afr. Agric. J. Vol 32-33. 29 (3). 243-260. Rensburg, H. J. van, 1953: Run-off and soil Pratt, D. J. & Knight, J., 1964: Reseeding erosion experiment. Ann. Rep. Dept. Vet. denuded land in Baringo district, Kenya. 3 Sci. Anim. Husb., Dar es Salaam. p. 39. . Techniques for capped Red Loam Soils. Rensburg, H. J. van, 1954: Run-off and soil E. Afr. Agric. J., Vol 30 (2). 117-125. erosion experiment. Ann. Rep. Dept. Vet. Pratt, D. J., Greenway, P. J. & Gwynne, M. Sci. Anim. Husb., Vol 11, Dar es Salaam, p. D., 1966: A classification of East African 28. rangeland. J. Appl. Ecol. Vol 3. 369-382. Rensburg, H. J, van, 1955: Run-off and soil Pratt, M. A. C., 1962: Relationship of erosion tests. MDW~DW~.Central Tan- run-off to rainfall. E. Afr. Agric. J. Vol27, ganyika, E. ~fr.'~Gic.J., Vol 20 (4). Special Issue. 73-74. 228-3 1. Prothero, R. M. (ed), 1972: People and land Rensburg, H. J. van, 1958: Gully utilization in Africa south of the Sahara. Oxford and erosion control. E. Afr. Agric. J., Vol Univ. Press, London, New York and To- 23 (3). 190-192. ronto. 344 p. Rigby, P,, 1969a: Pastoralism and prejudice: Pullan, R. A., 1975: Burning impact on Afri- Ideology and rural development in East can savannas. Geographical Magazine. Vol Africa. In: Society and social change in 47 (7). 432438. Eastern Africa. (R. J. Apthorpe & P. PM7D, 1958: Reservoir capacity tables (Un- Rigby, eds). Nkanga Editions No 4. pub].) Tang. Terr., Publ. Works Dept., Makerere Institute for Social Research, Dodoma. Kampala. 42-52. Rigby, P,, 1969b: Cattle and kinship among Quennel, A. M,,McKinlay, A. C. M. & Ait- the Gogo. A semi-pastoral society of cen- ken, W. G., 1956: Summary of the geology tral Tanzania. Cornell Univ. Press, Ithaca of Tanganyika. Part 1. Geol. Surv. Tang., Memoir No 1, Govt, Printer, Dar es and London. 355 p. Salaam. 264 p. Rigby, P,, 1977: Local participation in national politics: Ugogo, Tanzania. Africa, Rapp, A., 1974: A review of desertization in Journ. IAI, V0147 (1). 89-107. Africa-water, vegetation and man. Rep. Rodhe, H. & Virji, H., 1976: Trends and No 1, Secretariat for International Ecol- periodicities in East African rainfall data. ogy, Sweden (SIES),Stockholm. 77 p. Monthly Weather Review, Vol 104 (3). Rapp, A., 1975: Soil erosion and sedimenta- 307-3 15. tion in Tanzania and Lesotho. AMBIO, Roose, E. H. & Lelong, F., 1976: Les facteurs V014 (4),Royal Swed. Acad Sci. 154-163. de l'erosion hydrique en Afrique tropicale. Rapp, A., Berry, L. Temple, P. H., 1972: Etudes sur petites parcelles experimen- Soil erosion and sedimentation in tales de sol. Rev. de Geographie Physique Tanzania-the project. Geogr. Ann. Vol et de Gkologie Dynamique. Vol 18 (2). 54 A (34). 105-109. 365-374. Rapp, A., Axelsson, V., Berry, L. & Rounce, N. V., 1949: The agriculture of the Murray-Rust, D. H., 1972: Soil erosion cultivation steppe. Longmans, Green & Co, Cape Town, and Government Printer, Tanz. Min. Agric., Dar es Salaam. 58 p. Dar es Salaam. 105 p. Southon, E. J., 188 1: Notes on a ,journey Rounce, N. V., King, J. G. M. & Thornton. through Northern Ugogo in East Central D., 1942: A record of investigations and Africa. Proceedings of the Royal observations on the agriculture of the Geographical Society, Vol3. 547-553. cultivation steppe of Sukuma and Nyam- Speke, J. H.. 1863: Journal of the discovery wezi with suggestions as to the lines of of the source of the Nile. Blackwood, process. Pamphlet 30, Govt. Printer, Dar London. SW.transl.: C~~tacktenaf Xilens es Salaam. Rounce, N. V. & Thornton, D.. 1939: The ridge in native cultivation with special ref- Stallings, J. H.. 1957: Soil conservation. erence to the Mwanza district. E. Afi. Prentice Hall Inc., Engletvood Cliff's. 576 Agric. J., Vol4. 352-354. P. Rudengren, J., 1976: Ujamaa och bypolitik i Stanley. H. >l.. 1872: Hoxv I found Living- Tanzania. Internasjonal politikk 3 B. stone. Schribner, Armstrong & Co., New 757-770. York. Sampson Low, London. 736 p. Ruthenberg, H., 197 l : Farming systems in Stanley, H. M,, 1878: Through the dark the tropics. Oxford Univ. Press, London. continent (2 vols). Sa~npsonLow, London, 313 p. 522+566 p. Saggerson, E. P,, 1972: Geology. In: East Stanley, R. & Neame, A., 1961: The explo- Africa, its peoples and resources. (W. T. ration diaries of H. M. Stanley. W. Kimber W. Morgan, ed) Oxford Univ. Press, Nai- & Co. Ltd. London. 208 p. robi, London and New York. 67-94. Staples, R. R., 1934: Runoff and soil erosion Schaegelen, T., 1938: La tribu des Wagogo. tests. An11. Rep. Dept. Vet. Sci. Anim. Anthropos. Vol 33 (1). 195-217. (2). Husb. for 1933, Govt. Printer, Dar es 5 15-567. Salaam. 95-103. Schele, F. von, 1896: Uhehe. Mitt. a.d. Staples, R. R.. 1936: Runoff and soil erosion Deutsch. Schutzgeb. Bd 9.67-74. tests in semi-arid Tanganyika Territory, Schick, A. P,. 1967a: Suspended sediment .Inn. Rep. Dept. Vet. Sci. Anim. Husb. for sampler. Rev. Geomorph. dyn. Vol 17 (4). 1935, Govt. Printer, Dar es Salaam. 181-182. 134-142. Schick, A. P.. 1967b: On the construction of Staples, R. R., 1937a: Rotatio~lalgrazing. troughs. Rev. Geomorph. dyn. Vol 17 (4). Ann. Rep. Dept. Vet Sci. Anim. Husb. for p. 170-172. 1937, Govt. Printer, Dar es Salaam. 36-38. Schumm, S. A., 1967: Erosion measured by Staples. R. R.. 193%: Further studies on the stakes. Rev. Geomorph. dyn. Vol 17 (4). conservation of water supplies in semi-arid 161-162. East Africa: Some ctuantitative determina- Schumm, S. A. & Hadley, R. F., 196 1: Pro- tions on the percolation rate of the rainfall gress in the application of landform under different types of vegetation at analysis in studies of semi-arid erosion. Mp~vapwa. Ann. Rep., Dept. Vet. Sci. Geol. Surv. Circular 437, LVashington. ,4n1m. Husb. for- 1937. Govt. Printer. Dar Schynse, P. A., 1890: hIit Stanley unti Emin es Salaarn. 110 -1 32. Pascha dnrch Deutsch-Ost-Afrika, Reise- Staples. R. R.. 1939: Runoff and soil erosion tagebuch. K. Hespers & ,J. P. Bachem. rests in semi-arid Tanganyika. Third re- Koln. port. Ann. Rep.. Dept. Vet. Sci. Anim. Scott. R. 31.. 1962: The soils of Easr -4f'rica. Hush. for 1988. Vol 11, Govt. Printer, Dar In: The natural resources of' East Africa es Salaam. 109-1 19. (E. L\'. Russel, ed). Ha.i\.kins and E. Afr. Staples. R. R.. 1942: Combating soil erosion Lit. Bur.. Nairobi. 67-76. in the Central Province of Tanganyika Scott, R. 11.. 1972: Soils. In: East Africa, its 'I'erritory. E. '4f.r. .4gric. J. 1'01 7. Part I. peoples and resources. (W. T. LV. 11organ. 1.56-63. Part 11. 189-95. ed). Oxford Univ. Press, Sairobi, 1,ondon Staples. R. R.. Hornby. H. E. LG R. hl.. 1942: ancl New Yol-k. 9.5-105. stud\-of the c omparati~e effects of goats Sigl. Leutnant. 1892: Bericht iiber den Hall- ancl cattle on a ~nixetlgrass -hush pasture. delsverkehr \ on Tahora. 1)eutsches Ko- E:. .Ifl-. Agric. J. Vol 8 (2).62 -70. lonialblatt. \'ol 3. 164-166. Stocking. XI. .I. & Elb\ell, H. X.. 1973: Soil Skerman, P. ,l.. 1968: Reviexi of the pasture crosion hazard in Rhodesia. Rhodesia and forage situation on thc Central agric. 1.. \'ol 70 (4). 93-1 0 1. Plateau. Gogoland. Tanzania. (Unpubl. Strij~nc~uist.L.. 1976: Land S\-stems of the rep.) FAO/USDP Livestock Slission. (;reat Ruaha drairragr basin r~pstrcarnthe ~Mteradam site. In: Ecological studies of Tanz. 3linistr)- of Water Dev. and Power, the litera Basin (D. Johansson, ed). Research and Training Div., Dodoma. 57 TANESCOISWECO, Stockholm. 11-54. P. Sri-omquist. L. & Johansson, D., 1978: TMWD, 1972b: Dodoma Region Water Mas- Studies of soil erosion and sediment ter Plan, Semi-ann. progr. rep. No 3., ransport in the Mtera reservoir region, Tanz. Ministry of Water Dev. and Power, C:entral Tanzania. Zeitschrift fur Geo- Research and Training Div., Dodoma. 26 morph. N. F. Suppl. Bd 29. 43-51. P. Stuhlmann, F.. 1891 : Beohachtungen uber ThlM'D, 1974: Dodo~naRegion Water Mas- Geologe und Flora auf der Ror~teBaga - ter Plan I-XI1 (3 vols). (Unpubl. rep.) moyo-Tabora. Mitt. a.d. Deutsch. Schutz- Tanz. hlinistry of Water Dev. and Power, geb. Bd 4 (1). 48-33. Research and Training Div., Dodoma. Stuhlmann, F.. 1892: Dr Stuhln~annsRuck - TNA 1: Pastures (193249). Tanzania Na- reise von Victoria-Nyansa nach Baga- tional Archives, File 1251511. moyo. hlitt. a.d. Deutsch. Schutzgeb., Bd 5 TNA 2: Dodoma Regional Office. Dodoma (4) Berlin. 185-188. Soil Erosion (1929-37). Tanzania National Stuhlmann, F., 1894: Mit Ernill Pascha ins Archives, File 46120126. Vol I. Herz von Afrika. Deutsch Ostafrika, Bd 1, TN,4 3: Regional Office Dodoma. Dodoma Dietrich Reimer, Berlin. 901 p. Soil Erosion (193740).Tanzania National Stuhlmann, F., 1910: Hantlverk und In- Archives, File 46120126. Vol 11. dustrie in Ostafrika. Abhandlungen des Chlig. C., 1907: Der sogenannte Grosse hamburgischen Kolonialinstituts. Bd 1. Ostafrikanische Grahen. Geogr. Zeit- Friederichsen, Hamburg. 163 p. schrift, Vol 13. 430-500. Sutton, J. E. G., 1966: The East Afiican UNDPIFAO, 1968a: Masailand Develop- coast. Hist. Ass. of' Tanzania. Paper No 1. ment Report (Unpubl. rep.). 1 l p. E. A. Publishing House, Nairobi. 28 p. UNDPIFAO. 1968b: Report by the Temperle), B. N., 1938: The geology of the UNDPIFAO livestock mission follolving country around blpwapwa. Short paper visit to Dodoma and hIptvap.i\.a districts, No 19. Geol. Sun. Div.. Dar es Salaam. 61 May 1968. (Unpubl. rep.) 7 p. P. Vageler, P,, 191 2: Ugogo I. Die Vorbedin- Temple. P. H.. 1972: hleasurements of' gungen fiir die M'irtschaftliche Erschlies- runoff and soil erosion at an erosion plot sung der Landschaft Ugogo in Deutsch- scale ~iith particular reference to Ostafi-ika. Beihefte zum Tropen- Tan~ania.Geogr. .4nn.. 1-01.34 A (3-4). pflanzer, Vol 13 (1-2). 1-127. 203-220. Trade, F. B. & Oates, F., 1938: An explana- Temple, P. H. 8: XIurray-Rust, D. H., 1972: tion of Degree Sheet No 52 (Dodoma). Sheet wash measurements on erosion plots Tangan!-ika Terr.. Dept. of Lands and at Mfi~~nh~ve.Eastern Ulugnru mountains, hlines, Geological Division, Govt. Printer, Tanzania. Geogr. .4nn.. Vol 54 A (34). Dar es Salaam. 37 p. 195-202. LVapakala, CV. CV., 1974: Methods and .I'FNC;, 1978: Data report on the food ancl techniques of assessing the extent of en- qutririon situation in Tanzania vironmental damage in the rural areas of 1972/73-1976177 (2 vols). Tanzania Food East Africa. Pap. No 39. IDEP-KENDEV: and Nutrition Centre. Dept. of Planning, Workshop on environment and rural de- Dar es Salaam. velopment in Easr Africa. Nairobi. 13 p. ThlA, Land Use Schemes-(:cntral Pro- IVDD, 1935: Rep. on Ikowa Irrig. Scheme. vince. Tanzania hlinistr! of Agriculture. (Unpubl. rep.) Tang. Terr.. \Vater Dev. File P/SCH C:P. Dept, Dar es Salaam. T1LIA. 1977: The threat of desertification in \.VI)D, 19.57: Rep. on Ikowa Irrig. Scheme. central 'Tanzania. Technical paper pre- (Unpubl. rep.) Tang. Terr. Water Dev. pared for the United Nations C;onference Dept, Dar es Salaam. on Ilesertification. Nairobi 27th Au- \VD 8: ID. 1968: Bubu River Basin (Unpubl. qust-9th September 1977. Tanz. hfinistry rep.) File No 9/19. \Vater Dev. and Irriga- of Agric.. Dar es Salaam. 38 p. tion Div.. Dodorna. 4+4 p. ThlL. 1967: Atlas of Tanrania. 7an~. l\-D 8: ID. 1970: Xlat~~mbuluirrigation hlinistr! of Lands, Settlement and \Vater, scheme and water supply. (Unpubl. rep.) Surveys ,inrl >lapping l)i\-. Ilar es Salaam. File No 5) 5 1. \VD & ID. Dotloma. 8 p. 29 11. M'D & ID. 1971: hlatumbulu irrigation TbfIVD. 1972a: Dodoma Region \Vater hlas- scheme and water supply. (Unpubl. rep.) ter Plan. Semi-ann. pr(~gt-.rep. So 2.. File No 91.5 l. \VD & ID. Dodoma. Wenner, C. G., 1977: Soil conservation in Wissmann. Walther & Apolant, Berlin. Kenya. Kenya Ministry of Agriculture, 444 p. Land and Farm Management Division, Wolfel, K., 191 1: Viehhaltung im Tabora Nairobi. 206 p. Bezirk. Der Pflanzer VII. 77-85. Werther, C. W., 1898: Die mittleren Hoch- Wood, A., 1950: The Groundnut Affair. lander des nordlichen Deutsch-Ostafrika. Lane, London. 264 p. Hermann Paetel, Berlin. 493 p. Woodhead, T., 1968: Studies of potential Wigg, P. M., 1973: The role of sown pastures evaporation in Tanzania. Tanz. Min. of in semi-arid central Tanzania. E. Afr. Agr. Lands, Settlement and Water., Govt. J., V0138 (4). 375-382. Printer, Dar es Salaam. 60 p. Wissmann, H., 1889: Unter deutscher Young, A., 1972: Slopes. Oliver & Boyd, Flagge quer durch Afrika. Von 1880-1883 Edinburgh. 288 p. ausgefiihrt von Paul Pogge und Hermann Maps, air photographs and satellite images

Topographical maps Air Photographs 1:250,000 SB-37-9 Kilosa 1:10,000 Aug. 1953, 1080. Dodoma township 1:50,000 16211 Dodoma W and surroundings. 16212 Dodoma E 1:10,000 June 1970, CAN:8,106. Dodoma 16213 Luatu township and surroundings. 16214 Mvumi 1:12,500 Nov. 1960, 1546. Dodoma town- 16311 Chilonwa ship and surroundings. 16313 Handali 1: 14,000 Aug. 1957, 1387. Ikowa reser- 1:2,500 Sheet 14 Dodoma Town voir. Sheet 15 ,, 1:33,000 Aug. 1949, 82 Al276lTAN. Sheet 16 ,, Entire study area. 1:40,000 June 1960,49/TN/5and 6. Entire study area.

Geological maps Satellite images 1:250,000 Degree sheet Dodoma, 52 (1936) 1048-07184 9 Sept. 1972 1: 125,000 Quarter degree sheet Mpwapwa LANDSAT(ERTS 1). south, B 371M1 (1953). 1065-07125 26Sept. 1972 LANDSAT(ERTS 1). 1155-07133 25 Dec. 1972 LANDSAT(ERTS 1). 1155 -07 140 25 Dec. 1972 LANDSAT(ERTS 1). Abbreviations

Agr. D.: Agricultural Department. Tan- TANC: Tanzania African National ganyika Territory. Union. BRALUP: Bureau of Resource Assess- TMA: Tanzania Ministry of Agricul- ment and Land Use Planning, ture. University of Dar es Salaam. TML: Tanzania Ministry of Lands. CDA: Capital Development i\u- Settlement and Water. thority. TMM'D: Tanzania Ministry of Water DSAS: Die Deutschen Schutzgebiete in Development and ~orver. Afrika und der Siidsee. TNL4: Tanzania National Archives. DUSER: Dar es SalaamIUppsala Uni- UNDP: United Nations Development versities Soil Erosion Research. Programme. EAAFRO: East African Agriculture and VSD: Veterinary Science and Animal Forestry Research Organisa- Husbandry Department, Tan- tion. ganyika Territory. EAMD: East African Meteorological M'DD: Water Development Depart- Department. ment, Tanganyika Territory. FAO: Food and Agriculture Organi- M'D 8s ID: Water Development and Ir- ati ion of the United Nations. rigation Division, Ministry of FSL: Full Supply Level. Lands, Settlement and Water GSD: Geological Surve)- Division, Development. Tanganyika Territory. PWD: Public Works Department. Tanganyika Territory.

The semi-arid environments in East Africa face serious threats. Increased population, shortage of land suitable for cultivation, heavy grazing and increasing demands for food, firewood and charcoal have led to erosion, i.e. loss of productive soil, increased sediment load in streams and consequent rapid siltation of water reservoirs. An account is given of studies of contemporary soil erosion and sedimentation in the semi-arid savanna areas of central Tanzania, together with a review of the land degradation in a historical perspective. Detailed recording of the soil erosion complex was made within five selected catchments, four of which with reservoirs. The principal methods employed include field surveying and air photo interpretation. Thus, the main approach is physical geographical but it is not primarily an investigation of the processes as such but a study,of existing features of soil erosion and sedimentation and an analysis of the underlying causes of the processes. Thus, for instance, for the first time the 19th century East African caravan trade is analysed from the viewpoint of its direct and indirect impact on the natural environment and its possible role in the initiation of soil erosion. The traditional type of agriculture in the study area is shifting cultivation and animal husbandry based on seasonal transhumant movements between settlement areas and areas of dry season grazing and water supplies. This system of land use can be satisfactory if the ratio of land to people is high. Due to the increasing population pressure, mobility has become limited; the available land is being cultivated more intensely, and new fields are being cleared on the upper pediments around the inselbergs, Soils in this position are highly erodible. Thus the fields lose their top soil and rills and gullies dissect the slopes. Since the early 1970s a new factor has become involved as the entire population of the area has been moved from scattered homesteads into villages. This implies large concentrations of people and cattle; local overstocking arises causing overgrazing and soil erosion. This requires immediate attention from planners to avoid accelerated land degradation around the villages.

Published by:

Scandinavian Institute of African Studies P.O. Box 2126, S-750 02 UPPSALA, Sweden

Distributed by: Almqvist & Wiksell International P.O. Box 62, S-101 20 STOCKHOLM, Sweden ISBN 91-7 106-197-5