Zambezia (1978). VI (ii).RELATIONSHIP OF AGRICULTURAL HISTORY ANDSETTLEMENT TO SEVERE SOIL EROSION INRHODESIAM. A. STOCKINGSchool of Development Studies, University of East AngliaSOME 100 KILOMETRES south-west of Salisbury, well away from the mainroad, lies a tribal area distinctive because of its severe erosion and its fine,white Karoo sand. Amidst gently rolling relief are found gully scars up tosix metres deep and three kilometres long. The contrast in relief is remark-able; moving through the area, one meets slopes of no more than four de-grees at one moment and a ravine with vertical sides dissecting the landscapeat the next. More alarming than this static display of erosion, however, isthe way in which these gullies maintain themselves. Gully-head cuts havewithin living memory advanced hundreds of metres, cutting roads, destroyingvegetation, draining an already parched landscape and generally disruptingrural life. In short, it is a soil conservationist's nightmare; a badlands lands-cape reminiscent of a cross between the Grand Canyon and American dust-bowl conditions. One is tempted, as others have been, to point to the localpopulation to lay the blame for this severe erosion. Is this fair? Can theerosion and contemporary rates of advance of the gullies be linked to thepopulation and the way it utilizes the land?Man has long been recognized as an important control variable in theerosion process. The evidence is irrefutable. While soil erosion is essentiallya natural process governed by rainfall, soil conditions, slope and vegetation,man controls the system through the upsetting of the steady state. Usuallythis is through a destruction of the vegetation that promotes increased runoffand inadequate protection of the soil. The many means by which this canbe done have been reviewed elsewhere.1 In a Rhodesian context, aspects ofman's influence on erosion have been discussed in an earlier article in thisjournal by the present writer.2' L. D. .Meyer and J. V. Mannering, 'The influence of vegetation and vegetativemulches on soil erosion', in Biological Effects on the Hydrological Cycle (W. LafayetteIndiana, Purdue University. Dep. of Agricultural Engineering. Proceedings of the 3rdInternational Seminar for Hydrology Professors, 1971), 355-66; M.A. Stocking andIT. A. Elwell. 'Vegetation and erosion: A review', Scottish Geographical Magazine(1976), XCII, 4-16.2 M. A. Stocking, 'Aspects of the role of man in erosion', Zambezia (1971-2). II, ii,1-10.129'I3OAGRICULTURAL HISTORY AND EROSIONAgainst this view of man's over-riding influence on his soil resourcesmust be set the fact that quantitative evidence is lacking. The degree towhich man causes increased soil erosion is not known and, indeed, theexact mechanism by which the erosion takes place is little understood. Ero-sion modelling is in its infancy. Heinemann and Piest3 report that theemphasis in the United States is increasingly turning toward, developingcomprehensive models that consider soil detachment and transport mecha-nisms in detail. The modelling and simulation of sheet erosion has receivedparticular attention through a modified version of the Universal Soil LossEquation incorporating hydrological variables,4 and through mathematicalsimulation techniques.5 i! developed and co-ordinated, such models withappropriate and tested refinements hold promise of precise and logical ex-planation of both sheet and gully erosion, and we will have progressed to-ward 'the ultimate goal of watershed modelling',6 deterministic models inwhich all cause-effect linkages and feedbacks are known and understood.Since neither the quantitative evidence nor practicable models (testing pro-cedure) are to hand, there remains the real need to test emotive assumptionssuch as man's ubiquitous influence on erosion. Already evidence appears tofavour a mixture of man and climatic change as being responsible in partsof the United States for severe gullying,7 but local natural circumstances maystill be of significant importance. The tenuous basis of much research todate is well summarized by Cooke and Reeves: 'almost all investigators havemade an assumed relationship between vegetation and runoff, the comer-stone underlying their respective hypotheses'.6This paper will seek by two very different avenues to determine man'srole in promoting erosion in the headwaters of the Umsweswe River incentral Rhodesia. Conclusions may be applicable to other badly gullied situa-tions especially where most erosion is through deep, long gullies, variouslytermed as 'iavaka' by the French, as 'bocorocas' by South Americans, and3 II. G. Heinemann, and R. F. Piest. 'Soil erosion-sediment yield research inprogress", Eos Transactions (1975). LVI, 149-59.Ł* C. A. Onstad and G. R. Foster, 'Erosion modeling on a watershed'. Trans-actions of the American Society of Agricultural Engineers (1975), XVIII, 288-92.s W. P. David and C. E. Beer. 'Simulation of soil erosion Š Part I. Developmentof a mathematical erosion model', Transactions of the American Society of AgriculturalEngineers (1975), XVIII, 126-9, 133; 'Simulation of soil erosion ŠŁŁ Part II. Steamflowand suspended sediment simulation results', ibid., 130-3.e K. ]. Gregory and D. E. Wallinsr. Drainage Basin Form and Process (London,Arnold, 1973), 22!i. '~> See R. U. Gooke and R. W. Reeves, Arroyos and Environmental Change in theAmerican South-West (London, Oxford Univ. Press, 1976), which gives a wide-ranging-review of the problem.e Ibid., 7.M. A. STOCKING.131Chibikird _ 'School ;.'Ł'.-.MONDORORUST. LAND- ' ŁMatarutseSchool .'Hanan BrosStore IRUST /LAND-.:Ł:Ł* .32 34 16Maior gully lineGravel roadCattle dip tankPermanent buildingBusiness centreMondoro/NqeziApproximate extent ofcultivation in 197232 3420-22-Figure 1: SITE AND LOCATION OF THE UMSWESWE CATCHMENT\3ZAGRICULTURAL HISTORY AND EROSIONas 'arroyos' by North Americans,9 which are a peculiarly tropical, and sub-tropical phenomenon occurring in fine-grained sands of a cohesive nature.First, an examination is made of the history of early settlement andagriculture in Mondoro which includes the headwaters of the LJmswesv/eRiver to determine if there is any qualitative evidence to suggest man's in-volvement in erosion. Secondly, through a variable describing human pres-sure on the land, a statistical test is made of the relationship between gullygrowth and man. The study area is shown in Figure 1. It is a basin of deepKaroo sands underlain by granite. Archival records, interviews and experi-mental observation provide the main sources of information. The experimenton linking gully erosion to man is part of a broader study on the factors,both human and natural, that control rates of erosion; this wider study isunpublished'0 but the results are used in this account.EARLY SETTLEMENTEarly Agriculture, 1890-1950. Native Reserves were set up in Southern Rho-desia at an early date following European settlement. Although it couldfairly be said that in 1890 the whole country was available to the Africanpopulation, the beginnings of land apportionment were very soon manifest.The Lippert Concession of 1891 provided for the selling and leasing oflarge tracts of land to incoming settlers. It could do so with relative im-punity because the African population v/as small and the majority were ininhospitable country.11 After the Matebele and Mashona rebellions of 1896-7,a Southern Rhodesia Order in Council (1898) placed the responsibility ofproviding sufficient land for Africans in the hands of the British South Afri-ca Company. In practice this responsibility was devolved on to the districtNative Commissioners and there was no standard practice in allocating land.As Palmer notes,12 these African Reserves usually carry the connotation ofland left over after European exploitation. Nevetheless it was widely believedat the time that the Reserves 'were not only sufficient but liberal, and willprove to be ample for the requirements of the native inhabitants for manys M. Petit and F. Bourgeat, 'Les lavaka malgaches; Un agent naturel devolutiondes vcrsants'. Bulletin de I'Association de Geographes Francais, Paris (1965),CCCXXXII-III, 29-33; J. Tricart. Landforms of the Humid Tropics. Forests andSavannas (London, Longmans, 1972); F. L. Prandini, 'Occurrence of "bocorocas" insouthern Brazil Š geological conditioning of environmental degradation', Proceedingsof the 2nd International Congress of the International Association of Engineering Geo-logy (Sao Paulo, Institute) de- Pcsquisas Trcnologicas, 1974), I. iii. 36. l-.'Ui. 10: S A.Schumm, and R, F. lladley. 'Arroyos and the semiarid cycle of erosion'. American Jour-nal of Science (1957), CCXXV, 161-74.10 M, A. Stocking, 'Erosion of Soils on Karoo Stands in Central Rhodesia withParticular Reference to Gully Form and Process' (Univ. of London, impub]. Ph.D.thesis. 1977:.11 G. Kay. Rhodesia : A Human Geography (London, Univ. of London Press, 1970),47.'2 R. Palmer. Land and Racial Domination in Rhodesia ('London. Hcinemann.1977), 57.M. A. STOCKING133years to come, even taking into consideration the rapid rate at which theyand their stock increase, as well as their extensive methods of cultivation'.13By 1913 there were no less than 104 Reserves ranging in size from 2 000 to600 000 hectares.14 The situation is little changed today (although the areaset aside for exclusive African occupation is rather greater at about 47 percent of total land) and is enshrined constitutionally.Against this background of land alienation must be set the rapid in-crease in African population accompanied by an even more dramaticcrease in livestock over the whole country. In 1900 there were about 600 000Africans with, it is estimated. 55 000 cattle. Through control of both humanand cattle diseases by 1945 there were some .1,75 million Africans with 1,9million cattle.13 Today an African population of 4,5 million (40,4 per centin urban areas and, therefore, holding no cattle) has well over 2 millioncattle, about 650,000 goats and numerous other small stock. Both Kay andPalmer16 document the serious deterioration in the man-land ratio withoutany compensating improvement in man-land relationships. Indeed, changesin land usage such as the change from traditional foodstuffs and hunting tomaize monoculture and cattle keeping have added to pressures on the land.For instance in J900 it is estimated that the cultivated area was 0,43 ha perperson. By 1945 it was 0,48 and by 1962 it had risen to 0,71 ha per person.17Though little documentary evidence remains, the Umsweswe catchmentwas never settled by Europeans as the soils are relatively infertile and thearea is away from the main gold-producing districts which attracted mostearly attention. From what must have been a very sparse population at theturn of the century the area for African settlement was compressed into theMondoro Reserve with large tracts of land either side of European settlementalong the Salisbury-Hartley-Que Que-Bulawayo and Salisbury-Enkeldoorn-Fort Victoria axes. In the first two decades of this century therewere considerable influxes of people; not only those dispossessed oftheir land but probably more importantly those moving from districts in-fested by tsetse fly. For example it was noted in 1913 that fon account ofthe Fly people were moved from the Sanyati-Umfuli area to near the ZwezweRiver in Mondoro'.'e Several such population movements are mentioned,other factors being tribal disputes and the natural search for better lands,'3 Quoted in ibid., 60."» Kay. Rhodesia. 49.is Ibid.. 48, 79. Figures for African-owned rattle are problematic. Palmer, Landand Rncial Domination. 94. cites approximate figures of 44 000 in 1901 and 1,8 millionin 1(>M with specific connf of )^T-)WM ;n 1908 ;uid 40b' 180 six year--; later. The im-pression of exactness is spurious but the increase is nonetheless dramatic.'R Kay. Rhodesia. 79-93: Palmer, Land and Racial Domination, 96-8.17 Kay, Rhodesia. 80.is National Archives of Rhodesia, Salisbury, NSE/4/2/I (N[ative] Commissioner]and MaeMvitr. Hartley : Reports : Monthly : Scp\ 1912-Dcc. 1919). Momh ending'il Aim'. 1913. All documents cited hereafter are in the National Archives unless other-wise indicated.134AGRICULTURAL HISTORY AND EROSIONall of which add to a sizeable influx of people. In 1915 it was estimatedthat some 19 acres (7,5 ha) were available for each African and it wasremarked that this was 'exceedingly good';19 it does not, however, comparefavourably with the countrywide average in 1962 of 8 ha per person in theTribal Trust Lands. Some areas close by remained unsettled by Black orWhite farmers; the Native Commissioner, Hartley, noted in 1921 that an areato the south of the Umsweswe River and a little to the west of the studyarea was 'clear of natives and may be alienated'.20 Such alienation probablyonly served to further restrict available land and increase the effective popu-lation pressure.Comments in reports on early conditions and standards of agricultureare highly subjective but there are many references to the difficult physicalconditions in Mondoro, though no mention whatsoever is made of thegullies. In 1918 it was reported that 'the natives living in the sandy partsof Mondoro were not reaping much from their lands because it becameso wet and crops failed to mature'.21 The destruction of crops by gameanimals seemed an ever present problem and swarms of locusts were some-times noted in 1925. The area surrounding the Umsweswe River was des-cribed in 1925 as a 'moving mass of hoppers'.22 Apparently 2 453 separateswarms were destroyed that year with the help of adjacent European farmerswhose interests might have been prejudiced.23 All these 'natural' hazardsappear to have reduced the state of agriculture in the first three decades ofthe century to a poor and parlous condition, exacerbating man-land relation-ships.By the 1930s the situation appears to have become more stabilized. Atleast one agricultural demonstrator was active in the area and Native Com-missioner's reports note fairly good crops and good responses to attemptsat improving standards of agriculture.24 There was, however, mountingconcern over grazing land pressures. A letter from the Agriculturist, NativeDepartment, to the Chief Native Commissioner, of 18 April 1934, reported'sNSE/2/3/l (N.C. and Magistrate. Hartley : Out Letters : Superintendent ofNatives, Salisbury : 18 Nov. 1913-14 Mar. 1917), Ngizi Reserve, 19 Mar. 1915 : Aremark by the authorities not by the populace.2ONSE/2/3/2 (Out Letters : 15 Jan. 1920-23 Dec. 1922), Circular from N.C,Hartley, 'Reverted Reserves', 9 Mar. 1921.aiNSE/4/2/1, Report,'13 June 1918.22 S1032 (Ministry of Agriculture, Salisbury : Correspondence : N.C, Hartley,1906-25), letter from N.C, Hartley to Cfhief] N.C, Salisbury, 14 Jan. 1925.23 The whole, campaign against locusts in the Charter District is documented inSI827/1425/1/8 (Ministry of Agriculture, Botany and Plant Pathology, Correspondenceand Other Papers, Numerical Series, 1904-58 : Locust Campaign, Charter, 30 Sept.1932-24 Feb. 1943).24 For example, S138/72 (C.N.C, Correspondence, Numerical Series, 1923-33 :Agricultunal Activities within the Department of Native Development, 1927-33),D26/3O, Alvord to Director of Native Development, 27 May 1930. Considerable localinterest was being shown in 'Before Harvest Meetings'.M. A. STOCKING135that the overgrazing, especially in the northern sector of the Mondoro Re-serve, was particularly bad, and so recommended control of grazing andthe migration of people to south of the Umfuli River and into the Umsweswecatchment.25 At the same time there was also a general increase in aware-ness of soil erosion especially in the Native Reserves which culminated in theReport of the Soil Erosion Committee of the Rhodesian Agricultural Unionm 19322G which drew heavily on early American experience in erosion. Yetno spec1 fie nu-mic!-! is mnde of the p.i:ly erosion in (he Un-swerve catchment.Several indirect references are to be found and one report illustrates wellthe paternalism of the era:The natives are simply running wild and ploughing up large areasof land every second year. Not only are they ruining good soilthrough bad farming but destroying all the valuable Limber. This iscountry that can support a lot more natives if properly settled. Themain offenders here arc non-indigenous natives who have clearedacres and acres of the best soil, farming it very poorly and workingit as though it was a private farm . . . There is no rotation of cropsand no manure applied which means he will, soon be looking foranother 100 acres of good soil to ruin.27There were also other problems to contend with. At times the NativeDepartment agricultural demonstrators, though Africans themselves, wereunpopular, it is alleged because of their involvement with Mission schools.26However, Demonstrator Dawa in Mondoro seems to have overcome this pro-blem by 1953 and had 346 'co-operators' who were following approvedagricultural methods including manuring, using improved maize seed andpractising rudimentary conservation measures. Increased yields and the pro-duction of surplus maize gave rise to a further problem which exists eventoday: marketing and transport.29 The Reserves are usually some distancefrom main communication routes and the poor state of the roads and lackof marketing facilities obviously retarded any attempts at production of sur-plus produce. This general problem is seen today in declining productivityin the Tribal Trust Lands and a money economy unable to absorb the in-creasing rural population.Boundary Change and Administrative Responsibility. The limited awarenessof soil erosion and in particular of the gullies, coupled with a general lackof any documentary information south of the Umsweswe River has already25S1542/A4 (C.N.C., Correspondence. General. 1914-43 : Agriculture, 1933-9,Reports and Surveys on Reserves), I.26 Ibid., unpubl. report, minieo, 14pp.27 Ibid., Ill, Inspection Report of Assistant Agriculturist, G. R. Polner to C.N.G.Salisbury, 26 Mar. 1938.26 SI38/72, Agricultural Status of Reserve Natives, H. Jowitt, Director of NativeDevelopment to Supcrintendant of Natives, Fort Victoria, 14 Feb. 1930.za Ibid., Report on Before-Planting Inspection of Demonstrator's Plots on ReservesSeason 1932-3, Alvord to C.N.C., undated. Alvord notes that, 'demonstration work hastaken a set-back because plot owners have not been able to sell the improved maizegrown on plots and are wondering if it is worth while to give up the old easy methodsand do the hard work required on plots'.136AGRICULTURAL HISTORY AND EROSIONbeen noted. No obvious reason is to hand as the area was certainly settledby Africans prior to 1918 and even as early as 1899 administrative responsi-bility was apportioned to the various district offices. The answer to thelack of information probably lies in two related reasons; first, the distanceand poor communications from the nearest district offices (Hartley to thenorth-west and Fort Charter or Charter at the Range to the south-east);secondly the history of boundary changes In the area and later changes inresponsibility for different aspects of native agriculture.In 1899 the boundary between Hartley and Charter Districts was takento be the 'Uingezi' [Ngczi] River30 which runs parallel to the Umsweswe andto the south of the study area. The Umsweswe headwaters were thereforein the Hartley District. This boundary was reaffirmed in 190931 althoughsome documents in the interim appear to show some administrative functionscarried out by the Native Commissioner at the Range (Charter). Hartley issome 95 km from the study area but 175 km by road, most of it on poordirt. That is the situation today; in 1900 ail administrative communicationwas by horse and the Umsweswe catchment was on the very periphery ofthe Hartley District unconnected except by bush path and seldom visited.In 1920 the boundary between Hartley and Charter Districts was movednorth to the Umsweswe River32 and about this time all native land to theNgezi River was named Mondoro Reserve and land south of the Ngezinamed 'Unigesi' Reserve. By 1935 the Southern Rhodesia 1:1 millionmap shows the area south of the Umsweswe, now called 'Umgezi and Mondo-ro South', in 'Charter at the Range' District and the area north of theUmsweswe, called 'Mondoro North', as being in Hartley District.After 1935 the situation is confused and staff shortages especially inagricultural personnel necessitated the grouping of areas which took littleaccount of then existing boundaries. Because of its outlying position theUmsweswe catchment appears to have been sadly neglected, being rarelymentioned.The final documented changes in administrative responsibility arosefrom the 1961 Southern Rhodesia Constitution where all African lands weregrouped together to form larger blocks of land to be known as Tribal TrustLands.33 The area south of the Umsweswe River (Ngezi T.T.L.) now foundits administrative headquarters under the District Commissioner at Gatooma,some 40 km south-west of Hartley. Mondoro T.T.L. north of the Umswesweremained with Hartley although it does appear that one Agricultural Officerserved the two District Commissioners at Gatooma and Hartley for the whole30 Southern Rhodesia. Oovt Notice 1.V1RQQ.3t Southern Rhodesia, Govt Notice 131/1909.32 Southern Rhodesia, Govt Notice 146/1920.33 Southern Rhodesia, Govt Notice 518/1963.~>M. A. STOCKING137of Mondoro and Ngezi. At around this time also responsibility for Africanagriculture, which had for many years been with the Ministry of Agriculture(dealing both with African and European agriculture), was transferred tothe Ministry of Internal Affairs which now had sole responsibility for allaspects of the tribal areas. Many agriculturalists considered this a retrogradestep, especially as many of the experienced Agricultural Officers remainedwith the Ministry of Agriculture dealing with European farming alone. Theextra responsibility placed on Internal Affairs field staff wras great and in-evitably expertise in agricultural matters suffered.The Land Tenure Act of 1970 finally fixed the sizes of Ngezi T.T.L.as 56 900 ha and Mondoro as 130 500 ha, the situation pertaining today.The sorry history of administration of the Umsweswe headwaters con-trasts with the tighter control exercised in areas closer to the main districtoffices. A large part of the mismanagement and general neglect of the areacan perhaps be related to these administrative changes. How far this neglectthrough administrative change affected erosion is uncertain. It was not untilthe 1950s with the implementation of the Land Husbandry Act that therewas any concerted effort at finding out the problems of the area.LAND HUSBANDRY ACTThe Native Land Husbandry Act of 1951 was the product of mountingconcern over the deterioration of the tribal areas and unworkable earlierlegislation such as the Natural Resources Act (No. 9 of 1941). The newAct provided regulations for enforcing conservation and good farming withserious penalties for offenders. It also provided for the assessment of thecarrying capacity of each tribal area and the redistribution of arable hold-ings. Since the data base for much of the implementation of the Act did notexist, each area had to convene an Assessment Committee which met toconsider the overall picture. The information collected for each committeewould consist of a census of people and animals, areas of arable land, histori-cal background of tribes and general impressions of the state of the agricul-ture. Information varied in quality but it seems that that for Mondoro Southwas good.The Assessment Committee for Ngezi-Mondoro Reserve met on 12September 1956.34 They divided the Reserve into three, the northern zoneclosely corresponding to the study area south of the Umsweswe River. Thisnorthern zone was approximately 175 km2 with a population estimated at5 500 giving a density of about 51 persons per square kilometre. However, thepoint was made that some half of the male population was out of the Reserveand working in towns leaving a resident male population of only 532; theeffective population density is then less than 29 persons per square kilometre.Other pertinent 1956 data are shown in Table I. It should be noted that manyof the figures are probably grossly underestimated to reduce tax assessmentand cattle dipping fees. It is impossible to estimate the numbers of 'hidden'138AGRICULTURAL HISTORY AND EROSiONpeople and livestock but experience elsewhere indicates that the underestima-tion for cattle could be of the order of 50 per cent.Through, the Land Husbandry Act the only comprehensive count ofpopulations was carried out. As such it was an extremely useful exercisealthough many of the recommendations based, on the collected data failedto materialize into firm changes in land use and practice. Perhaps the onlyreal charges were firstly a maintenance of approximately 4 009 head ofcattle in the area until today (a ratio of one beast to 4,37 lin) and secondlya fairly drastic reduction in cultivated area. These moves seemed also tostem increase in human population at the expense of areas further to thesouth.Table ISTATISTICAL INFORMATION ON THE NORTHERN SECTOR OFNGEZI RESERVEHuman populationTotal, number of:cattlecalvesdonkeyssheen "goatslarge stock equivalenttaxpayersresident maleskraal headsstockownerslandholdersCultivated area (ha)Proportion cultivated land to grazing land5 5004 0295001622132874 294t 250532616861 0385 21630Number of stockowners owning:1 - 3 cattle4-67-910-1213-1516-1819-2112130016869189134 Relevant data for consideration by the Assessment Committee, was submitted inan unpublished report from R. E. Jenkinson, Land Deveiopemnt Officer, Aug. 1956(now hold by District Commissioner, Hartley). The findings of the Committee ocrurin : Minutes of Meeting of Assessment Commi'tee for Ngesi Mondoro Reserve in theCharter District, unpubl., Sept. 1956 (held by Minister of Internal Affairs, Salisbury),The latter is largely a copy of Jenkinson1 s report.M. A. STOCKINGRECENT EROSION AWARENESS139As the population of the Umsweswe catchment grew and the area wasopened up, the gullies did start to attract attention because of their sheersize. Other aspects of erosion were neglected.The curliest attempts at dealing with the gullies were as a result of ameeting of Native Department and conservation staff at Hartley in 1946 or1947,3S From this meeting a series of large contour banks designed to agradient of 1:600 were built around the heads of St Michael's Gully (thencalled Timba Gully) and a few of its tributaries. These banks were somassive that they are clearly evident today on the ground. Si Michael'sGully has progressed through the first bank and one of its tributaries isnow following the contour on the upstream side of the bank. Withoutdoubt it was the worst course of action that could have been taken sincethe contour bunks served only io concentrate surface waters and led to pre-ferential infiltration and the development of tunnelsAlso at this time three dams were built (none surviving now) for stockwatering to reduce concentrations of cattle along the river banks. The headsof the gullies were fenced but it appears that the fencing material was stolenwithin one or two years and the members of the Assessment Committee forthe Land Husbandry Act were unaware of its existence in 1956.It was not until 1969 that the authorities showed any firm resolve totackle the problem of advancing gullies. A series of Ministry of InternalAffairs departmental reports and memos in 1970 and 1971 leave a record ofunfortunate recommendations and conclusions.36 For example, despite anadverse report from the Rhodesia Forestry Commission on the planting oftrees in the immediate vicinity of gullies37 several belts of exotic trees wereplanted. Their growth has been minimal and has probably only retardedgrass growth within the belts. Much was also written on the mismanagementand overgrazing of the catchment with little evidence to support such acontention. The result was an expensive programme of fencing the gullyheads to allow regrowlh of grasses with minimum grazing; the D.C., Hartley,reported in 1969 that 39 miles of fencing were recommended at £100 permile.30 The fencing programme has continued through to 1975 with furtherconsiderable cost. Since the fencing depends on both regular maintenanceand the closing of the few main gates into the enclosures, the onus lor success35 F. Micklesfield, retired Agricultural officer, personal communication, 17 May1976.36 Ministry of Internal Affairs, Salisbury, AGR/5/50 (Correspondence and Reports,1967-71), Soil Conservation Special Project Ngezi Tribal Trust Land and MondoroTribal Trust Land.2v Rhodesia Forestry Commission, Ref. 399. Tree Planting for Gully Reclamation,.Memo from T. j. Hodgson, Forest Officer, 29 Jan. 1971,38 Ministry of Internal Affairs, Salisbury AGR/7/12/69. Gullies : Hartley andGatooma, Woollacott to Provincial Commissioner, Mashonaland South, 21 Nov. 1969.140AGRICULTURAL HISTORY AND EROSIONof the scheme was placed on local tribesmen, a responsibility not appreciatedby many. Certainly by 1974 the fencing was relatively ineffective and nomeasured differences in grass cover were found inside and outside fencedareas, indicating that cattle were effectively free to graze at will.Recently the Natural Resources Board has taken an interest in gully re-clamation. A series of stone bolsters have been placed in the base ofSt Michael's Gully and in several of the heads of the Chinove-Bhiza sectionwith finance raised by the Natural Resources Board. The bolsters consist ofironstone wrapped in chicken wire. Some considerable sedimentation hasoccurred upstream of the bolsters and the floor of St Michael's Gully nearits head cut is now 75 cm higher. However, the cost of construction of thebolsters and subsidiary efforts such as planting and watering of grass hasbeen exorbitant. While it has been demonstrated that some sedimentationhas occurred, the effective gain in land has been negligible and the advanceof the head cuts has hardly been retarded. Without the sizeable grants fromGovernment sources, any reclamation would be impossible and it is feltthat any workable scheme is financially impractical. To date the programmeof stone bolsters is continuing. Other less successful (or even disastrous)attempts have been discontinued, examples of which were the bulldozing ofhead cuts to smooth the profile (the head cut in one instance reinstated itselfafter one storm) and the planting of sisal.From all of these more recent experiences it may be concluded thatlittle real practical benefit in land reclamation has been achieved. The focusof attention has always been the head cut and the immediate gully base. Thecatchments have been largely neglected because perhaps by comparison withother tribal areas they are fairly well covered by vegetation. Finally, therehas been very little recognition that the gullies might conceivably be primarilya product of natural processes. The erodible nature of the soils is occasion-ally mentioned but the blame has all too readily been placed on local tribes-men for mismanagement of the land Š a factor which is not necessarily sup-ported by observational evidence and enquiry.QUANTITATIVE ASSESSMENTThe evidence from a study of the history of occupation of the area suggeststhat while mismanagement and administrative neglect have occurred, thepopulation has never been excessive and no extraordinary man-induced in-fluences may be invoked to account for severe erosion. It remains, however,to determine if contemporary erosion may be aided by pressure of popula-tion on the land. The most obvious means of doing this is to compare ratesof advance of gully head cuts between catchments with differing populationpressures. Since catchment variables such as area and vegetation cover arealso likely to be different between catchments, it is necessary to take theseinto account in a multiple regression model along with the human factor. Itis reiterated that the purpose of this paper is to test man's influence onM. A. STOCKING141erosion; the other variables are introduced only for statistical rigour andtheir significance is reported on fully in my unpublished thesis.39A Variable for Population Pressure. The choice of a parameter to measurepopulation pressure in problematic. On the one side, it is not known howman affects erosion; on the other, it is not known how much pressureland can sustain without irrevocable damage. Carrying capacity formulaehave been designed40 but all are based on untestable assumptions and datathat are not always available. A proxy variable for population pressure isnecessary which integrates the many unknown linkages. In a study of differ-ences between types of land tenure, population densities and severity oferosion, it was found that in most instances population density was relatedto erosion.41 In the present study all catchments to gullies are wholly withinone type of land tenure, the communal grazing and small arable plots of theTribal Trust Lands. Therefore, the major difference that may have a bearingon erosion betweer. catchments is the numbers of people making use of theland. The main assumption is that every person through his practices andthrough the grazing animals he keeps affects erosion equally. That this isnot so on an individi^l basis can be demonstrated by comparisons of sheeterosion and declining fertility on adjacent smallholdings; farming has multi-ple objectives and multiple paths to reach those objectives. Nevertheless overlarge groups of peoples the averaged effect should approximate to the assump-tion, and population density be a suitable proxy variable for man's influenceon erosion through farming practicePopulation density was measured on a square kilometre grid basis by100 per cent count. From Figure 2 it can be seen that the density is quitevariable (between 0 and 130 persons per square kilometre) depending onthe location of the larger village groupings on the edge of the arable landsand along the main watershed boundary roads. Mean population density in1974 was 31,4 persons/km2 indicating a slight increase over 1956. The totalpopulation was 5 645 with a rather greater density (34,1) south of the Urn-sweswe River than north of it (25.8).Other Variables of Gully Growth. In the multiple regression model, popula-tion density, D, represents one of the independent variables. The others are:39 See footnote 10.40 \V. Allan, 'How much land dors a man require?', in Studies in African LandUsage in Northern Rhodesia (Lusaka. Rhodos-LK-'.ng.itone Institute, Paper No. \b,1949), 1-23; R. Feachem 'A clarification of carrying capacity formulae', AustralianGeographical Studies (1973), XI, 234-6.41 Stocking, 'Aspects of the role of man in erosion'.42 I. Seginer, 'Gully development and sediment yield', Journal of Hydrology (1966),IV, 238-53; j. R. Thompson, 'Ouanlitave effect of watershed variables on rate of pully-head advancement', Transactions of the American Society of Agricultural Engineers(1964), VII, 54-5,142AGRICULTURAL HISTORY AND EROSIONKmUmsvveswe RivetMajor gully lineSt Michael s MissionPopulation Densitypersons/km^i 10-910-19?0-29:'Ł:} 30-3940-5960-7980 +Figure 2: POPULATION DFNSJTY OVTR THE UMSVVESWE CATCHMENT.M. A. STOCKING,143Precipitation, P mmAntecedent precipitation index. A mmCatchment area to gully head. A, knrMean vegetation cover. C °oHeight of head cut, K mSiopc oi* approach channel, S dc?rec>While many other variables could ha\e been u<cd. of tho::e chosen seme areones thai have been found imponi-n' ehvv. heiv:r- and others would ha\ea high depree of interdepcmlcncy with extra variables. Gully growth isme^ured by the volume cf soil reproved from the head cut after each storm,event, Y m\Results. Using a step-wise multiple regression procedure whereby the leastsignificant- contributory variables (based en t-value) are progressively dropped.the most significant variable remains. At the same time the contributionof the other variables towards explanation cf gu!ly growth may be monitored(Table II). A sample of gully heads retreating principally ihrough waterfalleros;o;i was used; she wider siuciy. of which this is a part, also consideredother head cuts at different time scales. Results are broadly similar to theones reported here in Table II.The most significant individual variable is A, (product-moment correla-tion coefficient, r = 0,686), followed by H (0,606), A., (0,535) and P(0,361). Variables C, D (population density) and S are nearly randomly cor-related with gully growth and are eliminated at an early stage in the step-wise regression, Interaction between the other variables. P, A,,, A. and Hgives an equation that explains 80,1 per cent of the variation (R = 0,895).With a 10 per cent loss in explanation (R = 0,844) A., may be eliminatedto give what is probably the optimum prediction equation for these gullyheads:Y = 6.S7X10-3 p1'34 Ac 1-°° H0'52Discussion. Population density fails to improve the prediction of head cutgrowth and. by ii-jvif. nn> p. \cry low correlation with gully growth (r = 0.0I8).The reason for ii^s is either that population density is an ineffective measureof population pressure on the land or that population pressure is not an im-portant contributory factor to erosion under the conditions investigated. Ithas already been argued that population density is likely to be a crude butsignificant parameter for population pressure, and, therefore, the latter ex-planation is preferred. Indeed, the results suggest an essentially natural basisfor gully formation in the area and no discrimination is found in rates oferosion between catchment areas with high population density and thosewith low density."3\I. A. Stocking. 'Tunnel erosion'. Rhodesia Autirti'tnrnl journal (197G),LXXIII, 35-9.f ..//GULLY GROWTH PREDICTION EQUATIONS FOR WATERFALL HEAD CUTSI-s 5 i£-RegressiontŠvalues of VariablesY = 2 55 x 10~2 P1-02 A °'56 A °'90 C~"°i28 D~°>09 H0>46P cY = 2,03 X 10~-2 P1'01 A 0,56 A 0,90 C-0,31H0.47S-0,29J) CY = 4,97 x 10~3 P1-06 Ap°>52 Ac0'85 H0-49Y = 4,14 X lO-Sp^^A 0,53 A 0,86 H0,50Y = 6,87 x jO--3P1-34Ac1'ooH0'53Y = 4,42 x lOŠ'P1-1-"^1'47Y = 3,5A,1-50P Ap Ac C D H S40 0,902 4,06 4,41 4,95 Š1,05 Š0,41 4,67 Š1,3041 0,901 4,08 4,52 4,99 Š1,19 Š 4,76 Š1,3142 0,898 4,30 4,34 4,82 Š Š 5,12 Š1,0543 0,895 4,40 4,39 4,86 Š Š 5,22 Š44 0,844 4,73 Š4,86 Š Š4,52 Š45 0,760 3,37 Š 6,91 Š Š Š46 0,686 Š Š 6,40 ______Note: Scries of regressions follows the step-wise procedure giving equations with the higher multiple correlation coefficient with a givennumber of variables.M. A. STOCKINGCONCLUSIONS145There is neither firm historical evidence nor contemporary experimental andstatistical evidence to support the contention that man has largely broughtabout the severe erosion in the Umsweswe catchment. Historically, it hasbeen shown that the population has remained fairly static over recent years,having risen gradually from the mid-nineteenth century. Cattle and livestockpopulations likewise have little changed. There are some documented agri-cultural difficulties, such as locust swarms, administrative neglect and unwiseconservation treatment of the gullies, but these cannot solely account forthe formidable erosion manifest today.This qualitative view is supported by experimental evidence of ratesof erosion compared to population pressures within the catchments to thegullies. From a large sample of gully heads and individual storm events thereis no basis for present high population densities to be the cause for measuredrates of gully advance. It may be that if population were to continue to in-crease above some threshold, gully advance would also increase but thisthreshold, if it exists, has not been reached by existing population densities.Care must be taken in applying the results to other situations. Certainlythe variety of valley bottom gully characterized by steep or vertical walls incohesive fine sediments and flat wide floors, and variously called Mavaka','arroyos' or 'bocorocas', should not immediately be taken as evidence formismanagement of the land. As in the case studied, these gullies are pro-bably a natural response to changing environmental conditions. This is notto say that man may not aggravate the conditions; he often does, but he isnot the principal cause. Similarly, other forms of erosion may be naturalresponses to particularly erosive conditions. In the tropics, soil types areoften responsible for serious tunnel erosion.43 Other cases may come to light.At the same time, man's role should not be diminished. He is undoubtedlyresponsible for many if not most cases of land degradation. But he may notbe responsible for all. There exists a wide field for research in erosional pro-cesses and forms in the tropics.ACKNOWLEDGEMENTFunds for this research were provided by the Research Board, Universityof Rhodesia.