Zambezia (1989), XVI (i).GULLY FORM AND DEVELOPMENT ON KAROOSEDIMENTS IN CENTRAL ZIMBABWE:A PRELIMINARY SURVEYR. WHITLOWDepartment of Geography, University of ZimbabweandC. FIRTHGeography Section, West London Institute of Higher Education, IsleworthABOUT ONE TENTH of Zimbabwe is characterized by sodic soils which areespecially prone to sheetwash erosion and gullying associated with subsurfacepiping (Wendelaar, 1976). Although localized patches of sodic soil occur in poorlydrained sites on granitic rocks, they are more widespread on Karoo sediments inthe north-western, western and central parts of the country (see inset in Fig. 1).Previous research on these soils has been directed mainly at the dynamics ofplant-habitat relationships, as a basis for devising suitable methods of reclamation(Dye, 1979; Dye and Walker, 1980), and at the definition of factors influencing themorphology and extension of headcuts (Stocking, 1977 and 1981). Consequently,a great deal is known already about the nature and current rates of erosion on sodicsoils. However, very little basic geomorphological research has been carried out onthe development and environmental significance of gully systems on the Karoosediments or, more specifically, the fine sandy colluvium overlying thesesediments.The present exploratory study was carried out in an 'island' of Karoosediments some 2 380 km2 in area, which lies to the south of Harare (Fig. 1). Theobjectives of the exercise were, firstly, to determine the distribution of active andstabilized gully systems in this locality and, secondly, to evaluate the morphologyand sedimentary sequence of selected gully systems. Although it would beunrealistic, at this stage, to propose an evolutionary model and chronologicalframework comparable with that described by Smith (1982) for large gullies innorthern Nigeria, there are good prospects for doing so in the future. In addition, itwould be desirable to link such research with the picture of late Quaternaryenvironmental changes identified from recent investigations in Swaziland (Price-Williams et ai, 1982).DISTRIBUTION OF GULLIESInitial mapping of the gullies was carried out by detailed examination of 1:80 000aerial photographs. This indicated that gully features are restricted to the westernGULLY FORM AND DEVELOPMENTFigure 1: DISTRIBUTION OF ACTIVE AND STABILIZED GULLIES ONKAROO SEDIMENTS IN CENTRAL ZIMBABWEhalf of the area underlain by Karoo sediments (Fig. 1 inset). More detailedmapping of this western area was carried out using 1:50 000 topographical basemaps and 1:25 000 aerial photographs taken in 1976. The results of the surveyare presented in Figure 1.A total of 116 gullies or gully segments (where active and stabilized channelswere linked) were identified, of which 62 appear to be active and the remainderR. WHITLOW & C. FIRTH 3have the plan forms of gullies, but are overgrown with herbaceous and snrutvegetation. The total length of active gullies is 73 575 m giving an average lengthper gully of 1 187 m and a density of 70 m per km2. Although there are fewerstabilized gullies, their total length is greater than that of the active systems givingan average length of 1605 m per gully and a density of 83 m per km2. The densityof natural streams in this area, excluding the main rivers shown in Figure 1, is inthe order of 425 m per km2. Therefore, the gullies represent a significant extensionof the overall drainage network.Three general observations can be made regarding the distribution of thegullies. Firstly, although active gullies occur throughout the area they areespecially common in the upper reaches of the Muzvezve River. Here gullydensity is around 237 m per km2 Š over five times that for active gullieselsewhere in the study area. Secondly, stabilized gullies are prevalent in the lowerreaches of the river along with the gullies draining into the northward flowingtributaries of the Mupfure River. Thirdly, there are several gullies where the lowerportions of the incised channels are now vegetated and stabilized but where theupper portions are still devoid of plant cover and apparently actively eroding. Thereasons for these differences are not known as yet but may relate to variations inthe nature of the Karoo sediments and rejuvenation of the major river systems.Human factors may also be involved in determining the occurrence of thesegullies since, with one major exception, all the gullies are located within theMhondoro and Ngezi Communal Lands. The gullies are thus located in areasused for peasant agriculture and are generally absent from the adjacentcommercial farmlands. On a national level there is certainly more widespreadand severe erosion in the Communal Lands due, in part, to the populationpressures and poor conservation methods (Whitlow, 1987). However, in this areageneral observations on aerial photographs suggest that there is a poor spatialcoincidence of gullies and human factors such as settlement density and extent ofcroplands, a fact demonstrated previously by Stocking (1978a) for the StMichael's Mission area. Nevertheless, as discussed later, the concentration ofrunoff along tracks and old plough furrows can have localized effects on the formand enlargement of gullies.Generally, the active gullies in this area are 'Type 3 gullies' as defined byImeson and Kwaad (1980); that is, they have U-shaped cross-sections, are formedin colluvial or weathered materials and are characterized by subsurface pipingdue to the influence of high levels of exchange sodium. Although the gullies aretypically dendritic in form (Plate 1), at least three other types are present. Firstly,at least 7 of the 62 active gullies have a compound form (Plate 2; Ireland etal,1939). This gully form is believed to represent an advanced stage in thedegradation of a dendrit ic system whereby tun nelling between adjace nt headcutsresults in the formation of residual ridges and the opening out of the gullies asGULLY FORM AND DEVELOPMENTPlate I: DENDRITIC GULLY SYSTEM 1:12 500Reproduced by courtesy of the Surveyor-General of Zimbabwechannels coalesce. Secondly, there are ten distinctly linear gullies (Plate 3), thesebeing relatively narrow channels with very few lateral headcuts (Ireland et al,1939). Thirdly, there are three examples of broad, shallow gullies with distinctivebulbous forms (Plate 4). In a detailed study of headcuts on dendritic gully systemsStocking (1978b) recognized five major plan forms (pointed, rounded, notched,digitate and bulbous), these being related to the processes of subsurface andoverland flow affecting the headcuts. Similarly, at the scale of the gully system as awhole, the variations in a plan form are likely to reflect dominant processes aswell as the nature and depth of the unconsolidated materials within which thegullies are formed. Further field observations are required to verify these points.CROSS-SECTIONAL FORMS OF GULLIESThe examination of cross-sectional forms and associated sediments along thelength of a gully can reveal evidence of infilling and trenching (e.g. Womack andSchumm, 1977; Hannam, 1983). Reconnaissance surveys were undertaken onR. WHITLOW & C. FIRTHPlate 2: COMPOUND GULLY SYSTEM 1:12 500Reproduced by courtesy of the Surveyor-General of Zimbabwetwo active gullies and one stabilized gully in the south of the study area, the localnames for these sites being indicated in Figure 1. Gullies were chosen in thislocality primarily because of the availability of 1:25 000 scale false colourimagery taken as part of an FAO/UNDP programme in May 1986. Field mapsbased on 15x enlargements of this imagery were drawn up prior to fieldwork andproved invaluable in the selection and location of cross-sections for surveying. Fiveother active gullies were examined briefly to ensure that the surveyed features weretypical of the gullies in the area.Manyewe Gully: This was selected as an example of a stabilized gully, nowchoked with sediment, in the lower Muzvezve basin. The outline plan of the gully(Fig. 2) defines the junction between the relatively flat gully floor covered in denseherbaceous vegetation, a vivid red colour on the false colour imagery, and theadjacent sparsely-wooded degraded slopes. The gully system is 2 810 m in length,up to 6 m in depth and, using the basal width of the infilled floor, between 6 and42 m wide. The former headcuts have a distinct broad lobate form and aredistinguished on the ground mainly through contrasts in vegetation rather thanGULLY FORM AND DEVELOPMENTPlate 3: LINEAR GULLY SYSTEM 1:12 500Reproduced by courtesy of the Surveyor-General of Zimbabweclear breaks in slope. For example, the fringes of the gully floor are characterizedby a narrow (3-8 m) zone of stunted, frosted shrubs, mainly Terminalia sericeaand Colophospermum mopane (Fig. 2).Three morphological units could be identified in this stabilized gully Š thesebeing degraded slopes, a sheetwash zone and the gully floor. The degraded slopes,lowered through processes of sheetwash and rilling, were characterized by a drymiombo woodland dominated by Brachystegia spp., very similar to that in areassurrounding the gully. The upper margins of these slopes were poorly defined, butgenerally convex in form. Slope angles (see sections 1-5 in Fig. 2) were generallybetween 8° and 12° along the main gully, although locally much greater.Extensive areas of bare ground are exposed beneath the discontinuous treecanopy, and numerous calcareous nodules 1-2 cm in diameter occur on thesedenuded surfaces. The bases of these degraded slopes are sometimes marked by aclear break in slope with the lower sheetwash zone. More commonly, however,deposition of sheetwash material gives rise to a gradual change in slope on ratherhummocky ground.R. WHITLOW & C. FIRTHPlate 4: BULBOUS GULLY SYSTEM 1:12 500Reproduced by courtesy of the Surveyor-General of ZimbabweThe sheetwash zone is defined mainly because of its distinctive shrubvegetation and via the thin veneer of sediments that mantle its surface. This zone isan area through which sediment is transported from the steeper adjacent slopes tothe low angled surfaces (under 4°) on the margins of the gully floor. However,since the grass cover is still very sparse in this zone it is unable to prevent overlandflow from removing the unconsolidated surface materials during heavy rain-storms. The gully floor is relatively flat, with dense plant cover and localized stripsof rushes (sections 3 and 5, Fig. 2) marking possible seepage lines. Shallowdepressions in the gully floor between sections 2 and 3, for example, may havebeen caused through livestock trampling. However, elongated pools in the lowerparts of the gully suggest that subsurface erosion may be undermining the infillsediments. An entrenched section just above the junction with the main streamindicates that the gully infill is at least 1,6 m deep. The exposure consists ofintercalated coarse to fine sandy materials and humic horizons, generally welldefined and 2-8 cm in thickness.If the interpretation of this channel as a former, now-stabilized gully is correct,GULLY FORM AND DEVELOPMENTFigure 2: MORPHOLOGY OF MANYEWE GULLYNote Scale of cross sections isditlereftt trwn those lorMusinanta and Miatonje gultysKey to symbols used to identify features on gully cross-sectionsdry miombo woodland 4 termite mound2 Terminalia sericea 5 poorly stratified channel infill3 Colophospermum mopane 6 sheetwash zoneR. WHITLOW & C. FIRTH 9it has some interesting clues to offer on the duration and stages of stabilization ofgully systems. Certainly, it would appear that it has taken a long time for infillingand lowering of the formerly steep gully banks to occur if one compares the formof Manyewe Gully with that of Muzhanje Gully described below. Since there aregood prospects for finding datable organic-rich horizons in the gully infills, it maybe possible to establish the period(s) when active sedimentation began in thesegully systems.Muzhanje Gully: This was selected as a relatively simple example of a deep, activegully system. Twelve cross-sections were surveyed in the upper reaches of thisgully over a length of 720 m, with more general observations being made in theremaining 630 m of the channel (Fig. 3).The gully is incised in a gently sloping valley floor (Plate 5), with generallywell-defined, somewhat irregular margins due to the existence of numerous smalllateral headcuts. Coalescence of active headcuts in the past has left a series ofresidual ridges and knolls (Fig. 3), between sections 8 and 10, for example. Thereare also several large subsidence hollows, over 2 m in diameter, near the gullyupstream of section 8 and these emerge as pipes at various levels in the walls of thegully below section 7. Downstream of section 9 several shallow lateral gulliesdrain into the main channel, but these are well vegetated and have poorly definedmargins.This incised channel is eroded into deep colluvium within which there areclearly developed discontinuous stonelines. Stocking (1978c) interpreted these aslag gravels buried by colluvium which is up to 3,5 m deep in this particular gully.The upper 1 m of the colluvium forms a distinct layer that probably originatedfrom sheetwash erosion (Fig. 3), but this origin requires further study.Six profiles were surveyed within 120 m of the headcut on the basis that thesewould show evidence of more recent (last 200 years) gully extension. Theheadcut itself is 2,8 m deep and 2,2 m wide with a 'square' plan form. A shallow,discontinuous channel drains into the headcut and most of the surface horizonabove the incision has been stripped away (Plate 5). At the base of the headcutthere is evidence of lateral undercutting associated with piping (Fig. 3). Thesections directly below the headcut are essentially U-shaped with a flat floorcovered in recently deposited sediments. Differential erosion of the friablesheetwash material and localized mass movements have resulted in irregularhummocky and stepped gully walls, with the steeper banks (over 40°) beingcharacterized by fluting. There is a terrace-like feature some 0,8 m above thepresent floor in this upper reach of the gully (Plate 6). It is composed of poorlystratified sediments which could have been produced from materials derived fromsidewall erosion rather than movement down the channel. It suggests a recentphase of localized entrenchment of the gully of up to 1,5 m into infill deposits thatextend 0,5-0,7 m below the present gully floor.10GULLY FORM AND DEVELOPMENTFigure 3- MORPHOLOGY OF MUZHANJE GULLY.. . . -10Key to symbols used to identify features on gully cross-secuons1 suspected sheetwash 6 channel infill2 loose surface material 7 infill with siliceous boulders3 slump blocks 8 siliceous gravels and boulders4 poorly stratified sediments 9 fluted gully bank5 well-stratified sediments 10 pipe outlet*R. WHITLOW & C. FIRTH11Plate 5: MAIN HEADCUT ON MUZHANJE GULLY WITH HANDLE OFSPADE MARKING THE SUSPECTED SHEETWASH HORIZONBelow the junction of a major right-bank tributary, the main gully widensappreciably to about 13,5 m at section 7, but does not exhibit a comparableincrease in depth (Fig. 3). There is still evidence of a terrace feature in this portionof the gully. Further downstream (section 8) the gully increases to a width of14,4 m and a depth of 4,4 m and the banks are strongly fluted with numerous pipeoutlets entering the gully about 2 m below the ground surface. This part of thegully is extremely dissected due to the coalescence of several right-banktributaries. Vertical incision has exposed the well-jointed, partially weatheredKaroo sediments (Plate 7) in the basal 3 m of section 9 and for 80 m either side ofthis section. Channel erosion has exploited the vertical joints in the Karoosandstone forming a cleft 1,3 mdeepand 1-1,2 m wide in the base of the channel(section 9, Fig. 3).Downstream of section 9 the right bank of the gully is indented with numerousheadcuts. However, within 120 m the gully regains a more regular form. The gullyis 21 m wide and 4,8 m deep at section 10, the distinct break in slope on the leftbank of the gully being interpreted as the former floor of the channel, and isindicative of 2,8 m of incision. Several wells excavated in the base of the gullyindicate that the contemporary infill is up to 1,2 m deep, and overlies weatheredKaroo sediments. Further down the gully (sections 11 and 12) there is clearevidence of sediment fills at higher levels, with stratified sediments forming12GULLY FORM AND DEVELOPMENTPlate 6: FLUTED VERTICAL BANK AND RESIDUAL BENCH BELOWSECTION 6, LOOKING UPSTREAMPlate 7: EXPOSED KAROO BEDS AT SECTION 9, LOOKING UPSTREAMR. WHITLOW & C. FIRTH13distinctive benches between 2,2 and 2,8 m above the present gully floor (Plate 8).These sediments comprise intercalated humic and mineral horizons which aregenerally underlain by a coarse gravel layer of mainly siliceous nodules. Theseterraces seem to have been produced as a result of deposition in a broad, flatchannel with localized fans from lateral gullies. Below section 12 the gully has nearvertical walls up to 8 m deep and a very flat floor, this form being typical of thelower reaches of gullies in this area. However, remnants of the infill terraces persist,sometimes on both sides of the channel, for some distance below section 12.Musinambi Gully: This was selected as an atypical gully system with distinctivebadland erosion along its right banks, giving rise to an asymmetric cross-sectionaland plan form (Fig. 4). The gully is also noteworthy because relatively recentheadward extension has breached a dam. The area surrounding the gully is usedmainly for grazing but the extensive network of ridges and furrows to thenorth-west of the gully indicate that cultivation was more widespread in the past(see Fig. 5A-E). The furrows may, in fact, have contributed towards the erosionof the right bank of the gully through concentration of surface and subsurfacerunoff.Sixteen cross-sections were surveyed along this gully as shown in Figure 4. Fordiscussion purposes these can be divided into three groups, these being a zone ofPlate 8: STRATIFIED INFILL IN LOWER REACHES OF MUZHANJEGULLY14GULLY FORM AND DEVELOPMENTFigure 4: MORPHOLOGY OF MUSINAMBI GULLYKey to symbols used to identify features on gully cross-sections1 sheetwasb2 loose surface material3 poorly stratified sediments4 well-stratified sediments5 channel infill6 siliceous gravels and boulders7 surface nodules8 fluted gully bankR. WHITLOW & C. FIRTH15incision (sections 1-6), a zone of deposition (sections 7-10) and zone ofentrenched infill (sections 11-16). Below section 5 the cross-section surveys wererestricted to the inner portion of the gully because of badland erosion extending30-80 m back from the main axis of the gully. Wherever possible the sections werelocated where intact remnants of the former ground surface existed, avoidingseverely eroded banks. A series of photographs was selected to demonstratevariations in morphology and infill sediments in this gully (Plates 9-12).Section 1 (Fig. 4) is at the head of the gully where there is a broad roundedheadcut some 0,8 m deep. Subsurface piping is evident at the base of the finesandy A-horizon and appears to be a major cause of headcut enlargement. Some30 m downstream of this there is an incision 1,4 m deep and 1,7 m wide, with nearvertical, fluted walls (section 2, Fig. 4), giving rise to a clear gully-in-gully form(Plate 9). The subsoil material which has been incised is pale grey-brown incolour with localized iron mottling and a distinctive prismatic structure, typical ofthe weathered Karoo sediments. Below the inner gully headcut the channelwidens rapidly to about 6 m and at section 4, below the breached dam, is nearly2,5 m deep. Deep vertical clefts and pipe outlets up to 0,35 m in diameter, alongwith loose fresh debris and blocks of soil on the gully floor, show that there is veryactive erosion, mainly subsurface in nature, in this reach of the gully. Small (under0,1 m diameter) holes are present up to 35 m away from the outer gully banks andPlate 9: GULLY-IN-GULLY FORM OF THE UPPER REACHES OFMUSINAMBI GULLY16GULLY FORM AND DEVELOPMENTindicate the existence of an extensive pipe network. Subsurface tunnelling couldwell have caused the breaching of the dam wall between sections 3 and 4 (Fig. 4).Sections 5 and 6 represent a more advanced stage of widening of the gully.Massive siliceous nodules at the base of the gully at section 5 form distinctive stepsin the gully wall and have inhibited basal widening. However, erosion ofmaterials overlying the nodules has given rise to lower angled (45-50°) slopes,with numerous deep vertical clefts on the left bank and rounded cavities providingevidence of active subsurface erosion. There is still some sign of the gully-in-gullyform at section 6 (Fig. 4), with the inner channel being nearly 9,5 m wide and1,8 m in depth. Stripping away of the surface soil has exposed material packedwith calcareous nodules, 1-2 cm in diameter, which in section display a series ofconcentric rings, suggesting gradual accretion in a weathered substrate. There isstill loose, fresh sediment on the gully floor at this site, but now with a sparse coverof stoloniferous grasses.Sections 7-10 represent a zone of deposition within the inner gully systemwhich itself is incised into infill sediments (Fig. 4). The channel is rectangular insection and about 7,5-12 m wide and about 1 m in depth, with a well-grassed flatfloor (Plate 10). Either side of the gully the surface soil has been removedcompletely giving rise to a series of residual knolls and ridges, especially on theŁPlate 10: CHANNEL INFILL IN MIDDLE REACH OF MUSINAMBI GULLYLOOKING DOWNSTREAM WITH C. FIRTH AT POSITION OFSECTION 7R. WHITLOW & C. FIRTH17right bank. Some of the materials in the exposed walls of the inner gully are poorlystratified, but of limited lateral extent, while sediments at section 8 are clearlybedded. This suggests that the present channel has been incised into a former gullyinfill and now the gully is being filled itself, a hypothesis supported byobservations in the remaining sections of the gully.The third reach of the gully, the zone of entrenchment, is in the order of 1,5-2 min depth, up to 23 m wide and very irregular in cross-sectional form (Fig. 4). There isa well-defined trench about 1 m deep and up to 3 m wide in the gully floor (Plate11), beginning with a narrow, vertical headcut at section 12 and extending forsome 80 m down the centre of the infilled channel. The trench becomesprogressively shallower downstream of section 13 until it eventually disappears.The remaining sections, 14-16, show signs of successive reworking of infillsediments which has left a very hummocky channel floor. The entrenchedsediments were clearly stratified (Plate 12) with several thin humic bandsseparated by fine, convoluted layers of fine sandy clays indicative, perhaps, ofgradual accretion of material. Shallow, elongated troughs upstream of the presentheadcut at section 11, for example, suggest that there is an element of subsurfaceerosion taking place within the infilled gully floor. The main cause of headwardmigration of the trench appears, however, to be overland flow.Plate 11: ENTRENCHED INFILL IN LOWER REACH OF MUSINAMBIGULLY, LOOKING UPSTREAM TOWARDS HEADCUT ATSECTION 1218GULLY FORM AND DEVELOPMENTPlate 12: STRATIFIED SEDIMENTS IN ALLUVIAL INFILL NEAR SECTION 12R. WHITLOW & C. FIRTH 19EXTENSION OF ACTIVE GULLIESConsiderable research has been conducted on the rates and processes of gullyheadcut extension in the Ngezi area. Stocking (1981), who monitored the StMichael's Mission gullies, has shown that headcut extension is a function ofdrainage area above each incision, precipitation amount, antecedent precipitationand the height of headcut. However, he noted that the order and significance ofthese variables alters between different head cuts and if the time-scale is changed.In contrast, there has been limited study of the ways in which the plan forms ofgullies alter through time, information which is complementary to an analysis ofthe cross-sectional forms within gully systems. Apart from the false colourimagery referred to earlier, five different dates of panchromatic photographs wereavailable for examination of Musinambi and Muzhanje Gullies as shown in thedetailed map sequence in Figure 5. Changes in these two gullies since 1956 arediscussed separately.Musinambi Gully: This gully, as noted earlier, is characterized by a sequence ofincision and infill and widespread badland erosion along its right bank. The 1956photographs show that there was localized degradation along the left bank of thegully (Fig. 5A), but this is not visible on later photographs. This is because theground surface is obscured by dense thickets of thorn shrubs, mainly Acacia andCombretum spp., which make it difficult to determine the margins of the badlanderosion precisely. This also applies, to a lesser extent, to the right bank where over14 000 m2 of degraded patches were mapped on the 1956 photographs (Fig. 5A).Over the period 1956-84 these patches extended and coalesced, increasing thearea to some 16 200 m2 by 1984 (Fig. 5A-E). During this time there was gradualextension of flat-floored tributary channels into the badlands. The grounddraining into these badlands is crossed by numerous old cultivation furrowswhich probably concentrated runoff (surface and subsurface) into the gullysystem, and thus aided the extension of badland topography.Data on the three main incisions in Musinambi Gully, based on themeasurements using 8x enlargements of the original 1:25 000 photographs, aresummarized in Table I. Headcut 1 represents the incision above the dam wall,which had been breached by 1956 but had not been subjected to gullyingupstream of the breach. Consequently, the estimated 6,5 m advancement of thisshallow headcut between 1956 and 1964 probably took place towards the end ofthis period, the material excavated from above the dam wall forming a smalloutwash fan below the breach (Fig. 5B). Over the period 1956-86 this headcutextended 43,9 m, being especially active in the two very wet seasons during1984-6, with an average rate of extension of n early 1,5 m per year; th is is a muchslower rate of advancement than the other two incisions (Table I). The mainincision in the gully, Headcut 2, was not visible on the 1956 and 196420GULLY FORM AND DEVELOPMENTFigure 5: PLAN FORM CHANGES IN MUSINAMBI GULLY ANDMUZHANJE GULLY. 1956-1984.HŠ-iMUSINAMBI GULLY = A to EMUZHANJE GULLY =FtoJ-O -2Ł-'- 4-11-12-14-15R. WHITLOW & C. FIRTH 21Key to symbols used to identify features ongully cross-sections in Figure 51 well-defined gully margin 9 boundary of croplands2 poorly defined gully margin 10 old cultivation furrows3 shallow channel 11 homesteads4 seepage line along old channel 12 kraals (cattle pens)5 residual surfaces within gully 13 main tracks6 badlands dissection 14 former dam wall7 entrenched gully infill IS large clumps of trees8 seepage zonephotographs, possibly because of the clogging of the channel by sediment erodedfrom the breached dam. Between 1971 and 1986, this headcut advanced 76,9 mat an average rate of 5,1 m per year, being most active in the 1984-6 period ofheavier rains. The fact that this headcut is nearly twice the height of Headcut 1and more prone to the effects of undercutting partially accounts for the differencesin the rates of advancement of the upper incisions (Stocking, 1981).The trench in the lower part of the gully system is a result of upstreammigration of a very active headcut or knick point. This was visible in the streamchannel below Musinambi Gully in 1971, entering the lower reaches of the gullyby 1976 (Fig. 5D). Since 1971 the headcut has advanced some 311 m up the gullyat an average rate of 19,4 m per year, nearly four times the rate of the mainheadcut (Headcut 2). This higher rate of headcut advancement is probably theproduct of two factors. Firstly, the infill sediments are very poorly consolidatedand thus more easily eroded than the surface colluvium and weathered Karoo.Secondly, the headcut in the infill sediments advances owing to channel flowrather than piping. Once a vertical headcut has formed in such infill it is likely tomove upstream very quickly. Consequently, even the stabilized gullies could beaffected by renewed erosion in the future.Muzhanje Gully: The changes in plan form of this gully are shown in Figure 5F-Jand details on the gully margins and the two main headcuts in the upper part ofthe gully are summarized in Tables II and III, respectively.As in the case of Musinambi Gully, there is evidence of old cultivationfurrows within close proximity of the gully. These may well have assisted in theformation of lateral headcuts along the left bank in the lower reaches of MuzhanjeGully (Fig. 5F and G), but these headcuts are now stabilized and barely visible onthe aerial photographs. More crucial in terms of runoff and erosion has been thegradual encroachment of ploughed fields into the valley bottom and theassociated restriction of tracks along the margi ns of the gully during the 1960s andTable IINCISIONS IN MUSINAMBI GULLYPeriodFrom 1956 lo 1964From 1964 to 1971From 1971 lo 1976From 1976lo 1984From I984lo 19X6*TOTALS/MEANHeadcul 1Total headwordextension (mi6,517,81.69,78,343,9Rate ofextension(m/year)0.82,50,31,24,11,5Headait 2Total headwordextension (mi16,242,218,576,9Rate ofextension(m/year)3,25,39,25,1Lower trenchTotal headwordextension (m)70,0180,061,0 t311,0Rate ofextension(mlyear)14,022.520.319.4* Data derived from Figure 3.t Period ol three years 19X4 to 19X7.ac|-r~-nOD>OomOmzTable IICHANGES IN BANK LENGTHS OF MUZHANJE GULLYYear19561964197119761984INCREASE1956-84Main gullxaxis fm)7.18780796809839101Estimated lengthof gullxmargin tin)13881635189816612070682Riflht ImnkRatio of lengthof main gullyaxis to gullymargin:l.88:2.l():2.38:2.O5:2.470.59*Estimated lengthi>t gullymargin tin)92094399110341098178Left hankRatio of lengthof main gullyaxis to gullymargin.25.21.24.28.310.06*3J£IZ\rO<OT3)x* Increase in the sinuosity ratio.Table IIIINCISIONS IN MUZHANJE GULLYocPeriodlleadait 1Main gullyHeadcut 2Ri^hl bank tributaryTotal headwordextension tintRale of extension(mlyear)Total headwordextension (m)Rale of extension(mlyear)Irom 1956 to 1964 42.2Irom 1964 lo 1971 16.2(Łtorn 1971 lo 1976 13.0Irom 1976 lo 19X4 29,2From 19X4 lo I9X6:: 5.55,32,32,63,72.822.X26,09.X26,01.92.83.71.93.21.0TOTAl.S/MKAN 106.13.5X6.52.9o2>zDOmOTJmI).II.i tlcn\t'tl llum lR. WHITLOW & C. FIRTH 251970s (Fig. 5G, H and I). Extension of croplands at the expense of grazing areascertainly contributed towards growing pressures upon and accelerated degra-dation of valley sites in the Communal Lands during these years (Cleghorn, 1966;Whitlow, 1979). As a result of this erosion, peasant farmers since Independence(1980) have been persuaded to abandon fields close to gullies as shown here forMuzhanje Gully (Fig. 5 J). This may slow down but is unlikely to prevent furthergrowth of this gully.It was not always easy to determine the exact position of the gully marginseven on aerial photo enlargements. This was partly because of small trees andshrubs growing along the edges of the gully and partly because of degradation ofthe banks by livestock. Consequently, the configurations of plan forms of thegully between 1956 and 1984 (Fig. 5F-J) are approximations. Changes in gullymargin over this 28-year period are given in Table II. The poorly defined,generally stabilized lateral gullies along the lower reaches of the left bank wereexcluded from these measurements since consistent identification of the marginsof these channels was not practicable. A sinuosity ratio (SR) based on comparingthe length of the main axis of the gully against gully margin provides an index ofthe irregularities of the right and left banks. The greater this SR value, the moreindented the gully bank.The left bank, allowing for the exclusion of the stabilized tributary gulliesmentioned earlier, is more regular than the opposite bank with relatively low SRvalues between 1,21 and 1,31 (Table II). Over the 1956-84 period the left bankincreased in length by only 178 m, mainly owing to the development andextension of two tributary gullies in the upper reaches of the main channel. Thelower of these tributary gullies was poorly defined on the 1976 and 1984 aerialphotographs (Fig. 51 and J) owing to the excessive trampling by cattle.In contrast, the right bank is much more sinuous with SR ratios between 1,88and 2,47. This is mainly because of the occurrence of three large tributary gulliesand numerous smaller lateral headcuts along this bank. The gully marginincreased in length by 682 m over the period 1956-84 with the extension of theseside channels, and the resulting increased irregularity of the margin is shown by anincrease of 0,59 in the SR value. The slight decrease in the length of the right bankin 1976 can be attributed to coalescence of headcuts, leaving isolated ridges andknolls within the main gully (Fig. 51).Over the 30-year period, 1956-1986, the main channel (Headcut 1 in Table HI)advanced 106,1 m at an average rate of 3,5 m per year, the highest rate of extensionbeing recorded during the first eight years of this period. The large right-bank gully,Headcut 2, increased in length by 86,5 m over the same 30-year period at anaverage rate of 2,9 m per year. Both headcuts were under 3,0 m in height andaffected by basal piping, hence the slightly faster rate of extension of Headcut 1 islikely to be a function of receiving greater volumes of runoff than Headcut 226 GULLY FORM AND DEVELOPMENTaccording to the relationships defined by Stocking (1981). When these rates ofadvancement are compared with the equivalent feature in Musinambi Gully(Headcut 2), they are significantly lower. However, the Musinambi gully has aheadcut of 1,4 m in height and a catchment area above the headcut nearly twicethat of the Muzhanje features Š hence the greater runoff entering this shallowerincision explains the higher rate of extension. This view is supported by Stocking's(1981) equations which highlight the role of drainage basin area as a key variableaccounting for differences in advancement of headcuts.CONCLUSIONAs indicated at the beginning of this article, it would be premature to formulate ageneral model on the development of gullies on the Karoo sediments in centralZimbabwe. Nevertheless, this reconnaissance study has shown that there areinteresting possibilities for further research on these gullies, particularly in terms ofevaluating hypotheses on episodic erosion (e.g. Womack and Schumm, 1977) andrates of gully growth (e.g. Graf, 1977). This preliminary survey has also providedsome useful information concerning the distribution, morphology and enlargementof gullies in this part of Zimbabwe.Mapping of gullies shows that these features are restricted almost entirely to thewestern area of the Karoo sediments in the central part of the country. This appearsto be related to differences in mineralogy of the Karoo sediments rather than to thefact that this western area is used for peasant agriculture (Stocking, 1978a). Thegullied areas are characterized by sodic soils produced by weathering of plagioclasefeldspars and poor drainage (Wendelaar, 1976). The eastern area of Karoosediments has more leached fersiallitic soils which are erodible and rarely subject topiping (Thompson and Purves, 1978). Both active and stabilized gullies wereidentified in the Mhondoro-Ngezi area, these constituting just over one quarter thelength of the total drainage network. Whilst active gullies were mainly dendritic inform, linear, compound and bulbous gullies were also recorded. The reasons forthese differences are not yet known.The three gullies chosen for field observations differ markedly in their plan andcross-sectional forms. As an apparently stabilized, infilled channel, ManyeweGully has a broad flat floor and low-angled banks. Musinambi Gully is a shallowgully with pronounced badland erosion along one bank and a gully-in-gully formdue to repeated phases of incision and sedimentation. The deeper and longerMuzhanje Gully is more typical of the gullies in the area, with a more cornpiexcross-sectional form and clear evidence of cut and fill sequences. Field observatj0.of several other gullies in this locality indicates that this pattern of infilling andentrenchment is common, but very little is known, as yet, about the factorsinfluencing these processes.R. WHITLOW & C. FIRTH 27Enlargement of the two active gullies was examined by means of aerialphotography for six different dates since 1956. In the case of Musinambi Gully, thephotos reveal gradual coalescence of areas of badland and differential extension ofheadcuts, dependent on factors such as headcut height and sediment type.Subsurface piping and concentration of runoff from old cultivation furrows andtracks affected both this gully and Muzhanje Gully. This latter gully is asymmetricin form in so far as the right bank has numerous active lateral headcuts while theleft bank is characterized by several stabilized headcuts. The rates of headwardextension on these two active gullies are comparable to those reported elsewhere(e.g. Gregory and Walling, 1973), but the reasons for differences in rates ofextension over the period 1956-86 have not been established. It is likely, however,that these variations are a result of rainfall fluctuations although the paucity ofgauging stations in this part of the country would make it difficult to verify thispossibility.AcknowledgementsWe would like to thank Mr and Mrs T. J. Roos for their hospitality during fieldwork,and the technical staff in the Geography Department, University of Zimbabwe, forpreparation of the illustrations. The research was carried out during a visit toZimbabwe by Dr Firth which was sponsored by the British Council.ReferencesCLEGHORN, W. B. 1966 'Report on the conditions of grazing in the TribalTrust Land', Rhodesia Agricultural Journal, LXIII, 57-67.DYE, P. J. 1979 'Feasibility of reclaiming and improving sodic soils',Rhodesia Agricultural Journal, LXXVI, 65-8.DYE, P. J. and WALKER, B. H. 1980 'Vegetation-environment relations onsodic soils of Zimbabwe Rhodesia', Journal of Ecology, LXVm, 589-606.GRAF, W. L. 1977 'The rate law in fluvial geomorphology', AmericanJournal of Science, CCLXXVTI, 179-91.GREGORY, K. J. and WALLING, D. E. 1973 Drainage Basin: Form andProcess: A Geomorphological Approach(London, Arnold).HANNAMJ.D. 1983 'Gully morphology in a Bathurst catchment', Journal ofSoil Conservation: New South Wales, XXXLX, 156-67.IMESON, A. C. and KWAAD, F. J. P. M. 1980 'Gully types and gullyprediction', Koninklijke Nederlandse Akademie van Geografie: GeografischTydschrift, XTV, 430-41.IRELAND, A. A., SHARP, C.F.S. and EARGLE,D.H. 1939 Principles of GullyErosion in the Pedimont of South Carolina (Washington, U.S. Dept. ofAgriculture, Technical Bulletin 63).28 GULLY FORM AND DEVELOPMENTPRICE-WILLIAMS, D., WATSON, A. and GOUDIE, A. S. 1982 'Quaternarycolluvial stratigraphy, archaeological sequences and palaeoenvironment inSwaziland, Southern Africa', Geographical Journal, CXLVIII, 50-67.SMITH, B. J. 1982 'Effects of climate and land-use change on gullydevelopment: An example from northern Nigeria', Zeitschrift fur Geomor-phologie, XLIV, 33-51.STOCKING, M. A. 1977 'Erosion of Soils on Karoo Sands in CentralRhodesia with Particular Reference to Gully Form and Process' (London,Univ. of London, Ph.D. thesis).STOCKING, M. A. 1978a 'Relationships of agricultural history and settlementto severe soil erosion in Rhodesia', Zambezia, VI, 129-45.STOCKING, M. A. 1978b 'Remarkable erosion in central Rhodesia',Proceedings of the Geographical Association of Rhodesia, XI, 42-56.STOCKING, M. A. 1978c 'Interpretation of stone-lines', South AfricanGeographical Journal, LX, 121-34.STOCKING, M. A. 1981 'Causes and prediction of the advance of gullies', in'Proceedings of the South-East Asian Regional Symposium on Problems ofSoil Erosion and Sedimentation' held in Bangkok, 27-29 Jan. 1981,37-47.THOMPSON, J. G. and PURVES, W. D. 1978 A Guide to the Soils of Rhodesia(Salisbury, Rhodesia Agricultural Journal, Technical Handbook 3).WENDELAAR, F. B. 1976 'Field identification of sodic soils', RhodesiaAgricultural Journal, LXXII, 77-82.WHITLOW, J. R. 1979 'A scenario of changes in subsistence land use and itsrelevance to the tribal areas of Zimbabwe', Zambezia, VII, 171-90.WHITLOW, J. R. 1987 'A national soil erosion survey for Zimbabwe'Journal of Soil and Water Conservation, XLII, 239-42.WOMACK, W. R. and SCHUMM, S. A. 1977 'Terraces of Douglas Creek,northwestern Colorado: An example of episodic erosion', Geology, V, 72-6.