16 AGRICOLA 2007

THE DIET SELECTED BY FREE-RANGING BEEF CATTLE ANDITS EFFECT ON THE CONDITION OF A SEMI-ARID SAVANNA IN NAMIBIA

A. ROTHAUGE1, G. N. SMIT2 and A. L. ABATE3

1Neudamm Agricultural College, Private Bag 13188, Windhoek, Namibia2Department of Animal, Wildlife and Grassland Sciences, University of the Free State,

P.O. Box 33�, �301 Bloemfontein, South Africa3Department of Animal Science, University of Namibia, Private Bag 13301, Windhoek, Namibia

SUMMARY

A long-termsystems trialsetup in1984 todetermine theoptimal stocking rate of beef cattle in the camelthornsavannaofeast-centralNamibiawasusedtodeterminedietselection of free-ranging cattle, and its effects on animalperformanceandrangelandconditionduringthreehot-wet,twocold-dryandonehot-dryseasonfrom2001to2003.

Two cattle frame sizes (CFS), the small-framed purebredSanga and the large-framed Afrikaner x Simmentalcrossbreed, and four systematically increasing stockingrates(SR),fromlow(15kgcowmass/ha)tohigh(45kgcowmass/ha),werecombined ina2x4 factorialdesign.Dietselection of cows was observed directly by bite-counting,replicated for cow, time of day, day and season. Dietaryabundance of forage species was calculated from bitestaken, and principal forages were identified. Rangelandcondition was determined by measuring canopy cover ofthesoil,botanicalcomposition,totalherbaceousyieldandthetuftvitalityofsixindicatorgrassesineverytreatmentplotbeforegrazingcommenced.

The dietary preference of the cattle was calculated forevery forage species by comparing its dietary to itsbotanicalabundance.Thenutrientcontentanddigestibilityof herbaceous forage was determined from randomlycollected samples, and compared to samples collectedfromeachforagespeciesindividuallybyhand-pluckinginamannerimitatingtheselectivityofcattle.Assumeddietarynutrientcontentwascalculatedfromthedietaryabundanceand nutritive value of each forage species, and related toanimalproductivity.DuringstatisticalanalysisusingGLMprocedures,datawaspooledforCFS,SRandseason.

Of all treatments, SR had the greatest effect on the dietselected by cattle and the condition of the rangeland. Itwasmostlyseasonthataffectedtheformer,whileCFShadlittleeffectoneither.Theprincipalforagespeciesweretheperennial grasses Schmidtia pappophoroides, Anthephora pubescens and Eragrostis lehmanniana. Together, theycontributed59%,onaverage,tocattlediet,butasmuchas74%atthelowestSR.ThesegrassspecieswerealsohighlypreferredforagesbuttheirutilisationdependedonSR(P<0,01). The utilization of forbs and woody plants increasedathigherSR(P<0,01)andduringdrierseasons(P<0,01).Small-framed cattle were less dependent on the highlypreferred grass species than were large-framed cattle

(P<0,01)andwerebetterabletoexploittheavailableforageresource.Thismayhavecontributedtotheirhigherfertility(P<0,05)andsystemsproductivity(P<0,01)comparedtolarge-framedcattle.

Thebotanicalabundanceand tuftvitalityof thepreferredgrassspeciesdeclinedwithincreasingSR,whilethatoftheless-preferredgrassspeciesincreased(P<0,01).Preferredgrassspeciesdifferedintheirtoleranceofgrazing(P<0,01)andonlyS. pappophoroideswasabletomaintainasizeablepresenceintheswardevenatthehighestSR.ThebotanicalabundanceofwoodyplantswashighestatthehighestSR(P<0,05),especially thatof theknown invasivespecies(P<0,01).TheeffectsofseasonandCFSonrangelandconditionweremuchsmaller.

The nutritive value of imitated forage samples was muchhigher(P<0,01)thanthatofrandomherbaceoussamples,especially in crude protein content, indicating that cattlewereabletoselectamorenutritiousdietthantheaverageofthevegetationonoffer.ThisabilitydecreasedathigherSR(P<0,05)andduringdrierseasons(P<0,01),whereasCFShadlittleeffect.Theassumeddietarynutrientconcentrationappeared adequate, especially for small-framed cattle,except that the completely inadequate forage phosphoruscontentandgrosslyimbalanceddietarycalcium:phosphorusratiorequiredsupplementation.

Inconclusion,cattlecouldpursuetheirdietarypreferencesat low SR only, and were forced to select previously less-preferred forage species at high SR. This resulted in aless nutritious diet and reduced animal performance; thiswas more pronounced during drier seasons and in large-framedcattle.FromanSRof25kgcowmass/ha,changesinthedietselectedbycattleinducedchangesinthespeciescompositionofthegrasssward,reducedthevitalityoftheindicatorgrassesandreducedtheproductivityofindividualanimals. The threshold towards bush-encroachment wasapproached at an SR of 45 kg cow mass/ha, and this SRshouldthusnotbeexceeded,eventhoughtheproductivityofthebeefsystemcontinuedtoincreaseacrossallSR.

TheperennialgrassS. pappophoroideswasagoodindicatorof the change in rangeland condition brought about byforaging cattle in the camelthorn savanna of Namibia. Itis recommended that stockbreeders pursuing optimumindividual animal production should limit their stockingrateto25kgcowmass/ha,whereasthiscanbeincreasedto

AGRICOLA 2007 17

45kgcowmass/haiftheaimistooptimisebeefproductionon the ranch while maintaining an acceptable rangelandcondition.

INTRODUCTION

Namibia is one of the driest and most sparsely populatedcountries in southern Africa, and very dependent on thelucrativeexportof range-fedbeef to theEuropeanUnion.However, the number of ranched beef cattle in Namibiahas declined by 50 % since the 1960s (Rawlinson, 1994),as a result of degradation of its semi-arid savannas into abush-encroachedstate.Bushencroachmentoccurswhenafew,butusuallyonlyoneof the indigenouswoodyspeciesopportunisticallyexploitsconditions,suchastheweakeningofthegrasslayer,thatarefavourabletoitandsoenableittodevelopgraduallyintoverydensestandsthatdominatetheremaining vegetation (Smit, Richter and Aucamp, 1999).Thisencroachmenthasreduced thegrass-basedcarryingcapacityofNamibiansavannasby20 to90% (AdamsandWerner, 1990; Bester, 1998). While the symptoms andtreatmentofrangelanddegradationhavebeenresearchedintensively, many of the underlying ecological processes,such as the interaction between free-ranging domesticruminantsandthevegetation,arenotproperlyunderstood(Rothauge,2000;Ward,2005).

Dietselectionofdomesticruminantsisatthenexusoftheplant–animalinterface(Forbes,1995).Itreferstotheabilityofananimaltoselectfromallthefoodsonofferthosethatitneedstosatisfy itsbodilyrequirementofnutrients.Thecomposition of the diet depends on factors relating to theforaginganimalandtotheforageresource(Forbes,1995).The ability of a free-ranging herbivore to select adequatefoodfromthevarietyofplantsonofferiscrucialtoitswell-being (RogersandBlundell, 1991),determines its levelofproductionandreflectsitshabitsandhabitat(Milne,1991).

At the landscape level, the choice of where to forage is aspatial one, whereas at the microsite level, the choice isbetween different forage species and even different partsofthesameplant(Stuth,LyonsandKreuter,1993).Inthesemi-arid savannas of Namibia, free-ranging beef cattleface a huge choice of potential forage species: more than4200taxaofplantsarefoundontherange(Craven,1999),including 391 different species of grass (Klaassen andCraven,2003).Foraging animals select preferred and principal foods.Preferredfoodsarethosethatananimalconsumesfirst ifgivena choice.Theyareproportionallymoreabundant inthedietthaninthefeedingarea(Petrides,1975).Principalfoodsarethosethatcontributemosttothetotaldietofananimal(Grunow,1980)andmaynotnecessarilybepreferred(Petrides,1975).Thedietarypreferenceratio(DPR)istheratiobetweentheabundanceofaplantspecies in thedietand its abundance in the herbage (Petrides, 1975; Senft,1989), andenablesus to rank foragespeciesaccording totheir preference. Forage species that are proportionallyusedmore frequently than theyoccur in the feedingarea

(DPR>1) arepreferred; those taken less frequently thanthey occur (DPR < 1) are not preferred or are avoided,while those taken in roughly thesameproportionas theyoccur(DPR~1)reflectaneutralappetiteonthepartoftheanimal.

Themannerinwhichplantsaredefoliatedbyanimals,thefeeding habits of animals and the diet they select exerta shaping influence on the characteristics of savannavegetation (Owen-Smith, 1999). African savannas co-evolvedwithherbivory(Skarpe,1991)andalthoughgrazingisnottheonlyparameterthatshapesvegetation(O’Connor,1994),itisamajorfactorinvegetationdynamics(Clements,1928;Westoby,WalkerandNoy-Meir,1989).Theconditionofarangeland,relatingtosomefunctionalcharacteristicoftherange,suchasitsproductivityorbotanicalcomposition(Tainton, 1999), also reflects the effect of herbivory onvegetation.

Trends in rangeland condition are commonly used todeterminewhetheranimalshaveapositiveornegativeeffectonthevegetation.Asdietselectionrepresentstheinterfacebetween animal production and vegetative processes thatreact to defoliation (Emmans, 1991; Prache, Gordon andRook,1998),itisinfluencedbythestockingrateofanimals,which determines the intensity of defoliation of plants aswell as the competition between foraging animals (JonesandSandland,1974;Skarpe,1991).

In Namibia, an existing systems trial of 20 years, whichsoughttodeterminetheoptimumstockingrateofdifferenttypes of beef cattle for beef production to be sustainablein a semi-arid savanna (Kruger, 1998), offered an idealopportunity to investigate the plant – animal interfacein greater detail. This trial served to elucidate the dietselectedbycattleanditseffectonrangelandcondition.Theobjectiveofthediet-selectiontrialbetween2001and2003wastoquantify thedietof free-rangingbeefcattleand itsnutritivevalue,aswellas thereactionof thevegetationtograzingandhowtheseparameterswereaffectedbygrazingpressure,differenceincattletypeandseasonoftheyear.

MATERIALS AND METHODS

Trial site

The long-term systems trial was initiated in 1984 on 5516 ha of the Sandveld Research Station (22º00,237’Sand19º09,226’Eatanaltitudeof1523mabovesea level)andwasterminated in2004.Thestation issituated inthecentralKalaharicamelthorntreesavanna,whichisthemostimportant commercial beef-producing region of Namibia(Rawlinson,1994).Ithadreceived392±182,4mmrainp.a.since1984.Typically,rainfallishighlyvariableintimeandspace.Theclimateischaracterisedbyashorthot-wet(HW)seasonwithavegetativegrowingperiodofonly48daysfromJanuaryonwards,followedbyacold-dry(CD)seasonwithfrostoccurringuntilAugust,andahot-dry(HD)seasonofvariablelengthfromSeptemberonwardsuntiltheadventofthesummerrains.

18 AGRICOLA 2007

Experimental treatments

Thelong-termsystemstrialwasa2x4factorialexperimentoftwocattleframesizes(CFS)andfourstockingrates(SR),resultingineightindividualtreatments.ThetwoCFSswerethe small-framed, purebred Sanga (S) and the relativelylarge-framedAfrikanerxSimmentalrotationalcrossbreed(L),weighing381±41,6kgand506±67,0kg,respectively(P < 0,01). The SR treatments were based on cow mass/haandstartedatalow(L)rate,increasingsystematicallytoalow-medium(LM),medium-high(MH)andhigh(H)SR (Table 1). The targeted SR was achieved by fixing thenumberofcowsinatreatmentherdattheoutsetofthetrialin1984,andtheactualSRthusfluctuatedaroundthetarget.

Table 1. Targeted and actual stocking rates (in kg cow mass/ha and ha/LSU) of the eight treatments, achieved by fixing the number of cows in a treatment herd

TreatmentTargetedstocking

rate

Herdsize

Actualstocking

rate

(CFS–SR) (kg/ha) (ha/LSU)

(number of cows) (kg/ha) (ha/

LSU)L-L 15 30,0 18 17,7±0,65 25,4

L-ML 25 18,0 28 27,6±0,67 16,3L-MH 35 12,� 40 35,8±0,12 12,6L-H 45 10,0 52 44,1±0,16 10,2S-L 15 30,0 25 17,2±0,32 26,2

S-ML 25 18,0 42 2�,1±1,11 15,5S-MH 35 12,� 60 40,5±0,10 11,1S-H 45 10,0 78 4�,8±0,0� �,0

Animalmanagementandthepreventivehealthroutinewereidenticalamongtreatments,aswasthebreedingandcullingpolicy.Cowswereweighedmonthlyafterovernightfasting,while thebodyconditionscore (BCS)wasdeterminedona5-pointscaleatfourcriticalstagesofacow’sproductioncycle,viz.beforethematingseasonstarted(December)andagainatitsend(April),whencalveswereweaned(July)andwhencowsstartedtocalve(October).Weanerswereraisedoutsidethegrazingareaofatreatment,butatanSRsimilarto that of their treatment-of-birth. Pregnant replacementheifers were returned to their treatment-of-birth shortlybeforecalving.Treatmenteffectsthusaccumulatedoverthegenerationsofcattle.

Rangeland management was also identical among thedifferent treatments, save for the SR. Each of the eightindividual treatments was allocated a grazing area of 690±4,4ha,subdividedintosixcamps(paddocks)distributedrandomlyovertheranch.Cattlewererotatedthroughthesesixcampsonasetcycleof7days’occupation followedby35days’restduringtheHWseasonand14days’occupationfollowedby70days’restduringthedryseasons.Drinkingwater was freely available at all times, as was a minerallickduringtheHWandaprotein,energyandminerallickduringthedryseasons.Fireandarboricideswereexcludedcompletelyasrangelandmanagementtools.

At the outset of the long-term systems trial, cows wereblocked for age and parity, and camps for condition and

productivity, so that all treatments started from the samebase. Beef-production parameters, such as calving andweaningratesandweaned-massproductionperarea,wererecorded routinely for every individual animal and everytreatment.

Duration of the diet-selection trial

Thediet-selectiontrialwasconductedduringthelaststageofthelong-termsystemstrialandconsistedofsixseasonalexperiments, viz. in the HW seasons of 2001, 2002 and2003,theCDseasonsof2001and2002andtheHDseasonof2002.Initially,thediet-selectiontrialhadbeenscheduledfortheHWandCDseasonsonly,butwhenitwasrealisedthatthenadirofgrazingconditionswasreachedintheHDseasononly,oneoftheCDseasonswasscrappedbelatedlyinfavourofasingleHDexperiment.

IntheHWseason,experimentswereconductedinMarch,whenthevegetationhaddevelopedmaximallyasmostoftherainshadalready fallen.Cowsat thisstageweresucklingcalvesandwerejoinedbythebulls.ExperimentationduringtheCDseasonwastimedtocoincidewiththecoldesttimeof the year, June – July, shortly after the calves had beenweaned off their pregnant dams, while experimentationoccurredinOctoberduringtheHDseason,whencowswerein late pregnancy. To minimise intra-seasonal variation,every one of the six seasonal experiments was completedwithinaperiodoffourweeks.

Treatment plots

Onlyonecampof thesixallocated toeach treatmentwasselected to serve as the treatment plot in which the diet-selection experiment was carried out. To eliminate inter-campvariation, thesamecampwasusedevery time.Theeight treatment plots averaged 142 ± 28,9 ha. They wereselected on the basis of their soil and vegetation type aswellastheirproximitytoeachother.Treatmentplotsweredominatedbyred,sandyHuttonsoils,whicharemorefertileandthussupportalargerbotanicaldiversityandherbaceousproductionthanthealternative,thewhite-greysoilsofthecentral Kalahari (Scholes, Dowty, Caylor, Parsons, FrostandShugart,2002).

Treatmentplotswererectangular,withawateringpointinone corner. A line transect was fixed by GPS coordinatesfromthewateringpointtothediagonallyoppositecornerofthecamp.Theeightdiagonallinetransectsaveraged1535±251,8minlengthandwereusedduringbotanicalsurveysbeforeandaftergrazingofthetreatmentplots.

Methodology of diet-selection observations

Dietselectionofcowsinthetreatmentplotswasobservedon four consecutive days per treatment, twice during themorningandtwiceduringtheafternoon,whencattle,whichare crepuscular feeders, are most active (Albright andArave,1997).Eachtime,sixcowswereselectedatrandomfromatreatmentandobservedatclosequarters(<5m)for

AGRICOLA 2007 1�

acontinuousperiodof10minutespercow.Observedcowswereidentifiedforretrospectivecouplingtotheirproductiondata and life history. All observations were performedwithinthefirsthalfoftheperiodofplotoccupation,becausetheplotatthisearlystageofoccupationstillofferedanimalsthe maximum choice of forage. Also, forage plants hadnot yet been defoliated to an extent that would limit theiridentification.

During observation, all bites taken by the cow werecounted and all plants utilised were identified, as werethe plant parts utilised, by following the bite-countingmethod developed by Narjisse (1991) for goats but foundby Ortega, Bryant and Drawe (1995) to be applicable tocattle too. The dietary abundance of every forage specieswascalculatedasapercentagebasedon the frequencyofitsoccurrenceinthediet.TheDPRofeveryforagespecieswascalculatedfromitsabundanceinthedietandvegetationrespectively. During bite counting, the height at whichforage was utilised was determined as being either aboveorbelow120cm,whichcorrespondsroughlywiththehead-heightofcattle.Thehabitatoftheutilisedgrasseswasalsodetermined, i.e.whether theywereutilised fromtheopenor fromunderneath thecanopyof a leguminousoranon-leguminouswoodyplant, inrecognitionof the importanceoftheassociationbetweensavannatreesandgrasses(SmitandSwart,1994).

Methodology of the botanical surveys

A botanical survey was conducted in every treatmentplot the day before cattle entered it. The survey wasconductedalong thediagonal line transect, andconsistedof determining the botanical composition, canopy coverof the soil and herbaceous productivity. The botanicalcompositionwasdeterminedbysystematicpointsampling(Tothill,1987),usingafree-falling,3m-longrodofsteeltoindicate accurately the point of impact. The plant whosecanopy covered the point of impact was identified or, if aplantcanopydidnotcoverthepoint,thenearestplantwasidentified.

The botanical abundance of plants was calculated as apercentage based on the frequency of occurrence in thepointsurveys.Multiplereadingsatonepointwerepossibleif a small plant at that point occurred under the canopyof a larger plant. In this manner, grasses were classifiedaccording to theirsub-canopyhabitat,occurringeither inthe open habitat or under the canopy of a leguminous ornon-leguminous woody plant. The height of woody plantswasalsorecordedasbeingeitheraboveorbelow120cm.

Theproportionofpointsof impactthat fellonbaresoiloron soil covered by a plant canopy was used to calculatethecanopycoverof thesoil.Herbaceousproductivitywasdeterminedbyclipping40quadratsof1m²spacedregularlyalong the diagonal transect. All rooted herbaceous plantswithin the quadrats were clipped at ground level (Bester,1988) and the material sorted immediately into tenfractions:

• the separate yields of six perennial grass species thathad been chosen to indicate rangeland condition, viz.Anthephora pubescens, Aristida stipitata, Brachiaria nigropedata,Eragrostis rigidior,Schmidtia pappophoroidesandStipagrostis uniplumis,

• theyieldofallotherperennialgrasses,• theyieldofallannualgrasses,• theyieldofalldicotyledonousherbsandforbsand• allmoribundherbaceousmatter.

Total herbaceous yield was obtained from the sum of thetenyieldfractions.Duringclipping,thenumbersoftuftsofeachofthesixindicatorgrassspeciesoccurringinsidethequadratswerecounted,tocalculatethetuftdensityandtuftyield of the indicator grasses, which indicated the vigourofthesegrasses.Theyieldfractionswereweighedaswas,sampledandthesampledriedtoconstantmassinaforced-draught oven at 65 °C to determine its dry matter (DM)content. The six samples of the indicator grasses wereretainedforlaboratoryanalysis.

The same clipping and tuft counting procedure wasrepeated in every treatment plot immediately after cattlehad completed their period of occupation of the plot.The difference between the before- and-after grazingmeasurements was presumed to be due to the grazing.Rangeland condition was deduced from the botanicalcompositionofthetreatmentplots,thecanopycoverofthesoil,herbaceousDMyieldand thevigourof the indicatorgrasses.

Forage sampling and analysis

Dried samples were obtained from the six indicatorgrass species from every treatment plot before grazingcommenced, as explained above. In addition, a seventhsample, representing the herbaceous bouquet on offer inthe treatmentplots,wascollected fromtheyield fractionsobtained during clipping, by reconstituting a compositesampleproportionaltothemassofthefractions,excludingonly the moribund herbaceous matter. All dried sampleswere ground through a 1 mm sieve and subjected tostandard laboratory analysis of their nutrient content anddigestibility(Menke,Raab,Salewski,Steingass,FritzandSchneider, 1979; Robertson and Van Soest, 1981; AOAC,1995).Sincethesesampleswerecollectedbyharvestingatground level, theyrepresented the totalnutrientsonofferin the herbaceous layer of the savanna and were termed‘random’samples.

After every seasonal bite-counting observation, the sixindicatorgrassspeciesandthesixprincipalforagespeciesin every treatment were sampled by hand-plucking in amanner imitating the selectivity displayed by cattle andin a manner representing the sub-canopy habitats – opensub-canopy (O), sub-canopy under a leguminous tree orbush (L) and sub-canopy under a non-leguminous tree orbush(NL)–fromwhichtheywereselectedbythecattle.Inaddition,everyotherforagespeciesthathadbeenutilisedbuthadnotyetbeensampledwassampledonce,fromany

20 AGRICOLA 2007

treatment,inamannerimitatingtheselectivitydisplayedbycattle.All imitatedsampleswerecollectedinthemorningto prevent diurnal variation in their nutrient content, andwere immediately sealed in plastic bags to obtain theirtrue field moisture content. They were then subjected tothesamelaboratoryanalysesas therandomsamples,andrepresented the nutrients selected by the cattle from theherbaceousbouquetonoffer.

Statistical analysis of the data

AlldataentrieswerepooledforthethreemaintreatmentsofCFS,SRandseasonoftheyear.Allpercentagefrequencydata was first transformed by arcsine, as recommendedby Zar (1999) to eliminate bias due to a large number ofsmall and a small number of large percentages, beforebeingsubjectedtoanalysisofvariance(anova)proceduresof the general linear model (GLM) using the StatisticalProceduresforSocialScientists(SPSS)package,version10(BrymanandCramer,1997).Thealphavaluewassetat0,05andconfidence limits at 95%.Tukey’smultiple range testwas used to evaluate differences between specific meansfurther,whilepartialeta-square(ç2)wasusedtoestimatethesizeofaparticulareffect.

RESULTS AND DISCUSSION

Diet selection of cattle

Thecowswereobservedtohavetaken436953bites,or37,9bites per cow per minute during the six seasonal experi-ments. The vast majority of bites were from the grasses(Table 2), confirming that cattle are grazers (Forbes,1995).Infact,theyarespecialistgrazers,becauseoveralltreatments,threegrassspecies,Schmidtia pappophoroides,Anthephora pubescens and Eragrostis lehmanniana con-stituted59,2%ofthediet.

Table 2 The diet selected by free-ranging beef cattle in a semi-arid savanna of Namibia (arranged in descending order of importance)

Forage group and species Abundance in diet (%)

Grasses: 83,�±11,72Schmidtia pappophoroides 33,7±18,23Anthephora pubescens 14,5±1�,65Eragrostis lehmanniana/E. trichophora * 11,0±10,50Stipagrostis uniplumis 7,5±�,54Melinis repens repens 5,8±5,11Eragrostis rigidior 5,0±6,75All other grasses (18 species) 6,4Woody plants: �,8±10,65Grewia flava/G. flavescens ** 2,6±4,34Tarchonanthus camphoratus 2,6±4,17All other woody species (�) 4,6Dicotyledonous herbs and forbs: 6,3±7,46Nidorella resedifolia 2,8±5,8�All other dicots (20 species) 3,5Total diet (56 species) 100,0

*E. lehmanniana can easily be distinguished in-hand from E. trichophora, but it was impossible to distinguish them at the distance and speed required during diet selection.

These species are the principal diet components of free-ranging beef cattle in the camelthorn savanna of east-centralNamibia.InotheraridsouthernAfricansavannas,S. pappophoroideshasalsobeenreportedtobeaprincipalcomponentofcattlediets(Mphinyane,2001).Cattleutilised24ofthe27grassspeciesoccurringinthetreatmentplots,12ofthe13woodyspeciesand21ofthe25dicots.

Relianceonthethreeprincipalforagegrassesincreasedto74,3 % when cows were stocked at a low rate ( Figure 1),indicating that these principal forage grasses were alsohighlypreferred.Evenat thehighestSR treatment, thesegrassesstillcontributed41,6%ofthediet(P<0,05).Cattlethus relied greatly on a small number of highly preferredgrass species under optimum grazing conditions, as hasalsobeenreportedfromKenya(Odadi,YoungandOkeyo-Owuor,2003)andhadtoadjusttheirdietdrasticallytolessfavourablegrazingconditionsbroughtaboutbyanincreaseintheSR.AtthehighestSR,thewoodyplants’contributionto thediethaddoubled to12,8±11,88%compared to thelowestSR(P<0,05),butthatofthedicotsremainedrelativelyconstantacrossallSRtreatments(P>0,05).

TheseasonoftheyearhadassimilarlyalargeeffectonthecompositionofthedietasSRhad,whereastheeffectofCFSwas small, affecting mainly the dietary abundance of thebulkygrassesEragrostis rigidiorandStipagrostis uniplumis,whichwereutilisedbetterbysmall- thanby large-framedcattle(P<0,05andP<0,01respectively).Cattlerespondedto increasing seasonal aridity by utilising more browsedforageattheexpenseofthegrasses(P<0,01)(Figure2).Inparticular,theshedleavesofAcacia melliferaincreasedfrom0%intheHWdietto4,4±6,06%intheCDdiet(P<0,01).Changes in the dietary abundance of individual foragespeciesweremagnifiedbyalargenumberofsignificanttwo-andthree-wayinteractionsbetweentreatments,indicatingthat the composition of cattle diets is sensitive to factorsundermanagerialcontrol,suchasthestockingrateofcattleandtonaturalfactors,suchastheseasonalityofclimate.

Rangeland condition

Thebotanicalcompositionoftreatmentplotswasbasedon477±68,2surveypointsperplot,safelyexceedingtheminimumnumber required for scientific monitoring (Hardy andWalker,1991).Morethan64speciesofplantswererecordedinthetreatmentplots,butsomeofthedicotyledonousherbscouldbeidentifiedatthegenuslevelonly,e.g.theIndigoferaspp. The rangeland was dominated by grasses (Table 3),ofwhichnearly99%wereperennial.Theprincipal foragegrasses of cattle, S. pappophoroides, A. pubescens andE. lehmanniana, made up 30,7 % of all rangeland plantsacrossalltreatments.

TheSRofcattle(P<0,01;r²=0,95)andtheseasonoftheyear(P<0,05;r²=0,92)hadthegreatesteffectonthebo-tanical composition of treatment plots, whereas the effectof CFS was negligible (P = 0,34; r² = 0,91). The botanicalabundanceofthethreeprincipalforagegrassesdecreasedfrom 43,7 % at the lowest to 20,0 % at the highest SR

AGRICOLA 2007 21

(P<0,01),withonlyE. lehmannianaabletoincreaseinabundance(P<0,01)despitetheincreased grazing pressure (Figure 3). Incontrast,thebotanicalabundanceofthelesspreferred, bulky grasses S. uniplumis andE. rigidiorincreasedfrom12,0%atthelow-estto30,1%(P<0,01)atthehighestSR.Dif-ferentresponsesbyindividualgrassspeciesto grazing had caused a transformation ofthegrassswardovertheyears,byweaken-ingthemorepreferredspeciestotheadvan-tageofthelesspreferredspecies.

Theimplicationisthatrangelandmanagerscannot simply evaluate the abundance of‘grass’ toestimaterangelandcondition,buthavetouseindividualgrassspeciestoobtainagoodindicationoftheeffectofgrazingonthe grass sward. Schmidtia pappophoroidesappears to be very suited to this purpose,as it is a principle and a preferred foragespeciesofcattleandisabundantinpristineand under-utilised rangeland, yet still abletomaintainasizeablepresenceatstockingrates that appear to be viable in ranchingpractice.

Figure 1. The abundance (%) of forage plants in the diet of free-ranging beef cattle as affected by the stocking rate of cattle (*: P < 0,05; **: P < 0,01).

Figure 2. The abundance (%) of forage plants in the diet of free-ranging beef cattle as affected by the season of the year (*: P < 0,05; **: P < 0,01).

Table 3. Botanical composition of a semi-arid rangeland in Namibia (arranged in descending order of importance)

Plant group and speciesBotanical

abundance(%)

Grasses: 71,4±5,1�Schmidtia pappophoroides 21,0±5,33Stipagrostis uniplumis 12,5±5,68Eragrostis rigidior 10,4±4,3�Anthephora pubescens 5,�±7,30Aristida stipitata 5,2±3,32Eragrostis lehmanniana/E. trichophora * 3,8±1,73

Melinis repens repens 3,6±2,�4Aristida congesta 2,2±1,81All other grasses (1� species) 6,8

Woody plants: 13,4±3,21Acacia erioloba 3,�±1,38Dichrostachys cinerea 2,0±1,01All other woody plants (10 species) 7,5

Dicotyledonous herbs and forbs: 15,3±5,44

Tylosema esculentum 5,6±2,43Nidorella resedifolia 2,7±2,21All other dicots (23 species) 7,0

Total (64 species) 100,0*These two grasses were grouped together because of difficulties

distinguishing them during diet selection observations. However, �5 % were E. lehmanniana and only 5 % were E. trichophora.

Figure 3. Botanical composition (%) of the rangeland as affected by the stocking rate of cattle (*: P < 0,05; **: P < 0,01).

22 AGRICOLA 2007

IncreasingtheSRinitiallyhadlittleeffectonthebotanicalabundanceofwoodyplants,butinplotsstockedatthehighestrate,theabundanceofwoodyplants increasedsignificantly(P<0,05) to16,8%,comparedto 12,1 % at the lowest SR. The abundance of known invasive speciessuchasA. melliferaandDichrostachys cinerea increasedby89%atthehighestcomparedtothelowestSR(Figure4).ItthusappearedthattheecologicalthresholdtowardsdensificationofwoodyplantsasaresultofgrazingpressurehadbeenreachedatthehighestSRappliedinthistrial,although the woody component still consisted of a mix of micro- andmacrophyllous, evergreen and deciduous species. Increased grazingpressureiswellknowntochangethebalancebetweenherbaceousandwoodycomponentsofasavanna(Smitet al.,1999).

The major effect of increasing seasonal aridity was to decrease thebotanicalabundanceofherbsandforbs(P<0,01),althoughmostwereperennial, and to increase the relative abundance of E. lehmanniana(P<0,01).ThecanopycoverofthesoilwasaffectedbytheSRofcattle(P<0,05),decreasingfrom81,1%to73,6%,butnotbyseasonoftheyearorCFS(P>0,05).

ThemeanherbaceousDMyieldbeforegrazingwas172,1±39,51g/m²over the six seasonal experiments. It was reduced by 8 % to 158,8 ±36,52 g/m² after grazing. Herbaceous yield before grazing was notaffectedbyCFSandseasonoftheyear(P>0,05),butSRreduceditby21%,from185,5±31,81gDM/m²to153,5±40,89gDM/m²(P<0,05).

The fractionalyieldsof theprincipalgrassesA. pubescensandS. pap-pophoroides declined sharply due to an increase in SR (by 99,5 % and25,5%respectively;P<0,01),while thatof thebulkygrassesAristida stipitata, E. rigidior and S. uniplumis increased just as noticeably (by83%,247%and233%respectively;P<0,01).Inaddition,thedensityandtuftyieldoftheprincipalforagegrassesdeclineddramaticallywithanincreaseintheSR(P<0,01),whereasthedensityandtuftyieldofthebulkygrassesincreased(P<0,01)(Figure5).Therefore,eventhoughthevigourofpreferredforagegrassesdeclinedovertheyearsasgrazingpressureincreased,aswasalsorecordedbyKirkmanandMoore(1995),itwaspartlycompensated forbyan increase in thevigourof the lesspreferred,bulkygrasses.

Although this compensatory effect may have partly obscured theweakening of the grass sward generally, it was still apparent thatincreasingtheSRofcattlereducedtheproductivityandconditionoftherangeland.Duetosignificanttwo-andthree-wayinteractionsbetween

Figure 4. The botanical abundance (%) of woody species as affected by the stocking rate of cattle (**: P < 0,01).

Figure 5. The effect of stocking rate of cattle on the vigour of six grass species, as represented by the yield of DM per tuft (top) and the density of tufts (bottom).

treatment effects on various grass species,the degradation of the range appearedto be worse in plots frequented by large-framedcattleandduring thedrier seasonsof the year. If, therefore, a grass such asS. pappophoroidesistobeusedasanindicatorof rangeland condition, not only should itsabundancebemonitored,butalso theyieldanddensityofitstufts.

Moribund herbaceous matter comprised24,3 % of the total herbaceous yield overall treatments (Figure 6). Its proportionalcontribution to total yield was not affectedby treatment (P > 0,05), but the absoluteyieldofmoribundmatter,ing/m²,declinedwith SR (P < 0,05) and increased duringthe HD season (P < 0,01), emulating theyieldoflivingherbaceousmatter.Thelargeamountofmoribundmatterproducedbytheherbaceous layer of this semi-arid savannadidnotappeartoincreasetheorganicmattercontent of the upper soil layer (Rothauge,Smit and Abate, 2003), possibly because itwasoxidisedabovethesoilsurfacebeforeitsnutrientscouldberecycledintothesoil.

AGRICOLA 2007 23

Preference for forage species

TheprincipalforagegrassspeciesA. pubescens,E. lehman-nianaandS. pappophoroideswerehighlypreferredbycattleirrespectiveof treatment,withanaverageDPRof2,5,2,9and1,6respectively.Theirpalatabilityresultedincattleuti-lisingthesegrassesuniformlytoaresidualstubbleheightthatbecamelowerwithincreasingSR.

ThebulkygrassspeciesE. rigidiorandS. uniplumiswerenotpreferredatanytreatment,andtheirDPRvariedfromvirtually zero at the lowest to close to 1,0 at the highestSR. As soon as the more-preferred grass species wereeatenoutoftheswardatthehigherSR,thebulkygrassesbecame more prominent in the diet of cattle; not becausetheywereintrinsicallypalatablebutmerelybecausemore-preferred grasses were no longer freely available in thetransformedgrasssward.Thebulkygrasseswereutilisedby cattle extremely selectively: large tufts were ignoredwhereas smaller tufts and those utilised before wereutilisedpreferentially(inter-tuftselection),andstemsandleaveswereutilisedonlyonceall inflorescenceshadbeenutilised (intra-tuftororganselection).The thirdcategoryofgrassspecieswasnotutilisedmuchatanytreatmentandthusremainedunpreferred,withaDPRclose tozero,e.g.A. stipitataandEragrostis pallens.

Thedietarypreferenceofcattleforthegrasseswasinfluencedbyseasonoftheyear,withgrassesbeingconsiderablymorepreferred during the HW season, when they were green;and CFS, with large-framed cattle displaying a slightlymorepronouncedpreferenceforgrassesthanthatdisplayedbysmall-framedcattle.Approximately84%of thegrassesoccurred in the open, but as the SR of cattle increased,grasses from the canopied habitats, and especially fromunderneathleguminoustreesandbushes,constituted20%andmoreofthedietofcattle(P<0,01;r²=0,89).However,thiseffectwasdependentontheparticulargrassspecies,asnotallspeciesoccurredinthecanopiedhabitats.

Increasing aridity of the season further enhanced theutilisation of grasses occurring in the canopied habitats

(P <0,01; r²=0,80). In turn,ahighlypreferredgrass such as A. pubescens increasingly foundrefuge in the canopied habitats at higher SR,when up to 63,7 % were in canopied habitatscomparedtoonly10,7%atthelowestSR(P<0,01),indicating that woody canopies offered highlypreferred grasses some protection againstintensive utilisation. Some of the most highlypreferredforagespecieswerewoodyplants,butinmostinstancestheyrepresentednicheforagesthatcontributedtodietonlyoccasionally.

ThemostpreferredofallforageswasA. mellifera,with an average DPR of 4,0, but only its fallenleaveswereutilisedlateintheCDseason.Itwasaseasonallylimitedresourceenrichedbyimportedmattersuchasbirddroppings,spiderwebsandsoil, and was soon exhausted. Grewia flava wasoneofthefewwoodyplantsthatwasutilisedatall

seasonsoftheyear:itsgreenleaveswereeatenintheHWseason,theshed,dryleavesintheCDseasonandthebudsintheHDseasonand,accordingly,ithadahighaverageDPRof2,2.Theheightatwhichcattlebrowsedwasinfluencedbyseasonoftheyearonly.Theevergreen,broad-leavedwoodyplantTarchonanthus camphoratusforexample,wasutilisedmainlyataheightlessthan120cmabovegroundintheHWseason,butasseasonalaridity increased,cattleutilised itprogressively higher up (P < 0,01) until it was completelystrippedduringtheHDseason.

The nutritive value of forage

Altogether1012foragesampleswerecollectedoverthesixseasonal experiments; of these roughly one-quarter werecollectedrandomlyandthree-quartersinamannerimitatingtheforagingselectivityofcattle.Imitatedforagesampleshadamuchhighernutritivevaluethanrandomsamples(Table4,nextpage),indicatingthatcattlewereabletoselectamorenutritious diet from the vegetation than the average thatwasonoffer.Thedifferencebetweenrandomandimitatedsamplesextendedtotheindividualforagespeciestoo.

The nutritive value of the randomly sampled herbaceousbouquetonofferwasseverelyaffectedbytheseasonoftheyear, as the protein (P < 0,01), energy (P < 0,01), mineralcontent (P < 0,05) and digestibility (P < 0,01) of herbagedeclined during the dry seasons, while the fibre contentincreased(P<0,01forNDF;P<0,05forCF).EventhoughtheeffectofSRwasnotsignificant,ittendedtoconsistentlydecreasethenutritivevalueof theherbaceousbouquetonoffer.TheeffectofCFSwasnegligible.

Theeffectofseasononthenutritivevalueofimitatedforagesamples was profound, and only the CF and ADF contentwerenotaffected(P>0,05)byseason.However,theeffectdependedontheindividualforagespecies.Ingeneral, thenutritivevalueofimitatedforagesamplesdeclinedfromtheHWtotheCDseasonbutrecoveredagainslightlyduringthe HD season, especially that of the principal foragegrasses.Thiswasa resultof renewedavailabilityof fresh

Figure 6. Proportional composition (%) of the total herbaceous DM yield before grazing of the treatment plots

24 AGRICOLA 2007

regrowth when the increase in temperature and daylightlengthcausedperennialgrasses tobreak theirdormancy.Since the sprouting is driven by stored reserves withinthegrass, it isunsustainableuntil the rainyseasonstarts(WolfsonandTainton,1999).

Incontrasttoitseffectonrandomsamples,SRsignificantlyaffectedthenutritivevalueof the imitatedforagesamples(Table5).ImitatedsamplescollectedinplotsstockedatthelowestSRcontained27%moreCa (P < 0,01), 6% lessCF(P<0,05),6%lessNDF(P<0,05),26%morefat(P<0,01)and6%moreME(P<0,01),andwere4%moredigestible(P < 0,05) than samples collected in plots stocked at thehighestSR.Thus,cattlestockedatalowSRwerebetterabletoexploitthenutritionalresourceonofferinthevegetationthanwerecattlestockedatahigherSR.ThiswasduemainlytotheeffectofSRongrasses,whoseP(P<0,05;r²=0,02),ADF(P<0,01;r²=0,03)andashcontent(P<0,01;r²=0,03)reactedsignificantlytoSR,asdidtheabovenutrients.EventheCPcontentofimitatedgrasssamplestendedtodecreasewithanincreaseintheSR,butincreasedagainatthehighestSR (P > 0,05), indicating that at the highest SR treatmentgrasses were grazed right down into the reproductivetilleringpartatthebaseofthetuft.ThiswasalsorecordedwhenseveregrazingwasappliedbyMufandaedza(1977),and it is highly detrimental to continued productivity andsurvivaloftheperennialgrasstuft.

Incontrast,SRhadasmalleffectonthenutritivevalueofwoodyplants,affectingonlytheirADF(P<0,05;r²=0,14),NDF (P < 0,05; r² = 0,12) and ash content (P < 0,05; r² =

0,13), and had no effect on dicots. Imitated samples fromwoodyplantsanddicotswerealsounaffectedbytreatmentinteractions, as was the case for many grass species,indicatingthatitwasthecombinationoftreatmentsthatputpressureongrasses.

Grasses from canopied habitats were utilised morefrequentlybycattlethatwereunderstressfromahighSRand/orthearidityoftheseasonthantheywereutilisedbycattlethatdidnotexperiencesuchstress.Grassesoccurringinthecanopiedhabitatshadasignificantlyhighernutritivevaluethanthoseoccurringintheopen,eveniftheywereofthesamespecies.Grassesfromtheleguminoushabitat,forexample,contained22%moreCP(P<0,01),12%morefat(P<0,01),2%lessCF(P<0,05)and3%lessNDF(P<0,01) than grasses from the open, indicating clearly that thegrass–treeassociationsocommonofsemi-aridsavannasextendednutritionalbenefits to foragingcattle (Rothaugeet al.,2003).

Animal production

When thechanges in thecompositionof thedietofcattlewerecombinedwithchangingnutritivevalueoftheutilisedforage, it was obvious that cattle stocked at a high rateexperiencedaworsenutritionalstatus thancattlestockedata lowerrate,and that thiseffectwasaggravatedby theincreasingaridityoftheseason.Inaddition,cattlestockedat a high rate had less herbage at their disposal in thetreatmentplotsthancattlestockedatalowerrate,althoughthis trialdidnotattempt toquantify theamountof forageactuallyingestedbycattle.AdropinanimalproductionwasthusexpectedatthehigherSRtreatments,especiallysincethenutritionalnadircoincidedwiththeweaningofcalvesintheCDseasonandrenewedcalvingintheHDseason.

Over the three years of the diet-selection trial and inaccordance with results obtained by Els (2002) over themuchlongerperiodofthesystemstrial,theincreaseinSRresulted in deteriorating performance of individual cows(Figure 7). As SR increased from lowest to highest, cowsbecame11%lighter(P<0,01),13%lowerinBCS(P<0,01),3%lessfertileintermsoftheircalvingrate(P<0,01)andinter-calvingperiod(ICP,P<0,01),whiletheaverageageofthecowherdwasreducedby9%(P>0,05).

Sincethecullingpolicywasthesameforalltreatments,thereductionincowageatthehigherSRtreatmentsimpliedashorterlifetimeinthetreatmentherdbecauseoftheneedto replace the individual sooner. The response curves forbodymassandICPhaveexactly thesameshapeas thosepredictedbyJonesandSandland(1974).

Productivityofthebeefsystem,measuredinkgofweanedmass produced per hectare (Figure 8), increased acrossallSRtreatments(P<0,01),albeitbyonly241%comparedto the 300 % increase in SR. The decrease in individualproductivity was evidently offset by the increase in thenumberofanimalsinasystem(treatment),andtheturningpointintheresponsecurveofthetotalsystem,aspredicted

Table 4. Nutrient content and digestibility of all random and imitated forage samples

Random samples

Imitated samples Difference

Number of samples 280 732 –

Field dry matter (DM, %) 76,8±15,0� 65,2±22,61 P < 0,01

Crude protein (CP, %) 4,5±1,63 7,7±3,74 P < 0,01

Calcium (Ca, %) 0,37±0,238 0,70±0,77� P < 0,01

Phosphorus (P, %) 0,03±0,018 0,05±0,032 P < 0,01

Crude fibre (CF, %) 37,�±3,21 33,6±7,43 P < 0,01

Acid detergent fibre (ADF, %)

45,1±3,74 40,7±6,24 P < 0,01

Neutral detergent fibre (NDF, %) 72,7±5,50 63,�±13,64 P < 0,01

Fat (%) 1,4±0,37 2,1±1,28 P < 0,01Ash (%) 8,2±2,41 �,3±5,8� P < 0,01Digestibility (DOM, %) 44,�±8,4� 50,4±�,27 P < 0,01

Metabolizable energy (ME, MJ/kg)

6,2±1,03 7,2±1,18 P < 0,01

AGRICOLA 2007 25

Tabl

e 5.

Th

e ef

fect

of s

tock

ing

rate

on

the

nutri

tive

valu

e an

d di

gest

ibili

ty o

f im

itate

d sa

mpl

es o

btai

ned

from

gra

sses

, woo

dy p

lant

s an

d di

coty

ledo

nous

her

bs a

nd fo

rbs

(mea

ns w

ithin

a s

truct

ural

gro

up, e

.g.

gras

ses,

with

diff

eren

t sup

ersc

ripts

diff

er s

igni

fican

tly a

t P <

0,0

1 if

supe

rscr

ipts

are

in c

apita

l and

at P

< 0

,05

if su

pers

crip

ts a

re in

sm

all l

ette

rs)

Sam

ple

nFi

eld

DM

%C

P %C

a %P %

CF %

AD

F%

ND

F%

Fat

%A

sh %D

OM

%M

EM

J/kg

Gra

sses

125

65,�

±23,

146,

8±2,

550,

60±0

,480

A0,

05±0

,026

a35

,8±3

,51A

41,6

±3,2

7A68

,6±4

,48A

2,0±

0,82

a8,

5±2,

33A

52,3

±7,8

�A7,

5±0,

�4A

147

64,7

±1�,

876,

4±2,

330,

54±0

,415

A0,

05±0

,026

a36

,5±3

,0�B

42,4

±3,5

8A,B

70,6

±4,�

1B1,

�±0,

75a

8,0±

2,37

A51

,0±�

,42A

,C7,

4±1,

21A

136

68,5

±21,

586,

1±2,

530,

30±0

,1�2

B0,

04±0

,022

b37

,3±3

,63B

,C43

,3±4

,51B

70,4

±6,7

�B1,

�±1,

06a

7,8±

2,43

A48

,8±8

,81B

7,0±

1,05

B

136

67,4

±20,

666,

5±2,

780,

43±0

,30�

C0,

04±0

,032

b37

,6±2

,�7C

43,3

±3,2

0B72

,7±4

,40C

1,6±

0,54

b7,

3±2,

48B

4�,7

±�,4

5B,C

7,1±

1,16

B

Woo

dy

plan

ts

2265

,8±2

6,07

11,1

±4,1

00,

87±0

,827

0,05

±0,0

2521

,4±4

,�3

37,8

±7,8

8a46

,4±8

,�5a

3,�±

2,41

8,�±

4,71

a46

,8±�

,38

7,1±

1,63

186�

,�±2

6,55

12,4

±4,1

21,

0�±0

,341

0,06

±0,0

3222

,4±4

,81

35,5

±7,3

3a41

,1±6

,42b

3,3±

2,31

11,0

±�,5

3a45

,�±1

1,44

6,�±

1,61

156�

,1±2

3,3�

12,0

±4,3

11,

1�±0

,213

0,07

±0,0

6420

,8±5

,52

33,1

±7,8

�a,b

40,2

±8,1

8b4,

1±2,

3310

,1±8

,1�a

45,7

±8,2

27,

2±1,

25

1782

,2±2

0,06

12,5

±3,�

61,

56±0

,180

0,05

±0,0

401�

,8±6

,55

2�,2

±�,7

7b38

,5±1

0,�5

b2,

8±2,

4�1�

,1±1

7,1�

b48

,1±1

0,20

6,7±

1,86

Dic

ots

2043

,1±1

2,��

11,4

±2,7

71,

73±1

,415

0,07

±0,0

2823

,3±8

,13

35,1

±7,4

241

,6±�

,36

3,3±

1,31

16,6

±8,8

055

,�±6

,71

8,0±

0,�5

1542

,5±1

2,�5

11,8

±2,7

21,

65±0

,742

0,08

±0,0

3326

,6±�

,85

35,6

±6,5

843

,3±�

,14

2,6±

0,70

14,7

±7,3

756

,0±7

,43

8,1±

1,04

1235

,4±8

,85

12,1

±2,0

11,

66±0

,7�0

0,10

±0,0

4322

,3±8

,03

32,1

±5,5

441

,0±7

,07

2,8±

0,66

21,2

±�,8

354

,4±1

0,�3

7,6±

1,17

1640

,4±1

1,44

12,2

±3,7

01,

60±0

,�45

0,0�

±0,0

3624

,�±7

,25

33,1

±4,8

442

,�±7

,36

2,8±

1,46

16,7

±6,8

854

,0±8

,�1

7,6±

0,�2

All

imita

ted

sam

ples

167

63,2

±23,

677,

�±3,

410,

85±0

,�78

A0,

05±0

,027

32,5

±7,4

1a40

,3±5

,25

62,5

±12,

26a

2,4±

1,40

A�,

5±4,

7652

,0±8

,26a

7,5±

1,07

A

180

63,4

±21,

077,

4±3,

410,

73±0

,718

A,B

0,05

±0,0

2834

,5±7

,50b

40,8

±6,7

465

,3±1

2,34

b2,

1±1,

10B

8,�±

4,63

50,�

±�,6

8a,b

7,4±

1,26

A

164

66,2

±22,

6�7,

2±3,

680,

48±0

,5�5

C0,

05±0

,036

34,5

±7,4

2b41

,4±6

,4�

65,2

±13,

38b

2,1±

1,37

B�,

0±5,

434�

,0±�

,17c

7,1±

1,10

B

170

66,5

±21,

�87,

7±3,

�40,

67±0

,768

B0,

05±0

,037

34,3

±6,3

�b41

,1±5

,0�

66,2

±14,

14b

1,�±

1,11

C�,

3±7,

4450

,1±�

,61b

,c7,

1±1,

25B

Sto

ckin

g ra

teLo

wLo

w–m

ediu

mM

ediu

m–h

igh

Hig

h

26 AGRICOLA 2007

byJonesandSandland(1974),wasnotreachedundertheconditionsofthistrial. However, it did appear as if an inflection point was reached in theresponsecurveoflarge-framedcattleatthemedium-highSR,whereastheresponsecurveofsmall-framedcattlecontinuedincreasinglinearly(effectof CFS on system productivity: P < 0,05). Thus, it appeared that the SRtreatment resulted in more stress for large-framed cattle than for theirsmall-framedcontemporaries.

CONCLUSION AND RECOMMENDATION

Thedietselectedbyfree-rangingcattleinasemi-aridsavannainNamibiawasseverelyaffectedbythestockingrateofcattleandtheseasonoftheyear. Only at a low stocking rate could cattle express their true dietarypreferences, relying heavily on a small number of highly preferredperennialgrassspecies.Asthestockingrateincreased,theywereforcedtomakeincreasinguseofpreviouslyunpreferredforages,totheextentthattheirnutritional statusand individualproductivitydeclinedsignificantly.Similarly, cattle were forced to make increasing use of browsed forageduringthedryseasonsoftheyear,whenthegrassesaredormantanddonotregrowafterdefoliation.Bycomparison,theeffectofcattleframesizeon diet composition was small, although it appeared that small-framedcattlewerebetterabletoexploittheavailableforageresourcesthanlarge-framedcattlewere.

Increasing the stocking rate of cattle hadsevere effects on rangeland condition. Assoonasthestockingratewasincreased,themostpreferredgrassspeciesbecameweakerand less competitive, and were replaced inthe grass sward by less preferred, bulkiergrassspecieswithalowernutritivevalue.Atthehigheststockingrateappliedinthistrial,the threshold towards bush encroachmentappeared to be reached as well. Degradedrangelandwaslessproductiveandnutritious,andthuslessabletosupportbeefproductionatahighlevel.

It is therefore recommended that beefproducersinthesemi-aridsavannasofsouth-westernAfricadonotexceedastockingrateof45kgcattlemass/ha.Atthisrate,theycanexpect relatively high system productivityonaproductive,iftransformed,grassswardinasavannathatisnotyetbush-encroached.At a higher stocking rate, degradation ofthe range will advance to a higher, moredeleterious level that will probably makebeefproductionlesssustainable.

Cattlestudbreedersthatdependonthesaleofhigh-performingliveanimalsareadvisednottoexceedastockingrateof25kgcattlemass/ha,asonlysuchalowratewillsupportmaximumproductionofindividualanimals.Ranchingwithanadapted,indigenoussmall-framed breed of cattle will allow ranchersto slightly exceed the above stocking raterecommendations without sacrificing pro-ductivity,as thesmall-framedbreed in thistrialwasmoreproductiveandbetterabletoexploitforagingconditionsthantheirlarge-framedcontemporaries.

ACKNOWLEDGEMENTS

ThefinancialassistanceoftheInternationalFoundation for Science (IFS), Stockholm,Sweden, through its grant B/3183-1, isgratefully acknowledged. The personnel oftheSandveldResearchStationare thankedfortheirassistanceinthefield,andthestaffoftheAEZLaboratoryfortheanalysesofthefeedsamples.

REFERENCES

ADAMS, F. & WERNER, W., 1��0. The Land Issue in Namibia: An Enquiry. Namibia Institute for Social and Economic Research, University of Namibia, Windhoek, Namibia.

ALBRIGHT, J.L. & ARAVE, C.W., 1��7. The Behaviour of Cattle. CAB International, Wallingford, UK.

Figure 7. Effect of the stocking rate of cattle on the performance of individual cows, such as their age, body mass, body condition score (BCS) and fertility

Figure 8. The response of beef system productivity to an increase in the stocking rate of cattle (the arrow indicates the inflection point in the response curve of large-framed cattle)

AGRICOLA 2007 27

AOAC, 1��5. Official Methods of Analysis (16th edition). Association of Official Analytical Chemists, Washington DC, USA.

BESTER, F.V., 1�88. Die bepaling van die grasproduksie van natuurlike veld [Determining the grass yield of natural veld]. Agricola 6: 26–30.

BESTER, F.V., 1��8. Major problem: bush species and densities in Namibia. Agricola 10: 1–3.

BRYMAN, A. & CRAMER, D., 1��7. Quantitative Data Analysis with SPSS for Windows: A Guide for Social Scientists. Routledge, London, UK.

CLEMENTS, F.E., 1�28. Plant Succession and Indicators: a Definitive Edition of Plant Succession and Plant Indicators. H.W. Wilson Co., New York, USA.

CRAVEN, P. (editor), 1���. A Checklist of Namibian Plant Species. Southern African Botanical Diversity Network Report no. 7. Pretoria, South Africa.

ELS, J.F., 2002. The influence of frame size/breed and stocking rate on: I. Reproduction and production of beef cattle (preliminary results). Proc. 1st Joint GSSA/SASAS Congress, 13–16 May 2002. p 32. Christiana, South Africa.

EMMANS, G.C., 1��1. Diet selection by animals: theory and experimental design. Proc. Nutr. Soc. 50: 5�–64.

FORBES, J.M., 1��5. Voluntary Food Intake and Diet Selection in Farm Animals. CAB International, Wallingford, UK.

GRUNOW, J.O., 1�80. Feed and habitat preferences among some large herbivores on African veld. Proc. Grassl. Soc. South Afr. 15: 141–146.

HARDY, M.B. & WALKER, R.S., 1��1. Determining sample size for assessing species composition in grassland. J. Grassl. Soc. S. Afr. 8 (2): 70–73.

JONES, R.J. & SANDLAND, R.L., 1�74. The relation between animal gain and stocking rate. Derivation of the relation from the results of grazing trials. J. Agric. Sci. (Cambridge) 83: 335–342.

KIRKMAN, K.P. & MOORE, A.C., 1995. Quantification of the effects of defoliation on grass vigour. Afr. J. Range For. Sci. 12: 8�–�1.

KLAASSEN, E. & CRAVEN, P., 2003. Checklist of Grasses in Namibia. Southern African Botanical Diversity Network Report no. 20. Pretoria, South Africa.

KRUGER, A.S., 1998. The influence of stocking rate and cattle type on veld and animal performance in the camel thorn savannah of Namibia. M.Sc.Agric. thesis. University of Pretoria, Pretoria, South Africa.

MENKE, K.H., RAAB, L., SALEWSKI, A., STEINGASS, H., FRITZ D. & SCHNEIDER, W., 1�7�. The estimation of the digestibility and metabolizable energy content of ruminant feedstuffs from the gas production when they are incubated with rumen liquor in vitro. J. agric. Sci., Camb. �3: 217–222.

MILNE, J.A., 1��1. Diet selection by grazing animals. Proc. Nutr. Soc. 50: 77–85.

MPHINYANE, W.N., 2001. Influence of livestock grazing within piospheres under free range and controlled conditions in Botswana. Ph.D. thesis. University of Pretoria, Pretoria, South Africa.

MUFANDAEDZA, O.T., 1�77. Diet preferences of cattle and defoliation patterns under grazing on legume-improved veld. Ann. Rep. 1976. pp175–180. Div. of Livestock and Pastures, Salisbury, Rhodesia.

NARJISSE, H., 1��1. Feeding behaviour of goats on rangeland. In: MORAND-FEHR, P. (editor). Goat Nutrition. ch 2. Pudoc, Wageningen, The Netherlands.

O’CONNOR, T.G., 1��4. Composition and population responses of an African savanna grassland to rainfall and grazing. J. Appl. Ecol. 31: 155–171.

ODADI, W.O., YOUNG, T.P. & OKEYO-OWUOR, J.B., 2003. The influence of large mammalian herbivores on cattle foraging

behaviour in an Acacia savanna in Laikipia, Kenya. Proc. VIIth Int. Rangeland Congress, 26 July–1 August 2003: 522–524. Durban, South Africa.

ORTEGA, I.M., BRYANT, F.C. & DRAWE, D.L., 1��5. Contrasts of oesophageal fistula versus bite-count techniques to determine cattle diets. J. Range Manage. 48: 4�8–502.

PETRIDES, G.A., 1�75. Principal foods versus preferred foods and their relation to stocking rate and range condition. Biol. Cons. 7 (3): 161–16�.

PRACHE, S., GORDON, I.J. & ROOK, A.J., 1��8. Foraging behaviour and diet selection in domestic herbivores. Annales de Zootechnie 47: 335–345.

RAWLINSON, J., 1��4. The Meat Industry of Namibia 1835 to 1994. Gamsberg Macmillan, Windhoek, Namibia.

ROBERTSON, J.B. & VAN SOEST, P.J., 1�81. The Analysis of Dietary Fibre in Food. Dekker, New York, USA.

ROGERS, P.J. & BLUNDELL, J.E., 1��1. Mechanisms of diet selection: the translation of needs into behaviour. Proc. Nutr. Soc. 50: 65–70.

ROTHAUGE, A., 2000. New ecological perceptions of arid rangelands. Agricola 11: 4�–56.

ROTHAUGE, A., SMIT, G.N. & ABATE, A.L., 2003. The effect of the sub-habitat below savanna trees on soil, grass and its utilization by cattle. Proc. VIIth Int. Rangeland Congress, 26 July–1 August 2003. pp 106–10�. Durban, South Africa.

SENFT, R.L., 1�8�. Hierarchical foraging models: effects of stocking and landscape composition on simulated resource use by cattle. Ecol. Model. 46: 283–303.

SCHOLES, R.J., DOWTY, P.R., CAYLOR, K., PARSONS, D.A.B., FROST, P.G.H. & SHUGART, H.H., 2002. Trends in savanna structure and composition along an aridity gradient in the Kalahari. J. Veg. Sci. 13: 41�–428.

SKARPE, C., 1��1. Impact of grazing in savanna ecosystems. Ambio 20 (8): 351–356.

SMIT, G.N., RICHTER, C.G.F. & AUCAMP, A.J., 1���. Bush encroachment: an approach to understanding and managing the problem. In: TAINTON, N.M. (editor). Veld Management in South Africa. ch 10. University of Natal Press, Pietermaritzburg, South Africa.

SMIT, G.N. & SWART, J.S., 1994. The influence of leguminous and non-leguminous woody plants on the herbaceous layer and soil under varying competition regimes in Mixed Bushveld. Afr. J. Range For. Sci. 11: 27–33.

STUTH, J.W., LYONS, R.K. & KREUTER, U.P., 1��3. Animal/plant interactions: nutrient acquisition and use by ruminants. In: POWELL, J.M., FERNANDEZ-RIVERA, S., WILLIAMS, T.O. & RENARD, C. (editors). Proceedings, Livestock and Sustainable Nutrient Cycling in Mixed Farming Systems of Sub-Saharan Africa. Vol 2. Technical Papers, 22–26 November 1993. pp 63–82. Addis Ababa, Ethiopia.

TOTHILL, J.C., 1�87. Measuring botanical composition of grasslands. In: t’MANNETJIE, L. (editor). Measurement of Grassland Vegetation and Animal Production. Bull. 52. CAB International, Berkshire, UK.

WARD, D., 2005. Do we understand the causes of bush encroachment in African savannas? Afr. J. Range Forage Sci. 22: 101–106.

WESTOBY, M., WALKER, B. & NOY-MEIR, I., 1�8�. Opportunistic management for rangeland not at equilibrium. J. Range. Manage. 42: 266–274.

WOLFSON, M.M. & TAINTON, N.M., 1���. The morphology and physiology of the major forage plants. Grasses. In: TAINTON, N.M. (editor). Veld Management in South Africa. ch 3. University of Natal Press, Pietermaritzburg, South Africa.

ZAR, J.H., 1���. Biostatistical Analysis (fourth edition). Prentice Hall Int., New Jersey, USA