in Vertisols of an East African savanna The effects of herbivory and nutrients on plant biomass and carbonstorage Agriculture, Ecosystems and Environment

The effects of herbivory and nutrients on plant biomass and carbon storage in Vertisols of an East African savanna

Lucy W. Ngatia

^a

, Benjamin L. Turner

^b

, Jesee T. Njoka

^c

, Truman P. Young

^d

, K. Ramesh Reddy

^a,

*

aSoilandWaterScienceDepartment,UniversityofFlorida,Gainesville,FL,USA

bSmithsonianTropicalResearchInstitute,Apartado0843-03092,Balboa,Ancon,Panama

cLARMATDepartment,UniversityofNairobi,Kenya

dDepartmentofPlantSciences,UniversityofCalifornia,Davis,CA95616,USA

ARTICLE INFO Articlehistory:

Received11June2014

Receivedinrevisedform27February2015 Accepted20April2015

Availableonlinexxx Keywords:

Grazing Kenya Nitrogen Nutrients Phosphorus Savanna

ABSTRACT

HerbivoryandnutrientsaremajorecosystemdriversinAfricantropicalsavanna.Althoughprevious studieshavedeterminedtheinfluenceofherbivoryoncarbonstorageinsavannaecosystems,littleis knownabouttheinteractiveeffectsofnutrientsandherbivory.Wedeterminedtheeffectsoflongterm grazingandshort-termfactorialnitrogen(N)andphosphorus(P)additionsonabovegroundbiomass,soil organicmatter(SOM)content,andplantnutrientstorage.Grazingreducedabovegroundbiomass,foliarP andNstocksby45%,38%and45%,respectively,comparedtoungrazedplots,althoughthefoliarP concentrationwas20%greateringrazedplots.Therewasnosignificantincreaseintheaboveground biomass afternutrient additiondespiteincreases infoliar Nand Pconcentrations, suggestingthat productivitywaslimitedbyadifferentresource(e.g.,moisture).Therewerenosignificantinteractions betweennutrientenrichmentandgrazing.Weconcludethatgrazingreducedabovegroundbiomass,but improvedgrassqualitythroughincreasedfoliarPconcentration.

ã2015ElsevierB.V.Allrightsreserved.

1.Introduction

Herbivoryandsoilnutrientsareamongthemajordeterminants of tropical savanna function, influencing both plant primary productivityandcarbonstorage.Nitrogen(N)andphosphorus(P) canlimitplantprimaryproductivityintropicalsavanna(Augustine et al., 2003; Thornley et al., 1991). The limitation to plant productivitybyaspecificnutrientisdiagnosedwhentheaddition of the given nutrient results in an increase in net primary production(LebauerandTreseder,2008).Thenutrientlimitation onplantproductioncanalsoimpacttheherbivoreproduction,asit hasbeenobservedthatsomegrazersexhibitNandPdeficiencyin theirdiet(Ngatia,2012).

Largemammalianherbivores couldhavepositive,neutral or negativeeffects onannualnetabovegroundplantproductionin differentecosystems,dependinginpartontheirdirecteffectson the availability of key nutrients (Bagchi and Ritchie, 2010).

Previous studies have indicated that grazing increases primary

productivityinsometropicalareas(PandeyandSingh,1992)while decreasingitinothers(Wilseyetal.,2002).Similarfindingswere reportedintemperategrasslands,withreportsofgrazingleading todecreases(Coughenour,1991;Puchetaetal.,1998;Singerand Schoenecker,2003)andincreases(Coughenour,1991;Frankand McNaughton, 1993; Pandey and Singh, 1992) in productivity.

Holland et al. (1992) argue that the capacity of herbivores to increaseprimaryproductionisduetoincreasednutrientturnover rates. However, the mechanisms/factors resulting to suchcon- trastingfindingsarenotyetclear.

Innaturalgrasslands,productivityandsoilfertilityaremainly maintainedbyrecyclingofnutrientsthroughplantlitterdecom- positionandherbivorefecalmatter(Grantetal.,1995).Nutrient limitationtoplantproductivityvariesspatiallyacrosstheAfrican savanna. Plant tissue N:P ratios have been widely used as an indicatorofnutrientlimitation,wherebyanN:Pratio>16indicates Plimitationtoplantgrowth,<14indicatesNlimitation,andratios between14and16suggestthatplantgrowthcanbeco-limitedby N and P (Gusewell, 2004; Koerselman and Meuleman, 1996).

Previousstudiesreportedcontrastingresultsonnutrientlimita- tion to plant productivity. Ludwig et al. (2001) reported N limitationunderopencanopyandPlimitationundersub-canopy.

Ries and Shugart (2008) indicatedNand P co-limitation,while

* Correspondingauthorat:POBox110290,2181McCartyHallAGainesville,FL 326110290,USA.Tel.:+13522943154.

E-mailaddress:krr@ufl.edu(K.R.Reddy).

http://dx.doi.org/10.1016/j.agee.2015.04.025 0167-8809/ã2015ElsevierB.V.Allrightsreserved.

ContentslistsavailableatScienceDirect

Agriculture, Ecosystems and Environment

j o u r n a l h o m e p a g e : w w w . e l s e vi e r . c o m / l o c a t e/ a g e e

O’Halloran et al. (2010) found luxury nutrient uptake. This indicatesthatNandPlimitationtoplantproductivityinsavanna iscasespecificandthereisneedformorestudiesinordertohavea betterunderstandingofnutrient limitationto plantproduction thatwouldinformthemanagementpractices.

AlthoughrainfallisakeydeterminantofproductivityinAfrican savanna(Sankaranetal.,2005)thereismuchless knownabout nutrientsandgrazing.Previousstudieshavefocusedonnutrients (Ludwig et al., 2001; Ries and Shugart, 2008) or herbivory (Graceetal.,2006)separatelyasmajorfactorsinfluencingplant productivityandC storage,but fewstudies haveconsideredthe interactionsofthetwofactorsinEastAfricansavanna.Considering theprojectedincreasesinatmosphericCO2 (Stokes etal.,2005), increasingherbivorespopulationsespeciallydomesticanimalsand somewildanimals(Kinnairdetal.,2010;Thornton,2010)andsoil nutrientlimitationtoplantproductioninthesavanna(Augustine

et al., 2003), it is important to consider theeffects of nutrient availabilityandherbivoryonplantquality,productionandCstorage.

Theobjectivesofthisstudywereto:(1)determinetheeffectsof longtermgrazingonthegrassabovegroundbiomass,soilCand nutrientstorage,and(2)todeterminehowNandPenrichment affect grass abovegroundprimary production, biomass and soil nutrient storage and (3) examine the interactive effects of herbivoryandnutrientenrichmentbetweengrazedandungrazed ecosystems.Wehypothesizedthat:(1)grazingreducesgrassCand nutrientstoragebutincreasessoilorganicmatter(SOM)andsoil nutrients,(2)grazingimprovesgrassqualitybyincreasedfoliarN and P due to herbivory accelerating nutrient cycling and stimulating newshoots regrowth and (3) N and P enrichment favors grass productivity in the non-grazed plots compared to grazedplots,becausenaturalnutrientcyclinghasbeenfacilitated throughfecalmatterdepositioningrazedplots.

Fig.1.Studysite.

Mapssource.Africamap;adaptedandmodifiedfromBooneetal.(2005).Quantifyingdeclinesinlivestockduetolandsubdivision(Page525,Fig.1).RangelandEcologyand ManagementJournalpublication.Kenyamap;adaptedfromGraham,(2006).PhDdissertation,(page8,Figs.1and2).King’sCollege,UniversityofCambridge.Laikipiadistrict andMpalaResearchCentremapsareunpublishedmapsfromMpalaResearchCentre.TheexperimentwasconductedinKLEEexclosureslocatedintheSWcorneroftheMpala ResearchCentreasindicatedbybluesquares.

2.Materialsandmethods

2.1.Studysite

MpalaResearchCentreandConservancyislocatedinasemi- aridtropicalsavannainLaikipiaCounty,Kenya(37E,0N;1800m elevation)andisassociatedwithMpalaranchcovering190km² (Fig. 1) (Augustine and McNaughton, 2004). The mean annual rainfall is 550mm at the study site (Veblen, 2012). Mpala Research Centre is managed for both livestock production and wildlifeconservation.Someoftheresidentwildlargeherbivores include elephants (Loxodonta africana), hartebeests (Alcelaphus buselaphus), giraffes (Giraffa camelopardalis), buffaloes (Syncerus caffer), Grant’sgazelles(Gazellagrantii),zebras(Equusburchelli), impalas (Aepyceros melampus) and elands (Taurotragus oryx) (Young et al.,1998). The livestock herbivoresare mainlycattle (Bostaurus)andcamels(Camelusdromedaries).

ThestudysiteisunderlainbyVertisols(blackcottonsoils)with a pH-H2O of 6.2 and bulk density of 1gcm ³. Vertisols cover approximately 43% of the Laikipia ecosystem (Ahn and Geiger, 1987).ThewhistlingthornAcacia(Vachellia)drepanolobiumisthe dominant tree species on Vertisols, accounting for 97% of the overstory,whilemorethan90%oftheunderstoryconsistsoffive grassspeciesandtwoforbspecies(Youngetal.,1998).Rainfallis weaklytrimodalwithadistinctdryseasonfromJanuarytoMarch (Georgiadisetal.,2007).

2.2.Experimentaldesign 2.2.1.Grazingexperiment

Thisstudywas conducted in theKenya long-termexclusion experiment (KLEE). KLEE is a herbivore exclusion experiment establishedin1995thatusessemi-permeablebarrierstoexclude differentguildsoflargemammalsarrangedina three replicate blocks(Youngetal.,1998).InOctober2010four16m²plotswere establishedineachofthethreeblocks,withinplotsthatexcludeall large herbivores (complete exclosure) and plots where all herbivores are allowed to graze (open). The large herbivores included both wild and domestic animals. At the start of the experiment, thecenter 1m² in each 16m² plot was clipped to groundlevel,andthegrassdriedandweighed.Thisprovidedan estimateoftheeffectof17yearsofherbivoreexclusiononstanding grassbiomassandfoliarnutrients.Atthesametime,acomposite soilsample(0–10cmdepth)wascollectedwithineach16m²plot (four separate cores in each composite) to estimatesoil C and nutrientconcentrations.

2.2.2.Nutrientenrichmentexperiment

After grass clipping and soil sampling, four fertilization treatmentswere established:N alone,Palone, a mixtureof N andP,andacontrol(hereafterreferredtoasN,P,NPandcontrol).

Thefourfertilizertreatmentswereappliedinthefour16m²plots thathadbeenestablishedinthegrazedandnon-grazedplots(i.e., eachreplicateherbivoryplotcontained one 16m² plotof each treatment).ThefertilizerapplicationincludedN(urea)at100kg Nha ¹and/orP(triplesuperphosphate)at50kgPha ¹,applied in late October 2010 and mid March 2011. After fertilizer applicationa1m² areainbothgrazedandungrazedplotswas cagedtoexcludeallvertebrateherbivores (i.e.,including hares androdents).Soilwassampledinthe16m²plotandaboveground regrowthbiomass in the1m² subplots washarvested in early May2011. Thegrassbiomasswasdriedat60C(Wrenchetal., 1996)toconstantweightandthenground.Soilsampleswereair driedfor12daysat25C(Wrenchetal.,1996).Allsampleswere analyzedinthe Universityof Florida WetlandBiogeochemistry Laboratory.

2.3.Nutrientanalysisofsoilandplanttissue

PlantandsoiltotalCandNweredeterminedusingaThermo- Electron Flash 1112 elemental analyzer. Organic matter was measured byloss onignitionfrom0.5gsamplesof driedsoils.

Thesampleswereplacedinamufflefurnaceandheatedto250C for30min.Thefurnacetemperaturewasthenincreasedto550C for4h(Anderson,1976).Organicmattercontentwascalculatedas themasslossonignitiononadryweightbasis(Luczaketal.,1997).

TotalPwasdeterminedusingignitionat550Cfollowedbyacid extraction in 1M H2SO4. Digested solutions were analyzed colorimetrically using Shimadzu UV–vis recording spectropho- tometerUV-160. ExtractableP,potassium, calcium,magnesium, ironandaluminumweredeterminedusingMehlich1methodas outlined by(Kuo,1996).Nutrient ratiosC:N, C:Pand N:Pwere calculatedona massbasis.NH⁺4-Nand NO 3-N wereextracted using 2M KCl. The samples were filtered through a 0.45-

m

^m

membranefilter(PallCorporation)andanalyzedcolorimetrically (WhiteandReddy,2000).

2.4.Statisticalanalysis

Statisticalanalyses wereconductedusing JMP(version7.02;

SASInstitute,2007).Significantdifferencesamongthetreatments forthevariablesweredeterminedbyone-wayanalysisofvariance forgrazingversusnon-grazedtreatmentbefore nutrientenrich- mentandtwo-wayANOVAafternutrientenrichmentusingTukey HSDtestandt-testat

a

^=0.05.

2.5.Determinationofnutrientuseefficiency(NUE)andapparent nutrientrecovery(ANR)

The difference method was used to determine apparent nutrientrecoveryand nutrientuseefficiency.Apparentnutrient recoveryreflectsplantabilitytoacquireappliednutrientfromsoil (Baligaretal.,2001)andisdeterminedas(UN UO)/FN,whereUN

and UOare thenutrient uptake bygrass withand withoutthe appliednutrient,andF_Nistheamountofnutrientappliedallin kgha ¹;theresultsareexpressedasa percentage.Nutrientuse efficiency(NUE)isdefinedastheamountofforage(drymatter) thatisproducedforeachunitofappliedNorP(FageriaandBaligar, 1999; Zemenchik and Albrecht, 2002). Nutrient use efficiency (NUE)isdeterminedas(YN YO)/FN,whereYNandYO,arethegrass Table1

Foliarnutrientconcentrationunderherbivoryandnutrientenrichment.

N(gkg ¹) P(gkg¹) C:N C:P N:P

Herbivory(beforefertilization)

Grazed 9.70.5^a 0.580.01^a 422^a 70010^a 170.83^a Ungrazed 9.40.3^a 0.530.02^b 421^a 76038^a 180.85^a

t-ratio 0.54 5.7 0.02 2.4 0.74

P-value 0.47 0.03 0.90 0.14 0.40

Nutrientenrichment(afterfertilization)

Control 191^c 10.03^b 21.40.9^a 39310^a 18.50.9^b N 251^b 10.05^b 16.00.7^b 40118^a 25.10.9^a P 170.6^c 2.60.3^a 23.40.8^a 16218^b 7.00.8^d NP 280.8^a 2.10.1^a 14.00.4^b 1907^b 13.60.5^c

F-ratio 40 26 38 81 91

P-value <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 MeansoftheNandPconcentrationandC:N,C:PandN:Pratiosbetweengrazedand ungrazedplotsbeforeandafterfertilization.Beforefertilizationthegrazedand ungrazedplotmeanswerereplicatesoftwelve(n=12).Thefertilizationincludes additionofNonly,Ponly,N+Pandcontrol(wherenonutrientswereadded).After fertilizationthegrazedandungrazedplotswerenotsignificantlydifferentand hence resultswerecombinedfor statisticalanalysis.Themeans werefrom6 replicates(n=6).ThedataindicatemeanSEM(Standarderrorofmean).The different superscript letters after SEM within a column indicate significant differencebetweentreatmentmeansatP<0.05.

biomasswithandwithoutthenutrientbeingtestedallinkgha ¹ andFNisasindicatedabove(Guillardetal., 1995;Syersetal.,2008).

3.Results

3.1.Effectsoflongtermgrazingonabovegroundbiomassandsoil nutrients

SeventeenyearsofgrazingincreasedthefoliarPconcentration ingrassby20%comparedtotheungrazedplots(t=5.7P=0.03;

Table 1). Foliar N was also higher in grazed plots, but not significantlyso(P>0.05;Table1).Grazingreducedaboveground grass biomass by 45% compared to ungrazed plots (t=15;

P=0.0008;Table2).ThestocksofabovegroundgrassbiomassC (t=16; P=0.0006),N (t=15; P=0.0008) and P(t=10; P=0.004) weretherefore,significantlygreaterinthenon-grazedplotsthan in the grazed plots, despite a reduction in foliar nutrient concentrations; grazed plots contained 55% of the biomass N and60%ofthebiomassPoftheungrazedplots(Table2)Inthis studybiomassisconsideredasthetotalamountofaboveground grassclippedatgroundlevel.ThegrassC:N,C:PandN:Pratiosdid, notdiffersignificantlybetweenthegrazedandungrazedplots.The C:NandC:Pratioswere>40and>700,respectively,inbothgrazed andungrazedplots,whiletheN:Pratiowas17(Table1).

Therewerenosignificantdifferencesinsoilnutrientsbetween grazedandungrazedplots,includingtotalC,N,P,extractableP,K, Ca,Mg,Al,NO3 -N,andNH4+

-N(Table3).OnlyextractableFewas greaterintheungrazedplots(by38%;t=6.6,P=0.02)thangrazed plots(Table3).

3.2.Nutrientenrichmenteffectsonsoilsandabovegroundbiomass A two-way ANOVA indicated foliar nutrient concentrations, biomassC,NandPstocksandsoilsnutrientswerenotsignificantly different between grazed and ungrazed plots after nutrient addition.Therewerenosignificantinteractionsbetweennutrient

enrichmentandgrazing.Therefore,datafromaboveparameters fromgrazedandungrazedplotswerecombinedwhenconducting thestatisticalanalysis.

AfternutrientenrichmentsoiltotalP,availableP,NH4+-Nand NO3 -N changed significantly (Fig. 2). However, there wereno significantchangesinsoilCortotalN(Table4).SoiltotalPdoubled underPandNP treatments(t=16;P<0.0001)comparedtothe controlandNtreatments(Table4;Fig.2A).AvailablePwasalso significantlygreaterunderPandNPtreatments(F=15;P<0.0001) than incontrol andN treatments(Fig.2B). TheNH4+-N (F=31;

P<0.0001) and NO3 -N (F=50; P<0.0001) were significantly greater in N and NP treatments compared to control and P treatment(Fig.2CandD).AfternutrientenrichmentthesoilC:N was significantly lower in the N treatment (F=7; P=0.0023) compared toothertreatments (Table 4).Thesoil C:Pratiowas significantlylowerinPandNPtreatments(F=30;P=0.0001)than othertreatments(Table4).

Two-wayANOVAindicatedthatthenutrientenrichmenthad a significant effect on foliar N (F=40; P<0.0001) and P (F=26;P<0.0001)concentrations(Table1).FortheNP,NandP treatments, foliar N concentration changed by +53%, +33%

and –10%, respectively, and foliar P concentration changed by +110%, +0% and +155%, respectively compared to the control plotgrass.

ApplicationofNand/orPfertilizerdidnothaveasignificant effectongrassbiomassorbiomassC,butsignificantlyincreased NandP stocks(Fig. 3A–D).WithN+Pfertilizeradditionthere was a significant increase in biomass N and P, addition of N alonedidnothaveasignificantdifferencefromcontrol orN+P treatment in termsofbiomass N.Also additionofP alonedid nothaveasignificantdifferencefromcontrolorN+Ptreatment in terms of biomass P (Fig. 3C and D) indicating interactions between Nand P in improvingbiomass N andP.The N and P interactionwas further supported by the N:Pratio decreasing significantly whenPandNP were added (<14) andincreasing whenNwasadded(>16)(Table1).ThegrassC:N,andC:Pratios weresignificantlydifferentacrossthetreatmentsafternutrient enrichment (Table 1). The grass C:N ratio was significantly lowerintheNandNPtreatments(<20)thaninthecontroland P treatments (>20; F=38; P<0.0001; Table 1). The grass C:P ratiowasreducedbyPaddition(<200)comparedtono-Pplots (>300;F=81; P<0.0001;Table 1).

3.3.Apparentnutrientrecovery(ANP),nutrientuseefficiency(NUE) andphysiologicalnutrientefficiency(PNE)

After nutrientenrichment apparent N and P recovery were not significantly different between N and NP treatments (Table 5) although apparent N was more than two times higher under NP treatment compared toN treatment. BothN andPuseefficiencywerenotsignificantlydifferent betweenN and NP treatments (Table 5); however, P addition had a negative effect while NP addition had a positive effect on nutrient useefficiency. Nitrogenuse efficiency wasmore than three times higher under the NP treatment compared to N treatment.

Table2

Grassbiomassandbiomassnutrientsunderherbivory.

Abovegroundbiomass(kgha ¹) C(kgha ¹) P(kgha ¹) N(kgha ¹)

Grazed 2214212^a 88984^a 1.30.1^a 273^a

Ungrazed 4030417^b 1569148^b 2.10.3^b 495^b

t-ratio 15 16 10 15

P-value 0.0008 0.0006 0.004 0.0008

Meansofgrassbiomassandbiomassnutrientsbetweengrazedandungrazedplotsbeforefertilization.Themeanwerefrom12replicatesthedataindicatemeanSEM.The differentsuperscriptlettersafterSEMwithinacolumnindicatesignificantdifferencebetweentreatmentmeansatP<0.05.

Table3

Soilparametersunderherbivorybeforenutrientenrichment.

Grazed Ungrazed t-ratio P-value

SOM(%) 15.60.3^a 15.00.2^a 2.1 0.16

TotalC(gkg ¹) 211^a 241^a 3.5 0.08

TotalN(gkg¹) 2.00.1^a 2.20.1^a 2.3 0.15 TotalP(mgkg ¹) 1395^a 1422^a 0.3 0.62 ExtractableK(mgkg ¹ 84750^a 87534^a 0.2 0.65 ExtractableP(mgkg ¹) 6.00.5^a 6.10.4^a 0.03 0.86 NH4+

-N(mgkg ¹) 5.90.3^a 6.10.3^a 0.2 0.68 NO3 -N(mgkg¹) 6.21.0^a 5.80.7^a 0.1 0.8 ExtractableCa(mgkg¹) 4139200^a 3999190^a 0.26 0.62 ExtractableMg(mgkg ¹) 88325^a 89417^a 0.1 0.70 ExtractableFe(mgkg ¹) 303^a 424^b 6.6 0.02 ExtractableAl(mgkg ¹) 47821^a 53217^a 4.0 0.058 Means of the soil parameters between grazed and ungrazed plots before fertilization.Therewere12replicatesforeachtreatmentmean.Thedataindicate meanSEM.ThedifferentsuperscriptlettersafterSEMwithin arowindicate significantdifferencebetweentreatmentmeansatP<0.05.

4.Discussion

4.1.Grazingeffectsonsoilandplantnutrientsandabovegroundgrass production

Thisobservationofdecreasedabovegroundgrassbiomass(45%) undergrazingisconsistentwithsimilarobservationinSouthAfrica semi-arid savanna (Mbatha and Ward, 2010) and in Rocky Mountain National Park, Colorado (Singer and Schoenecker, 2003).Puchetaetal.(1998)reporteda33%decreaseinstanding biomassundergrazingincentralArgentinacomparedwithanarea thathadbeenexcludedfromgrazingfortwoyears.Cuietal.(2005) dataindicated that grazing decreased aboveground biomassby 65–79%after25yearsofherbivoreexclusioninMongolia.Thelarge decrease was suggested to be due to the intensity of current grazing, although grazing reduced biomass by 18–32% in a degraded/overgrazed site where herbivores had been excluded for 10years(Cuiet al.,2005).Thissuggeststhat theperiod of herbivore exclusion, quality of the grass and the intensity of grazing determine the quantity of the standing aboveground biomass(Pandeyand Singh,1992).However,thestudyfindings contrast with those of Frank and McNaughton, (1993), who

reporteda47%increaseinabovegroundgrassbiomassingrazed plots in Yellowstone National Park, Wyoming, USA. However, Yellowstone NationalPark isa temperateareaand grazingonly occurredoverthewinterseason(7months),unlikeinourstudyin savannawhich indicatedlowergrassbiomassproductioninthe grazedplotswheregrazingoccurredthroughouttheyear.

Our study findings did not agree with our hypothesis that grazingwouldimprovesoilorganicmatter.Thiscouldbedueto soilcoverdifferencesbetweengrazedandungrazedplotswhich could influence the soil temperature. Grazing reduces the soil cover, leading to increasedsoil temperature, which is likely to acceleratetheorganicmatterdecompositionrateandreducethe SOM(Haynesetal.,2014).Thedepositionoffecalmatterandurine withhighNcontentduringgrazingcouldalsohaveaprimingeffect onexistingsoilC,acceleratingtheorganicmatterdecomposition andreducingsoilC(Fontaineetal.,2004;Ngatiaetal.,2014).In addition,itisnotablethatcattlearethedominantherbivoresinthe studysiteandafterfeedingonafreerangesystemduringtheday theyareenclosedincattlecorrals(bomas)atnightwherethesoil accumulatesalotofCandothernutrients(Augustineetal.,2003).

ThismeanssomeoftheCistranslocatedbycattlefromthegrazing sitetothebomas.

Fig.2.Soilnutrientsbeforeandafternutrientenrichment.(A)TotalP.(B)AvailableP.(C)Nitrate-N.(D)Ammonium-Nbeforeandafterfertilization.Barsrepresentmeanand errorbarsrepresentSEM.Meansarefromsixreplicates.Thedifferentlettersindicatesignificantdifferencebetweentreatmentsforbeforeandafterfertilizationmeansat P<0.05.

Table4

Soilsparametersafternutrientenrichment.

Treatment Totalcarbon(gkg ¹) Totalnitrogen(gkg ¹) Totalphosphorus(gkg ¹) C:N C:P

Control 222^a 2.10.2^a 0.170.01^b 10.50.4^a 1307^a

N 201^a 2.30.1^a 0.150.01^b 8.40.2^b 13510^a

P 231^a 2.20.1^a 0.340.05^a 10.00.4^a 7411^b

NP 212^a 2.250.1^a 0.330.02^a 9.10.4^ab 633^b

F-ratio 1.1 0.7 16 7.0 30

P-value 0.39 0.58 <0.0001 0.0023 <0.0001

Bothgrazedandungrazedwerestatisticallyanalyzedtogetherbecausetheywerenotsignificantlydifferent.Therewere6replicatesforeachtreatmentmean.Thedata indicatemeanSEM.ThedifferentsuperscriptlettersafterSEMwithinacolumnindicatesignificantdifferencebetweentreatmentmeansatP<0.05.

ThisstudyfoundnosignificantdifferenceinfoliarNconcen- trationsbetween grazed and ungrazed plots, which contrasted studies showing increased/decreased foliar N with grazing (Coughenour,1991;SingerandSchoenecker,2003;Turneretal., 1993).However,therewasatrendofgreaterfoliarNconcentration inthegrazedgrass.GrazingcanincreasesoilNthroughNrichurine andfeces(Augustineetal.,2003;Starketal.,2002)inamoreeasily decomposableform,whichbypasstheslowlitterdecomposition pathway(Coughenour,1991).Thisincreases nutrientuptake by plants (Ruess,1984) and increases the shoot nitrogen content (Knappetal.,1999).However,inthisstudysiteitispossiblethat thegrazersareminingthenutrientsfromthegrazedplotgrasses anddepositing themelsewhere,as indicatedbyatrend toward

lower soil N concentration in the grazed plots. The dominant grazersat thestudy site arecattle, which feedthroughout the rangelandallday,but thenarekeptinbomasatnight,wherea large amount of dung accumulates. Hence there is a flow of nutrientsfromthenutrientpoorbushlandtonutrientrichbomas.

Overnightcattlebomasareabandonedafteraperiodofuseand theynaturallyconverttonutrientrichglades,whosehighnutrient statusismaintainedbywildanimalsthatspendlotoftimeonthem whilefeedingorresting.(Augustineetal.,2003;Veblen,2012).

TheobservationofhigherfoliarPingrazedplotsissimilarto patternsfoundbyTurneretal.(1993)inKansaswhereherbivores wereexcludedfor10yearsandChanetonetal.(1996)intemperate subhumidgrasslandinArgentinawhereherbivoreswereexcluded foreightyears.Thelightgrazingintensityemployedinthisstudy site does not change the grass species composition. Therefore, therecouldbetworeasonsleadingtohigherfoliarPobservedin grazed plots. First, higher P could bedue tograzer defoliation stimulatinggrassregrowth(Luoetal.,2012)producingnewshoots withhigherPwhiletheungrazedplotsretainoldergrassshoots thatpersistforalongtimeuntillitterfall.Second,herbivorycould acceleratePmineralization(VadigiandWard,2014).Soilnutrient cycling through litter decomposition is referred toas the slow cycle,herbivoresshortcircuittheslowcyclewhenfeedingand acceleratenutrientreleasebacktothesoils(BelovskyandSlade, 2000).It has beenobserved that approximately75–90% of the nutrientsconsumedbygrazinganimalsarecycledbacktothesoil inurineandfeces(McKenzieetal.,2003).Approximately50–90%

ofPinmanureisplantavailable(Douetal.,2001).UnlikefecalN whichcanbevolatilizedfromthesoilsafterdeposition,Pisnot volatilized or easily lost due to poor drainage system in the Vertisols,andisthereforeeasilyavailabletotheplants(Miolaetal., Fig.3.Grassbiomassandnutrientsstorageafternutrientenrichment.(A)Abovegroundbiomass.(B)BiomassC.(C)BiomassN.(D)BiomassPinahectareoflandpermonth afterfertilization.Barsrepresentmeananderrorbarsrepresentstandarderrorofmeanfromsixreplicates.Thedifferentlettersindicatesignificantdifferencebetween treatmentsmeansatP<0.05.

Table5

Apparentnutrientrecovery(ANR;%)andnutrientuseefficiency(NUE)

Treatment ANR(%) NUE

Nitrogen

N 4.51.9^a 0.81^a

NP 12.05.1^a 2.92.1^a

F-ratio 1.9 0.8

P-value 0.20 0.39

Phosphorus

P 1.40.5^a 0.121.6^a

NP 2.10.8^a 5.74.2^a

F-ratio 0.6 1.7

P-value 0.44 0.22

NUE(kgofgrassbiomassproducedforeachunitofappliedNorP)following nutrientadditiontreatments.Valuesaremeansofsixreplicates,meanSEMare presented.ThedifferentsuperscriptlettersafterSEMwithinacolumnindicate significantdifferencebetweentreatmentmeansatP<0.05.

2015). Phosphorus availability to plants can be reduced by adsorptionontoclaysormetaloxides(PerkinsandUnderwood, 2001).Thisseemsnotthecaseinthisstudybecausetheincreased foliarPsuggestthatgrazingenhancedtherateofPcycling,with morePallocatedabovegroundsupportingshootregrowth.Hence, theyoungertissuesgeneratedaftergrazinghavehigherPcontents thanmatureorsenescenttissuesthatpersistinabsenceofgrazing (Zheng et al., 2012). In this P limited savanna ecosystem, the capacity of grazing to enhance vegetation quality in terms of improvingfoliarPisimportant.

Thelargevariationinreportedeffectsofgrazingofrangeland soilsand vegetation in this studyand others (Cuiet al., 2005;

Puchetaetal.,1998;SingerandSchoenecker,2003;Turneretal., 1993)suggestsstrongsiteeffectsandstudyeffectsthatwedonot yetunderstand,and thenumber of studiesis still toosmall to discerngeneralizedexplanationsforthisvariation.Clearly,more dataisneededfromothersitestohelpunderstandtheeffectsof grazingonrangelandsoilsandvegetation.

4.2.Effectsofnutrientenrichmentonsoilandplantnutrient concentration,grassprimaryproductionandcarbonstorage

Soilandbiomassparametersinthegrazedandungrazedplots didnotrespond significantlydifferentlytonutrientenrichment.

Thiswasduetosimilarityinsoilparametersevenbeforenutrient enrichment,andalsoperhapsbecausetheabovegroundbiomass wasclippedtogroundlevelbeforenutrientenrichment.

ThefoliarN:Pratio haswidely beenusedasanindicatorof nutrientlimitation(KoerselmanandMeuleman,1996).Thisstudy findings of 17–18 foliar N:P ratio before nutrient enrichment suggested N and P co-limitation to primary productivity.

However, the luxurious uptake of both N and P in this study withoutsignificantincreaseinbiomass productionwasconsis- tentwithpreviousfindingsinSouthAfricansavanna(O’Halloran etal.,2010),butcontrastedpreviousfindingsbyAugustineetal.

(2003)inneighboringAlfisols.whereplantproductionincreased withincreasedplantnutrientuptake.Thesefindingssuggestthat the plants are either inefficient in nutrient utilization, or are adapted to low soil nutrients status, or that there are other factors,suchasmoisturethat limitplantproduction (Berendse andAerts,1987).BaligarandBennett(1986)suggestedthatforan efficient plant, increased nutrient uptake should translate to increasedbiomassproduction,andfailuretoincreaseproduction indicatesplantinefficiencyandthepossibilitythattheplanthas slowgrowthrate.

ThestudyobservationofincreasedfoliarPwithincreasingsoilP concurredwiththeresultsofpreviousstudies(Ludwigetal.,2001;

RiesandShugart,2008),whilethestudyobservationofincreased foliarNconcentrationwithincreasingsoilNwassimilartoLudwig etal.(2001)inTanzaniansavannabutcontrastedRiesandShugart (2008)findingsinAfricanwoodlandsavannainBotswana.

Although the NP treatment in this study increased above- groundbiomassby+42%, thisincreasewaslowcomparedtoa +220%abovegroundbiomassincreaseinBotswanasavannaafter applyingsimilarquantitiesofNandPasinthisstudy(Riesand Shugart,2008).However,RiesandShugart(2008)reportedthat the foliar N concentration in their study did not differ significantly from the control, while in this study there was luxuryuptake of bothN andP after nutrientenrichment. The study findings of reduced C:N ratio after soil N enrichment contrasted with the results in Botswana savanna (Ries and Shugart, 2008). The Botswana study sitewas dominated by a different grass species and had higher rainfall (698mm). The differentgrassspeciesinthetwositescouldbehavingdifferent physiologicalresponsestoresourcesavailabilityandnutrientuse efficiency(ZemenchikandAlbrecht,2002).

4.3.EffectsofnutrientenrichmentonANRandNUE

Thisstudyfoundrelatively higherNuseefficiencyintheNP thantheNtreatment,whichcontrastswiththefindingsofSnyman (2002)thatNuseefficiencywashigherintheNfertilizationthan NP fertilization in South Africa savanna. We observed lower apparentNrecoverycomparedtoanaverageof21–27%whenN wasaddedaloneand36–45%whenNandPwereaddedtogetherin maizefarmincoastalsavannaofWestAfrica(Fofanaetal.,2005).

Inaddition,Fofanaetal.(2005)indicatedthatadditionofNandP togethersignificantlyimprovedtheapparentNuptakecompared to adding N alone, which was not the case in this study. We broadcastedthe fertilizer without manuallymixing it withthe soils.Hencebecause therootoccupiesaround1–2%of thesoil surfacevolumeandtheamountandproportionofaddednutrients that reach roots determines the efficiency of nutrient uptake (Baligar and Bennett, 1986), there is a possibility that this application method limited the apparent nutrient recovery.

Furthermore,inthetropicslargeNlossesoccurthroughleaching, denitrificationandammoniumvolatilization,whilePisfixedby adsorption on amorphous Fe and Al oxides and hydroxides preventingimmediateavailabilitytoplants.BaligarandBennett (1986) furtherindicated that nutrient uptake capacity isdeter- mined by the ability of the soils to supply nutrients and the capacityofplanttouptakethem.Thisisinfluencedbythegenetic makeupoftheplantandinteractionswithenvironmentalfactors, such as rainfall, solar radiation and temperature, since NUE dependsonthe(1)uptakeefficiency(acquiringnutrientfromsoil), (2)incorporationefficiency(transporttoshootsandleaves)and(3) utilizationefficiency(Baligaretal.,2001).TheincreasedfoliarN and Pandunaffectedprimary productionafternutrientenrich- ment indicatethat thespecies in this studywereinefficient in utilizingtheaddedandabsorbednutrients.

5.Conclusion

Grazingimprovesthequality(intermoffoliarP)whilereducing thequantityoftheabovegroundbiomassinthisecosystem.Inthis study, aboveground biomass, foliar nutrientsand soil nutrients respondedsimilarlyuponNand/orPadditioninbothgrazedand ungrazedplots(thisisbasedontheassumptionofbiomassclipped to ground level before addition of the nutrientsin grazed and ungrazed plots). The luxury uptake of both N and P without increasing plant biomass suggests that other resources, for example soilmoisturelimitationor adaptationofplant species to low nutrient conditions, could limit plant production in response tonutrient addition.Comparingour studywithother previousstudiesitisclearthereareinconsistentresponsesbyboth soilsand vegetation toherbivoryand N or/andPaddition. The mechanism/s behind the inconsistenciesis notyet clear; more studies in different rangelands are required to address this ambiguity.Inadditionitwouldbehelpfulforfutureresearchto focus on the interactions of the belowground biomass and abovegroundbiomassunderherbivoryandnutrientenrichment.

Acknowledgements

WethankMargaretKinnairdandstaffattheMpalaResearch Centre forlogistical support,Korir, Kariuki and Joseph forfield assistance,StuartDaviesforhelpfuldiscussion,JamesColee, for statistical support, CETRAD, University of Nairobi and Wetland Biogeochemistry Laboratory, Universityof Florida for access to laboratoryfacilities,andtheOfficeofthepresidentinKenyafor permissiontoconducttheresearchinKenya.LucyNgatiathanks FrankLevinsonandtheSmithsonianTropicalResearchInstitutefor fundingthroughaSTRI–Levinsonfellowship,aswellasfinancial