Agriculture, Ecosystems and Environment

(1)

Agriculture,EcosystemsandEnvironment188(2014)12–19

ContentslistsavailableatScienceDirect

Agriculture, Ecosystems and Environment

jo u r n al ho m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / a g e e

Seasonal patterns in decomposition and nutrient release from East African savanna grasses grown under contrasting nutrient conditions

Lucy W. Ngatia

^a,b

, K. Ramesh Reddy

^a,∗

, P.K. Ramachandran Nair

^a,c

, Robert M. Pringle

^b,d

, Todd M. Palmer

^b,e

, Benjamin L. Turner

^f

aSoilandWaterScienceDepartment,UniversityofFlorida,Gainesville,FL32611USA

bMpalaResearchCentre,P.O.Box555,Nanyuki,Kenya

cSchoolofForestResourcesandConservation,UniversityofFlorida,118Newins-ZieglerHall,Gainesville,FL32611,USA

dDepartmentofEcologyandEvolutionaryBiology,PrincetonUniversity,Princeton,NJ08544,USA

eDepartmentofBiology,UniversityofFlorida,Gainesville,FL32611,USA

fSmithsonianTropicalResearchInstitute,Apartado0843-03092,Balboa,Ancon,Panama

a r t i c l e i n f o

Articlehistory:

Received20September2013

Receivedinrevisedform27January2014 Accepted4February2014

Keywords:

Abovegroundbiomass EastAfrica

Nutrientcycling Nrelease Prelease Savanna

a b s t r a c t

Litterdecompositionandnutrientreleaseisoneofthekeybiogeochemicalprocessesthatregulateplant productivityandnutrientcyclinginAfricansavannaecosystems.Weexaminedtheinfluenceofnitro- genandphosphorusadditionsongrassdecompositionandnutrientreleaseratesinanAcaciasavanna ecosystemincentralKenya.Grasswasclippedfromafactorialnitrogen×phosphorusexperimentand decomposedinacommonplotthathadnotreceivedfertilizer.After20weeks,includingonedrysea- sonandonewetseason,50–65%ofcarbon,68–75%ofnitrogenand73–83%ofphosphorushadbeen releasedfromthelitter.Decompositionwasslowinthedryseason(massloss1–2%wk⁻¹)comparedto thewetseason(7–11%wk⁻¹).Wetseasondecompositionwasmorerapidforgrassesthathadbeenfer- tilizedwithnitrogen,eventhoughtissuenitrogenwasnotsignificantlydifferentfromthecontrolgrass, indicatingthatfactorsotherthanlitternitrogenconcentrationinfluenceddecompositionratesunder nitrogenenrichment.Surprisingly,nutrientlossfromdecomposinglitterwasrelativelyhighduringthe dryseason,suggestingarolefordewinleachingnutrientsfromdrylitter.Weconcludethatseasonalrain andnitrogenaddition(butnotphosphorusaddition)acceleratedecompositionofgrasslitter,butthat nutrientleachingduringthedryseasoncanbeconsiderable.

1. Introduction

Thebalancebetweennetprimaryproductivityanddecompo- sitionoforganicmatterdeterminescarbon(C)stocksinsavannas andotherterrestrialecosystems(Valentinietal.,2000).Thedecom- positionofannuallitterfallcontributesapproximatelyhalfofthe CO2 releasedfromsoils(soil+litterCO2)(Couteauxetal.,1995) andrecyclesnutrientsforplantuptake(Aertsetal.,1992).Under- standingthefactorscontrollinglitterdecompositionistherefore importanttodeterminetheinﬂuenceongreenhousegasesemis- sionandglobalwarming(Fearnside,2000).

Bothabioticandbioticprocessesregulatelitterdecomposition, includingphotodegradation,andmicrobialbreakdown(Facelliand Pickett,1991)andalsophysicalfragmentation(leachingofsoluble

∗ Correspondingauthor.Currentaddress:POBox110290,2181McCartyHallA, Gainesville,FL32611-0290,USA.Tel.:+13522943154;fax:+13523923399.

E-mailaddress:krr@uﬂ.edu(K.R.Reddy).

C)duringrainfallevents(Ferreiraetal.,2006).Decompositionrates arecontrolledprimarilybyextrinsicdriverssuchasclimate(Aerts, 2006;AustinandVitousek,2000;GillandBurke,2002)andsoil properties(Gilland Burke,2002).Intrinsicdriversincludeplant litterquality(Aerts,2006;Gindabaetal.,2004;MugendiandNair, 1997)andlitterphysicalproperties(Meentemeyer,1978).Temper- atureisconsideredtobeamajorregulatoroflitterdecompositionin coldclimates,whilelitterqualityismoreimportantunderwarmer conditions(Couteauxetal.,1995).

Moisturecontentis a majorregulator ofdecomposition and nutrientleachinginthearidenvironments.Rainandirrigationare consideredtobethemainsourcesofmoistureinagriculturalareas, whilerainisconsideredtobethemainsourceinnaturalecosys- tems.Intheseplacesothersourcesofwaternamelyfog,mistand dewareoftenoverlooked(Went,1955).Dewismainlydeposited inthedryseasonwhentheskyisclearandhenceanimportant sourceofmoistureinthesemi-aridareasduringthedryseason comparedtootherclimates(Went,1955).Duvdevani(1953)con- ductinganexperimentinthecoastalplainsofIsrael,wheredew http://dx.doi.org/10.1016/j.agee.2014.02.004

(2)

Fig.1. RainfallattheMpalaResearchCentre,Laikipia,Kenya.Recordsarepresentedfora)thelong-termrainfallaverage(1999–2011)andb)duringthestudyperiodfrom 12/22/2010to5/11/2011.In(b)thedataarepresentedin4weeksintervalstoemulatetheﬁeldlitterincubationperiod.Thelitterwasincubatedfor20weeksandthesamples werecollectedfromtheﬁeldafterevery4weekswithoutreplacement.

occurredfrequentlyindicatedthatmostplantsgrewabouttwice asmuchwhentheyreceiveddewduringthenight.Thedewwater couldrunofftheleavesandcollectinthesoilonwhichitdrips (Duvdevani,1953;Ruinen,1961).Thedrippingdewcontainsboth mineralandorganicconstituentsinvaryingcomposition,making dewaleachingagent(Ruinen,1961).

InAfricansavannas,mostlitterfalloccursinthedryseason,yet mostdecompositionstudieshavebeeninitiatedattheonsetofthe wetseason(Deshmukh,1985;Jamaand Nair,1996;Mafongoya etal.,1997;MugendiandNair,1997;Mugendietal.,1999)with fewinthedryseason(e.g.,FornaraandDuToit,2008).Inaddi- tion,moststudiesinvolvedleguminousforbsandtrees(Fornaraand DuToit,2008;Fosuetal.,2007;JamaandNair,1996;Mafongoya etal.,1997;MugendiandNair,1997;Mugendietal.,1999;Oladoye etal.,2008),andfewincludedgrasses(Deshmukh,1985;Ohiagu andWood,1979),eventhoughgrassesdominatetheaboveground biomassofsavannaecosystems(Bond,2008).Finally,fewstudies haveexaminedtheroleofnutrientsindeterminingthedecompo- sitionratesoftropicalsavannagrasses,yetfoliarnutrientscanhave amarkedimpactonlitterdecompositioninotherecosystemssuch astemperategrassland(KochyandWilson,1997;Morettoetal., 2001),tundra(Bryantetal.,1997;HobbieandGough,2004),tem- perateforest(Melilloetal.,1982),montaneforests(Hobbieand Vitousek,2000)andtropicalforests(GonzalezandSeastedt,2001;

Kasparietal.,2008).

Theobjectivesofourstudyweretodeterminetheinﬂuenceof NandPadditiononplantlitterdecomposition,NandPreleasein savannaecosystem.Todothis,wedecomposedgrassesfromafac- torialN×Pfertilizationexperimentinacommonunenrichedsite throughoutonedryseasonandonewetseason.Wehypothesized thatdecompositionrate,NandPreleasewouldbegreaterinthe wetseasonthaninthedryseason,andthatgrassesthathadbeen fertilizedwithNand Pwoulddecomposefaster thanuntreated grasses.

2. Methods 2.1. Sitedescription

Thestudywasconductedin2010and2011atMpalaResearch Centre,Kenya,whichencompasses190km²ofsemi-aridsavanna

withintheLaikipiaCounty oftheRift Valley Province(37^◦53E, 0^◦17N).The site is a conservancywhere wildlife and livestock co-exist and share resources. The dominant woody vegetation includesSenegalia(Acacia)brevispica,Vachellia(Acacia)etbaica,S.

(Acacia)mellifera,V.(Acacia)niloticaandV.(Acacia)gerrardii,Croton dichogamus,Grewiaspp.andRhusvulgaris(Youngetal.,1995).The herbaceousvegetationconsistsofadiscontinuouslayerofmostly perennialgrasses,whichincludePennisetummezianum,Pennisetum stramineum,Digitariamilanjiana,andCynodondactylon.

Thesoilsareredsandyloams(TypicHaplustalfsinSoilTaxon- omy)derivedfrommetamorphicbasementrock(AhnandGeiger, 1987; Goheen et al., 2013). Prior to fertilization, soil pH was 6.3(measuredinwater),totalsoilPwas230mgkg⁻¹ andavail- ablePwas11mgkg⁻¹ (determinedbyresinbagsasoutlinedby Kouno et al. (1995)).Total soil C and N contents were 10 and 1.1gkg⁻¹,respectivelyandthemeanannualrainfallwasapprox- imately640mmover a13years period(Fig.1a)(Goheenetal., 2013).Monthlymaximumtemperaturesrangefrom25to33^◦C, whileminimumtemperaturesrangefrom12to17^◦C(Youngetal., 1998).

This studywas conductedin the Ungulate HerbivoryUnder RainfallUncertainty(UHURU)experiment,establishedin2008at theMpalaResearchCentre(Goheenetal.,2013).Thestudywascon- ductedinthesouthern(wettest)siteoftheUHURUexperiment, whichreceivedanaverageof638mmrainyear⁻¹ from2009to 2011.

2.2. Experimentaldesign

Four fertilizer addition plots (16m² each; hereafter “plots”) wereestablishedinFebruary2010ineachofthreereplicateexclo- sures (1ha each) which fenced out all herbivores larger than Lepusspp.(∼2–3kg).Thus,therewereatotalof12plotsarranged acrossthethreeexclosures.The 12plotsentailedfourfertilizer treatments and three replications of each fertilizer treatment.

Within each ofthe 12 plots,grass wasclipped toground level ina1m²patchanddiscardedandfertilizerswereappliedtothe entire16m²plotontheonsetofrainfall(March2010).Thisallowed regrowthstandingdeadgrassfromthe1m²plottobeusedforthe decompositionexperiment.Thefertilizertreatmentsincluded:(1) Nonly,(2)Ponly,(3)amixtureofNandP,and(4)unfertilized

(3)

14 L.W.Ngatiaetal./Agriculture,EcosystemsandEnvironment188(2014)12–19

control(hereafterreferredasN,P,NPandcontrol).Thefertilizer applicationconsistedofN(urea)at100kgNha⁻¹ and/orP(triple superphosphate) at50kgPha⁻¹.Alltreatments were arranged inarandomizedcompleteblockdesignwithineachoftheexclo- sures.InNovember/December2010,grassregrowthfromthe1m² fertilizer-enrichedpatcheswasclippedforuseindecomposition study.

2.3. Experimentalprocedure

After fertilization, aboveground standing dead grass was selectedfromtheregrowth,wasclippedfromeachplotandair- driedtoa constant weight.Litterbags (15×20cm) weremade from2mmmeshpolyestertoallowentryofmicro-andmesofauna.

Threegrams(dryweightbasis)ofamixtureofgrassesfromeach fertilizer-enrichedplotandeachofthethreetreatmentreplicates wasputintoeachof10litterbags,whichwereheatsealedand placedinthethatchlayerinacommonunenricheddecomposition areaawayfromtheplotsoforigin.Afterplacementontheground, thebagswerelightlycoveredwithlitterfromthecommonunen- richedplot.Thisexperimentwasmaintainedfrom22December 2010to11May2011(5months).Twolitterbags(duplicates)from eachofthethreetreatmentsreplicateswereretrievedafter0,4,8, 12,16,and20weeks.Thebagsandtheircontentswereair-driedto aconstantweight.Priortoweighingthecontentsofeachbag,soil particlesandanyotherextraneousmatterwerecarefullyremoved.

Thecontentfromtheduplicatelitterbagsineachtreatmentrepli- catewascompositedforchemicalanalysisafterweighing.Litter sampleswerethen groundtopassthrougha 1-mmscreen and analyzedfortotalC,N,Pandligninconcentrations.TheNandP concentrationandremaininglitterbiomassatanygiventimewere usedtoestimatetheNandPrelease.TheNorPreleasewasthebal- ancebetweentheoriginalNorPmassandthemassatanygiven time.

2.4. Litterprocessingandanalysis

TotalCandNweredeterminedusingaCostechModel4010Ele- mentalAnalyzer(CostechAnalyticalIndustries,Inc.,Valencia,CA) coupledtoaFinniganMATDeltaplusXLmassSpectrometer(CF- IRMS,ThermoFinnigan)viaaFinniganConﬂoIIinterface.TotalP wasdeterminedbyignitionat550^◦Cfollowedbyextractionin1M H₂SO₄ acidanddetectionbyautomatedmolybdatecolorimetry.

DigestedsolutionswereanalyzedcolorimetricallyusingShimadzu UVvisiblerecordingspectrophotometerUV-160.Ligninwasdeter- minedbyamodifiedsequentialfiberextractionmethod(Ankom Technology,Fairport,NY)modifiedfromthefeedandforageanal- ysisbyVanSoest(1970).

2.5. Statisticalanalysis

Statistical differences in treatments were determined using analysisof variance (ANOVA)for a randomizedcomplete block design using Tukey HSD test at ˛=0.05. Sidak adjustment for multiplecomparisonstestwasusedtodeterminethesigniﬁcant differenceforthemultiplephaseregressionmodelat˛=0.05.A nonlinearprocedureof SAS(SAS 9.2)for multiphaseregression modelswasﬁttedforCreleaseduringdecomposition;parameters includeinterceptandslopeofeachphase,andthespline.Theslope ofthelinearregressionrepresentsthedecompositionrateconstant (k).Thesplinepointistheendofphase1andbeginningofphase 2(JamaandNair,1996).ForCreleasehalf-lifewascalculatedfrom multipleregressionmodelbydeterminingthetime(x)thathalfof theoriginalCmasshadbeenreleased.

3. Results

3.1. Patternoflitterdecomposition

Plant litter decomposition followed a biphasic pattern, with aninitialphase ofslow decompositionrates followedby faster rates(Fig.2a–d)aftertheonsetoftherainsinApril.Withinnutri- entenrichmenttreatments,thedecompositionconstants(k1and k2)forthetwophasesdifferedsignificantlybetweendryandwet seasons(phaseoneishereafterreferredtodryphaseandphase two as wetphase). Sidak adjustmentfor multiplecomparisons testindicatedthatthedecompositionrateconstantwashigherin thewetphasethan thedry phase foralltreatments(P<0.0001 for thecontroland N enrichmenttreatments P=0.0003 forthe P enrichment treatment, P=0.04 for the NP enrichment treatment)(Fig.2).Thedecompositionratewastwotimesgreaterfor theenrichedtreatmentsthanforthecontrol(Fig.2).Forthewet phase,thedecompositionconstantwassignificantlyhigherforN enrichedgrasses(11%wk⁻¹;P=0.0008),whereasP,NP-enriched grassandcontrolgrassdidnotdiffersignificantlyfromoneanother (Fig.2a–d).

Thedecompositionrateofgrassfromthecontroltreatmentwas 10gkg⁻¹week⁻¹inthedryphaseand70gkg⁻¹week⁻¹inthewet phase(Fig.2a).Thiscorrespondstoahalf-lifeof 39weeks.The wetphaseofdecompositionstartedat13weeksat88%initialC mass.Attheendofthestudy(after 20weeks),50%ofthetotal Cremained.ThedecompositionrateofN-onlyenrichedgrasswas 20gkg⁻¹week⁻¹inthedryphaseand110gkg⁻¹week⁻¹inthewet phase,(Fig.2b),correspondingtoahalf-lifeof35weeks.Thewet phaseoflitterdecompositioncommencedonweek14with75%of theinitialCremaining.Attheendofthestudy,35%oftheinitialC remained.

The decomposition rate of the P-only enriched grass was 20gkg⁻¹week⁻¹inthedryphase,and90gkg⁻¹week⁻¹inthewet phase(Fig.2c),withahalf-lifeof40weeks.Thewetphaseofdecom- positionbeganat16weekswith73%oftheinitialCmassremaining.

Attheendofthestudy,45%oftheinitialCmassremained.

The NP-enriched grass decomposition rate was 20gkg⁻¹week⁻¹ in the dry phase and 80gkg⁻¹week⁻¹ in thewetphase (Fig.2d)withahalf-lifeof42 weeks.Wetphase decomposition started at 16 weeks (Fig. 2d) with 73% of the initialCremaining.Althoughtherewerenosigniﬁcantdifferences betweentheinitialnutrientconcentrationsoftheenrichedgrasses (Table1),Sidakadjustmentformultiplecomparisonstestindicate thatN-enrichedgrasseshadhigherdecompositionratesinthesec- ond(wet)phase(P=0.0008;Fig.2b)comparedtoothertreatments, withtheN-enrichedgrasseshavingtheleastamountofCmass remainingattheendofthestudy.Theseasonalrainfallpositively correlatedwithdecompositionrate(R²=0.80;P<0.0001).

3.2. Trendsinnitrogenandphosphorusreleasefromthegrass litter

Nitrogenreleasewashighintheinitial8weeksofthedecom- position(dryseason).Thecontrol,N,PandNPtreatmentsreleased 45%,48%,41%and55%respectively,oftheinitialNmass(Fig.3).

Bytheendofthestudy(20weeks),theN,PandNPtreatmentshad eachreleased75%oftheinitialNmass,whilethecontrolreleased 68%ofinitialNmass(Fig.3a–d).

Phosphorusreleasewasrapidintheinitial8weeks(dryseason) ofthestudy.Inthisperiod,thecontrol,N,P,andNPtreatments released49%,61%,58%and70%oftheinitialPmass,respectively.

By theend ofthestudy(20 weeks),thesefourtreatments had released73%, 80%, 77%, and 83% of initialP mass, respectively (Fig.3e–f).

(4)

Fig.2.Decompositionpatternovertimefora)control,b)nitrogen-enriched,c)phosphorus-enriched,andd)nitrogen+phosphorusenrichedgrassplacedontheunenriched soilsurface.Themeansarereplicatesofthreecompositesamples±onestandarderror.Thesampleswerecollectedfromtheﬁeldafterevery4weeksforaperiodof20weeks withoutreplacement.

Table1

Abovegroundbiomass,biomassC(carbon),N(nitrogen)andP(phosphorus)storage.Chemicalcompositionoftheinitialgrassesbiomassafterfertilizationandratiosof initialabovegroundbiomass.Thevaluesaremeansofthreecompositereplicates±onestandarderror.Therewerenosigniﬁcantdifferencesbetweenthetreatmentsforall parameters.

Treatment Control N P NP

Abovegroundstorage(kgha⁻¹)

Abovegroundbiomass 1155±187 1494±407 1602±490 2411±735

Carbon 472±77 620±170 663±193 974±313

Nitrogen 14.7±2.7 25.6±3.6 27.3±5.2 33.4±17.6

Phosphorus 1.4±0.2 1.9±0.6 2.0±0.4 4.1±0.9

Chemicalcompositionoftheinitialabovegroundbiomass(gkg⁻¹)

Carbon 404±5 415±7 414±6 404±9

Nitrogen 13±1 17±2 17±2 14±2

Phosphorus 1.3±0.1 1.5±0.3 1.4±0.1 1.2±0.2

Lignin 250±30 210±25 250±5 230±13

Nutrientratiosoftheinitialabovegroundbiomass

Lignin:N 19±0.8 13±3.1 15±1.8 17±3

C:N 32±3 25±3 25±3 30±3.6

C:P 326±23 276±49 300±22 335±46

N:P 10.2±0.8 12.2±3.3 12.5±1.4 12.2±3.0

AlthoughtheC:N ratioofthelitter was>25 throughoutthe study,exceptfortheNandPtreatmentintheinitiallitter(Fig.4a), theC:Nratioincreasedthroughoutthestudy.However,theC:N ratioofthemassloss(<25)wasmuchlowerthantheC:Nofthe remaininglitterovertime(Fig.5a)andtheNandPlosswasclosely correlated(P<0.0001;R²=0.88)(Fig.5b).Foralltreatments,theC:P ratiowas>300throughoutthedecompositionstudy,apartfrom theinitiallitterfromtheN-enrichedtreatment(whichwas275;

Fig.4b).However,there wasa continuousincrease inC:Pratio throughoutthestudyperiod.

4. Discussion

The studyassessed litter decomposition in thedry and wet seasonsofasemi-aridAfricansavannabyemulatingthetiming of processesinvolved in naturaldecomposition. Ourresultsare

(5)

Fig.3. Cumulativepercentnitrogenandphosphorusreleasepatternovertime.Fornitrogen;a)control,b)nitrogen,c)phosphorus,d)nitrogen+phosphorusenrichedgrasses.

Forphosphorus;e)control,f)nitrogen,g)phosphorus,h)nitrogen+phosphorusenrichedgrasses.Themeansarereplicatesofthreecompositesamples±onestandarderror.

DifferentlettersalongthelinegraphindicatesigniﬁcantdifferencebetweenthemeansatP<0.05.Thesampleswerecollectedfromtheﬁeldafterevery4weeksforaperiod of20weekswithoutreplacement.

consistentwithpreviousworkinitiatedinawetseason(Dubeux et al.,2006; Hamadi et al.,2000; Jama and Nair, 1996), which reporteda biphasiclitter decomposition pattern. However,the orderwasreversedinourstudy,initiatedinthedryseason,with aninitialslowdecompositionrateinthedryseasonfollowedbya fasterdecompositionrateontheonsetofrainfall.Thisisconsistent withmoisturebeingthemain factorinﬂuencing decomposition rates(AustinandVitousek,2000;Epsteinetal.,2002;Mugendiand Nair,1997;Vitouseketal.,1994),furthersupportedbytheposi- tivecorrelationobservedbetweenthedecompositionconstant(k) andrainfall.Notably,however,CandNinourstudydidnotmin- eralizesimultaneouslyasobservedinseveralpreviousstudiesin

whichpatternsofbothClossandNreleasefollowedamultiphase regressionmodel(Hamadietal.,2000;JamaandNair,1996).In ourstudy,litterdecompositionfollowedamultiphaseregression model,whereas Nreleasedidnot;thiswassimilartothefind- ingsofDubeuxetal.(2006)insub-tropicalFlorida,USA.Minimal Cwaslostinthedryseason:inthefirst4weeks,only6–24%C waslost,whereas33–47%Nwasreleasedinthesameperiod.This contrasts withthefindings ofAustin andVitousek (2000),who reportedthatproportionalreleaseofCwasfasterthanreleaseof nutrients,whichisexpectedinadecompositionprocess.Onepossi- bilityisthat,althoughmoisturelimitsthedecompositionprocessin thedryseason,dewpromotesnutrientleachingfromdecomposing

(6)

Fig. 4.C:N and C:P ratio over time for a) carbon/nitrogen ratio. b) Car- bon/phosphorusratioofthelitterovertime.Thelitterisfromfertilizedplotsand decomposedinacommonunfertilizedsite.Themeansarereplicatesofthreecom- positesamples.Thesampleswerecollectedfromtheﬁeldafterevery4weeksfora periodof20weekswithoutreplacement.

litterinthedryseason(seealsoLeonardandFluckiger,1989;Tukey, 1970).

Dewisimportantinleachingnutrientsfromlitterinseason- allydryclimatesanddeserts(Tukey,1970,1969).Inthedryseason, plantlitterthatisotherwisedryduringthedayisfrequentlywet withdewatnight (Yang et al.,2001).Nash(1996)arguedthat althoughitisrarelyquantiﬁed,dewwatercanrepresentasignif- icanthydrologicalinputindryplacesorseasons.Ruinen(1961) indicatedthatinbothIndonesiaandSurinamwettingbydewand groundmisttakesplacefor10–12hperdaywherebyintheearly morningthevegetationdripswetandintheshadethedewremains visibleuntilmid-morning.Dataontheincidenceofdewarescarce andspatiallyvariable;however,Ruinen(1961)reportedthatdew occurredonanannualmeanof249nightsinYangambi,Congoand 254nightsinSouth Cameroon.Further,Ruinen(1961)reported thatnitrogenconcentrationsindewwere25–80mgL⁻¹,compared to9–19mgL⁻¹ inrain.Inourstudy,overthedryseasonwhich lastedforﬁrst12weeksofthestudy,rainfellforonly3daysand eachdaytherainwas∼10mm.However,thedewisexpectedto formalmosteverynightoverthedryseasonbecausetheskyisclear (Went,1955).Hence,inthedryseasonthereisahigherfrequency ofdewformationcomparedtorainfall.Tukeyetal.(1957)indicated acloserelationshipbetweenlightintensityandcarbohydratesloss byleachingfrombeanleaveswithandthelossesbeingcumulative withtimeofwetting.Weconcludethatleachingbydewisaplausi- blemechanismtoexplainthelossofnitrogenandphosphorusfrom litterduringthedryseasoninourstudy.

WiththeinitiallittercontainingC:Nratios>25andC:Pratios

>300, immobilization was expected to take place (Palm and

Fig.5.Nutrientslossesovertimefora)carbonversusnitrogenlosses.b)Nitrogen versusphosphoruslosses.Thelitterisfromfertilizedplotsanddecomposedina commonunfertilizedsite.Themeansarereplicatesofthreecompositesamples.

Thesampleswerecollectedfromtheﬁeldafterevery4weeksforaperiodof20 weekswithoutreplacement.

Sanchez,1990;Swiftetal., 1979), yettheratiosincreasedcon- tinuouslythroughouttheexperiment,whiletheC:Nratioofthe mass losswas<25. Dubeux et al.(2006) suggestedthat higher tissueC:NratiosindicatelowerNconcentrationintissuesrather thanchanges in Cconcentrations. Thedecreasein litter Nover thestudy period contrastedwiththe ﬁndings of severalprevi- ousstudies(Deshmukh,1985;Dubeuxetal.,2006;Hamadietal., 2000;Liuetal.,2011),whichreportedincreasednitrogenmasses overthestudyperiod.The68–75%Nand73–83%Poftheinitial massreleasedbytheendofour20-weekstudywerehigherthan the20–30%Nand60%PmineralizationreportedbyDubeuxetal.

(2006)for18-weekincubationinsub-tropicalFlorida,USA.How- ever,theNreleaseinourstudycontrastedtheNimmobilization reportedbyMtambanengweandKirchmann(1995)throughoutthe 75daysstudyinZimbabwewoodlandsavanna.Thissuggeststhat microbialdecompositionwasnotthedominantprocessleadingto NandPrelease.

Consideringthattheinitialgrassbiomasshad400–410gCkg⁻¹, ofwhich21–25%waslignin,theinitialslowrateofdecomposi- tioninthedryseasonlikelyresultsfromthelossofverysoluble Ccompoundsthatcandecomposeevenunderminimalmoisture.

Thesubsequentphaseofhigherdecompositionrateswiththeonset ofrainsuggeststhatthemorerecalcitrantcompoundsneedmore moistureforbreakdowntobeaccomplished(Meentemeyer,1978).

Overall,thedisappearanceofCwasgreaterinthewetseasonthan thedry season.Theseﬁndings contrastwithOhiaguand Wood (1979),whoreported69%ofgrasslitterdisappearanceovera4- monthdryseasoninNigeriansavanna,andindicatedthislosswas

(7)

duetoconsumptionbyfungus-growingtermites.Termiteswere excludedinourstudy,whichmightexplainthisdiscrepancy.

However,inasimilarstudyinKenya’sNairobiNationalPark, where termites wereabsent and the decomposition studywas initiated in the wet season, only 50% of the grass litter was lost over a 23-month study period, in spite of a greater rain- falland theuseoflitterbagswithlargeraperturemesh(5mm) that would have allowed in arthropods (Deshmukh, 1985). In the current study, 50% of C in the control treatment was lost in only 20 weeks. This suggests the inﬂuence of factors other thanjust rainfallandtheabsenceof termitesonthedecompo- sitionprocess,suchasthemixtureof grassspecies usedinthe study. In our study site, the grass mixture was dominated by P.mezianum,P.stramineum,D.milanjianaand C.dactylon,while thestudysiteof Deshmukh(1985)wasdominatedbyThemeda triandraandSetaria phleoides,whichmight haverespondeddif- ferentlytodecomposition.Bothphysicalandchemicalchangesin leafmixescaninﬂuencerateofdecompositiondirectly(physically) andindirectlythroughthedecomposercommunityanditsactiv- ities(GartnerandCardon,2004).Likewise,ratesofClossinour studywerehigherthanthe50%Zoysiajaponicagrasslittermassloss reportedbyNakagamietal.(2010)inJapanovera1yearincubation period.

Nevertheless,ourfindingsof50–65%Cbiomasslosswerecom- parable to thefindings of Dubeux et al. (2006), who reported 40–60%biomasslossofPaspalumnotatumFlueggé(Bahiagrass) inFloridainastudyperiodofthesamelength.Thefindingswere alsocomparableto36–55%Cynodondactylon(L.)Pers.(Bermuda grass)organicmatterlossreportedbyLiuetal.(2011)within18 weeksofincubationinFlorida.

Weacknowledgethattheuseofthe2-mmlitterbagsinthis studycouldhavealteredthelittermicro-climate(Meentemeyer, 1978;WitkampandOlson,1963)andexcludedthelargermacro- fauna(ElkinsandWhitford,1982),thusreducingthefragmentation oftheleavesandcausingthebaggedleavestodecomposeatadif- ferentratecomparedtonon-conﬁnedleaves.Ourstudyassumes thattheseeffectsdidnotmaskthedifferencesresultingfromcli- maticconditionsandlitterquality.

Nutrientenrichmentincreasedgrass-biomassproduction,but there was no significant difference in nutrient concentration betweeninitial incubatedlitters from different treatments.For example,theN-enrichedlitterlostthehighestamountofbiomass C(65%)andhadsignificantlyhigherrateofdecompositioninthe second(wet)decompositionphase(11%).Thesefindingsagreewith previousobservations(Liuetal.,2011;LupwayiandHaque,1999) thatNfertilizationincreasesbiomassdecomposition,butsuggest thatfactorsotherthanlitterNconcentrationacceleratedecompo- sitioninN-enrichedlitter.Anexampleofsuchafactorisanincrease inClabilityresultingfromNenrichment(PeyraudandAstigarraga, 1998),whichisaresultofNpromotinggrowthofsucculentherbage withalowcellwallcontent(Waite,1970).

ThehighNandPreleaseinthedryperiodcomparedtothewet- terperiodcontrastedwiththesuggestionbySwiftetal.(1979)that precipitationcancontrolthephysicalprocessofleachingthrough acceleratedbreakdownofsurfacelitterbyincreasedrainfall.After theﬁrst8weeksoftheincubationperiod(i.e.inthedryseason), 40–50%ofNand50–70%ofPhadbeenreleased.TherapidNand PreleasethatdidnotmatchtheClossandsuggestthatinthelong term,both NandParelikelytolimit themicrobial decomposi- tionprocessunlessthereisanexternalsourceofnutrientstodrive thedecompositionprocessorphotodegradationwouldprevailand short-circuitthedecompositionprocessleadingtodirectClossto theatmosphere (Austinand Vivanco,2006).On theotherhand, Couteauxet al. (1995)suggested that decrease in litter quality couldleadtoincreasedproductionofligninolyticenzymes,lead- ingtoincreaseddegradationofrecalcitrantcompounds.However,

theextenttowhichloweringtheNconcentrationwouldinﬂuence lignindegradationisstillunknown.

5. Conclusions

OurstudyshowedthatbothNenrichmentandrainfallacceler- atethedecompositionofplantlitterinAfricansavanna,indicating theimportanceofnutrientreleaseinsupportofplantproductiv- ity.TheeffectofNadditionappearstooccurindependentlyofan increaseintheNconcentrationofthelitter,suggestinginternal cyclingandrapidNturnoverinthesystem.Previousstudiespro- jectedincreasedatmosphericNdepositionandrainfallvariability inthesavannawherebothsoilmoistureandNavailabilitylimit plantproductivity.

ThelitterNandPreleasedduringthedryseasonaredeposited inthesoilsforplantuptakeontheonsetofwetseason,supporting plantgrowthandproductioninthenextgrowingseason.Releaseof NandPduringthedryseasonwasnotinparallelwithCreleasesug- gestinganenhancedCstorageinlitterwhenNandPreleaselimits thedecompositionratesinthelongterm.However,futurestudies needtoconsidertheroleofphotodegradationoncarbonlossand ligninolyticenzymesonlignindegradation.Thebalancebetween plantproductionandlitterdecompositionwilldeterminethecar- bonsequestration,carbondioxideemissionsandthefeedbackto globalclimatechange.

Acknowledgments

Fundingfor the researchwas provided by theUniversity of Florida, the Swiss National Centre of Competence in Research (NCCR)andaSmithsonianTropicalResearchInstitute–Levinson Fellowship.WethankElkanahKorirforﬁeldassistance,theMpala ResearchCentreforlogisticalsupport,CETRADandtheUniversityof Nairobiforprovidingaccesstolaboratories,JamesColeeforstatisti- calsupport,theWetlandBiogeochemistryLaboratoryforanalytical support,andtheOfﬁceofthePresidentinKenyaforallowingusto conductthisresearchinKenya.JacobR.Goheenprovidedhelpful commentsonthepaper.

References

Aerts,R.,2006.Thefreezerdefrosting:globalwarmingandlitterdecomposition ratesincoldbiomes.J.Ecol.94,713–724.

Aerts,R.,Bakker,C.,Decaluwe,H.,1992.Rootturnoverasdeterminantofthecycling ofC,N,andPinadryheathlandecosystem.Biogeochemistry15,175–190.

Ahn,P.M.,Geiger,L.C.,1987.SoilsofLaikipiaDistrict.KenyaSoilSurveyProject, ReconnaissanceSoilSurveyReportno.R-13.MinistryofAgriculture,National AgriculturalLaboratories,Nairobi,pp.128,p.ISN:11427.

Austin,A.T.,Vitousek,P.M.,2000.Precipitation,decompositionandlitterdecom- posabilityofMetrosiderospolymorphainnativeforestsonHawaii.J.Ecol.88, 129–138.

Austin,A.T.,Vivanco,L.,2006.Plantlitterdecompositioninasemi-aridecosystem controlledbyphotodegradation.Nature442,555–558.

Bond,W.J.,2008.WhatlimitstreesinC4grasslandsandsavannas?Annualreview ofecology.Evol.Syst.39,641–659.

Bryant,D.M.,Holland,E.A.,Seastedt,T.R.,Walker,M.D.,1997.Analysisoflitter decompositioninanalpinetundra.Can.J.Bot.76,1295–1304.

Couteaux,M.M.,Bottner,P.,Berg,B.,1995.Litterdecomposition,climateandlitter quality.TrendsEcol.Evol.10,63–66.

Deshmukh,I.,1985.DecompositionofgrassesinNairobi-National-Park,Kenya.

Oecologia67,147–149.

Dubeux,J.C.B.,Sollenberger,L.E.,Interrante,S.M.,Vendramini,J.M.B.,Stewart,R.L., 2006.LitterdecompositionandmineralizationinBahiagrasspasturesmanaged atdifferentintensities.CropSci.46,1305–1310.

Duvdevani,S., 1953. Dewgradients in relation to climate soil and topogra- phy,indesertresearch.In:ProceedingsoftheDesertSymposium,Jerusalem, pp.136–152.

Elkins,N.Z.,Whitford,W.G.,1982.Theroleofmicroarthropodsandnematodesin decompositioninasemi-aridecosystem.Oecologia55,303–310.

Epstein,H.E.,Burke,I.C.,Lauenroth,W.K.,2002.Regionalpatternsofdecomposition andprimaryproductionratesintheUSGreatPlains.Ecology83,320–327.

Facelli,J.M.,Pickett,S.T.A.,1991.Plantlitter–itsdynamicsandeffectsonplant communitystructure.Bot.Rev.57,1–32.

(8)

Fearnside, P.M., 2000. Globalwarming and tropical land-use change: green- housegasemissionsfrombiomassburning,decompositionandsoilsinforest conversion,shiftingcultivationandsecondaryvegetation.Clim.Change46, 115–158.

Ferreira,V.,Grac¸a,M.A.S.,deLima,J.L.M.P.,Gomes,R.,2006.Roleofphysicalfragmen- tationandinvertebrateactivityinthebreakdownrateofleaves.Arch.Hydrobiol.

165,493–513.

Fornara,D.A.,DuToit,J.T.,2008.Browsing-inducedeffectsonleaflitterqualityand decompositioninasouthernAfricansavanna.Ecosystems11,238–249.

Fosu,M.,Kuhne,R.F.,Vlek,P.L.G.,2007.Mineralizationandmicrobialbiomass dynamicsduringdecompositionoffourleguminousresidues.J.Biol.Sci.7, 632–637.

Gartner,T.B.,Cardon,Z.G.,2004.Decompositiondynamicsinmixed-speciesleaf litter.Oikos104,230–246.

Gill,R.A.,Burke,I.C.,2002.InﬂuenceofsoildepthonthedecompositionofBouteloua gracilisrootsintheshortgrasssteppe.PlantSoil241,233–242.

Gindaba,J.,Olsson,M.,Itanna,F.,2004.Nutrientcompositionandshorttermrelease fromcrotonmacrostachyusDel.andMillettiaferruginea(Hochst)Bakerleaves.

Biol.Fert.Soils40,393–397.

Goheen,J.R.,Palmer,T.M.,Charles,G.K.,Helgen,K.M.,Kinyua,S.N.,etal.,2013.Piece- wisedisassemblyofalarge-herbivorecommunityacrossarainfallgradient:

TheUHURUExperiment.PLoSONE8(2),e55192,http://dx.doi.org/10.1371/

journal.pone.0055192.

Gonzalez,G.,Seastedt,T.R.,2001.Soilfaunaandplantlitterdecompositionintropical andsubalpineforests.Ecology82,955–964.

Hamadi,Z.,Steinberger,Y.,Kutiel,P.,Lavee,H.,Barness,G.,2000.Decompositionof Avenasterilislitterunderaridconditions.J.AridEnviron.46,281–293.

Hobbie,S.E.,Gough,L.,2004.Litterdecompositioninmoistacidicandnon-acidic tundrawithdifferentglacialhistories.Oecologia140,113–124.

Hobbie,S.E.,Vitousek,P.M.,2000.NutrientlimitationofdecompositioninHawaiian forests.Ecology81,1867–1877.

Jama,B.A.,Nair,R.K.R.,1996.Decompositionandnitrogenmineralizationpatterns ofLeucaenaleucocephalaandCassiasiameamulchundertropicalsemi-aridcon- ditionsinKenya.PlantSoil179,275–285.

Kaspari,M.,Garcia,M.N.,Harms,K.E.,Santana,M.,Wright,S.J.,Yavitt,J.B.,2008.

Multiplenutrientslimitlitterfallanddecompositioninatropicalforest.Ecology 11,35–43.

Kochy,M.,Wilson,S.D.,1997.Litterdecompositionandnitrogendynamicsinaspen forestandmixed-grassprairie.Ecology78,732–739.

Kouno,K.,Tuchiya,Y.,Ando, T.,1995.Measurementofsoilmicrobialbiomass phosphorusbyananionexchangemembranemethod.SoilBiol.Biochem.27, 1353–1357.

Leonard,S.,Fluckiger,W.,1989.Effectsofcationleachingonmineralcyclingand transpiration:investigationswithbeechseedlingsFagussylvaticaL.NewPhytol.

111,173–179.

Liu,K.,Sollenberger,L.E.,Silveira,M.L.,Vendramini,J.M.B.,Newman,Y.C.,2011.

Grazingintensityandnitrogenfertilizationaffectlitterresponsesin‘Tifton85 Bermudagrasspastures:II.Decompositionandnitrogenmineralization.Agron.

J.103,163–168.

Lupwayi,N.Z.,Haque,I.,1999.Leucaenahedgerowintercroppingandcattlemanure applicationintheEthiopianhighlands.I.Decompositionandnutrientrelease.

Biol.Fert.Soils40(28),182–195.

Mafongoya,P.L.,Nair,P.K.R.,Dzowela,B.H.,1997.Multipurposetreepruningsasa sourceofnitrogentomaizeundersemiaridconditionsinZimbabwe.Agroforest.

Syst.35,57–70.

Meentemeyer,V.,1978.Macroclimateandlignincontroloflitterdecomposition rates.Ecology59,465–472.

Melillo,J.M.,Aber,J.D.,Muratore,J.F.,1982.Nitrogenandlignincontrolofhardwood leaflitterdecompositiondynamics.Ecology63,621–626.

Moretto,A.S.,Distel,R.A.,Didone,N.G.,2001.Decompositionandnutrientdynamic ofleaflitterandrootsfrompalatableandunpalatablegrassesinasemi-arid grassland.Appl.SoilEcol.18,31–37.

Mtambanengwe,F.,Kirchmann,H.,1995.Litterfromatropicalsavannawoodland (miombo):chemicalcompositionandCandNmineralization.SoilBiol.Biochem.

27,1639–1651.

Mugendi,D.N.,Nair,P.K.R.,1997.Predictingthedecompositionpatternsoftree biomassintropicalhighlandmicroregionsofKenya.Agroforest.Syst. 35, 187–201.

Mugendi,D.N.,Nair,P.K.R.,Mugwe,J.N.,O’Neill,M.K.,Swift,M.J.,Woomer,P.L.,1999.

Alleycroppingofmaizewithcalliandraandleucaenainthesub-humidhigh- landsofKenyaPart2.Biomassdecomposition,Nmineralization,andNuptake bymaize.Agroforest.Syst.46,51–54.

Nakagami,K.,Sakanoue,S.,Takahashi,S.,2010.Estimationofnitrogenreleasefrom decomposingZoysiajaponicalitterusingitsrelationshiptomassloss.Jpn.Grassl.

Sci.56,177–182.

NashIII,T.H.,1996.Nutrientelementalaccumulationandmineralcycling.In:Nash III,T.H.(Ed.),Lichens.Cambridge,UnitedKingdom,pp.136–153.

Ohiagu,C.E.,Wood,T.G.,1979.GrassproductionanddecompositioninSouthern GuineaSavanna,Nigeria.Oecologia40,155–165.

Oladoye,A.O.,Ola-adams,B.A.,Adedire,M.O.,Agboola,D.A.,2008.Nutrientdynamics andlitterdecompositioninLeucaenaleucocephala(Lam.)DeWitplantationin theNigerianderivedsavanna.WestAfr.J.Appl.Ecol.13,96–103.

Palm,C.A.,Sanchez,P.A.,1990.Decompositionandnutrientreleasepatternsofthe leavesof3tropicallegumes.Biotropica22,330–338.

Peyraud,J.L.,Astigarraga,L.,1998.Reviewoftheeffectofnitrogenfertilizationon thechemicalcomposition,intake,digestionandnutritivevalueoffreshherbage:

consequencesonanimalnutritionandNbalance.Anim.FeedSci.Tech.72, 235–259.

Ruinen,J.,1961.Thephyllosphere.PlantSoilXV(2),81–109.

Swift,M.J.,Heal,O.W.,Anderson,J.M.,1979.DecompositioninTerrestrialEcosys- tems.BlackwellScientiﬁcPublications,Oxford.

TukeyJr.,H.B.,Wittwer,S.H.,Tukey,H.B.,1957.Leachingofcarbohydratesfromplant foliageasrelatedtolightintensity.Science126,120.

TukeyJr.,H.B.,1969.Implicationsofallelopathyinagriculturalplantscience.Bot.

Rev.35,1–16.

TukeyJr.,H.B.,1970.Theleachingofsubstancesfromplants.Annu.Rev.Psychol.21, 305–324.

VanSoest,J.P.,1970.ForageandFiberAnalysisHandbookNo.379.USDA,Washing- ton,DC.

Valentini,R.,Matteucci,G.,Dolman,A.J.,etal.,2000.Respirationasthemaindeter- minantofcarbonbalanceinEuropeanforests.Nature404,861–865.

Vitousek,P.M.,Turner,D.R.,Parton,W.J.,Sanford,R.L.,1994.Litterdecomposition ontheMauna-Loaenvironmentalmatrix,Hawaii–patterns,mechanisms,and models.Ecology75,418–429.

Waite,R.,1970.Thestructuralcarbohydratesandtheinvitrodigestibilityofarye- grassandacocksfootattwolevelsofnitrogenousfertilizer.J.Agric.Sci.74, 457–462.

Went,F.W.,1955.Fog,Mist,Dew,andOtherSourcesofWater.YearbookofAgricul- ture.

Witkamp,M.,Olson,J.S.,1963.Breakdownofconﬁnedandnonconﬁnedoaklitter.

Oikos14,138–147.

Yang,C.,Crowley,D.E.,Borneman,J.,Keen,N.T.,2001.Microbialphyllospherepopu- lationsaremorecomplexthanpreviouslyrealized.PNAS98,3889–3894.

Young,T.P.,Patridge,N.,Macrae,A.,1995.Long-termgladesinacaciabushlandand theiredgeeffectsinLaikipiaKenya.Ecol.Appl.5,97–108.

Young,T.P.,Okello,B.,Kinyua,D.,Palmer,T.P.,1998.KLEE:along-termmulti-species herbivoreexclusionexperimentinLaikipiaKenya.Afr.J.RangeForageSci.14, 94–102.