JournalofAridEnvironments74(2010)1240e1247
ContentslistsavailableatScienceDirectJournalofAridEnvironmentsjournalhomepage:www.elsevier.com/locate/jaridenvAbovegroundbiomassinthreeSonoranDesertcommunities:VariabilitywithinandamongsitesusingreplicatedplotharvestingA.Búrquez,A.Martínez-Yrízar*,S.Nú?ez,T.Quintero1,A.AparicioDepartamentodeEcologíadelaBiodiversidad,InstitutodeEcología,UniversidadNacionalAutónomadeMéxico,ApartadoPostal1354,Hermosillo83000,Sonora,MéxicoarticleinfoArticlehistory:Received8December2009Receivedinrevisedform12February2010Accepted13April2010Availableonline14May2010Keywords:AllometryAridecosystemsCarbonstocksDesertscrubCanopyvolumeThornscrubabstractTotalabovegroundbiomass(TAGB)valuesarereportedforthe?rsttimefortheSonoranDesert.Har-vestingofreplicatedplotsinthreesitesdifferingingeomorphologyandvegetationstructureallowedthestatisticalmeasurementofspatialvariabilitywithinandamongsites.Linear,powerlog-transformed,andpowernon-linearregressionswereusedtorelateTAGBwithplantmetrics.CanopyvolumeexplainedthelargestproportionofTAGBvariance(r2?0.74e0.94)inthethreemodels.Allmodelswerehighlysigni?cant,butthenon-linearwasmorerobust,hadabetterdistributionofresiduals,anddidnotrequiredataback-transforming,allowingtheaccurateestimationofbiomassattheplotlevelusingsimplemeasurementsofvegetation.TAGBrangedfrom6.99Mghaà1(Plains)to29.24Mghaà1(Arroyos)inthedesertscrub,andwasintermediateinthethornscrub(Hillsides:13.03Mghaà1).Within-sitevariabilityofTAGBwashighestinthePlainsandlowestintheHillsides.TheweightedmeanAGBwas16Mghaà1.Assuminga1:1root:shootratio,averagetotalbiomasscouldbe32Mghaà1.ThisremarkablevaluesetstheSonoranDesertasapotentialreservoirofupto4.4%ofthetotalworlddesertsplantbiomassinlessthan1%ofthedeserts’globalarea.ó2010ElsevierLtd.Allrightsreserved.1.IntroductionAccuratequanti?cationsofcarbonstorageinnaturalvegetationareimportanttodeterminethemagnitudeofbiomassandcarbonstocksinterrestrialecosystemsandtoimproveourunderstandingofhowecosystemsmightrespondtoclimatechange(LioubimtsevaandAdams,2004;LittonandKauffman,2008).Theyarealsoimportanttoproperlyassesstheimpactsofland-use/land-coverchangeontheterrestrialcarbonbalance(Jaramilloetal.,2003;Bonino,2006;Piaoetal.,2009).Differentmethodshavebeenusedtodeterminetheamountofabovegroundbiomass(AGB)andcarbonstocksinterrestrialvegetation.Moststudieshavereliedongeneralizeddimensionanalysis(sensuAllaby,1998),anindirectmethodconsistingofestablishingsize-biomassallometricrela-tionshipsofselectedspecies.Bythistechnique,thesizeoftheplantspriortodestructivesampling(harvestingandweighing)ismeasured,andlater,speci?cplantsize-biomassrelationshipsareestablished(Baskerville,1972;LiandXiao,2007;LittonandKauffman,2008;Návar,2009).Thismethodiseasytoapplyin*Correspondingauthor.Tel.:t526622139303;fax:t525556226536.E-mailaddress:angelina@unam.mx(A.Martínez-Yrízar).1Presentaddress:DepartamentodeInvestigacionesCientí?casyTecnológicas,UniversidaddeSonora,Hermosillo83250,Sonora,México.0140-1963/$eseefrontmatteró2010ElsevierLtd.Allrightsreserved.doi:10.1016/j.jaridenv.2010.04.004plantcommunitiescomposedofoneorafewcodominantspecies(Patten,1978;Rundeletal.,1982;Návaretal.,2002;Northupetal.,2005).However,inmorediverseandstructurallycomplexcommunities,alargernumberofsamplesareoftenneededtoaccountforthehigherspatialvariabilityinbiomass.Directmethodsofbiomassmeasurementinvolvingtheharvestingofwholeplotsallowthevalidationofothermethodssuchasdimensionanalysismodelsorremotesensingtechniquesaimedtogeneralizebiomassestimates(Houghtonetal.,2009).Awidespreadperceptionofdrylandsisthatthetotalamountofbiomassissmallandevenlydistributed.Thisideaisbasedontheassumptionthatvegetationcoverinaridandsemi-aridecosystemsethatrepresentasizableportionoftheworld’slandarea(29.8%;Lal,2004)eissparsecomparedtothemorehumidregions.However,inmanydesertareas,biomassandproductivitycanbeconsiderablyhighandvariable,evenatsmallspatialscales(hundredsofmetres;HuennekeandSchlesinger,2006).Thisvariationinthedistributionofbiomassisproximallyaffectedbydifferencesinsoilwateravail-ability(Noy-Meir,1973;EvenariandNoy-Meir,1985;Búrquezetal.,1999;López-PortilloandMonta?a,1999;Whitford,2002;Rangoetal.,2006),although,atsmallerscales,itisknownthatbioticin?uencesplayamajorroleindeterminingvegetationdynamics(Valiente-BanuetandEzcurra,1991;BúrquezandQuintana,1994;Callaway,1995;Bisigatoetal.,2009).A.Búrquezetal./JournalofAridEnvironments74(2010)1240e12471241
IntheSonoranDesert,spatialheterogeneityinvegetationstruc-tureeevenamongadjacentsitesehasbeenrelatedtostrongdiffer-encesinprimaryproductivity(MayaandArriaga,1996;Martínez-Yrízaretal.,1999;SponsellerandFisher,2006),microbialpotentialofsoils(Nú?ezetal.,2001)anddecompositionrates(ArriagaandMaya,2007;Martínez-Yrízaretal.,2007).Plantbiomassisalsolikelytoexhibitstrongspatialvariability,butitsquanti?cationhasbeenaneglectedaspectofecosystemstudiesinthisregion.TheaimsofthepresentstudyweretomeasuredesertTAGBanditsspatialvariabilityapplyingthedirectmethodofplotharvesting.Speci?cobjectiveswere:(1)toquantifyTAGBandcarbonstocksinthreeextensiveSonoranDesertsitesstronglydifferentiatedbyvegetationstructure,topographyandthepresenceofsmalldrainagesystems,(2)todevelopthebestmultispeciesregressionmodelsfortheestimationofTAGBharvestingreplicatedplots,and(3)tobrie?ydiscusstherelevanceoftheSonoranDesertcarbonstocksonlocalandglobalscenariosofcarbonbalance.2.Materialsandmethods2.1.StudyareaThestudywasconductedinthesouthernSonoranDesertinanareaclosetoHermosillo,Sonora,Mexico(29??010N,110??570W,200e600melevation).Theclimateiswarmanddry,withmostoftheannualrainfall(348mm;mean1966e2003)concentratedduringsummertimeasmonsoonrains(JulyeSeptember,237mm;ComisiónNacionaldelAgua,Hermosillo).Meanannualtemperatureis24.6??C,withameanmonthlyrangeof16.6??C(January)to32.3??C(July).ThesoilsareAridisols(USDA,1999)mainlyderivedfromdecompositionofgranitesandvolcanicrocks.Thecontentofclayishighondepositionplains.Sandysoilspredominatealongthearroyos,whiletheslopesarecharacterizedbycoarse,stonysoilsamongthegraniteoutcropsandassociateddikes.Thevegetationbelongsmostlytothe“PlainsofSonora”subdivisionoftheSonoranDesert(ShreveandWiggins,1964)butslopeswithnorthernandeasternorientationsupportFoothillsThornscrubat250e600melevation.Theplainsaredrainedbyanetworkofephemeraldesertarroyosthatsupportspeciesrich,arborescentvegetationwithadensercanopycover(Búrquezetal.,1999;Martínez-Yrízaretal.,1999).2.2.StudysitesThreesites(namedPlains,ArroyosandHillsideshereafter)stronglydifferentiatedbytopographyandthepresenceofminorwatercourseswerechosenwithinawell-preservedareaofabout800ha.TheseincludethethreemajorvegetationtypesfoundthroughoutcentralSonora(Martínez-Yrízaretal.,2007).ThePlainsarecharacterizedbyvegetationtypicaloftheextensivearborescentplainswhichsupportthesparsestplantcoverinthearea.Afewdominanttrees,mostlyironwoodOlneyatesotaA.Gray,andpalo-verdeParkinsoniamicrophyllaTorr.,occuraslong-livedisolatedindividualsinamatrixofshort-livedherbaceousplants,spacedshrubs,andbareground(ShreveandWiggins,1964).TheArroyosarerestrictedtotheneighbourhoodofdry,sandy,shallowchannels3e10mwidethatdraintheplainsduringshortperiodsoftime,andrunonlyinresponsetoprecipitation(channeltype1e3ofSponsellerandFisher,2006).Theyarethemosthumidsitesandhavethedensestvegetationintheareawiththeexceptionofthelargerdrainageswithmesquiteriparianforests.ThedominantspeciesarethesameasthosefoundinthePlains,butmanyothertreespeciesarecodominant(CoursetiaglandulosaA.Gray,Eysen-hardtiapolystachya[C.G.Ortega]Sarg.),severalshrubs,vines,andherbaceousplantswithtropicalaf?nitiesareonlyfoundhere.TheHillsidesarerepresentativeoftheFoothillsThornscrubvegetation(FelgerandLowe,1976;Búrquezetal.,1999)whichmainlydevelopsonnorth-facingslopesofdeserthills.Here,thevegetationismoretropicalincharacter,morediverse,andmoreuniformincoverthanontheplains.ProminentspeciesincludeAcaciawill-ardianaRose,CrotonsonoraeTorr.,Jatrophacordata(C.G.Ortega)Muell.Arg.,MimosadistachyaandParkinsoniamicrophylla,alongwithseveralspeciesofCylindropuntia,manyvinesandshrubs.Beinglocatedlessthan1kmfromeachother,allthreeplantcommunitiesshareacommonmacro-andmeso-climate,butdiffermarkedlyinmicroclimaticconditions.Thesoilsareslightlyalkaline(pHH2O7.3e7.8)anddifferinchemicalcomposition(Martínez-Yrízaretal.,1999).Comparedtothetwoothersites,thesoilsinthePlainsarethreetimeslowerintotalcarbon(4.6vs11.4Arroyos,and15.8gkgà1Hillsides)andnitrogen(0.3vs1.0Arroyos,and1.1gkgà1Hillsides),and?vetimesricherinMg(53.3vs12.3Arroyos,and10.2mgkgà1Hillsides)andCa(127.1vs30.8Arroyos,and22.5mgkgà1Hillsides).2.3.Harvestofbiomassand?elddesignOnanaerialimage,thestudyareawaspartitionedintoeachofthethreecommunitytypes,andtwenty5?5msamplingplotswererandomlyselectedwithineachcommunity.Allperennialplantsrootedwithineachplotwereidenti?edandassignedintooneofsixcategories:trees(largeplants4tmtall),largeshrubs(largemultipletrunks1.5e4mtall),mediumshrubs(0.5e1.5mtall),smallshrubs(lowerthan0.5mtall),vines,andcacti.Theirmaximumheight(h)wasdeterminedusingameasuringpole,andtheircanopycoverwasestimatedbymeasuringtwoorthogonaldiametersandtheirradii(r1andr2),thencalculatingthecanopyareaby?ttingtoanellipseoftheform:A?pr1r2.Canopyvolumewasestimatedasanellipticalcylinder:V?pr1r2h.Eachindividualplantoneachplotwascutatgroundlevel,sortedbyclippingintofractions(trunkandwoodystems,deadwood,andcurrentyear’stwiggrowth),andweightedfreshtothenearest5gusingabeamscale.Allleavesstillattachedtotheplantsatharvesttime(autumn)werediscarded.Forlargeplants,asubsampleofonequarterofthetotalmasswassortedintocomponentsandweighed.Thisportionwasusedtocalculatetherelativecontributionofeachcomponenttothetotalmassoftheplantandtherestwasdiscarded.Subsamplesofknownfreshweightfromallfractionsweretakentothelaboratoryanddriedinanovenat95??Ctoconstantmassandweighedtothenearest0.1g.Speci?cwoodgravitywasdeterminedinsamplesofwood(hexahedraabout2.5cmeachside)fromselectedindividualsofeachspecies.Itwasmeasuredastheratiooftheovendrymassdividedbythemassofthewaterdisplacedwhenthesample(coveredwithathinlayerofwax)wascompletelysubmerged.Carboncontentwasestimatedasabout50%ofthedrymass(Jaramilloetal.,2003;Houghtonetal.,2009).Biomasswascalculatedasthesumofallindividualswithineachplotweighedbytheplotarea(drymass,gmà2).Thesameprocedurewasappliedforeachoftheothervariableswithineachplot.2.4.DataanalysesTheminimumnumberofplotsneededtoobtainarepresenta-tivesamplingwasdeterminedbythecalculationofthecumulativeaveragedrymassperplotbyrandomlyselectingplots(withoutreplacement)andprogressivelyaddingsamples.Weiteratedthisproceduremanytimestosimulatecumulativeaveragecurveswithdifferentsamplingorders.Theminimumnumberofsampleswasdeterminedasthepointwhereallcurvesconvergedwithinthe95%con?denceintervalofthefullcomplementofsamples.OnewayANOVAwasusedtotestforsitedifferencesinmeanTAGB(afterappropriatetransformations).Levene’stestwasusedtotestforhomocedasticity,andpost-hocmultiplecomparisonswere1242A.Búrquezetal./JournalofAridEnvironments74(2010)1240e1247
performedusingaDuncan’stest,additionallyarobustT2Tam-hane’stestwasusedwhenvarianceswerenothomogeneous.Thebestpossiblepredictiveequationwasdeterminedbyrelatingthetotalplantcover,meanandmaximumheight,andthecompositevalueofcanopyvolumewithTAGBusingplotsasreplicates.Astherewerelargedifferencesamongspeciesinwooddensity,meanplotspeci?cwoodgravitywasalsoincludedasapredictorvariable.Torelatebiomasswithplantdimensions,weusedthemeanvaluesofbiomassandvegetationwithineachplotasreplicatesandamultipleregressionwasperformedto?ndthebestlinearcombinationofvariables.Forpredictionpurposes,we?ttedthedatatosimpletwo-parameterequationsusingthreeapproaches:(1)bytheordinaryleastsquares(OLS)methodforasimplelinearrelationship,(2)bytheOLSonthelog10-trans-formeddataforalinear?ttoapowerequation,and(3)bymeansofanon-linearmodelto?tatruenon-linearpowerequationemployingtheLevenbergeMarquardtalgorithm(Pressetal.,2007).AllanalysesweremadeusingSPSS13(SPSSInc.Chicago,Illinois).3.Results3.1.MinimumnumberofplotsDespitethelargevariationinTAGBamongindividualplotswithineachplantcommunity,thecumulativeaveragecurvesconvergedwithinthe95%con?denceintervalofthemeanofallobservationswhen14ormoresamplesareaveraged(Fig.1).Thelargestcoef?cientofvariationinTAGBwasfoundintheArroyos,andtheleastintheHillsides(Fig.1).3.2.Live,dead,andtotalabovegroundbiomassAtotalof1949individualperennialplants(trees,shrubs,vinesandcacti)representing55speciesand22familiesweremeasuredandharvestedontheplots(Table1).Fromthistotal,26werestandingdeadplants,mainlyofAbutilonincanum(Link)SweetintheHillsides(19individuals).Thetotalnumberofplantsharvestedperplotwashighlyvariableoneachsite:2e57individualsperplotinthePlains,6e67intheArroyosand11e74intheHillsides.Asexpected,thenumberoflow-densityplots(<25plants)wasmuchhigherinthesparservegetationofthePlains(14plots)thanintheothertwosites(6plotsinArroyos,4plotsinHillsides).ThedistributionofthearithmeticvaluesofTAGBbyplotwasskewedtowardssmallvaluesinallthreesites(Fig.2aec),buttherewasagradientfromhigh,inthePlains,toalmostnoskewnessintheHillsides.TheArroyosexhibitedamarkedleptokurticdistri-bution,andthePlainsandtheHillsideswereclosetomesokurtic(Fig.2aec).Withalog10-transformationofthedata,thePlainsandtheArroyossitesfollowedcloselyanormaldistribution,buttheHillsidesweresomewhatnegativelyskewed,andleptokurtic.Within-sitevariabilitywasmuchlargerinthePlainswherethestandarderrorofthemeanwasabouttwotimeslargerthanintheothertwoplantcommunities(Fig.2def).ThePlainshadthelowestmeanTAGB(640gmà2),andtheArroyosthelargest(2567gmà2).Bygrowthform,treesmadeupthelargestTAGBcomponentinthePlains(71%)andArroyos(59%)fol-lowedbytheshrubs(25%and29%,respectively),whileintheHill-sides,treesandshrubshadasimilarcontributiontoTAGB(43%and44%,respectively).Vinescontributedwithonlyabout2%ofTAGBintheHillsidesandArroyos,andevenlessinthePlains(0.7%).ThecontributionofcactitoTAGBvariedgreatlyamongsites,fromalmost3%inthePlainstoabout11%intheArroyosandHillsides(Table2).Thedistributionofthearithmeticvaluesofliveabovegroundbiomass(LAGB,notincludingleavesandherbs)wasalmostiden-ticalasthatofTAGB(livetdead).ThetotalmeanLAGBvaried4 PLAINS
)m gk( 3 egarev A B2 GA T evital1 umuC0 4
ARROYOS
)mg k( 3
egarev A B2
GA T e vital1
umuC0 4
HILLSIDES
)mg k( e3
garev A B2
GA T e vitalu1
muC0
0 5 Number of Plots
10 15 20
Fig.1.Cumulativeaveragecurvesforthemeantotalabovegroundbiomass(TAGB)inthreeplantcommunitiesoftheSonoranDesert,Mexico(onlyteniterationsshown).Theheavydash-dotlineisthemeanofcumulativemeans.Thedottedlinesineachplotindicatethe95%con?denceintervalofthemean.
signi?cantlywithsite(afteralog10-transformationofdata:F2,57?7.525,P<0.001)from578gmà2inthePlains,to2091gmà2intheArroyos.ThePlainshadaLAGBvalue1.7timeslowerthantheHillsides,and3.6timeslowerthantheArroyos(Table2).Themeanamountofstandingdeadstemsandbranches(i.e.,necromass)alsovariedsigni?cantlywithsite(Table2),makingup9.7,13.5and18.5%oftheTAGBinthePlains,HillsidesandArroyossites,respectively.Thecurrentyear’sgrowthoftwigsrepresentedasmallfractionofthetotalLAGBineachsite(1.3e2.3%),withthehighestfractionfoundinthePlains.A.Búrquezetal./JournalofAridEnvironments74(2010)1240e12471243
Table1Totalnumberofindividualsperspeciesharvestedin20plots5?5minthreeplantcommunitiesintheSonoranDesert,Mexico.P?Plains,A?Arroyos,H?Hillsides.GF?growthform:H?herb,P?parasite,T?tree,V?vine,LC?largecacti,LS?largeshrub,SC?smallcacti,SS?smallshrub.FamilyGFSpeciesSite/IndividualsPAHAcanthaceaeSSCarlowrightiaarizonicaA.Graye122AcanthaceaeSSJusticiacalifornica(Benth.)D.N.Gibs.ee36AchatocarpaceaeLSPhaulothamnusspinescensA.Graye1eAsclepiadaceaeVMateleacordifolia(A.Gray)Woodson157AsteraceaeSSBebbiajuncea(Benth.)Greenee1eAsteraceaeSSBrickelliacoulteriA.Grayee30AsteraceaeSSEnceliafarinosaA.GrayexTorr.2672995AsteraceaeSSTrixiscalifornicaKell.e31BurseraceaeTBurserafagaroides(H.B.K.)var.ee2elongataMcVaugh&RzedowskiBurseraceaeTBurseralaxi?oraS.Watson2714BurseraceaeTBurseramicrophyllaA.Grayee3CactaceaeSCMammilariagrahamii132Engelm.tM.microcarpaEngelm.CactaceaeSCMammilariamainaeK.Brandegeee1eCactaceaeLCOpuntiafulgidavar.mammillata354e(Schott.exEngelm.)J.M.Coult.CactaceaeLCOpuntiagosselinianaF.A.C.Weberee10CactaceaeLCOpuntialeptocaulisD.C.9151CactaceaeLCOpuntiathurberiEngelm.722817CactaceaeLCOpuntiaversicolorEngelm.e3220exJ.M.CoulterCactaceaeLCPeniocereusstriatusee2(Brandegee)Buxb.ConvolvulaceaeHEvolvulusalsinioidesvar.ee2acapulcensisOoststr.ConvolvulaceaeVMerremiapalmeri(S.Watson)Halliere2eCucurbitaceaeVIbervilleasonorae(S.Watson)Greenee11EuphorbiaceaeLSAdeliabrandegeeiV.W.Stainm.ee1EuphorbiaceaeLSCrotonsonoraeTorr.e2193EuphorbiaceaeLSJatrophacardiophylla(Torr.)33581Muell.Arg.EuphorbiaceaeTJatrophacordataee51(C.G.Ortega)Muell.Arg.FabaceaeTAcaciaconstrictaBenth.e2eFabaceaeTAcaciawillardianaRoseee1FabaceaeLSCaesalpiniapalmeriS.Watson110eFabaceaeTCoursetiaglandulosaA.Grayee18FabaceaeSSDesmanthuscovilleie4e(Britton&Rose)WigginsFabaceaeTEysenhardtiaorthocarpae45(A.Gray)S.WatsonFabaceaeHMarinaparryi(Torr.&A.e11Gray)BarnebyFabaceaeLSMimosadistachyaCav.var.622104laxi?ora(Benth.)BarnebyFabaceaeVNissoliaschottii(Torr.)A.Graye29FabaceaeTOlneyatesotaA.Gray13eFabaceaeTParkinsoniamicrophyllaTorr.ee11FabaceaeHSennacovesii(A.Gray)H.S.ee1Irwin&BarnebyFouquieriaceaeTFouquieriamacdougaliiNashe3eKrameriaceaeSSKrameriaerectaWilld.exSchulte2eMalpighiaceaeVJanusiacalifornicaBenth.tJ.41555gracilisA.GrayMalpighiaceaeVJanusialinearisWiggins51413MalpighiaceaeVMascagniamacroptherae3e(Moc.&SesséexD.C.)Mied.MalvaceaeSSAbutilonincanum(Link)Sweet8343Passi?oraceaeVPassi?orafoetidaL.ee1RubiaceaeLSRandiaobcordataS.Watson19eSapindaceaeVCardiospermumcorindumL.21925SolanaceaeLSLyciumspp.e19SolanaceaeLSNicotianaglaucaGraham31eSterculiaceaeHAyenia?liformisS.Watsonee62ViscaceaePPhoradendroncalifornicumNutt.e1eVitaceaeVCissustrifoliata(L.)L.e1eZygophyllaceaeLSGuaiacumcoulteriA.Gray1eeTotalnumberofplantsharvested3877837793.3.BestmodelforpredictingbiomassfromharvestingplotsMultipleregressionanalysissingledoutcanopyvolumeasthebestpredictorofTAGB(eitherusingthearithmeticorlog-trans-formedvalues);anyotherstructuralvariableaddednosigni?cantexplainedvariance(r2).AstheinteractiontermSite*Canopyvolumewashighlysigni?cantwhentestingfordifferencesforTAGBamongcommunities,anANCOVAshowedsigni?cantdifferencesamongslopesusingthearithmeticdata(F2,54?13.59,p<0.001),butnotforthelog-transformeddata(F2,54?0.43,p?0.65).Theinterceptswerealsosigni?cantlydifferentamongsitesinbothcases(arith-metic:F3,54?4.14,p?0.01;logelog:F3,54?1720.93,p<0.001).Thelinearmodelexplainedconsiderablelessvariancethantheothertwomodels(Table3).Agraphicalexaminationofthedistri-butionofresidualsindicatedapoor(nonrandom)errordistribu-tionforthelinearmodel,amoderate,butstillcurvilineardistributionoferrorsforthepowerlogelogmodel,andaverygooddistributionofpredictionerrorsforthenon-linearmodel(Fig.3).Thus,theequationsderivedfromthenon-linearmodelarerec-ommendedasthebestmethodforpredictingbiomassfromhar-vestingplots.4.Discussion4.1.VariationinTAGBamongandwithinplantcommunitiesDesertecosystemsarespatiallyhighlyheterogeneous(Huennekeetal.,2001;Lal,2004;Bisigatoetal.,2009);therefore,variableamountsofplantbiomassarestoredindifferentvegetationphasesthataremainlyrelatedtochangesingeomorphology(ShreveandWiggins,1964;Noy-Meir,1973;EvenariandNoy-Meir,1985;Bisigatoetal.,2009).ThedriestofourstudysitesarethePlains,alandformthatduringthebriefdesertthunderstorms,rapidlydrainin“sheet?oods”(ShreveandWiggins,1964)thatreducein?ltrationandremovemostofthedepositedlitter.Undertheseconditions,fewspeciescanrecruitandgrow,leadingtoamarkedskewnessoftheTAGBdistribution.Skewnesscanbeexplainedbythepresenceoflong-livedscatteredphreatophytessuchasO.tesotathatonceestablishedefollowingrarerecruitmenteventsecanmodifytheenvironmenttowardsgreaterproductivityandbiomass.Theseoddindividualsallowtheestablishment,throughprocessesoffacilitation,ofmanyotherspeciesbytheeffectsofhydrauliclift,protectionfromdirectirradiance,and/orbythebuild-upofsoilandnutrientandpropaguleimports(BúrquezandQuintana,1994;Callaway,1995;Suzánetal.,1996).Inasampling,thenetresultofthesephenomenaisre?ectedbytheabundanceofmanyplotswithlittlevegetationandafewplotswithadisproportionateshareoftheTAGB.Morebalanced,butstillskeweddistributions,arefoundintheArroyos,whiletheHillsidesshowsdistributionsclosertonormality.Variancesdecreaseaccordingly.VegetationintheArroyosandHillsidesformsanalmostcontinuouscanopy,aresitesricherinsoilnutrients,andhave,atleastduringtherainyseason,abettersoilwatersupply(Búrquezetal.,1999;SponsellerandFisher,2006),allowingamuchhigherstandingcropbiomass,andmakinglessobviousanyfacili-tationprocesses.Variabilitywithinandamongsites,demandsextensive,strati?edsamplingtodeterminemeanvaluesinthepatchyvegetationofdrylands;aproblemhighlightedbyHuennekeandSchlesinger(2006).Althoughitinvolvesgreatphysicaleffortofharvestingandweighing,ourrandomsampleoftwenty5?5mquadrats(500m2)isclosetotheminimumrequiredtoprovidesoundestimatesofTAGBindesertsystemsdominatedbyarbo-rescentelements.Althoughshrubsandsmalltreesareprominentinallthreesites,themajorcomponentsofTAGBvarygreatlyamongthem.Inthe1244A.Búrquezetal./JournalofAridEnvironments74(2010)1240e1247
PLAINS
a15
Frequency1050
X=639.9±330.01K=3.19±0.51S=10.26±0.99
d
6
4
X=2.0970±0.1920K=1.57±0.99S=-0.39±0.51
2
02468
0
01234
ARROYOS
b
Frequency6
X=2567.04±815.64K=13.38±051S=3.46±0.99
e
6
44
X=3.1657±0.1048K=0.87±0.99S=-0.05±0.51
22
0
024616
0
0
1234
HILLSIDES
c
10
X=1146.4±164.7K=2.36±0.51S=1.33±0.99
f
10
X=2.9227±0.0967K=7.99±0.99S=-2.38±0.51
88
Frequency66
44
2
2
0
0
4268
TotalAGB(g10m)0
0
4123
LogTotalAGB(gm)Fig.2.Frequencydistributionsofthemeantotalabovegroundbiomass(TAGB;gmà2)in5?5mplotsinthreeplantcommunitiesoftheSonoranDesert,Mexico.Mean(X),skewness(S)andkurtosis(K)(?SEM)areindicatedineachgraph.Arithmeticdata(plotsaec),log-transformeddata(plotsdef).Thedottedlineindicatesa?ttednormaldistribution.
PlainsandArroyos,treesmakethehighestcontribution(71%and59%,respectively)whileintheHillsides,treesandlargeshrubsarecodominant(about44êch).Thedistributionofgrowthformsseemstoin?uencethevarianceinTAGB;thus,plantcommunitieswithhigherrelativeproportionoftreesshowlargervariancesthanthosewithamoreevendistributionofgrowthforms.TherelativelylargecontributionofcactitototalTAGBintheArroyosandHillsides,addcomplexitytothefunctionalrelationshipsofthevegetationinthesesites.TheunexpectedlowbiomasscontributionofcactiinthePlainsisremarkablesincecactiareconspicuousdrought-resistantplantsthatsuccessfullythriveintheselandforms(ShreveandWiggins,1964;GibsonandNobel,1986).Despitethefunctionaland?oristicrelevanceofvinesindeserts(Krings,2000),theyaredif?culttomodelintermsofdimensionanalysisastheyhavelargecanopyvolumeswithlowTAGB.However,theimpactoftheirexclusioncanleadtoasmallerroraddingnomorethan2%ofthetotalbiomass(Table2).Arelativelyhighproportionofstandingdeadbiomass(10e20%oftheTAGB),mainlyfrompartialtocompletemortalityofwoodyindividualseprobablyresultingfromepisodicdroughts(BowersandTurner,2001)dindicatesthatthisslowlydecomposingfrac-tionisanimportantcomponentofthecarbondynamicsinthesedesertecosystems.OurvaluesarecomparabletothecloselyrelatedtropicaldryforestsofthePaci?ccoastofMexico,where12%oftheTABGwasstandingnecromass(Maassetal.,2002).ThehigherproportionofnecromassintheArroyosisprobablytheconse-quenceofdifferentialratesofdecompositionthatleadtolitterandstandingdeadbiomassaccumulation;inthesesitesbothphoto-degradationandtermiteactivityarelowerthaninthePlains(Martínez-Yrízaretal.,2007).SinceourestimateofTAGBonlyincludedthewoodycompo-nent,thenon-woodybiomass(i.e.,leaves,reproductivestructuresandherbs)producedduringthegrowingseasonateachstudysitewasobtainedfromMartínez-Yrízaretal.(1999)whoreportedthis