Biogeosciences,6,2759–2778,2009 Biogeosciences www.biogeosciences.net/6/2759/2009/ ©Author(s)2009. Thisworkisdistributedunder theCreativeCommonsAttribution3.0License. Above- and below-ground net primary productivity across ten Amazonian forests on contrasting soils L.E.O.C.Araga˜o1,2,Y.Malhi1,D.B.Metcalfe1,2,J.E.Silva-Espejo3,E.Jime´nez4,D.Navarrete4,5,S.Almeida6, A.C.L.Costa7,N. Salinas1,3,O.L.Phillips9,L.O.Anderson1,E.Alvarez4,T.R.Baker9,P.H.Goncalvez7,8, J.Huama´n-Ovalle3,M.Mamani-Solo´rzano3,P.Meir12,A.Monteagudo13,S.Patin˜o4,M.C.Pen˜uela4,A.Prieto14, C.A.Quesada9,10,11,A.Rozas-Da´vila3,A.Rudas15,J.A.SilvaJr.7,andR.Va´squez13 1EnvironmentalChangeInstitute,SchoolofGeographyandtheEnvironment,UniversityofOxford,SouthParksRoad, OX13QY,Oxford,UK 2ClimateChangeandSustainabilityGroup,SchoolofGeography,UniversityofExeter,AmoryBuilding,RennesDrive, Exeter,Devon,EX44RJ,UK 3UniversiadadNacionalSanAntonioAbad,Cusco,Peru 4GrupodeEstudiodeEcosistemasTerrestresTropicales,UniversidadNacionaldeColombia,Leticia,Colombia 5Fundacio´nPuertoRastrojo,Bogota´,Colombia 6MuseuParaenseEmilioGoeldi,66077-530Belem,Brazil 7UniversidadeFederaldoPara,Belem,Para,Brazil 8UniversidadeFederaldeVicosa,Vicosa,MinasGerais,Brazil 9EarthandBiosphereInstitute,SchoolofGeography,UniversityofLeeds,LS29JT,UK 10InstititoNationaldePesquisasAmazoˆnicas,Manaus,Brazil 11DepartamentodeEcologia,UniversidadedeBras´ılia,Brazil 12SchoolofGeography,UniversityofEdinburgh,Edinburgh,UK 13JardinBotanicodeMissouri,Oxapampa,Pasco,Peru 14InstitutoAlexandervonHumboldt,ClaustrodeSanAgust´ın,VilladeLleva,Boyaca,Colombia 15UniversidadNacionaldeColombia,InstitutodeCienciasNaturales,Apartado7495,Bogota,Colombia Received: 18December2008–PublishedinBiogeosciencesDiscuss.: 25February2009 Revised: 11November2009–Accepted: 11November2009–Published: 1December2009 Abstract. The net primary productivity (NPP) of tropi- sure the major elements of productivity, we show that NPP cal forests is one of the most important and least quan- variesbetween9.3±1.3MgCha−1yr−1(mean±standarder- tified components of the global carbon cycle. Most rel- ror), at a white sand plot, and 17.0±1.4MgCha−1yr−1 at evant studies have focused particularly on the quantifica- a very fertile Terra Preta site, with an overall average of tion of the above-ground coarse wood productivity, and lit- 12.8±0.9MgCha−1yr−1. The studied forests allocate on tle is known about the carbon fluxes involved in other el- average 64±3% and 36±3% of the total NPP to the above- ements of the NPP, the partitioning of total NPP between and below-ground components, respectively. The ratio of its above- and below-ground components and the main en- above-ground and below-ground NPP is almost invariant vironmental drivers of these patterns. In this study we with total NPP. Litterfall and fine root production both in- quantify the above- and below-ground NPP of ten Amazo- creasewithtotalNPP,whilestemproductionshowsnoover- nian forests to address two questions: (1) How do Ama- all trend. Total NPP tends to increase with soil phospho- zonian forests allocate productivity among its above- and rus and leaf nitrogen status. However, allocation of NPP to below-ground components? (2) How do soil and leaf nu- below-ground shows no relationship to soil fertility, but ap- trientstatusandsoiltextureaffecttheproductivityofAma- pearstodecreasewiththeincreaseofsoilclaycontent. zonian forests? Using a standardized methodology to mea- Correspondenceto: L.E.O.C.Araga˜o ([email protected]) PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. 2760 L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests 1 Introduction stratedthatwoodproductionvariesbyuptoafactorofthree across Amazonian forests. The lowest wood productivities Plants are able to capture and accumulate atmospheric car- are found on heavily weathered oxisols (United States De- bonviaphotosynthesisorgrossprimaryproductivity(GPP), partment of Agriculture soil classification – USDA), or fer- andsynthesisoforganiccompounds. Theamountoforganic ralsols (World Reference Base soil classification – WRB) carbon retained in plant biomass over time, which results in lowland eastern Amazonia; the highest on fertile alluvial fromthedifferencebetweenGPPandautotrophicrespiration soils and inceptsols (USDA classification system) or fluvi- (Ra),isknownasnetprimaryproductivity(NPP).Globally, sols and cambisols (WRB classification system) in western ithasbeenestimatedthattheterrestrialbiospherefixesannu- Amazonia (Ecuador and Peru). More fertile sites in west- allybetween46PgC(DelGrossoetal., 2008)and63PgC ern Amazonia tend to favour fast growing, low wood den- (Grace,2004)throughNPP,approximatelythesameamount sity species (Baker et al., 2004), which are likely to allo- thatisfixedbyoceans(Fieldetal.,1998). Despitecovering caterelativelymoretowoodandleafproductionandlessto only around 13% of the total land cover area (Bartholome´ structural and chemical defences and their associated con- andBelward,2005;DelGrossoetal.,2008),tropicalforests struction and maintenance metabolic costs. A companion alonehaveamajorimpactonglobalcarboncycling,account- paperinthisissue(Chaveetal.,2009)examinespatternsof ing for about a third of overall terrestrial NPP (Field et al., canopy NPP across Amazonia in greater detail. This study 1998; Malhi and Grace, 2000; Grace, 2004; Del Grosso et suggests that soil type is not a major determinat of litterfall al.,2008). patternsacrossAmazonia,however,infertilewhitesandsoils Detailed understanding of the total NPP of tropical (5.42±1.91Mgha−1yr−1) have significantly lower litterfall forests, including both above- and below-ground productiv- productionthanothersoiltypesandseemstoprioritizecar- ity(NPPAGandNPPBG,respectively),islimitedbychalleng- bon allocation to photosynthetic organs over that to repro- ing logistics and elevated research costs. Hitherto, most of duction. the on-site measurements of NPP for tropical forests have Another (Quesada et al., 2009a) relates Amazon above- beenbasedonfewsitesanddonotpresentadequatedataon ground productivity to its potential edaphic and climate below-groundNPP(Clarketal.,2001a).Amazonia,hometo drivers. This analysis revealed that forest structure and dy- overhalfoftheworld’stropicalforestarea,isnoexception. namicsarestronglyrelatedtophysicalandchemicaledaphic Most studies that attempted to measure NPP in this ecosys- conditions. On one hand tree turnover rates were mostly temfocusedexclusivelyonabove-groundwoodproductivity influenced by soil physical properties, on the other hand, (e.g. Chambers et al., 2001; Malhi et al., 2004; Quesada et forest growth rates were mainly related to available soil al.,2009a). Malhietal.(2009)compiledasynthesisofcar- phosphorus,suggestingthatsoilsmaybeadeterminantfac- bonproductionforthreeAmazonianforests(keysitesofthe tor on forest functioning and composition at a Basin wide Large Scale Biosphere-Atmosphere Experiment in Amazo- scale. However, beyond the stand-level wood and leaf pro- nia – LBA) based on detailed measurements of the individ- duction pattern across the region, almost nothing is known ual above and below-ground C cycling components. GPP abouttheamountofcarbonbeingallocatedbyothercompo- estimatefromthecomponentstudieswasinagreementwith nentsoftheNPP,thepartitioningoftotalNPP(NPP )be- total estimates derived from ecosystem flux measurements, giv- tweenitsabove-andbelow-groundcomponents,NPP and AG ing increased confidence in both approaches to estimating NPP respectively, and the main environmental drivers of BG tropicalforest’sGPP.Moreover,thisstudyindicatedthatold- anysite-to-sitevariation. Therefore,inthispaperweprovide growthorinfertiletropicalforestsmayhavelowcarbonuse thefirstinter-sitequantificationofthemajorcomponentsof ∼ efficiency(CUE=0.3)incomparisontorecentlydisturbedor NPP forAmazonianforeststandsoncontrastingsoilsus- total ∼ fertile tropical forests (CUE=0.5), highlighting the hetero- ing standardized on-site measurements of the major above- geneityofforestprocessesintheAmazon. andbelow-groundcomponentsofforestNPP. Inordertoimproveourunderstandingofthebiogeochem- The total NPP of a tropical forest stand can be broken ical function of Amazonian forests, and model simulations downas: of their vulnerability to climate change and human-induced NPP =NPP +NPP (1) impacts,thereisaneedtoexpandourknowledgeonthepri- total AG BG mary productivity of these ecosystems, taking into account EachofthecomponentsofNPP canbedescribedasthe total their spatial heterogeneity. The understanding of how these sumofitssubcomponents(Clarketal., 2001a; Malhietal., processesvaryacrosstheregionandacrosssoiltypeswould 2009). Thus,NPP canbeexpressedas: AG thereforeassistinplanningthedevelopmentoftheAmazon NPP =NPP +NPP +NPP +NPP (2) regionwithintheglobalclimatechangemitigationandadap- AG canopy branch stem VOC tationframework. where NPP is the canopy production (leaves, twigs canopy Based on the analysis of 104 forest plots across Amazo- <2cm diameter, flowers and fruits), NPP is the pro- branch niafromtheRAINFORnetwork(AmazonForestInventory ductionofbranches>2cmdiameter,NPP istheproduc- stem Network; Malhi et al., 2002), Malhi et al. (2004) demon- tion of coarse woody biomass, calculated as the change in Biogeosciences,6,2759–2778,2009 www.biogeosciences.net/6/2759/2009/ L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests 2761 thestembiomassoftrees>10cmdiameterplusthebiomass recruited during the measurement interval. NPP is the VOC emissionofvolatileorganiccarboncompounds(seeMalhiet al.,2009,forgreaterdiscussionoftheseterms). Thecanopy production,NPP isestimatedtobeequaltotherateof canopy litterfall; this assumes the forest is in near-steady state and thatthereislittlelossofthisproductionthroughinsecther- bivoryordecompositionbeforethelitterhitstheground. NPP can be divided into three major subcomponents BG (Eq.3): NPP =NPP +NPP +NPP (3) BG fineroot coarseroot exudates Where NPP is the fine root (<2mmdiameter) pro- fineroot duction, NPP is the production of coarse roots coarseroot (>2mmdiameter)andNPP isthecarbonlossthrough exudates exudatesandmycorrhizae, which ischallengingtomeasure andisnotconsideredhere. Fig. 1. Location of the study area within South America (lower- In this study we quantify the above- and below-ground rightpanel)andlocationofthestudiedplotswithinthestudiedarea. NPPoftenAmazonianforeststoaddresstwogeneralques- Plotlocationsareapproximatedinordertodisplayallplots. tions: (1) how do Amazonian forests allocate productivity among its above- and below-ground components? (2) How dosoilandleafnutrientstatusandsoiltextureaffectthepro- 2 Studysites ductivityofAmazonianforests? Based on the concepts above, these two questions can be We analysed the NPPtotal and its above- and below-ground decomposedintofivespecificquestions,whichwetacklein subcomponentsfortenforestplotsacrossAmazonia(Fig.1). thispaper: WedirectlyquantifiedNPPemployingaconsistentmethod- ology in eight plots: three plots at Caxiuana˜, Brazil (CAX- 1. How do NPP and NPP and their subcomponents AG BG 03, CAX-06 and CAX-08), two at Tambopata, Peru (TAM- varywithNPP ? total 05 and TAM-06), two at Amacayacu, Colombia (AGP-01 2. IsthepartitioningbetweenNPP andNPP invariant and AGP-02) and one at Zafire, Colombia (ZAR-01). In AG BG withchangesinNPP ? addition, we used published data compiled from two other total Brazilian sites in Manaus (MAN-05) and Tapajo´s (TAP-04) 3. IsthepartitioningbetweenNPPstemandNPPcanopycon- (Malhietal.,2009)tosupportouranalysis. stant? Thesesitesarepartoftheintensivesurveyedplotswithin the RAINFOR network where measurements of all the ma- 4. HowdoesNPPvarywithsoilandleafnutrientsstatus? jorcomponentsoftheCcyclearebeingmeasuredsincethe 5. How does the partitioning of NPP vary with soil and endof2004orbeginningof2005(seemethodssection),and leafproperties? wherestemproductivityhasbeenmeasuredsinceasearlyas the 1980s. All directly studied plots were one-hectare (ha) WethereforeaimtoinvestigateinthisstudyhowNPP and total inarea;theresultsfromManausandTapajo´swereasynthe- its subcomponents vary across a wide range of Amazonian sisfromseveralstudyplots(Malhietal.,2009). Asummary forests on different soil types. Specifically, our objectives ofplotsname, locationandbasicclimatedata(Malhietal., areto: 2004)isgiveninTable1. 1. Quantify and describe the patterns of NPPtotal across a Allforestssurveyedinthisstudyare“primary”old-growth gradientofsoilconditions. rainforests with the exception of CAX-08, which is a well preserved late successional forest growing on a very fer- 2. Quantify the partitioning of NPP among its major total tileIndianDarkEarth(orTerraPretadoIndio)soil(Hortic aboveandbelow-groundcomponents. Archeo-Anthrosol,Kampfetal.,2003).Thissoilwasformed 3. Investigate if the partitioning between NPPcanopy and byhumanactivitiesfromancientinhabitantsthathaveoccu- NPPstemisconstant. piedthisareabetween720–300yearsBP(RuivoandCunha, 2003; Lehmann et al., 2003). This was selected as one of 4. Determinehowsoilfertilityandtexture,basedonavail- thefewTerraPretasitesintheregioncoveredbyforestthat able soil phosphorus, nitrogen and clay content data has remained largely undisturbed for at least 40 years since (Quesadaetal.,2009a,b)influenceNPPinAmazonian the creation of Caxiuana˜ National Forest reserve. More de- forests. tailsabouteachsitesurveyedaregivenbelow. Foreachsite www.biogeosciences.net/6/2759/2009/ Biogeosciences,6,2759–2778,2009 2762 L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests Table1. Sitecodes,locationsandclimaticcharacteristicsofthetenAmazoniansitesevaluatedinthisstudy. Theclimatedatapresentedin thistablearemeanvaluesfrom1960–1998derivedfromtheUniversityofEastAngliaObservationalClimatology(Newetal.,1999)and publishedinMalhietal.(2004). Cumulativeannualrainfallisgiveninmmyr−1,dryseasonlength(DSL)inmonths,correspondstothe sumofconsecutivemonthswithrainfall<100mmmonth−1,andtemperatureisthemeanannualtemperature(MAT)inCelsiusdegrees. Studysites Climate RAINFORsitescode Name Country Location Rainfall DSL MAT Lat Long mmyr−1 months Celsiusdegrees AGP-01 AguaPudreplotE Colombia −3.72 −70.3 2723 0.0 25.5 AGP-02 AguaPudreplotU Colombia −3.73 −70.4 2723 0.0 25.5 CAX-03 Caxiuana˜droughtexperimentcontrolplot Brazil −1.72 −51.5 2314 4.0 26.9 CAX-06 Caxiuana˜fluxtowersite Brazil −1.72 −51.5 2314 4.0 26.9 CAX-08 Caxiuana˜TerraPretasite Brazil −1.72 −51.5 2314 4.0 26.9 MAN-05 Manaus Brazil −2.5 −60.0 2272 3.0 27.1 TAM-05 TambopataRAINFORplot3 Peru −12.8 −69.7 2417 3.5 25.2 TAM-06 TamboapataRAINFORplot4 Peru −12.9 −69.8 2417 3.5 25.2 TAP-04 Tapajo´sfluxtowersite Brazil −2.5 −55.0 1968 4.5 26.1 ZAR-01 Zafire,Varillal Colombia −4.0 −69.9 2723 0.0 25.5 wealsocompileddataonleafnitrogenandphosphoruscon- forest, with a 25m height canopy, on relatively fertile clay centrations(N andP ,respectively;Fyllasetal.,2009), plinthosols(aquicentisolsinUSDAsoiltaxonomy). leaf leaf soiltypes,followingtheWRBsoiltaxonomytobeconsistent AtZafire,Colombiaourfocuswasona1-haplotlocated withQuesadaetal.(2009b), soiltexture(claycontent), and onawhitesandsite,ZAR-01(Jime´nezetal.,2009). Thisis soilnitrogenandphosphorusconcentrations(N andP , part of the Rio Caldero´n Forest Reserve, around 50km east soil soil respectively; Quesada et al., 2009a). These data are shown oftheAmacayacusite. Theplotwassetupinanareaofpri- inTable2. maryforestonwhitesandpodzol(spodosolintheUSDAsoil At Caxiuana˜, Brazil, we surveyed three 1-ha plots taxonomy), locally known as Varillal. This forest is shorter (100m×100m). All plots are located at the Caxiuana˜ Na- thantheforestatAmacayacu,withanaverageheightof20m. tional Forest in Para´ State. The plot CAX-06, is a tall pri- ThisforesttypeisscarceinwesternAmazoniabutmorefre- mary forest (35m height canopy) situated on a clay ferral- quent in the Guyana Shield and is similar to the formations sol (oxisol in USDA soil taxonomy) near a flux tower site alongtheRioNegro.Thesoilinourplothasanimpermeable (Malhi et al., 2009). The CAX-03 plot is a sandier site lo- hardpanlayerat∼100cmdepth(Jime´nezetal.,2009). cated2kmtofurthersouth,whichwasthecontrolplotfora AtTapajo´s,Brazil,wefocusonthesitesreportedbyMalhi droughtexperiment(Metcalfeetal.,2007a).TheTerraPreta et al. (2009). The main site is the km 67 flux tower (TAP- site (CAX-08) is a late successional forest on an Archaeo- 04, Hutyra et al., 2007; Saleska et al., 2003) and its vicin- Anthrosol (this classification was modified from the WRB ity. ThissiteislocatedwithintheboundariesoftheTapajo´s soiltaxonomybyKampfetal.(2003)toencompassthevari- National Forest in Para´ State. The plots are established in abilityofTerraPretasoilsinAmazonia). TheCAX-08site old-growthforestwithcanopyheightaround35m. Thesoils islocatedabout15kmtothesouthoftheprimarystudyarea, are very clay-rich Belterra clay ferralsols, interspersed with bytheedgeofalargeriverbay. sandiersoilpatches. Thekeyplotsarefour1-hatransectses- At Tambopata, Peru, we surveyed two pre-existing long- tablishedin1999immediatelytotheeastofthetower(Rice term1-haplots(100m×100m)locatedattheTambopataBi- etal.,2004;Pyleetal.,2008). ologicalReserve,inMadredeDiosRegion. TheplotTAM- AtManaus, Brazil, thestudysiteswerealsofocusonthe 05 was set up on relatively infertile Pleistocene cambisols sites reported in Malhi et al. (2009). The main sites are the (inceptsolsinUSDAsoiltaxonomy),withanaveragecanopy K34 flux tower site (Arau´jo et al., 2002), and the various heightof30m. TheplotTAM-06wasonalisols(ultisolsin studies that have been conducted in its vicinity. The plots USDAsoiltaxonomy)onafertileHolocenealluvialterrace. arestablishedinold-growthTerraFirmeforestsonclay-rich ThecanopyinthisplothasthesameaverageheightasTAM- ferrasols,extensivelydissectedbyrivervalleyshostinglower 05,butagreaterdensityofpalms. biomassforestonfrequentlywaterloggedpodzols. Thekey At Amacayacu, Colombia, our focus was on two 1-ha forestplotsintheareaarethethree1-ha“Bionte”plotsonthe terrafirmeforestplotsAGP-01andAGP-02(Jime´nezetal., plateaux, providing annual census data since 1989, and the 2009). Both are located at the Amacayacu National Nat- two5-ha(20×2500m)“Jacaranda”transectplotsthatdrape ural Park, near the border between Colombia, Brazil and acrosstheplateau-valleylandscape. Peru. Theplotsweresetupinanareaofprimaryold-growth Biogeosciences,6,2759–2778,2009 www.biogeosciences.net/6/2759/2009/ L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests 2763 Table2. Leafnutrientconcentration(mgg−1)andsoilavailablephosphorusconcentration(mgkg−1),nitrogenconcentration(%)andclay content(%)forthetenAmazoniansitesevaluatedinthisstudy.LeafdataarederivedfromFyllasetal.(2009)andsoildatafromQuesadaet al.(2009b).NotethatthesoiltypeisinaccordancewiththeWorldReferenceBasesoilclassificationsystem(WRB). Leafnutrient Soil RAINFORcode Nitrogen Phosphorus Type Clay Nitrogen Phosphorus AGP-01 20.87 1.06 EndostagnicPlinthosol(Alumic,Hyperdystric) 42.12 0.16 25.36 AGP-02 19.17 0.96 EndostagnicPlinthosol(Alumic,Hyperdystric) 43.10 0.16 25.43 CAX04 20.63 0.55 VeticAcrisol(Alumic,Hyperdystric) 16.30 0.07b 12.31b CAX-06 19.13 0.53 GericAcricFerralsol(Alumic,Hyperdystric,Clayic) 47.53 0.13 12.31b CAX-08 HorticArchaeo-Anthrosol(Ebonic,Clayic,Mesothropic,Mesic,Ferralic) 41.41 0.17 80.00 MAN-05 19.89a 0.54a GericFerralsol(Alumic,Hyperdystric,Clayic) 66.21a 0.16a 7.28a TAM-05 23.99 1.05 HaplicCambisol(Alumic,Hyperdystric,Clayic) 7.41 0.16 32.34 TAM-06 24.80 1.88 HaplicAlisol(Hyperdystric,Siltic) 9.66 0.17 33.06 TAP-04 22.58 0.75 GericFerralsol(Alimic,Hyperdystric,Clayic,Xanthic) 89.25 0.14 15.45 ZAR-01 OrtseincPodzol(Oxyaquic) 0.64 0.11 14.36 aValuesarefromMAN-05plot. bValuesareaveragesfromnearbyplotsCAX-01andCAX-02. Foramoredetailedviewofthelandscapeattributes,such structures(flowers,fruitandseeds)andunidentifiedmaterial, as vegetation type and structure, topography and plot loca- andweighed. tions,ofmanyoftheseplotsseeAndersonetal.(2009). At Tapajo´s, litterfall values compiled by Malhi et al.(2009)werefrom: (1)Riceetal.(2004),whocalculated fine litterfall from 30 circular mesh traps (0.43m diameter, 3 Materialsandmethods 0.15m2), randomly located throughout the 19.75-ha survey area;(2)Silveretal.(2000),whoestimatedfinelitterfallrates 3.1 AbovegroundNPP in six to ten 4m×12m plots, using six 0.9m2 baskets per plot;and(3)Nepstadetal.(2002),whoused0.5m2 trapsat 3.1.1 Litterfall 100 points arrangedas a regular grid within two 1-ha plots. At all sites litter was collected at biweekly intervals. For At Caxiuana˜ and Tambopata, one litter trap with an area of Manaus, following Malhi et al. (2009), we used the mean 0.25m2 (0.5m×0.5m) was installed in the centre of each valuesfromLuiza˜oetal.(2004). of the twenty-five 20 by 20m subplots in the plots CAX- 06andCAX-08inAugust2004. InJanuary2005thesame 3.1.2 Branchproduction design was installed at Tambopata (TAM-05 and TAM-06). Litterfallinthesefourplotswascollectedeveryfifteendays The production of large branches was not measured in this from September 2004 to December 2006 in Caxiuana˜ and experiment. ForcompletenessofourNPPestimateweused from February 2005 to December 2006 in Tambopata. Lit- a branchfall average rate for all plots of 1MgCha−1yr−1 terfall in the third 1-ha plot in Caxiuana˜ (CAX-03) was based upon data from two studies carried out in Amazonia: recorded monthly from November 2001 to December 2006 one in Manaus that reported a rate of 0.4MgCha−1yr−1 usingtwentycirculartraps(area=1m2)randomlyplacedin (Chambersetal.,2001)andasecondoneintheTapajo´sthat November2001. estimatedabranchfallrateof1.6±0.8MgCha−1yr−1(Nep- At Amacayacu and Zafire one litter trap with an area of stadetal., 2002). Fortheanalysisoferrorpropagation(see 0.50m2 (0.5m×1.0m) was installed in 2005 in the centre below)weusedaconservativeuncertainty±100%(Malhiet of each of the twenty-five 20 by 20m subplots in the plots al., 2009). It is possible that sites with higher NPPstem and AGP-01, AGP-02 and ZAR-01. In all three sites litterfall alsohigherstembreakageinwesternAmazoniawouldhave was collected biweekly for two years at AGP-01 and AGP- higherNPPbranchrates. 02,andfor1.5yearsatZAR-01. 3.1.3 Coarsewoodybiomassproduction For all of these sites sites, traps were made with a PVC frameanda1mmnylonmeshandwereplacedat1mabove Woodproductivity(NPP )wasestimatedbyrepeatedcen- stem the ground surface. Litter retrieved from the traps was im- suses of tree diameters and stems newly recruiting into the mediatelysundriedandsubsequentlydriedinthelaboratory 10cmdiametersize-class,takingintoaccounttaxon-specific oven at 60◦C until constantweight. Each dried samplewas variationinwooddensity. separated into leaves, twigs (<2cm diameter), reproductive www.biogeosciences.net/6/2759/2009/ Biogeosciences,6,2759–2778,2009 2764 L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests At Caxiuana˜, the censuses at CAX-06 and CAX-08 were 3.2 BelowgroundNPP carriedoutfrom2004–2006withanaveragecensusinterval of 0.78±0.11 years (CAX-06) and 1.16±0.23 years (CAX- 3.2.1 Coarserootproductivity 08). AtCAX-03recensusestookplaceannuallyfrom2001– 2006(Metcalfeetal.,2009). Productionofcoarserootsisaverydifficulttermtoquantify. At Tambopata, a much longer time interval was avail- Forthisstudyweassumedthattheproductivityperunitmass able for both plots (TAM-05 and TAM-06). The censuses ofallrootsthatarenotincludedinthefinerootproductivity at TAM-05 were carried out from 1985 to 2005 with an (i.e. roots >2mm diameter) is the same as the productivity average census interval of 2.79±0.50 years, and at TAM- of the above ground biomass and that belowground coarse 06 from 1987 to 2005 with average census interval of root biomass is 21±3% of above-ground biomass (Malhi 3.19±0.87years. et al., 2009). Hence, NPPcoarseroot=0.21 (±0.03)×NPPstem. At Amacayacu and Zafire, plots AGP-01, AGP-02 and This may underestimate NPPcoarseroot as the productivity of ZAR-01censuseswerecarriedoutannuallybetween2004– smaller roots is likely to be greater than the productivity of 2006(Jime´nezetal.,2009). themassivestructuralroots. AtTapajo´sandManausweusedvaluespublishedinMalhi et al. (2009). For Tapajo´s, NPP is an average of two 3.2.2 Finerootproductivity stem surveys at km 67 reported by Pyle et al. (2008) (20ha in total), and for Manaus values are an average of the Bionte, This term was directly quantified in CAX-06, CAX-08, JacarandaandBDFplots(usingtherecentvaluesfromPyle CAX-03,TAM-05,TAM-06,AGP-01,AGP-02andZAR-01 etal.,2008,forthelatter). using the ingrowth cores technique (Vogt et al., 1998; Ste- Forallplotsthesamemethodologywasappliedtoderive ingrobeetal., 2000; Hendricksetal., 2006; Metcalfeetal., biomassandstemgrowth. Initially,foreachplottheabove- 2008). Fine-roots considered here are defined as <2mm in groundbiomass(AGB,kgdryweightha−1)ofalltreeswith diameter. diameter at breast height (DBH) ≥10cm, including palms, At Caxiuana˜ and Tambopata, sixteen root-free ingrowth wascalculatedusingtheequationproposedbyChamberset cores with an area of 154cm2 and depth of 30cm were in- al. (2001) for central Amazonian forests, and modified by stalled in each plot (CAX-06, CAX-08, CAX-03, TAM-05, Baker et al. (2004) to allow for variation in wood density TAM-06)ina20mby20mregulargrid,20mfromtheedge (Eq.4). of the plot. Ingrowth cores were installed at the beginning ofNovember2004intheCaxiuana˜plots(CAX-06,CAX-08, AGB=Xn σi (4) CAX-03)andatthebeginningofJune2005intheTambopata 0.67 plots (TAM-05 and TAM-06). At each sampling point, one 1 ( ) soil core were extracted, keeping the soil layers separated h i × exp 0.33(lnD )+0.933(lnD )2−0.122(lnD )3−0.37 to avoid the transference of soil with high nutrient content i i i fromthetoptothebottomlayers. Rootsweremanuallyre- moved and the remaining soil was reinserted into the hole where D and σ are, respectively, the diameter (cm) and i i wood density (gcm−3) of tree i, n is the number of stems surroundedbya1cmplasticmeshbag. Thisprocedurewas repeatedeverythreemonthsinallplotsfromNovember2004 per plot. To be consistent with other studies we considered toNovember2005inCaxiuana˜ andfromJune2005toApril thecarbonfractioninthedrybiomasstobe0.5. 2007 in Tambopata. At each three-month interval ingrowth FollowingMalhietal.(2004)weestimatedthewoodypro- cores were removed from the soil and all fine roots manu- ductivity for each plot at each census interval by subtract- ally extracted following the method described by Metcalfe ing the AGB of the surviving trees at second census (t ) by 1 et al. (2007c) which corrects for underestimates in, partic- the first census (t ) and added to this total the AGB that 0 ularly fine, root mass. Afterwards, the root free soil cores of new recruits. However, variation in census intervals af- were reinserted into the same holes. Roots were then dried fects estimates of wood productivity, as trees that grew but at 60◦C until constant weight, subsequently cleaned for re- diedbeforethesecondcensusaremissed. Specifically,esti- moval of soil particles and weighed to determine dry root matesofwoodproductivitydecreaseapproximatelylinearly massproducedduringeachthree-monthinterval. withincreasingcensusintervallength.Wethereforeusedthe AtAmacayacu,thirteeningrowthcoresandatZafirefour- plot-specificcensus-intervalcorrectionproposedbyMalhiet teen ingrowth cores with an area of 23.6cm2 and depth of al. (2004), adjusted for AGB values instead of basal area 20cmwereinstalledintheplotsAGP-01,AGP-02andZAR- (Phillips et al., 2009). The correction varies approximately 01 (Jime´nez et al., 2009). The ingrowth cores at AGP-01 asthesquareofthegrowthrate. andAGP-02wereinstalledatthreedifferentdates: February 2004 (Experiment 1), using cores surrounded by a 4mm2 mesh bag, and September 2004 (Experiment 2) and Febru- ary 2006 (Experiment 3), using cores with no mesh. At Biogeosciences,6,2759–2778,2009 www.biogeosciences.net/6/2759/2009/ L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests 2765 plotZAR-01installationtookplaceinSeptember2004(Ex- Soilsampleswereair-driedandhadroots, detritus, small periment 2) and February 2006 (Experiment 3) using no rocksandparticlesover2mmremoved. Sampleswerethen meshcores. Thecollection,extractionandprocessingofthe analysed for: (1) complete phosphorus fractionation (mod- roots were fairly similar to the one presented for Caxiuana˜. ified from Hedley et al. 1982), (2) nitrogen (Pella, 1990; Detailed description of the methods is given in Jime´nez et Nelson and Sommers, 1996), and (3) particle size analysis al. (2009), so here we give a summary of the methods. Ex- usingtheBoyoucosmethod(GeeandBauder,1986). periment1wascarriedoutfromFebruary2004toJuly2006, In this study we used the readily available P, which is Experiment 2 from September 2004 to July 2006 and Ex- defined as sum of the P immediately accessible in solution periment3fromFebruary2006toDecember2006. Thefirst (measured as the resin P), plus that organic and inorganic collectionwascarriedout5–7monthsafterinstallationofthe phosphorusthatcanbeextractedbybicarbonate. Itappears ingrowthcoresforthethreeexperiments.Subsequentcollec- a good measure of the phosphorus that plant roots are able tionswerecarriedout2–4monthsafterthere-installationof toextractatlittlecost,andhenceagoodmeasureofplantP theroot-freeingrowthcores. Forthepresentstudyweused availability(Quesadaetal.,2009a). an average value of all experiments to better represent inter The leaf phosphorus and nitrogen concentrations used in andintra-annualvariabilityoffine-rootproduction. thisstudywereobtainedfromFyllasetal.(2009). Leafsam- AtTapajo´s,weusedthevaluesoffinerootproductionre- plingandanalysisprotocolsaredescribedindetailinLloyd portedinMalhietal.(2009). Finerootproductionatthissite etal.(2009). Inbrief, 12to40treesper1haplotwereran- was calculated by Silver et al. (2000) using the sequential domlyselectedandleafsamplesforeachtreewerecollected coring method at TAP-04, sampling 0–10cm depth with a fromfullsuncanopypositionswiththehelpofatreeclimber. 6cmdiametercorereverytwomonthsfortwoyears. Subse- Leaveswerecleanedanddried,milledandstoredforlabora- quently,rootproductionwasestimatedusingacompartment toryanalysis. flowmodel (Sanantonio andGrace, 1987), with decayrates Samples from Peruvian and Colombian sites were anal- calculated directly from a trenching experiment. The study ysed at the Central Analytical and Stable Isotopes Facili- wasconductedinsixplotsapproximately4m×12minsize. ties at the Max-Planck Institute for Biogeochemistry (MPI- All data from Caxiuana˜, Tambopata, Amacayacu, Zafire BGC)inJena,Germany. SamplesfromBraziliansiteswere and Tapajo´s were adjusted following Malhi et al. (2009) to analysed for phosphorus at Instituto Nacional de Pesquisas accountforrootproductiontoone-meterdepth. Noexisting daAmazoˆnia(INPA)andfornitrogenatEmpresaBrasileira dataonfine-rootproductionwerefoundforManaus. There- de Pesquisas Agropecua´rias (EMBRAPA), both in Manaus, fore,anaveragevalueofplotsonsimilarsoiltypes(acrisols Brazil. Leafnitrogenconcentrationsweredeterminedinthe andferrasols)atCAX-06,CAX-03,TAP-04wasusedtoes- BrazilianandGermanlaboratoriesusingfinelygroundplant timatefine-rootproductioninthissite. material in an elemental analyser (Elementar Analysensys- In addition, we compiled fine root stand biomass data, teme, Hanau, Germany). At INPA, phosphorus concentra- adjusted to one meter depth, in order to estimate fine root tionsweredeterminedafteraciddigestionoftheleafmaterial turnover (stand biomass/productivity) as a complementary inaUVvisiblespectrophotometer(Model1240,Shimadzu, measure of fine root dynamics across the fertility gradient. Kyoto,Japan). AttheMPI-BGCsampleextractswereanal- For Caxiuana˜, values of fine root stand biomass for all the ysedbyanICP-OES(ModelOptima3300DV,PerkinElmer, three plots are reported in Metcalfe et al. (2008). For the Norwalk, CT, USA). For all chemical analyses the equip- twoplotsatAmacayacuandtheoneatZafire,valuesoffine mentswereinter-calibrateforconsistencyoftheresults. root biomass are from Jime´nez et al. (2009). For Tapajo´s weusedthevaluesoffinerootturnoverreportedbySilveret 3.4 Propagationofuncertaintiesanddataanalyses al. (2005). Fine root stand biomass for Tambopata was di- rectly estimated in this study (TAM-05=7.7MgCha−1 and Alluncertaintyestimatesthroughoutthetextarereportedas TAM-06=5.0MgCha−1). standarderrorsassumingnormaldistributions. Uncertainties arepropagatedbyquadratureofabsoluteerrors(Malhietal., 3.3 Soilandleafdata 2009). Therefore, if y is the sum of a number of variables x ...x ...x ,eachwithassociateduncertainty1x ,thenthe 1 i n i The soil available phosphorus and nitrogen concentrations absoluteuncertaintiesarepropagatedinquadrature(Eq.5): (0–30cm depth) and texture used in this study were ob- n X tained from Quesada et al. (2009c), and methods are only (1y)2= (1x )2 (5) i briefly summarized here. For each 1-ha plot, five to twelve i=1 soil cores were collected using an undisturbed soil sam- Thisassumesthatuncertaintiescanbeconsideredtobeinde- pler (Eijkelkamp Agrisearch Equipment BV, Giesbeek, The pendentandnormallydistributed. Itisimportanttonotethat Netherlands). All sampling was done following a standard uncertainty values represent uniquely the precision of mea- protocol (see http://www.geog.leeds.ac.uk/projects/rainfor/ surements and that biases related to methodological issues projdocs.html). were not accounted for in the error analysis due to the lack www.biogeosciences.net/6/2759/2009/ Biogeosciences,6,2759–2778,2009 2766 L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests Table3. NetprimaryproductivityofeachsubcomponentofthetotalNPP(NPPtotal)foreachoneofthestudiedsites. TheNPPvalues aregiveninMgCha−1yr−1 withitsassociatedstandarderror(S.E.). NotethatforManausanaveragevalueofplotsonsimilarsoiltypes (acrisolsandferrasols)atCAX-06,CAX-03,TAP-04wasusedtoestimatefine-rootproductioninthissite. Site CAX-06 CAX-03 CAX-08 TAP-04 MAN-05 TAM-05 TAM-06 AGP-01 AGP-02 ZAR-01 NPP S.E. NPP S.E. NPP S.E. NPP S.E. NPP S.E. NPP S.E. NPP S.E. NPP S.E. NPP S.E. NPP S.E. NPPcanopy 3.8 0.10 3.5 0.10 5.4 0.20 6.5 0.70 3.6 0.70 5.6 0.30 4.6 0.24 3.9 0.20 3.7 0.20 2.7 0.10 NPPbranch 1.0 1.00 1.0 1.00 1.0 1.00 1.0 1.00 1.0 1.00 1.0 1.00 1.0 1.00 1.0 1.00 1.0 1.00 1.0 1.00 NPPstem 1.7 0.21 2.6 0.20 2.5 0.26 3.8 0.07 2.6 0.06 2.8 0.24 2.6 0.42 3.4 0.30 3.8 0.30 1.3 0.30 NPPcoarseroot 0.4 0.04 0.4 0.03 0.5 0.05 1.0 0.30 0.8 0.20 0.6 0.10 0.6 0.09 0.7 0.06 0.8 0.06 0.3 0.06 NPPfineroot 3.9 0.40 4.0 0.90 7.6 0.93 2.0 0.30 3.3 0.34 6.8 1.00 4.8 0.57 2.2 0.40 2.2 0.40 3.9 0.68 NPPvoc 0.1 0.13 0.1 0.13 0.1 0.13 0.1 0.13 0.1 0.13 0.1 0.13 0.1 0.10 0.1 0.10 0.1 0.10 0.1 0.10 NPPAG 6.7 1.04 7.2 1.03 9.0 1.06 11.4 1.20 7.3 1.23 9.5 1.08 8.4 1.11 8.4 1.07 8.6 1.07 5.1 1.05 NPPBG 4.2 0.40 4.4 0.90 8.1 0.93 3.0 0.40 4.1 0.40 7.4 1.00 5.4 0.58 3.0 0.41 3.0 0.40 4.2 0.68 NPPtotal 10.9 1.11 11.6 1.37 17.0 1.41 14.4 1.30 11.4 1.29 16.9 1.47 13.8 1.26 11.3 1.14 11.7 1.14 9.3 1.26 of available information. Probable sources of unaccounted variedbetween3.0±0.4MgCha−1yr−1atTAP-04,AGP-01 errors are discussed throughout the text and their potential andAGP-02and8.1±0.9MgCha−1yr−1atCAX-08. magnitudeisquantifiedinClarketal.(2001a). To investigate if NPP in our analysis presents the same All the results of the estimated fluxes reported in this trend as proposed by Malhi et al. (2004), where plots on study are in MgCha−1yr−1. 1MgCha−1yr−1 is equal to morefertilesoilsinwesternAmazoniahavegreaterNPP , stem 100gCm−2yr−1,or0.264µmolCm−2s−1. we ordered the data presented in Table 3 by increasing soil The partitioning of NPP into its subcomponents, the availablephosphorus(Fig.2). Asexpected,themorefertile total relationship between NPP and NPP and the effect sites sit to the west, with exception of the “artificially” fer- canopy stem of soil and leaf nutrient status and soil texture on patterns tileTerraPretasite,CAX-08intheeast(themostfertileof ofabove-andbelow-groundNPPacrossthetenAmazonian all our sites), and the anomalous white sand ZAR-01 in the sites was assessed using linear regression analysis. In ad- west. ThetrendofincreasedNPPstem towardsthemorefer- dition, we ran a set of correlation analyses, including the tilesites,assuggestedbyMalhietal.(2004)isnotobviousin Pearson product-moment, Spearman rank and Kendall tau, oursmallerdataset. However,thereisatendencyofgreater between NPPtotal, NPPAG, and NPPBG and climate and en- NPPAG, NPPBG and NPPtotal towards the more fertile sites, vironmental variables. We opted for presenting the results whichwillbefurtherexploredinthefollowingsections. of both parametric and non-parametric tests because of our NPP varied between 2.7±1.1MgCha−1yr−1 at the canopy small sample size (n between 7 and 9). A z-test was ap- white sand plot (ZAR-01) and 6.5±0.7MgCha−1yr−1 plied, when necessary, to compare mean values and regres- at Tapajo´s (TAP-04) (Fig. 3). NPP varied between stem sion slopes. For the regression analyses, plots AGP-01 and 1.3±0.3MgCha−1yr−1 at the white sand plot (ZAR-01) AGP-02 were merged into a single plot average to avoid and 3.8±0.1MgCha−1yr−1 at Tapajo´s (TAP-04), while spatial autocorrelation issues. Moreover, for the analyses NPP variedbetween2.0±0.3MgCha−1yr−1 atTAP- fineroot wherebelow-groundproductivityisexplicitlytakenintoac- 04and7.6±0.9MgCha−1yr−1attheanthrosolCAX-08. count,plotMAN-05wasremovedasplotlevelestimatesof WiththenotableexceptionoftheTAP-04site,NPP canopy NPPfinerootisinexistent. was greater in the more fertile sites. NPPstem, on the other hand,ismaximalintheplotsTAP-04,AGP-01andAGP-02, which are in the middle of our soil fertility gradient. Con- 4 Results versely, these same plots have the lowest NPPfineroot among all plots. Overall, sites on the most fertile soils tended to 4.1 Patternsofnetprimaryproductivityacrossthesites have higher NPPfineroot, with values being similarly high in theveryfertileanthrosolCAX-08andinthemoderatelyfer- The estimated NPP for the ten plots analysed ranged tilecambisolsandalisolsatTambopata(TAM-05andTAM- total between 9.3±1.3MgCha−1yr−1 at the white sand 06)(Fig.3). plot, ZAR-01, and 17.0±1.4MgCha−1yr−1 at the The TAP-04 site has distinct patterns of extremely high Terra Preta site, CAX-08 (Table 3). Similarly, the NPP and low NPP in comparison to other plots with AG BG above-ground NPP (NPP ) has its lowest value at similarsoilandclimatecharacteristics. TheforestatTapajo´s AG ZAR-01 (5.1±1.1MgCha−1yr−1), however, the high- isallocatingadisproportionalamountofCtoabove-ground est NPP was estimated for the Tapajo´s site, TAP-04 biomass gain, which may be a consequence of large-scale AG (11.4±1.2MgCha−1yr−1). Below-ground NPP (NPP ) recentnaturaldisturbanceorchronicon-goingmortality(see BG Biogeosciences,6,2759–2778,2009 www.biogeosciences.net/6/2759/2009/ L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests 2767 Fig.2. Thenetprimaryproductivityofthetenstudiedsites. Bars correspondtotheNPPtotalofeachsite.Eachbargivestheabsolute contributionofeachsubcomponenttotheNPPtotal.Negativevalues wereusedtorepresentthebelow-groundproductivity,andtoobtain thecorrectpositivevaluesbelow-groundcomponentsmustbemul- tipliedby−1. Sitesareorderedinanincreasedorderofsoilavail- ablephosphorus(Table2)andallvaluesareinMgCha−1yr−1. discussion section). For subsequent analyses we therefore presentsomeoftheresultsbothwithandwithouttheTapajo´s site. 4.2 PartitioningofNPP intoitssubcomponents total Theregressionanalysisshowedthatthereisasignificantin- Fig.3. Thenetprimaryproductivityofthemajorsubcomponents crease of both NPPAG (R2=0.58, p=0.02, n=9) and NPPBG of the NPPtotal in the ten studied sites: (a) fine litter production; (R2=0.55, p=0.03, n=8) with increasing NPP (Fig. 4a). (b) stem production; and (c) fine root production. All values are total Thisrelationshipbecomesmuchmoresignificantwithoutthe in MgCha−1yr−1with their respective standard errors. Sites are plot TAP-04 forboth NPP (R2=0.76, p=0.005, n=8) and rankedbyincreasingsoilavailablephosphorus(Table2). AG NPP (R2=0.83, p=0.004, n=7) (Fig. 4a). This indicates BG that the partitioning pattern may differ between old-growth systemsin“quasi-equilibrium”andrecentlydisturbedforests However, Fig. 4b shows that the proportion of NPPtotal al- (Malhietal.,2009). located below-ground is inversely correlated with the total There is no significant difference (z-test) in the slopes of amount of C allocated to stem growth (R2=0.58, p=0.03, therelationofNPP andNPP withNPP whetheror n=8). However,thisresultcouldbesomewhatbiasedbythe AG BG total not TAP-04 is included. This is indicative of fairly invari- factthatourestimationofNPPcoarseroot accountedinthees- antallocationbetweenNPPAG andNPPBG acrossthebroad timationofNPPBG iscalculatedasaproportionofNPPstem. NPP gradient. Evenusingonlytheproportionallocatedtofinerootinstead total By dividing the NPP of each component by NPP we ofthetotalbelow-groundallocation,whichisanindependent total quantified the relative proportion of NPP allocated to measurement, the relationship is strong (R2=0.61, p=0.02, total each component. Our results show that the ten Amazo- n=8). nian forests studied allocate between 53% (CAX-08) and NPP is very strongly correlated with NPP total canopy 79% (TAP-04) of the NPPtotal into NPPAG and between (R2=0.73, p=0.003, n=9), especially when TAP-04 is ex- 21% (TAP-04) and 37% (CAX-08) into NPPBG. We found cluded (R2=0.98, p<0.001, n=8). As NPPcanopy is a ma- that NPPcanopy, NPPstem and NPPfineroot are on average jor component of NPPtotal the result of the previous anal- 33.5±1.5%,21.3±2.2%,31.4±3.5%oftheNPPtotal,respec- ysis may be redundant. To check the validity of this test tively. we regressed NPP against the result of the difference canopy TherelativeproportionofNPPallocatedtobelow-ground between NPP and NPP . The correlation between total canopy (NPP /NPP ) was not significantly related to NPP these two completely independent variables was significant BG total total assuggestedabovebytheanalysisoftheregressionslopes. when excluding TAP-04 (R2=0.38, n=9, p=0.08; R2=0.90, www.biogeosciences.net/6/2759/2009/ Biogeosciences,6,2759–2778,2009 2768 L.E.O.C.Araga˜oetal.: NetprimaryproductivityinAmazonianforests with and without the Tapajo´s site respectively) (Fig. 5a, c). Interestingly, stem production did not vary much along the NPP gradient (Fig. 5b). NPP hence seems to be a canopy strong predictor of NPP in old-growth, low disturbance total Amazonianforestswiththeinvertedlinearregressionbeing NPP =2.81(±0.25)×NPP −1.22(±1.06). total canopy In summary, the sites with higher NPP tend to have total higher NPP (mainly higher NPP ) and also higher AG canopy NPP (mainly NPP ). The fractional allocation of BG fineroot NPP tobelow-groundisinvariantacrosstheNPP gra- total total dientbutvariesinverselywithNPP ,decreasingfrom47% stem whenNPP =2.5MgCha−1yr−1 (CAX-08)to21%when stem NPP =3.8MgCha−1yr−1(TAP-04). stem 4.3 RelationshipbetweenNPP andNPP canopy stem Malhi et al. (2004) reported a strong proportionality be- tween NPP and NPP . They combined data on lit- canopy stem terfall and stem production from sites investigated in their study and published by Clark et al. (2001a) and found that: NPP =1.73 (±0.09)×NPP . Despite the lin- stem canopy ear trend, we did not find a significant relationship using only our dataset (NPP =1.63 (±0.14)×NPP , n=9, stem canopy p=0.2). The weak relationship observed in our data alone is mainly caused by Colombian site (AGP) with high stem production and relatively low litterfall production. Without this plot, the relationship is similar to the one presented by Malhi et al. (2004), with no significant difference between theslopes(z-test)(NPP =1.78(±0.11)×NPP , n=8, canopy stem R2=0.64, p<0.02). A significant relationship is also re- Fig. 4. (a) The relationship between the above-ground NPP tained (NPP =1.67 (±0.07)×NPP , n=29, R2=0.67, canopy stem and the total NPP (dark grey squares), with the regression lines p<0.001) when we merge all data together, including this for all data (black) and without the Tapajo´s site (black dashed). studyandMalhietal.(2004)(Fig.6). The linear regression equations are (all data – NPPAG=0.51 (±0.16)×NPPtotal+1.55 (± 2.17), n=9, R2=0.58, p=0.02; with- 4.4 Effectofsoilandleafphosphorusandnitrogenand outTapajo´s–NPPAG=0.43(±0.10)×NPPtotal+2.15(±1.32),n=8, R2=0.76, p=0.005) and the relationship between below-ground soiltextureonpatternsofabove-andbelow-ground NPP and the total NPP (light grey triangles), with the regression NPPinAmazonia linesforallsite(grey)andwithouttheTapajo´ssite(greydashed). The linear regression equations are (all data – NPPAG=0.49 Theobservedresponsesinourstudyareconsistentwiththe (±0.18)×NPPtotal−1.55(±2.43),n=8,R2=0.55,p=0.03;without predictionsofthehypothesisofsoilphosphoruslimitationof Tapajo´s – NPPBG=0.55 (±0.11)×NPPtotal−2.05 (± 1.49), n=7, NPPinAmazoniantropicalforests(seealsoQuesadaetal., R2=0.83, p=0.004). (b) The relationship between the propor- 2009a). We demonstrated that NPP is significantly cor- total tion of NPPtotal allocated below-ground (NPPBG/NPPtotal) and relatedtosoilavailablephosphorus(R2=0.54,n=9,p=0.03) the NPPAG (dark grey squares), NPPcanopy (black triangles) and (Fig. 7a). The correlation between these two variables was NPPstem (light grey diamonds). The regression line in black also significant according to the Spearman rank correlation indicates a significant relationship between (NPPBG/NPPtotal) (R=0.73, n=9, p=0.03)andtheKendalltau(tau=0.54, n=9, and NPPstem ((NPPBG/NPPtotal)=−0.08 (±0.03)×NPPstem+0.59 (±0.08),n=8,R2=0.58,p=0.03). p=0.05)(Table4). Ontheotherhand,wedidnotfindaclear relationship between NPP and soil nitrogen (R2=0.29, total p=0.13)(Fig.7b). Wealsodidnotfindsignificantrelation- ships between NPP and any other soil and climate vari- total n=8, p<0.001, with and without considering the Tapajo´s able,excepttotalextractableP(seeQuesadaetal.,2009c,for siterespectively),reinforcingthesignificanceofthecorrela- method) according to both Spearman (p=0.01) and Kendall tionbetweenNPP andNPP . Moreover, NPP is tau (p=0.02) (Table 4). These results indicate that soil total canopy total alsosignificantlycorrelatedwithNPP whenexcluding phosphorus is likely to be more important than nitrogen in fineroot thesiteTAP-04(R2=0.44,p=0.07andR2=0.77,p<0.009, thedeterminationofNPP acrossAmazonia. Theslopeof total Biogeosciences,6,2759–2778,2009 www.biogeosciences.net/6/2759/2009/
Description: