Journal of Ecology. 2018;106:157–167. wileyonlinelibrary.com/journal/jec  | 157© 2017 The Authors. Journal of Ecology © 2017 British Ecological Society

Received:2April2017  |  Accepted:31May2017DOI:10.1111/1365-2745.12823

R E S E A R C H A R T I C L E

Functional groups, species and light interact with nutrient limitation during tropical rainforest sapling bottleneck

Cleo B. Chou  | Lars O. Hedin | Stephen W. Pacala

DepartmentofEcologyandEvolutionaryBiology,PrincetonUniversity,Princeton,NJ,USA

CorrespondenceCleo B. ChouEmail:cleochou@princeton.edu

Present addressCleoB.Chou,PrincetonEnvironmentalInstitute,PrincetonUniversity,Princeton,NJ,USA

Funding informationNationalScienceFoundation,Grant/AwardNumber:GraduateResearchFellowshipProgram;PrincetonUniversity,Grant/AwardNumber:CarbonMitigationInitiative

HandlingEditor:NataliaNorden

Abstract1. Potentialvariabilityinnutrientlimitationamongtreesizeclasses,functionalgroupsandspeciescallsforanintegratedcommunity-andecosystem-levelperspectiveoflowlandtropicalrainforestnutrientlimitation.Inparticular,canopytreesdetermineecosystem nutrient conditions, but competitive success for nutrients and lightduringthesaplingbottleneckdeterminescanopycomposition.

2. Weconductedaninsitumulti-nutrientsaplingfertilizationexperimentatLaSelvaBiologicalStation,CostaRica,todeterminehowfunctionalgroupidentity,speciesidentityandlightavailabilitycanimpactnutrientlimitationofstemgrowthinthreefunctionalgroupsandninespecies.

3. Despitehighsoilfertility,wefoundnutrient-lightlimitationintwofunctionalgroupsandfourspecies.Unexpectedly,thenitrogen-fixing(“N2fixers”)andshade-tolerantfunctional groups were significantly nutrient limited, while the light-demandingfunctionalgroupwasnot.

4. Thiswaspartiallyexplainedbyspecies-levelvariationinnutrientlimitationwithinthesefunctionalgroups,withonlysomespeciesconformingtothepredictionofstrongernutrientlimitationinlightdemanderscomparedtoshade-tolerants.

5. Mostsurprisingly,wefoundstrongnutrientlimitationatlow-lightlevelsintheN2 fixers (whichwereshade-tolerant),butnot in theshade-tolerantnon-fixers.Wehypothesize that theN2 fixerswere actually nitrogen limited at low-light levelsbecauseoftheirnitrogen-richleavesandthehighcarboncostoftheirsymbionts.

6. Thisfindingsuggestsahighlyshade-tolerant,N2fixationstrategy,inadditiontotheperceptionthatN2fixationismostlyadvantageousinhigh-lightenvironmentsdur-ingearlyandgapsuccession.Theshade-tolerant,N2fixationstrategymaybepartofasaplingandcanopytreefeedback,wherethecanopyN2fixersenrichthesoilN,enhancinggrowthof theirshade-tolerantsaplings relative tonon-fixingcom-petitors,enablingfurthercanopydominationbyshade-tolerantN2fixers,asseenatLaSelva.

7. Synthesis.Thepervasivenessoffunctionalgroup-andspecies-specificnutrientandlight co-limitation in our saplings indicates that these interactions likely play animportant role in thedynamicsof lowland tropical rainforestnutrient limitation,potentially via other such sapling and canopy tree feedbacks as the onehypothesized.

PaperpreviouslypublishedasStandardPaper

158  |    Journal of Ecology CHOU et al.

1  | INTRODUCTION

Understanding tropical tree growth limitation by nutrients such asnitrogen (N), phosphorus (P) and potassium (K) is fundamental forpredicting the dynamic response of lowland tropical rainforests tofutureclimaticconditionsandtheirpersistenceaslargecarbonsinks(Huntingfordetal.,2013;Körner,2009;Santiago,2015).Asmallnum-berofinsitufertilizationexperimentshavefoundmixedevidenceoftheextenttowhichnutrientslimittreegrowthintheseforestsanddif-ferencesintheidentityofthelimitingnutrient(s)(Alvarez-Clare,Mack,&Brooks,2013;Fisheretal.,2013;Mirmanto,Proctor,Green,Nagy,&Suriantata,1999;Newberyetal.,2002;Wrightetal.,2011).Theseinconsistenciesmaybeduetothehypothesized“heterogeneousnu-trientlimitation”(sensuAlvarez-Clareetal.,2013)inlowlandtropicalrainforests,wherevariabilityinnutrientresponsesdependsondiffer-encesamongtreetaxaandsizeclasses,butthesedifferences,espe-ciallyamongtaxa,haveyettobecomprehensivelytested.

Althoughheterogeneityofnutrient limitationmightbeexpectedgiventhehighdiversityoflowlandtropicalrainforests,mostofthesepreviousinsitustudiesevaluatedgrowthresponsesattheecosystemscale.Thepotentialthattreepropertiessuchassizeclass,taxonomicidentityorfunctionalgroupidentitymaycomplicateforestresponsetonutrientsindicatetheneedtoexaminelimitationalsoatthecom-munity, population and individual scales.At these scales, there is acentral nutrient-light feedback between saplings and canopy trees,wheresuccessincompetingfornutrientsandlightatthesaplingstagedetermineswhichindividualssurvivethebottleneckpassageintothecanopy, and in turn these canopy trees determine ecosystem-levelnutrientcyclingandunderstoreylightavailability, influencingsaplingsuccess(Figure1).

Therefore, to understand the dynamics of lowland tropicalrainforestnutrientlimitation,itisessentialtoexaminehownutrientsand light interact todeterminethesuccessof individualsaplingsastheyexperiencethebottlenecktransitiontothecanopy,with>90%ofsaplingmortalityeventsoccurringbeforetheyreach4cmindiam-eter(Clark&Clark,1992).Thistransitionistypicallyassociatedwithtreefallgaps,whichprovidetheelevatedlightlevelsthatamajorityofspeciesneedatsomepointduringtheirontogenyinordertoreachthecanopy(Brokaw,1985;Denslow,1980,1987).Duetotheasymmetryof lightavailabilityfromthetopofthecanopytotheshadedforestfloor,comparedtolargertreesinthecanopyandsub-canopy,saplingsintheunderstoreyexperienceafullrangeoflightavailabilities,fromdesirable gap environments to undesirable non-gap environments(Wrightetal.,2010;Yoda,1974).

Thisuncertainavailability,butnecessity,ofgapsforindividualsuc-cessduringthesaplingbottleneckhasselectedforrapidsaplinggrowth

rates under favourable high-light conditions (Clark & Clark, 1992;Denslow, 1987). Rapid growth and biomass accumulation increasesplantnutrientdemand(Montagnini,2000),raisingfundamentalques-tions about the interaction between nutrient and light limitation atthesaplingstage.Previousstudiesofunderstoreynutrientlimitationand light interactions in lowlandtropical rainforestsfocusedontreeseedlingsorshrubcuttingsinshadehousesandcommongardens,andfound either no response to nutrients (Denslow, Schultz, Vitousek,& Strain, 1990) or potentially species-specific responses (Fetcheretal.,1996;Palow&Oberbauer,2009).Morerecently,insitustudiesshowedlightbutnotnutrientlimitationofunderstoreysaplinggrowth(Magalhães,Marenco,&Camargo,2014),nutrient limitationof low-light understorey tree seedling growth (Pasquini & Santiago, 2012;Santiagoetal.,2012),andapproximatelyequalcontributionsbylight

K E Y W O R D S

co-limitation,CostaRica,fertilization,gapsuccession,LaSelvaBiologicalStation,lightlimitation,lowlandtropicalrainforest,nitrogenfixation,plant–soil(below-ground)interactions,tropicaltrees

F IGURE  1 Saplingandcanopytreefeedbacksarecentraltolowlandtropicalrainforestnutrientdynamics.Canopytreesdominatefeedbackstoecosystem-levelnutrientcyclingbyprovidingalargeproportionofecosystemfoliar,woodandrootlitterinputswithfunctionalgroup-orspecies-specificnutrientconcentrations(1).Thisinturncanresultinfunctionalgroup-orspecies-specificimpactsondecompositionratesandtotalfluxesofnutrientinputsfromlitterpoolstothesoil(2).Additionally,thesecanopytreesmayalsoimpactunderstoreylightavailabilityinfunctionalgroup-orspecies-specificwaysbasedontheircrownstructure,andthislightavailabilityinteractswithunderstoreynutrientdynamicsaswell(3).However,nutrient-lightlimitationofsaplingsduringthebottlenecktoreachthecanopy(dottedbox)determineswhichindividualsbecomecanopytrees(4).Saplingresponsetosoilnutrientsandlight,andcorrespondingcompetitivesuccessduringthisbottleneck,mayalsobefunctionalgroup-orspecies-specific

     |  159Journal of EcologyCHOU et al.

and nutrients to understoreywoody plant seedling growth (Holste,Kobe,&Vriesendorp, 2011).These studies suggest that saplings ofat least some functional groups or speciesmay be nutrient limitedeveninlow-lightunderstoreyconditions,althoughthestrengthofthislimitation likely increasesasgreater lightavailabilityelevatessaplinggrowthratesandnutrientdemand.

Furthermore,thesestudiesindicatethatnutrientandlightlimita-tionofsaplinggrowthmaydifferacrossthewidearrayoftreestrate-giesforresourceacquisition(Reich,Walters,&Ellsworth,1997),whichcanbeobservedatthelevelofspecies,oratacoarserscale,functionalgroupsofspeciesthatrespondtoenvironmentalvariablessimilarly.Awell known, but complicated gradient of resource acquisition strat-egies is tied to shade tolerance (Clark & Clark, 1992; Pacala etal.,1996),withamajortrade-offbetweengrowth inhigh lightandsur-vivalinlowlight(Wrightetal.,2010).Speciesinthelight-demandingfunctionalgrouparelessshade-tolerantandtendtohavetraitsthatallowforquickgrowthbutlowernutrientuseefficiency(NUE),suchasshortleaflifespan,lowleafmassperarea,highleafnutrientcon-centrationandlowwooddensity(Poorter&Bongers,2006;Swaine&Whitmore,1988).

Incontrast,shade-tolerantspeciestendtohavetraitsattheop-positeendof the spectrum that result in slowergrowthandhigherNUE,withleavesthatarewelldefendedagainstherbivoryandenvi-ronmentalstress.Thus,althoughsaplingsinboththelight-demandingandshade-tolerant functionalgroupsmaybenutrient limited in thelow-lightunderstorey (and toan increasingdegreewithhigher lightavailability), the strength of this limitation is likely greater in light-demandingsaplingsacrossalllightlevelsduetotheirlowerNUE.

A tree resourceacquisition strategywithadirect impactonnu-trientcycling issymbioticN2 fixation,which in the tropics is largelycarriedoutby species (hereafter “N2 fixers”) in theFabaceae familythatcanhostN2-fixingrhizobialbacteriainrootnodulestoaccessat-mosphericN2.TheabilitytofixN2givesN2fixersacompetitiveadvan-tageinenvironmentswhereNdemandishighrelativetosupply,suchasduringsecondaryorgapsuccession(Battermanetal.,2013;Menge&Chazdon,2016).AsN2fixersarenotdirectlyconstrainedbysoilN,theyarelikelylimitedbyothernutrients,particularlyPbecausetheN2 fixationprocessraisesdemandforP(Vitousek&Howarth,1991),andalsomolybdenum(Mo)whichisaco-factorinthenitrogenaseenzyme(Barronetal.,2009).Beyondtheirsymbioticrelationshipitself,otheraspectsofN2fixerphysiologythatmayberelatedtotheirN2fixationlifestyle are distinctive compared to that of non-fixing trees.ThesecharacteristicsincludehighleafNconcentrationsandthushighNre-quirements,aswellasgreaterwateruseefficiency(Adams,Turnbull,Sprent,&Buchmann,2016;McKey,1994).AlthoughthissuiteoftraitsmaycomplicateN2fixerresponsestonutrients,itisplausiblethattheyare also nutrient limited in the low-light understorey (and to an in-creasingdegreeashigherlightavailabilityelevatesgrowthandnutri-entdemand).However,N2fixersmaybelessnutrientlimitedacrossalllightlevelsthannon-fixingsaplingsofsimilarshadetoleranceduetotheirabilitytofixN2inresponsetoNlimitation.

Weconductedaninsitumulti-nutrientfertilizationexperiment(N,P,Kandmicronutrients)ofnaturallyoccurringsaplings ina lowland

rainforest todirectly test for interactions amongnutrient limitation,lightavailabilityandfunctionalgrouporspecies identities.Toexam-ine sapling responses to fertilization and light availability, we usedstem growth, themost commonmetric ofwhole tree performanceandalsothemostpracticalmetric inthiscase,duetothechallengeofmeasuringbelow-groundgrowthbothat the individual scaleandinaninsituexperiment.Specifically,ourexperimentwasdesignedtotestthefollowinghypotheses:H1:Light-demandingsaplingsaremorenutrientlimitedthanshade-tolerant,non-fixingsaplingsacrossalllightlevels;H2:Shade-tolerant,non-fixingsaplingsaremorenutrient lim-itedthanshade-tolerant,N2-fixingsaplingsacrossalllightlevels;andH3:Nutrient-limitedfunctionalgroupsandspeciesareco-limitedbylight,withgreater lightavailabilityamplifying thedegreeofnutrientlimitation.

2  | MATERIALS AND METHODS

2.1 | Study site

Weconducted the experiment in the lowland tropical rainforest ofnortheastern Costa Rica at La Selva Biological Station (10°26′N,83°59′W).TheforestatLaSelvaisclassifiedastropicalwetforestintheHoldridge life-zone system (Hartshorn, 1983;Holdridge, 1967).Meanannualtemperatureis25.8°Candmeanannualprecipitationis3,962mm,withnotruedryseasonasnomonthsreceive<100mmrainfall(Sanford,Paaby,Luvall,&Phillips,1994).SoilsatLaSelvaareamong themost fertile found in neotropical lowland rainforests intermsofNandP,buthavelowerbasecationavailabilitythanmanyother tropical soils (Powers, Treseder, & Lerdau, 2005; Vitousek &Matson,1988).Weestablishedtheexperimentonprimarilyresidualultisol soils that have consistent chemical and morphological char-acteristics (Sollins, Sancho,Mata,& Sanford, 1994), in amix of oldgrowthandregeneratingforest,withanaverageelevationofapproxi-mately100m.

2.2 | Experimental design

Weselectednine common speciesof canopy trees at La Selvabe-longing to three functional groups—light-demanding, shade-tolerantandN2-fixing(O.Vargas,pers.comm.).Duetothecomplexityoftheshade tolerance growth-mortality trade-off, we used a single trait,seedgerminationshadetolerance,tosortspeciesintogeneralshadetolerancecategories(Clark&Clark,1992;Swaine&Whitmore,1988).Wedefinedlight-demandingspecies(Casearia arborea,Laetia procera and Simarouba amara)asrequiringgaplightconditionsforseedger-minationandshade-tolerantspecies (Hernandia didymantha,Protium pittieri,Virola koschnyi)ascapableofgerminatinginshadedundersto-rey.AsLaSelvahasanunusualabundanceofshade-tolerantN2fixersinthecanopy(Hartshorn&Hammel,1994;Lieberman&Lieberman,1987),wechosethreeN2-fixingspecies (Inga pezizifera, Inga thibau-diana and Pentaclethra macroloba) that were shade-tolerant by ourclassificationscheme.GiventhatspecificN2 fixationrateswerenotafocusofourstudyandthatourspeciesareknowntonodulateand

160  |    Journal of Ecology CHOU et al.

activelyfixN2atLaSelvaandinthebroaderCentralAmericanregion(Batterman etal., 2013; Carpenter, 1992), we did not measure N2 fixation rates. Doing so would have required repeated destructiverootsamplingthroughoutthestudy,whichwouldhaveimpactedthegrowthresponsesofinterest.

InAugustandSeptember2012,weidentified235naturallygrow-ingsaplingsintheforest,approximately26individualsperspecies,inagradientoflightconditionsrangingfromclosedcanopytothelargestcanopygapswecouldfind.Saplingsrangedfrom2.5to27mmindi-ameterand32to312cminheight.Wefertilizedapproximatelyhalfoftheindividualsofeachspeciesacrossthelightgradientwithaslow-release fertilizer (Miracle-Gro® Tree & Shrub Fertilizer Spikes; TheScottsCompany,Marysville,OH,USA)containingN(15%),P(5%),K(10%)plusmicronutrients:sulphur,ironandmanganese,andrepeatedfertilizationevery6monthsduring the2.5yearsof theexperiment.Saplingsreceivedonefertilizerstakeperapplicationevent,whichwasbrokenintofourevenlysizedpiecesandburied5cmbelowthesur-face0.6mawayfromthesteminthecardinaldirections.Thisresultedintheapplicationof0.0340kgN,0.0113kgP,and0.0227kgKpersaplingperyear, andassuming thenutrients spread to2m2 around each sapling, theapplication ratewasapproximately170kgNha−1 year−1,57kgPha−1 year−1and114kgKha−1 year−1,whichscalestoabout142%ofN inputs,1,256%ofP inputsand757%ofK inputsfromlitterfallmeasuredinthisforest(Wood,Lawrence,&Clark,2006).

2.3 | Census measurements

Foreverysapling,wemeasuredstemdiameter,stemheightandlightavailabilityevery6monthsover the2.5yearsof thestudy, thussixtimestotalforeachvariable.Wealsomeasuredfoliarnutrientconcen-trationsforeachindividual,buttheresponseswerecomplexandwethereforetreattheminaseparatecontribution.Duringeachcensus,wemeasuredstemdiametertothenearest0.1mmusingcallipersatamarkedpointofmeasurementbelowthelowestbranchandawayfromstemirregularitiesatheightsof0,40or130cmwhenpossible(Clark&Clark,1992).Forsaplings>4cmindiameterorforthosethathadhighlynon-cylindricalstems,weusedadiametertapetomeasurethestemtothenearestmillimetre.Wemeasuredallstemsofmulti-stemmedsaplingsat the samepointofmeasurement tocalculateadiameterequivalenttothatofasingle-stemmedtreeofequalbasalarea.

Additionally,duringeachcensus,wemeasuredsaplingheight tothenearestmillimetreusinga folding2-m ruler, orwhennecessaryto the nearest centimetre using an extendable 3- or 15-mmeasur-ingpole.Wedefinedheight as theperpendiculardistancebetweenthegroundandtallestmeristem,except inapproximately3%of thesaplings, where due to architectural form, growthwas consistentlyinabentdirectionthroughoutthestudyperiod,causingasaplingtobecome shorterwith time absent any breakage. In these cases,wemeasuredthebentstemlengthbetweenthegroundandfurthestmer-istem,andfoundthistobeanappropriateproxyforheightgrowth,asinclusionor exclusionof thesepointsdidnot fundamentally impactourresultsorconclusions.

Finally,wealsoquantifiedlightavailabilityforeachsaplingateachcensusbytakingahemisphericalphotographattheheightofitstallestleafusingaNikonCoolpix4500cameraequippedwiththeNikonFC-E8FisheyeConverter(Nikon,Tokyo,Japan),whichwasmountedonagyroscopicpoletoallowforlevelpicturesatgreaterheights.Photosweretakenpre-dawnoronuniformlycloudydays,andwereanalysedusingGap LightAnalyzerVersion 2.0 (Frazer, Canham,& Lertzman,1999)toquantifytotaltransmittedradiation.

2.4 | Tree growth analysis

We analysed tree growth responses to fertilization and light avail-ability using total growth between the first and last census in the2.5-studyperiodtocapturethestrongestsignaloftreeresponsetotheseresourceswhileminimizingmeasurementerrorsthatmaybeas-sociatedwiththeshortcensusintervals.Wealsofoundsimilarresultsfromamorecomplexrepeatedmeasuresanalysisthatusedthedatafromeachcensus(seeAppendixS1).Althoughweexaminedbothdi-ameterandheightgrowth,wecentreourinterpretationonthediame-terresultsasmeasurementsofsaplingdiametergrowthareinherentlylessvariablethanmeasurementsofsaplingheightgrowth,whichtendtoincludebreakageandheightloss.

Forbothdiameterandheightgrowth,weusedrelativegrowthrate(RGR)as the responsevariable inorder toaccount for theeffectoftreesizeongrowthrate,whereRGR=ln(sizefinal/sizeinitial)/(numberofstudydays/365). Individuals thatdidnotsurvivetheentire2.5-yearstudy periodwere excluded from all analyses, and individualswithmultiplestemsthathadnegativediametergrowthduetostemdeathand individuals that had negative height growth due to observedstembreakagewereexcludedfromthediameter(n=202)andheight(n=200)growthanalysesrespectively.

Weusedstepwiselinearregressiontoassesswhethereachfunc-tionalgroupandeachspecieswasnutrient limited in itsRGRand ifthis nutrient limitation interacted with light availability, which wecalculatedforeachsaplingasitsmeanlightavailabilityacrossthesixcensuses.Foreachfunctionalgroupandspecies,webeganwiththemaximalmodel,whereRGR~fertilizationtreatmentxlightavailability,and simplified to theminimal adequatemodel,which contains onlysignificantexplanatoryvariablesandinteractions.WeconfirmedthatregressionassumptionsweremetintheresidualsofeachmodelandalsotestedforinfluentialpointsusingCook’sdistance.

Functional groups or species that had a significant growth re-sponsetofertilization(withorwithoutlightinteractions)intheirmini-maladequatemodelwereconsiderednutrientlimited.Althoughtherewasvariabilityintherangesoflightavailabilityamongthefunctionalgroups and species due to the natural experimental design (light-demanding7.62%–26.33%;shade-tolerant5.70%–21.01%;N2-fixing5.83%–24.18%;C. arborea7.62%–17.17%;L. procera8.63%–22.29%;S. amara 7.84%–26.33%; H. didymantha 5.70%–12.83%; P. pittieri 6.62%–21.01%; V. koschnyi 7.44%–15.34%; I. pezizifera 5.83%–24.18%; I. thibaudiana 8.80%–17.09%; P. macroloba 8.31%–14.15%;allrangesinpercenttotaltransmittedradiation),thelinearityofthedatareassuresusthatthelinearregressionmodelsweresuitablefor

     |  161Journal of EcologyCHOU et al.

understandingtherelativeresponsestofertilizationandlightavailabil-ityamongthefunctionalgroupsandmostspecies.Allstatisticalanaly-seswereperformedinr3.0.2(RCoreTeam,2013).

3  | RESULTS

3.1 | Variable nutrient- light responses across functional groups

Wefoundauniquediametergrowthresponsetonutrientsandlightin each of the functional groups (Figure2, see Table1 for detailedresults from all diameter RGR models). All groups responded sig-nificantlyandpositivelytotheeffectof lightalone(p<.001)andinaddition, somegroups respondedpositivelyandsomenegatively tofertilization×lightinteractions.

Thelight-demandingfunctionalgroupdidnotrespondsignificantlyto fertilization (Figure2a). In contrast, the shade-tolerant functionalgroup showed a significant positive growth response to a fertiliza-tion×light interaction (p<.001; Figure2b).As a result, fertilization

increasedtheslopeofthepositiverelationshipbetweenRGRandlightavailabilityby2.5times,sothattheresponsetofertilizationincreasedwith light availability.At very low light, therewas a slight negativeinfluenceof fertilizationon growth thatwas likely the result of thestrongpositiveinteractionterm,althoughitisalsopossiblethatfertil-izationmildlysuppressedgrowthintheseconditions.

TheN2-fixing functional group also responded significantly to afertilization×lightinteraction,butdifferedfromshade-tolerantgroupinthatthisinteractionwasnegative(p=.01;Figure2c).ForN2-fixingsaplings,fertilizationdecreasedtheslopeofthepositiverelationshipbetweenRGRandlightavailabilitytoonequarteroftheunfertilizedslope,withsaplingsrespondingpositivelytofertilizationat lowlightandnegativelyathighlight.

3.2 | Variable nutrient- light responses across species

Although we did not find a response to fertilization in thelight-demandingfunctionalgroupasawhole,C. arboreadidrespondsignificantly and positively to a fertilization×light interaction

F IGURE  2 Relationshipbetweendiameterrelativegrowthrate(RGR)andlightavailabilitybyfunctionalgroupandfertilizationtreatment(redcircles=fertilizedandblackcrosses=unfertilized).Significancevaluesandlinesrepresenttheminimaladequatemodelforeachfunctionalgroup(red=fertilized,black=unfertilizedandblue=nosignificantfertilizationtreatment)[Colourfigurecanbeviewedatwileyonlinelibrary.com]

××

(a) (b) (c)

TABLE  1 Diameterrelativegrowthrateregressiontable:samplesize(n),regressionparameterestimates,adjustedmultipleR2 and whole model p-valuefortheminimaladequatemodelofeachfunctionalgroupandspecies.Speciesarearrangedbyfunctionalgroups(LD=light-demanding,ST=shade-tolerantandNF=N2-fixing).NAindicatesthatthefactororinteractionwasnotincludedintheminimaladequatemodel. †p<.1;*p<.05,**p<.01;***p<.001

Functional group n Intercept Light Fertilization Light × Fertilization Adjusted R2 Model p- value

Light-demanding 67 −0.14 0.029*** NA NA .36 <.001

Shade-tolerant 69 −0.050 0.015*** −0.19** 0.022*** .56 <.001

Nitrogen-fixing 66 −0.20 0.028*** 0.25** −0.021* .23 <.001

Species

Casearia arborea(LD) 24 −0.086 0.015 −0.43* 0.057** .55 <.001

Laetia procera(LD) 20 −0.034 0.024† NA NA .15 .051

Simarouba amara(LD) 23 −0.13 0.027*** NA NA .56 <.001

Hernandia didymantha(ST) 24 0.12 NA NA NA NA NA

Protium pittieri(ST) 22 −0.071 0.018** −0.28* 0.028** .77 <.001

Virola koschnyi(ST) 23 −0.073 0.015† −0.21 0.026† .57 <.001

Inga pezizifera(NF) 24 −0.14 0.021** 0.26* −0.021* .24 .038

Inga thibaudiana(NF) 19 −0.38 0.044*** NA NA .48 <.001

Pentaclethra macroloba(NF) 23 −0.17 0.028* NA NA .15 .041

162  |    Journal of Ecology CHOU et al.

(p=.007; Figure3a). Fertilization increased the response slopebetweenRGRand lightavailabilityby4.8times,sothatthegrowthincrease from fertilizationwas greaterwith higher light availability,although(asdiscussedabove)therewasaslightnegativeresponseatverylowlight.Incontrast,wedidnotfindanysignificantresponsestofertilizationinL. procera or S. amara,theothertwolight-demandingspecies(Figure3d,g).However,bothL. procera(p=.05)andS. amara (p<.001)showedasignificantpositivegrowthresponsetolightalone,while C. arboreadidnot.

Intheshade-tolerantfunctionalgroup,wefoundthatP. pittieri and V. koschnyirespondedsignificantlyandpositivelytofertilization×lightinteractions (p=.003 and p=.05, respectively; Figure3e,h), asobservedforthefunctionalgroupasawhole.FertilizationincreasedtheRGRvs.lightslopeforP. pittieriby2.6timesandforV. koschnyi by 2.7times.Forbothspecies,thereagainwasaslightnegativefertiliza-tioneffectatverylowlight.Protium pittierialsoshowedanadditional,significant positive response to the effect of light alone (p=.007),while V. koschnyi did not. In contrast, the third shade-tolerant spe-cies,H. didymantha,didnotrespondtoeitherresource,althoughthisspeciesdidhavearestrictedlightavailabilityrangeinourexperimentthatmayhaveobstructedtheobservationofitscompleteresponsetotheseresources(Figure3b).

Finally,asseenfortheN2-fixingfunctionalgroupasawhole,thegrowthofI. peziziferarespondednegativelytoafertilization×lightin-teraction(p=.03;Figure3c).FertilizationdecreasedthepositiveRGRvs.lightslopetonearzero,resultinginapositivefertilizationresponseatlowlightandanegativeresponseathighlight.Inga peziziferaalsorespondedpositively ingrowth to lightalone (p=.007). In contrast,thegrowthofbothI. thibaudiana(p<.001)andP. macroloba(p=.04)respondedpositivelytolightalone,butdidnotsignificantlyrespondtofertilization(Figure3f,i).However,P. macrolobaalsohadarestrictedlight availability range inourexperiment,whichmayhaveobscuredourunderstandingofitsresponsetobothresources.

Theresultsforthreespeciesweresensitivetotheinfluenceofasinglesaplinginthehighestlightenvironment(Cook’sdistancesof1.53forC. arborea,5.8forS. amaraand2.25forP. pittieri),inthatremovingtheinfluentialpointchangedtheminimaladequatemodel.However,eachofthesepointsisvaluableforrevealingthegrowthresponsesweareassessing,ashighlightcanbecriticalforsaplingsuccess,butitcanbeexceedinglydifficulttofindnaturallyoccurringsaplingsofcertainspecies invery large forestgaps. Innocasecouldwe finda reasontoexcludethepoints,evenfollowingathoroughexaminationofdataaccuracyandanevaluationofthebiologicalfeasibilityoftheobservedgrowthrates.

F IGURE   3 Relationshipbetweendiameterrelativegrowthrate(RGR)andlightavailabilitybyspeciesandfertilizationtreatment(redcircles=fertilizedandblackcrosses=unfertilized).Significancevaluesandlinesrepresenttheminimaladequatemodelforeachspecies(red=fertilized,black=unfertilizedandblue=nosignificantfertilizationtreatment).Speciesarearrangedbyfunctionalgroupcolumns(LD=light-demanding,ST=shade-tolerantandNF=N2-fixing)[Colourfigurecanbeviewedatwileyonlinelibrary.com]

× ×

×

×

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

     |  163Journal of EcologyCHOU et al.

3.3 | Comparable height growth nutrient responses

Our results from the analyses of the height and diameter growthdata were similar, despite the inherently larger variance of theheightgrowthdata: (1)significantresponsestofertilization× lightinteractionsfortheshade-tolerantandN2-fixingfunctionalgroups(Figure4,seeTable2fordetailedresultsfromallheightRGRmod-els); (2) significant or near-significant responses to fertilization orfertilization×light interactions in C. arborea, P. pittieri, V. koschnyi and I. pezizifera (Figure5);and(3)significantornear-significantre-sponsestofertilization×lightinteractionsinP. pittieri and I. pezizif-era (Figure5). However, the nature of the diameter and heightgrowth responses to fertilizationwas dissimilar forC. arborea and V. koschnyi,becauseheightgrowthrespondedtofertilizationwith-outanyinteractionswithlightavailability.Finally,aswiththediam-etergrowthresults,almostallfunctionalgroupsandspeciesshowedasignificantpositivegrowthresponselightalone(p<.05;Figures4and5).

4  | DISCUSSION

Lowland tropical rainforest saplings employ a variety of strategiestocompetefornutrientsandlightduringthesaplingbottleneck.Wefound significant nutrient limitation in two out of the three func-tionalgroupsandfouroutoftheninespeciesweexamined,aswellas generally positive growth responses to increasing light avail-ability, indicating that nutrient and light co-limitationmay exist inmanyfunctionalgroupsandspeciesatLaSelva.Thepervasivenessof strong growth responses tonutrients inour saplings, even in asiteasnutrientrichasLaSelva,confirmstheimportanceofnutrientsin addition to light availability for sapling growth and emphasizesthesignificanceofsaplingnutrientandlightco-limitationinlowlandtropical rainforest nutrient dynamics. Additionally, this study re-vealedfunctionalgroup-andspecies-specific interactionsbetweennutrientlimitationandlightavailability,someofwhichcounterpre-vailinghypothesesoftreeresourceacquisitionstrategiesandnutri-enteconomies.

F IGURE  4 Relationshipbetweenheightrelativegrowthrate(RGR)andlightavailabilitybyfunctionalgroupandfertilizationtreatment(redcircles=fertilizedandblackcrosses=unfertilized).Significancevaluesandlinesrepresenttheminimaladequatemodelforeachfunctionalgroup(red=fertilized,black=unfertilizedandblue=nosignificantfertilizationtreatment)[Colourfigurecanbeviewedatwileyonlinelibrary.com]

(a) (b) (c)××

TABLE  2 Heightrelativegrowthrateregressiontable:samplesize(n),regressionparameterestimates,adjustedmultipleR2 and whole model p-valuefortheminimaladequatemodelofeachfunctionalgroupandspecies.Speciesarearrangedbyfunctionalgroups(LD=light-demanding,ST=shade-tolerantandNF=N2-fixing).NAindicatesthatthefactororinteractionwasnotincludedintheminimaladequatemodel. †p<.1;*p<.05,**p<.01,***p<.001

Functional group n Intercept Light Fertilization Light × Fertilization Adjusted R2 Model p- value

Light-demanding 64 −0.17 0.033*** NA NA .34 <.001

Shade-tolerant 66 −0.050 0.017** −0.17† 0.020* .37 <.001

Nitrogen-fixing 70 −0.28 0.036*** 0.29* −0.024* .27 <.001

Species

Casearia arborea(LD) 23 −0.47 0.052*** 0.15** NA .52 <.001

Laetia procera(LD) 20 −0.022 0.027† NA NA .13 .070

Simarouba amara(LD) 21 −0.16 0.028*** NA NA .57 <.001

Hernandia didymantha(ST) 24 −0.067 0.020* NA NA .14 .042

Protium pittieri(ST) 20 0.049 0.012 −0.39† 0.032† .40 .011

Virola koschnyi(ST) 22 −0.16 0.023* 0.086* NA .46 .0011

Inga pezizifera(NF) 26 −0.22 0.030*** 0.31** −0.025** .45 <.001

Inga thibaudiana(NF) 21 −0.55 0.060** NA NA .40 .0012

Pentaclethra macroloba(NF) 23 −0.15 0.026* NA NA .14 .043

164  |    Journal of Ecology CHOU et al.

4.1 | Counterintuitive functional group nutrient limitation

ThefunctionalgroupresultsfalsifiedH1,thehypothesisthatlight-demandingsaplingsaremorenutrientlimitedthanshade-tolerant,non-fixingsaplings.First,wefoundnoevidenceofnutrientlimita-tioninthelight-demandingsaplings(Figure2a),despitetheirpro-pensity to have traits that lower their NUE (Poorter & Bongers,2006;Swaine&Whitmore,1988).Inaddition,wedidfindsignifi-cant nutrient limitation in the shade-tolerant, non-fixing saplings(Figure2b), although we expected these saplings to have traitsthat allow for greaterNUE.Thenutrient limitation in the shade-tolerant, non-fixing saplings did increase with light availabilityas hypothesized inH3, with fertilizationmore than doubling theslopeofthepositiverelationshipbetweendiameterRGRandlightavailability.

Asecondsurprisewasthatthefunctionalgroupanalysisalsofal-sifiedH2.Althoughwe found significantnutrient limitation inboththeshade-tolerant,non-fixingsaplingsandshade-tolerant,N2-fixingsaplings, the strength of nutrient limitation was not consistentlygreaterintheshade-tolerantnon-fixers,aspredictedinH2.Whiletheshade-tolerant, non-fixing saplings followed the pattern predictedin H3, unexpectedly, the N2-fixing saplings displayed the opposite

pattern,withfertilizationincreasingtheRGRofsaplingsinlowlightbutthestrengthofthisnutrient limitationdecreasingas lightavail-abilityincreasedsothattherewasanegativeresponsetofertilizationathighlight(Figure2c).Althoughtheshade-tolerantfunctionalgroupappearsmorenutrientlimitedthantheN2-fixingfunctionalgroupathigh-lightlevels,supportingH2,therewererelativelyfewsaplingsintheselightconditions.Thus,thedifferenceintheresponsebetweenthetwogroupsisdrivenprimarilybythelowerlightsaplings,wheretheN2fixersweremorenutrientlimitedthanthenon-fixers,falsify-ingH2.

We were surprised by these results because N2 fixation isthought to provide the greatest competitive benefits either earlyin succession or during gap succession in mature forests, whenrapidgrowthcreatesthehighestNdemand(Battermanetal.,2013;Menge&Chazdon,2016).Thus, ifN2fixerswerenutrient limited,wewouldexpectthislimitationtobestrongestathigh-lightlevels(aspredicted inH3), and that this limitationwouldbebyPorMo(Barronetal.,2009;Vitousek&Howarth,1991).Wewouldalsoex-pect thatnutrient limitationof shade-tolerantN2 fixerswouldbelower thanshade-tolerantnon-fixers (aspredicted inH2), asnon-fixerscannotfixtheirownN2.Weexplorethesaplingandcanopytree feedbacks impliedby thiscomplexpatternofnutrient limita-tionindetailbelow.

F IGURE  5 Relationshipbetweenheightrelativegrowthrate(RGR)andlightavailabilitybyspeciesandfertilizationtreatment(redcircles=fertilizedandblackcrosses=unfertilized).Significancevaluesandlinesrepresenttheminimaladequatemodelforeachspecies(red=fertilized,black=unfertilizedandblue=nosignificantfertilizationtreatment).Speciesarearrangedbyfunctionalgroupcolumns(LD=light-demanding,ST=shade-tolerantandNF=N2-fixing)[Colourfigurecanbeviewedatwileyonlinelibrary.com]

×

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

×

     |  165Journal of EcologyCHOU et al.

4.2 | Variable species nutrient limitation within functional groups

Whenexaminingnutrientlimitationbyspecies,wefoundvariablere-sponses to nutrientswithin functional groups. The light-demandingfunctionalgroupwasnotsignificantlynutrientlimitedasawhole,butone out of the three species, C. arborea, was significantly nutrientlimited(Figure3a).Intheshade-tolerantfunctionalgroup,twooutofthe threespecies,P. pittieri and V. koschnyi,weresignificantlynutri-ent limited,andtheirpatternsofnutrient limitationwereconsistentwiththatobserved inthefunctionalgroupasawhole (Figure3e,h).Finally, in theN2-fixing functional group,onlyoneoutof the threespecies,I. pezizifera,wassignificantlynutrientlimited,andonceagainthepatternofnutrientlimitationinthisspecieswasconsistentwiththefunctionalgroup-levelnutrientlimitation(Figure3c).

Notably,thepatternsofnutrientlimitationwefoundinC. arborea vs.P. pittieri and V. koschnyiwereexactlywhatweexpectedforlight-demanding saplings relative to shade-tolerant saplings, aspredictedin H1.All three species had increasing nutrient limitationwith lightavailability as predicted inH3, and the strength of nutrient limita-tionwasmuchgreater inthe light-demandingC. arborea than intheshade-tolerant P. pittieri and V. koschnyi (Figure3a vs. Figure 3e,h).FertilizationincreasedtheslopeofthepositiverelationshipbetweendiameterRGRand lightby4.8 times inC. arborea, compared to2.6timesinP. pittieriand2.7timesinV. koschnyi.

AlthoughtherestrictedrangesoflightavailabilityforH. didyman-tha and P. macrolobamay have limited a few species-level compari-sons,itisclearinothercases,forexamplewiththelight-demandingC. arborea and L. procera (Figure3a,d), that specieswithin the samefunctional group can have entirely different responses to nutrients.Thus,althoughfunctionalgroupclassificationscanbequiterepresen-tativeforsomespecies,theyarenotforothers.

4.3 | Nutrient limitation in shade- tolerant N2 fixers: A case study of sapling- canopy feedbacks

In addition to our unexpected finding that shade-tolerantN2 fixerswerestronglynutrientlimitedatlow-lightlevelswhileshade-tolerantnon-fixerswerenot, the forestatLaSelvahas threeotherunusualcharacteristics: (1)Relative tootherneotropical lowlandrainforests,LaSelvaisknownforitshighabundanceofanddominancebyshade-tolerant, N2-fixing species, with P. macroloba alone accounting for12.4%–13.7% of stems and 34.6%–36.0% of basal area in matureforest (Hartshorn&Hammel,1994;Lieberman&Lieberman,1987);(2)LaSelvasoilsareknowntobehighlyNrichrelativetosoilsfromotherneotropicallowlandrainforests(Powersetal.,2005;Vitousek&Matson,1988);and(3)LaSelvasoilsarealsoknowntobehighlyPrichrelative to soils fromother neotropical lowland rainforests (Powersetal.,2005).

Together,theselinesofevidenceimplyanicheforashade-tolerant,N2 fixer strategy that functions through a sapling and canopy treefeedback.UnlikethepredominantperspectivethatN2fixationismostbeneficial inhigh-light, successionalenvironmentswhereNdemand

ishighrelativetosupply(Battermanetal.,2013;Menge&Chazdon,2016),N2fixationmayalsohelpshade-tolerantN2fixersinlow-lightenvironments,withthebenefitseennotonlywithinindividualsfixingN2fortheirowngainbutalsoacrosslife-historystageswithcanopyN2 fixersmodifyingtheenvironmentfavourablyfortheirsaplings.

Consider this feedback at the ecosystem scale, where shade-tolerant N2 fixers in the canopy are able to fix large quantities ofN2 and enrich soil N via their N-rich foliage and litterfall, whichthen helps their shade-tolerant, N2-fixing saplings grow faster thanshade-tolerant, non-fixing competitors, which in turn increases theabundanceofshade-tolerant,N2-fixingcanopytrees(Figure1).Thereis evidence for this feedback cycle at La Selva, as shade-tolerant,N2-fixing treesaredominant in thecanopy, theN2-fixing functionalgroup had significantly higher foliar N content in this experiment(ANOVAF2,179=51.08,p<.001;TukeyHSDp<.001forallcompar-isons;C.B.Chou,unpubl.data),thereishighsoilNandtheN2-fixingsaplings significantly increased growth rates in response to fertil-ization in thisexperiment.Additionally, a clue tohow this feedbackemergedatLaSelvaandnotatotherneotropicallowlandrainforestsmaybethehighsoilP,whichcouldpotentiallybeone(butcertainlynot theonly) factor that allowed for the selectionofN2 fixerswithhighlyN-demandinglifestyles(Vitousek&Howarth,1991).

GiventhelinesofevidenceatLaSelvasupportingourhypothesizedshade-tolerant,N2fixernichewhereshade-tolerant,N2-fixingsaplingsbenefitfromhighsoilN,wehypothesizethattheshade-tolerant,N2-fixingsaplingsinourstudywerelikelyco-limitedbylightandN,ratherthan lightandPorMo.Specifically,theadditionofNfromfertilizermayhavedown-regulatedN2fixationinlow-lightsaplingswheretheprocesswas carbon costly, allowing them to shift the carbon theywereusingtofeedtheirrhizobiatogrowthinstead(Hedin,Brookshire,Menge,&Barron,2009).Aslightlimitationdecreased,makingfixationrelativelylesscarboncostly,theN2fixersmayhavebeenabletomeettheelevatedNdemandof theirhigh-lightgrowth rates themselves,diminishingtheimpactofthefertilizerNongrowth.Inaddition,thediscretefertilizationeventsmayhaveunintentionallycausedanega-tivegrowthresponsetofertilizationathighlightbytriggeringdown-regulationofN2fixationwithoutmeetingthefullNdemandofthesefast-growingindividuals,whileatlowlight,theentireNdemandoftheslowergrowingindividualswasmetbythefertilizer.

Alternatively,iftheN2fixerswerelimitedbyPorMoandlight,theadditionofPortraceamountsofMofromfertilizermayhaveallowedlow-lightsaplingstofixmoreN2andincreasetheirlightcaptureeffi-ciencyandRGRbygrowingmorenutrient(especiallyN)-richleavesormoreleavesoverall.Inthiscase,thehigh-light,N2-fixingsaplingswerelikely still P orMo limited, but the lack of high-light individuals didnotallowustosufficientlytestforafertilizationresponse.However,given the statistically significant divergent responses to fertilizationbetween shade-tolerant non-fixers and shade-tolerant N2 fixers(Figure2b,c),andtheirsimilarlysmallnumbersofhigh-lightsaplings,thisexplanationislessparsimonious.

Incontrast,thenon-fixingfunctionalgroupsappearedpurelylightlimitedatlow-lightlevels,likelybecausetheyhadgreaterNUEandalessN-demandinglifestylethanN2-fixingspecies(McKey,1994).This

166  |    Journal of Ecology CHOU et al.

resultemphasizesthecostlinessofhighleafNconcentrationsintheN2fixers,whichatlow-lightlevelsoutweighedthegrowthadvantagetheyshouldhaveseenfromtheirabilitytofixN2.

5  | CONCLUSIONS

This study revealed pervasive nutrient and light co-limitation ofsaplings growing in a lowland tropical rainforest with highly fertilesoils, emphasizing the importance of sapling nutrient-light interac-tions in thenutrientdynamicsof theseecosystems.Moreover, thisco-limitation was functional group- and species-specific, providingevidence for “heterogeneousnutrient limitation” by tree taxonomicidentity (Alvarez-Clare etal., 2013) as well as functional identity,althoughfurtherstudiescanenhanceourunderstandingofeffectivetaxonomicorfunctionalgroupingsforpredictingnutrientresponses.Withinthefunctionalgroupsweused,wefoundstrongnutrientlimi-tationatlow-lightlevelsintheshade-tolerantN2fixers,butnotintheshade-tolerant non-fixers. This is evidence for a shade-tolerant,N2 fixationnichethroughasaplingandcanopytreefeedbackcyclewhereshade-tolerant,N2-fixingcanopytreesenrichsoilNtothebenefitoftheirsaplings,allowingthemtodominateforestcanopycomposition,asseenatLaSelva.Therearelikelyadditional,variedsapling-canopynutrientandlightfeedbacksinlowlandtropicalrainforests,andmorestudiesof these feedbacks, combinedwith careful considerationofappropriate taxonomic or functional groupings, can aid our under-standingofnutrientlimitationdynamicsintheseecosystems.

ACKNOWLEDGEMENTS

WethankDeborahA.Clarkforadviceonexperimentaldesignandwis-domaboutLaSelva.WethankLeoCampos,AdemarHurtado,WilliamMirandaandOrlandoVargasforplantidentification;GeraldCampos,JohnnyFlores,ZoeC.Sims,TimothyL.H.Treuer,RubenVargasandChhayaM.Wernerforfieldassistance;AdamF.A.Pellegrini,SamS.Rabin,AnnetteM.Trierweiler,IsaacK.Uyehara,MarcoD.VisserandS.JoeWrightforfeedback;andmanyotherLaSelvastaffmembersfor their help. This work was supported by the CarbonMitigationInitiative at Princeton University and an NSF Graduate ResearchFellowshiptoC.B.C.

AUTHORS’ CONTRIBUTIONS

C.B.C.,L.O.H.andS.W.P.conceivedtheideasanddesignedmethodol-ogy;C.B.C.collectedthedata,analysedthedataandledthewritingofthemanuscript.Allauthorscontributedcriticallytothedraftsandgavefinalapprovalforpublication.

DATA ACCESSIBILITY

Data available from the Dryad Digital Repository https://doi.org/10.5061/dryad.k2152(Chou,Hedin,&Pacala,2017).

REFERENCES

Adams,M.A.,Turnbull,T.L.,Sprent,J.I.,&Buchmann,N.(2016).Legumesaredifferent:Leafnitrogen,photosynthesis,andwateruseefficiency.Proceedings of the National Academy of Sciences of the United States of America,113,4098–4103.

Alvarez-Clare,S.,Mack,M.C.,&Brooks,M.(2013).Adirecttestofnitro-genandphosphoruslimitationtonetprimaryproductivityinalowlandtropicalwetforest.Ecology,94,1540–1551.

Barron,A.R.,Wurzburger,N.,Bellenger,J.P.,Wright,S.J.,Kraepiel,A.M.L.,&Hedin,L.O.(2009).Molybdenumlimitationofasymbioticnitrogenfixationintropicalforestsoils.Nature Geoscience,2,42–45.

Batterman,S.A.,Hedin,L.O.,vanBreugel,M.,Ransijn,J.,Craven,D.J.,&Hall,J.S. (2013).Keyroleofsymbioticdinitrogenfixation in tropicalforestsecondarysuccession.Nature,502,224–227.

Brokaw, N. V. L. (1985). Gap-phase regeneration in a tropical forest.Ecology,66,682–687.

Carpenter,E.J.(1992).NitrogenfixationintheepiphyllaeandrootnodulesoftreesinthelowlandtropicalrainforestofCostaRica.Acta Oecologica,13,153–160.

Chou,C.B.,Hedin, L.O.,&Pacala, S.W. (2017).Data from: Functionalgroups,species,andlightinteractwithnutrientlimitationduringtrop-ical rainforestsaplingbottleneck.Dryad Digital Repository,https://doi.org/10.5061/dryad.k2152.

Clark, D.A., & Clark, D. B. (1992). Life history diversity of canopy andemergenttreesinaneotropicalrainforest.Ecological Monographs,62,315–344.

Denslow, J. S. (1980). Gap partitioning among tropical rainforest trees.Biotropica,12,47–55.

Denslow,J.S. (1987).Tropical rainforestgapsand treespeciesdiversity.Annual Review of Ecology and Systematics,18,431–451.

Denslow,J.S.,Schultz,J.C.,Vitousek,P.M.,&Strain,B.R.(1990).Growthresponsesoftropicalshrubstotreefallgapenvironments.Ecology,71,165–179.

Fetcher, N., Haines, B. L., Cordero, R. A., Lodge, D. J., Walker, L. R.,Fernández,D.S.,&Lawrence,W.T.(1996).ResponsesoftropicalplantstonutrientsandlightonalandslideinPuertoRico.Journal of Ecology,84,331–341.

Fisher,J.B.,Malhi,Y.,Torres,I.C.,Metcalfe,D.B.,vandeWeg,M.J.,Meir,P.,…Huasco,W.H.(2013).Nutrientlimitationinrainforestsandcloudforests along a 3,000-m elevation gradient in the Peruvian Andes.Oecologia,172,889–902.

Frazer,G.W.,Canham,C.D.,&Lertzman,K.P.(1999)GapLightAnalzyer(GLA),Version2.0: Imagingsoftwaretoextractcanopystructureandgap light indices from true-colour fisheyephotographs,users manual and program documentation. Millbrook, NY: Simon Fraser University,Burnaby,BritishColumbia,andtheInstituteofEcosystemStudies.

Hartshorn,G.S. (1983).Plants. InD.H.Janzen (Ed.),Costa Rican natural history(pp.118–157).Chicago,IL:UniversityofChicagoPress.

Hartshorn,G. S.,&Hammel, B. E. (1994).Vegetation types and floristicpatterns. In L. A.McDade, K. S. Bawa, H. A. Hespenheide, & G. S.Hartshorn (Eds.), La Selva: Ecology and natural history of a neotropical rain forest(pp.73–89).Chicago,IL:TheUniversityofChicagoPress.

Hedin,L.O.,Brookshire,E.N.J.,Menge,D.N.L.,&Barron,A.R.(2009).Thenitrogenparadox intropical forestecosystems.Annual Review of Ecology, Evolution, and Systematics,40,613–635.

Holdridge, L. R. (1967).Life zone ecology, revised. SanJose,CA:TropicalScienceCenter.

Holste,E.K.,Kobe,R.K.,&Vriesendorp,C.F.(2011).Seedlinggrowthre-sponses to soil resources in the understory of awet tropical forest.Ecology,92,1828–1838.

Huntingford,C.,Zelazowski,P.,Galbraith,D.,Mercado, L.M., Sitch,S.,Fisher,R.,…Cox,P.M.(2013).Simulatedresilienceoftropicalrainfor-eststoCO2-inducedclimatechange.Nature Geoscience,6,268–273.

     |  167Journal of EcologyCHOU et al.

Körner,C.(2009).ResponsesofhumidtropicaltreestorisingCO2. Annual Review of Ecology, Evolution, and Systematics,40,61–79.

Lieberman,D.,& Lieberman,M. (1987). Forest tree growth anddynam-icsatLaSelva,CostaRica(1969-1982).Journal of Tropical Ecology,3,347–358.

Magalhães,N.S.,Marenco,R.A.,&Camargo,M.A.B.(2014).Dosoilfertil-izationandforestcanopyfoliageaffectthegrowthandphotosynthesisofAmazoniansaplings?Scientia Agricola,71,58–65.

McKey,D. (1994). Legumesandnitrogen:Theevolutionaryecologyof anitrogen-demandinglifestyle.InJ.I.Sprent&D.McKey(Eds.),Advances in legume systematics 5. The nitrogen factor (pp. 211–228). Kew,UK:RoyalBotanicGardens.

Menge,D.N.L.,&Chazdon,R.L.(2016).Highersurvivaldrivesthesuccessofnitrogen-fixingtreesthroughsuccessioninCostaRicanrainforests.New Phytologist,209,965–977.

Mirmanto,E.,Proctor,J.,Green,J.,Nagy,L.,&Suriantata(1999).Effectsof nitrogen and phosphorus fertilization in a lowland evergreenrainforest. Philosophical Transactions of the Royal Society B, 354,1825–1829.

Montagnini, F. (2000). Accumulation in above-ground biomass and soilstorageofmineralnutrientsinpureandmixedplantationsinahumidtropicallowland.Forest Ecology and Management,134,257–270.

Newbery,D.M.,Chuyong,G.B.,Green,J.J.,Songwe,N.C.,Tchuenteu,F.,&Zimmermann,L.(2002).DoeslowphosphorussupplylimitseedlingestablishmentandtreegrowthingrovesofectomycorrhizaltreesinacentralAfricanrainforest?New Phytologist,156,297–311.

Pacala, S.W.,Canham,C.D., Saponara,J., Silander,J.A.,Kobe,R.K.,&Ribbens, E. (1996). Forest models defined by field measurements:Estimation, error analysis and dynamics. Ecological Monographs, 66,1–43.

Palow,D.T.,&Oberbauer, S. F. (2009). Soil type affects seedling shaderesponseat lowlightfortwo IngaspeciesfromCostaRica.Plant and Soil,319,25–35.

Pasquini, S. C., & Santiago, L. S. (2012). Nutrients limit photosynthesisinseedlingsofa lowlandtropical forest treespecies.Oecologia,168,311–319.

Poorter,L.,&Bongers,F. (2006).Leaf traitsaregoodpredictorsofplantperformanceacross53rainforestspecies.Ecology,87,1733–1743.

Powers,J.S.,Treseder,K.K.,&Lerdau,M.T.(2005).Fineroots,arbuscu-larmycorrhizalhyphaeandsoilnutrients infourneotropicalrainfor-ests:Patternsacrosslargegeographicdistances.New Phytologist,165,913–921.

RCoreTeam.(2013).R: A language and environment for statistical comput-ing.Vienna,Austria:RFoundationforStatisticalComputing.Retrievedfromhttp://www.R-project.org/.

Reich,P.B.,Walters,M.B.,&Ellsworth,D.S.(1997).Fromtropicstotundra:Global convergence in plant functioning. Proceedings of the National Academy of Sciences of the United States of America,94,13730–13734.

Sanford, R. L., Paaby, P., Luvall, J.C.,&Phillips, E. (1994).Climate, geo-morphology,andaquaticsystems.InL.A.McDade,K.S.Bawa,H.A.Hespenheide,&G.S.Hartshorn(Eds.),La Selva: Ecology and natural his-tory of a neotropical rain forest(pp.19–33).Chicago,IL:TheUniversityofChicagoPress.

Santiago,L.S.(2015).Nutrientlimitationofeco-physiologicalprocessesintropicaltrees.Trees,29,1291–1300.

Santiago,L.S.,Wright,S.J.,Harms,K.E.,Yavitt,J.B.,Korine,C.,Garcia,M.N.,&Turner,B.L. (2012).Tropicaltreeseedlinggrowthresponsestonitrogen,phosphorusandpotassiumaddition.Journal of Ecology,100,309–316.

Sollins,P.,Sancho,F.,Mata,R.,&Sanford,R.L.(1994).Soilsandsoilpro-cessresearch.InL.A.McDade,K.S.Bawa,H.A.Hespenheide,&G.S.Hartshorn(Eds.),La Selva: Ecology and natural history of a neotropical rain forest(pp.34–53).Chicago,IL:TheUniversityofChicagoPress.

Swaine,M.D.,&Whitmore,T.C. (1988).Onthedefinitionofecologicalspeciesgroupsintropicalrainforests.Vegetatio,75,81–86.

Vitousek,P.M.,&Howarth,R.W.(1991).Nitrogenlimitationonlandandinthesea:Howcanitoccur?Biogeochemistry,13,87–115.

Vitousek,P.M.,&Matson,P.A.(1988).Nitrogentransformationsinarangeoftropicalforestsoils.Soil Biology and Biochemistry,20,361–367.

Wood,T.E.,Lawrence,D.,&Clark,D.A.(2006).Determinantsofleaflitternutrientcyclinginatropicalrainforest:Soilfertilityversustopography.Ecosystems,9,700–710.

Wright,S.J.,Kitajima,K.,Kraft,N.J.B.,Reich,P.B.,Wright,I.J.,Bunker,D.E.,…Zanne,A.E.(2010).Functionaltraitsandthegrowth-mortalitytrade-offintropicaltrees.Ecology,91,3664–3674.

Wright,S.J.,Yavitt,J.B.,Wurzburger,N.,Turner,B.L.,Tanner,E.V.J.,Sayer,E.J.,…Corre,M.D.(2011).Potassium,phosphorus,ornitrogenlimitrootallocation,treegrowth,or litterproduction ina lowlandtropicalforest.Ecology,92,1616–1625.

Yoda, K. (1974). Three-dimensional distribution of light intensity in atropical rainforestofWestMalaysia.Japanese Journal of Ecology,24,247–254.

SUPPORTING INFORMATION

Additional Supporting Information may be found online in thesupportinginformationtabforthisarticle.

How to cite this article:ChouCB,HedinLO,PacalaSW.Functionalgroups,speciesandlightinteractwithnutrientlimitationduringtropicalrainforestsaplingbottleneck.J Ecol. 2018;106:157–167. https://doi.org/10.1111/1365-2745.12823