Embed Size (px)
Citation preview
J Appl Ecol. 2019;56:733–744. wileyonlinelibrary.com/journal/jpe | 733© 2018 The Authors. Journal of Applied Ecology © 2018 British Ecological Society
Received:15January2018 | Accepted:23October2018DOI:10.1111/1365-2664.13308
R E S E A R C H A R T I C L E
Identifying the tree species compositions that maximize ecosystem functioning in European forests
Lander Baeten1 | Helge Bruelheide2,3 | Fons van der Plas4,5 | Stephan Kambach2,3 | Sophia Ratcliffe4,6 | Tommaso Jucker7,8 | Eric Allan9 | Evy Ampoorter1 | Luc Barbaro10 | Cristina C. Bastias11 | Jürgen Bauhus12 | Raquel Benavides11 | Damien Bonal13 | Olivier Bouriaud14 | Filippo Bussotti15 | Monique Carnol16 | Bastien Castagneyrol17,18 | Yohan Charbonnier19 | Ewa Chećko20 | David A. Coomes7 | Jonas Dahlgren21 | Seid Muhie Dawud22 | Hans De Wandeler23 | Timo Domisch24 | Leena Finér24 | Markus Fischer9 | Mariangela Fotelli25 | Arthur Gessler26 | Charlotte Grossiord27 | Virginie Guyot17,18 | Stephan Hättenschwiler28 | Hervé Jactel17,18 | Bogdan Jaroszewicz29 | François-Xavier Joly28 | Julia Koricheva30 | Aleksi Lehtonen31 | Sandra Müller32 | Bart Muys23 | Diem Nguyen33,34 | Martina Pollastrini15 | Kalliopi Radoglou35 | Karsten Raulund-Rasmussen36 | Paloma Ruiz-Benito37 | Federico Selvi15 | Jan Stenlid33 | Fernando Valladares11 | Lars Vesterdal36 | Kris Verheyen1 | Christian Wirth3,4,38 | Miguel A. Zavala37 | Michael Scherer-Lorenzen32
Abstract1. Forestecosystemfunctioninggenerallybenefitsfromhighertreespeciesrichness,butvariationwithinrichnesslevelsistypicallylarge.Thisismostlyduetothecon-trasting performances of communities with different compositions. Evidence-basedunderstandingofcompositioneffectsonforestproductivity,aswellasonmultipleotherfunctionswillenableforestmanagerstofocusontheselectionofspeciesthatmaximizefunctioning,ratherthanondiversityperse.
2. Weusedadatasetof30ecosystemfunctionsmeasuredinstandswithdifferentspeciesrichnessandcompositioninsixEuropeanforesttypes.First,wequantifiedwhetherthecompositionsthatmaximizeannualabove-groundwoodproduction(productivity)generallyalsofulfilthemultipleotherecosystemfunctions(multi-functionality).Then,wequantifiedthespeciesidentityeffectsandstrengthofin-terspecificinteractionstoidentifythe“best”and“worst”speciescompositionformultifunctionality.Finally,weevaluatedthereal-worldfrequencyofoccurrenceofbest and worst mixtures, using harmonized data from multiple national forestinventories.
CorrespondenceLanderBaetenEmail:[email protected]
Funding informationSeventhFrameworkProgramme,Grant/AwardNumber:265171
HandlingEditor:AkiraMori
734 | Journal of Applied Ecology BAETEN ET Al.
1 | INTRODUC TION
During the last25years, awealthof studiesaimed toanswer thequestion:doesplantbiodiversitymatterforthefunctioningofeco-systemsandfor theirpotential todeliverservices tohumanity? Inessence, these studies showed that changes in species diversityusually result in changes inmultiple ecosystem processes, includ-ing those related to productivity, nutrient cycling, and stability,
aswell as to trophic interactions andassociatedbiodiversity (e.g.,Isbelletal.,2017;Schulze&Mooney,1993;Tilman,Isbell,&Cowles,2014).Thesegeneral patternsweremainlyderived fromcompari-sonsofmeanvaluesofecosystemfunctioningamongdifferentlev-elsofspeciesrichness.However,withineachlevelofrichness,thereis typicallyahighvariation in functioning,mostlydue todifferentspecies compositionprovidingdifferent levelsof functioning.Thiscompositionalvariationmayhavea similarorevengreater impact
1DepartmentofEnvironment,GhentUniversity,Gontrode,Belgium;2InstituteofBiology/GeobotanyandBotanicalGarden,MartinLutherUniversityHalle-Wittenberg,Halle,Germany;3GermanCentreforIntegrativeBiodiversityResearch(iDiv)Halle-Jena-Leipzig,Leipzig,Germany;4DepartmentofSystematicBotanyandFunctionalBiodiversity,UniversityofLeipzig,Leipzig,Germany;5SenckenbergGesellschaftfürNaturforschung,BiodiversityandClimateResearchCentre,Frankfurt,Germany;6NationalBiodiversityNetworkTrust,Nottingham,UK;7ForestEcologyandConservation,DepartmentofPlantSciences,UniversityofCambridge,Cambridge,UK;8CSIROLandandWater,Floreat,WesternAustralia,Australia;9InstituteofPlantSciences,UniversityofBern,Bern,Switzerland;10Dynafor,INRA-INPT,UniversitédeToulouse,Auzeville,France;11MNCN-CSIC,Madrid,Spain;12ChairofSilviculture,FacultyofEnvironmentandNaturalResources,UniversityofFreiburg,Freiburg,Germany;13UniveristédeLorraine,AgroParisTech,INRA,UMRSilva,Nancy,France; 14FacultyofForestry,StefancelMareUniversityofSuceava,Suceava,Romania;15LaboratoryofEnvironmentalandAppliedBotany,DepartmentofAgri-FoodandEnvironmentalScience(DISPAA),UniversityofFirenze,Firenze,Italy;16InBioS—PlantandMicrobialEcology,UniversityofLiège,Liège,Belgium;17INRA,UMR1202BIOGECO,Cestas,France;18UniversityBordeaux,BIOGECO,UMR1202,Pessac,France;19LPO,leBourg,Bourrou,France;20DepartmentofForestryandForestEcology,UniversityofWarmiaandMazury,Olsztyn,Poland;21DepartmentofForestResourceManagement,SwedishUniversityofAgriculturalSciences,Umeå,Sweden;22DepartmentofForestry,CollegeofAgriculture,WolloUniversity,Dessie,Ethiopia;23DepartmentofEarthandEnvironmentalSciences,UniversityofLeuven,Leuven,Belgium;24NaturalResourcesInstituteFinland(Luke),Joensuu,Finland;25ForestResearchInstituteofThessaloniki,GreekAgriculturalOrganization-Dimitra,Vassilika,Thessaloniki,Greece;26ResearchUnitForestDynamics,SwissFederalResearchInstituteWSL,Birmensdorf,Switzerland;27EarthandEnvironmentalSciencesDivision,LosAlamosNationalLaboratory,LosAlamos,NewMexico;28CentreofEvolutionaryandFunctionalEcology,CNRS—UniversityofMontpellier—UniversityPaul-ValéryMontpellier—EPHE,Montpellier,France;29BiałowieżaGeobotanicalStation,FacultyofBiology,UniversityofWarsaw,Białowieża,Poland;30SchoolofBiologicalSciences,RoyalHollowayUniversityofLondon,Egham,UK;31NaturalResourcesInstituteFinland(Luke),Helsinki,Finland;32DepartmentofGeobotany,FacultyofBiology,UniversityofFreiburg,Freiburg,Germany;33DepartmentofForestMycologyandPlantPathology,SwedishUniversityofAgriculturalSciences,Uppsala,Sweden;34DepartmentofOrganismalBiology,UppsalaUniversity,Uppsala,Sweden;35DepartmentofForestryandManagementoftheEnvironmentandNaturalResources,DemocritusUniversityofThrace(DUTH),NeaOrestiada,Greece;36DepartmentofGeosciencesandNaturalResourceManagement,UniversityofCopenhagen,FrederiksbergC,Denmark;37GrupodeEcologíayRestauraciónForestal,DepartamentodeCienciasdelaVida,UniversidaddeAlcalá,Madrid,Spainand38Max-Planck-InstituteforBiogeochemistry,Jena,Germany
3. Themostproductivetreespeciescombinationsalsotendedtoexpressrelativelyhighmultifunctionality,althoughwefoundarelativelywiderangeofcompositionswith high- or low-averagemultifunctionality for the same level of productivity.Monoculturesweredistributedamongthehighestaswellasthelowestperform-ingcompositions.Thevariationinfunctioningbetweencompositionswasgener-allydrivenbydifferencesintheperformanceofthecomponentspeciesand,toalesserextent, byparticular interspecific interactions.Finally,we found that themostfrequentspeciescompositionsininventorydataweremonospecificstandsandthatthemostcommoncompositionsshowedbelow-averagemultifunctional-ityandproductivity.
4. Synthesis and applications.Speciesidentityandcompositioneffectsareessentialtothedevelopmentofhigh-performingproductionsystems,forinstanceinforestryandagriculture.Theythereforedeservegreatattentionintheanalysisanddesignoffunctionalbiodiversitystudiesiftheaimistoinformecosystemmanagement.Amanagement focus on tree productivity does not necessarily trade-off againstotherecosystemfunctions;highproductivityandmultifunctionalitycanbecom-binedwithaninformedselectionoftreespeciesandspeciescombinations.
K E Y W O R D S
ecosystemmultifunctionality,forestmanagement,forestry,FunDivEUROPE,overyielding,productivity,speciesinteractions,treespeciesmixtures
| 735Journal of Applied EcologyBAETEN ET Al.
on ecosystem functioning compared with variation in diversity(Hectoretal.,2011;Ratcliffeetal.,2017),butitisoftenoverlookedorevenconsideredtobeunwantednoise.Speciesdifferstronglyintheir functionaleffects,meaningthatcompositionscontainingdif-ferentspeciesprovidedifferentlevelsoffunction(“speciesidentityeffect”;Kirwanetal.,2009). Inaddition, functionaleffectsofmix-turesmaydifferfromtheexpectedeffectsoftheindividualspeciesmonoculturesduetointerspecificinteractions(“speciesinteractioneffect”),whichcanbesynergistic,neutral,orantagonisticdependingontheparticularspeciesinvolved.Ifwecanidentifywhichidentityandinteractioneffectsprovidehighestfunction,thenwecouldde-liberatelyselectcertainspeciescombinationsthatoptimizeoneormultipleecosystemfunctions(Storkeyetal.,2015).Inthiscontext,biodiversity–ecosystemfunctioningresearchcouldhelptodevelophigh-performingproductionsystems,forinstance,inmultifunctionallow-inputagriculture(Barotetal.,2017),incarbonplantings(Hulveyetal., 2013) and in the context of sustainable forestmanagement(Mori,Lertzman,&Gustafsson,2017).
By favouring different tree species throughmanagement (e.g.,selective thinning), foresters have been following this approachforcenturies.However,forestryhastraditionallyfocusedonwoodproduction as the main management goal, rather than on the si-multaneous provision ofmultiple ecosystem functions or services(ecosystemfunctionorservice“multifunctionality”;Manningetal.,2018).Itisoftenassumedthatafocusonwoodproductionwill,quasi automatically,fulfilallotherfunctionsaswell.Thisreasoningevenhas its own name in German forestry (the “Kielwassertheorie” or“waketheory”;Rupf,1961),wherehabitat,regulation,andrecreationfunctionsareassumedtobeboostedinthe“wake”ofusefunctions,thatis,woodproduction.Yet,thispremisehasbeenchallengedbystudiesshowingtrade-offsbetweendifferentfunctionsorservices.Forexample,a focusontreebiomassproductionwasfoundtobedetrimental for dead wood occurrence, bilberry production, andfoodforgameinborealandtemperateproductionforests(Gamfeldtetal., 2013). In general, species effects ondifferent functions arenotwellcorrelated,sothatno“super-species”fulfilsmanyfunctionsatthesametimeandunderallconditions(vanderPlasetal.,2016).In sum, there is a need for evidence-basedunderstandingof howdifferent tree species compositions promote multiple ecosystemfunctionsandservices, including,butnot restricted to,woodpro-duction. Such insightswill help tobridge thegapbetween funda-mentalbiodiversity-functioningtheoryandecosystemmanagementandcould,forinstance,betterinformforestmanagersaboutwhichtreesshouldbeplantedtogetherinordertomaximizeforestmulti-functionalitywithinstands.
Researchonbiodiversity–ecosystemfunctioningrelationships,aswellasontreespeciesmixtureeffectsinforestry(reviewedinPretzsch,Forrester,&Bauhus,2017), still often relieson single-site experiments or case studies, limiting our capacity for syn-thesisandgeneralizationacrossspatialandtemporalscales.TheFunDivEUROPE exploratory platform was established as a net-workof researchplots in sixEuropean forest types, selected todifferintreespeciesrichnessanddifferentspeciescompositions
(Baetenetal.,2013).Theplatformprovidedacommonhypothesis-drivendesign indifferentgeographical locations,usedstandard-ized methodology and measurement protocols and coordinateddata acquisition and management. Using data on 30 ecosystemfunctionsmeasuredinthisplatform,wecanperformanin-depthanalysis of tree composition effects on forest ecosystemmulti-functionality.Weaimto(i)assesstowhatdegreeamanagementfocusontreeproductivityalsoboostsotherecosystemfunctionsor whether there are trade-offs between production and otherfunctions;(ii)quantifytheindividualspecieseffectsandstrengthof interactions among particular species and species groups toidentify the “best” and “worst” species compositions for multi-functionality; and (iii) evaluate the frequency of occurrence ofbestandworstmixturesbasedonNationalForestInventories.Wehypothesizethat(i)treeproductivityisnotstronglypositivelyre-latedwithecosystemmultifunctionality,refutingthewaketheory;(ii) interspecific interactions can explain ecosystem functioningbetter than species identity effects alone, and that these inter-actionsarespeciesspecific;and(iii)treecompositionssupportinghigh ecosystem multifunctionality are rare in European forestsduetothehistoricalfocusonproductionforests.
2 | MATERIAL S AND METHODS
2.1 | FunDivEUROPE exploratory platform design
The FunDivEUROPE exploratory platform is a coordinated net-work of 209 forest plots in six European regions, covering agradient of different climates and forest types (see SupportingInformationAppendixS1).Itwasestablishedin2011tostudytheeffect of tree diversity on ecosystemmultifunctionality (http://project.fundiveurope.eu/).ThefieldsitesincludeborealforestsinFinland,hemi-borealforestsinPoland,beechforestsinGermany,mountainousbeechforestsinRomania,thermophilousdeciduousforestsinItaly,andMediterraneanmixedforestsinSpain.Ineachforesttype,plotswith locallydominantandeconomically impor-tanttreespecieswereselectedtocoverarangeinspeciesrichnessfrom1to3inboreal(numberofplots:28),1to4inmountainousbeech(28),beech(38),andMediterraneanmixed(36),and1to5inthermophilousdeciduous(36)andhemi-boreal(43)(SupportingInformation TableS1.1). Each richness level was replicated withdifferentspeciescompositions.Furthermore,thetreespecieshadsimilar abundances inmixtures (high evenness), all specieswererepresentedinallspeciesrichnesslevels,andnoneofthespecieswaspresentineveryplotsothatspeciesidentityanddiversityef-fectscouldbeseparated.Thestudyplotswerelocatedinmatureforestsstandsandsharedsimilarenvironmentalconditionswithinforesttypes(e.g.,geology,soiltype,topography),sothatcovaria-tionbetweenthesefactorsandspeciesrichnesslevelswasmini-mized.Thus,thediversitygradientmainlyresultedfromhistoricalmanagementor stochastic events.Moredetails about the studysites, the selectionprocedure, andplot-level information canbefoundinBaetenetal.(2013).
736 | Journal of Applied Ecology BAETEN ET Al.
2.2 | Ecosystem property and function measurements
Weusedplot-levelmeasurementsof30ecosystemproperties,func-tions,orserviceproxies,whichforsimplicitywerefertoas“func-tions”orpropertieshereafter (Supporting InformationTableS1.2).Theseincludethesetof26functionsanalysedinapreviousstudylooking at the relative importanceof compositionvs. diversity ef-fects (Ratcliffe etal., 2017). Four additional functions, represent-ingdiversitymeasurementsoffourtaxonomicgroups,wereaddedto thedataset: bat, bird, earthworm, andunderstoreyplantdiver-sity.As ameasureof treeproductivity,weused themean annualabove-groundwoodproductionestimatedfromwoodcores(Jucker,Bouriaud,Avacaritei,&Coomes,2014).Toaidintheinterpretation,thefunctionswereaprioriclassifiedintosixgroupsreflectingbasicecological processes (Supporting Information TableS1.2): nutrientandcarboncycling-relateddrivers(e.g.,earthwormbiomass,micro-bialbiomass),nutrientcycling-relatedprocesses(e.g.,litterdecompo-sition,nitrogenresorptionefficiency),primaryproduction(includingnot only tree productivity, but also photosynthetic efficiency andtreebiomass),regeneration(e.g.,treeseedlingregeneration,saplinggrowth),resistancetodisturbance(e.g.,resistancetodrought,resist-ancetoinsectdamage),andthevalueoftheforeststandsashabitatforotherspecies (e.g.,batandbirddiversity).Amajor strengthoftheFunDivEUROPEprojectwasthegeneralphilosophytomeasureallecosystemfunctionsinallplots,followingthesameprotocolbythesameobserversacrossthesixforesttypes.Measurementsarethusdirectlycomparableacrossplotsandshowhighcoverage;24functionsweremeasuredinatleast207ofthe209plots.Detailsonthemeasurementsofthevariousfunctionscanbefoundinprevioussynthesispapersof theFunDivEUROPEproject (e.g., vanderPlasetal.,2016;Ratcliffeetal.,2017).
2.3 | National forest inventory data
WithintheFunDivEUROPEproject,wecompiledharmonizedforestplotdatafromthenationalforestinventoriesofFinland,Sweden,Germany,Belgium (Wallonia), andSpain (fordetails seeRatcliffeetal., 2016).These inventories included three forest types fromtheexploratoryplatform:borealforest,beech(-dominated)forest,andMediterraneanmixedforest(whichcomprisedMediterraneanconiferous,broadleavedevergreen,andthermophilousdeciduousforest).Determinationof the forest typewasbasedon theEEATechnicalReport9 (Barbati,Corona,&Marchetti,2017). Ineachinventory, we used the twomost recent surveys and extractedbasalarea(BA,m²/ha)foralltreeswithadiameteratbreastheightofmorethan10cm.Plotswithsinglemeasurementsoranyindi-cationof harvest activities between surveyswereomitted fromthe dataset. For each of the remaining plots, we calculated theproportionalBApertreespecies.Treespeciesnameswereharmo-nizedfollowingtheAtlasFloraeEuropaeae.Inordertoidentifythespeciescompositionofaplot,weadoptedthefollowingapproach:onlyspecieswithaBAexceeding10%wereconsideredandonly
plots inwhich the summedproportionof all component speciesexceeded90%wereincluded.Plotsthatdidnotmeetthesecrite-riawerediscardedfromthedataset.ThisapproachisinagreementwiththeselectioncriteriaoftheFunDivEUROPEexploratoryplat-form.Furthermore,weonlyretainedtheplotswithcompositionsthatcouldbeassignedtooneofthethreeforesttypesmentionedabove.Nodistinctionwasmadebetweenplantedandspontane-ously regeneratedstands.Our finaldataset included64.8% (bo-real), 22.3% (beech), and 70.8% (Mediterranean mixed) of theavailableNFIplots.
2.4 | Data analyses
2.4.1 | Quantifying multifunctionality and its relationship with productivity across different species compositions
Wequantifiedthemultifunctionalityofeachtreespeciescomposi-tionwithamodel-basedapproach. Ineachplot,wehaveavalueforeachof the30functions.Theseestimatesweremodelledto-gether in a hierarchicalmeta-analyticmodelwith group-level ef-fects for plot identity (209 plots) and species composition (103compositions).We considered species combinations occurring inmultipleforesttypesasdifferentcompositions,becausethesamespecies combinationmay have different functioningwhen grow-ing on different soils or in different climates and we wanted toaccount for the fact that the same compositionmay behave dif-ferentlyamongforesttypes. Inaddition,compositionswithinthesame forest typewere related to each other because theyweremeasuredmorecloselytogetherintimeandspace.However,only8outof92uniquespeciescompositionsoccurredinmultipleforesttypes:sixwererepresentedintwoforesttypesandmonoculturesofPinus sylvestris and Picea abieswerepresent in threeand fourtypes,respectively.
The estimated effects of composition from the hierarchicalmodelwereusedhereasmeasureofmultifunctionalityforagiventreespeciescomposition.Theeffectquantifiesthedegreetowhichthe functioning of a particular composition deviates from the av-erage, takingall functions intoaccount.Positiveandnegativeval-uesexpressabove-averageandbelow-average functioningof thatspecies combination, respectively. An alternative single thresholdapproach (Byrnes etal., 2014) provided a very similarmeasure ofmultifunctionality; so,weexpectqualitativelysimilar resultswhenusing alternative measures (Supporting Information FigureS2.1).Themodel-basedapproachwaspreferredherebecause itdirectlyquantifiesthedependencyoffunctioningoncomposition(withouttheneedtoderiveametricfirst)andallowsustoextendtheanaly-sestodiversity-interactionmodels(seeSection2.4.2).AfullmodeldescriptionisgiveninSupportingInformationAppendixS2andad-ditionalsensitivityanalysesareprovidedinSupportingInformationAppendixS4(e.g.,reducingthenumberoffunctionstocalculatethemultifunctionalitymeasure,eitherrandomlyorbyecosystemfunc-tiongroup).
| 737Journal of Applied EcologyBAETEN ET Al.
Werelatedthemultifunctionalitytothemeanproductivityofeachcompositionwithalinearregressionmodel,totestwhetherselecting composition for high productivity also ensures highmultifunctionality. In this analysis, we quantified the measureofmultifunctionality after excluding productivity, that is, mul-tifunctionality was calculated with 29 functions. This analysiswasfirstperformedonthefulldatasetandthenforeachforesttypeseparately.Differencesinproductivityandmultifunctional-itybetweencompositionswithdifferentspeciesrichnessvalues(monoculturevs.mixed)ordifferent leafphenologies (pureev-ergreen,puredeciduous,ormixed)weretestedwithananalysisofvariance.
2.4.2 | Diversity- interaction models
To identify the individual species and pairs of species that in-creased functioning, we used a diversity-interaction modellingframework (Kirwanetal.,2009).This testshow theabundanceofindividualtreespecies,andtheinteractionsbetweenthem,af-fectecosystemfunctioning.Theapproachusesalinearmodeloftheformf= ID+DE+BA+ residual,wherefistheestimateoffunc-tioninginaplot,IDisthespeciesidentityeffects,DEisthediver-sityeffects,BA is theeffectofvariation inplot-levelbasalarea(averagecentredtozerowithinforesttypes),andaresidualistheerrorterm.Thespeciesidentityeffectsequaltheaveragemono-culture performances, weighted by the species’ relative abun-dances. The diversity effects result from species interactions,which causesmixture functioning to differ from that expectedfrom monoculture functioning. Kirwan etal. (2009) proposedalternative patterns of interactions based on different ecologi-calassumptions,correspondingtodifferentformulationsofthediversityeffectsterm.SeeSupportingInformationAppendixS2for a full model description and explanation of the alternativediversityterms.
Weconfrontedfivealternativemodelswiththedata.Afirstnullmodelassumesthatallspeciesidentityeffectsareequal(model0),while a second assumes thatmonoculture functioning differs andonly the relative abundances of the species influence function-ing in mixtures (identity-effect model; model 1). Three additionalmodelscombinetheidentityeffectwithdifferentdiversityeffects,corresponding to the alternative types of species interactions: apairwise interactionseffect (model2), an additive species-specificcontributionseffect (model3),orafunctional-groupeffect (model4).The importanceofthedifferenttypesof interactionswasthenexploredbycomparing themodelsdiffering in theirecological as-sumptions(Kirwanetal.,2009).WeusedAICvaluesandlikelihoodratioteststocomparemodels.First,wefittedthealternativemodelsforeachecosystemfunctionandforesttypeseparately.Second,wemodelled the 30 functions together, using a similarmeta-analyticmodel described above (Quantifying multifunctionality), replacingthe composition effect with the identity and diversity effects ofthediversity-interactionmodels.Thevaluesforeachfunctionwere normalizedbeforemodelling.
2.4.3 | Relationship between multifunctionality and frequency of occurrence of tree species compositions
We calculated the frequency of occurrence of all tree speciescompositions for each of the three forest types (boreal, beech,andMediterraneanmixedforest)fromthenationalforestinven-torydata.So,foreachofthecompositionsofthesethreeforesttypesstudiedintheexploratoryplatform,wehaveameasureoftheirfrequencyamongallothercompositionsinthesameforesttype.Wedrewgraphsrankingcompositionsbyfrequency,multi-functionality,andproductivitytoexplorewhethercompositionssupportinghighecosystemmultifunctionalitywererareinagivenforesttype.Weareawarethatthespeciescombinationsencoun-tered in the exploratoriesmay have different effects onmulti-functionality in the different contexts (e.g., climates, soil typesor stand development stages) encountered in the inventories(Ratcliffeetal.,2017).Nevertheless,ourassessmentprovidesanindicationofwhethercompositionslikelytopromotehighmulti-functionalityoccurmoreoftenintheinventoriesthanthosewithlowmultifunctionality.
3 | RESULTS
3.1 | Relationship between productivity and ecosystem multifunctionality
Across all plots, the multifunctionality (excluding productivity) oftreespeciescompositionswaspositively relatedto theirmeanpro-ductivity (Figure1;slope=0.028,p < 0.001,R²=0.22),althoughforagivenlevelofproductivitytherewasaconsiderablerangeinmul-tifunctionality between compositions.Within the forest types, theproductivity–multifunctionalityrelationshipwassignificantlypositiveinthreetypes(beech,thermophilousdeciduous,andMediterraneanmixed)andpositivebutnon-significantinthethreeothers(SupportingInformation FigureS3.1). Patterns at the level of individual ecosys-tem functionswere consistent: inbeech, thermophilousdeciduous,andMediterraneanmixedforest,themostproductivecompositionsalsohadabove-average(withinregion)valuesofthemajorityoftheotherfunctions(>20outof29functions),whereaslessthanhalfofthefunctionsexceededtheaverageintheleastproductivecompositions(Supporting InformationFiguresS3.2andS3.3).Monocultureswerenotconsistentlydifferentfrommixtures:theyweredistributedamongthehighest aswell as the lowestperforming compositions, both interms of productivity (F=0.62, p = 0.43) and multifunctionality(F=2.19,p = 0.14). Similarly, the leafphenology (evergreen,decidu-ous,ormixed)wasnotimportantinexplainingdifferencesinproduc-tivity(F=1.83,p = 0.17)ormultifunctionality(F=1.09,p = 0.34).
Sensitivity analyses showed that the tree productivity–multi-functionalityrelationshipdidnotchangewhenweclassifiedallspe-cies combinationsoccurring indifferent forest types as the same,for example, rather than considering Picea abies monocultures asbeingfourseparatecompositionsbecausetheyoccurredinfourfor-esttypes,weregroupedthemasasinglecomposition (Supporting
738 | Journal of Applied Ecology BAETEN ET Al.
Information FigureS4.1). While the productivity–multifunctional-ity relationship remained the same ifwe randomlyexcluded func-tionsfromourmultifunctionalitymeasure (Supporting InformationFigureS4.2), when we excluded particular ecosystem functiongroups then the strength of the relationship altered (SupportingInformationFigureS4.3).For instance,excludingall functionssup-portingprimaryproductionweakened theproductivity–multifunc-tionalityrelationship,however,itremainedsignificantlypositive.
3.2 | Identifying the best mixtures
Looking at individual functions, diversity-interaction models showedthat pairwise species interactions often influenced functioning, posi-tivelyaswellasanegatively(Figure2).Interactionsindicate,forparticu-larspeciespairs,whethergrowingthetwospeciesinamixtureincreasedor decreased functioning compared with growing them in separatemonocultures.Ecosystemfunctiongroupsdidnotshowconsistentpat-terns:production-relatedfunctionsweremoreoftenfoundtobenefitfrommixing(26positivevs.11negativeinteractioneffects)andpositive
interactionsalsooutnumberednegativeinteractionsinresistance-andregeneration-related functions (27 vs. 17 and 10 vs. 3, respectively).Interactions tended to be positive in thermophilous deciduous andMediterraneanmixedandnegativeinborealforest.Resultsfortheindi-vidualfunctionsareshowninSupportingInformationFigureS3.4.
When multifunctionality was modelled with all 30 functionstogether, including productivity, we often found tree species tohaveverydifferenteffectsonfunctioning (identity-effectsmodel;Supporting Information FigureS3.5). Furthermore, functioninglevels generally also increasedwith plot-level basal area.We alsolookedatvariation infunctioningacrossforesttypesforthesmallnumber of composition present inmultiple types.We found thatPicea abies had higher functioning, compared with the averagemonoculture, in hemi-boreal and mountainous beech forest, butbelow-averagefunctioninginborealandbeechforests(SupportingInformationFigureS3.5).Pinus sylvestrishadhigher(Mediterraneanmixed),lower(boreal),oraverage(hemi-boreal)monocultureperfor-mance.Incontrast,monoculturesofQuercus robur/petraeatendedtohaveconsistentlylowermultifunctionalitythanothermonocultures,acrossforesttypes(hemi-boreal,beech,thermophilousdeciduous).
Species interactions were important in explaining multi-functionality in all forest types except formountainousbeech(likelihood ratio tests of models with interaction effects vs.identity-effects models; p < 0.05). We found that mixing ev-ergreen and deciduous species reduced functioning in boreal(functional group vs. identity model; p = 0.029) but increasedfunctioning inhemi-boreal forest (p = 0.025). Inboreal forests,the negative effectwasmainly because of an antagonistic in-teraction between Picea abies and Betula pendula leading tolower multifunctionality than expected based on their mono-culture functioning. In beech, thermophilous deciduous andMediterraneanmixedforest,therewasnosuchfunctionalgroupeffect, as here the species interacted similarlywith all others,illustratingthatthemaineffectofmixingwasthecontrastbe-tween intra- and interspecific interactions (additive contribu-tionsvs.identitymodel;p < 0.05).
Thelistoftopfivecompositionsineachforesttypeintermsoftheir multifunctionality (Table1), reflected this: only 6 out of thetotal 28 best compositions listed in Table1 were monocultures.Someof thebestcompositions includedup to fourspeciesand insometypesnoneofthefivebestcompositionsweremonocultures(hemi-boreal and thermophilous deciduous). Finally, the composi-tionswiththehighestmultifunctionalitywerealsonotdominatedbypureevergreenordeciduouscompositionsand15outofthe22mul-tispeciescompositionsweremixturesofdeciduousandevergreenspecies.Thespeciescombinationswiththehighestmultifunctional-itywerealsoamongthemostproductiveones.
3.3 | Frequency of the best mixtures in forest inventory data
Thespeciescompositionsstudiedintheexploratoryplatformwerealsowellrepresentedinthenationalforestinventoriesofthethree
F IGURE 1 RelationshipbetweenthetreeproductivityandmultifunctionalityofdifferenttreespeciescompositionsacrosssixEuropeanforesttypes.Pointsshowtheperformanceofindividualcompositions(N=103):filledpointsrepresentmonoculturesandcolouringrepresentsfunctionalcompositionintermsofleafphenology(onlydeciduousspecies,onlyevergreenspecies,oramixtureofboth).Thefulllineshowsthefitofalinearmodel,withthedashedlinesdelimitingthe95%confidenceinterval.Productivitycorrespondstotheannualabove-groundwoodproductionandwasnormalizedwithinforesttypestoallowforacross-regionalcomparison;absolutemeanproductivityvaluesarepresentedinSupportingInformationFigureS3.1.Themultifunctionalityexpressesthedegreetowhichthefunctioningofaparticularcompositiondeviatesfromtheaverage,takingallfunctionsintoaccount(positivevaluesindicateabove-averageperformance).Forthisanalysis,theproductivitywasexcludedfromthemultifunctionalitymeasure
–0.1
0.0
0.1
–2 –1 0 1 2Normalized productivity
Mul
tifun
ctio
nalit
yAll deciduous
All evergreen
Deciduous−evergreen
Mixture
Mono
| 739Journal of Applied EcologyBAETEN ET Al.
studiedforest types (boreal,beech,andMediterraneanmixedfor-est)(Figure3).Inallthreetypes,themostwidelyoccurringtreespe-ciescompositionsweremonospecificstands.Furthermore,themostfrequentcompositionshadbelow-averagemultifunctionalityscores,thatis,belowzero.Especiallyinbeechforest,thecompositionswithabove-average multifunctionality were rare (frequency<1%). Wefoundessentiallythesamepatternwhenfocussingonproductivityratherthanmultifunctionality(SupportingInformationFigureS3.6):themostproductivecompositionswerenotthemostfrequentones.
4 | DISCUSSION
Despitetheimportanceofspeciescompositioninexplainingvari-ationinecosystemfunctioning(Hectoretal.,2011;Ratcliffeetal.,2017),speciesidentityeffectsaregenerallynotthefocusofbiodi-versityandecosystemfunctioningstudies,wheretheyareinsteadtreatedasanuisancevariabletobeaccountedfor.Here,weaimed
tounpackthevariationinfunctioningbetweencompositionsandtounderstandwhichparticularspeciesorspeciespairssustainedthehighestmultifunctionality.Our findings show that itmattersconsiderablywhich particular combinations are promotedwithinagivenrichnesslevel.Thisiscriticalfromanappliedperspective,as forestmanagersaremuchmore likely to focusonspecies se-lection(e.g.,whenreplantingafteraregenerationcut)ratherthandiversityperse.
4.1 | Managing for productivity can also promote multifunctionality
Afundamentalmanagementgoalinforestryistoproducewood,andso,manystudieslookingatthefunctionalimportanceofmixingtreespecies focused on tree productivity. There is evidence that treespeciesdiversityincreasestheproductivityofforestsglobally(Liangetal.,2016;Piotto,2008).Inclosedcanopyforests,thisisprimarilyduetomoreefficientlightusewhenspecieswithcontrastingcanopy
F IGURE 2 Synthesisoftreespeciesinteractioneffectsonecosystemfunctioning(30functions)insixEuropeanforesttypes.Foreachfunction,pairwisespeciesinteractionmodelswerefittedtoquantifythedegreetowhichtreespeciesinteractionscausemixtureperformancetodifferfromthatexpectedfromthemonoculturespeciesperformances.Foreachspeciespair,thegraphshowsthetotalnumberofpositive(andnegative)effects,indicatingthenumberoftimesthespeciesmixtureisprovidingmore(orless)functioningthanthecorrespondingmonocultures(onlyeffectswithp < 0.1werecounted).Functionsweregroupedintoaprioriclassestoaidintheinterpretation;seemethodsandSupportingInformationTableS1.2.Forresultsforsinglefunctions,seeSupportingInformationFigureS3.4.Notethatthegraphcompareswithintreespeciescombinations(performanceofmixturesvs.themonoculturesoftwoparticularspecies)anddoesnotallowadirectcomparisonbetweencompositions,becausethespeciesidentityeffectswerenotaccountedforinthisanalysis.FullspeciesnamesaregivenbelowTable1
−6
−4
−2
0
2
4
6
Pic
ea:P
inus
Pin
us:B
etul
a
Pic
ea:B
etul
a
Num
ber o
f fun
ctio
ns
Bet
ula:
Car
pinu
s
Bet
ula:
Que
rcus
Pic
ea:B
etul
a
Pin
us:Q
uerc
us
Pic
ea:C
arpi
nus
Pic
ea:Q
uerc
us
Pin
us:B
etul
a
Pin
us:C
arpi
nus
Car
pinu
s:Q
uerc
us
Ace
r:Que
rcus
Pic
ea:Q
uerc
us
Fagu
s:Q
uerc
us
Pic
ea:F
agus
Frax
inus
:Que
rcus
Pic
ea:A
cer
Fagu
s:Fr
axin
us
Pic
ea:F
raxi
nus
Ace
r:Fag
us
Ace
r:Fra
xinu
s
−6
−4
−2
0
2
4
6
Abi
es:A
cer
Abi
es:F
agus
Ace
r:Fag
us
Pic
ea:A
cer
Abi
es:P
icea
Pic
ea:F
agus
Num
ber o
f fun
ctio
ns
Ost
rya:
Que
rcus
_il
Cas
tane
a:Q
uerc
us_c
e
Ost
rya:
Que
rcus
_rp
Que
rcus
_rp:
Que
rcus
_ce
Cas
tane
a:Q
uerc
us_i
l
Que
rcus
_ce:
Que
rcus
_il
Ost
rya:
Que
rcus
_ce
Cas
tane
a:Q
uerc
us_r
p
Cas
tane
a:O
stry
a
Pin
us_n
i:Que
rcus
_il
Que
rcus
_fa:
Que
rcus
_il
Pin
us_n
i:Que
rcus
_fa
Pin
us_s
y:Q
uerc
us_i
l
Pin
us_s
y:Q
uerc
us_f
a
Pin
us_s
y:P
inus
_ni
Boreal Hemi−boreal Beech
Mountainous beech Thermophilous deciduous Mediterranean mixed
Nutrient and C cycling related driver
Nutrient cycling related process
Primary production
Regeneration
Resistance to disturbance
Habitat for species
740 | Journal of Applied Ecology BAETEN ET Al.
traitsco-occur(Fichtneretal.,2017;Pretzsch,2014;Zhang,Chen,&Reich,2012).Theseinsightsproviderelevantinformationformakinginformedtreespecieschoices in forestrybut theydonot indicatewhetherselectingspecies tomaximizehighproductivityalsoben-efits multiple other functions.While trade-offs between produc-tivityandotherfunctionshavepreviouslybeenreported inborealforests(Gamfeldtetal.,2013),ourstudyevaluatedagreaternumberoffunctionsacrossabroadrangeofforesttypes,andshowedthatthemostproductivetreespeciescombinationsalsotendtoproviderelatively highmultifunctionality. In the context of recent discus-sions about the sensitivity of multifunctionality measures to thenumberandidentityoftheircomponentfunctions(e.g.,Gamfeldt&Roger,2017;Meyeretal.,2018),weshowedthatourfindingswererobust when randomly reducing the number of functions consid-ered.Deletingparticulargroupsoffunctionsdidchangethestrengthoftherelationshipbetweenproductivityandmultifunctionality,al-though itwasalwayspositive.Sincepreviousanalysesofourdatashowedfewtrade-offsbetweenarangeofmultifunctionalitymeas-uresreflectingalternativestakeholderobjectives(sensuAllanetal.,2015; van der Plas etal., 2018), changing our multifunctionalitymeasuretorepresentspecificmanagementscenario’sisalsounlikelytochangetheconclusions.
Rankingthespeciescompositionswithinforesttypes,basedoneitherproductivityormultifunctionality,resultedinasimilarsetofbestcompositions (Table1,Supporting InformationFigureS3.1).Anotablepatterntoemergefromouranalysis isthatforfourofthesix forest typeswe identifiedat leastone species that repeatedlyoccurred across the best compositions that characterize that par-ticular forest type (hemi-boreal:Picea abies, beech:Fraxinus excel-sior, thermophilousdeciduous:Quercus ilex and Quercus cerris, andMediterranean mixed: Pinus sylvestris) (Table1). In beech forests,
thecombinationF. excelsior–Acer pseudoplatanusevenappearedfourtimes inthistopfive.Atthesametime,mixturescontainingtheseparticularspecieswerenotalwaysthemostproductiveones.Thisinformation may already provide useful empirical evidence whendecidingamongseveralmanagementoptions,suchastheselection(orexclusion)ofspecieswhenplantingorregeneratingnewstands.
Wedonotproposetousetreeproductivityasanintegratedmea-sureofforestperformanceinageneralwaybecauseforthesamelevelofproductivitywefoundarelativelywiderangeofcompositionswithhigh or low average performance across functions. For instance, inMediterraneanmixedforest,monoculturesofPinus sylvestris and Pinus nigrahadnearlythesameproductivity,butvariedstronglyinmultifunc-tionality.Furthermore,themostproductivecompositionshadabove-average values formany, but certainly not all functions (SupportingInformationFiguresS3.2andS3.3).Therelativeimportanceoftheseexisting trade-offs between individual ecosystem functions need tobeevaluatedbasedonsocio-ecologicalperspectives,includingthede-siredmanagementgoalsandland-useschemes(Morietal.,2017),andinthisrespect,ourdatacanhelpinformthesedecisions.Thus,ourre-sultsshouldnotbeusedasageneralconfirmationofthe“waketheory”thatallforestfunctionsareautomaticallyfulfilledbyafocusontimberproductiononly.Rather,weconcludethatamanagementfocusonpro-ductivitydoesnotnecessarilytrade-offagainstotherecosystemfunc-tions and high productivity andmultifunctionality can be combinedwithaninformedselectionoftreespeciescombinations.
4.2 | The identity of co- occurring tree species matters
Wefound that thevariation in functioningbetweencompositionswasgenerallydrivenbyidentifyeffectsand,toa lesserextent,by
TABLE 1 Topfivespeciescompositionforeachforesttype,rankedaccordingtodecreasingmultifunctionality(fromthetopdown).Compositionswithanasteriskwerealsoidentifiedamongthebestfiveincaserankingwasdonebasedonproductivityonly.Underlinedspeciesareevergreentrees.Thenumberofdifferentcompositionsstudiedineachtypeisgiveninbrackets.Inborealforest,onlysevencompositionswerestudied,sothatonlythreeperformedaboveaverage
Boreal (7) Hemi- boreal (25) Beech (18)Mountainous beech (14)
Thermophilous deciduous (27)
Mediterranean mixed (12)
*P. abies *C. betulus, P. abies A. pseudoplatanus, F. sylvatica, F. excelsior
P. abies *C. sativa, O. carpinifolia, Q. cerris, Q. ilex
*P. nigra, P. sylvestris
B. pendula B. pendula, C. betulus, P. abies, Q. robur
A. pseudoplatanus, F. sylvatica, F. excelsior, Q. petraea
A. alba, A. pseudoplatanus, F. sylvatica, P. abies
*Q. cerris, Q. ilex *P. sylvestris, Q. faginea
*B. pendula, P. abies, P. sylvestris
*P. abies, P. sylvestris *F. excelsior *F. sylvatica, P. abies O. carpinifolia, Q. cerris, Q. ilex
*P. sylvestris
*C. betulus, P. abies, Q. robur
*A. pseudoplatanus, F. excelsior, Q. petraea
*A. alba *C. sativa, Q. cerris *P. nigra, P. sylvestris, Q. faginea
B. pendula, P. abies, P. sylvestris, Q. robur
*A. pseudoplatanus, F. sylvatica, F. excelsior, P. abies
A. pseudoplatanus, F. sylvatica
C. sativa, O. carpinifolia, Q. ilex, Q. petraea
*P. nigra, P. sylvestris, Q. faginea, Q. ilex
Notes.Fullspeciesnames.Coniferousspecies:Abies alba,Picea abies,Pinus nigra,Pinus sylvestris. Broadleaved species: Acer pseudoplatanus,Betula pen-dula,Carpinus betulus,Castanea sativa,Fagus sylvatica,Fraxinus excelsior,Ostrya carpinifolia,Quercus robur,Quercus petraea,Quercus cerris,Quercus faginea,Quercus ilex.
| 741Journal of Applied EcologyBAETEN ET Al.
F IGURE 3 Frequencyofoccurrenceofparticulartreespeciescompositionsinnationalforestinventorydataforborealforests,beechforest,andMediterraneanmixedforests.GreybarsindicatethecompositionsthatwerealsostudiedinthecorrespondingforesttypesintheFunDivEUROPEexploratoryplatform;thewhitebarsrepresentcompositionsthatwerenotincludedintheexploratoryplatform.Thecolouredcirclesindicatethedegreeofmultifunctionalityofthecompositionsbasedontheestimatesintheexploratoryplatform(soonlyforgreybars).Thismultifunctionalityexpressesthedegreetowhichthefunctioningofaparticularcompositiondeviatesfromtheaverage,takingall30functionsintoaccount(positivevaluesindicateabove-averageperformance).Thedottedlinesindicateathresholdfrequencyof0.01belowwhichrarecombinationsoftreespeciesarenotshown,unlesstheywerestudiedintheexploratoryplatform
742 | Journal of Applied Ecology BAETEN ET Al.
particularinterspecificinteractions.Intryingtoexplainwhatmakesupahigh-performingspeciescombination,welookedatdifferencesbetweenpuredeciduous,pureevergreen,andmixeddeciduous–ev-ergreenmixtures.While heterogeneityof canopy traits related tolightcaptureanduse,includingleafphenology,isoftenfoundtoin-creaseproductivity(Jucker,Bouriaud,Avacaritei,Dǎnilǎ,etal.,2014;Lu,Mohren,denOuden,Goudiaby,&Sterck,2016;Zhang,Chen,&Taylor, 2015), mixing species from these broad functional groupsdidnotalways increasemultifunctionality.Manyoftheecosystempropertiesincludedherearenotdirectlyrelatedtolightavailability(e.g.,nutrientcyclingrelateddriversorprocesses;Rothe&Binkley,2001)andour findings showthat themechanisms responsible foroveryieldingofmixtures(foranoverview,seeForrester&Bauhus,2016),donotnecessarilyincreaseotherfunctions.Moregenerally,whilestudiesonidentityeffectshavemostlylookedatcommunity-weightedmeansoftraitsasawayofgeneralizingresults (Ratcliffeetal.,2016),suchanapproachisnotthebestchoicewhensearch-ingforhigh-performingtreespeciescompositionsbecausewelacktheorylinkingtraitstomultifunctionality.Inaddition,manyspeciesinteractionsarenot related tocommonlymeasured traits (suchaspathogensorherbivory),anditwouldbedifficulttotranslatetrait-basedidentityeffectsintoconcretemanagementdecisionswithrealspecies.
Our studywas designed using a pool of regionally abundantandeconomicallyimportanttreespecies(Baetenetal.,2013)andthereforeprovidescomprehensivedataonmultifunctionalityval-uesinmanyrelevantspeciescombinations.Anextstepwouldbetoexplorewhenandwherespecificcombinationsofinterestprovidemaximummultifunctionality,sothatmanagerscanmakeinformeddecisions as towhich combinationsof species to favouron theirland.Thisrequiresdeterminingthevariation inmultifunctionalityforparticularspeciescompositionsacrossdifferentenvironments(e.g.,climates,soil types)andtryingtoexplaintheprincipalenvi-ronmentaldriversofthisvariation.Anothercomprehensiveanaly-sisinourstudyplotsshowedthattreediversityeffectsonvariousecosystem functions are highly context dependent: stronger di-versityeffectsonmultifunctionalitywerefoundinforesttypesindrierclimates,withlongergrowingseasons,andmorefunctionallydiverse tree speciespools (Ratcliffeetal.,2017).A similaranaly-sis of the context dependency of species composition effects isnotstraightforwardbecausecompositionsarenoteasilyreplicatedin very different environments and forest types, unlike diver-sity gradients that can be replicatedwith very different speciespools. Focusing on productivity, Pretzsch etal. (2010, 2013) al-readyshowedthatspecifictwo-speciescombinations(oak–beech,spruce–beech)changefromoveryielding,duetofacilitation,toun-deryielding, driven by competitive interference, along a gradientfrompoortorichsoilsacrosscentralEurope.Focusingonmultipleotherfunctions,hereweshowedthatforthesubsetofspeciesthatoccurredinmultipletypestheiridentityeffectsonmultifunction-alitytendedtovaryconsiderably.ThepresenceofPicea abies and Pinus sylvestris, for instance, increasedordecreasedmixtureper-formance,dependingontheforesttype.
This calls for a new generation of forestry-oriented scientificexperimentsorsilvicultural trials tailoredtostudyspecies identityand composition effects in different environments (e.g., Paquetteetal.,2018),especiallyfocusingonthedriversofthecontextdepen-dencyindiversityeffects(wateravailability,growingseasonlength;Ratcliffeetal.,2017).Compositionscanbenotonlyreplicatedwithinforesttypesunderdifferentsoilconditionsandlevelsofwatersup-ply, but also across different forest types to cover regional-scalegradientssuchasclimate(seeBruelheideetal.,2014foradiversity-orientedexample) .Ofcourse, thegeographicscopeofamultisiteexperimentwillnotbeglobalandshouldstaywithinthecurrentorpredicteddistributionalrangeofthespeciesinvolved(e.g.,Verheyenetal.,2013),asstudyingfunctioningwelloutsidethespeciesrangeisprobablynotrelevantforforesters.Settinguppracticaltrialsobvi-ouslyrequirestheinvolvementofforesters,policymakers,resourcemanagers,andconservationists.Theycanuseour identificationofthebestspeciescombinationsasagoodstartingpointtocarefullyselectcompositionsfromthelargepoolofavailablespecies.
4.3 | Low multifunctionality of the most common species compositions
By ranking tree species compositions of three forest types ac-cordingtohowoftentheyoccurredininventorydata,weshowedthat the most frequent compositions were monospecific standsand that themost frequentspeciescombinationsmostlyshowedbelow-average performance in terms of multifunctionality andproductivitybasedontheexploratoryplatformdata.Severalmix-tureswithhighperformancewereveryrareinthenationalinven-toriesorevenabsentfromourselection.Weshouldacknowledge,however,thattheinventorydataspanmuchlargerenvironmentalgradients than the exploratory platform and that the samemix-turemayperformdifferentlyunderdifferentenvironmentalcondi-tions.Compositionsshowingpoorperformanceintheexploratoryplatformmaythusperformbetterindifferentclimaticorsoilcon-ditions.Whilethismaylimitthegeneralityofanyconclusionsre-garding specificmixtures, the under-representation of numerousabove-averageperformingmixturesintoday’sforestsandthehighproportionofmonoculturesisaclearindicationthatthepotentialofmixingdifferenttreespecies inforeststandshasnotyethavebeenfullyrealizedinEurope.
ACKNOWLEDG EMENTS
TheideasdevelopedforthisstudycamefromaFunDivEUROPEsyn-thesisworkshopheldon20–23February2017inFreiburg,Germany.The FunDivEUROPE project received funding from the EuropeanUnion’s Seventh Programme (FP7/2007–2013) under grant agree-mentNo.265171.We thank the followingorganizationsandpeo-pleforaccesstotheForestInventorydata:MAGRAMA(Spain),theJohannHeinrich vonThünen-Institut (Germany),Hugues LecomtefromtheWalloonForestInventory(Wallonia),theNaturalResourcesInstitute Finland LUKE (Finland), and the Swedish University of
| 743Journal of Applied EcologyBAETEN ET Al.
Agricultural Sciences (Sweden). Two anonymous reviewers helpedtoimproveanearlierdraftofthispublication.
AUTHORS’ CONTRIBUTIONS
L.B., H.B., F.V.D.P., and M.S.L. conceived the ideas and analysedthedata;L.B.ledthewritingofthemanuscript,togetherwithH.B.,F.V.D.P., S.K., S.R., T.J., andM.S.L. All authors collected the data,contributed critically to the drafts, and gave final approval forpublication.
DATA ACCE SSIBILIT Y
Dataoftheecosystempropertyandfunctionmeasurementsareavail-able via Figshare https://doi.org/10.6084/m9.figshare.5368846.v1(Ratcliffeetal.,2017).
ORCID
Lander Baeten https://orcid.org/0000-0003-4262-9221
Fons Plas https://orcid.org/0000-0003-4680-543X
Stephan Kambach https://orcid.org/0000-0003-3585-5837
Evy Ampoorter https://orcid.org/0000-0001-7676-0290
Raquel Benavides https://orcid.org/0000-0003-2328-5371
Charlotte Grossiord https://orcid.org/0000-0002-9113-3671
François-Xavier Joly https://orcid.org/0000-0002-4453-865X
Julia Koricheva https://orcid.org/0000-0002-9033-0171
R E FE R E N C E S
Allan,E.,Manning,P.,Alt,F.,Binkenstein,J.,Blaser,S.,Blüthgen,N.,…Fischer,M.(2015).Landuseintensificationaltersecosystemmulti-functionalityvialossofbiodiversityandchangestofunctionalcom-position. Ecology Letters, 18(8), 834–843. https://doi.org/10.1111/ele.12469
Baeten, L., Verheyen, K.,Wirth, C., Bruelheide, H., Bussotti, F., Finér,L., … Scherer-Lorenzen, M. (2013). A novel comparative researchplatformdesigned todetermine the functional significanceof treespecies diversity in European forests.Perspectives in Plant Ecology, Evolution and Systematics,15(5),281–291.https://doi.org/10.1016/j.ppees.2013.07.002
Barbati,A.,Corona,P.,&Marchetti,M. (2017).Europeanforest types:Categories and types for sustainable forest management report-ing and policy, European Environment Agency, Technical ReportNo9/2006,111pp.
Barot,S.,Allard,V.,Cantarel,A.,Enjalbert,J.,Gauffreteau,A.,Goldringer,I.,…Porcher,E.(2017).Designingmixturesofvarietiesformultifunctionalagriculturewiththehelpofecology.Areview.Agronomy for Sustainable Development,37(2),13.https://doi.org/10.1007/s13593-017-0418-x
Bruelheide,H.,Nadrowski,K.,Assmann,T.,Bauhus,J.,Both,S.,Buscot,F.,…Schmid,B. (2014).Designing forestbiodiversityexperiments:Generalconsiderationsillustratedbyanewlargeexperimentinsub-tropicalChina.Methods in Ecology and Evolution,5(1),74–89.https://doi.org/10.1111/2041-210X.12126
Byrnes, J. E. K., Gamfeldt, L., Isbell, F., Lefcheck, J. S., Griffin, J. N.,Hector, A., … Duffy, J. E. (2014). Investigating the relationship
betweenbiodiversityandecosystemmultifunctionality:Challengesandsolutions.Methods in Ecology and Evolution,5,111–124.https://doi.org/10.1111/2041-210X.12143
Fichtner,A.,Härdtle,W.,Li,Y.,Bruelheide,H.,Kunz,M.,&vonOheimb,G. (2017). From competition to facilitation: How tree species re-spondtoneighbourhooddiversity.Ecology Letters,20(7),892–900.https://doi.org/10.1111/ele.12786
Forrester,D.I.,&Bauhus,J.(2016).Areviewofprocessesbehinddiver-sity—Productivity relationships in forests.Current Forestry Reports,2(1),45–61.https://doi.org/10.1007/s40725-016-0031-2
Gamfeldt,L.,&Roger,F. (2017).Revisiting thebiodiversity–ecosystemmultifunctionality relationship. Nature Ecology & Evolution, 1(7),0168.https://doi.org/10.1038/s41559-017-0168
Gamfeldt,L.,Snäll,T.,Bagchi,R.,Jonsson,M.,Gustafsson,L.,Kjellander,P., … Bengtsson, J. (2013). Higher levels of multiple ecosys-tem services are found in forests with more tree species.Nature Communications,4,1340.https://doi.org/10.1038/ncomms2328
Hector,A.,Bell,T.,Hautier,Y., Isbell,F.,Kéry,M.,Reich,P.B.,…Schmid,B.(2011).BUGSintheanalysisofbiodiversityexperiments:Speciesrichnessand composition are of similar importance for grassland productivity. PLoS ONE,6(3),e17434.https://doi.org/10.1371/journal.pone.0017434
Hulvey, K. B., Hobbs, R. J., Standish, R. J., Lindenmayer, D. B., Lach,L.,&Perring,M.P. (2013).Benefitsof treemixes in carbonplant-ings. Nature Climate Change, 3(10), 869. https://doi.org/10.1038/nclimate1862
Isbell, F., Gonzalez, A., Loreau, M., Cowles, J., Díaz, S., Hector, A., …Larigauderie, A. (2017). Linking the influence and dependence ofpeopleonbiodiversityacrossscales.Nature,546,65–72.https://doi.org/10.1038/nature22899
Jucker,T.,Bouriaud,O.,Avacaritei,D.,&Coomes,D.A.(2014).Stabilizingeffectsofdiversityonabovegroundwoodproductioninforesteco-systems: Linking patterns and processes. Ecology Letters, 17(12),1560–1569.https://doi.org/10.1111/ele.12382
Jucker,T.,Bouriaud,O.,Avacaritei,D.,Dǎnilǎ,I.,Duduman,G.,Valladares,F., & Coomes, D. A. (2014). Competition for light and water playcontrasting roles in driving diversity-productivity relationships inIberian forests. Journal of Ecology, 102(5), 1202–1213. https://doi.org/10.1111/1365-2745.12276
Kirwan, L., Connolly, J., Finn, J., Brophy, C., Lüscher, A., Nyfeler,D., & Sebastia, M.-T. (2009). Diversity-interaction modeling:Estimating contributions of species identities and interactionsto ecosystem function. Ecology, 90(8), 2032–2038. https://doi.org/10.1890/08-1684.1
Liang, J., Crowther, T.W., Picard, N.,Wiser, S., Zhou,M., Alberti, G.,… Reich, P. B. (2016). Positive biodiversity-productivity relation-ship predominant in global forests. Science, 354(6309), aaf8957. https://doi.org/10.1126/science.aaf8957
Lu, H.,Mohren, G.M. J., den Ouden, J., Goudiaby, V., & Sterck, F. J.(2016).Overyieldingoftemperatemixedforestsoccursinevergreen-deciduousbutnotindeciduous-deciduousspeciesmixturesovertimeintheNetherlands.Forest Ecology and Management,376(September),321–332.https://doi.org/10.1016/j.foreco.2016.06.032
Manning, P., Plas, F., Soliveres, S., Allan, E., Maestre, F. T., Mace, G.,… Fischer, M. (2018). Redefining ecosystem multifunctionality.Nature Ecology & Evolution, 2(3), 427. https://doi.org/10.1038/s41559-017-0461-7
Meyer, S. T., Ptacnik, R., Hillebrand, H., Bessler, H., Buchmann, N.,Ebeling, A., … Klein, A. M. (2018). Biodiversity–multifunctionalityrelationships depend on identity and number of measured func-tions.Nature Ecology & Evolution,2(1),44.https://doi.org/10.1038/s41559-017-0391-4
Mori,A.S.,Lertzman,K.P.,&Gustafsson,L.(2017).Biodiversityandeco-systemservicesinforestecosystems:Aresearchagendaforappliedforest ecology. Journal of Applied Ecology,54(1), 12–27.https://doi.org/10.1111/1365-2664.12669
744 | Journal of Applied Ecology BAETEN ET Al.
Paquette,A.,Hector,A.,Castagneyrol,B.,Vanhellemont,M.,Koricheva,J.,Scherer-Lorenzen,M.,&Verheyen,K.(2018).Amillionandmoretrees for science.Nature ecology & evolution,2(5), 763. https://doi.org/10.1038/s41559-018-0544-0
Piotto,D. (2008).Ameta-analysis comparing treegrowth inmonocul-tures andmixed plantations. Forest Ecology and Management,255,781–786.https://doi.org/10.1016/j.foreco.2007.09.065
Pretzsch, H. (2014). Canopy space filling and tree crown morphol-ogy in mixed-species stands compared with monocultures. Forest Ecology and Management, 327, 251–264. https://doi.org/10.1016/j.foreco.2014.04.027
Pretzsch,H.,Bielak,K.,Block,J.,Bruchwald,A.,Dieler,J.,Ehrhart,H.P.,…Zingg,A.(2013).Productivityofmixedversuspurestandsofoak(Quercus petraea (Matt.) Liebl. andQuercus robur L.) and Europeanbeech (Fagus sylvatica L.) along an ecological gradient. European Journal of Forest Research,132(2),263–280.https://doi.org/10.1007/s10342-012-0673-y
Pretzsch, H., Block, J., Dieler, J., Dong, P. H., Kohnle, U., Nagel, J., …Zingg, A. (2010). Comparison between the productivity of pureandmixedstandsofNorwayspruceandEuropeanbeechalonganecologicalgradient.Annals of Forest Science,67(7),712.https://doi.org/10.1051/forest/2010037
Pretzsch,H.,Forrester,D. I.,&Bauhus,J. (2017).Mixed-species forests. Berlin,Heidelberg:Springer.
Ratcliffe, S., Liebergesell, M., Ruiz-Benito, P., Madrigal González, J.,Muñoz Castañeda, J. M., Kändler, G., …Wirth, C. (2016). Modesof functional biodiversity control on tree productivity across theEuropeancontinent.Global Ecology and Biogeography,25,251–262.https://doi.org/10.1111/geb.12406
Ratcliffe, S., Wirth, C., Jucker, T., van der Plas, F., Scherer-Lorenzen,M.,Verheyen,K.,…Baeten, L. (2017).Biodiversity andecosystemfunctioningrelationsinEuropeanforestsdependonenvironmentalcontext.Ecology Letters,20(11),1414–1426.https://doi.org/10.1111/ele.12849
Rothe, A., & Binkley,D. (2001).Nutritional interactions inmixed spe-ciesforests:Asynthesis.Canadian Journal of Forest Research,31(11),1855–1870.https://doi.org/10.1139/x01-120
Rupf, H. (1961). Wald und Mensch im Geschehen der Gegenwart.Allgemeine Forstzeitschrift. Der Wald,16,545–546.
Schulze, E.-D., & Mooney, H. A. (1993). Biodiversity and ecosys-tem function. New York, NY: Springer-Verlag. https://doi.org/10.1007/978-3-642-58001-7
Storkey, J.,Ring,T.D.,Baddeley, J.,Collins,R.,Roderick,S., Jones,H.,& Watson, C. (2015). Engineering a plant community to deliver
multiple ecosystem services. Ecological Applications, 25(4), 1034–1043.https://doi.org/10.1890/14-1605.1
Tilman,D.,Isbell,F.,&Cowles,J.M.(2014).Biodiversityandecosystemfunctioning.Annual Review of Ecology, Evolution, and Systematics,45(1),471–493.https://doi.org/10.1146/annurev-ecolsys-120213-091917
vanderPlas,F.,Manning,P.,Allan,E.,Scherer-Lorenzen,M.,Verheyen,K.,Wirth, C.,… Fischer,M. (2016). Jack-of-all-trades effects drivebiodiversity-ecosystemmultifunctionalityrelationshipsinEuropeanforests.Nature Communications,7, 11109.https://doi.org/10.1038/ncomms11109
van der Plas, F., Ratcliffe, S., Ruiz-Benito, P., Scherer-Lorenzen, M.,Verheyen,K.,Wirth,C.,…Allan,E. (2018).Continentalmappingofforest ecosystem functions reveals a highbut unrealisedpotentialforforestmultifunctionality.Ecology letters,21(1),31–42.https://doi.org/10.1111/ele.12868
Verheyen, K., Ceunen, K., Ampoorter, E., Baeten, L., Bosman, B.,Branquart,E.,…Ponette,Q.(2013).Assessmentofthefunctionalroleof treediversity:Themulti-siteFORBIOexperiment.Plant Ecology and Evolution,146(1),26–35.https://doi.org/10.5091/plecevo
Zhang,Y.,Chen,H.Y.H.,&Reich,P.B. (2012).Forestproductivity in-creaseswithevenness,speciesrichnessandtraitvariation:Aglobalmeta-analysis. Journal of Ecology, 100(3), 742–749. https://doi.org/10.1111/j.1365-2745.2011.01944.x
Zhang,Y.,Chen,H.Y.H.,&Taylor,A.R.(2015).Abovegroundbiomassofunderstoreyvegetationhasanegligibleornegativeassociationwithoverstorey tree species diversity in natural forests.Global Ecology and Biogeography,25,141–150.
SUPPORTING INFORMATION
Additional supporting information may be found online in theSupportingInformationsectionattheendofthearticle.
How to cite this article:BaetenL,BruelheideH,vanderPlasF,etal.IdentifyingthetreespeciescompositionsthatmaximizeecosystemfunctioninginEuropeanforests.J Appl Ecol. 2019;56:733–744. https://doi.org/10.1111/1365-2664.13308
