Approaches to defining a planetary boundary for biodiversity

Approaches to defining a planetary boundary for biodiversity

Georgina M. Mace

*, Belinda Reyers

, Rob Alkemade

, Reinette Biggs

,

F. Stuart Chapin III

, Sarah E. Cornell

, Sandra Dı´az

, Simon Jennings

, Paul Leadley

, Peter J. Mumby

, Andy Purvis

, Robert J. Scholes

, Alistair W.R. Seddon

, Martin Solan

, Will Steffen

, Guy Woodward

aCentreforBiodiversityandEnvironmentResearch,DepartmentofGenetics,EvolutionandEnvironment,UniversityCollegeLondon,LondonWC1E6BT,UK

bNaturalResourcesandEnvironment,CSIR,POBox320,Stellenbosch7599,SouthAfrica

cPBLNetherlandsEnvironmentalAssessmentAgency,Bilthoven,TheNetherlands

dEnvironmentalSystemsAnalysisGroup,WageningenUniversity,TheNetherlands

eStockholmResilienceCentre,StockholmUniversity,SE-10691Stockholm,Sweden

fCentreforStudiesinComplexity,StellenboschUniversity,SouthAfrica

gInstituteofArcticBiology,UniversityofAlaskaFairbanks,Fairbanks,AK99775,USA

hInstitutoMultidisciplinariodeBiologı´aVegetal(IMBIV-CONICET)andDepartamentodeDiversidadBiolo´gicayEcologı´a,FCEFyN,UniversidadNacionalde Co´rdoba,CC495,5000Co´rdoba,Argentina

iCentreforEnvironment,FisheriesandAquacultureScience,LowestoftNR330HT,UK

jSchoolofEnvironmentalSciences,UniversityofEastAnglia,NorwichNR47TJ,UK

kUnivParis-Sud,LaboratoireESE,UMR8079UnivParis-Sud/CNRS/AgroParisTech,F91405Orsay,France

lMarineSpatialEcologyLab.,SchoolofBiologicalSciences,GoddardBuilding,Room170,UniversityofQueensland,StLuciaCampus,Brisbane, Queensland4072,Australia

mDepartmentofLifeSciences,NaturalHistoryMuseum,CromwellRoad,LondonSW75BD,UK

nNaturalResourcesandEnvironment,CSIR,POBox395,Pretoria0001,SouthAfrica

oDepartmentofBiology,UniversityofBergen,Postbox7803,N-5020Bergen,Norway

pOceanandEarthScience,NationalOceanographyCentre,UniversityofSouthampton,SouthamptonSO143ZH,UK

qTheFennerSchoolofEnvironmentandSociety,TheAustralianNationalUniversity,Canberra,AustralianCapitalTerritory0200,Australia

rDepartmentofLifeSciences,ImperialCollegeLondon,SilwoodParkCampus,BuckhurstRoad,Ascot,BerkshireSL57PY,UK

ARTICLE INFO Articlehistory:

Received24April2014

Receivedinrevisedform16June2014 Accepted26July2014

Availableonline27August2014 Keywords:

Biodiversity Planetaryboundary Phylogeneticdiversity Functionaldiversity Biomeintegrity

ABSTRACT

Theideathatthereisanidentifiablesetofboundaries,beyondwhichanthropogenicchangewillputthe Earth systemoutside asafe operatingspace for humanity, is attracting interestin thescientific communityandgainingsupportintheenvironmentalpolicyworld.Rockstrometal.(2009)identifynine suchboundariesandhighlightbiodiversitylossasbeingthesingleboundarywherecurrentratesof extinctionputtheEarthsystemfurthestoutsidethesafeoperatingspace.Herewereviewtheevidenceto supportaboundarybasedonextinctionratesandidentifyweaknesseswiththismetricanditsbearing onhumanity’sneeds.Whilechangestobiodiversityareofundisputedimportance,weshowthatboth extinctionrateandspeciesrichnessareweakmetricsforthispurpose,andtheydonotscalewellfrom localtoregionalorglobal levels.Wedevelopalternativeapproachesto determinebiodiversityloss boundariesandextendouranalysistoconsiderlarge-scaleresponsesintheEarthsystemthatcould affectitssuitabilityforcomplexhumansocietieswhichinturnaremediatedbythebiosphere.We suggestthreefacetsofbiodiversityonwhichaboundarycouldbebased:thegeneticlibraryoflife;

functionaltypediversity;andbiomeconditionandextent.Foreachoftheseweexplorethescience neededtoindicatehowitmightbemeasuredandhowchangeswouldaffecthumansocieties.Inaddition tothesethreefacets,weshowhowbiodiversity’sroleinsupportingasafeoperatingspaceforhumanity maylieprimarilyinitsinteractionswithotherboundaries,suggestinganimmediateareaoffocusfor scientistsandpolicymakers.

ß2014TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/3.0/).

* Correspondingauthor.Tel.:+442031081125;fax:+4402076797193.

E-mailaddress:[email protected](G.M.Mace).

1Jointfirstauthors.

ContentslistsavailableatScienceDirect

Global Environmental Change

j ou rna l hom e pa ge : w w w. e l s e v i e r. c om/ l o ca t e / gl oe n v cha

http://dx.doi.org/10.1016/j.gloenvcha.2014.07.009

0959-3780/ß2014TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/3.0/).

1. Introduction

The identification of global-scale thresholds underpins the planetary boundaries concept introduced by Rockstrom et al.

(2009a, 2009b). Nine boundaries were proposed, representing specificthresholdsofclimatechange,oceanacidification,strato- sphericozone,globalnitrogenandphosphoruscycles,atmospheric aerosolloading,freshwateruse,land-usechange,biodiversityloss, andchemicalpollution(Rockstrometal.,2009a)thatcollectively delimit‘asafeoperatingspaceforhumanity’.Whilesomeofthe proposedboundarieswererelativelysimpletodefinebecauselocal inputs or changes make a predictable contribution to global processeswithknownthresholds,others(suchasland-usechange andbiodiversityloss)arerecognisedascomplexhumansystem- ecosystemprocessesnoteasilyassociatedwithknown globalor continentalthresholds(Rockstrometal.,2009a).

Transgressinganyofthenineboundariesisexpectedtoleadtoan increasedrisktooneormoreaspectsofhumanwellbeing,orwould underminetheresilienceoftheEarthsystemasawhole.Whilesome boundariesoperateinatop-downmannerdrivenbysystemicglobal processes(e.g.climatechange),othersmaybebottom-upprocesses drivinglarge-scaleresponsessothattheprocessesmightbelocalor regional only, but sufficiently widespread to have significant aggregateconsequencesattheglobal-scale(e.g.NandPnutrient pollution)(Rockstrometal.,2009a).

Theplanetaryboundaryforbiodiversityhasbeenparticularly problematic. The original analysis emphasised the difficulty of describingandquantifyingaboundaryforbiodiversityloss,noting thatitisa slowprocesswithoutknownglobal-levelthresholds, thatthereisincompleteknowledgeontheroleofbiodiversityfor ecosystem functioning across scales, and that the suggested boundary position was therefore highly uncertain (Rockstrom etal.,2009a).Howevertherearemorefundamentalproblemswith thebiodiversityboundarythanjustsettingitsposition.Brooketal.

(2013)questionedtheexistenceofaglobalbiodiversitythreshold, notingthelargespatialheterogeneityinthedriversandresponses associatedwithbiodiversityloss,thelackofabruptshiftsatglobal scale, and theabsence of thelarge-scale interconnectivity that would be needed to propagate local ecosystem regime shifts globally. In contrast, Hughes et al. (2013) suggest that local changescouldscaleuptoregionalorglobal-level,especiallygiven theinterconnectedness of human systems, and Barnosky et al.

(2012)notethatslowdriversoverhumantimescalescanstilllead tothresholds.Theextenttowhichabiodiversityboundarymight be experienced at local, regional or global scales, and indeed whetheraccumulationsoflocalbiodiversity changecanimperil large-scaleprocesses,iscurrentlyunresolved.

The discussionof a biodiversityboundary is alsocloudedby confusionovertheuseoftheterm‘biodiversity’,whichcansimply meanspeciesrichness,butisoftenusedforfunctionalorecosystem diversity,ormoregenerallytorepresentthewholevarietyoflifeon Earth,sometimeswithconnotationsofnaturalnessorintactness (ConventiononBiologicalDiversity,2010;DeLong,1996;Fischer andYoung,2007). Biodiversitylossis generally manifestedas a reductioninspeciesnumbers(ultimatelytoglobalextinctionrates), butitismoreoftentheextentandbiomassofthebiospherethathas adominantinfluenceonEarthsystemprocessesandtheecosystem servicesonwhichpeopledepend(Dı´azandCabido,2001;Dı´azetal., 2006;Mace, 2005; Mace et al., 2012). The broad definitionsof biodiversityincurrentusagedonotallowsuchdistinctionstobe drawn,despitetheirimportance.

Discussionsofthresholdsforbiodiversityarefurtherconfused aboutwhethertheproposedthresholdisintendedtorepresent(i) changesinelementsofbiodiversitythatcausealarge-scalechange inother processesin the Earthsystem, (ii)physical or biogeo- chemicalchangesintheEarthsystemthatcauserapid,large-scale

biodiversitychange,or(iii)localisedecosystemchangesthatmay propagateandscaleuptolarge-scaleorevenglobalbiodiversity change. There is someevidence that each of these takes place (Barnoskyetal.,2012;Leadleyetal.,2010;LentonandWilliams, 2013), but none is quite what is defined by the planetary boundariesconceptwithitsclearimplicationthattheboundary positionissetbythelevelofadriver(inthiscasebiodiversityloss) wherethereisaraisedriskofimpactonhumanwelfare.

Here,we reviewthecurrentbiodiversityboundaryas deter- minedbyRockstrometal.(2009b).Drawingonrecentresearchwe develop a conceptual basis for a biodiversity boundary which proposes alternative approaches that could delimit the safe operating spacefor humanity. We usethis conceptual basisto identify key research directions needed to move towards the identification of actual metrics and quantitative boundaries or thresholds.

2. Acritiqueofthebiodiversityboundary

Accordingtotheplanetaryboundariesconcept,allboundaries are defined in terms of response and control variables (Fig. 1) (Rockstrom et al., 2009a). Response variables are measures of Earth-system responses relevant to humans. Control variables represent the metric(s) related to the specific boundary that determines the Earth system response, while the boundary is definedasahuman-determinedlevelofthecontrolvariablesetat a‘‘safe’’distancefromaglobalthresholdorapotentiallydangerous level. Currently the planetary boundary for biodiversity uses ecosystem functioning as the response variable and the global speciesextinctionrateasitscontrolvariable.Theboundaryissetat 10 times the average background extinction rate, which is 10 extinctionspermillionspeciesperyear(E/MSY),roughlyequiva- lenttoHolocenerates.Speciesextinctionrateisarguablythemost fundamentalmeasureofglobalbiodiversityloss,butnotanideal metricinthiscontextforanumberofreasons.First,ittendstobe estimatedmostoftenforvertebratespecies(anunrepresentative

<2%ofalldescribedspecies).Second,itisinsensitivetoimportant

changes in species abundance, community composition and distribution of species (Balmford et al., 2003; Millennium Ecosystem Assessment, 2005; Pereira et al., 2012). Third, it is hardtoestimatewithhighcertaintyuntillongaftertheextinction hasoccurred(Heywoodet al.,1994).Finally itis notclearhow globalspeciesextinctionrateswillinfluenceecosystemfunction- ingatscalesrelevanttothesafeoperatingspace.

Recentreviewsoftherelationshipsbetweenspeciesrichness andecosystemfunctionsshowthatasspecieslossincreasesand thesystemapproachesamonoculture,ecosystemprocessessuch asprimaryproductionanddecompositiononaveragedecline,but alsoshowastrongly increasingvariancein response(Cardinale et al., 2012; Hooper et al., 2012). These reviews are based on multipleexperimentalstudieswheretheeffectsofbiomassand sampling of species are controlled so that the effects can be attributedtorichnessalone.Thebestmonocultureoftenoutper- formsthemostdiverse systembecausecertainspeciesarevery effective at a particular process on their own, but in general, especiallyatintermediatelevelsofrichness,lowerspeciesrichness leadstoreducedecosystemfunctions(Cardinaleetal.,2012).Such compositerelationshipshoweverhaveonlylimitedapplicabilityto thebroaderissueconcerningtheimpactof biodiversitylosson peoplebecausetheyarebasedoncontrolled mesocosmorfield experiments,usuallyconducted inrelativelysimple ecosystems overyearsor–occasionally–decades.Theycannotrepresentthe additionalcontributionsfromrichnessordiversitytoecosystem functionsovertime(Reichetal.,2012)andplace(Godboldetal., 2011;Spehnetal.,2005)orthefactthat,althoughcertainspecies mayappearredundantwhenaparticularfunctionisconsidered

under one set of environmental conditions, many species are needed to guarantee multiple functions in a changing world (Gamfeldtetal.,2008;HectorandBagchi,2007;Isbelletal.,2011).

Additionally,species’rolesanddominancechangeovertime,and functionsareenhancedwhenthediversityincludesspecieswith complementary interactions (Allan et al., 2011). Furthermore, despitemuchevidenceforlossofregionalandglobaldiversityand abundance(Butchartetal.,2010;Pereiraetal.,2012),thereislittle evidenceofdeclinesinlocalspeciesrichnessovertime(Dornelas etal.,2014;Vellendetal.,2013),mostnotablyinplantsthatare significant for ecosystem processes (de Bello et al., 2010) and which are the focus of study in many biodiversity-ecosystem functionexperiments.Changestolocal,comparedtoregionaland globalspeciesrichness,arealsonotcloselyrelatedforreasonstodo withthelocaldisappearanceofrarespeciesoftenbeingmaskedby morewidespreadspecies(Thomas,2013).

Ingeneral,therefore,whilethereisalargebodyofworkshowing thelinksbetweenspeciesrichnessandecosystemprocesses,mostof thesestudiesfocusonlyonspeciesrichness,havebeencarriedoutat veryfine scales and, because they do not consider other, more relevant,componentsofbiodiversity,theirfindingsareinsufficient topredictabiodiversityboundary(Brooketal.,2013).Whilearecent

commentarystatesthat70%ofspeciesinanyecosystemshouldbe retainedforsecuringhealthyandproductiveecosystems(Griggs etal.,2013),thereisnoevidencetosupportthisfigure.

Moreover,aboundarybasedonspeciesrichnessalonemisses many more fundamental and persistent roles of broader sense biodiversity,especiallyrelatedtoabundance,communitycomposi- tion, functional traits and ecosystem-level interactions. Existing global biodiversity measures such as Mean Species Abundance (Alkemadeetal.,2009),theRedListIndex(Butchartetal.,2004, 2007),theBiodiversityIntactnessIndex(ScholesandBiggs,2005) andothersreviewedbyVackaretal.(2012)areunabletoreflectthe keyfeaturesofbiodiversityimportantforhumanity.Themostrecent effort to agree even a minimum set of biodiversity indicators essential for study, reporting, and management of biodiversity change,includes22variables,mostofwhichrequiresamplingover multipletaxaorlocations(GEOBON,2013;Pereiraetal.,2013).

We have identified certain problems that arise from using speciesextinctionratesasthemetricforabiodiversityboundary.

Inadditiontherearequestionsabouthowanybiodiversity-related boundariesmightscaleupanddownfromlocaltoglobal.Belowwe explore these issues and suggest three possible approaches (Table 1). In addition to the role biodiversity loss plays in

Table1

Summaryofproposalsforplanetaryboundariesforbiodiversity.

Proposedboundary Controlvariable Responsevariable Relationship Boundary

Speciesnumbersrelatedto ecologicalfunctions

(fromRockstrometal.2009a,b)

Extinctionrate Ecosystemfunctioning atcontinentalandocean basinscales

Thresholdslikelyat localandregionalscales.

Boundaryposition highlyuncertain Thegeneticlibraryoflife uPSV–ameasurerelated

tophylogeneticdiversity

Longterminnovationand resilienceofecosystem formandfunction

Probablyroughlylinearbut potentiallywithstepsassociated withcladeandbiogeographic sensitivity

Couldberatherarbitrary ifrelationshipiscloseto linear.Likelytobemore thresholdsatlocaland regionalscales Levelsoffunctionaldiversity Measuresoffunctional

diversityrelevanttokey ecosystemprocesses

Ecosystemfunctionsand processeslinkedtohuman wellbeing

Likelytobenon-linearwith discontinuitiesaskeysetsof functionsarelost

Pointsofcriticalfunctional losspotentiallylinkedto majorecologicalprocesses (e.g.trophiclevels,production, nutrientcycles)

Biomeintegrity Biomespecificdrivers Biomeconditionandextent Eachbiomewillhavedistinct andhighlynon-linearform

Compositeofseveralbiomes (seeFig.2)

Fig.1.Conceptualdescriptionofplanetaryboundaries.InthelefthandpaneltheboundaryisdesignedtoavoidthecrossingofacriticalcontinentaltoglobalthresholdinanEarth systemprocess.Intherighthandpanelthereisnoglobalthresholdeffectasfarasweknow,butexceedingtheboundarylevelwillleadtosignificantinteractionswithregionaland globalthresholdsand/ormaycausealargenumberofundesiredthresholdeffectsatthelocaltoregionalscale,whichinaggregateadduptoaseriousglobalconcernforhumanity.

FromRockstrometal.(2009a).

supportinghumanity, italso playsa key rolein mediating and respondingtootherplanetaryprocesses;wethereforeconclude withanexaminationoftheinteractionsbetweenbiodiversityloss andtheotherplanetaryboundaries.

3. Alternativestotheextinctionrateboundary

3.1. Geneticdiversityforlong-termevolutionary-ecologicalpotential

In the long term – over centuries to millennia – human wellbeingwilldependonthebiota’scontinuedabilitytosupport desired ecosystem services and processes in the face of often rapidly-changingselectivepressures.Becauseitisnotpossibleto predictwhichfunctionaltraitcombinationswillbemostneeded,it isnotpossibletoidentifythemostimportantspeciesaprioriwith anycertainty.Instead,we shouldaimtomanagetherisks,and somefeaturesofspeciescanbeusedtopredicthowmuchtheyadd to humanity’s ‘‘portfolio of biodiversity insurance’’. For this purpose,a species’future importancedepends on howmuchit adds to the overall diversity of unspecified functional traits, ultimatelyreflectedintheextentofphylogeneticdiversity(Faith, 1992) representing future option values of biodiversity (sensu Faith,1994)

Theexpectedvarianceofaneutrally-evolvingphenotypictrait canbeestimatedfromthephylogenyoftheevolvingspeciesor populations. When branch lengths are expressed in terms of geneticchange(Helmusetal.,2007),theunscaledversionofthis PhylogeneticSpeciesVariability(PSV)measurereflectstheoverall diversityof unspecifiedtraits, underthe assumptionof neutral evolution,andcouldincorporatemorecomplicatedandrealistic modelsoftraitevolution(reviewedbyO‘Meara,2012).However,it canbehardtotellwhichofthesemodelsbestapproximatesreality.

Evenwithlargedatasets(CooperandPurvis,2010;HoandAne, 2013),ratesofphenotypicevolutioncanbelargelydecoupledfrom ratesofnucleotidesubstitutioningenesequences(Janeckaetal., 2012)andempiricalanalysesindicatethatphylogeneticdiversity does not capture the most diverse sets of biological features, especially among more distantly related lineages (Kelly et al., 2014). Analternativeapproachmightbetousemacroecological approachestomodelhowratesofphenotypicchangedependon featuresoflineagesandenvironments.Modelssofarhavelacked explanatorypowerorgenerality(Janeckaetal.,2012),butricher datasets–capturingalargersetoftraitsformorepopulationsin morespecies –mightimprovethesituation.Afurthercaveatis that,asspeciesarelostphylogeneticdiversitywillnotdeclineas rapidlyasfunctionaltraitvarianceiftheriskofextinctiondepends onspecies’valuesoffunctionaltraits.Suchadecouplinghasbeen shown for body size variance and other phylogenetic diversity measuresinmammals(FritzandPurvis, 2010).Lastly,phyloge- netic diversity does not incorporate within-population genetic variation, on which adaptability depends. A coarse way of accommodating this could be to exclude species that are threatenedwithextinction(perhapsusingtheIUCNRedList):a meta-analysisshowed that heterozygositywas on average 35%

lower in a set of 170 threatened species than in close non- threatenedrelatives(Spielmanetal.,2004).

Ifsuchametricofphylogeneticdiversityisviewedasacontrol variableonlong-termevolutionary-ecologicalpotential,howdoes it respond to human pressures? It declines when a species disappearsfromitscalculation,throughbeingdeclaredextinctor (with the modification suggested above) being declared to be threatened;theextentofthedeclinethendependsonthespecies’

evolutionarydistinctiveness. Smaller declines occur when local populationsareextirpated.Increasesoccurasspecies’conserva- tionstatusesimproveand,morepassively,simplyaspopulations surviveovertime.

Whatwouldconstituteathreshold?Lossofspeciesatrandom willreducephylogeneticdiversityonlyslowly,asmostspecieslost will have close relatives that remain (Nee and May, 1997).

However,bothextinctionandextinctionrisktendtobeclumped withinphylogenies(Maceetal.,2003)andareoftengeographically correlated,increasingthelikelihoodthatentirecladeswillceaseto exist,andthereforepotentiallydefininglocalorregionalthresh- olds. Habitat conversion has been the dominant driver of biodiversity loss over recent decades (Hoffmann et al., 2010) and is projected to continue to be important throughout this century(Pereiraetal.,2010).However,itspotentialtoconstitutea thresholdmaybereducedbecausehabitatconversionisa local process,operatingatdifferentratesandaffectingdifferentsetsof speciesindifferentsettings;globalresponsesmightthereforebe expectedtoberather smooth,eveniflocalor regionalchanges exhibitthresholds.

3.2. Levelsoffunctionaldiversity

Functional diversityrepresents thevalue,range, distribution andrelativeabundanceofthefunctionaltraitsoftheorganisms present in an ecosystem or biota (Dı´az and Cabido, 2001).

Organismswithdifferentfunctionaltraitscandifferentiallyaffect ecosystempropertiessuchasprimaryproduction,decomposition or detoxification and also react differently to changes in the environment (Dı´az and Cabido,2001; Diaz etal.,2007; Lavorel et al., 2011; Suding et al., 2008). As certain combinations of functionaltraitsarelostinthefaceofenvironmentalchange,key functions couldbe at risk, especially if thetraits of individual speciesco-varywiththeirriskofextinction(Solanetal.,2004),or couldsimplybeperformedless efficientlythaninmorediverse systems.Whilethegeneticdiversityboundarymightrepresentall diversityanditspotentialfutureutility,thefunctionaldiversity hererepresentscurrentfunctionsthatareknowntobesignificant forhumanity.

There may be a control variable of this sort that captures enough of the relationship between ecosystem structure and functiontobeusefulattheregionaltogloballevel.Itisunlikelyto beascrudeasasimplecountoffunctionalgroupspresent(i.e.,

‘functional type richness’), since some of those functions are irreplaceablewhereasothersmaybepartlyorfullyredundant.It could,however,bebasedonasystematicassessmentofrelevant traits(Diazetal.,2013;Kattgeetal.,2011;Lavoreletal.,2013), potentially geared to key functions (Craine et al., 2002), after recognisingand addressingthepotential circularityinvolvedin assigningfunctionaltraitstogroupsofspecies(Wrightetal.,2006).

A measurebased on functional traitswould operate effectively over short time scales and be relevant to the maintenance of adaptive variabilityand resilienceatlocal-to-regionalscalesfor key ecosystem functions(Table 1). It could be designed to be sensitivetolarge-scalelossesinparticularfunctionalgroups,such asimportant food and fibrespecies, toppredators (Estes et al., 2011),plantswithtraitsforhighcarbonassimilation,transferand storagebelowground(DeDeynetal.,2008),orforcertaingroups withsignificantregulatoryfunctions,suchasinArcticpeatlandsor marinephytoplankton.Althoughthereis nowawide varietyof functional diversityindices applicableat thelocal tolandscape scales(seeforexampleMasonetal.,2013),currentlythereareno generalmetricsthat couldbeusedtoassessstatus ortrendsin functional diversity at broader scales. The identification of thresholdsmightbechallengingiftherearemultiple,non-linear functionalrelationshipsthataredifficulttoaggregatewithinand acrossecosystems.Somepilotapproaches,for instancethrough adapting a functional-type-based version of the Biodiversity IntactnessIndex(ScholesandBiggs, 2005)couldbeconsidered, andbothrelevanttheoryandobservationsprovideafoundationfor

suchanapproach(e.g.,GEOBON)(Kattgeetal.,2011;Pereiraetal., 2013).

3.3. Biomeintegrity

Instead of asking about the consequences of progressive biodiversity loss,we nowconsider thelarge-scalebiodiversity- mediated responses in Earth systems that affect the planet’s suitability for complex human societies. For example, the biospheredrivesglobalbiogeochemistry, whichin turngoverns atmosphericcomposition,soilfertilityandoceanproductivity.The lossordegradationofentirebiomes(e.g.,coralreefs),or ofthe biodiversity components associated with large-scale ecological processes(e.g.predation,nutrientcycling)wouldhavesubstantial impacts on regional and distant social and ecological systems (Barnosky et al., 2012; Leadley et al., 2010). Changes in these biosphericprocessescouldbe large enoughtocompromise the Earth’sabilitytosustainhumansocietiesasweknowthem,and offera potentially simpler routetodeveloping a boundary.For example,Running (2012) proposes theuse ofNPP for a global boundary, which would reflect changes to the fundamental biosphereprocessesofproductionandnutrientcycling.Herewe developasimilarlogicforsubsetsofthebiosphere.

Biomesareglobal-scalesystems,suchastundra,coralreefsor tropicalgrasslandsandsavannas,distinguishedfromoneanother bythecollectionsofecosystemsandspeciesassemblagesfound there.Thereareseveralclassifications,butonewidelyusedscheme recognises14terrestrial,sevenfreshwaterandfivemarinebiomes (Olson and Dinerstein, 2002). A biome-based approach to planetary boundaries rests on the notion that biomes embed functional diversityand someaspects of phylogeneticdiversity andaretherefore,atleasttoadegree,biophysicallycoherentsub- unitsofthewholeEarth.Theycouldconnectmeaningfullytoother planetaryboundaries.Changesintheconditionandspatialextent ofbiomesmaybeappropriateforaplanetaryboundarybecause theconsequencesofbiome-levelchangesonecosystemprocesses, services, and humanwell-being are relatively well understood (Heyder et al., 2011), asare impacts on speciesand ecological communitieswithinthem(Alkemadeetal.,2009;Kennedyetal., 2013).Forexample,shiftsoftropicalforesttosavannaorpastures, iftheyoccuratsufficientlylargescales,canaffectregionalrainfall patternsandglobal-scalecarbonstorage(Davidsonetal.,2012) andradicallyalterthediversityofparticularfunctionalgroupsand species(Gibsonetal.,2011).

Ageneralindicatorofbiome-levelchangecouldbeidentifiedby defining a control variable that determines the integrity and functioningofspecificbiomesat broadscales.Forexample,the marinecarbonatebudgetcontrolsthepersistenceofcoralreefs, precipitationlevelsdeterminetheexistenceofdryland–rainforest systems, nitrogen deposition controls the existence of some forests, and carbon dioxide concentrations can control the existenceofgrasslandswithaC4photosyntheticpathway.

Usingthisapproachwouldrequirebiophysicalcontrolvariables to be identified for each biome and then for them all to be aggregated to the global-level. Box 1 describes the general approachandpresentsaworkedexampleforcoralreefintegrity.

Multiplecontrolvariables needtobeidentified;ideally oneper biome.Giventhestrikingdifferencesamongbiomes,suchasarctic tundraandcoralreefs,asinglecontrolvariablewouldnotapply meaningfully to all, but one control variable could assess the security of biome function on a per-biome basis. Sub-global patternswithin a biome, suchas those due tobiogeographical differences,couldbeaccountedforbyscalingthecontrolvariable tolevelsexpectedforeachprovince.Incertaincases,controland response variables could also be identified in long-term (i.e.

palaeoecological,palaeolimnological)datasetstoprovideinsights

Box1. Biomeintegrityasaboundary

Oneapproachistosetbiodiversityboundariesthatmaintainthe functioningoftheEarth’smajorbiomes(rainforest,savannah, coralreefs,etc.;seemain text).Biomesprovideawealth of regulatingecosystemservicesthatmaintainEarthsystempro- cesses,includingthecyclingof freshwaterandcarbon. We suggestthatsecurefunctioningofbiomeswouldsupportEarth systemfunctioninaHolocene-likestate,largelybymaintaining thestructureandfunctionaldiversityofecosystems.

Abiomeisconsideredfunctionallysecureifitcanmaintainits keystructuresandfunctionsinthelongterm(e.g.toatleast 2100,butgiventheHolocenefocusoftheboundariesconcept, potentiallyformillennia).Clearly,animportantconsiderationis theselectionofappropriateresponseandcontrolvariablesfor eachbiome.Thecontrolvariable(orsetofvariables)pertainsto thefunctioningofthesystem.Itshouldbemeasurable,ideally haveabioticbasisandfulfiltherationale,‘ifcontrolvariableX exceeds(orisbelow)thresholdA,thenthebiomeisintactand functioning’.Thethresholdmightbebasedonlevelsobserved inrelativelypristineenvironments,ormightbedefinedwith respecttosomekeyattribute(e.g.permafrostinthetundra,or stabilityincoralreefs).Theresponsevariablerepresentsthe securityoffunctioningwithrespecttochangesinthecontrol variable.Insomecasesthismightbethelong-termprobability ofstayingwithinthecriticalthresholdofthecontrolvariable giventhepresentvalueofacontrolvariable.

Anexamplebasedoncoralreefs

Thisapproachisparticularlyrelevantforbiomesthatexperi- encesignificantlong-termimpactsfromslowdriverssuchas subsidence or climate change. Carbonate budgets in coral reefsareanexampleofabiome-levelcontrolvariablewith thresholdswhich,ifcrossed,resultinthelossofthatbiome (Kennedyetal.,2013).Thecarbonatebudgetreflectswhether theprocessesthatconstructareefhabitat,suchasthecalcifi- cationoflivingcorals,outweighthosethaterodethehabitat, suchastheboring actionof sponges.If areef’s carbonate budgetremainsnegativethenthereefeventuallyerodesaway causing a directloss of calcifying taxa, an indirectloss of biodiversity associated with reef habitat, and a decline in ecosystemfunctions,suchasfisheriesproductivity.Thus,a singlemetric,thecarbonatebudget(kgCaCO3m ²y ¹)serves asaproxyforthemaintenanceofbiodiversityandmuchofthe functionofanentireecosystem.Carbonatebudgetsaremea- surableinthefield(Perryetal.,2013)andcanbelinkedtoboth climate change andlocalmanagement measures(Kennedy etal.,2013).Thusforcoralreefs,thebalancebetweengrowth anddissolutionofthecalciumcarbonateskeletonsofthereef- buildingorganisms is anexample of abiome-level control variablewiththresholdsthat,ifpersistentlycrossed,resultin thedegradationandeventuallossofthatbiome.Inthisexam- ple,thethresholdlevelofcontrolvariable mightbe, say,a carbonate budgetof1kgm ²y ¹.However,withrisingsea temperatureandincreasedoceanacidification,areefattaining thisthresholdtodaymightfailtomaintainthatthresholdinthe longterm.Inthiscase,ahigherthresholdmightbeneeded todayinorderto sustainecosystem functioninto thenext century (e.g., 3kgm ²y ¹). The response variable might thereforebecalculatedastheprobabilityofachievingacar- bonatebudgetof1kgm ²y ¹intheyear2100.

Whiledatacurrentlydonotexisttopopulatethisboundaryacross biomes,asimulationoffuturecarbonatebudgetsofCaribbean reefsundervariousscenariosoflocalmanagementaction,major changesinbiodiversity(recoveryofkeystonespecies),andgreen- housegasemissions(Kennedyetal.,2013)showsthatrecoveryof keyspecies(fromregion-wideepizootics)whileimportant,hadto becomplementedbylowemissionsscenarios(RCP2.6fromIPCC AR5)andlocalmanagementofpollutionandfishing.Thisexam- ple highlights the potentialimplementation of such abiome boundary,aswellastheimportanceofboundaryinteractions indeterminingasafeoperatingspace.

intoabiome’sfunctioningovertheHolocene.Biomesecuritycould thenbeaggregatedtoaglobal-scaletoencompassthefunctioning ofallmajorbiomes(Fig.2).Moresophisticatedmeansofscalingup mightfirstdisaggregatescoreswithineachbiomeintocategories of degradation (highly degraded, partly degraded, etc.). Global aggregationmight then include the distribution of the Earth’s surface that falls within a particular category overall (e.g., percentageofEarthbiomescategorisedashighlydegraded).

The advantage of this approach is that it is a system-based approach,fullyconsistentwiththeplanetaryboundariesconcept, with clear relevance for environmental stability in terms of maintainingasafeoperatingspaceoftheEarthsystem.Itprovides asub-globalscalethatismeaningfulintermsofhumanpressures, biophysicalresponsesandtheconsequencesforpeople,andwould appropriatelyincludelossofspeciesrichnessorextinctionratesas aresponsevariable,ratherthanasadriverascurrentlyformulated.

Thisdealswithonepotentialdisadvantagetothisapproach(thatit mightmissthevariabilitycomponentofbiodiversityatspeciesand organismlevels).Howeverthelinktomanagementinterventions anddecision-supportatthelocalscale,couldberelativelyweak.

While genetic and functional diversity metrics and their thresholdsappearnottofunctionorscaleupasglobalbiodiversity boundaries,thepossibilityof abiosphere integrity boundaryat biomescalesappearsapromisingavenuefordetermininghuman wellbeingat globaland sub-global scales.Furthermorethere is plentyofevidencethatthestateofthebiosphereiscriticalforthe safeoperatingspaceconceptualisedbyRockstrometal.(2009b) throughmechanisticinteractionswiththeothereightplanetary boundaries.

4. Thebiodiversityboundaryandotherplanetaryboundaries

Thebiodiversityboundaryinteractswithallotherboundaries andcouldbeframedexplicitlyasaresponsevariabletochangesin otherboundaries(e.g.oceanacidification),orasacontrolvariable forothers(e.g.climatechange).Fig.3presentsaschematicofthe magnitudeoftheeffectofagivenchangeinbiodiversitystatefrom its current position, on the position of the other planetary boundariesinrelationtothestateoftheirindicatorvariables. A positivefeedbackexistswithmostinteractionsbetweenboundary types.Asbiodiversity lossmoves closertoitsown boundaryit reducestheconditionofothers,movingthemclosertotheirown boundarieswithfeedbacksontobiodiversity(seeFig.3fordetails).

Thisself-amplifyingperturbationwillpushthecoupledsystemat anacceleratingrateintoanewstate(thusconstitutingapotential tippingpoint)ifthefeedbackisbothpositiveandstrong.

Fig. 3 shows that the net effects from all the interactions involving biodiversity are weakly positive. Biodiversity has no knownthresholdclosetoitscurrentstate,butdoeshaveapositive feedbackonitselfduetotheinterdependenciesbetweenspecies.

Forinstance,atrophiccascademaycauselossofspeciesatone trophiclevel(orfunctionalgroup)toresultinafurtherlossatother trophiclevels(orfunctionalgroups).Theeffectsofbiodiversityloss have a weakly aggravating effecton theproximityof all other boundaries to their thresholds (mostly because the loss of biodiversity-based adaptive capacity brings those boundaries closer).Conversely,theeffectsofexceedingtheoceanacidification, land use, climatechange, nutrientsand water boundarieshave large impacts on biodiversity since theyareprimary drivers of Fig.2.Hypotheticalexampleofhowtheaggregationofbiome-specificmetricscouldcontributetoaglobalstatisticonthesecurityofbiomefunction.Eachbiomeis consideredindependentlywithitsowncontrolandresponsevariablereflectingrelevantpressuresandresponses(seetext).Anoverallreportwouldhighlightthosebiomes closesttothresholdsthatrelatedtotheirownsecurityofformandfunction.TheinterpretationinthisexamplemightbethattherewasaparticularriskinArctictundra becauseofitslowoverallsecurityscore(60%)andthat,incommonwithcoralreefsitisclosetoathreshold.

biodiversity change. However, to the best of our knowledge, exceedinganyoftheseboundariesdoesnottriggerabiodiversity feedbackstrongenoughtoprecipitateathresholdcrossing,oran acceleratingcascadeinbiodiversityloss,althoughsuchscenarios canbeenvisaged(Barnoskyetal.,2012).

Ontheotherhand,biome-levelbiodiversitychangescanhave major feedbacks to climate change, water cycles and nutrient cycles.Forexample,ongoingregimeshiftsinArctictundrabiomes willlikelyresultinbiodiversity-mediatedfeedbackstotheclimate systemthroughmassivereleasesofgreenhousegasesandchanges in albedo (Leadley et al., 2010; Myers-Smith et al., 2011);

degradation of tropical forest biomes can alter precipitation patterns and river flow at sub-global scales (Davidson et al., 2012); and biome degradation often alters the capacity of ecosystemstoretainNandP(Clowetal.,2011).Whilesomeof thesebiome-levelbiodiversitychangesaredrivenbyandtherefore confounded with land use conversion to croplands (land use boundary), they are also being driven by other anthropogenic impacts on biomes, including climate change, use of fire, and logging,thatarenotincludedinthelanduseboundary(Anderegg etal.,2013;Davidsonetal.,2012;Leadleyetal.,2010).

Thebiodiversityboundarymaybemostsignificantthroughits interactionswithotherplanetaryboundarieswhenviewedatthe biomelevel.Thereareperhapsadditionalsignificantconnections betweenbiodiversitychangeatbiomelevelsandotherplanetary boundaries,and we recommend developing this understanding further as these interactions may suggest more urgent and important boundaries and may prove more useful for policy interventionsthanbiodiversityalone.

5. Conclusions

The planetary boundary for biodiversity as proposed by Rockstrometal.(2009b)wasbasedonglobalspeciesextinction rates, a metric of iconic significance in traditional biodiversity measurement,andonethatactsasbothacauseandconsequence ofglobalchange.However,thelackofwell-established,universal, scale-able orappropriate relationships and thresholds prevents

this metric fromeffectively defining a safeoperating space for humanity. It is rather the abundance, diversity, distribution, functionalcompositionandinteractionsofspeciesinecosystems thatunderliepersistentandproductivelifesupportsystemsand whichprovidetheconceptualbasisforthethreeproposedmetrics onwhichwesuggestaboundarywouldbebetterbased:ameasure ofphylogeneticdiversityrepresentingthegeneticlibraryoflife;

functional-diversity;andbiomeconditionandextent.Focusingon these aspects of biodiversity and their implications for human welfareofferspromisinglinesofenquiryforfutureresearchwith new data streams becoming available with which to develop metricsandthresholdsandtestthem.Noneoftheapproachesyet providesanoperationaldefinitionofaglobalboundaryatpresent, butthefirsttwocouldshowthresholdeffectsatlocalandregional scales while the third could represent a global-scale planetary boundary. The genetic library and the functional diversity approaches proposed here are both likely to show more and differentthresholdslocally,andtoexhibitsmootherandpossibly muchshallower responsesat global level.The biome approach allows for sub-global boundaries that aremeaningful but have weak linkstothediversitycomponentof biodiversityreflecting instead the role of biosphere process in determining a safe operatingspace.Theboundariessetprecautionarylimitstohuman perturbations in Earth system processes, avoiding potential thresholds(andmayserveashighlevelpolicytargets),butthey infactsayverylittleaboutdriversofchangeinthoseEarthsystem processesandhowtomanagethem.Thisisindeedachallengefor management and decision-makers and will require careful interpretation oftheboundaries intomanagementand practice contexts.

Thesethree metricsofferinformation atdifferenttimescales (Fig.4).Theboundarybasedonthegeneticlibraryrelatestothe long-termconsequencesforpeople, wherelossesareeffectively irreversible.Anypartoftheexistinggeneticvariationmayoneday provide unanticipated benefits, but predicting exactly which element will be essential in future is not possible. In contrast the humanbenefits related tothe biome extentand condition metric should be relevant for millennia, but changing global systems will lead over time to changes in stable, functioning biomes. Within the thousand-year time frames, however, the benefitstopeoplearerelativelypredictable.Thefunctionaltraits boundaryisprobablythemostproximateandpredictablebecause itisdefinedbytraitsknowntodaytobesignificant.

In moving forward, the conceptual basis developed here highlightstheroleofphylogeneticdiversity,functionaldiversity Fig.3.Theinteractionbetweenthebiodiversityplanetaryboundaryandother

proposedplanetaryboundaries.Asagivenfactor(i.e.boundarytype,suchas biodiversityorclimate)movesfurtherawayfromitsownsafespace,thearrows indicatechangesinthefactor(anotherboundarytype).Inallcaseswesuggestthat positivefeedbacksexist,soachangeinthefactorawayfromthesafespacewillalso movetheaffectedfactorawayfromthesafespace.Thickerarrowsdenotestronger andmorecloselyrelatedeffects.Thinnerarrowsindicateweakerandlessclosely relatedeffectswhilebrokenarrowsindicateanegligibleand/orsmallandvariable effect.

Fig.4.Roughrepresentationofthetimescalesandpredictabilityofcontributionof thethreeproposedbiodiversityboundaries.