REVIEW SUMMARY

EARTH SYSTEMS

Climate ecosystems and planetaryfutures The challenge to predict lifein Earth system modelsGordon B Bonan and Scott C Doney

BACKGROUNDEarth systemmodels (ESMs)simulate physical chemical and biological pro-cesses that underlie climate and are themostcomplex in a hierarchy of models of Earthrsquosinteracting atmospherendashlandndashoceanndashsea ice sys-tem As terrestrial andmarine ecosystems havebeen added to ESMs the distinction betweenthe physical basis for climate changemitigationand vulnerability impacts and adaptation (VIA)no longer necessarily holds The same globalchange stresses that affect terrestrial and ma-rine ecosystems are critical processes that deter-mine the magnitude and trajectory of climatechange andmany of the interventions thatmightlessen anthropogenic climate change pertainto the biosphere Here we describe environ-mental changes that are stressing terrestrialand marine ecosystems We discuss how thesestressors are being included in ESMs initiallywith an emphasis on climate processes butalso show their emerging utility for VIA an-alyses and examine them in the context ofEarth system prediction

ADVANCESTerrestrial ecosystems face stressesfrom changing climate and atmospheric com-position that alter phenology growing seasonlength and community composition thesestresses enhance productivity and water-useefficiency in some regions but also lead tomortality and increased disturbances fromwild-fires insects and extreme events in other re-gions The addition of reactive nitrogen elevatedlevels of tropospheric O3 and anthropogenicland-use and land-cover change stress ecosys-tems as well The terrestrial biosphere modelsincluded in ESMs simulate the ecological im-pacts of these stresses and their effects on Earthsystem functioning Ocean ecosystems and livingmarine resources face threats from ocean warm-ing changing large-scale circulation increasedvertical stratification declining oxygen andacidification which alter nutrient supply thelight environment and phytoplankton produc-tivity result in coral bleaching and producenovel marine communities Three-dimensionalocean models simulate the carbon cycle and

associated biogeochemistry Plankton ecosystemmodels both drive biogeochemistry models andcharacterize marine ecological dynamics

OUTLOOK The untapped potential of ESMsis to bring dispersed terrestrial andmarine eco-system research related to climate processesVIA andmitigation into a common frameworkESMsoffer an opportunity to move beyond phy-sical descriptors of atmospheric and oceanic

states to societally relevantquantities such as habitatlosswateravailabilitywild-fire risk air quality andcrop fishery and timberyields Todo so the scienceof climate predictionhas to

be extended to amoremultifaceted Earth systemprediction including the biosphere and its re-sources ESMs provide the means not just toassess the potential for future global changestresses but also to determine the outcome ofthose stresses on the biosphere Such Earth sys-tem prediction is necessary to inform soundpolicy that maintains a healthy biosphere andprovides the food energy and freshwaterneededfor a growing global population without furtherexacerbating climate change Substantial impedi-ments that must be overcome include advancingour knowledge of biosphere-related climate pro-cesses reducing model uncertainty and effec-tively communicating among rather than acrossthe disparate science communities of climateprediction global biospheremodeling VIAanaly-ses and climate change mitigation

RESEARCH

Bonan et al Science 359 533 (2018) 2 February 2018 1 of 1

The various modelsused for climate pro-jections and mitigationand VIA analysesoverlap in scope andwould benefit from abroad perspectiveof Earth system predic-tion Shown are thedomains of ESMs mitiga-tion models and VIAmodels along axes fromVIA to climate processes(horizontal) and from pri-marily serving the researchcommunity to informingsocietal needs (vertical)Panels show forests andagriculture (left) andmarine ecosystems (right)as represented acrossmodeling domains

Water use

Society

Research

ClimateVIA

CO2 removal

Forestmanagement

ESMs

Mitigation

Ecology hydrology amp economic models

Timber amp forest products

Carbonstorage

Habitat loss

Forest fires

Tree mortality

Albedo

Roughness

Evapotranspiration

Carbon

Reactive nitrogen

BVOCs

Low O2-hypoxia

Society

Research

ClimateVIA

CO2 injection

ESMs

Mitigation

Ecology amp waterquality models

Fisheries

Diseases

Acidification

Coral bleaching

Harmful agal blooms

Coastalwetlands

AcceleratedweatheringCrop

management

No-till

Land Forest and agriculture Ocean Marine ecosystems

Crop yield

Reduced CO2 emissions

Ironfertil-ization

Ecosystems ampbiogeochemistry

Carbon cycle

Shortwave radiation

Vertical stratificationBiomass burning aerosols

The list of author affiliations is available in the full article onlineCorresponding author Email bonanucaredu (GBB)sdoneyvirginiaedu (SCD)Cite this article as G B Bonan S C Doney Science 359eaam8328 (2018) DOI 101126scienceaam8328

ON OUR WEBSITE

Read the full articleat httpdxdoiorg101126scienceaam8328

on Decem

ber 6 2020

httpsciencesciencemagorg

Dow

nloaded from

REVIEW

EARTH SYSTEMS

Climate ecosystems and planetaryfutures The challenge to predict lifein Earth system modelsGordon B Bonan1 and Scott C Doney2

Many global change stresses on terrestrial and marine ecosystems affect not onlyecosystem services that are essential to humankind but also the trajectory of futureclimate by altering energy and mass exchanges with the atmosphere Earth system modelswhich simulate terrestrial and marine ecosystems and biogeochemical cycles offer acommon framework for ecological research related to climate processes analyses ofvulnerability impacts and adaptation and climate change mitigation They provide anopportunity to move beyond physical descriptors of atmospheric and oceanic states tosocietally relevant quantities such as wildfire risk habitat loss water availability and cropfishery and timber yields To achieve this the science of climate prediction must beextended to a more multifaceted Earth system prediction that includes the biosphere andits resources

Human activities are transforming Earthrsquosatmosphere ocean and land surfaces at ascale and magnitude not previously seenduring the past several thousand years ofhuman history These changes threaten

healthy planetary functions and socioeconomicwell-being (1 2) Fossil fuel combustion indus-trialized agriculture urbanization and other facetsof modern human societies are changing climateand atmospheric compositionmelting permafrostglaciers ice sheets and Arctic sea ice raising sealevels warming and acidifying the oceans pollut-ing air water and soils altering biogeochemicalcycles and freshwater availability increasing thecycling of reactive nitrogen reducing forest coverand degrading land and destroying habitats andreducing biodiversity (3ndash5) The ecological con-sequences of these changes are apparent in in-dividual organisms the communities they inhabitand the ecosystems in which they function (6ndash8)The interconnectedness and global scope of

this changing environment have transformedthe scientific study of Earth as a system It is nowunderstood that climate change must be studiedin terms of a myriad of interrelated physicalchemical biological and socioeconomic pro-cesses This broadening basis for climate changeresearch underlies the transformation fromglobalclimate models to Earth system models (ESMs)Thesemodels have shown that the biosphere notonly responds to climate change but also directlyinfluences the direction and magnitude of cli-mate change Terrestrial andmarine ecosystemsand their uses by humans are fundamental to

addressing the climate change problem Howdowe provide the food energy and fresh waterneeded for a growing global population withoutfurther exacerbating climate change Can terres-trial andmarine ecosystemsbemanaged to reducegreenhouse gas emissions With the advent ofESMs climate science is no longer limited to thephysical basis for climate projections but alsoincludes projections of the biospheremdashfor exam-ple regarding carbon storage on land and in theocean forest dieback wildfires crop yield andfisheries and marine resourcesHowever the study of climate change is still

often parsed into separate activities of observingchanges and deducing causes (3) assessing thevulnerability impacts and adaptation (VIA) ofnatural and human systems to these changes(6 7) and determining the socioeconomic trans-formations needed to mitigate them (9) Theuntapped potential of ESMs is to bring thesedispersed activities into a common frameworkThere has been success for example in coordi-nating climate projections with the integratedassessment models that identify the societal trans-formations needed tomitigate climate change (10)and even some initial attempts at directly couplingESMs and integrated assessment models (11) Asterrestrial andmarine ecosystemshavebeen addedto ESMs the distinction between the physical basisfor climate change VIA andmitigation no longernecessarily holds Land-use and land-cover changefor example is driven by socioeconomic needs forfood fiber and fuel but is also an ecological prob-lem that alters habitat and biodiversity and ameansto mitigate anthropogenic CO2 emissions (4)In this Review we discuss the treatment of the

biosphere in ESMs considering terrestrial andmarine ecosystems as they are now representedin themodels exploring how ESMs can be used

to study the biosphere and highlighting oppor-tunities for future research We then describeenvironmental changes that are occurring glob-ally and that are stressing terrestrial and marineecosystems and show how these stresses are in-cluded in ESMs in the past primarily with anemphasis on climate processes but now with ad-ditional utility for VIA and mitigation researchLast we examine these stresses in the context ofEarth system prediction Our list of stressors isnot meant to be exhaustive Rather we highlightseveral key stressors and their coincidence amongclimate processes VIA andmitigationwith the goalof initiating a dialog among the scientific commun-ities that study climate change This Review istimely because it identifies synergies across theclimate and ESM research communities involvedin the next CoupledModel Intercomparison Project(CMIP6) (12) which provides an unparalleled op-portunity tomodel and analyze the Earth system

Earth system models

ESMs simulate physical chemical and biologicalprocesses that underlie climate They are themost complex in the ongoing evolution of globalmodels of Earthrsquos atmosphere ocean cryosphereand land (Fig 1) Climate models focus on thephysical climate system as represented by atmo-sphere ocean and sea ice physics and dynamicsand land surface hydrometeorology In climatemodels land and ocean are coupled with theatmosphere through energy andmomentum fluxesand the hydrologic cycle ESMs have the samerepresentation of the physical climate system butadditionally include the carbon cycle terrestrialand marine ecosystems and biogeochemistry at-mospheric chemistry and natural and humandisturbances ESMs typically couple distinct com-ponentmodules for land atmosphere and oceanphysics and ecosystem dynamics and biogeo-chemistry are embedded into these modulesA prominent feature of ESMs is their inclusion

of the biosphere and abiotic interactions thattogethermake up an ecosystem On land terres-trial ecosystems are represented in ESMs by thetype of vegetation the amount of leaf area thestomata on leaves and carbon and nitrogen pools(13) Similarly ESMs simulate oceanphytoplanktonproduction of chlorophyll that influences thevertical profile of light absorption in the upperocean which in turn affects model sea surfacetemperature andmixed layer dynamics as wellas large-scale ocean circulation heat transportand climate variability (14 15)Biogeochemical cycles were added to ESMs

because of the potential for large climate feed-backs arising from the carbon cycle Terrestrialecosystems and the ocean together absorb aboutone-half of the annual anthropogenic CO2 emis-sions (16) but the future efficacy of these sinksis uncertain (17) Biogeochemical processes on landencompass spatial scales from leaves to plant cano-pies and from ecosystems to landscapes to biomes(13) Temporal scales include near-instantaneousphysiological responses (eg stomatal conduct-ance photosynthesis and respiration) to prevailingenvironmental conditions the seasonal emergence

RESEARCH

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 1 of 9

1National Center for Atmospheric Research (NCAR) BoulderCO 80307 USA 2Department of Environmental SciencesUniversity of Virginia Charlottesville VA 22904 USACorresponding author Email bonanucaredu (GBB)sdoneyvirginiaedu (SCD)

on Decem

ber 6 2020

httpsciencesciencemagorg

Dow

nloaded from

and senescence of leaves and changes in eco-system structure and biogeography over decadesand centuries in response to natural disturbances(eg wildfires) anthropogenic disturbances (egland-use transitions) and climate change Ongoingmodel development aims to more authenticallyrepresent plant demography and life history char-

acteristics using cohorts of individuals of similarfunctional traits in vertically structured plantcanopies (18)The three-dimensional carbon cycle models

used to estimate ocean uptake of anthropogenicCO2 evolved from model tracer studies of oceanphysical circulation Biogeochemical models ad-

ditionally track natural cycling of inorganic carbonalkalinity macronutrients (nitrogen phosphorusand silicon) and often O2 net organic matterand CaCO3 production and export from the sur-face ocean particle sinking and respiration andremineralization at depth and air-sea CO2 (and O2)gas exchange (19) Plankton ecosystem models

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 2 of 9

SST

Sub-surface drainage

Solarinput

Solarinput

Glaciers

Ocean-atmosphereexchanges

Sea ice

Biosphere-atmosphereexchanges

Runoff

Snow

Wetlands

Deforestation

BVOCsDustSmoke

CO2 CH4 chemistry aerosols

GPP

RA

RH

Permafrost

Nr

Soil carbon

CO2 CO2

Large phytoplankton

Small phytoplankton

Bacteria Microzooplankton

Zooplankton

Consumers

Org

anic

car

bon De

ep w

ater

form

atio

nVe

ntila

tion

(upw

ellin

g)

Deep ocean

Bacteria

Deep consumers

Sea floor

Surface ocean

3700

m10

0 m

Agriculture

AfforestationCO2

CO2

CO2

CO2

CH4

Runoff

FeNr

Ocean-atmosphereexchanges

SST

Sea ice

Biosphere-atmosphereexchanges

Infiltration

Runoff

Groundwater

Soil water

Infiltration

Groundwater

Soil water

Sub-surface drainage

EnergyH2O momentum

EnergyH2O momentum

EnergyH2O momentum

EnergyH2O momentum

Runoff

Climate model

Earth system model

Snow

Fig 1 Representation of the biosphere in Earth system models (ESMs) The toppanel shows land and ocean as included in climate models and the bottom panel showsthe additional processes included in ESMs ESMs simulate atmospheric CO2 in responseto fossil fuel emissions and terrestrial and marine biogeochemistry Some ESMs alsosimulate atmospheric chemistry aerosols and CH4 Terrestrial processes shown on theleft side of the diagram include biogeophysical fluxes of energy water and momentumbiogeochemical fluxes the hydrologic cycle and land-use and land-cover change (13)The carbon cycle includes component processes of gross primary production (GPP)autotrophic respiration (RA) litterfall heterotrophic respiration (RH) and wildfire Carbonaccumulates in plant and soil pools Additional biogeochemical fluxes include dustentrainment wildfire chemical emissions biogenic volatile organic compounds (BVOCs)the reactive nitrogen cycle (Nr) and CH4 emissions from wetlands Ocean processes are shown on the right side of the diagram Physical processesinclude sea ice dynamics ocean mixing and circulation changes in sea surface temperature (SST) and ocean-atmosphere fluxes The gray shadedarea depicts the marine carbon cycle consisting of the phytoplankton-based food web in the upper ocean export and remineralization in the deep seaand sediments and the physiochemical solubility pump controlled by surface-deep ocean exchange (100)

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

that simulate interactions of phytoplankton zoo-plankton nutrients and detrital pools arose bothto drive biogeochemistry models and to character-ize marine ecological dynamics such as seasonalphytoplankton blooms Recent biogeochemicaldevelopments include incorporation of iron andother trace elements (20) iron limitation being amajor controlling factor for phytoplankton growthin much of the ocean more sophisticated treat-ment of marine biological nitrogen fixation anddenitrification (21) coastal inputs of nutrientsand ocean acidification Major model advancesunder way involve expansion of plankton biologicalcomplexity to incorporate functional groups trait-based dynamics and biodiversity (22ndash24) andefforts to integrate simulated plankton produc-tivity with fisheries catch (25)An active research frontier for ESMs is incor-

poratingmore extensive chemistry-climate inter-actions Additional reactive nitrogen affects climatethrough enhanced terrestrial carbon storage emis-sions of N2O and chemical reactions that deter-mine the amount of tropospheric O3 CH4 andaerosols (5 26) Atmospheric deposition of nitro-gen to the surface ocean can enhance biologicalproductivity in low-nutrient subtropical regionsglobally however marine biogeochemistry maybe more sensitive to anthropogenic iron deposi-tion (27 28) Increased concentrations of tropo-spheric O3 decrease plant productivity and reducethe terrestrial carbon sink but the appropriateway to parameterize this in models is uncertain(29 30) Emissions of biogenic volatile organiccompounds from terrestrial ecosystems influenceatmospheric concentrations of O3 CH4 and sec-ondary organic aerosols (31) Wetlands are an im-portant source of CH4 as are permafrost soils andhydrates Global models of wetlands and CH4

emissions are being developed (32) and someESMs include methane chemistry in their cli-mate projections (33) In the ocean more di-verse biogeochemistry is needed (eg trace gasessuch as dimethyl sulfide) inmodels to link ocean-atmosphere chemistryNatural and human disturbances are a con-

tinuing research priority for ESMsWildfires affectclimate and air quality through emissions of long-lived greenhouse gases short-lived reactive gasesand aerosols and by altering surface albedo (34)Wildfires are included in ESMs but our ability tomodel the precise details of fire regimes is limited(35) Themountain pine beetle epidemic in west-ern North American forests has reduced terres-trial carbon uptake (36) increased surface albedo(37) and warmed the surface by reducing evapo-transpiration (38) Efforts to represent insect out-breaks in ESMs are promising but still nascent(39) Human disturbances include land-use tran-sitions (eg deforestation reforestation farmabandonment and shifting cultivation) andwoodharvest (40) Global models of crop growth areincluded in ESMs but specific cultivars time ofplanting crop rotation and other managementpractices are lacking (41 42) Nor is forest man-agement and commercial timber production in-cluded despite having an effect on temperaturesimilar to that of land-cover change (43)

Planetary stresses and climate feedbacksThe inclusion of terrestrial and marine ecosys-tems in ESMs enables study of the global changestresses on the biosphere and feedbacks with cli-mate change (Table 1) A prominent signal on landis a ldquogreeningrdquo of the biosphere although this ispartially countered by increased tree mortalityand disturbance Novel community assemblagesare likely to emerge that depend on the magni-tude and rate of climate change and the ability ofspecies to adjust to these changes through dis-persal Marine ecosystems also face numerousthreats fromclimate change (44ndash46) Surface oceanwaters are warming globally and freshening athigh latitudes together these trends act to in-crease vertical physical stratification resulting inaltered regional patterns of nutrient supply lightenvironment and phytoplankton productivityOcean warming leads to shifts in plankton sea-sonal phenology and polewardmigration of plank-ton invertebrate and fish species coral bleachingand sea ice loss in polar marine ecosystems Cli-mate change is projected to alter the spatial pat-terns and size of marine wild-caught fisheries (47)and may potentially change marine disease out-breaks (48) Marine communities and ecosystemsalsomay be reorganizing into novel assemblagesrequiringmore sophisticated ESMs that can trackin more detail the effects of warming on planktoncommunity composition and trophic interactions (49)Human activities imperil ecosystems and biota

inways other thandirect climate change (Table 1)Additional reactive nitrogen alters biodiversityterrestrial freshwater andmarine biogeochemistry

andwater and air quality Anthropogenic aerosolsincrease the amount of diffuse solar radiationwhich can enhance terrestrial productivity Highconcentrations of troposphericO3 can cause stomatato close and thus decrease plant productivity andtranspirationVast areas of forests havebeen clearedover the industrial era and many of the remain-ing forests are managed or secondary rather thanold-growth primary forests About one-third of theice-free land is covered by cropland or pasturelandandmuch of the terrestrial productivity has beenappropriated for human usesMarine impacts arise from ocean acidifica-

tion owing to increasing atmospheric CO2 anddeoxygenation from climate-related circulationchanges Regional stresses particularly on conti-nental shelves and some parts of the open oceanare occurring from overfishing anthropogenicnoise seabed habitat destruction pollution andcoastal eutrophication as well as loss of coastalwetlands mangrove forests and seagrass bedsowing to development and sea level rise (50)

Vulnerability impacts and adaptation

Agriculture and food security forest and waterresources terrestrial ecosystems and fisheries andmarine ecosystems are facets of the biosphere thatsustain socioeconomic well-being Assessing theimpact of climate change on these goods andservices their vulnerability to disruption in a chang-ing climate and the adaptations needed tomain-tain their future availability is critical for informingsound climate policies (6 7) Such assessments arecommonly obtained by using climate projections

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 3 of 9

Table 1 Planetary stresses faced by terrestrial and marine ecosystems

Stress Reference

Terrestrial ecosystems

Greening of the biosphere

Earlier springtime and longer growing season (101)

Higher leaf area index (55)

Greater productivity (56 91)

Higher water-use efficiency (102)

Increased nitrogen deposition (5 26 28)

Diffuse radiation (103)

Browning of the biosphere

Tree mortality (104)

Extreme events (105)

Wildfire and insects outbreaks (36 106)

Ozone damage (29 30)

Community assemblages (107)

Land-use and land-cover change (4 40)

Human appropriation of net primary production (108)

Marine ecosystems

Vertical stratification nutrient supply and phytoplankton productivity (86 109)

Plankton seasonal phenology (46)

Coral bleaching (110)

Polar marine ecosystems and sea ice loss (111 112)

Community assemblages (46 49 113)

Acidification (114)

Deoxygenation (57)

Aerosol deposition (28)

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

to drivemodels of terrestrial ecosystems (51) cropyield (52) water availability (53) and fisheries andmarine ecosystems (54) This indirect two-stepapproach has deficiencies because archived cli-matemodel outputmay not capture variable typesor temporal resolutionneeded for someVIAmodelsand restricts the ability to study feedbacks onclimate and biogeochemistryWith inclusion of the biosphere in ESMs VIA

canbe investigateddirectly For example theESMsused to quantify future carbon-climate feedbackscan also be used in retrospective studies to assessresponses of terrestrial and marine ecosystemsto historical anthropogenic forcings and climatevariability (16 55 56) The ocean component ofESMs provides a tool for reconstructing the var-iability trends andmechanismsof historical oceanbiogeochemistry (57) and plankton dynamics (58)ESMswithhigh-resolutionoceancirculationmodelscan be used to track larval dispersal and connec-tivity among coral reefs subject to bleaching (59)ESMs provide further opportunity to move

beyond physical descriptors of atmospheric andoceanic states (such as temperature and precip-itation) to societally relevant quantities relatedto food energy and freshwater For example thecroplands in ESMs allow direct study of the im-pacts of climate change on agricultural produc-tion the vulnerability of the food supply to futureclimate disruption and adaptations to make foodproduction more resilient (42) Some ESMs in-clude urban land cover which allows study ofextreme heat waves and facilitates assessmentof heat-stress mortality in cities (60) ESMs areparticularly useful for assessing air quality andhuman health issues because of the interactionsamong agriculture wildfire nitrogen gaseous fluxesand biogenic volatile organic compounds that af-fect regional air qualityAchieving this potential requires effective com-

municationbetween the scientists developingESMsand those using climate projections to study VIA(61) One result of better collaboration would beto identify and reconcile discrepancies betweenESMs and VIA models such as are evident intheir assessments of water availability in a futureclimate ESMs account for the effects of elevatedatmospheric CO2 on stomatal conductance andevapotranspiration but many VIAmodels do notresulting in an inconsistency in projections ofwater availability (62) Other examples of processesincluded in ESMs but not VIA models are the ef-fect of O3 on stomata (29 30) and the effect ofvegetation greening and land use on runoff (63)Closer collaboration between the communitieswould help to identify capabilities relevant toVIA define impact-relevant metrics that ESMsshould produce and develop data sets and pro-tocols for validation of simulated impacts (61)ESMs remain just one of several means to

study VIA A suite of specialized research toolsincluding statistical models and process-basedcrop ecosystem and hydrologymodels is required(6 7) Thesemodels have the advantage that theycan be run at the fine-scale spatial resolutionneeded to inform decisionmaking Also they areless computationally expensive than ESMs and

can therefore be used in an ensemble of simu-lations to assess uncertaintyESMs cannot yet represent the rich ecological

detail needed to capture spatial heterogeneity atlocal scales Similarly the ocean ecosystemmodelsused in ESMs typically incorporate the lowesttrophic levels of the marine food web (phyto-plankton herbivorous zooplankton) and have onlya limited representation of biodiversity OftenESMs lack the ecological complexity required topredict outcomes in higher trophic levels andfisheries The spatial resolution of global modelsis too coarse to capture regional dynamics of highlyproductive coastal ecosystems and coral reefs andmodels are just beginning to incorporate adequateland-ocean connectivity to assess nutrient eutroph-ication water quality and harmful algal blooms(64) Variable-resolution globalmodels with a hor-izontal resolution that refines froma 1deg global gridto a regional 0125deg (14-km) grid help to bridgethe gap between coarse-scale ESMs and the finerscales needed for VIA research (65 66)

Climate change mitigation

Reducing the sources and enhancing the sinks oflong-lived greenhouse gases are the most directmeans tomitigate anthropogenic climate change(67) However many interventions that mightreduce greenhouse gas emissions affect the bio-sphere and have other effects on climate andecosystem services Afforestation reforestationor avoided deforestation for example enhance theterrestrial carbon sink but also warm climate an-nually by decreasing surface albedo cool climatethrough evapotranspiration and turbulentmixingwith the atmosphere and have additional effectsthroughatmospheric chemistry andaerosols (1368)These biogeophysical effects can counter the car-bon mitigation benefits of forests so that evenmore extensive forested land may be required toachieve climate stabilization at a target that avoidsdangerous climate change (eg 2degC) ESMs are animperfect but necessary tool to study the net cli-mate effects of forests (68)Agriculture is another example of the need to

consider mitigation in an ESM context Efficientapplication of nitrogen fertilizer tillage and othermanagement can enhance carbon storage and re-duce N2O emissions (69) Crops also affect climatethrough biogeophysical coupling with the atmo-sphere it is likely that expansion of agriculturallands over the industrial era has cooled climatebecause of these changes (70) Intensification ofagriculture is thought to have cooled summertemperatures in the Midwest United States (71)No-till agriculture can increase surface albedo andcool climate (72) and other increases in surfacealbedo may have geoengineering potential (73)Production of bioenergy for carbon capture andstorage (BECCS) can alsomitigate climate changebut land use for BECSS must be balanced by ara-ble land for food production (74) ESMs provide anecessary tool to investigate the multidisciplinaryoutcomes of BECCS for climate food energy andfresh water ESMs are also being used to deter-mine the effects on ecosystems of geoengineeringtechniques involving solar radiation modification

such as stratospheric aerosol injection cloud bright-ening and surface albedo manipulation (75 76)At present the ocean removes roughly a quarter

of anthropogenic CO2 emissions to the atmo-sphere with the magnitude modulated by chem-ical dissolution into surface seawater and thephysical rate of exchange between surface anddeep waters (16) Over the centuries-long timescales of ocean-overturning circulation an increas-ing fraction of anthropogenic CO2 will be storedin the deep ocean reservoir Several geoengineer-ing approaches have been proposed to increaseocean carbon uptake by injecting CO2 directly atdepth fertilizing phytoplankton to speed up themarine biological pump that transports organiccarbon from the surface layer into the deep oceanand accelerating weathering processes to add al-kalinity to seawater (67) A number of studieswith ocean-only models and full ESMs have ex-plored the possible efficacy of these approachesas well as the potential for ecological impactsand biogeochemical feedbacks (75) Substantialalterations to marine ecosystems could also arisefrom solar radiation modification

Earth system prediction

Atmospheric science has long embraced modelsto make predictions of near-term weather andlong-term climate ESMs enable predictions ofthe biosphere but the atmospheric-centric viewof prediction needs to be extended to the bio-logical realm In this section we introduce termi-nology and concepts specific to weather forecastsand climate prediction and then show extensionto the biosphereForecasting the weather on time scales of a few

hours to 2 weeks is a classic prediction probleminwhich amodel that describes the atmospherendashlandndashoceanndashsea ice system is stepped forward intime from initial conditions The predictabilitymdashthe capability to make a skillful forecastmdashislimited by uncertainty in the exact initial condi-tions imperfections in the model and under-standing of the underlying physics and dynamicsand the degree of randomness or chaotic behaviorin the system The same concept applies to pre-dicting climate at seasonal interannual or dec-adal time scales (77ndash79) Climate projection overseveral decadesmust consider additional long-termEarth systemprocesses such as ocean circulationice sheet melting and changes in vegetation ter-restrial andmarine biogeochemistry and humanbehavior The lattermost is particularly poorlyknown and is imposed using anthropogenicforcing scenarios At multidecadal time scalesthe exact initial state is less critical Insteaduncertainty in climate projections is largely dom-inated by the choice of an anthropogenic forcingscenario althoughuncertainties also remainwithregard to climate sensitivity and feedback pro-cesses (Fig 2)The term Earth system prediction is used to

capture this spectrum of temporal scales fromsubseasonal to multidecadal mostly in the con-text of weather and climate (77ndash81) In a broaderperspective however the scope of Earth systemprediction can be expanded to include other facets

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 4 of 9

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

of the Earth system The predictability of Arcticsea ice loss is a prominent such example (82) Asclimatemodels have evolved intomore complexESMs the predictability of biosphere states andprocesses needs to be considered jointly withthat of weather and climateThe predictive capability of a model depends

on the sources of errors in the forecast For cli-mate these errors are broadly classified into ini-tial conditions boundary conditions and modeluncertainty including both model structure andparameters (83) Uncertainty in exact initial con-ditions manifests in unforced variability internalto the climate system (also known as natural var-iability) in which small differences in initial con-ditions produce different climate trajectories Theimportance of natural variability can be assessedthrough a multimember ensemble of simulationswith a singlemodel initializedwith different statesThe second source of uncertainty is model error

seen in the model response to the imposed forc-ing scenarios Models are imperfect and differ intheir forced response owing to their spatial reso-lution and imprecision in their parameterizationof the various physical chemical and biologicalprocesses Model uncertainty is assessed throughmultimodel ensemble studiesThe third source of uncertainty is the forcing

scenarios and their depiction of the time evolutionof greenhouse gases land use and other anthro-pogenic forcings of climate which are imposedas boundary conditions to the models For tem-perature projections at the global scale modeluncertainty and natural variability dominate atnear-termdecadal time scales (10 to 30 years) (83)Scenarios are the major source of uncertainty atmultidecadal lead times Total uncertainty is larger

at regional scales mostly from natural variabilityand model structureA related concept is time of emergence Deter-

mining the time at which the forced climatechange signal emerges from the noise of naturalvariability is a necessary requirement in assess-ing when an expected change can be detected orwhether observed changes can be attributed toanthropogenic forcings (84 85) Time of emer-gence has been mostly studied for temperatureand can vary from a few decades inmid-latitudeswith low natural variability to several decades orlonger in regions with larger natural variabilityCan these concepts of predictability uncer-

tainty and time of emergence be extended tostudy the biosphere in the Earth system ESMspredict prominent changes in terrestrial andmarine biogeochemistry but only recently havethe necessary large multimodel and multimem-ber ensembles become available to distinguishthe forced response from natural variability andmodel uncertainty Such analyses give importantinsights into theuseofESMs tounderstandchangesin Earth system biogeochemistry In addition towarmer temperatures the ocean has trends ofincreased carbon uptake acidification lower O2and reducednet primary production over the nextseveral decades in the absence of mitigation (86)The forced trend in air-sea CO2 flux emergesrapidly in some ocean regions but large naturalvariability precludes detection of changes in therate of carbon uptake until mid-century or laterin many regions (87) Other aspects of ocean bio-geochemistry such as pH O2 concentration andnet primary production also have large region-ally dependent natural variability (88ndash90) Oceanacidification has a rapid time of emergence driven

by the accumulation of anthropogenic CO2 in thesurface layer and the sea surface temperaturewarming signal also emerges within a few decadesin many regions but forced changes in O2 con-centration and net primary production do notemerge from natural variability until mid- to late-century if at all (Fig 3 A to C)Time of emergence has been less studied in

the terrestrial biosphere Observational andmodel-ing analyses support an enhanced terrestrial car-bon sink arising from global change (17 56 91)Unforced variability in the land-atmosphere CO2

flux is large and precludes detection of change atdecadal time scales (92) There is considerablevariability within and among models and theforced response statistically emerges only afterseveral decades in many regions of the worldThe HadGEM2-ES model for example shows theforced signal of terrestrial carbon gain as emerg-ing from internal variability in many regions by2030 but other models show a weaker signalthat has yet to statistically emerge and even car-bon loss rather than gain (Fig 3 D to F)The various contributions to uncertainty differ

depending on the quantity of interest lead timeand spatial scale The uncertainty from naturalvariability is particularly large at small spatialscales and short lead times for pH O2 concen-tration and net primary production in the ocean(89) and air-sea carbon flux (93) By the end ofthe 21st century scenario uncertainty dominatestotal uncertainty for these states and fluxes at theglobal scale (except for net primary production)but natural variability and model uncertainty re-main large at the regional or biome scale Simu-lations of the terrestrial carbon cycle are muchmore variable among models and largely domi-nated bymodel uncertainty (94) Comparisons ofocean and land carbon cycle projections point toa markedly different assessment of uncertainty(Fig 4) For ocean carbon uptake model uncer-tainty is initially large but scenario uncertaintydominates during the latter part of the 21st centuryIn contrast model structure contributes 80 oftotal uncertainty throughout the 21st century forthe terrestrial carbon cycleMuch of the study of Earth system prediction

is focused on climate and decadal climate predic-tion is particularly focused on the dynamics andthermal characteristics of the ocean because ofits prominent role in climate variability Whensoil moisture and vegetation are considered byatmospheric modelers it is often in the contextof how these affect climate predictability ratherthan whether they can be predicted (79) In aglobal change perspective ecological predictionsare as essential as those of the physical climatesystem ESMs provide the means not just to as-sess the potential for future stresses engenderedby a changing climate but also to determine theoutcome of those stresses on crop yield treemortality fisheries and other aspects of thebiosphere For example characterizing whenand where wildfires might occur would be valu-able to aid governmental agencies charged withwildfire protection Particular fire events are nearlyimpossible to forecast especially because somany

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 5 of 9

Scenarios

Climatefeedbacks

Internalvariability

Ecosystemimpacts

Initialization Subseasonal to seasonal forecast(2 weeks ndash 12 months)

Decadal prediction(1 ndash 30 years)

Earth system projection (30 ndash 100+ years)

Initial value problem

Boundary value problem

Earth system model

Sources of uncertainty

Initial condition

Model uncertainty

Scenario uncertainty

Fig 2 Schematic depiction of Earth system prediction of the biosphereThe synergies betweenclimate feedback processes internal climate variability and ecosystem impacts determine modeloutcomes Subseasonal to seasonal forecasts and decadal climate prediction are initial value problemsEarth system projections are a boundary value problem driven by anthropogenic forcing scenariosUncertainty arises from inexactness of initial conditions model imperfections and scenarios

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

fires are caused by people but fire behavior onseasonal time scales can be forecast on the basisof relationships with sea surface temperatures(95) Prediction of future fire behavior requiresmodels that accurately depict fire occurrenceand severity but wildfire prediction will alsorequire an understanding of the predictabilityof the climate that drives fire behavior A similarargument pertains to crop yieldmarine resourcesand dust emissions simulated in the current gen-eration of ESMs and to forestmortality and habitloss that will be simulated by the next generationof models Such forecasts may be particularly rel-evant at the subseasonal to seasonal time scale (79)

Research needs

As this Review highlights the biosphere is centralto understanding why and how the Earth systemis changing and to adapting to and mitigatingfuture changesMany of the global change stress-ors that terrestrial and marine ecosystems face

need to be understood not only for their impactson ecosystem services that are essential to human-kind but also as processes that affect the mag-nitude and trajectory of climate change A strategyis needed to extend the study of subseasonal andseasonal forecasts and decadal climate predictionto a moremultifaceted Earth system predictionincluding the biosphere and its resources The ex-tension of seasonal to decadal climate forecasts toliving marine resources for example has consid-erable potential to aid marine management (96)A similar extension to terrestrial ecosystemswouldaid land resource managementToward this goal this Review has presented

several pathways to further define Earth systemprediction First is continued advancement ofterrestrial and marine science in light of climateprocesses and the many ways in which thebiosphere influences climate A prominent exam-ple is the carbon cycle its feedback with climatechange and whether terrestrial and marine eco-

systems can be purposely managed to mitigateanthropogenic CO2 emissions There is consid-erable uncertainty in ocean carbon cycle pro-jections particularly at regional or biome scales(89 93) and land carbon model uncertainty pre-cludes distinguishing among various alternativescenarios (94) Moreover if planting forests andbiofuels are essential to maintaining atmosphericCO2 concentrations within some planetary warm-ing target how confident are we in our abilityto know the net climate outcome of these pol-icies (68)The current generation of ecosystem models

are abstractions of complex systems Many eco-logical and biogeochemical processes are rep-resented but the challenge of representing thebiospheremdashwith its rich diversity of life formstheir assemblage into communities and ecosys-tems and the complexity of ecological systemsmdashisdaunting as is evident in large model uncertaintyin terrestrial carbon cycle projections Theoretical

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 6 of 9

Fig 3 Ocean and land forcedtrends relative to internal vari-ability and model uncertaintyData are from a range ofESMs contributed to CMIP5(A to C) Multimodel time ofemergence for SST O2 and netprimary production (NPP) (89)Time of emergence is defined asthe year at which the signalexceeds the noise which as usedhere includes both internal vari-ability and model uncertainty Theforced SSTsignal emerges rapidlyin many locations O2 time ofemergence is regionally variableand the forced NPP signal doesnot statistically emerge by 2100(D to F) Signal-to-noise ratio forcumulative land carbon uptake ina business-as-usual scenario at2030 for three different ESMs(92) Positive (negative) valuesindicate carbon gain (loss) Inthese panels the noise is strictlyinternal variability and a ratiogreater than 2 or less than ndash2indicates that the signal hasemerged from the internal varia-bility There are considerable dif-ferences among models in thesign of the terrestrial carbon fluxand whether the change hasemerged from natural variabilityby 2030 CCSM4 CommunityClimate System Model version 4HadGEM2-ES Hadley CentreGlobal Environmental Modelversion 2 CanESM2 second-generation Canadian EarthSystem Model

F

D CCSM4

HadGEM2-ES

CanESM2

Signal-to-noise ratio

-30 -15 -8 -4 -2 2 4 8 15 30

NPP

A SST

C

B O2 E

2020

gt2095

lt2015

2090

2030

2040

2050

2060

2070

2080

Year

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

advances are needed but theremay be a limit tohowmuchmodel uncertainty can be reduced (94)More complexity does not necessarily lead to bet-ter predictions or reduce uncertaintyA second pathway is to better integrate ESMs

and VIA models The gap between models arisesfromdisciplinary expertise (atmospheric and oceansciences for ESMs and hydrology ecology biogeo-chemistry agronomy forestry andmarine sciencesfor VIA models) but effective communicationamong rather than across disciplines is not trivialThere are also pragmatic considerations partic-ularly with regard to spatial scale and processcomplexity that limit collaboration between globalESMs and VIA models with a more local to re-gional domain However just as the science ofEarth systemprediction is seen as ameans to unitethe weather and climatemodeling communities(8081) so too can the broadening ofEarth systemprediction to include the biosphere stimulate col-laborations with the VIA communityA third promising researchpathway is to expand

the concepts andmethodology of seasonal to dec-adal climate prediction to include terrestrial andmarine ecosystems and to quantify prediction un-certainty at spatial and temporal scales relevantto stakeholders The predictability of the terres-trial carbon cycle can be considered from an eco-logical perspective (97) but only recently has itbeen considered in an Earth system perspectiveof natural climate variability the forced climateresponse and model uncertainty (92 94) Anal-ysis of natural variability model uncertainty andscenario uncertainty is similarly informingmarinebiogeochemistry (87ndash90 93) Whether the bio-sphere is a source of climate predictability is notnecessarily the right question to pose A morefruitful research pathway may be to investigatehow to predict the biosphere and its resourcesin a changing environment as identified specif-ically for marine living resources (96) and con-sidered also for atmospheric CO2 (98) Initialcondition uncertainty and the difficulty in separat-ing natural variability from the forced trend likelyproduces irreducible uncertainty in climate pre-diction (99) At the regional or biome scale nat-ural variability is large for the ocean and land

carbon cycles (89 92 93) Whether a similar ir-reducible uncertainty manifests in terrestrial andmarine ecosystems remains to be exploredWith their terrestrial and marine ecosystems

biogeochemical cycles and simulation of plantsmicrobes and marine life ESMs challenge ter-restrial and marine ecologists and biogeochem-ists to think in terms of broad generalizationsand to find the mathematical equations to de-scribe the biosphere its functioning and its re-sponse to global change ESMs similarly challengegeoscientists to think beyond a physical under-standing of climate to include biology Themodelsshowmuch promise to advance our understand-ing of global change but must move from thesynthetic world of an ESM toward the realworldBridging the gap between observations and theoryas atmospheric CO2 rises climate changesmorenitrogen is added to the system forests are clearedgrasslands are plowed or converted to pasturescoastal wetlands and coral reefs are degraded orlost andoceanswarmandare increasinglypollutedposes challenging opportunities for the next gener-ation of scientists to advance planetary ecology andclimate science

REFERENCES AND NOTES

1 J Rockstroumlm et al A safe operating space for humanityNature 461 472ndash475 (2009) doi 101038461472apmid 19779433

2 W Steffen et al Planetary boundaries Guiding humandevelopment on a changing planet Science 347 1259855(2015) doi 101126science1259855 pmid 25592418

3 Intergovernmental Panel on Climate Change (IPCC) ClimateChange 2013 The Physical Science Basis Contribution ofWorking Group I to the Fifth Assessment Report of theIntergovernmental Panel on Climate Change T F StockerD Qin G-K Plattner M Tignor S K Allen J BoschungA Nauels Y Xia V Bex P M Midgley Eds (CambridgeUniv Press 2013)

4 J A Foley et al Global consequences of land use Science309 570ndash574 (2005) doi 101126science1111772pmid 16040698

5 J N Galloway et al The nitrogen cascade Bioscience 53341ndash356 (2003) doi 1016410006-3568(2003)053[0341TNC]20CO2

6 IPCC Climate Change 2014 Impacts Adaptation andVulnerability Part A Global and Sectoral AspectsContribution of Working Group II to the Fifth AssessmentReport of the Intergovernmental Panel on Climate ChangeC B Field V R Barros D J Dokken K J MachM D Mastrandrea T E Bilir M Chatterjee K L Ebi

Y O Estrada R C Genova B Girma E S Kissel A N LevyS MacCracken P R Mastrandrea L L White Eds(Cambridge Univ Press 2014)

7 IPCC Climate Change 2014 Impacts Adaptation andVulnerability Part B Regional Aspects Contribution ofWorking Group II to the Fifth Assessment Report of theIntergovernmental Panel on Climate Change V R BarrosC B Field D J Dokken M D Mastrandrea K J MachT E Bilir M Chatterjee K L Ebi Y O Estrada R C GenovaB Girma E S Kissel A N Levy S MacCrackenP R Mastrandrea LL White Eds (Cambridge UnivPress 2014)

8 B R Scheffers et al The broad footprint of climatechange from genes to biomes to people Science 354aaf7671 (2016) doi 101126scienceaaf7671pmid 27846577

9 IPCC Climate Change 2014 Mitigation of Climate ChangeContribution of Working Group III to the Fifth AssessmentReport of the Intergovernmental Panel on Climate ChangeO Edenhofer R Pichs-Madruga Y Sokona E FarahaniS Kadner K Seyboth A Adler I Baum S BrunnerP Eickemeier B Kriemann J Savolainen S SchloumlmerC von Stechow T Zwickel J C Minx Eds (Cambridge UnivPress 2014)

10 R H Moss et al The next generation of scenarios for climatechange research and assessment Nature 463 747ndash756(2010) doi 101038nature08823 pmid 20148028

11 W D Collins et al The integrated Earth system modelversion 1 Formulation and functionality Geosci Model Dev 82203ndash2219 (2015) doi 105194gmd-8-2203-2015

12 V Eyring et al Overview of the Coupled ModelIntercomparison Project Phase 6 (CMIP6) experimentaldesign and organization Geosci Model Dev 9 1937ndash1958(2016) doi 105194gmd-9-1937-2016

13 G B Bonan Ecological Climatology Concepts andApplications (Cambridge Univ Press ed 3 2016)

14 C Sweeney et al Impacts of shortwave penetration depth onlarge-scale ocean circulation and heat transport J PhysOceanogr 35 1103ndash1119 (2005) doi 101175JPO27401

15 M Jochum S Yeager K Lindsay K Moore R MurtuguddeQuantification of the feedback between phytoplankton andENSO in the Community Climate System Model J Clim 232916ndash2925 (2010) doi 1011752010JCLI32541

16 C Le Queacutereacute et al Global carbon budget 2016 Earth Syst SciData 8 605ndash649 (2016) doi 105194essd-8-605-2016

17 P Friedlingstein et al Uncertainties in CMIP5 climateprojections due to carbon cycle feedbacks J Clim 27511ndash526 (2014) doi 101175JCLI-D-12-005791

18 R A Fisher et al Taking off the training wheels Theproperties of a dynamic vegetation model without climateenvelopes CLM45(ED) Geosci Model Dev 8 3593ndash3619(2015) doi 105194gmd-8-3593-2015

19 J L Sarmiento N Gruber Ocean Biogeochemical Dynamics(Princeton Univ Press 2006)

20 A Tagliabue et al How well do global ocean biogeochemistrymodels simulate dissolved iron distributions GlobalBiogeochem Cycles 30 149ndash174 (2016) doi 1010022015GB005289

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 7 of 9

Fig 4 Ocean and land carbon cycleuncertainty The percentage of total varianceattributed to internal variability modeluncertainty and scenario uncertainty inprojections of cumulative global carbon uptakefrom 2006 to 2100 differs widely between(A) ocean and (B) land The ocean carbon cycleis dominated by scenario uncertainty by themiddle of the century but uncertainty inthe land carbon cycle is mostly frommodel structure Data are from 12 ESMsusing four different scenarios (94)

Model uncertainty Scenario uncertaintyInternal variability

2020

100

80

60

40

20

0

100

80

60

40

20

02040 2060 2080 2100 2020 2040 2060 2080 2100

Perc

enta

ge o

f tot

al v

aria

nce

()

Time Time

A BOcean Land

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

21 J K Moore K Lindsay S C Doney M C Long K MisumiMarine ecosystem dynamics and biogeochemical cyclingin the Community Earth System Model [CESM1(BGC)]Comparison of the 1990s with the 2090s under the RCP45and RCP85 scenarios J Clim 26 9291ndash9312 (2013)doi 101175JCLI-D-12-005661

22 R R Hood et al Pelagic functional group modeling Progresschallenges and prospects Deep Sea Res Part II Top StudOceanogr 53 459ndash512 (2006) doi 101016jdsr2200601025

23 E Litchman C A Klausmeier Trait-based communityecology of phytoplankton Annu Rev Ecol Evol Syst 39615ndash639 (2008) doi 101146annurevecolsys39110707173549

24 M J Follows S Dutkiewicz Modeling diverse communitiesof marine microbes Annu Rev Mar Sci 3 427ndash451(2011) doi 101146annurev-marine-120709-142848pmid 21329212

25 C A Stock et al Reconciling fisheries catch and oceanproductivity Proc Natl Acad Sci USA 114 E1441ndashE1449(2017) doi 101073pnas1610238114 pmid 28115722

26 R W Pinder et al Climate change impacts of US reactivenitrogen Proc Natl Acad Sci USA 109 7671ndash7675 (2012)doi 101073pnas1114243109 pmid 22547815

27 A Krishnamurthy et al Impacts of increasing anthropogenicsoluble iron and nitrogen deposition on oceanbiogeochemistry Global Biogeochem Cycles 23 GB3016(2009) doi 1010292008GB003440

28 N M Mahowald et al Aerosol deposition impacts on landand ocean carbon cycles Curr Clim Change Rep 3 16ndash31(2017) doi 101007s40641-017-0056-z

29 S Sitch P M Cox W J Collins C Huntingford Indirectradiative forcing of climate change through ozone effects onthe land-carbon sink Nature 448 791ndash794 (2007)doi 101038nature06059 pmid 17653194

30 D Lombardozzi S Levis G Bonan P G Hess J P SparksThe influence of chronic ozone exposure on global carbonand water cycles J Clim 28 292ndash305 (2015) doi 101175JCLI-D-14-002231

31 N Unger Human land-use-driven reduction of forest volatilescools global climate Nat Clim Change 4 907ndash910 (2014)doi 101038nclimate2347

32 J R Melton et al Present state of global wetland extentand wetland methane modelling Conclusions from a modelinter-comparison project (WETCHIMP) Biogeosciences 10753ndash788 (2013) doi 105194bg-10-753-2013

33 D T Shindell et al Interactive ozone and methane chemistryin GISS-E2 historical and future climate simulationsAtmos Chem Phys 13 2653ndash2689 (2013) doi 105194acp-13-2653-2013

34 D S Ward et al The changing radiative forcing of firesGlobal model estimates for past present and futureAtmos Chem Phys 12 10857ndash10886 (2012) doi 105194acp-12-10857-2012

35 S Hantson et al The status and challenge of global firemodelling Biogeosciences 13 3359ndash3375 (2016)doi 105194bg-13-3359-2016

36 W A Kurz et al Mountain pine beetle and forest carbonfeedback to climate change Nature 452 987ndash990 (2008)doi 101038nature06777 pmid 18432244

37 B C Bright J A Hicke A J H Meddens Effects of barkbeetle-caused tree mortality on biogeochemical andbiogeophysical MODIS products J Geophys Res Biogeosci118 974ndash982 (2013) doi 101002jgrg20078

38 H Maness P J Kushner I Fung Summertime climate responseto mountain pine beetle disturbance in British Columbia NatGeosci 6 65ndash70 (2013) doi 101038ngeo1642

39 S L Edburg J A Hicke D M Lawrence P E ThorntonSimulating coupled carbon and nitrogen dynamics followingmountain pine beetle outbreaks in the western United StatesJ Geophys Res 116 G04033 (2011) doi 1010292011JG001786

40 G C Hurtt et al Harmonization of land-use scenarios for theperiod 1500ndash2100 600 years of global gridded annualland-use transitions wood harvest and resulting secondarylands Clim Change 109 117ndash161 (2011) doi 101007s10584-011-0153-2

41 S Levis et al Interactive crop management in theCommunity Earth System Model (CESM1) Seasonalinfluences on land-atmosphere fluxes J Clim 254839ndash4859 (2012) doi 101175JCLI-D-11-004461

42 S Levis A Badger B Drewniak C Nevison X RenCLMcrop yields and water requirements Avoided impacts bychoosing RCP 45 over 85 Clim Change (2016)doi 101007s10584-016-1654-9

43 S Luyssaert et al Land management and land-cover changehave impacts of similar magnitude on surface temperatureNat Clim Change 4 389ndash393 (2014) doi 101038nclimate2196

44 S C Doney et al Climate change impacts on marineecosystems Annu Rev Mar Sci 4 11ndash37 (2012)doi 101146annurev-marine-041911-111611 pmid 22457967

45 O Hoegh-Guldberg J F Bruno The impact of climatechange on the worldrsquos marine ecosystems Science 3281523ndash1528 (2010) doi 101126science1189930pmid 20558709

46 E S Poloczanska et al Global imprint of climate change onmarine life Nat Clim Change 3 919ndash925 (2013)doi 101038nclimate1958

47 A B Hollowed et al Projected impacts of climate change onmarine fish and fisheries ICES J Mar Sci 70 1023ndash1037(2013) doi 101093icesjmsfst081

48 C A Burge et al Climate change influences on marineinfectious diseases Implications for management andsociety Annu Rev Mar Sci 6 249ndash277 (2014) doi 101146annurev-marine-010213-135029 pmid 23808894

49 J G Molinos et al Climate velocity and the future globalredistribution of marine biodiversity Nat Clim Change 683ndash88 (2016) doi 101038nclimate2769

50 B S Halpern et al Spatial and temporal changes incumulative human impacts on the worldrsquos oceanNat Commun 6 7615 (2015) doi 101038ncomms8615pmid 26172980

51 J M Melillo et al Vegetationecosystem modeling andanalysis project Comparing biogeography andbiogeochemistry models in a continental-scale study ofterrestrial ecosystem responses to climate change and CO2

doubling Global Biogeochem Cycles 9 407ndash437 (1995)doi 10102995GB02746

52 C Rosenzweig et al Assessing agricultural risks of climatechange in the 21st century in a global gridded cropmodel intercomparison Proc Natl Acad Sci USA 1113268ndash3273 (2014) doi 101073pnas1222463110pmid 24344314

53 J Schewe et al Multimodel assessment of water scarcityunder climate change Proc Natl Acad Sci USA 1113245ndash3250 (2014) doi 101073pnas1222460110pmid 24344289

54 C A Stock et al On the use of IPCC-class models toassess the impact of climate on Living Marine ResourcesProg Oceanogr 88 1ndash27 (2011) doi 101016jpocean201009001

55 Z Zhu et al Greening of the Earth and its drivers Nat ClimChange 6 791ndash795 (2016) doi 101038nclimate3004

56 S Sitch et al Recent trends and drivers of regional sourcesand sinks of carbon dioxide Biogeosciences 12 653ndash679(2015) doi 105194bg-12-653-2015

57 T Ito S Minobe M C Long C Deutsch Upper ocean O2

trends 1958ndash2015 Geophys Res Lett 44 4214ndash4223(2017) doi 1010022017GL073613

58 S A Henson J P Dunne J L Sarmiento Decadal variabilityin North Atlantic phytoplankton blooms J Geophys Res 114C04013 (2009) doi 1010292008JC005139

59 J A Kleypas et al Larval connectivity across temperaturegradients and its potential effect on heat tolerance in coralpopulations Glob Change Biol 22 3539ndash3549 (2016)doi 101111gcb13347 pmid 27154763

60 G B Anderson K W Oleson B Jones R D Peng Projectedtrends in high-mortality heatwaves under different scenariosof climate population and adaptation in 82 US communitiesClim Change (2016) doi 101007s10584-016-1779-x

61 A C Ruane et al The Vulnerability Impacts Adaptation andClimate Services Advisory Board (VIACS AB v10)contribution to CMIP6 Geosci Model Dev 9 3493ndash3515(2016) doi 105194gmd-9-3493-2016

62 A L S Swann F M Hoffman C D Koven J T RandersonPlant responses to increasing CO2 reduce estimates ofclimate impacts on drought severity Proc Natl Acad SciUSA 113 10019ndash10024 (2016) doi 101073pnas1604581113 pmid 27573831

63 S Piao et al Changes in climate and land use have a largerdirect impact than rising CO2 on global river runoff trendsProc Natl Acad Sci USA 104 15242ndash15247 (2007)doi 101073pnas0707213104 pmid 17878298

64 Q Sun M M Whitney F O Bryan Y Tseng A box model forrepresenting estuarine physical processes in Earth systemmodels Ocean Model 112 139ndash153 (2017) doi 101016jocemod201703004

65 X Huang A M Rhoades P A Ullrich C M ZarzyckiAn evaluation of the variable-resolution CESM for modelingCaliforniarsquos climate J Adv Model Earth Syst 8 345ndash369(2016) doi 1010022015MS000559

66 A M Rhoades X Huang P A Ullrich C M ZarzyckiCharacterizing Sierra Nevada snowpack using variable-resolution CESM J Appl Meteorol Climatol 55 173ndash196(2016) doi 101175JAMC-D-15-01561

67 National Research Council Climate Intervention CarbonDioxide Removal and Reliable Sequestration (NationalAcademies Press 2015)

68 G B Bonan Forests climate and public policyA 500-year interdisciplinary odyssey Annu Rev Ecol EvolSyst 47 97ndash121 (2016) doi 101146annurev-ecolsys-121415-032359

69 P Smith et al ldquoAgriculture forestry and other land use(AFOLU)rdquo in Climate Change 2014 Mitigation of ClimateChange Contribution of Working Group III to the FifthAssessment Report of the Intergovernmental Panel onClimate Change O Edenhofer et al Eds (Cambridge UnivPress 2014) pp 811ndash922

70 N de Noblet-Ducoudreacute et al Determining robust impacts ofland-use-induced land cover changes on surface climate overNorth America and Eurasia Results from the first set ofLUCID experiments J Clim 25 3261ndash3281 (2012)doi 101175JCLI-D-11-003381

71 N D Mueller et al Cooling of US Midwest summertemperature extremes from cropland intensificationNat Clim Change 6 317ndash322 (2016) doi 101038nclimate2825

72 E L Davin S I Seneviratne P Ciais A Olioso T WangPreferential cooling of hot extremes from cropland albedomanagement Proc Natl Acad Sci USA 111 9757ndash9761(2014) doi 101073pnas1317323111 pmid 24958872

73 K Caldeira G Bala L Cao The science of geoengineeringAnnu Rev Earth Planet Sci 41 231ndash256 (2013)doi 101146annurev-earth-042711-105548

74 P Smith et al Biophysical and economic limits to negativeCO2 emissions Nat Clim Change 6 42ndash50 (2016)doi 101038nclimate2870

75 L M Russell et al Ecosystem impacts of geoengineeringA review for developing a science plan Ambio 41350ndash369 (2012) doi 101007s13280-012-0258-5pmid 22430307

76 P J Irvine et al Towards a comprehensive climate impactsassessment of solar geoengineering Earthrsquos Future 593ndash106 (2017) doi 1010022016EF000389

77 G A Meehl et al Decadal prediction Can it be skillfulBull Am Meteorol Soc 90 1467ndash1485 (2009) doi 1011752009BAMS27781

78 G A Meehl et al Decadal climate prediction An update fromthe trenches Bull Am Meteorol Soc 95 243ndash267 (2014)doi 101175BAMS-D-12-002411

79 National Academies of Sciences Engineering and MedicineNext Generation Earth System Prediction Strategies forSubseasonal to Seasonal Forecasts (National AcademiesPress 2016)

80 W Hazeleger et al EC-Earth A seamless Earth-systemprediction approach in action Bull Am Meteorol Soc 911357ndash1363 (2010) doi 1011752010BAMS28771

81 J C Carman et al The national Earth system predictioncapability Coordinating the giant Bull Am Meteorol Soc98 239ndash252 (2017) doi 101175BAMS-D-16-00021

82 A Jahn J E Kay M M Holland D M Hall How predictableis the timing of a summer ice-free Arctic Geophys Res Lett43 9113ndash9120 (2016) doi 1010022016GL070067

83 E Hawkins R Sutton The potential to narrow uncertainty inregional climate predictions Bull Am Meteorol Soc 901095ndash1107 (2009) doi 1011752009BAMS26071

84 C Deser R Knutti S Solomon A S Phillips Communicationof the role of natural variability in future North Americanclimate Nat Clim Change 2 775ndash779 (2012) doi 101038nclimate1562

85 E Hawkins R Sutton Time of emergence of climate signalsGeophys Res Lett 39 L01702 (2012) doi 1010292011GL050087

86 L Bopp et al Multiple stressors of ocean ecosystems inthe 21st century Projections with CMIP5 modelsBiogeosciences 10 6225ndash6245 (2013) doi 105194bg-10-6225-2013

87 G A McKinley et al Timescales for detection of trends in theocean carbon sink Nature 530 469ndash472 (2016)doi 101038nature16958 pmid 26911782

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 8 of 9

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

88 K B Rodgers J Lin T L Froumllicher Emergence of multipleocean ecosystem drivers in a large ensemble suitewith an Earth system model Biogeosciences 123301ndash3320 (2015) doi 105194bg-12-3301-2015

89 T L Froumllicher K B Rodgers C A Stock W W L CheungSources of uncertainties in 21st century projections ofpotential ocean ecosystem stressors Global BiogeochemCycles 30 1224ndash1243 (2016) doi 1010022015GB005338

90 S A Henson et al Rapid emergence of climate change inenvironmental drivers of marine ecosystems Nat Commun8 14682 (2017) doi 101038ncomms14682pmid 28267144

91 J E Campbell et al Large historical growth in globalterrestrial gross primary production Nature 544 84ndash87(2017) doi 101038nature22030 pmid 28382993

92 D Lombardozzi G B Bonan D W Nychka The emerginganthropogenic signal in landndashatmosphere carbon-cyclecoupling Nat Clim Change 4 796ndash800 (2014)doi 101038nclimate2323

93 N S Lovenduski G A McKinley A R Fay K LindsayM C Long Partitioning uncertainty in ocean carbonuptake projections Internal variability emission scenario andmodel structure Global Biogeochem Cycles 30 1276ndash1287(2016) doi 1010022016GB005426

94 N S Lovenduski G B Bonan Reducing uncertainty inprojections of terrestrial carbon uptake Environ Res Lett12 044020 (2017) doi 1010881748-9326aa66b8

95 Y Chen D C Morton N Andela L Giglio J T RandersonHow much global burned area can be forecast on seasonaltime scales using sea surface temperatures Environ ResLett 11 045001 (2016) doi 1010881748-9326114045001

96 D Tommasi et al Managing living marine resources in adynamic environment The role of seasonal to decadalclimate forecasts Prog Oceanogr 152 15ndash49 (2017)doi 101016jpocean201612011

97 Y Luo T F Keenan M Smith Predictability of the terrestrialcarbon cycle Glob Change Biol 21 1737ndash1751 (2015)doi 101111gcb12766 pmid 25327167

98 S M Polavarapu et al Greenhouse gas simulations with acoupled meteorological and transport model Thepredictability of CO2 Atmos Chem Phys 16 12005ndash12038(2016) doi 105194acp-16-12005-2016

99 E Hawkins R S Smith J M Gregory D A StainforthIrreducible uncertainty in near-term climate projectionsClim Dyn 46 3807ndash3819 (2016) doi 101007s00382-015-2806-8

100 S W Chisholm Stirring times in the Southern Ocean Nature407 685ndash687 (2000) doi 10103835037696 pmid11048702

101 R Buitenwerf L Rose S I Higgins Three decades ofmulti-dimensional change in global leaf phenology Nat ClimChange 5 364ndash368 (2015) doi 101038nclimate2533

102 T F Keenan et al Increase in forest water-use efficiencyas atmospheric carbon dioxide concentrations riseNature 499 324ndash327 (2013) doi 101038nature12291pmid 23842499

103 L M Mercado et al Impact of changes in diffuse radiationon the global land carbon sink Nature 458 1014ndash1017(2009) doi 101038nature07949 pmid 19396143

104 C D Allen et al A global overview of drought and heat-induced tree mortality reveals emerging climate change risksfor forests For Ecol Manage 259 660ndash684 (2010)doi 101016jforeco200909001

105 D Frank et al Effects of climate extremes on the terrestrialcarbon cycle Concepts processes and potential futureimpacts Glob Change Biol 21 2861ndash2880 (2015)doi 101111gcb12916 pmid 25752680

106 W M Jolly et al Climate-induced variations in global wildfiredanger from 1979 to 2013 Nat Commun 6 7537 (2015)doi 101038ncomms8537 pmid 26172867

107 A Ordonez J W Williams J-C Svenning Mapping climaticmechanisms likely to favour the emergence of novel

communities Nat Clim Change 6 1104ndash1109 (2016)doi 101038nclimate3127

108 H Haberl et al Quantifying and mapping the humanappropriation of net primary production in Earthrsquosterrestrial ecosystems Proc Natl Acad Sci USA 10412942ndash12947 (2007) doi 101073pnas0704243104pmid 17616580

109 C Laufkoumltter et al Drivers and uncertainties of future globalmarine primary production in marine ecosystem modelsBiogeosciences 12 6955ndash6984 (2015) doi 105194bg-12-6955-2015

110 T P Hughes et al Global warming and recurrent massbleaching of corals Nature 543 373ndash377 (2017)doi 101038nature21707 pmid 28300113

111 P Wassmann C M Duarte S Agustiacute M K Sejr Footprintsof climate change in the Arctic marine ecosystem GlobChange Biol 17 1235ndash1249 (2011) doi 101111j1365-2486201002311x

112 A J Constable et al Climate change and Southern Oceanecosystems I How changes in physical habitats directlyaffect marine biota Glob Change Biol 20 3004ndash3025(2014) doi 101111gcb12623 pmid 24802817

113 T Wernberg et al Climate-driven regime shift of a temperatemarine ecosystem Science 353 169ndash172 (2016)doi 101126scienceaad8745 pmid 27387951

114 J P Gattuso L Hansson Ed Ocean Acidification (OxfordUniv Press 2011)

ACKNOWLEDGMENTS

We acknowledge funding from the National Institute of Food andAgricultureUS Department of Agriculture (2015-67003-23485)and the NASA Ocean Biology and Biogeochemistry Program(NNX14AL86G) We thank C Tebaldi and J Kleypas (NCAR) forcomments on the manuscript and figures NCAR is sponsored bythe National Science Foundation

101126scienceaam8328

Bonan et al Science 359 eaam8328 (2018) 2 February 2018 9 of 9

RESEARCH | REVIEWon D

ecember 6 2020

httpsciencesciencem

agorgD

ownloaded from

modelsClimate ecosystems and planetary futures The challenge to predict life in Earth system

Gordon B Bonan and Scott C Doney

DOI 101126scienceaam8328 (6375) eaam8328359Science

this issue p eaam8328Sciencesome of which may be unavoidable and to better translate observations into abstract model representationscrop yields wildfire risk and water availability Further research is needed to better understand model uncertaintiesbiological aspects of the Earth system This provides insight into climate impacts of societal importance such as altered system models that include the terrestrial and marine biosphere Such models capture interactions between physical andscenarios and for mitigating and adapting to the resulting climatic changes Bonan and Doney review advances in Earth

High-quality climate predictions are crucial for understanding the impacts of different greenhouse gas emissionIntegrating the biosphere into climate models

ARTICLE TOOLS httpsciencesciencemagorgcontent3596375eaam8328

REFERENCES

httpsciencesciencemagorgcontent3596375eaam8328BIBLThis article cites 102 articles 13 of which you can access for free

PERMISSIONS httpwwwsciencemagorghelpreprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAASScienceScience 1200 New York Avenue NW Washington DC 20005 The title (print ISSN 0036-8075 online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Copyright copy 2018 American Association for the Advancement of Science

on Decem

ber 6 2020

httpsciencesciencemagorg

Dow

nloaded from