Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades

S. Sitch; P. Friedlingstein; N. Gruber; S. D. Jones; G. Murray-Tortarolo; A. Ahlström; S. C. Doney; H. Graven; C. Heinze; C. Huntingford; S. Levis; P. E. Levy; M. Lomas; B. Poulter; N. Viovy; S. Zaehle; N. Zeng; A. Arneth; G. Bonan; L. Bopp; J. G. Canadell; F. Chevallier; P. Ciais; R. Ellis; M. Gloor; P. Peylin; S. Piao; C. Le Quéré; B. Smith; Z. Zhu; R. Myneni

doi:10.5194/bgd-10-20113-2013

Sitch, S.;Friedlingstein, P.;Gruber, N.;Jones, S. D.;Murray-Tortarolo, G.;Ahlström, A.;Doney, S. C.;Graven, H.;Heinze, C.;Huntingford, C.;Levis, S.;Levy, P. E.;Lomas, M.;Poulter, B.;Viovy, N.;Zaehle, S.;Zeng, N.;Arneth, A.;Bonan, G.;Bopp, L.;Canadell, J. G.;Chevallier, F.;Ciais, P.;Ellis, R.;Gloor, M.;Peylin, P.;Piao, S.;Quéré, C. Le;Smith, B.;Zhu, Z.;Myneni, R.

2013-12-23 00:00:00

Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ doi:10.5194/bg-12-653-2015 © Author(s) 2015. CC Attribution 3.0 License. Recent trends and drivers of regional sources and sinks of carbon dioxide 1 1 2 3 1 4 5 6 S. Sitch , P. Friedlingstein , N. Gruber , S. D. Jones , G. Murray-Tortarolo , A. Ahlström , S. C. Doney , H. Graven , 7,8,9 10 11 12 13 14 15 16 17 C. Heinze , C. Huntingford , S. Levis , P. E. Levy , M. Lomas , B. Poulter , N. Viovy , S. Zaehle , N. Zeng , 18 11 15 19 15 15 10 20 15 A. Arneth , G. Bonan , L. Bopp , J. G. Canadell , F. Chevallier , P. Ciais , R. Ellis , M. Gloor , P. Peylin , 21 3 4 22,23 24 S. L. Piao , C. Le Quéré , B. Smith , Z. Zhu , and R. Myneni University of Exeter, Exeter EX4 4QF, UK Institute of Biogeochemistry and Pollutant Dynamics, ETH Zurich, Zurich, Switzerland Tyndall Centre for Climate Change Research, University of East Anglia, Norwich NR4 7TJ, UK Lund University, Department of Physical Geography and Ecosystem Science, Sölvegatan 12, 223 62 Lund, Sweden Marine Chemistry and Geochemistry Department, Woods Hole Oceanographic Institution, 266 Woods Hole Road, Woods Hole, MA 02543, USA Department of Physics and Grantham Institute for Climate Change, Imperial College London, London SW7 2AZ, UK Geophysical Institute, University of Bergen, Bergen, Norway Bjerknes Centre for Climate Research, Bergen, Norway Uni Climate, Uni Research AS, Bergen, Norway Centre for Ecology and Hydrology, Benson Lane, Wallingford OX10 8BB, UK National Center for Atmospheric Research, Boulder, Colorado, USA Centre for Ecology and Hydrology, Bush Estate, Penicuik, Midlothian EH26 0QB, UK Department of Animal & Plant Sciences, University of Shefﬁeld, Shefﬁeld S10 2TN, UK Institute on Ecosystems and Department of Ecology, Montana State University, Bozeman, MT 59717, USA Laboratoire des Sciences du Climat et de l’Environnement, CEA CNRS UVSQ, 91191 Gif-sur-Yvette, France Biogeochemical Integration Department, Max Planck Institute for Biogeochemistry, P.O. Box 10 01 64, 07701 Jena, Germany Department of Atmospheric and Oceanic Science, University of Maryland, College Park, MD 20740, USA Karlsruhe Institute of Technology, Garmisch-Partenkirchen, Germany Global Carbon Project, CSIRO Oceans and Atmosphere Flagship, Canberra, Australia University of Leeds, School of Geography, Woodhouse Lane, Leeds LS9 2JT, UK College of Urban and Environmental Sciences, Peking University, Beijing 100871, China State Key Laboratory of Remote Sensing Science, Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China Center for Applications of Spatial Information Technologies in Public Health, Beijing 100101, China Department of Geography and Environment, Boston University, 675 Commonwealth Avenue, Boston, MA 02215, USA Correspondence to: S. Sitch ([email protected]) Received: 21 November 2013 – Published in Biogeosciences Discuss.: 23 December 2013 Revised: 30 November 2014 – Accepted: 19 December 2014 – Published: 2 February 2015 Published by Copernicus Publications on behalf of the European Geosciences Union. 654 S. Sitch et al.: Recent trends and drivers of regional sources Abstract. The land and ocean absorb on average just over “missing” carbon and the identiﬁcation of the processes driv- half of the anthropogenic emissions of carbon dioxide (CO / ing carbon sinks has been one of the dominating questions every year. These CO “sinks” are modulated by climate for carbon cycle research in the past decades (e.g. Tans et al., change and variability. Here we use a suite of nine dynamic 1990; Sarmiento and Gruber, 2002; and others). While much global vegetation models (DGVMs) and four ocean biogeo- progress has been achieved (e.g. Prentice et al., 2001; Sabine chemical general circulation models (OBGCMs) to estimate et al., 2004; Denman et al., 2007; Le Quéré et al., 2009), trends driven by global and regional climate and atmospheric and estimates have converged considerably (Sweeney et al., CO in land and oceanic CO exchanges with the atmo- 2007; Khatiwala et al., 2013; Wanninkhof et al., 2013), the 2 2 sphere over the period 1990–2009, to attribute these trends spatial attribution of recent sink rates for the ocean and land, to underlying processes in the models, and to quantify the and particularly their changes through time, remain uncer- uncertainty and level of inter-model agreement. The mod- tain. To balance the global carbon budget, the combined sinks els were forced with reconstructed climate ﬁelds and ob- by land and ocean must have increased over recent decades served global atmospheric CO ; land use and land cover (Keeling et al., 1995; Canadell et al., 2007; Raupach et al., changes are not included for the DGVMs. Over the pe- 2008; Sarmiento et al., 2010; Gloor et al., 2010; Ballantyne riod 1990–2009, the DGVMs simulate a mean global land et al., 2012). Sarmiento et al. (2010) showed that some of carbon sink of 2.4 0.7 Pg C yr with a small signiﬁ- the increasing sinks are driven by the ocean, but also iden- cant trend of0.06 0.03 Pg C yr (increasing sink). Over tiﬁed an even more substantial increase in the net uptake by the more limited period 1990–2004, the ocean models sim- the land biosphere between the 1980s and the 1990s. This in- ulate a mean ocean sink of 2.2 0.2 Pg C yr with a crease in the global land and ocean sink has been sustained trend in the net C uptake that is indistinguishable from zero to date (Ballantyne et al., 2012). (0.01 0.02 Pg C yr /. The two ocean models that ex- There are several studies on the trends in carbon exchanges tended the simulations until 2009 suggest a slightly stronger, at the regional level based on atmospheric CO observations but still small, trend of0.02 0.01 Pg C yr . Trends from (top-down approach) (Angert et al., 2005; Buermann et al., land and ocean models compare favourably to the land green- 2007; Chevallier et al., 2010; Sarmiento et al., 2010) and ness trends from remote sensing, atmospheric inversion re- changes in high-latitude greenness on land (Nemani et al., sults, and the residual land sink required to close the global 2003; Myneni et al., 1997) and changes in sea surface tem- carbon budget. Trends in the land sink are driven by increas- perature in the ocean (Park et al., 2010). Atmospheric CO - ing net primary production (NPP), whose statistically sig- based top-down approaches provide large-scale constraints niﬁcant trend of 0.22 0.08 Pg C yr exceeds a signiﬁcant on the land and ocean surface processes, but they cannot trend in heterotrophic respiration of 0.16 0.05 Pg C yr – unambiguously identify the underlying processes or the re- primarily as a consequence of widespread CO fertilisation gions driving these changes. Bottom-up studies using dy- of plant production. Most of the land-based trend in simu- namic global vegetation models (DGVMs) or ocean biogeo- lated net carbon uptake originates from natural ecosystems chemical general circulation models (OBGCMs) mechanis- in the tropics (0.04 0.01 Pg C yr /, with almost no trend tically represent many of the key land (Prentice et al., 2007) over the northern land region, where recent warming and and ocean processes (Le Quéré et al., 2005), and offer the reduced rainfall offsets the positive impact of elevated at- opportunity to investigate how changes in the structure and mospheric CO and changes in growing season length on functioning of land ecosystems and the ocean in response carbon storage. The small uptake trend in the ocean mod- to changing environmental conditions affect biogeochemi- els emerges because climate variability and change, and in cal cycles. Therefore DGVMs and OBGCMs potentially al- particular increasing sea surface temperatures, tend to coun- low for a more comprehensive analysis of surface carbon teract the trend in ocean uptake driven by the increase in at- trends and provide insight into possible mechanisms behind mospheric CO . Large uncertainty remains in the magnitude regional trends in the carbon cycle. and sign of modelled carbon trends in several regions, as well There is a growing literature on regional carbon budgets as regarding the inﬂuence of land use and land cover changes for different parts of the world (Ciais et al., 1995; Phillips on regional trends. et al., 1998; Fan et al., 1998; Pacala et al., 2001; Janssens et al., 2003; Stephens et al., 2007; Piao et al., 2009; Lewis et al., 2009a; Ciais et al., 2010; Pan et al., 2011; Tjipu- tra et al., 2010; Roy et al., 2011; Schuster et al., 2013; Lenton et al., 2013), using bottom-up (inventory, carbon 1 Introduction cycle models) and top-down methodologies, although they Soon after the ﬁrst high-precision measurements of atmo- typically cover different time intervals. To date, no glob- spheric CO started in the late 1950s, it became clear that ally consistent attribution has been attempted for regional the global-mean CO growth rate is substantially lower than sources and sinks of atmospheric CO . This paper attempts 2 2 expected if all anthropogenic CO emissions remained in the to ﬁll this gap by combining top-down and bottom-up ap- atmosphere (e.g. Keeling et al., 1976). The search for this proaches discussed in the regional syntheses of the REgional Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 655 Carbon Cycle Assessment and Processes project (RECCAP; Canadell et al., 2013) and by using factorial simulations to elucidate the processes that drive trends in the sources and sinks of atmospheric CO . This study has two major aims. The ﬁrst of these is to es- timate the regional trends in the carbon exchange over the period 1990–2009, associated with changes in climate and at- mospheric CO concentration, for three land regions (north- ern land, tropical land, and southern land) and seven ocean regions (North Paciﬁc, equatorial Paciﬁc, South Paciﬁc, North Atlantic, equatorial/South Atlantic, Indian Ocean, and Southern Ocean) (Fig. 1). The second aim is to determine Figure 1. Land and ocean regions. The three land regions: north- which factors and processes among those included in the ern land, tropical land, and southern land. Northern land com- models are driving the modelled/observed trends in the re- prises boreal North America (navy blue), Europe (light blue), bo- gional land/ocean to atmosphere net CO ﬂuxes. For the land real Asia (blue), temperate North America (pale red), and tem- models, those factors and processes included are the CO fer- perate Asia (red). Tropical land comprises tropical South Ameri- tilisation effect on productivity and storage, as well as cli- can forests (sea green), northern Africa (sand), equatorial Africa mate effects on productivity, respiration, and climate-caused (green), and tropical Asia (dark green). Southern land comprises natural disturbances (see Table S1 in the Supplement for de- South American savanna (pale green), temperate South America (violet), southern Africa (orange), and Australia and New Zealand tails represented in individual models). A particular focus (yellow). Ocean regions comprise North Paciﬁc (dark red), equa- is on the impacts of climate variation and change on land torial Paciﬁc (orange-red), South Paciﬁc (orange), North Atlantic ecosystems at the regional scale, as extreme climate events (orchid), equatorial/South Atlantic (slate blue), Indian Ocean (this- occurred during the period of 1990–2009 across many re- tle), Southern Ocean (sky blue), and Arctic Ocean and Antarctica gions of the world, including North America (southwestern (white). USA, 2000–2002), Europe (2003), Amazonia (2005), and eastern Australia (2001–2008), raising considerable attention in the ecological community regarding the consequences of for individual land and ocean regions over the period 1990– recent climate variability on ecosystem structure and func- 2009 (see RECCAP special issue; Canadell et al., 2013, tion (Allen et al., 2010) and the carbon cycle (Ciais et al., http://www.biogeosciences.net/special_issue107.html). 2005; Van der Molen et al., 2011; Reichstein et al., 2013). Trends and variability in the air–sea CO ﬂuxes simulated This study addresses the changes in the magnitude of the by the employed OBGCMs are driven by the increase in at- global carbon sink but does not discuss the efﬁciency of the mospheric CO and by variability and change in ocean tem- sinks, which is widely discussed elsewhere (Raupach et al., perature, circulation, winds, and biology largely governed 2014; Gloor et al., 2010; Ciais et al., 2013). These DGVMs by climate variability. The air–sea CO ﬂux arising from have been extensively evaluated against observation-based the increase in atmospheric CO is often referred to as the gross primary production (GPP), land to atmosphere net CO ﬂux of anthropogenic CO , while the remainder, induced ﬂux, and CO sensitivity of net primary production (NPP) by changes in the natural cycling of carbon in the ocean– compared to results from free-air CO enrichment (FACE) atmosphere system, is called the “natural” CO component experiments (Piao et al., 2013). (e.g. Gruber et al., 2009). Although this conceptual separa- Consideration of land use and land cover change (LULCC) tion has its limits (McNeill and Matear, 2013), it provides on regional trends is beyond the scope of the present for a powerful way to understand how different forcings af- study, and therefore models assume a ﬁxed present-day fect the net ocean sink. land use throughout the simulation period. Thus our re- DGVM results are compared with estimates of the resid- sults presented should be interpreted with this caveat in ual land sink (RLS) and remote sensing products indicat- mind. There are large uncertainties in the global LULCC ing trends of greening and browning in the northern region. ﬂux and its change through time, with an estimated decrease Regional sources and sink trends are attributed to processes 1 1 from 1.6 0.5 Pg C yr (1990–1999) to 1.0 0.5 Pg C yr based on factorial simulations. (2000–2009) (LeQuéré et al., 2013). In addition, the net land use (LU) ﬂux for the period 1990–2009 will be inﬂuenced by earlier LULCC (i.e. legacy ﬂuxes), confounding the analysis. 2 Methods The response of the large ﬂuxes associated with net primary productivity and heterotrophic respiration to climate variabil- 2.1 Dynamic global vegetation models ity and CO are the focus of this study. Other companion pa- pers investigate ecosystem response to interannual and sea- Following the studies of Le Quéré et al. (2009) and Sitch et sonal timescales (Piao et al., 2013), and the carbon balance al. (2008), a consortium of DGVM groups set up a project www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 656 S. Sitch et al.: Recent trends and drivers of regional sources to investigate further the spatial trends in land–atmosphere transfer velocity of Broecker et al. (1985) was too high (Pea- ﬂux and agreed to perform a factorial set of DGVM simu- cock, 2004; Sweeney et al., 2007; Müller et al., 2008). lations over the historical period, 1901–2009. These simula- 2.3 Data sets tions have contributed to the RECCAP activity (Canadell et al., 2011, 2013). There are now a variety of DGVMs with ori- 2.3.1 Land gins in different research communities that typically contain alternative parameterisations and a diverse inclusion of pro- Climate forcing is based on a merged product of Cli- cesses (Prentice et al., 2007; Piao et al., 2013). DGVMs have mate Research Unit (CRU) observed monthly 0.5 clima- emerged from the land surface modelling (LSM), forest ecol- tology (v3.0, 1901–2009; New et al., 2000) and the high- ogy, global biogeography, and global biogeochemical mod- temporal-resolution NCEP reanalysis. The merged product elling communities. Representative of these research strands has a 0.5 spatial and 6 h temporal resolution. A coarse- are the following nine DGVMs, which are applied here: Hy- resolution 3.75 2.5 version at monthly timescales was land (Levy et al., 2004), JULES (Cox, 2001; Clark et al., also produced (see Table 1 for spatial resolution of individ- 2011), LPJ (Sitch et al., 2003), LPJ-GUESS (Smith et al., ual DGVMs). Global atmospheric CO was derived from ice 2001), NCAR-CLM4 (Thornton et al., 2007, 2009; Bonan core and NOAA monitoring station data, and provided at and Levis, 2010; Lawrence et al., 2011), ORCHIDEE (Krin- annual resolution over the period 1860–2009. As land use ner et al., 2005), OCN (Zaehle and Friend, 2010), SDGVM and land cover change was not simulated in these model (Woodward et al., 1995; Woodward and Lomas, 2004), and experiments, models assume a constant land use (invariant VEGAS (Zeng, 2003; Zeng et al., 2005). In this study we fo- agricultural coverage) throughout the simulation period. At- cus on two aspects of land surface modelling: the carbon and mospheric nitrogen deposition data for CLM4CN and OCN the hydrological cycles. In the case of land surface models were sourced from Jean-Francois Lamarque (personal com- coupled to GCMs, energy exchange between the land surface munication, 2012) and Dentener et al. (2006), respectively. and atmosphere is also simulated. Gridded ﬁelds of leaf area index (LAI) are used in the eval- uation of DGVM northern greening trends. These LAI data 2.2 Ocean biogeochemical general circulation models sets were based on remote sensing data and were generated from the AVHRR GIMMS NDVI3g product using an artiﬁ- cial neural network (ANN)-derived model (Zhu et al., 2013). A total of four different groups have conducted the fac- The data set has a temporal resolution of 15 days over the torial simulations over the analysis period with three- period 1981–2011, and a spatial resolution of 1=12 . dimensional OBGCMs and submitted their results to the RECCAP archive. These are MICOM-HAMOCCv1 (BER) 2.3.2 Ocean (Assmann et al., 2010), CCSM-WHOI using CCSM3.1 (BEC) (Doney et al., 2009a, b), CCSM-ETH using CCSM3.0 Unlike how the land models simulations were set up, no com- (ETH) (Graven et al., 2012), and NEMO-PlankTOM5 (UEA) mon climatic forcing data set was used for the ocean model (Buitenhuis et al., 2010). Details of the models are given simulations. In fact, some models provided several simula- in the respective publications cited and in Table 2. Not all tion results obtained with different climatic forcings. Models model simulations are independent of each other, as sev- were forced by the NCEP climatic data (Kalnay et al., 1996) eral of them share components. BEC and ETH employ the in their original form, or in the modiﬁed CORE (Common same OBGCM, but differ in their spin-up and surface forc- Ocean-ice Reference Experiments – Corrected Normal Year ing. The employed models have relatively similar horizontal Forcing (CORE-CNYF; Large and Yeager, 2004)) form (Ta- resolution of the order of 1 to 3 in longitude and latitude, ble 2). i.e. none of them is eddy-permitting or eddy-resolving. The four ecosystem/biogeochemical models are also of compara- 2.3.3 Atmospheric inversion ble complexity, i.e. including explicit descriptions of at least one phytoplankton and zooplankton group, with some mod- Simulated trends in land to atmospheric net CO ﬂux are els considering up to three explicitly modelled groups for compared with those from version 11.2 of the CO inver- phytoplankton and two for zooplankton. All models use the sion product from the Monitoring Atmospheric Composi- same gas exchange parameterisation of Wanninkhof (1992), tion and Climate – Interim Implementation (MACC-II) ser- although with different parameters. In particular, the ETH vice (http://copernicus-atmosphere.eu/). The horizontal res- model used a lower value for the gas exchange coefﬁcient olution of the inversion is 3.75 2.5 square degrees (longi- than originally used in the CCSM standard conﬁguration, tude latitude), and weekly temporal resolution, with night- yielding a global-mean gas transfer velocity that is more than time and daytime separated. The accuracy varies with the pe- 25 % lower than those of the other models (Graven et al., riod and the location over the globe, depending on the den- 2012). This reduction reﬂects the mounting evidence based sity and the information content of the assimilated data, and on radiocarbon analyses that the original global-mean gas usually decreases with increasing the resolution. Uncertainty Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 657 Figure 2 shows the historical changes in climate, atmo- spheric CO concentration, and nitrogen deposition over the period 1990–2009 used to force the DGVMs. A summary of DGVM characteristics is given in Table 1. A more detailed description of DGVM process representations is given in Ta- ble S1. 2.4.2 Ocean The ocean models employed two different approaches for creating the initial conditions for the experiments. The ﬁrst approach, followed by CCSM-ETH, CCSM-WHOI, and BER, involved ﬁrst a multiple-century-long spin-up with cli- matological forcing and with atmospheric CO held constant at its pre-industrial value, bringing these models very close to a climatological steady state for pre-industrial conditions (in some models 1750; in others 1850). In the second step, the models were then integrated forward in time through the historical period until 1948, with atmospheric CO pre- scribed to follow the observed trend and a climatological forcing. The length of the spin-up varies from a few hun- Figure 2. Global trends in environmental driving variables: (a) land dred years to several thousand years, resulting in differing temperature, (b) land precipitation, (c) ocean temperature, (d) wind global integrated drift ﬂuxes, although their magnitudes are speed, (e) N deposition, and (f) atmospheric [CO ]. 2 1 substantially smaller than 0.05 Pg C yr with essentially no rate of change. The second approach, followed by NEMO- PlankTOM5 (UEA), was to initialise the model with recon- numbers at various scales can be found in Table 2 of Peylin structed initial conditions in 1920, and then also run it for- et al. (2013). The inversion covers years 1979–2011, and ward in time until 1948 with prescribed atmospheric CO , a previous release has been documented by Chevallier et repeating the daily forcing conditions of a single year (1980). al. (2010). It uses a climatological prior without interannual The modelled export production was tuned to obtain an ocean variability, except for fossil fuel CO emissions. CO sink of 2.2 Pg C yr in the 1990s. This second method 2.4 Experimental design offers the advantage that the model’s carbon ﬁelds remain closer to the observations compared to the long spin-up ap- 2.4.1 Land proach, but it comes at the cost of generating a drift that af- fects the mean conditions and to a lesser extent the trend. Model spin-up consisted of recycling climate mean and Tests with the model runs of Le Quéré et al. (2010) suggest variability from the early decades of the 20th century 1 the drift in the mean CO sink is about 0.5 Pg C yr and (1901–1920) with 1860 atmospheric CO concentration of 2 the drift in the trend is about 0.005 Pg C yr globally, and is 287.14 ppm until carbon pools and ﬂuxes were in steady state largest in the Southern Ocean. (zero mean annual land to atmospheric net CO ﬂux). The From 1950 onward, the models performed two separate land models were then forced over the 1861–1900 transient simulations: simulation using varying CO and continued recycling of cli- mate as in the spin-up. The land models were then forced – S_O1: CO only, i.e. atmospheric CO increases, but 2 2 over the 1901–2009 period with changing CO , climate, and models are forced with climatological atmospheric ﬁxed present-day land use according to the following simu- boundary conditions (referred to as ACO2 in the REC- lations: CAP archive); – S_L1: changing CO only (i.e. time-invariant present- – S_O2: CO and climate, i.e. as S_O1, but models are day land use mask, ﬁxed pre-industrial climate); forced with “realistic” year-to-year variability in atmo- spheric boundary conditions (ANTH). – S_L2: changing CO and climate (i.e. time-invariant present-day land use mask). In these runs, both S_O1 and S_O2 are affected by the same For DGVMs including the N cycle, N deposition was a drift, and their differences thus remove the drift. The CCSM- time-variant forcing in both simulations, such that the differ- based models performed an additional experiment to bet- ence between S_L2 and S_L1 includes the synergistic effects ter separate between the ﬂuxes of natural and anthropogenic of N deposition on CO fertilisation (Zaehle et al., 2010). CO : 2 2 www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 658 S. Sitch et al.: Recent trends and drivers of regional sources Table 1. Characteristics of the nine dynamic global vegetation models. Model name Abbreviation Spatial Land surface Full nitrogen River export Fire Harvest/grazing Source resolution model cycle ﬂux simulation ﬂux Community Land CLM4CN 0.5 0.5 Yes Yes No Yes No Oleson et al. (2010); Model 4CN Lawrence et al. (2011) Hyland HYL 3.75 2.5 No No No No Yes Friend et al. (1997); Levy et al. (2004) Lund–Potsdam–Jena LPJ 0.5 0.5 No No No Yes Yes Sitch et al. (2003) LPJ-GUESS LPJ-GUESS 0.5 0.5 No No No Yes No Smith et al. (2001) ORCHIDEE-CN OCN 3.75 2.5 Yes Yes No No Yes Zaehle and Friend (2010); Zaehle et al. (2010) ORCHIDEE ORC 0.5 0.5 Yes No No No No Krinner et al. (2005) Shefﬁeld-DGVM SDGVM 3.75 2.5 No No Yes Yes No Woodward et al. (1995) TRIFFID TRI 3.75 2.5 Yes No No No No Cox (2001) VEGAS VEGAS 0.5 0.5 Yes No Yes Yes Yes Zeng et al. (2005) Table 2. Characteristics of the four ocean biogeochemical general circulation models (OBGCMs). All include NPZD-type ecosystem models and N, P, Si, and Fe nutrient components. Model name Abbreviation Spatial resolution Meteorological Gas transfer Years used Source forcing formulation MICOM-HAMOCCv1 BER 2.4 0.82.4 NCEP Wanninkhof (1992) 1990 to 2009 Assmann et al. (2010) CCSM-WHOI BEC 3.6 0.81.8 NCEP Wanninkhof (1992) 1990 to 2009 Doney et al. (2009a, b) CCSM-ETH ETH 3.6 0.9 1.9 CORE Wanninkhof (1992) 1990 to 2007 Graven et al. (2012) NEMO-PlankTOM5 UEA 2 0.52 NCEP Wanninkhof (1992) 1990 to 2009 Buitenhuis et al. (2010) – S_O3: pre-industrial CO and climate, i.e. atmospheric results for simulation S_L2 are compared against the global CO is ﬁxed at its pre-industrial level, but atmospheric RLS, calculated as the annual anthropogenic CO emissions 2 2 boundary conditions vary as in S_O2 (PIND). (fossil fuel, cement manufacture, and land use C ﬂux) mi- nus the annual CO growth rate and model mean ocean C From these simulations, only the results from 1990 through sink as given by Friedlingstein et al. (2010). The ocean up- to 2009 were analysed. Only the UEA and CCSM-WHOI take is from the same OGGCMs as the ones used here, and models made results available for the S_O1 and S_O2 simu- the land use C ﬂux is based on a book-keeping approach from lations for the entire analysis time. The results for the BER Houghton (2010). Note the RLS depends on a LULCC model model for 2009 are incomplete, and the CCSM-ETH simula- of emissions (the one of Houghton). Strictly speaking, com- tions extend only to 2007. In order to maintain a sufﬁciently parison of model land to atmosphere net CO ﬂux with RLS large set of models, we decided to focus our analysis primar- is therefore inconsistent because these models treat areas af- ily on the 1990–2004 period, but occasionally also include fected by LUC as pristine ecosystems, and these areas are the results through to 2009, with the important caveat that generally associated with a high land carbon sinks. Simulated the latter are based only on two models. net carbon ﬂux from S2 is therefore likely to overestimate the RLS sink, by construction. 2.5 Output variables The regional analysis will focus on three large land regions 2.5.1 Land (Fig. 1), and within these regions, trends at a ﬁner spatial res- olution, from multi-grid-cell to the sub-region, are analysed. In this study we focus primarily on the simulated carbon cy- The comparison of DGVM simulated trends in the north- cle variables, net NPP, RH (heterotrophic respiration), and ern growing season against satellite-derived NDVI (nor- LAI, a measure of vegetation greenness. The land to atmo- malised difference vegetation index) observations was based sphere net CO ﬂux is on eight models (JULES, LPJ, LPJ-GUESS, NCAR-CLM4, ORCHIDEE, OCN, SDGVM, VEGAS), which provided land to atmosphere net CO ﬂuxDNBP LAI outputs. The means and trends in the onset, end, and D RHC wildﬁre ﬂux riverine C ﬂuxC harvest NPP; length of growing season were computed. Growing season where we have adopted the atmospheric perspective with re- variables were calculated using the methodology of Murray- gard to the sign of the ﬂuxes, i.e. negative numbers indicate Tortarolo et al. (2013). Leaf onset is deﬁned as the day when a sink for atmospheric CO and a negative trend indicates an LAI begins to increase above a critical threshold (CT), de- increasing sink or a decreasing source. ﬁned as DGVMs typically do not represent all these processes; a list for each individual DGVM is given in Table 1. DGVM CTD LAI C 0:2 .LAI LAI /; min max min Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 659 where LAI and LAI represent the minimum and max- tions, and @F=@ FS is the change in the air–sea CO ﬂux in re- min max 2 imum LAI over the annual cycle. Similarly, leaf senescence, sponse to freshwater ﬂuxes. This latter term includes not only or offset, or end of growing season, is deﬁned as the day the sensitivity of oceanic pCO to changes in salinity but also when LAI decreases below the CT. The length of the grow- the dilution effects of freshwater on DIC and Alk (see Doney ing season in days is calculated as the end minus the onset. et al., 2009a, for details). The partial derivatives were com- This calculation was made for each grid cell above 30 N puted directly from the model equations for the mean condi- (i.e. northern extratropics) from the models and the satellite tions in each region. The changes in the driving components data. In addition, any grid cell where LAI varied by less than were derived from the trend computed via a linear regression 0.5 over the annual cycle from the satellite data was consid- of the model results and then multiplied by the length of the ered to be predominantly evergreen (e.g. boreal forest), and time series. thus excluded from the analysis. We also masked out regions where LAI decreases in the summer (drought deciduous veg- etation). In addition, when the growing season spans over the 3 Results end of year (e.g. Mediterranean and some pixels particularly on the southern margin of the domain), we include the ﬁrst 3 3.1 Global Trends months of the second year in our analysis. Means and trends were calculated using a linear model over the period 1990– 3.1.1 Land The ensemble mean global land to atmosphere net carbon 2.5.2 Ocean dioxide ﬂux from S_L2 is 2.38 0.72 Pg C yr over the period 1990–2009 (P D 0.04, where P is the probability of a The modelling groups provided output on a monthly basis trend statistically indistinguishable from zero; a signiﬁcance for the years 1990 through to 2004 and 2009 at two levels level of 0.05 is selected) (Fig. 3, Fig. S1 in the Supplement, of priority. Tier-one data included the surface ocean ﬁelds of Table 3). The numbers behind signs are the 1 standard de- the air–sea CO ﬂux, oceanic pCO , dissolved inorganic car- 2 2 viation of 20-year means for nine DGVMs. This compares bon (DIC), alkalinity (Alk), temperature (T ), salinity (S), and to the global RLS of 2.45 1.17 Pg C yr , inferred from mixed layer depth. The second-tier data included the biologi- the global carbon budget by Friedlingstein et al. (2010) over cal export at 100 m, the vertically integrated net primary pro- the same period. All DGVMs agree on an increasing land duction, and the surface chlorophyll a concentration. Some sink with a net ﬂux trend over this period ranging between models also supplied three-dimensional climatological ﬁelds 0.02 and 0.11 Pg C yr , corresponding to the OCN and of DIC, Alk, T , and S. Hyland DGVMs, respectively (Table 3). DGVMs simulate To determine the different factors contributing to the mod- an increase in the land C sink with an ensemble mean trend elled trends and variations, we undertook two (linear) sepa- 2 of 0.06 0.03 Pg C yr (P < 0.05) over the period 1990– rations: 2009 (Table 3) in response to changes in climate and atmo- spheric CO content. The two DGVMs with a fully cou- – The contribution of climate variability and change on pled carbon and nitrogen cycle (CN) also simulate an in- the ocean carbon cycle: X_varD X(S_O2) X(S_O1), crease in the land sink, at 0.02 (P D 0.6) for OCN and X is any variable or ﬂux, where the expression in paren- 0.05 Pg C yr (P D 0.06) for CLM4CN. DGVMs suggest theses represents the results of the corresponding sim- the increase in global land sink between 1990 and 2009 is ulation, and X_var represents the impact of climate driven by increases in simulated global NPP (Fig. 3). change and variability on the ocean carbon cycle. DGVMs simulate an ensemble mean global NPP of – The contribution of anthropogenic CO : 1 62.9 8.73 Pg C yr over the period 1990–2009 (Table 3). X_antD X(S_O2) X(S_O3). All DGVMs simulate an increase in NPP over this pe- riod, with an ensemble mean DGVM trend in NPP of For each of the integrations, but particularly for the changing 0.22 0.08 Pg C yr (P D 0.00) (Table 3). Models with a CO and climate simulation S_O2, we analysed the factors higher NPP trend also produce a higher land to atmosphere contributing to the temporal change in the air–sea CO ﬂux net CO ﬂux trend (Fig. S2 in the Supplement). The ensem- F by a linear Taylor expansion (see e.g. Lovenduski et al., ble mean NPP trend of 0.22 0.08 Pg C yr (P < 0.01) from 2007 and Doney et al., 2009a): simulation S_L2 (CO and climate forcing) contrasts with 1F D @F=@ ws 1wsC @F=@T 1T C @F=@ ice an ensemble trend of 0.19 0.08 Pg C yr (P < 0.01) and 0.03 0.05 Pg C yr (P D 0.24) over the same period for 1iceC @F=@ sDIC 1sDIC the S_L1 (CO only) and S_L2–S_L1 (the climate effect), C @F=@ sAlk 1sAlkC @F=@ FS 1S; respectively (Tables S2, S3 in the Supplement). These re- where ws is the wind speed, ice is the sea-ice fraction, sDIC sults suggest that the simulated increase in global NPP is and sAlk are the salinity normalised DIC and Alk concentra- mainly in response to increasing atmospheric CO (direct www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 660 S. Sitch et al.: Recent trends and drivers of regional sources Figure 4. Global trends in ensemble ocean model ﬂuxes. Black line: results from simulation S_O2 with variable “climate” and increas- ing CO . Red line: results from simulation S_O1 with constant “cli- mate” and increasing CO . The dashed grey and dashed red lines indicate the uncertainty bands given by the four models that con- tribute to the ensemble mean. 0.16 0.05 Pg C yr (P < 0.01) over the period 1990–2009 (Table 3). This is lower than the trend in global NPP, resulting in a trend towards increasing net land carbon uptake. This is unsurprising as there is a lagged response in increases in RH relative to NPP, reﬂecting the turnover time of the newly in- corporated plant material. The ensemble mean trend in RH is 2 2 0.12 0.06 Pg C yr (P < 0.01) and 0.04 0.02 Pg C yr (P D 0.09) over the same period for the S_L1 (CO only) and S_L2–S_L1 (the climate effect), respectively (Tables S2, S3). This implies the dominant effect on RH is increased substrate for microbial respiration, with the additional lit- Figure 3. Global trends in ensemble land model responses. ter input into soils, as a consequence of enhanced NPP, (a) DGVM mean model land to atmosphere net CO ﬂux and stan- rather than enhanced rates of microbial decomposition with dard deviation (grey lines); (b) component ﬂuxes, NPP; and (c) RH rising temperatures. Nevertheless, the simulated mean resi- (D RHC wildﬁreC riverine C ﬂux); and (d) remotely sensed trends dence time (MRTD soil carbon / RH) of soil organic mat- in annual mean NDVI (crosses), a measure of vegetation greenness, ter decreases, in response to warming, which is especially and a linear regression through the data points (bold line). pronounced in high-latitude regions (Fig. S3 in the Sup- plement). The difference in land–atmosphere ﬂux trend be- tween the CN models OCN (0.02 Pg C yr / and CLM4CN CO fertilisation of photosynthesis, in addition to the indirect 2 2 (0.05 Pg C yr / is largely due their difference in RH trends beneﬁts from an improved water balance in water-limited at 0.14 and 0.11 Pg C yr , respectively, rather than differen- ecosystems due to the physiological effects of CO on wa- tial responses of simulated NPP to elevated CO (Table 3). ter use efﬁciency). VEGAS, CLM4CN, and OCN simulate Only four DGVMs simulated wildﬁre ﬂuxes (CLM4CN, the smallest positive trends in NPP among the DGVMs in re- LPJ, LPJ-GUESS, SDGVM). No signiﬁcant trends in the sponse to elevated CO forcing (Table S2). This suggests that global wildﬁre ﬂux were reported by any of the DGVMs. the potential CO fertilisation effect may be already strongly limited by present-day nitrogen availability in some ecosys- 3.1.2 Ocean tems (Vitousek and Howarth, 1991). There is more uncer- tainty among models on the impact of climate changes on global NPP, with only two models simulating a signiﬁcant The global ocean is simulated to have acted as a very sub- positive trend (Table S3). stantial sink for atmospheric CO but one that has increased DGVMs simulate an ensemble mean global RH of only slightly over the last two decades (see also discus- 57.5 9.8 Pg C yr over the period 1990–2009 (Table 3). sion in Wanninkhof et al., 2013). The mean ocean sink in All DGVMs simulate an increase in RH for S_L2 the four models (CCSM-ETH, CCSM-WHOI, UEA, and (CO and climate), with an ensemble mean trend of BER) increased from2.0 Pg C yr in the early 1990s to Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 661 2.1 Pg C yr during the ﬁrst 5 years of the 21st century 3.2 Regional trends (Fig. 4). We separate the mean and variable components by us- 3.2.1 Land ing our factorial experiments, i.e. by using S_O1 results to identify the ocean uptake in the absence of climate variabil- Northern land ity and change, and the difference between S_O2 and S_O1 as measure of the impact of climate change. This separa- All DGVMs agree on a land C sink over the north- tion reveals that, in the absence of climate variability and ern land region, with a mean land–atmosphere ﬂux of change, the global ocean uptake would have increased from 1.03 0.30 Pg C yr over the period 1990–2009 (Fig. S4 about 1.98 0.04 Pg C yr for the 1990–1994 period to in the Supplement, Table 3). The ensemble mean land– 2.3 0.09 Pg C yr for 2000–2004 (for the two models atmosphere ﬂux trend is near zero for this region between that provided S_O1 results up to 2009 (CCSM-WHOI and 1990 and 2009 (Fig. S5 in the Supplement). Of particu- UEA), the uptake ﬂux would have increased from 1.99 lar interest are sub-regions with a simulated positive land– to 2.56 Pg C yr for 2005–2009). This global net uptake atmosphere ﬂux trend (Fig. 5), implying a diminishing sink ﬂux and its substantial trend in time (0.03 Pg C yr for of atmospheric CO or an increasing source of CO to the at- 2 2 1990–2004, and 0.04 Pg C yr for 1990–2010) is entirely mosphere. At least six models out of nine agree on a decreas- driven by the increase in atmospheric CO and is – integrated ing regional land sink across some areas in temperate North globally – numerically equivalent to the ocean uptake ﬂux America, eastern Europe, northeastern China, and Mongolia of anthropogenic CO . Climate variability and change mod- (Fig. 5). These largely correspond to regions with negative iﬁed these ﬂuxes, and particularly the trend in these mod- trends in precipitation (Fig. 6). els. The four models suggest an enhancement of the uptake Over the northern region, which covers almost 50 % of in the early 1990s (1990–1994) of about 0.2 Pg C yr , the land surface, DGVMs simulate an ensemble mean NPP turning into a reduction of the uptake in the subsequent pe- of 24.1 4.48 Pg C yr , which represents almost 40 % of riod (1995–1999), followed by a further reduction in the the global total (Table 3). All DGVMs simulate an increase 2000–2004 period of C0.1 Pg C yr . This trend toward in northern NPP over this period, with a trend in NPP reduced uptake in response to climate variability and change of 0.06 0.02 Pg C yr (P < 0.01) (Table 3). However, en- of C0.03 Pg C yr nearly completely compensates for the hanced productivity in the northern land region accounts anthropogenic CO driven increase in uptake, causing the for only around 29 % of the simulated global trend in NPP. overall uptake of CO to have a nearly ﬂat trend over the The ensemble mean NPP trend of 0.06 0.02 Pg C yr 1990–2004 period of < 0.01 Pg C yr The same tendencies (P < 0.01) from simulation S2 (CO and climate forcing) are found for the two models that extend over the entire compares to a trend of 0.07 0.03 Pg C yr (P < 0.01) and 1990–2009 period: in these models, climate change and vari- 0.00 0.04 Pg C yr (P D 0.85) for the S_L1 (CO only) ability reduces the CO -driven trend of 0.04 Pg C yr by and S_L2–S_L1 (the climate effect), respectively (Tables S2, 2 2 more thanC0.02 Pg C yr , to around0.02 Pg C yr . S3). All DGVMs simulate a positive trend in NPP in response With consideration of the different factors affecting the to elevated CO across the northern land region, and trends ocean carbon sink following our Taylor expansion, we ﬁnd are all signiﬁcant at the 95 % conﬁdence level with the ex- increasing sea surface temperature to be a globally im- ception of CLM4CN (P D 0.21). portant driver for the positive trends (reduced sinks) in- Large areas in temperate North America and Asia ex- duced by climate change and variability. Over the 1990– perienced warming combined with reductions in precipita- 2004 period, the surface ocean warmed, on average, by tion over the period 1990–2009 (Fig. 5). Indeed, although 1 1 0.004 C yr (0.005 C yr from 1990 through to 2009). DGVMs simulate larger mean NPP in temperate compared Isochemically, this leads to an increase in the oceanic pCO to boreal regions (Table S5 in the Supplement), they simu- of 0.06 μatm yr , which appears small. However, it needs late signiﬁcant positive trends in boreal North America and to be compared with the trend in the global-mean air– boreal Asia, whereas trends in both temperate North Amer- sea pCO difference of about 0.1 μatm yr that is re- ica and Asia are smaller and not signiﬁcant at the 95 % con- quired in order to generate a trend in the ocean uptake ﬁdence level (Table S5). of 0.03 Pg C yr (see e.g. Matsumoto and Gruber, 2005; In response to warming, models simulate an earlier onset Sarmiento and Gruber, 2006). The overall sink is therefore (ensemble mean model trendD0.078 0.131 days yr / largely a consequence of the increase in atmospheric CO and delayed termination of the growing season (i.e. it mostly corresponds to the uptake ﬂux of anthropogenic (0.217 0.097 days yr / based on LAI, and thus a CO /, but it includes a substantial perturbation ﬂux stem- trend towards a longer growing season in the north- ming from the impact of climate variability and change on ern extratropics (0.295 0.228 days yr / (Fig. 7). This the ocean carbon cycle. is in broad agreement with observed greening trends (Zhu et al., 2013; Murray-Tortarolo et al., 2013): on- 1 1 setD0.11 days yr , offsetD 0.252 days yr , and www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 662 S. Sitch et al.: Recent trends and drivers of regional sources Table 3. Mean and trends in NPP, RH, and land–atmosphere ﬂux as simulated by individual DGVMs and the ensemble mean. MODEL NPP Trend P value RH Trend P value Land–atm CO Trend P value 1 2 1 2 1 1 (Pg C yr ) (Pg C yr ) (Pg C yr ) (Pg C yr ) ﬂux (Pg C yr ) (Pg C yr ) Global_Land CLM4CN 51.508 0.148 0.000 47.668 0.106 0.000 1.459 0.052 0.059 HYLAND 73.422 0.319 0.000 68.835 0.203 0.000 3.466 0.109 0.000 LPJ 59.306 0.216 0.000 47.612 0.117 0.000 2.251 0.068 0.061 LPJ-GUESS 62.506 0.174 0.000 55.448 0.145 0.000 1.802 0.043 0.346 OCN 53.941 0.155 0.000 50.611 0.135 0.000 2.272 0.015 0.568 ORCHIDEE 75.516 0.293 0.000 72.037 0.208 0.000 3.479 0.086 0.046 SDGVM 60.965 0.240 0.000 53.778 0.190 0.000 2.127 0.044 0.170 TRIFFID 71.929 0.305 0.000 69.167 0.244 0.000 2.762 0.061 0.265 VEGAS 57.308 0.113 0.006 51.930 0.092 0.000 1.783 0.018 0.551 Ensemble 62.934 0.218 0.000 57.454 0.160 0.000 2.378 0.055 0.048 SD 8.729 0.076 9.791 0.053 0.721 0.030 Northern_Land CLM4CN 17.523 0.043 0.003 16.215 0.036 0.000 0.670 0.007 0.612 HYLAND 19.139 0.098 0.000 17.591 0.080 0.000 0.876 0.014 0.311 LPJ 24.566 0.079 0.001 19.578 0.062 0.006 1.168 0.006 0.735 LPJ-GUESS 28.484 0.039 0.085 25.883 0.067 0.009 0.634 0.023 0.521 OCN 21.008 0.044 0.035 19.264 0.047 0.008 1.117 0.007 0.632 ORCHIDEE 30.337 0.070 0.007 29.112 0.063 0.000 1.226 0.006 0.740 SDGVM 25.144 0.063 0.006 22.598 0.065 0.006 0.828 0.004 0.762 TRIFFID 28.476 0.088 0.009 27.006 0.103 0.001 1.470 0.016 0.455 VEGAS 21.895 0.048 0.012 18.914 0.043 0.001 1.322 0.000 0.968 Ensemble 24.064 0.063 0.001 21.796 0.063 0.001 1.034 0.002 0.865 SD 4.484 0.022 4.562 0.020 0.295 0.012 Tropical_Land CLM4CN 26.400 0.090 0.000 24.464 0.058 0.000 0.692 0.039 0.110 HYLAND 34.489 0.112 0.000 32.695 0.067 0.000 1.560 0.044 0.001 LPJ 25.830 0.100 0.001 21.224 0.035 0.001 0.817 0.049 0.031 LPJ-GUESS 21.922 0.078 0.000 19.332 0.051 0.000 0.785 0.036 0.038 OCN 22.750 0.084 0.000 21.476 0.065 0.000 0.982 0.017 ORCHIDEE 31.313 0.151 0.000 29.640 0.108 0.000 1.673 0.043 0.084 SDGVM 23.505 0.118 0.000 20.677 0.075 0.000 0.984 0.038 0.030 TRIFFID 29.801 0.141 0.000 28.925 0.096 0.000 0.876 0.045 0.218 VEGAS 23.472 0.041 0.061 21.994 0.033 0.004 0.278 0.010 0.527 Ensemble 26.609 0.102 0.000 24.492 0.065 0.000 0.961 0.036 0.045 SD 4.350 0.034 4.752 0.025 0.428 0.013 Southern_Land CLM4CN 7.617 0.014 0.187 7.017 0.011 0.036 0.098 0.005 0.719 HYLAND 19.875 0.109 0.000 18.623 0.056 0.000 1.035 0.051 0.000 LPJ 8.940 0.037 0.074 6.833 0.021 0.004 0.267 0.013 0.355 LPJ-GUESS 12.124 0.058 0.003 10.255 0.026 0.001 0.385 0.031 0.192 OCN 10.222 0.027 0.165 9.909 0.023 0.053 0.174 0.004 0.744 ORCHIDEE 13.884 0.073 0.002 13.304 0.037 0.000 0.581 0.036 0.027 SDGVM 12.358 0.059 0.034 10.539 0.050 0.000 0.317 0.010 0.701 TRIFFID 13.707 0.077 0.020 13.290 0.045 0.000 0.417 0.032 0.269 VEGAS 11.971 0.024 0.382 11.049 0.016 0.140 0.182 0.009 0.656 Ensemble 12.300 0.053 0.011 11.202 0.032 0.000 0.384 0.021 0.196 SD 3.528 0.031 3.597 0.016 0.285 0.017 Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 663 Figure 5. (a) Average land to atmosphere net CO ﬂux over the period 1990–2009 for the ensemble mean and model disagreement, with stippling representing agreement for < 66 % of DGVMs , and (b) standard deviation across DGVMs. (c) The trend in land to atmosphere net CO ﬂux across the ensemble, and model disagreement, with stippling representing agreement of < 66 % of the DGVMs , and (d) the standard deviation of the trend. Figure 6. Trends in land climate drivers and process responses. (a) Trend in temperature ( C yr /, (b) trend in precipitation 1 2 2 2 2 (% yr /, (c) trend in land to atmosphere net CO ﬂux (gC m yr /, (d) trend in NPP (gC m yr /, and (e) trend in RH 2 2 (D RHC wildﬁreC Riverine C ﬂux) (gC m yr /. www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 664 S. Sitch et al.: Recent trends and drivers of regional sources growing season lengthD 0.361 days yr . There is less agreement among models on reproducing the observed browning trends in some regions of the boreal forest. DGVMs simulate an ensemble mean RH of 21.8 4.6 Pg C yr across the northern land region (Table 3). All DGVMs simulate an increase in northern RH over the period 1990–2009, with a signiﬁcant trend in RH of 0.063 0.02 Pg C yr (P < 0.01) (Table 3). DGVMs simulate larger mean RH in temperate compared to boreal regions, yet smaller positive trends for Asia (Table S6 in the Supplement). This is because of relatively smaller increases in substrate (i.e. NPP) in temperate regions and greater warming in boreal regions stimulating microbial decomposition, reducing mean residence time of carbon in soils (MRTD soil carbon / RH; see Fig. S3). No signiﬁcant trends in the wildﬁre ﬂux were reported by any of the DGVMs for the northern land region. However, DGVMs agree on simulating a small negative trend in wild- ﬁre ﬂux across boreal North America and tundra. Figure 7. Ensemble-mean trends in the onset (a, b), offset (c, d), Tropical land and length of growing season in days (e, f) for the ensemble mean (left) compared with satellite-derived estimates (right). All DGVMs simulate an increasing land C sink over recent decades, in response to changes in climate and atmospheric CO concentration over the tropical 0.065 0.025 Pg C yr (P < 0.01). This can be largely land region, with an ensemble mean land–atmosphere attributed to the response of ecosystems to elevated CO ﬂux of 0.96 0.43 Pg C yr (Table 3, Fig. S4) (Table S2). and trend of 0.04 0.01 Pg C yr (P D 0.05), or No signiﬁcant trends in the wildﬁre ﬂux were reported by 2 2 0.88 0.33 g C m yr on an area basis (Table 3, any of the DGVMs for the tropical land region. However, Table S4 in the Supplement Fig. S5). This represents 65 % DGVMs agree on simulating a negative trend in wildﬁre ﬂux of the increase in global land sink over the last two decades across equatorial Africa and tropical Asia. across the tropical land, which covers 27 % of the land sur- face (Table S4). DGVMs simulate signiﬁcant negative trends Southern land (i.e. increasing sinks) across tropical Asia and equatorial Africa (Table S4). All DGVMs agree on a net land sink over the southern DGVMs simulate an ensemble mean NPP of land during the last two decades, with an ensemble mean 1 1 26.6 4.35 Pg C yr averaged over the tropical re- land–atmosphere ﬂux of 0.38 0.29 Pg C yr (Table 3, gion, representing 42 % of the global total (Table 3). All Fig. S4). Although all DGVMs simulate an increase in the DGVMs simulate a signiﬁcant increase in tropical NPP land sink over the southern extratropics, with an ensem- over this period, with an ensemble mean trend in NPP of ble mean land–atmosphere trend of 0.02 0.02 Pg C yr 2 2 2 0.10 0.03 Pg C yr (P D 0.00) for S_L2 (Table 3). This (P D 0.20) (Fig. S5) or 0.58 0.45 g C m yr on an compares to a trend of 0.09 0.03 Pg C yr (P < 0.01) and area basis, only trends for HYL and ORC are signiﬁcant at 0.02 0.02 Pg C yr (P D 0.33) over the same period for the 95 % conﬁdence level (Table 3). Ensemble mean trends the S_L1 (CO only) and S_L2–S_L1 (the climate effect), are signiﬁcant for temperate South American and south- respectively (Tables S2, S3). Again, the simulated trend in ern African regions at 0.005 0.005 Pg C yr (P D 0.05) NPP is dominated by the simulated response of ecosystems and 0.022 0.011 Pg C yr (P D 0.01), respectively (Ta- to elevated atmospheric CO content. DGVMs simulate ble S4). For southern Africa, all DGVMs simulate an in- positive NPP trends across tropical South American forests, crease in the land sink in response to climate variability and tropical Asia, equatorial Africa, and North African savanna change over this period (ﬁve out of nine are signiﬁcant at the (Table S5). Nevertheless there are some areas within tropical 90 % conﬁdence level) (Table S7 in the Supplement, Fig. 6). South America and North African savanna regions with In contrast, the simulated decrease in land sink for temperate negative trends in NPP (Fig. 6). South America is associated with a decrease in precipitation All DGVMs simulate an increase in RH over over 1990–2009 (Table S8 in the Supplement). the period 1990–2009, with an ensemble mean RH DGVMs simulate an ensemble mean NPP of 1 1 of 24.49 4.75 Pg C yr (Table 3) and trend of 12.3 3.53 Pg C yr over the southern extratropics, Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 665 which represents 20 % of the global total (Table 3) western USA, southern Asia, northern boreal China, south- across 24 % of the land surface. All DGVMs simulate eastern South America, and western and southern Australia an increase in NPP over this period, with a signiﬁcant are simulated to have negative NPP trends over the last two ensemble mean trend of 0.05 0.03 Pg C yr (P D 0.01), decades, as a result of reduced rainfall, and there is a less i.e. the southern land region accounts for around 25 % of negative trend in RH, possibly due to a reduction in micro- the simulated global trend in NPP. Southern Africa is the bial respiration rates with increased soil dryness. The warm- only southern sub-region with a signiﬁcant trend in NPP ing and drying in central Asia (northeastern China and Mon- of 0.041 0.018 Pg C yr (P < 0.01) (Table S5), due to golia) and southern Australia is simulated to reduce the rate a positive response of plant production to both CO and of microbial decomposition in these regions (Fig. S3), which climate, and is likely in response to increases in precipitation partly opposes the NPP-driven lagged decrease in RH. The over the last two decades (Table S7, Fig. 5). source trend in eastern Europe is simulated as a combina- DGVMs simulate an ensemble mean RH of tion of a negative trend in NPP, as a result of a combination 11.20 3.60 Pg C yr over the southern land region of elevated temperatures and reduced precipitation (i.e. soil (Table 3). All DGVMs simulate an increase in RH over drying), and a positive trend in RH driven by increasing tem- the period 1990–2009, with a signiﬁcant trend in the perature, despite reduced plant litter input. ensemble mean RH of 0.03 0.02 Pg C yr (P < 0.01). This is only partly explained by the response of ecosys- .2 Ocean tems to elevated CO ; over southern Africa the ensemble mean trend in RH from S_L1 is 0.01 0.01 Pg C yr Regional ﬂuxes (P < 0.01), and a climate-induced positive trend in RH of The large-scale distribution of the modelled mean surface 0.01 0.00 Pg C yr (P < 0.01) (Table S2, S7). No signiﬁcant trends in the wildﬁre ﬂux were reported by ﬂuxes consists of strong outgassing in the tropical regions, any of the DGVMs for the southern land region. However especially in the Paciﬁc, and broad regions of uptake in the DGVMs agree on simulating a negative trend in wildﬁre ﬂux mid-latitudes, with a few regions in the high latitudes of par- across southern Africa. ticularly high uptake, such as the North Atlantic (Fig. 9). This In summary, the globally increasing trend in land carbon pattern is largely the result of the exchange ﬂux of natural sink is about two-thirds due to tropical ecosystems and one- CO that balances globally to a near-zero ﬂux, but exhibits third due to the southern land region, with zero contribution regionally strong variations (Gruber et al., 2009). Superim- from northern land. This partitioning in trend is quite differ- posed on this natural CO ﬂux pattern is the uptake of an- ent from the mean carbon sink ﬂuxes themselves, which is thropogenic CO , which is taken up everywhere, but with more like 43V 41V 16 (northern : tropical : southern). substantial regional variation. Large anthropogenic CO up- take ﬂuxes occur in the regions of surface ocean divergence, Qualitative change in processes such as the equatorial Paciﬁc and particularly the Southern Ocean (Sarmiento et al., 1992; Gloor et al., 2003; Mikaloff A qualitative assessment of the differential responses of Fletcher et al., 2006). This is a result of the divergence caus- the underlying land processes to changes in environmental ing waters to upwell to the surface which have not been ex- conditions, and their contribution to the sink–source land– posed to the atmosphere for a while, thereby permitting them atmosphere ﬂux trends is shown in Fig. 8. Many regions to take up a substantial amount of anthropogenic CO . This are simulated to have a negative land–atmosphere ﬂux trend, reduces the outgassing that typically characterises these re- with increases in NPP leading increases in RH. However gions as a result of these upwelling waters also bringing with there are locations with positive trends over the period 1990– them high carbon loads from the remineralisation of organic matter. 2009, i.e. red colours in Fig. 8. In some regions models sim- Over the analysis period, the air–sea CO ﬂuxes exhibit ulate a positive trend in NPP but an even larger positive trend in RH (eastern Europe, southeastern USA, Amazonia, south- only a remarkably small trend in most places, with some re- ern China, North America tundra). Warming is likely to en- gions increasing in uptake, while others show a positive ﬂux hance both NPP and RH in high-latitude ecosystems, but pri- anomaly, i.e. lesser uptake. Thus the small global trend in marily RH in low latitudes. Reduced precipitation may par- ocean uptake over the 1990–2004 analysis period is a result tially or fully offset the beneﬁts of elevated atmospheric CO of also the individual regions having relatively modest trends. abundance on NPP, and the response of RH to changes in precipitation is not obvious, as this is inﬂuenced by the ini- Process analysis tial soil moisture status. This is because microbial activity increases with increasing soil moisture at low moisture lev- The regional ﬂux trends are, however, much smaller than ex- els, before reaching a maximum activity, and then begins to pected from an ocean with constant circulation that is only decline as water completely ﬁlls the soil pore spaces and oxy- responding to increasing atmospheric CO , and hence would gen becomes more limiting to respiration. Locations in the tend to increase its uptake of anthropogenic CO through www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 666 S. Sitch et al.: Recent trends and drivers of regional sources Figure 8. Qualitative change in processes over the period 1990–2009. Negative trend in land–atmosphere net CO ﬂux: enhanced NPP > enhanced RH (D RHC wildﬁreC riverine C ﬂux) (pale blue); enhanced NPP, reduced RH (turquoise); and reduced NPP < reduced RH (dark blue). Positive trend in land–atmosphere net CO ﬂux: enhanced NPP < enhanced RH (dark red); reduced NPP, enhanced RH (red); and reduced NPP > reduced RH (pink). Figure 9. Gridded maps of the ensemble mean sea–air CO ﬂux over the period 1990–2004 (a), standard deviation of the mean ﬂux across the four OBGCMs (b), the trend in the net ﬂux across the ensemble (c), and the standard deviation of the trend (d). Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 667 time (Fig. 10). In the absence of climate variability and change, all regions would have ﬂux density trends of more 2 2 than 0.05 g C m yr , with some regions, such as the 2 2 Southern Ocean, exceeding 0.15 g C m yr . However climate variability and change compensate for these neg- ative trends in every single region by increasing them by 2 2 C0.04 g C m yr or more (with the exception of the South Paciﬁc), such that the overall trends ﬂuctuate from region to region around zero (Fig. 10). The largest reductions in trends are simulated to occur in the North and equatorial Paciﬁc and in the North Atlantic, where they even cause a change in the Figure 10. Regional ocean ﬂux trends from 1990 through to 2004 sign of the overall trend. A similar, although slightly more for the standard case, i.e. variable climate and increasing CO (sim- moderate, pattern is seen if the analysis is undertaken for the ulation S_O2), and for the constant climate case (simulation S_O1), entire 1990–2009 period with two models only. The most im- and their difference (S_O2–S_O1). Ocean regions comprise North portant difference is found in the North Atlantic, where the Paciﬁc (NP), equatorial Paciﬁc (EP), South Paciﬁc (SP), North At- climate variability impact is substantially smaller, and not lantic (NAT), equatorial/South Atlantic (EQ), Indian Ocean (IO), offsetting the anthropogenic CO trend when analysed for and Southern Ocean (SO), and world oceans (W). 1990–2009. The mechanisms driving the oceanic ﬂux trends differ be- tween the analysed regions. Attribution of regional trends to is consistent with the DGVM projections presented here. speciﬁc processes or changes in speciﬁc state variables in Lewis et al. (2009b) found broad agreement between biomass the different models is a work in progress, and is difﬁcult trends from observations and from a suite of carbon cycle to achieve with high conﬁdence as yet. This is due to the an- models applied with 20th century forcing of climate and at- tagonistic effect of ocean warming on CO solubility and on mospheric CO content, using a similar protocol to the cur- 2 2 dissociation of carbonic acid into bicarbonate and carbon- rent analysis. DGVMs suggest a large component of the up- ate, as well as to the complex changes in ocean circulation take trend is associated with a positive NPP response to ele- and mixing, which themselves inﬂuence the biological car- vated CO , which is broadly consistent with the enhancement bon pumps of the ocean. of forest production due to CO observed in FACE experi- ments (Norby et al., 2005), although they are largely located in temperate forest ecosystems. However, recent studies have highlighted the role of nitrogen in limiting the long-term CO 4 Discussion 2 response (Canadell et al., 2007; Norby et al., 2010) in these ecosystems. The long-term plant response to elevated CO is 4.1 Land 2 likely affected by nutrients and its impact on plant C alloca- The DGVMs used in this study simulate an increase in tion (Zaehle et al., 2014), however only two out of the nine land carbon uptake over the period 1990–2009. The re- models used here (CLM4CN and OCN) include interactive sult is consistent with the earlier ﬁndings of Sarmiento et nutrient cycling (see DGVM characteristics, Table S1). al. (2010), who suggested a large increase in the RLS be- In contrast to the large trend in net C uptake across the tween 1960 and 1988 and between 1989 and 2009 (Table S9 tropics, DGVMs simulate no statistically signiﬁcant trend in the Supplement). The ensemble mean land–atmosphere over the northern land region. In particular, trends in NPP ﬂux increased by1.11 Pg C yr for the same period, com- over temperate regions are smaller than those in boreal re- pared to the estimated RLS increase of0.88 Pg C yr from gions, and are also not signiﬁcant. Many temperate areas ex- Sarmiento et al. (2010). The DGVM ensemble trends in land perienced a decrease in rainfall between 1990 and 2009, and uptake for the globe, northern, tropical, and southern land suffered periods of prolonged and severe drought. Examples regions of 0.06 0.03, 0.00 0.01, 0.04 0.01, and include the drought in the western USA of 2000–2004 (Mc- 0.02 0.02 Pg C yr , respectively, compare favourably Dowell et al., 2008; Anderreg et al., 2012) and the 2003 sum- with the inversion estimates of 0.06 0.04, 0.01 0.01, mer heatwave in Europe (Ciais et al., 2005). Zscheischler et 0.04 0.02, and 0.01 0.01 Pg C yr over the period al. (2014) suggest that negative productivity extremes dom- 1990–2009. Although encouraging, these results should be inated interannual variability in productivity during the pe- interpreted with caution because the inversion accounts for riod 1982–2011; these extremes are evident particularly over any trend in the land use change ﬂux over this period, temperate latitudes. whereas DGVMs had ﬁxed land use. Satellite observations suggest a general greening trend There is empirical evidence of a large increase in biomass in high latitudes, with an earlier onset and longer growing in intact forest in tropical South America and Africa (Pan et season in high-latitude ecosystems, which is reproduced by al., 2011; Baker et al., 2004; Lewis et al., 2009a, b), which the DGVMs. Observations suggest a greening tundra and a www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 668 S. Sitch et al.: Recent trends and drivers of regional sources slower greening and possible browning in some regions of mafrost and active layer thickness) (Dolman et al., 2012). the boreal forest (Tucker et al., 2001; Bhatt et al., 2010), es- In South America, DGVMs agree with inventory-based es- pecially in North America (Beck and Goetz, 2011). In tun- timates on a sink in natural forests (Gloor et al., 2012). dra ecosystems, an earlier onset is attributed to warming DGVMs also agree with other data sources on the sign and and earlier snowmelt. In these ecosystems, the start of the magnitude of the natural land sink over Australia (Harverd growing season corresponds to near peak in radiation. Thus et al., 2013). Over Europe DGVMs simulate a smaller mean any temperature-induced earlier snowmelt (McDonald et al., land sink than the synthesis study suggests (Luyssaert et al., 2004; Sitch et al., 2007a) is likely to enhance plant produc- 2012). However, the regional synthesis was conducted over tion. Warming may not have such a great effect on the end the shorter time period 2001–2005. For the Arctic, DGVMs of the growing season in Arctic tundra ecosystems, as this tend to simulate a lower sink than regional process-based may be driven primarily by radiation. DGVMs simulate a models (McGuire et al., 2012). However, over the 1990– signiﬁcant positive trend in NPP in boreal North America 2006 period, DGVMs are in line with observations and in- and boreal Asia and the circumpolar tundra. Nitrogen limi- versions on the magnitude and sign of the natural land sink, tation is also likely to constrain the productivity at high lati- and DGVM results also suggest a sink trend in line with ob- tudes, but it was not possible to quantify N-limitation effects servations. DGVMs simulate a land sink over South Asia in on regional trends in this study. agreement with inversions; however there were limited data DGVMs simulate decreasing NPP across northeastern to compare trends from DGVMs and other products (Patra China and Mongolia, contributing to the overall decreas- et al., 2013). For East Asia, DGVM results agree remark- ing land uptake trend, in response to recent climate. In ably well with remote sensing model–data fusion and inverse a regional study, Poulter et al. (2013) investigated the dif- models on the magnitude of the land sink over the period ferential response of cool semi-arid ecosystems to recent 1990–2009. Finally, for Africa, DGVMs are broadly consis- warming and drying trends across Mongolia and northern tent with inventory- and ﬂux-based estimates in simulating a China, using multiple sources of evidence, including the LPJ land sink over Africa, albeit of lower magnitude (Valentini et DGVM, FPAR remotely sensed data (derived from GIMMS al., 2014). NDVI3g), and tree-ring widths. They found coherent patterns of high sensitivity to precipitation across data sources, which 4.2 Ocean showed some areas with warming-induced springtime green- ing and drought-induced summertime browning, and limita- The investigated OBGCMs consistently simulate an ocean tions to NPP explained mainly by soil moisture. characterised by a substantial uptake of CO from the at- Browning has occurred as a consequence of regional mosphere, but with a global integrated trend in the last drought, wildﬁre, and insect outbreak, and their interaction, two decades (0.02 0.01 Pg C yr / that is substantially especially in North America (Beck and Goetz, 2011). Distur- smaller than that expected based on the increase in at- bance plays a key role in the ecology of many global ecosys- mospheric CO . Results based on the predictions from tems. For example, wildﬁre plays a dominant role in the car- ocean inversion and ocean Green function methods (Mikaloff bon balance of boreal forest in central Canada and other Fletcher et al., 2006; Gruber et al., 2009; Khatiwala et al., regions (Bond-Lamberty et al., 2007), and insect outbreaks 2009) suggest an increase in ocean uptake with a trend of like the mountain pine beetle epidemic between 2000 and the order of 0.04 Pg C yr over the analysis period (see 2006 in British Colombia, Canada, resulted in the transition also Wanninkhof et al., 2013). These latter methods assume of forests from a small carbon sink to a source (Kurz et al., constant circulation, while our simulations here include the 2008). In general, disturbance and forest management are in- impact of climate variability and change. adequately represented by the current generation of DGVMs, Our analyses reveal that recent climate variability and even though several models include simple prognostic wild- change has caused the ocean carbon cycle to take up less ﬁre schemes (Table S1), while some are starting to include CO from the atmosphere than expected on the basis of the other disturbance types such as insect attacks (Jönsson et al., increase in atmospheric CO , i.e. it reduces the efﬁciency of 2012) and windthrow (Lagergren et al., 2012). The exten- the ocean carbon sink. Globally, we ﬁnd that this efﬁciency sion of DGVMs to include representations of globally and reduction is primarily a result of ocean warming, while, re- regionally important disturbance types and their response to gionally, many more processes (e.g. wind changes, alkalin- changing environmental conditions is a priority. ity/DIC concentration changes) are at play. In Table 4, DGVM results are compared with the REC- Is this reduction in uptake efﬁciency over the analy- CAP synthesis papers documenting carbon sources and sinks sis period the ﬁrst sign of a positive feedback between for individual regions. Note that DGVMs provided one global warming and the ocean carbon cycle – or, alterna- source of evidence for some regional papers. Over Russia, tively, could it just reﬂect natural decadal-scale variability DGVMs agree on a sink yet underestimate that sink’s mag- in air–sea CO ﬂuxes? Without a formal attribution study, nitude, likely related to soil respiration (which is unsurpris- it is not possible to provide a ﬁrm answer. We suspect ing, as many DGVMs have a limited representation of per- that the majority of the trend in the efﬁciency is due to Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 669 Table 4. Ensemble DGVM regional NBP mean comparison with RECCAP regional chapter analyses. Region DGVM mean Region Inventory-based Flux-based Inversion Best estimate NBP (TgC yr ) processed-based models Russia 199 761 709 653 South America (forest) 472 211 570 170 (1990–94) 530 140 (1995–99) 450 250 (2000–04) 150 230 (2005–09) Africa 410 310 740 1190 1340 1320 50 280 (LULCC 510 280) Australia & New Zealand 70 78 36 29 (LULCC 18 7) Europe 179 92 891 155 (2001–05) Arctic (1990–2006) 86 177 96 South Asia 210 164 35.4 (1997–06) 317 to88.3 (2007–08) East Asia 224 141 293 33 combined 270 507 inventory–EO-ﬂux approach “natural” decadal-scale variability; however, largely based tion with the atmosphere. Changes in biogeochemical and on the results of McKinley et al. (2011) and Fay and McKin- ecosystem processes, such as locally varying gas exchange ley (2013), who showed that whereas trends in oceanic pCO velocities, phytoplankton blooms, and associated particle (and air–sea CO ﬂuxes) are variable on a decadal timescale, ﬂux pulses, can lead to regional interannual variations in air– they do converge towards atmospheric pCO trends when sea CO ﬂuxes, but may partially cancel for averages over 2 2 analysed over a longer 30-year period for most global re- larger regions. With ocean observations only over about a gions. Nevertheless, they also show that warming (partly two-decade time frame, it is difﬁcult to quantify longer-term driven by anthropogenic climate change) in the permanently trends due to other proposed mechanisms: a gradual slowing- stratiﬁed subtropical gyre of the North Atlantic has started to down of meridional overturning circulation due to a strength- reduce ocean uptake in recent years. In the Southern Ocean, ening of density stratiﬁcation; redissolution of CaCO sedi- where Le Quéré et al. (2007) and Lovenduski et al. (2008) ment from the seaﬂoor associated with fossil fuel neutraliza- used models to suggest a reduction in ocean carbon uptake tion; and potential changes in biogenic particle ﬂuxes due to efﬁciency over the past 25 years in response to increasing carbon overconsumption and changing ballasting (cf. Keller Southern Ocean winds, Fay and McKinley (2013) concluded et al., 2014). Whether more complex models will render bet- that the data are insufﬁcient to draw any conclusions. ter results will depend on how well the additional free pa- We should note that the associated uncertainties remain rameters in more complex biogeochemical models can be large. Of particular concern is the moderate success of the constrained by measurements. So far, more complex – and models in simulating the time-mean ocean sinks and their hence potentially more realistic – models do not necessarily long-term seasonal cycle (e.g. McKinley et al., 2006). Fur- give better results than the present nutrient-phytoplankton- thermore, some of the models underestimate the oceanic up- zooplankton-detritus (NPZD)-type models models as applied take of transient tracers such as anthropogenic radiocarbon here (Le Quéré et al., 2005; Kriest et al., 2010). (see e.g. Graven et al., 2012). Such a reduction in the oceanic 4.3 Reducing uncertainty in regional sinks uptake efﬁciency is also not suggested by independent mea- sures of oceanic CO uptake, such as the atmospheric O / N 2 2 2 In order to better quantify the regional carbon cycle and its method (Manning and Keeling, 2006; Ishidoya et al., 2012), trends, DGVM and ocean carbon cycle models need to im- although the large uncertainties in these estimates make the prove both process representations and model evaluation and determination of trends in uptake highly uncertain. benchmarking (Luo et al., 2012). There is a need for up- All the models have been tuned to reproduce data syn- to-date global climate and land use and cover change data thesis on ocean surface pCO (Pfeil et al., 2013; Takahashi sets to force the DGVMs, as well as a deeper investigation et al., 2009) and deep ocean (Key et al., 2004) reasonably of the quality and differences between the different reanaly- well. Speciﬁc systematic data assimilation procedures, how- sis products used to force ocean carbon cycle models. Also, ever, have not been applied. On decadal timescales, the ocean techniques such as detection and attribution can be applied to CO ﬂux feedback to climate change (change in hydrogra- elucidate trends in the regional carbon cycle and their drivers. phy and circulation) and rising ambient CO (change in CO 2 2 buffering) reacts only slowly on the global average due to the long timescales of oceanic motion and marine CO equilibra- www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 670 S. Sitch et al.: Recent trends and drivers of regional sources 4.3.1 Model evaluation and benchmarking driven overturning circulation in the Southern Ocean, where non-eddy-resolving models indicate a strong sensitivity of Piao et al. (2013) evaluated the DGVM model results for the overturning circulation and ocean carbon uptake to sur- their response to climate variability and to CO trends, and face wind stress (Le Quéré et al., 2007; Lovenduski et al., the seasonal cycle of CO ﬂuxes were benchmarked in Peng 2008). Some eddy-resolving models, in contrast, suggest that et al. (2014). Piao et al. (2013) found DGVMs to simulate enhanced wind stress is dissipated by increased eddy activ- higher mean and interannual variations (IAVs) in gross pri- ity, leading to only a small increase in overturning (Böning, mary production than a data-driven model (Jung et al., 2011), et al., 2008), although more recent results indicate a larger particularly in the tropics; however, this is the region where response (Gent and Danabasoglu, 2011; Matear et al., 2013). the data-driven model is most uncertain. DGVMs were able to capture the IAVs in RLS, although the simulated land– 4.3.3 Model structure atmosphere net CO ﬂux appears too sensitive to variations in precipitation in tropical forests and savannas. However, Poul- There is a need for improved representation of ecological ter et al. (2014) found an increase in the sensitivity of the net processes in land and ocean models, e.g. nutrient cycling ﬂux to precipitation over the last three decades across conti- (N, P), demographic dynamics, disturbance (wildﬁre, wind- nental Australia. Piao et al. (2013) found that the simulated throw, insects), land use and land cover change in land mod- net CO ﬂux was more sensitive than productivity to tem- els, and better representation of the key functional diversity perature variations. When compared to ecosystem warming in ocean and land biogeochemical models. DGVMs need to experiments the DGVMs tend to underpredict the response represent land use and land cover changes, forest manage- of NPP to temperature at temperate sites. DGVMs simulated ment, and forest age in order to improve estimates of the an ensemble mean NPP enhancement comparable to FACE regional and global land carbon budget. There have been experiment observations (Piao et al., 2013). However, mod- recent developments to include nutrient dynamics, mostly elling of ecosystem function in water-stressed environments nitrogen, in global land biosphere models (as reviewed by and changes in plant water use with elevated CO remains a Zaehle and Dalmonech, 2011). Too few model simulations challenge for DGVMs (Morales et al., 2005; Keenan et al., are available to date to allow for an ensemble model trend 2009; De Kauwe et al., 2013). assessment. However, a few general trends appear robust: There is a critical need for comprehensive model bench- as evident from Table 3, CN models generally show less marking, as a ﬁrst step to attempt to reduce model un- of a response to increasing atmospheric CO due to nitro- certainty. Several prototype carbon cycle benchmarking gen limitation of plant production. N dynamics further al- schemes have been developed (Randerson et al., 2009; Cad- ter the climate–carbon relationship, which tend to reduce the ule et al., 2010). A more in-depth evaluation and community C loss from temperate and boreal terrestrial ecosystems due benchmarking set needs to be agreed upon and implemented to warming – but with a considerable degree of uncertainty which also evaluates models for their implicit land response (Thornton et al., 2009; Sokolov et al., 2008; Zaehle et al., timescales (especially relevant in the discussion on future tip- 2010). Changes in the nitrogen cycle due to anthropogenic ping elements and non-linear future responses) and for the reactive nitrogen additions (both fertiliser to croplands and simulated carbon, water, and nutrient cycles. New emerging N deposition on forests and natural grasslands) further mod- frameworks now exist (Blyth et al., 2011; Abramowitz, 2012; ify the terrestrial net C balance and contribute with 0.2 Luo et al., 2012; Dalmonech and Zaehle, 2013; Harverd to 0.5 Pg C yr to the current land sink (Zaehle and Dal- et al., 2013). One example within RECAPP is a multiple- monech, 2011). Zaehle et al. (2011), using the OCN model, constraint approach applied to reduce uncertainty in land car- estimated the 1995–2005 trend in land uptake due to N de- bon and water cycles over Australia (Haverd et al., 2013). position to be1.1 1.7 Tg C yr , with strong regional dif- ferences depending on the regional trends in air pollution and 4.3.2 Model resolution reactive N loading of the atmosphere and the nitrogen status of the ecosystems, which are generally lower in less respon- Simulated ocean carbon dynamics may be sensitive to sive ecosystems close to nitrogen saturation highly polluted horizontal resolution, particularly as model resolution im- regions. The DGVMs applied here do not consider the P cy- proves sufﬁciently to adequately capture mesoscale eddies. cle; P limitation on land carbon uptake may be particularly Mesoscale turbulence inﬂuences the ocean carbon cycle in important in tropical forests and savannas (Edwards et al., a variety of ways, and the present eddy parameterisations 2005; Wang et al., 2010; Zhang et al., 2014). may not adequately capture the full range of effects and There are several additional land processes that have not the responses to climate variability and change. For exam- been considered in this current multi-model analysis. These ple, mesoscale processes are thought to modulate biologi- include the effects of aerosols and tropospheric ozone on the cal productivity by altering the supply of limiting nutrients carbon cycle. Unlike a global forcing agent such as CO , (Falkowski et al., 1991; McGillicuddy et al., 1998; Gruber the effects of air pollutants (aerosols, NO , and O /, with x 3 et al., 2011). A particularly crucial issue involves the wind- their shorter atmospheric lifetimes, are at the regional scale. Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 671 Aerosol-induced changes in radiation quantity and quality certainty in simulated regional-scale GPP (Jung et al., 2007; (i.e. the ratio of diffuse to direct) affect plant productivity and Quaife et al., 2008) and 3.5 % uncertainty for global NPP. the land sink (Mercado et al., 2009). From around 1960 until Climate forcing uncertainty tends to have larger effects on the 1980s, radiation levels declined across industrialised re- carbon ﬂux uncertainty than land cover (Hicke, 2005; Poul- gions, a phenomenon called “global dimming”, followed by ter et al., 2011), with up to 25 % differences in GPP reported a recent brightening in Europe and North America with the over Europe (Jung et al., 2007) and a 10 % difference for adoption of air pollution legislation. Reductions in acid rain global NPP (Poulter et al., 2011). Climate forcing uncertainty have been found to greatly inﬂuence trends in riverine DOC, and land cover (i.e. PFT distributions) can alter long-term vegetation health, and rates of soil organic matter decompo- trends in land to atmosphere net CO ﬂux and interannual sition. Tropospheric ozone is known to be toxic to plants and variability of carbon ﬂuxes to climate (Poulter et al., 2011). lead to reductions in plant productivity, and thus reduce the The DGVMs applied here did not consider LULCC. This efﬁciency of the land carbon sink (Sitch et al., 2007b; Anav et is an active area of research; models need a consistent im- al., 2011). Drivers of the land carbon sink related to air pollu- plementation of LULCC. Uncertainties in the simulated net tion – e.g. N deposition, acid precipitation, diffuse and direct land use ﬂux are associated with assumptions on the imple- radiation, and surface O – have varied markedly in space mentation of LULCC gridded maps (e.g. whether conversion and time over recent decades. Although likely important for to cropland in a grid-cell is taken preferentially from grass- regional carbon cycle trends, quantifying these effects is be- land, forest, or both), simulated biomass estimates, and sub- yond the scope of the present study. sequent decomposition rates. However DGVMs offer the ex- The Pinatubo eruption in 1991, at the start of the study pe- citing prospect of disentangling the component ﬂuxes asso- riod, had a major inﬂuence on many carbon cycle processes, ciated with land use (e.g. direct emissions and legacy ﬂuxes) leading to an enhanced land sink over the period 1991–1993. and separating the environmental and direct human impacts This has been attributed to a combination of cooling-induced on the net LU ﬂux (Gasser and Ciais, 2013; Pongratz et al., reductions in high-latitude respiration and enhanced produc- 2014; Stocker et al., 2014). tivity associated with changes in diffuse radiation (Jones and Cox, 2001; Lucht et al., 2002; Peylin et al., 2005; Mercado et al., 2009; Frölicher et al., 2013). The direct effect of aerosols 5 Conclusions on climate drivers is implicitly included in this study (i.e. re- sponses to high-latitude cooling, tropical drying, reduced net Land models suggest an increase in the global land net C incoming solar radiation); however diffuse radiation effects uptake over the period 1990–2009, with increases in trop- are not included. ical and southern regions and negligible increase in north- Similar gaps need to be addressed in ocean biogeochemi- ern regions. The increased sink is mainly driven by trends in cal models. The ecosystem modules in the current generation NPP, in response to increasing atmospheric CO concentra- of OBGCMs lack the ability to assess many of the suggested tion, and modulated by change in climate. Over the same pe- mechanisms by which climate and ocean acidiﬁcation could riod, ocean models suggest a negligible increase in net ocean alter marine biogeochemistry and ocean carbon storage. Pro- C uptake – a result of ocean warming counteracting the ex- posed biological processes that could inﬂuence ocean car- pected increase in ocean uptake driven by the increase in at- bon uptake and release involve, for example, decoupling of mospheric CO . At the sub-regional level, trends vary both in carbon and macronutrient cycling, changes in micronutrient sign and magnitude, particularly over land. Areas in temper- limitation, variations in elemental stoichiometry in organic ate North America, eastern Europe, and northeastern China matter, and changes in the vertical depth scale for the res- show a decreasing regional land sink trend, due to regional piration of sinking organic carbon particles (e.g. Boyd and drying, suggesting a possibility for a transition to a net car- Doney, 2003; Sarmiento and Gruber, 2006). Some advances bon source in the future if drying continues or droughts be- have been made with the incorporation of dynamic iron cy- come more severe and/or frequent. In the ocean, the trends cling and iron limitation, multiple plankton groups, calciﬁ- tend to be more homogeneous, but the underlying dynamics cation, and nitrogen ﬁxation (Le Quéré et al., 2005). How- differ greatly, ranging from ocean warming, to winds, and to ever, the evaluation of these aspects of the models is cur- changes in circulation/mixing and ocean productivity, mak- rently hindered by both data- and process-level information ing simple extrapolations into the future difﬁcult. limitations. Our conclusions need to be viewed with several important caveats: only a few models include a fully coupled carbon– 4.3.4 Climate and land use and cover data sets nitrogen cycle, and no model included land use and land cover changes. Ocean models tend to be too coarse in reso- In addition to model structure, the choice of climate forc- lution to properly represent important scales of motions and ing and model initial conditions can also contribute to dif- mixing, such as eddies and other mesoscale processes, and ferences in the simulated terrestrial carbon sink. At regional coastal boundary processes. Furthermore, their representa- scales, differences in land cover can introduce 10 % un- tion of ocean ecosystem processes and their sensitivity to www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 672 S. Sitch et al.: Recent trends and drivers of regional sources climate change and other stressors (e.g. ocean acidiﬁcation, Castro, J., Allard, G., Running, S., Semerci, A., and Cobb, N.: A global overview of drought and heat-induced tree mortality deoxygenation, etc.; Gruber, 2011; Boyd, 2011) is rather reveals emerging climate change risks for forests, Forest Ecol. simplistic. Manag., 4, 660–684, doi:10.1016/j.foreco.2009.09.001, 2010. There is a need for detailed model evaluation and bench- Anav, A., Menut, L., Khvorostyanov, D., and Viovy, N.: Impact of marking in order to reduce the uncertainty in the sinks in tropospheric ozone on the Euro-Mediterranean vegetation, Glob. the land and ocean and, particularly, in how these sinks have Change Biol., 17, 2342–2359, 2011. changed in the past and how they may change in the fu- Anderegg, W. R. L., Berry J. A., Smith, D. D., Sperry J. S., An- ture. For land ecosystems, a concerted effort is needed in the deregg, L. D. L., and Field, C. B.: The roles of hydraulic and car- DGVM community to incorporate nutrient cycling as well as bon stress in a widespread climate-induced forest die-off, P. Natl. land use and land cover change. For the oceans, models need Acad. Sci. USA, 109, 233–237, doi:10.1073/pnas.1107891109, to improve their representation of unresolved physical trans- port and mixing processes, and ecosystem models need to Angert, A., Biraud, S., Bonﬁls, C., Henning, C. C., Buermann, W., Pinzon, J., Tucker, C. J., and I. Fung: Drier summers cancel out evolve to better characterise their response to global change. the CO uptake enhancement induced by warmer springs, P. Natl. Acad. Sci., 102, 10823–10827, 2005. Assmann, K. M., Bentsen, M., Segschneider, J., and Heinze, C.: The Supplement related to this article is available online An isopycnic ocean carbon cycle model, Geosci. Model Dev., 3, at doi:10.5194/bg-12-653-2015-supplement. 143–167, doi:10.5194/gmd-3-143-2010, 2010. Baker, T. R., Phillips, O. L., Malhi, Y., Almeida, S., Arroyo, L., Di Fiore, A., Erwin, T., Higuchi, N., Killeen, T. J., Laurance, S. G., Laurance, W. F., Lewis, S. L., Monteagudo, A., Neill, D. A., Nunez Vargas, P., Pitman, N. C. A., Silva, J. N. M., and Vasquez Acknowledgements. S. Sitch acknowledges ﬁnancial support by Martinez, R.: Increasing biomass in Amazonian forest plots, Phi- RCUK through NERC (grant no. NE/J010154/). N. Gruber and los. T. R. Soc. B, 359, 353–365, 2004. C. Heinze acknowledge ﬁnancial support by the European Com- Ballantyne, A. P., Alden, C. B., Miller, J. B., Tans, P. P., and White, mission through the EU FP7 projects CARBOCHANGE (grant J. W. C.: Increase in observed net carbon dioxide uptake by no. 264879) and GEOCARBON (grant no. 283080). N. Gruber land and oceans during the past 50 years, Nature, 488, 70–72, was additionally supported through ETH Zurich. S. C. Doney doi:10.1038/nature11299, 2012. acknowledges support from the US National Science Foundation Beck, S. A. and Goetz, S. J.: Satellite observations of high north- (NSF AGS-1048827). P. Friedlingstein, A. Arneth, and S. Zaehle ern latitude vegetation productivity changes between 1982 and acknowledge support by the European Commission through the EU 2008: ecological variability and regional differences, Environ. FP7 project EMBRACE (grant no. 282672). A. Arneth and S. Sitch Res. Lett. 6, 045501, doi:10.1088/1748-9326/6/4/045501, 2011. acknowledge the support of the European Commission-funded Bhatt, U. S., Walker, D. A., Raynolds, M. K., Comiso, J. C., Epstein, project LUC4C (grant no. 603542). The research leading to these H. E., Jia, G., Gens, R., Pinzon, J. E., Tucker, C. J., Tweedie, results received funding from the European Community’s Seventh C. E., and Webber, P. J.: Circumpolar Arctic tundra vegetation Framework Programme (FP7 2007–2013) under grant agreement change is linked to sea ice decline, Earth Interact., 14, 1–20, no. 238366. A. Ahlström and B. Smith acknowledge funding through the Mistra Swedish Research Programme on Climate, Im- Blyth, E., Clark, D. B., Ellis, R., Huntingford, C., Los, S., Pryor, pacts and Adaptation (SWECIA). This study is a contribution to the M., Best, M., and Sitch, S.: A comprehensive set of benchmark Lund Centre for Studies of Carbon Cycle and Climate Interactions tests for a land surface model of simultaneous ﬂuxes of water (LUCCI) and the Strategic Research Area Modelling the Regional and carbon at both the global and seasonal scale, Geosci. Model and Global Earth System (MERGE). This is a contribution to the Dev., 4, 255–269, doi:10.5194/gmd-4-255-2011, 2011. Bjerknes Centre for Climate Research (BCCR) and core project Bonan, G. B. and Levis, S.: Quantifying carbon-nitrogen feedbacks BIOFEEDBACK of the Centre for Climate Dynamics (SKD) at in the Community Land Model (CLM4), Geophys. Res. Lett., 37, BCCR. C. Heinze acknowledges support from NOTUR/NorStore L07401, doi:10.1029/2010GL042430, 2010. projects NN2980K and NS2980K. The authors wish to thank Bond-Lamberty, B., Peckham, S. D., Ahl, D. E., and Gower, S. T.: Jonathan Barichivich for discussions on greening trends. Fire as the dominant driver of central Canadian boreal forest car- bon balance, Nature, 450, 89–92, 2007. Edited by: F. Joos Böning, C. W., Dispert, A., Visbeck, M., Rintoul, S. R., and Schwarzkopf, F. U.: The response of the Antarctic Circumpo- lar Current to recent climate change, Nat. Geosci., 1, 864–869, References 2008. Boyd, P. and Doney, S. C.: The impact of climate change and feed- Abramowitz, G.: Towards a public, standardized, diagnostic bench- back process on the ocean carbon cycle, Ocean Biogeochemistry, marking system for land surface models, Geosci. Model Dev., 5, edited by: Fasham, M., Springer, 157–193, 2003. 819–827, doi:10.5194/gmd-5-819-2012, 2012. Boyd, P. W.: Beyond ocean acidiﬁcation, Nat. Geosci., 4, 273–274, Allen, C. D., Macalady, A., Chenchouni, H., Bachelet, D., Mc- Dowell, N., Vennetier, M., Gonzales, P., Hogg, T., Rigling, A., Breshears, D., Fensham, R., Zhang, Z., Kitzberger, T., Lim, J., Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 673 Broecker, W. S., Peng, T.-H., Ostlund, G., and Stuiver, M.: The dis- Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., Canadell, J., tribution of bomb radiocarbon in the ocean, J. Geophys. Res., 90, Chhabra, A., DeFries, R., Galloway, J., Heimann, M., Jones, C., 6953–6970, 1985. Le Quéré, C., Myneni, R. B., Piao, S., and Thornton, P.: Carbon Buermann, W., Lintner, B. R., Koven, C. D., Angert, A., Pinzon, and Other Biogeochemical Cycles, in: Climate Change 2013: J. E., Tucker, C. J., and Fung, I. Y.: The changing carbon cycle The Physical Science Basis, Contribution of Working Group I at Mauna Loa observatory, P. Natl. Acad. Sci., 104, 4249–4254, to the Fifth Assessment Report of the Intergovernmental Panel 2007. on Climate Change, edited by: Stocker, T. F., Qin, D., Plattner, Buitenhuis, E. T., Rivkin, R. B., Sailley, S., and Le Quéré, C.: G.-K., Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Biogeochemical ﬂues through microzooplankton, Global Bio- Y., Bex, V., and Midgley, P. M., Cambridge University Press, geochem. Cy., 24, GB4015, doi:10.1029/2009GB003601, 2010. Cambridge, United Kingdom and New York, NY, USA, 465– Cadule, P., Friedlingstein, P., Bopp, L., Sitch, S., Jones, C. D., 570, 2013. Ciais, P., Piao, S. L., and Peylin, P.: Benchmarking cou- Clark, D. B., Mercado, L. M., Sitch, S., Jones, C. D., Gedney, N., pled climate-carbon models against long-term atmospheric Best, M. J., Pryor, M., Rooney, G. G., Essery, R. L. H., Blyth, E., CO measurements, Global Biogeochem. Cy., 24, Gb2016, Boucher, O., Harding, R. J., Huntingford, C., and Cox, P. M.: The doi:10.1029/2009gb003556, 2010. Joint UK Land Environment Simulator (JULES), model descrip- Canadell, J. G., Pataki D., Gifford, R., Houghton, R. A., Lou, Y., tion – Part 2: Carbon ﬂuxes and vegetation dynamics, Geosci. Raupach, M. R., Smith, P., and Steffen, W.: Saturation of the Model Dev., 4, 701–722, doi:10.5194/gmd-4-701-2011, 2011. terrestrial carbon sink, in: Terrestrial Ecosystems in a Changing Cox, P. M.: Description of the “TRIFFID” dynamic global vegeta- World, edited by: Canadell, J. G., Pataki, D., and Pitelka, L., 59– tion model, Hadley Centre Technical Note, 24, 1–16, 2001. 78, The IGBP Series, Springer-Verlag, Berlin Heidelberg, 59–78, Dalmonech, D. and Zaehle, S.: Towards a more objective evalua- 2007. tion of modelled land-carbon trends using atmospheric CO and Canadell, J. G., Ciais, P., Gurney, K., Le Quéré, C., Piao, S., Rau- satellite-based vegetation activity observations, Biogeosciences, pach, M. R., and Sabine, C. L.: An international effort to quantify 10, 4189–4210, doi:10.5194/bg-10-4189-2013, 2013. regional carbon ﬂuxes, EOS T. Am. Geophys. Un., 92, 81–82, De Kauwe, M. G., Medlyn, B. E., Zaehle, S., Walker, A. P., Dietze, 2011. M. C., Hickler, T., Jain, A. K., Luo, Y., Parton, W. J., Prentice, Canadell, J. G., Ciais, P., Sabine, C., and Joos, F. (Eds.): REgional C., Smith, B., Thornton, P. E., Wang, S., Wang, Y-P., Wårlind, Carbon Cycle Assessment and Processes (RECCAP), Special D., Weng, E. S., Crous, K, Y., Ellsworth, D. S., Hanson, P. J., issue, Biogeosciences, http://www.biogeosciences-discuss.net/ Seok-Kim, H., Warren, J. M., Oren, R., and Norby, R. J.: Forest special_issue83.html, 2013. water use and water use efﬁciency at elevated CO : a model-data Chevallier, F., Ciais, P., Conway, T. J., Aalto, T., Anderson, B. E., intercomparison at two contrasting temperate forest FACE sites, Bousquet, P., Brunke, E. G., Ciattaglia, L., Esaki, Y., Fröhlich, Glob. Change Biol., 19, 1759–1779, 2013. M., Gomez, A., Gomez-Pelaez, A. J., Haszpra, L., Krummel, P. Denman, K. L., Brasseur, G., Chidthaisong, A., Ciais, P., Cox, P. B., Langenfelds, R. L., Leuenberger, M., Machida, T., Maignan, M., Dickinson, R. E., Hauglustaine, D., Heinze, C., Holland, E., F., Matsueda, H., Morguí, J. A., Mukai, H., Nakazawa, T., Peylin, Jacob, D., Lohmann, U., Ramachandran, S., da Silva Dias, P. L., P., Ramonet, M., Rivier, L., Sawa, Y., Schmidt, M., Steele, L. P., Wofsy, S. C. , and Zhang, X.: Couplings Between Changes in the Vay, S. A., Vermeulen, A. T., Wofsy, S., and Worthy, D.: CO Climate System and Biogeochemistry, in: Climate Change 2007: surface ﬂuxes at grid point scale estimated from a global 21-year The Physical Science Basis, Contribution of Working Group I reanalysis of atmospheric measurements, J. Geophys. Res., 115, to the Fourth Assessment Report of the Intergovernmental Panel D21307, doi:10.1029/2010JD013887, 2010. on Climate Change, edited by: Solomon, S., Qin, D., Manning, Ciais, P., Tans, P. P., Trolier, M., White, J. W. C., and Francey, R. J.: M., Chen, Z., Marquis, M., Averyt, K. B., Tignor, M., and Miller, A large northern hemisphere terrestrial CO sink indicated by the H. L., Cambridge University Press, Cambridge, United Kingdom 13 12 C = C ratio of atmospheric CO , Science, 269, 1098–1102, and New York, NY, USA, 499–588, 2007. doi:10.1126/science.269.5227.1098, 1995. Dentener, F., Stevenson, D., Ellingsen, K., van Noije, T., Schultz, Ciais, P., Reichstein, M., Viovy, N., Granier, A., Ogee, J., Al- M., Amann, M., Atherton, C., Bell, N., Bergmann, D., Bey, I., lard, V., Aubinet, M., Buchmann, N., Bernhofer, C., Carrara, Bouwman, L., Butler, T., Cofala, J., Collins, B., Drevet, J., Do- A., Chevallier, F., De Noblet, N., Friend, A. D., Friedlingstein, herty, R., Eickhout, B., Eskes, H., Fiore, A., Gauss, M., Hauglus- P., Grunwald, T., Heinesch, B., Keronen, P., Knohl, A., Krin- taine, D., Horowitz, L., Isaksen, I. S. A., Josse, B., Lawrence, M., ner, G., Loustau, D., Manca, G., Matteucci, G., Miglietta, F., Krol, M., Lamarque, J. F., Montanaro, V., Muller, J. F., Peuch, V. Ourcival, J. M., Papale, D., Pilegaard, K., Rambal, S., Seufert, H., Pitari, G., Pyle, J., Rast, S., Rodriguez, J., Sanderson, M., G., Soussana, J. F., Sanz, M. J., Schulze, E. D., Vesala, T., Savage, N. H., Shindell, D., Strahan, S., Szopa, S., Sudo, K., Van and Valentini, R.: Europe-wide reduction in primary productivity Dingenen, R., Wild, O., and Zeng, G.: The global atmospheric caused by the heat and drought in 2003, Nature, 437, 529–533, environment for the next generation, Environ. Sci. Technol., 40, doi:10.1038/nature03972, 2005. 3586–3594, doi:10.1021/es0523845, 2006. Ciais, P., Canadell, J. G, Luyssaert, S., Chevallier, F., Shvidenko, Dolman, A. J., Shvidenko, A., Schepaschenko, D., Ciais, P., A., Pousii, Z., Jonas, M., Peylin, P., Wayne King, A., Schulze, Tchebakova, N., Chen, T., van der Molen, M. K., Belelli March- E.-D., Piao, S., Roedenbeck, C., Wouters, P., and Breon, F.-M.: esini, L., Maximov, T. C., Maksyutov, S., and Schulze, E.-D.: Can we reconcile atmospheric estimates of the Northern terres- An estimate of the terrestrial carbon budget of Russia using trial carbon sink with land-based accounting?, Current Opinion inventory-based, eddy covariance and inversion methods, Bio- in Environmental Sustainability, 2, 225–230, 2010. geosciences, 9, 5323–5340, doi:10.5194/bg-9-5323-2012, 2012. www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 674 S. Sitch et al.: Recent trends and drivers of regional sources Doney, S. C., Lima, I., Feely, R. A., Glover, D. M., Lindsay, K., Graven, H. D., Gruber, N., Key, R., Khatiwala, S., and Giraud, Mahowald, N., Moore, J. K., and Wanninkhof, R.: Mechanisms X.: Changing controls on oceanic radiocarbon: New insights on governing interannual variability in upper-ocean inorganic car- shallow-to-deep ocean exchange and anthropogenic CO uptake, bon system and air-sea CO ﬂuxes: Physical climate and atmo- J. Geophys. Res., 117, C10005, doi:10.1029/2012JC008074, spheric dust, Deep-Sea Res.-Pt. II, 56, 640–655, 2009a. 2012. Doney, S. C., Lima, I., Moore, J. K., Lindsay, K., Behrenfeld, M. J., Gruber, N.: Warming up, turning sour, losing breath: Ocean bio- Westberry, T. K., Mahowald, N., Glover, D. M., and Takahashi, geochemistry under global change, Philos. T. R. Soc. A, 369, T.: Skill metrics for confronting global upper ocean ecosystem- 1980–1996, doi:10.1098/rsta.2011.0003, 2011. biogeochemistry models against ﬁeld and remote sensing data, Gruber, N., Gloor, M., Fletcher, S. E. M., Doney, S. C., Dutkiewicz, J. Marine Syst., 76, 95–112, doi:10.1016/j.jmarsys.2008.05.015, S., Follows, M. J., Gerber, M., Jacobson, A. R., Joos, F., Lind- 2009b. say, K., Menemenlis, D., Mouchet, A., Muller, S. A., Sarmiento, Edwards E., S. McCaffery, S., and Evans, J.: Phosphorus status de- J. L., and Takahashi, T.: Oceanic sources, sinks, and transport termines biomass response to elevated CO in a legume: C-4 of atmospheric CO , Global Biogeochem. Cy., 23, GB1005, 2 2 grass community, Glob. Change Biol., 11, 1968–1981, 2005. doi:10.1029/2008GB003349, 2009. Falkowski, P. G., Ziemann, D., Kolber, Z., and Bienfang, P. K.: Role Gruber, N., Lachkar, Z., Frenzel, H., Marchesiello, P., Munnich, M., of eddy pumping in enhancing primary production in the ocean, McWilliams, J., Nagai, T., and Plattner, G.-K.: Eddy-induced re- Nature, 352, 55–58, 1991. duction of biological production in Eastern Boundary Upwelling Fan, S., Gloor, M., Mahlmam, J., Pacala, S., Sarmiento, J., Taka- Systems, Nat. Geosci., 4, 787–792, doi:10.1038/ngeo1273, 2011. hashi, T., and Tans, P.: A large terrestrial carbon sink in North Haverd, V., Raupach, M. R., Briggs, P. R., J. G. Canadell., Davis, America implied by atmospheric and oceanic carbon dioxide S. J., Law, R. M., Meyer, C. P., Peters, G. P., Pickett-Heaps, C., data and models, Science, 282, 442–446, 1998. and Sherman, B.: The Australian terrestrial carbon budget, Bio- Fay, A. R. and McKinley, G. A.: Global trends in surface ocean geosciences, 10, 851–869, doi:10.5194/bg-10-851-2013, 2013. pCO2 from in situ data, Global Biogeochem. Cy., 27, 541–557, Hicke, J. A.: NCEP and GISS solar radiation data sets available for doi:10.1002/gbc.20051, 2013. ecosystem modeling: Description, differences, and impacts on Friedlingstein, P., Houghton, R. A., Marland, G., Hackler, J., Bo- net primary production, Global Biogeochem. Cy., 19, GB2006, den,T. A., Conway, T. J., Canadell, J. G., Raupach, M. R., Ciais, doi:10.1029/2004GB002391, 2005. P., and Le Quéré, C.: Update on CO emissions, Nat. Geosci., 3, Houghton, R. A.: How well do we know the ﬂux of CO from 2 2 811–812, doi:10.1038/ngeo1022, 2010. land-use change? Tellus B, 62, 337–351, doi:10.1111/j.1600- Friend, A. D., Stevens, A. K., Knox, R. G., and Cannell, M. G. 0889.2010.00473.x, 2010. R.: A process-based, terrestrial biosphere model of ecosystem Ishidoya, S., Aoki, S., Goto, D., Nakazawa, T., Taguchi, S., dynamics (Hybrid v3.0), Ecol. Model., 95, 249–287, 1997. and Patra, P. K.: Time and Space variations of O /N ratio 2 2 Frölicher, T. L., Joos, F., Raible, C. C., and Sarmiento, J. L.: Atmo- in the troposphere over Japan and estimation of the global spheric CO response to volcanic eruptions: the role of ENSO, CO budget for the period 2000–2010, Tellus B, 64, 18964, 2 2 season, and variability, Global Biogeochem. Cy., 27, 239–251, doi:10.3402/tellusb.v64i0.18964, 2012. 2013. Janssens, I. A., Freibauer, A., Ciais, P., Smith, P., Nabuurs, G.-J., Gasser, T. and Ciais, P.: A theoretical framework for the net land- Folberth, G., Schlamadinger, B., Hutjes, R. W. A., Ceulemans, to-atmosphere CO ﬂux and its implications in the deﬁnition of R., Schulze, E.-D., Valentini, R., and Dolman, A. J.: Europe’s ter- “emissions from land-use change”, Earth Syst. Dynam., 4, 171– restrial biosphere absorbs 7 to 12 % of European anthropogenic 186, doi:10.5194/esd-4-171-2013, 2013. CO emissions, Science 300, 1538–1542, 2003. Gent, P. R. and Danabasoglu, G.: Response to Increasing Southern Jones, C. D. and Cox, P. M.: Modelling the volcanic signal in the Hemisphere Winds in CCSM4, J. Climate, 24, 4992–4998, 2011. atmospheric CO record, Global Biogeochem. Cy., 15, 453–465, Gloor, M., Gruber, N., Sarmiento, J. L., Sabine, C., Feely, D., doi:10.1029/2000GB001281, 2001. and Rödenbeck, C.: A ﬁrst estimate of present and preindus- Jönsson, A. M., Schröder, M., Lagergren, F., Anderbrandt, O., and trial air-sea CO ﬂux patterns based on ocean interior car- Smith, B.: Guess the impact of Ips typographus – an ecosystem bon measurements and models, Geophys. Res. Lett., 30, 1010, modelling approach for simulating bark beetle outbreaks, Agr. doi:10.1029/2002GL015594, 2003. Forest Meteorol., 166–167, 188–200, 2012. Gloor, M., Sarmiento, J. L., and Gruber, N.: What can be learned Jung, M., Vetter, M., Herold, M., Churkina, G., Reichstein, M., Za- about carbon cycle climate feedbacks from the CO airborne ehle, S., Cias, P., Viovy, N., Bondeau, A., Chen, Y., Trusilova, fraction?, Atmos. Chem. Phys., 10, 7739–7751, doi:10.5194/acp- K., Feser, F., and Heimann, M.: Uncertainties of modelling GPP 10-7739-2010, 2010. over Europe: A systematic study on the effects of using differ- Gloor, M., Gatti, L., Brienen, R., Feldpausch, T. R., Phillips, O. ent drivers and terrestrial biosphere models, Global Biogeochem. L., Miller, J., Ometto, J. P., Rocha, H., Baker, T., de Jong, B., Cy., 21, GB4021, doi:10.1029/2006GB002915, 2007. Houghton, R. A., Malhi, Y., Aragão, L. E. O. C., Guyot, J.-L., Jung, M., Reichstein, M., Margolis, H. A., Cescatti, A., Richardson Zhao, K., Jackson, R., Peylin, P., Sitch, S., Poulter, B., Lomas, A. D., Arain, M. A., Arneth, A., Bernhofer, C., Bonal, D., Chen M., Zaehle, S., Huntingford, C., Levy, P., and Lloyd, J.: The car- J., Gianelle, D., Gobron, N., Kiely, G., Kutsch, W., Lasslop, G., bon balance of South America: a review of the status, decadal Law, B. E., Lindroth, A., Merbold, L., Montagnani, L., Moors, trends and main determinants, Biogeosciences, 9, 5407–5430, E. J., Papale, D., Sottocornola, M., Vaccari, F., and Williams, doi:10.5194/bg-9-5407-2012, 2012. C.: Global patterns of land-atmosphere ﬂuxes of carbon diox- ide, latent heat, and sensible heat derived from eddy covariance, Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 675 satellite, and meteorological observations, J. Geophys. Res., 116, 4 of the Community Land Model, J. Adv. Model. Earth Syst., 3, G00J07, doi:10.1029/2010JG001566, 2011. M03001, doi:10.1029/2011ms000045, 2011. Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Lenton, A., Tilbrook, B., Law, R. M., Bakker, D., Doney, S. C., Gandin, L., Iredell, M., Saha, S., Woollen, J., Zhu, Y., Chelliah, Gruber, N., Ishii, M., Hoppema, M., Lovenduski, N. S., Matear, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K. C., Ro- R. J., McNeil, B. I., Metzl, N., Mikaloff Fletcher, S. E., Monteiro, pelewski, C., Wang, J., Leetmaa, A., Reynolds, R., Jenne, R., P. M. S., Rödenbeck, C., Sweeney, C., and Takahashi, T.: Sea- and Joseph, D.: The NCEP/NCAR 40-year reanalysis project, B. air CO ﬂuxes in the Southern Ocean for the period 1990–2009, Am. Meteorol. Soc., 77, 437–471, 1996. Biogeosciences, 10, 4037–4054, doi:10.5194/bg-10-4037-2013, Keeling, C. D., Bacastow, R. B., Bainbridge, A. E., Ekdahl Jr, C. 2013. A., Guenther, P. R., Waterman, L. S., and Chinm J. F. S.: At- Le Quéré, C. L., Harrison, S. P., Colin Prentice, I. C., Buitenhuis, mospheric carbon dioxide variations at Mauna Loa observatory, E. T., Aumont, O., Bopp, L., Claustre, H., Cotrim Da Cunha, L., Hawaii, Tellus, 28, 538–551, 1976. Geider, R., Giraud, X., Klaas, C., Kohfeld, K. E., Legendre, L., Keeling, C. D., Whorf, T. P., Wahlen, M., and van der Plicht, J.: Manizza, M., Platt, T., Rivkin, R. B., Sathyendranath, S., Uitz, J., Interannual extremes in the rate of rise of atmospheric carbon Watson, A. J., and Wolf-Gladrow, D.: Ecosystem dynamics based dioxide since 1980, Nature, 375, 666–670, 1995. on plankton functional types for global ocean biogeochemistry Keenan, T., García, R., Friend, A. D., Zaehle, S., Gracia, C., and models, Glob. Change Biol., 11, 2016–2040, doi:10.1111/j.1365- Sabate, S.: Improved understanding of drought controls on sea- 2486.2005.1004.x, 2005. sonal variation in Mediterranean forest canopy CO and wa- Le Quéré, C, Rödenbeck, C., Buitenhuis, E. T., Conway, T. J., ter ﬂuxes through combined in situ measurements and ecosys- Langenfelds, R., Gomez, A., Labuschagne, C., Ramonet, M., tem modelling, Biogeosciences, 6, 1423–1444, doi:10.5194/bg- Nakazawa, T., Metzl, N., and Gillett, N.: Saturation of the South- 6-1423-2009, 2009. ern ocean CO sink due to recent climate change, Science, 316, Keller, K. M., Joos, F., and Raible, C. C.: Time of emergence 1735–1738, doi:10.1126/science.1136188, 2007. of trends in ocean biogeochemistry, Biogeosciences, 11, 3647– Le Quéré C, Raupach M. R., Canadell, J. G., Marland, G., Bopp, 3659, doi:10.5194/bg-11-3647-2014, 2014. L., Ciais, P., Conway, T. J., Doney, S. C., Feely, R. A., Foster Key, R. M., Kozyr, A., Sabine, C. L., Lee, K., Wanninkhof, R., P, Friedlingstein, P., Gurney, K., Houghton, R. A., House, J. I., Bullister, J. L., Feely, R. A., Millero, F. J., Mordy, C., and Peng, Huntingford, C., Levy, P. E., Lomas, M. R., Majkut, J., Metzl, T.-H.: A global ocean carbon climatology: results from Global N., Ometto, J. P., Peters, G. P., Prentice, I. C., Randerson, J. T., Data Analysis Project (GLODAP), Global Biogeochem. Cy., 18, Running, S. W., Sarmiento, J. L., Schuster, U., Sitch, S., Taka- GB4031, doi:10.1029/2004GB002247, 2004. hashi, T., Viovy, N., van der Werf , G. R., and Woodward, F. I.: Khatiwala, S., Primeau, F., and Hall, T.: Reconstruction of the his- Trends in the sources and sinks of carbon dioxide, Nat. Geosci., tory of anthropogenic CO concentrations in the ocean, Nature, 2, 831–836, 2009. 462, 346–349, doi:10.1038/nature08526, 2009. Le Quéré, C., Takahashi, T., Buitenhuis, C. E., Rödenbeck, C., Khatiwala, S., Tanhua, T., Mikaloff Fletcher, S., Gerber, M., and Sutherland, S.: Impact of climate change and variability on Doney, S. C., Graven, H. D., Gruber, N., McKinley, G. A., the global oceanic sink of CO , Global Biogeochem. Cy., 24, Murata, A., Ríos, A. F., and Sabine, C. L.: Global ocean stor- GB4007, doi:10.1029/2009GB003599, 2010. age of anthropogenic carbon, Biogeosciences, 10, 2169–2191, Le Quéré, C., Andres, R. J., Boden, T., Conway, T., Houghton, R. doi:10.5194/bg-10-2169-2013, 2013. A., House, J. I., Marland, G., Peters, G. P., van der Werf, G. R., Kriest, I., Khatiwala, S., and Oschlies, A.: Towards an assessment of Ahlström, A., Andrew, R. M., Bopp, L., Canadell, J. G., Ciais, simple global marine biogeochemical models of different com- P., Doney, S. C., Enright, C., Friedlingstein, P., Huntingford, C., plexity, Progr. Oceanogr., 86, 337–360, 2010. Jain, A. K., Jourdain, C., Kato, E., Keeling, R. F., Klein Gold- Krinner, G., Viovy, N., de Noblet-Ducoudre, N., Ogee, J., Polcher, ewijk, K., Levis, S., Levy, P., Lomas, M., Poulter, B., Raupach, J., Friedlingstein, P., Ciais, P., Sitch, S., and Prentice, I. C.: M. R., Schwinger, J., Sitch, S., Stocker, B. D., Viovy, N., Zaehle, A dynamic global vegetation model for studies of the cou- S., and Zeng, N.: The global carbon budget 1959–2011, Earth pled atmosphere-biosphere system, Global Biogeochem. Cy., 19, Syst. Sci. Data, 5, 165–185, doi:10.5194/essd-5-165-2013, 2013. Gb1015, doi:10.1029/2003gb002199, 2005. Levy, P. E., Cannell, M. G. R., and Friend, A. D.: Modelling the Kurz, W. A., Dymond, C. C., Stinson, G., Rampley, G. J., Neilson, impact of future changes in climate, CO concentration and land E. T., Carroll, A. L., Ebata, T., and Safranyik, L.: Mountain pine use on natural ecosystems and the terrestrial carbon sink, Global beetle and forest carbon feedback to climate change, Nature, 452, Environ. Chang., 14, 21–30, 2004. 987–990, doi:10.1038/nature06777, 2008. Lewis, S. L., Lopez-Gonzalez, G., Sonké, B., Affum-Baffoe, K., Lagergren, F., Jönsson, A. M., Blennow, K., and Smith, B.: Imple- Baker, T. R., Ojo, L. O., Phillips, O. L., Reitsma, J. M., White, menting storm damage in a dynamic vegetation model for re- L., James, A., Comiskey, K., Djuikouo, M.-N., Ewango, C. E. N., gional applications in Sweden, Ecol. Model., 247, 71–82, 2012. Feldpausch, T. R., Hamilton, A. C., Gloor, M., Hart, T., Hladik, Large, W. G. and Yeager, S.: Diurnal to decadal global forcing for A., Lloyd, J., Lovett, J. C., Makana, J.-R., Malhi, Y., Mbago, ocean and sea-ice models: The data sets and ﬂux climatologies, F. M., Ndangalasi, H. J., Peacock, J., Peh, K. S.-H., Sheil, D., NCAR Tech. Rep. TN-460 _STR, 105 pp., 2004. Sunderland, T., Swaine, M. D., Taplin, J., Taylor, D., Thomas, Lawrence, D. M., Oleson, K. W., Flanner, M. G., Thornton, P. E., S. C., and Wöll, R. V. H.: Increasing carbon storage in intact Swenson, S. C., Lawrence, P. J., Zeng, X., Yang, Z.-L., Levis, S., African tropical forests, Nature, 457, 1003–1006, 2009a. Sakaguchi, K., Bonan, G. B., and Slater, A. G.: Parameterization improvements and functional and structural advances in version www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 676 S. Sitch et al.: Recent trends and drivers of regional sources Lewis, S. L., Lloyd, J., Sitch, S., Mitchard, E. T. A., and Lau- McGuire, A. D., Christensen, T. R., Hayes, D., Heroult, A., Eu- rance, W. F.: Changing Ecology of Tropical Forests: Evidence skirchen, E., Kimball, J. S., Koven, C., Laﬂeur, P., Miller, P. A., and Drivers, Annu. Rev. Ecol. Evol. Syst., 40, 529–49, 2009b. Oechel, W., Peylin, P., Williams, M., and Yi, Y.: An assessment Lovenduski, N. S., Gruber, N., Doney, S. C., and Lima, I. D.: En- of the carbon balance of Arctic tundra: comparisons among ob- hanced CO outgassing in the Southern Ocean from a positive servations, process models, and atmospheric inversions, Biogeo- phase of the Southern Annular Mode, Global Biogeochem. Cy. sciences, 9, 3185–3204, doi:10.5194/bg-9-3185-2012, 2012. 21, GB2026, doi:10.1029/2006GB002900, 2007. McKinley, G. A., Takahashi, T., Buitenhuis, E., Chai, F., Chris- Lovenduski, N. S., Gruber, N., and Doney, S. C.: Toward a mecha- tian, J. R., Doney, S. C., Jiang, M.-S., Le Quéré, C., Lima, I., nistic understanding of the decadal trends in the Southern Ocean Linday, K., Moore, J. K., Murtugudde, R., Shi, L., and Wetzel, carbon sink, Global Biogeochem. Cy., 22, 1–9, 2008. P.: North Paciﬁc carbon cycle response to climate variability on Lucht, W., Prentice, I. C., Myneni, R. B., Sitch, S., Friedlingstein, seasonal to decadal timescales. J. Geophys. Res., 111, C07S06, P., Cramer, W., Bousquet, P., Buermann, W., and Smith, B.: Cli- doi:10.1029/2005JC003173, 2006. mate control of the high-latitude vegetation greening trend and McKinley, G. A., Fay, A. R., Takahashi, T., and Metzl, N: Con- Pinatubo effect, Science, 296, 1687–1689, 2002. vergence of atmospheric and North Atlantic carbon dioxide Luo, Y. Q., Randerson, J. T., Abramowitz, G., Bacour, C., Blyth, trends on multidecadal timescales, Nat. Geosci, 4, 606–610, E., Carvalhais, N., Ciais, P., Dalmonech, D., Fisher, J. B., Fisher, doi:10.1038/NGEO1193, 2011. R., Friedlingstein, P., Hibbard, K., Hoffman, F., Huntzinger, D., McNeil, B. I. and Matear, R. J.: The non-steady state oceanic CO Jones, C. D., Koven, C., Lawrence, D., Li, D. J., Mahecha, M., signal: its importance, magnitude and a novel way to detect it, Niu, S. L., Norby, R., Piao, S. L., Qi, X., Peylin, P., Prentice, I. Biogeosciences, 10, 2219–2228, doi:10.5194/bg-10-2219-2013, C., Riley, W., Reichstein, M., Schwalm, C., Wang, Y. P., Xia, J. 2013. Y., Zaehle, S., and Zhou, X. H.: A framework for benchmarking Mercado, L. M., Bellouin, N., Sitch, S., Boucher, O., Huntingford, land models, Biogeosciences, 9, 3857–3874, doi:10.5194/bg-9- C., and Cox, P. M.: Impact of Changes in Diffuse Radiation on 3857-2012, 2012. the Global Land Carbon Sink, Nature, 458, 1014–1017, 2009. Luyssaert, S., Abril, G., Andres, R., Bastviken, D., Bellassen, V., Mikaloff Fletcher, S. E., Gruber, N., Jacobson, A. R., Doney, S. Bergamaschi, P., Bousquet, P., Chevallier, F., Ciais, P., Corazza, C., Dutkiewicz, S., Gerber, M., Follows, M., Joos, F., Lindsay, M., Dechow, R., Erb, K.-H., Etiope, G., Fortems-Cheiney, A., K., Menemenlis, D., Mouchet, A., Müller, S. A., and Sarmiento, Grassi, G., Hartmann, J., Jung, M., Lathière, J., Lohila, A., May- J. L.: Inverse estimates of anthropogenic CO uptake, transport, orga, E., Moosdorf, N., Njakou, D. S., Otto, J., Papale, D., Peters, and storage by the oceans, Global Biogeochem. Cy., 20, GB2002, W., Peylin, P., Raymond, P., Rödenbeck, C., Saarnio, S., Schulze, doi:10.1029/2005GB002530, 2006. E.-D., Szopa, S., Thompson, R., Verkerk, P. J., Vuichard, N., Morales, P., Sykes, M. T., Prentice, I. C., Smith, P., Smith, B., Wang, R., Wattenbach, M., and Zaehle, S.: The European land Bugmann, H., Zierl, B., Friedlingstein, P., Viovy, N., Sabaté, and inland water CO , CO, CH and N O balance between S., Sánchez, A., Pla, E., Gracia, C. A., Sitch, S., Arneth, A., 2 4 2 2001 and 2005, Biogeosciences, 9, 3357–3380, doi:10.5194/bg- and Ogee, J.: Comparing and evaluating process-based ecosys- 9-3357-2012, 2012. tem model predictions of carbon and water ﬂuxes in major Manning, A. C. and Keeling, R. F.: Global oceanic and land biota European forest biomes, Glob. Change Biol., 11, 2211–2233, sinks from the Scripps atmospheric oxygen ﬂask sampling net- doi:10.1111/j.1365-2486.2005.01036.x, 2005. work, Tellus B, 58B, 95–116, 2006. Myneni, R. B., Keeling, C. D., Tucker, C. J., Asrar, G., and Nemani, Matear, R. J., Chamberlain, M. A., Sun, C. and Feng, M.: Cli- R. R.: Increased plant growth in the northern high latitudes from mate change projection of the Tasman Sea from an Eddy- 1981 to 1991, Nature, 386, 698–702, doi:10.1038/386698a0, resolving Ocean Model, J. Geophys. Res.-Oceans, 118, 2961– 1997. 2976, doi:10.1002/jgrc.20202, 2013. Murray-Tortarolo, G., Anav, A., Friedlingstein, P., Sitch, S., Piao, Matsumoto, K. and Gruber, N.: How accurate is the estima- S., Zhu, Z., Poulter, B., Zaehle, S., Ahlstrom, A., Lomas, M., tion of anthropogenic carbon in the ocean?, An evaluation Levis, S., Viovy, N., and Zeng, N.: Evaluation of DGVMs of the DC* method, Global Biogeochem. Cy., 19, GB3014, in reproducing satellite derived LAI over the Northern Hemi- doi:10.1029/2004GB002397, 2005. sphere, Part I: Uncoupled DGVMs, Remote Sens., 5, 4819–4838; McDonald, K. C., Kimball, J. S., Njoke, E., Zimmermann, R., and doi:10.3390/rs5104819, 2013. Zhao, M.: Variability in springtime thaw in the terrestrial high Müller, S. A., Joos, F., Plattner, G.-K., Edwards, N. R., and Stocker, latitudes: monitoring a major control on the biospheric assimi- T. F.: Modeled natural and excess radiocarbon: Sensitivities lation of atmospheric CO with spaceborne microwave remote to the gas exchange formulation and ocean transport strength, sensing, Earth Interact., 8, 1–23, 2004. Global Biogeochem. Cy., 22, doi:10.1029/2007GB003065, McDowell, N. G, Pockman, W. T., Allen, C. D., Breshears, D. D., 2008. Cobb, N., Kolb, T., Plaut, J., Sperry, J., West, A., Williams, D. Nemani, R. R., Keeling, C. D., Hashimoto, H., Jolly, W. M., Piper, G., and Yepez, E. A.: Mechanisms of plant survival and mor- S. C., Tucker, C. J., Myneni, R. B., and Running, S. W.: Climate- tality during drought: why do some plants survive while others driven increases in global terrestrial net primary production from succumb to drought?. New Phytol., 178, 719–739, 2008. 1982 to 1999, Science, 300, 1560–1563, 2003. McGillicuddy Jr, D. J., Robinson, A. R., Siegel, D. A., Jannasch, H. New, M. G., Hulme, M., and Jones, P. D.: Representing twentieth- W., Johnson, R., Dickey, T. D., McNeil, J., Michaels, A. F., and century space-climate variability, Part II, Development of 1901– Knap, A. H.: Inﬂuence of mesoscale eddies on new production 1996 monthly grids of terrestrial surface climate, J. Climate, 13, in the Sargasso Sea, Nature, 394, 263–266, 1998. 2217–2238, 2000. Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 677 Norby, R. J., DeLucia, E. H., Gielen, B., Calfapietra, C., Giardina, Pfeil, B., Olsen, A., Bakker, D. C. E., Hankin, S., Koyuk, H., Kozyr, C. P., King, J. S., Ledford, J., McCarthy, H. R., Moore, D. J. P., A., Malczyk, J., Manke, A., Metzl, N., Sabine, C. L., Akl, J., Ceulemans, R., De Angelis, P., Finzi, A. C., Karnosky, D. F., Ku- Alin, S. R., Bates, N., Bellerby, R. G. J., Borges, A., Boutin, biske, M. E., Lukac, M., Pregitzer, K. S., Scarascia-Mugnozza, J., Brown, P. J., Cai, W.-J., Chavez, F. P., Chen, A., Cosca, C., G. E., Schlesinger, W. H., and Oren, R.: Forest response to ele- Fassbender, A. J., Feely, R. A., González-Dávila, M., Goyet, vated CO2 is conserved across a broad range of productivity, P. C., Hales, B., Hardman-Mountford, N., Heinze, C., Hood, M., Natl. Acad. Sci. USA, 102, 18052–18056, 2005. Hoppema, M., Hunt, C. W., Hydes, D., Ishii, M., Johannessen, Norby, R. J., Warren, J. M., Iversen, C. M., Medlyn, B. E., and T., Jones, S. D., Key, R. M., Körtzinger, A., Landschützer, P., McMurtrie, R. E.: CO enhancement of forest productivity con- Lauvset, S. K., Lefèvre, N., Lenton, A., Lourantou, A., Merlivat, strained by limited nitrogen availability, P. Natl. Acad. Sci., 107, L., Midorikawa, T., Mintrop, L., Miyazaki, C., Murata, A., Naka- 19368–19373, 2010. date, A., Nakano, Y., Nakaoka, S., Nojiri, Y., Omar, A. M., Padin, Oleson, K. W., Lawrence, D. M., Bonan, G. B., Flanner, M. G., X. A., Park, G.-H., Paterson, K., Perez, F. F., Pierrot, D., Poisson, Kluzek, E., Lawrence, P. J., Levis, S., Swenson, S. C., and Thorn- A., Ríos, A. F., Santana-Casiano, J. M., Salisbury, J., Sarma, V. V. ton, P. E.: Technical description of version 4.0 of the Community S. S., Schlitzer, R., Schneider, B., Schuster, U., Sieger, R., Skjel- Land Model (CLM), NCAR Tech. Note NCAR/TN-478CSTR, van, I., Steinhoff, T., Suzuki, T., Takahashi, T., Tedesco, K., Tel- 257 pp., 2010. szewski, M., Thomas, H., Tilbrook, B., Tjiputra, J., Vandemark, Pacala, S. W., Hurtt, G. C., Baker, D., Peylin, P., Houghton, R. A., D., Veness, T., Wanninkhof, R., Watson, A. J., Weiss, R., Wong, Birdsey, R. A., Heath, L., Sundquist, E. T., Stallard, R. F., Ciais, C. S., and Yoshikawa-Inoue, H.: A uniform, quality controlled P., Moorcroft, P., Caspersen, J. P., Shevliakova, E., Moore B., Surface Ocean CO Atlas (SOCAT), Earth Syst. Sci. Data, 5, Kohlmaier, G., Holland, E., Gloor, M., Harmon, M. E., Fan, S.- 125–143, doi:10.5194/essd-5-125-2013, 2013. M., Sarmiento, J. L., Goodale, C. L., Schimel, D., and Field, C. Phillips, O. L., Malhi, Y., Higuchi, N., Laurance, W. F., Nunez, P. B.: Consistent land- and atmosphere-based US carbon sink esti- V., Vasquez, R. M., Laurance, S. G., Ferreira, L. V., Stern, M., mates, Science, 292, 2316–2320, 2001. Brown, S., and Grace, J.: Changes in the carbon balance of trop- Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P. E., Kurz, ical forests: Evidence from long-term plots, Science, 282, 439– W. A., Phillips, O. L., Shvidenko, A., Lewis, S. L., Canadell, J. 442, 1998. G., Ciais, P., Jackson, R. B., Pacala, S., McGuire, A. D., Piao, S., Piao, S. L., Fang, J. Y., Ciais, P., Peylin, P., Huang, Y., Sitch, S., and Rautiainen, A., Sitch, S., Hayes, D., and Watson, C.: A large and Wang, T: The carbon balance of terrestrial ecosystems in China, persistent carbon sink in the World’s forests, 1990–2007, Sci- Nature, 458, 1009–1014, 2009. ence, doi:10.1126/science.1201609, 2011. Piao, S. L., Sitch, S., Ciais, P., Friedlingstein, P., Peylin, P., Wang, Park, G.-H., Wanninkhof, R., Doney, S. C., Takahashi, T., Lee, K., X. H., Ahlström, A., Anav, A., Canadell, J. G., Huntingford, C., Feely, R. A., Sabine, C. L., Triñanes, J., and Lima, I. D.: Variabil- Jung, M., Levis, S., Levy, P. E., Li, J. S., Lin, X., Lomas, M. R., ity of global net sea-air CO ﬂuxes over the last three decades Lu, M ., Luo, Y. Q., Ma, Y. C., Myneni, R. B., Poulter, B., Sun, Z. using empirical relationships, Tellus B, 62, 352–368, 2010. Z., Wang, T., Viovy, N., Zaehle, S., and Zeng, N.: Evaluation of Patra, P. K., Canadell, J. G., Houghton, R. A., Piao, S. L., Oh, N.-H., terrestrial carbon cycle models for their response to climate vari- Ciais, P., Manjunath, K. R., Chhabra, A., Wang, T., Bhattacharya, ability and to CO trends, Glob. Change Biol., 19, 2117–2132, T., Bousquet, P., Hartman, J., Ito, A., Mayorga, E., Niwa, Y., Ray- 2013. mond, P. A., Sarma, V. V. S. S., and Lasco, R.: The carbon budget Pongratz, J., Reick, C. H., Houghton, R. A., and House, J. I.: Ter- of South Asia, Biogeosciences, 10, 513–527, doi:10.5194/bg-10- minology as a key uncertainty in net land use and land cover 513-2013, 2013. change carbon ﬂux estimates, Earth Syst. Dynam., 5, 177–195, Peacock, S.: Debate over the ocean bomb radiocarbon sink: doi:10.5194/esd-5-177-2014, 2014. Closing the gap, Global Biogeochem. Cy., 18, GB2022, Poulter, B., Frank, D. C., Hodson, E. L., and Zimmermann, N. doi:10.1029/2003GB002211, 2004. E.: Impacts of land cover and climate data selection on un- Peng S., Ciais, P., Chevallier, F., Peylin, P., Cadule, P., Sitch, derstanding terrestrial carbon dynamics and the CO airborne S., Piao, S., Ahlström, A., Levy, P., Lomas, M., Poulter, B., fraction, Biogeosciences, 8, 2027–2036, doi:10.5194/bg-8-2027- Viovy, N., Wang, T., Zaehle, S., Zeng, N., Zhao, F. and Zhao, 2011, 2011. H.: Benchmarking the seasonal cycle of CO ﬂuxes simulated Poulter, B., Pederson, N., Liu, H., Zhu, Z., D’Arrigo, R., Ciais, P., by terrestrial ecosystem models, Global Biogeochem. Cy., 29, Davi, N., Frank, D., Leland, C., Myneni, R., Piao, S., and Wang, doi:10.1002/2014GB004931, 2014. T.: Recent trends in Inner Asian forest dynamics to temperature Peylin, P., Bousquet, P., Le Quéré, C., Sitch, S., Friedling- and precipitation indicate high sensitivity to climate change, Agr. stein, P., McKinley, G., Gruber, N., Rayner, P., and Ciais, Forest Meteorol., 178–179, 31–45, 2013. P.: Multiple constraints on regional CO ﬂux variations over Poulter, B., Frank, D., Ciais, P., Myneni, R., Andela, N., Bi, J., Bro- land and oceans, Global Biogeochem. Cy., 19, GB1011, quet, G., Canadell, J. G., Chevallier, F., Liu, Y. Y., Running, S. doi:10.1029/2003GB002214, 2005. W., Sitch, S. and van der Werf, G. R.: Contribution of semi-arid Peylin, P., Law, R. M., Gurney, K. R., Chevallier, F., Jacobson, ecosystems to interannual variability of the global carbon cycle, A. R., Maki, T., Niwa, Y., Patra, P. K., Peters, W., Rayner, P. Nature, doi:10.1038/nature13376, 2014. J., Rödenbeck, C., van der Laan-Luijkx, I. T., and Zhang, X.: Prentice I. C., Farquhar, G. D., Fasham, M. J. R., Goulden, M. L., Global atmospheric carbon budget: results from an ensemble of Heimann, M., Jaramillo, V.J., Kheshgi, H. S., Le Quere, C., Sc- atmospheric CO inversions, Biogeosciences, 10, 6699–6720, holes, R. J., Wallace, D. W. R., Contributing Authors Archer, doi:10.5194/bg-10-6699-2013, 2013. D., Ashmore, M. R., Aumont, O., Baker, D., Battle, M., Bender, www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015 678 S. Sitch et al.: Recent trends and drivers of regional sources M., Bopp, L. P., Bousquet, P., Caldeira, K., Ciais, P., Cox, P. M., Sarmiento, J. L., Gloor, M., Gruber, N., Beaulieu, C., Jacobson, A. Cramer, W., Dentener, F., Enting, I. G., Field, C. B., Friedling- R., Mikaloff Fletcher, S. E., Pacala, S., and Rodgers, K.: Trends stein, P., Holland, E. A., Houghton, R. A., House, J. I., Ishida, A., and regional distributions of land and ocean carbon sinks, Bio- Jain, A. K., Janssens, I. A., Joos, F., Kaminski, T., Keeling, C. geosciences, 7, 2351–2367, doi:10.5194/bg-7-2351-2010, 2010. D., Keeling, R. F., Kicklighter, D. W., Kohfeld, K. E., Knorr, W., Schuster, U., McKinley, G. A., Bates, N., Chevallier, F., Doney, Law, R., Lenton, T. , Lindsay, K., Maier-Reimer, E., Manning, S. C., Fay, A. R., González-Dávila, M., Gruber, N., Jones, S., A. C., Matear, R. J., McGuire, A. D., Melillo, J. M., Meyer, R., Krijnen, J., Landschützer, P., Lefèvre, N., Manizza, M., Mathis, Mund, M., Orr, J. C., Piper, S., Plattner, K., Rayner, P. J., Sitch, J., Metzl, N., Olsen, A., Rios, A. F., Rödenbeck, C., Santana- S., Slater, R., Taguchi, S. , Tans, P. P., Tian, H. Q., Weirig, M. Casiano, J. M., Takahashi, T., Wanninkhof, R., and Watson, A. F., Whorf, T., and Yool, A.: The carbon cycle and atmospheric J.: An assessment of the Atlantic and Arctic sea-air CO ﬂuxes, carbon dioxide, in. Climate Change 2001: The Scientiﬁc Basis 1990–2009, Biogeosciences, 10, 607–627, doi:10.5194/bg-10- edited by: Houghton, J. T., Cambridge Univ. Press, New York, 607-2013, 2013. 183–237, 2001. Sitch, S., Smith, B., Prentice, I. C., Arneth, A., Bondeau, A., Prentice, I. C., Bondeau, A., Cramer, W., Harrison, S. P., Hickler, Cramer, W., Kaplan, J. O., Levis, S., Lucht, W., Sykes, M. T., T., Lucht, W., Sitch, S., Smith, B., and Sykes, M. T.: Terrestrial Thonicke, K., and Venevsky, S.: Evaluation of ecosystem dynam- Ecosystems in a changing world, IGBP Book Series, edited by: ics, plant geography and terrestrial carbon cycling in the LPJ dy- Canadell. J., Pitelka, L., and Pataki, D., Springer, Heidelberg, namic global vegetation model, Glob. Change Biol., 9, 161–185, Germany, pp. 336, 2007. doi:10.1046/j.1365-2486.2003.00569.x, 2003. Quaife, T., Quegan, S., Disney, M., Lewis, P., Lomas, M. R., and Sitch, S., McGuire, A. D., Kimball, J., Gedney, N., Gamon, J., En- Woodward, F. I.: Impact of land cover uncertainties on esti- gstrom, R., Wolf, A., Zhuang, Q., Clein, J. S., and McDonald, mates of biospheric carbon ﬂuxes, Global Biogeochem. Cy., 22, K. C.: Assessing the carbon balance of circumpolar arctic tun- GB4016, doi:10.1029/2007GB003097, 2008. dra using remote sensing and process modeling, Ecol. Appl., 17, Randerson, J. T., Hoffman, F. M., Thornton, P. E., Mahowald, N. 213–234, 2007a. M., Lindsay, K., Lee, Y.-H., Nevison, C. D., Doney, S. C., Bo- Sitch, S., Cox, P. M., Collins, W. J., and Huntingford., C.: nan, G., Stoeckli, R., Covey, C., Running, S. W., and Fung, I. Indirect radiative forcing of climate change through ozone Y.: Systematic assessment of terrestrial biogeochemistry in cou- effects on the land-carbon sink, Nature, 448, 791–794, pled climate-carbon models, Glob. Change Biol., 15, 2462–2484, doi:10.1038/nature06059, 2007b. 2009. Sitch, S., Huntingford, C., Gedney, N., Levy, P. E., Lomas, M., Piao, Raupach, M. R., Canadell, J. G., and Le Quéré, C.: Anthropogenic S., Betts, R., Ciais, P., Cox, P., Friedlingstein, P., Jones, C. D., and biophysical contributions to increasing atmospheric CO2 Prentice, I. C., and Woodward, F. I.: Evaluation of the terres- growth rate and airborne fraction, Biogeosciences, 5, 1601–1613, trial carbon cycle, future plant geography, and climate-carbon doi:10.5194/bg-5-1601-2008, 2008. cycle feedbacks using 5 Dynamic Global Vegetation Models Raupach, M. R., Gloor, M., Sarmiento, J. L., Canadell, J. G., (DGVMs), Glob. Change Biol., 14 1–25, doi:10.1111/j.1365- Frölicher, T. L., Gasser, T., Houghton, R. A., Le Quéré, C., 2486.2008.01626.x, 2008. and Trudinger, C. M.: The declining uptake rate of atmospheric Smith, B., Prentice, I. C., and Sykes, M. T.: Representation of CO by land and ocean sinks, Biogeosciences, 11, 3453–3475, vegetation dynamics in the modelling of terrestrial ecosystems: doi:10.5194/bg-11-3453-2014, 2014. comparing two contrasting approaches within European climate Reichstein, M., Bahn, M., Ciais, P., Frank, D., Mahecha, M. D., space, Global Ecol. Biogeogr., 10, 621–637, 2001. Seneviratne, S. I., Zscheischler, J., Beer, C., Buchmann, N., Sokolov, A. P., Kicklighter, D. W., Melillo, J. M., Felzer, B. S., Frank, D. C., Papale, D., Rammig, A., Smith, P., Thonicke, K., Schlosser, C. A., and Cronin, T. W.: Consequences of consid- van der Velde, M., Vicca, S., Walz, A., and Wattenbach, M.: Cli- ering carbon-nitrogen interactions on the feedbacks between cli- mate extremes and the carbon cycle, Nature, 500, 287–295, 2013. mate and the terrestrial carbon cycle, J. Climate, 21, 3776–3796, Roy, T., Bopp, L., Gehlen, M., Schneider, B., Cadule, P., Frölicher, doi:10.1175/2008JCLI2038.1, 2008. T. L., Segschneider, J., Tjiputra, J., Heinze, C., and Joos, F.: Stocker, B. D., Feissli, F., Strassmann, K. M., Spahni, R., and Regional Impacts of Climate Change and Atmospheric CO on Joos, F.: Past and future carbon ﬂuxes from land use change, Future Ocean Carbon Uptake: A Multimodel Linear Feedback shifting cultivation and wood harvest, Tellus B, 66, 23188, Analysis, J. Climate, 24, 2300–2318, 2011. doi:10.3402/tellusb.v66.23188, 2014. Sabine, C. S., Feely, R. A.., Gruber, N. , Key, R. M., Lee, K., Bullis- Stephens, B. B., Gurney, K. R., Tans, P. P., Sweeney, C., Peters, W., ter, J. L., Wanninkhof, R., Wong, C. S., Wallace, D. W. R., Till- Bruhwiler, L., Ciais, P., Ramonet, M., Bousquet, P., Nakazawa, brook, B., Millero, F. J., Peng, T.-H., Kozyr, A., Ono, T., and T., Aoki, S., Machida, T., Inoue, G., Vinnichenko, N., Lloyd, Rios, A. F.: The oceanic sink for anthropogenic CO , Science, J., Jordan, A., Heimann, M., Shibistova, O., Langenfelds, R. L., 305, 367–371, 2004. Steele, L. P., Francey, R. J., and Denning, A. S.: Weak northern Sarmiento, J. L. and Gruber, N.: Sinks for anthropogenic carbon, and strong tropical land carbon uptake from vertical proﬁles of Phys. Today, 55, 30–36, 2002. atmospheric CO , Science, 316, 1732–1735, 2007. Sarmiento, J. L. and Gruber, N.: Ocean Biogeochemical Dynamics, Sweeney, C., Gloor, E., Jacobson, A. R., Key, R. M., McKin- Princeton Univ. Press, 526 pp., 2006. ley, G., Sarmiento, J. L., and Wanninkhof, R.: Constrain- Sarmiento, J. L., Orr, J. C., and Siegenthaler, U.: A perturbation ing global air-sea gas exchange for CO with recent bomb simulation of CO uptake in an ocean general circulation model, C measurements, Global Biogeochem. Cycl., 21, GB2015, J. Geophys. Res., 97, 3621–3645, 1992. doi:10.1029/2006GB002784, 2007. Biogeosciences, 12, 653–679, 2015 www.biogeosciences.net/12/653/2015/ S. Sitch et al.: Recent trends and drivers of regional sources 679 Takahashi, T., Sutherland, S. C., Wanninkhof, R., Sweeney, C., Wanninkhof, R., Park, G. -H., Takahashi, T., Sweeney, C., Feely, R., Feely, R. A., Chipman, D., Hales, B., Friederich, G., Chavez, Nojiri, Y., Gruber, N., Doney, S. C., McKinley, G. A., Lenton, F., Sabine, C., Watson, A., Bakker, D. C. E., Schuster, U., Metzl, A., Le Quéré, C., Heinze, C., Schwinger, J., Graven, H., and N., Yoshikawa-Inoue, H., Ishii, M., Midorikawa, T., Nojiri, Y., Khatiwala, S.: Global ocean carbon uptake: magnitude, variabil- Kortzinger, A., Steinhoff, T., Hoppema, M., Olafsson, J., Arnar- ity and trends, Biogeosciences, 10, 1983–2000, doi:10.5194/bg- son, T. S., Tilbrook, B., Johannessen, T., Olsen, A., Bellerby, R., 10-1983-2013, 2013. Wong, C. S., Delille, B., Bates, N. R., and de Baar, H. J. W.: Cli- Woodward, F. I. and Lomas, M. R.: Vegetation-dynamics- matological mean and decadal changes in surface ocean pCO , simulating responses to climate change, Biological Reviews, 79, and net sea-air CO2 ﬂux over the global oceans, SOCOVV Sym- 643–670, 2004. posium Volume, Deep Sea Res.-Pt. II, 810, 554–577, 2009. Woodward, F. I., Smith, T. M., Emanuel, W. R.: A global land Tans, P. P., Fung, I. Y., and Takahashi, T.: Observational constraints primary productivity and phytogeography model. Global Bio- on the global atmospheric CO budget, Science, 247, 1431– geochem. Cy., 9, 471–490, 1995. 1439, doi:10.1126/science.247.4949.1431, 1990. Zaehle, S. and Friend, A. D.: Carbon and nitrogen cycle dynamics Thornton, P. E., Lamarque, J. F., Rosenbloom, N. A., and in the O-CN land surface model, I: Model description, site- scale Mahowald, N. M.: Inﬂuence of carbon-nitrogen cycle cou- evaluation and sensitivity to parameter estimates, Global Bio- pling on land model response to CO fertilization and geochem. Cy., 24, GB1005, doi:10.1029/2009GB003521, 2010. climate variability, Global Biogeochem. Cy., 21, GB4018, Zaehle, S. and Dalmonech, D.: Carbon – nitrogen interactions on doi:10.1029/2006GB002868, 2007. land at global scales: current understanding in modeling climate Thornton, P. E., Doney, S. C., Lindsay, K., Moore, J. K., Ma- biosphere feedbacks, Current Opinion in Environmental Sustain- howald, N., Randerson, J. T., Fung, I., Lamarque, J.-F., Fed- ability, 3, 311–320, 2011. dema, J. J., and Lee, Y.-H.: Carbon-nitrogen interactions regu- Zaehle, S., Friedlingstein, P., and Friend, A. D.: Terrestrial nitrogen late climate-carbon cycle feedbacks: results from an atmosphere- feedbacks may acccelerate future climate change, Geophys. Res. ocean general circulation model, Biogeosciences, 6, 2099–2120, Lett., 37, L01401, doi:10.1029/2009GL041345, 2010. doi:10.5194/bg-6-2099-2009, 2009. Zaehle, S., Ciais, P., Friend, A. D., and Prieur, V.: Carbon beneﬁts Tjiputra, J. F., Assmann, K., and Heinze, C.: Anthropogenic car- of anthropogenic reactive nitrogen offset by nitrous oxide emis- bon dynamics in the changing ocean, Ocean Science, 6, 605–614, sions, Nat. Geosci., 4, 601–605, doi:10.1038/ngeo1207, 2011. 2010. Zaehle, S., Medlyn, B. E., De Kauwe, M. G., Walker, A. P., Dietze, Tucker, C. J., Slayback, D. A., Pinzon, J. E., Los, S. O., Myneni, M. C., Hickler, T., Luo, Y., Wany, Y.-P., El-Masri, B., Thornton, R. B., and Taylor, M. G.: Higher northern latitude normalized P., Jain, A., Wang, S., Warlind, D., Weng, E., Parton, W., Iversen, difference vegetation index and growing season trends from 1982 C. M., Gallet-Budynek, A., McCarthy, H., Finzi, A., Hanson, to 1999, Int. J. Biometeorol., 45, 184–190, 2001. P. J., Prentice, I. C., Oren, R., and Norby, R. J.: Evaluation of Valentini, R., Arneth, A., Bombelli, A., Castaldi, S., Cazzolla Gatti, eleven terrestrial carbon-nitrogen cycle models against obser- R., Chevallier, F., Ciais, P., Grieco, E., Hartmann, J., Henry, M., vations from two temperate Free-Air CO Enrichment Studies, Houghton, R. A., Jung, M., Kutsch, W. L., Malhi, Y., Mayorga, New Phytologist, 202, 803–822, doi:10.1111/nph.12697, 2014. E., Merbold, L., Murray-Tortarolo, G., Papale, D., Peylin, P., Zeng, N.: Glacial-interglacial atmospheric CO change – The Poulter, B., Raymond, P. A., Santini, M., Sitch, S., Vaglio Laurin, glacial burial hypothesis, Adv. Atmos. Sci., 20, 677–673, 2003. G., van der Werf, G. R., Williams, C. A., and Scholes, R. J.: A full Zeng, N., Mariotti, A., and Wetzel, P.: Terrestrial mechanisms of in- greenhouse gases budget of Africa: synthesis, uncertainties, and terannual CO variability, Global Biogeochem. Cy., 19, GB1016, vulnerabilities, Biogeosciences, 11, 381–407, doi:10.5194/bg- doi:10.1029/2004gb002273, 2005. 11-381-2014, 2014. Zhang, Q., Wang, Y. P., Matear, R. J., Pitman, A. J., and Dai, Y. Van der Molen, M. K., Dolman, A. J., Ciais, P., Eglin, T., Gob- J.: Nitrogen and phosphorous limitations signiﬁcantly reduce fu- ron, N., Law, B. E., Meir, P., Peters, W., Phillips, O. L., Reich- ture allowable CO emissions, Geophys. Res. Lett, 41, 632–637, stein, M., Chen, T., Dekker, S. C., Doubková, M., Friedl, M. A., doi:10.1002/2013GL058352, 2014. Jung, M., van den Hurk, B. J. J. M., de Jeu, R. A. M., Kruijt, B., Zhu, Z., Bi, J., Pan, Y., Ganguly, S., Anav, A., Xu, L., Samanta, A., Ohta, T., Rebel, K. T., Plummer, S., Seneviratne, S. I., Sitch, S., Piao, S., Nemani, R. R., and Myneni, R. B.: Global Data Sets of Teuling, A. J., van der Werf, G. R., and Wang, G.: Drought and Vegetation Leaf Area Index (LAI)3g and Fraction of Photosyn- ecosystem carbon cycling, Agr. Forest Meteorol., 151, 765–773, thetically Active Radiation (FPAR)3g Derived from Global In- doi:10.1016/j.agrformet.2011.01.018, 2011. ventory Modeling and Mapping Studies (GIMMS) Normalized Vitousek, P. M. and Howarth, R. W.: Nitrogen limitation on land Difference Vegetation Index (NDVI3g) for the Period 1981 to and in the sea: How can it occur?, Biogeochemistry, 13, 87–115, 2011, Remote Sens., 5, 927–948, 2013. 1991. Zscheischler, J., Mahecha, M. D., von Buttlar, J., Harmeling, Wang, Y. P., Law, R. M., and Pak, B.: A global model of carbon, S., Jung, M., Rammig, A., Randerson, J. T., Schoelkopf, B., nitrogen and phosphorus cycles for the terrestrial biosphere, Bio- Seneviratne, S. I., Tomelleri, E., Zaehle, S., and Reichstein, M.: geosciences, 7, 2261–2282, doi:10.5194/bg-7-2261-2010, 2010. A few extreme events dominate global interannual variability Wanninkhof, R.: Relationship between wind speed and gas ex- in gross primary production, Environ. Res. Lett., 9, 035001, change over the ocean, J. Geophys. Res., 97, 7373–7382, 1992. doi:10.1088/1748-9326/9/3/035001, 2014. www.biogeosciences.net/12/653/2015/ Biogeosciences, 12, 653–679, 2015

http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Biogeosciences Discussions Unpaywall

http://www.deepdyve.com/lp/unpaywall/trends-and-drivers-of-regional-sources-and-sinks-of-carbon-dioxide-J4YyzDCcA7

Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades

Loading next page...

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher: Unpaywall
ISSN: 1810-6285
DOI: 10.5194/bgd-10-20113-2013
Publisher site: See Article on Publisher Site

Abstract

Journal

Biogeosciences Discussions – Unpaywall

Published: Dec 23, 2013

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades

Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades

Trends and drivers of regional sources and sinks of carbon dioxide over the past two decades

References

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies