Environmental Research Letters

LETTER • OPEN ACCESS

Freshwater requirements of large-scale bioenergy plantations for limitingglobal warming to 1.5 °CTo cite this article: Fabian Stenzel et al 2019 Environ. Res. Lett. 14 084001

 

View the article online for updates and enhancements.

This content was downloaded from IP address 128.183.2.70 on 22/07/2019 at 14:29

https://ntrs.nasa.gov/search.jsp?R=20190029171 2020-06-18T21:06:30+00:00Z

Environ. Res. Lett. 14 (2019) 084001 https://doi.org/10.1088/1748-9326/ab2b4b

LETTER

Freshwater requirements of large-scale bioenergy plantationsfor limiting global warming to 1.5°C

Fabian Stenzel1,2,3 , DieterGerten1,2,3 , ConstanzeWerner1,2,3 and Jonas Jägermeyr1,4,5

1 Potsdam Institute for Climate Impact Research, D-14473 Potsdam,Germany2 Department ofGeography, Humboldt-Universität zu Berlin, D-12489 Berlin, Germany3 Integrative Research Institute onTransformations ofHuman-Environment Systems, D-10117Berlin, Germany4 Department of Computer Science, University of Chicago, Chicago, IL 60637,United States of America5 NASAGoddard Institute for Space Studies, NewYork,NY 10025,United States of America

E-mail: stenzel@pik-potsdam.de

Keywords:BECCS, water demand, irrigation, negative emissions, environmental flow requirements, climate change, bioenergy plantations

Supplementarymaterial for this article is available online

AbstractLimitingmean global warming towell below 2 °Cwill probably require substantial negative emissions(NEs)within the 21st century. To achieve these, bioenergy plantationswith subsequent carboncapture and storage (BECCS)may have to be implemented at a large scale. Irrigation of theseplantationsmight be necessary to increase the yield, which is likely to put further pressure on alreadystressed freshwater systems. Conversely, the potential of bioenergy plantations (BPs) dedicated toachievingNEs throughCO2 assimilationmay be limited in regions with low freshwater availability.This paper provides a first-order quantification of the biophysical potentials of BECCS as a negativeemission technology contribution to reaching the 1.5 °Cwarming target, as constrained by associatedwater availabilities and requirements. Using a global biospheremodel, we analyze the availability offreshwater for irrigation of BPs designed tomeet the projectedNEs to fulfill the 1.5 °C target, spatiallyexplicitly on areas not reserved for ecosystem conservation or agriculture.We take account of thesimultaneouswater demands for agriculture, industries, and households and also account forenvironmental flow requirements (EFRs) needed to safeguard aquatic ecosystems. Furthermore, weassess towhat extent different forms of improvedwatermanagement on the suggested BPs and oncroplandmay help to reduce the freshwater abstractions. Results indicate that global waterwithdrawals for irrigation of BPs range between∼400 and∼3000 km3 yr−1, depending on the scenarioand the conversion efficiency of the carbon capture and storage process. Consideration of EFRsreduces theNEpotential significantly, but can partly be compensated for by improved on-fieldwatermanagement.

Introduction

With the Paris Agreement (UNFCCC 2015), theinternational community has agreed to aim for aglobal mean temperature (GMT) (see table 3 for a fulllist of abbreviations) increase of well below 2 °Ccompared to preindustrial levels, and pursue efforts tolimit it to 1.5 °C. Since the remaining carbon emissionsbudget for such ambitious climate goals is very small(Fuss et al 2014), the use of negative emissiontechnologies (NETs) seems almost inevitable(Rockström et al 2017, Minx et al 2018, Rogelj et al

2018). The necessity for NET deployment might evenincrease, should efforts of decarbonization be lesspronounced or come into action later than envisionedtoday.

The NET most widely used in projections for the21st century is bioenergy plantations (BPs) with sub-sequent carbon capture and storage (BECCS) (Fusset al 2014, Schleussner et al 2016). BECCS utilizes fastgrowing plant species to convert atmospheric CO2 tobiomass, which is regularly harvested and burned forenergy generation or fermented to produce bio-fuels.The CO2 from the exhaust or by-product of

OPEN ACCESS

RECEIVED

4March 2019

REVISED

19 June 2019

ACCEPTED FOR PUBLICATION

20 June 2019

PUBLISHED

22 July 2019

Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2019TheAuthor(s). Published by IOPPublishing Ltd

fermentation is captured, compressed, stored perma-nently (e.g. in geologic reservoirs), and thus removedfrom the natural carbon cycle (Lenton 2010, Caldeiraet al 2013). BECCS could potentially provide largeamounts of negative emissions (NEs), but in turncompetes with agriculture and other uses such as eco-system conservation for land requirements. Different(portfolios of)NETs (Minasny et al 2017, Werner et al2018) or alternative mitigation pathways (van Vuurenet al 2018) are receiving more and more attention, butbioenergy utilization will likely be significant duringthe 21st century (Masson-Delmotte et al 2018), since itis relatively cheap, compared to direct-air-capture andmore land-effective than afforestation (Smith et al2016). Therefore, our study provides additional valuein support of making deployment decisions based notonly on economic, but also eco-hydrologic reasoning.

The cultivation of plants to generate biomass at thelevel needed to satisfy high NE demands requiresextensive plantation areas (Boysen et al 2017), andeven more so, if realized under rainfed conditions(Beringer et al 2011). Because of the land scarcity,future BPs are likely to be irrigated to a significantamount in order to expand into more marginal ter-rain. In view of already existing water stress in manyregions (Wada et al 2011, Schewe et al 2014), the quan-tification of freshwater demands for large-scaleBECCS is critical but remains largely unknown—especially under the assumption not to constrain exist-ing demands from agriculture, industry, and domesticusers. Furthermore, there is a need to more system-atically explore the NE constraints imposed by fresh-water limitations (including the trade-off with flowrequirements to sustain freshwater ecosystems), andto what extent such limitations could be alleviated byoptimal water management on agricultural and BPareas.

Previous studies have provided first assessments offreshwater demands corresponding to large-scaleBECCS deployment required to constrain GMT rise.Berndes (2002) projected 2281 km3 yr−1 of additionalwithdrawals for biomass-based energy production of304 EJ yr−1 in 2100 (mainly from first generation BPs),while more recent estimates from Smith et al (2016)suggest 720 km3 yr−1 of additional water use toachieve NEs of 3.3 Gt C yr−1 in 2100. A further modelstudy by Bonsch et al (2016) arrived at an additionalwater demand of 3,362–5860 km3 yr−1 for generating300 EJ yr−1 in 2100. The large range of these estimatesresults from different assumptions on productivityincreases, the associated BP area demand, and irriga-tion water productivity levels. Accounting for diversespatially explicit nature protection areas, Beringer et al(2011) estimate a bioenergy water demand in the rangeof 1481–3880 km3 yr−1 to generate 130–270 EJ.More recently, Yamagata et al (2018) suggested1910 km3 yr−1 of consumptive water demand forbioenergy crops to achieve NEs of 3.3 Gt C yr−1, whileSéférian et al (2018) estimate the water demand

for producing 220–270 EJ in 2100 to be only178 km3 yr−1, which is probably a result of strongrestriction of irrigation and model limitations. Janset al (2018) project a demand of 1,500–5000 km3 yr−1

to generate 200–1000 EJ, while also securing environ-mental flow requirements (EFRs) with the prospect ofmaintaining freshwater ecosystems in a good state.

The large span in projected water demands as aresult of the diverse methodologies applied motivatesa more systematic and internally consistent approach.The present study comprehensively quantifies howmuch freshwater for irrigation of BPs will potentiallybe needed to constrain GMT rise to 1.5 °C above pre-industrial levels by the end of the century. It advancesprevious studies through process-based and spatio-temporally explicit simulations of water use and waterconsumption of BPs (in addition to other sectors),considering a range of irrigation intensities (includinga rainfed option), water management improvements,EFR protection goals, a range of carbon conversionefficiencies (percentage of carbon from harvest of BPsthat is permanently removed from the carbon cycle),and their combinations (table 1). The water require-ments under each of these setups are evaluated foryearly carbon sequestration demands simulated to fol-low a prescribed trajectory based on NE trajectoriesfor a 1.5 °C climate from Rogelj et al (2015) (see SIfigure S1 available online at stacks.iop.org/ERL/14/084001/mmedia), representative of the upper end ofthe set of exclusive 1.5 °C scenarios (those that are notwithin the ranges of likely or medium 2 °C scenarios).The respective NE demands ramp up from 0.54 Gt Cin 2030 to 5.45 Gt C in 2100. The scenarios analyzed inRogelj et al (2015) already take into account a widerange of technologies to reduce emissions, includingan increasing global carbon price which is assumed tolead to a lowering of total energy demand, increasingenergy efficiency, carbon capture and storage inremaining fossil fuel energy generation plants, greateruse of bioenergy in primary energy generation, elec-trification of the transport sector, and fossil fuel repla-cement (especially in the transport sector) by biofuels(Bauer et al 2018). By applying the NE demand curve,we implicitly incorporate these underlying modelassumptions of the socio-economic scenarios con-sistent with 1.5 °C. The focus of our analysis is on thesequestration of carbon via BECCS that could serve toachieve the prescribed NE targets, above and beyondthe effects of these other transformations, and specifi-cally on the associated water requirements. We do nothowever consider the economic aspects of imple-mentation of such strategies, which are beyond thescope of the current analysis.

The total sequestration demand corresponding tothis target is 255 Gt C over 2030–2100. To account forthe possibility of partial or failed mitigation (Werneret al 2018), and, thus, a higher NE demand for com-pensating remaining emissions, a more ambitioustotal sequestration demand of 355 Gt C is also

2

Environ. Res. Lett. 14 (2019) 084001

explored, obtained by linearly upscaling the originalyearly demand.

To account for limited land availability, only areasoutside of current urban and agricultural land as wellas areas of conservation interest are considered forconversion. All simulations are performed with theDynamic Global Vegetation Model LPJmL, whichcomputes terrestrial water cycling coupled to the car-bon balance and vegetation growth of BPs alongsideagricultural and natural vegetation, at daily time stepson a global 0.5° grid (Schaphoff et al 2018). LPJmLdynamically represents land surface processes such asdischarge routing, crop growth, and water use effi-ciency, as well as yield responses to various stresses inany given grid cell. These features allow to dynamicallychoose the most productive BP type, based on localsoil type, climate, andmanagement options available.

Analysis is driven by the research question whe-ther and under which constellations (degree of irriga-tion, consideration or neglection of EFRs, on-fieldwater management) the targeted NE demands can bemet while minimizing the additional pressure on glo-bal freshwater resources.

Methods

ScenariosWe compare the water requirements associated withthe two sequestration demands (cumulative 255 Gt Cand 355 Gt C between 2030 and 2100, with annualcontributions as in figure S1) for four different wateruse scenarios: rainfed only (RF), unconstrained irriga-tion withdrawals (IRR), availability-constrained irri-gation respecting environmental flow requirements(EFR), and the latter combined with improved cropwater management (WM). For each of them, sub-scenarios are evaluated, considering a basic para-meter setting representing low-technology BECCSwith only a fraction of the yield being irrigated,

and two technologically more ambitious pathways(increased conversion efficiency—TechUp and irri-gation expansion—IrrExp) (see table 1). BPs wereonly considered to be grown on areas outside ofurban and agricultural land as well as areas ofconservation interest. The remaining areas wereconsecutively (starting with the highest ratio of netbiomass yield per irrigationwater per area) convertedto BP plantations until the respective sequestrationgoal was reached (see below). The scenarios were allcomputed independently of each other.

In scenario RF, only rainfed BPswere allowed to becultivated; the extent of food cropland (see potentialarea extent of BPs) and assumptions on irrigation sys-tem and extent of irrigated area (Jägermeyr et al 2015)were fixed at the state of 2015 in this and all other sce-narios, as it is beyond the scope of this study to accountfor simultaneous changes in food demand and agri-cultural area. RF also serves as a reference scenario forglobal water withdrawals for purposes other than BPs(households, industries, livestock (HIL), and irrigatedagriculture). As irrigation of BPs is absent, this sce-nario has the least additional impact on freshwaterresources (aside from indirect impacts on streamflowdue to a change in evapotranspiration (ET) of BPs,compared to the previous land-use).

In IRR, sprinkler-irrigated herbaceous BPs anddrip-irrigated woody BPs (for more information onBP types, see the LPJML mode) can be grown in anysuitable grid cell as long as there is enough freshwateravailable in rivers, lakes, and reservoirs (Jägermeyr et al2017). However, if irrigationwould not increase yieldsbymore than 50% (determined in an extra simulation,see below), rainfed BPs are assumed instead in order toirrigate only those BPs, where irrigation increases theyield significantly. Irrigation, as for crops, is applied ona daily basis, when soil moisture falls below a plant-type specific threshold. HIL demand is assumed to beprioritized over irrigation water demand in all scenar-ios, using data from Flörke et al (2013). In case there is

Table 1.Parameterization of BP simulations and respective watermanagement assumptions (RF, IRR, EFR,WM). Eachwatermanagement scenario is simulated for five different BP parameterizations (basic, TechUp, IrrExp, TechUp355,IrrExp355). The latter two refer to a higher sequestration target of 355 Gt C. irrfrac—maximumglobally irrigated BP yieldshare (1.0—all BPs can potentially be irrigated; 0.33—atmost a third of the BPs can be irrigated); ceff—fraction of thecarbon from the harvested biomass, which can be permanently removed from the carbon cycle (50%or 70%).

Scenario RF IRR EFR WM

Rainfed Unconstrainedwithdrawals Respect environmental Watermanagement

Flow requirements

Irrigation of BPs No Yes Yes Yes

Environmental No No Yes Yes

flowprotection

Watermanagement No No No Yes

Parameter set Basic TechUp IrrExp TechUp355 IrrExp355

MaximumBP irrigation fraction (irrfrac) 0.33 0.33 1.0 0.33 1.0

Carbon conversion efficiency (ceff) 50% 70% 50% 70% 50%

Carbon sequestration goal (seq) 255 Gt C 255 Gt C 255 Gt C 355 Gt C 355 Gt C

3

Environ. Res. Lett. 14 (2019) 084001

not enough water left for meeting the demand of agri-cultural crops and BPs, the allocation of the availablewater is distributed according to the ratio of therespective areas. In this scenario, there are no con-straints to water withdrawals, thus representing a casewith the largest potential withdrawals and the highestNEpotential.

In the EFR scenario, the daily amount of availablewater for irrigation in a grid cell is capped. The EFRsare calculated according to the variable monthly flowestimation method (Pastor et al 2014), which classifiesmonths as low-, medium-, and high-flowmonths andallocates 60%, 45%, and 30%of the flow for ecosystempurposes, respectively. EFRs are determined as 30-yraverages from a simulation based on historical land-use (Jägermeyr et al 2017) and the climate of the period1970–1999. Hence, only water in excess of these refer-ence EFRs is allowed to be used for BP irrigation in thefuture period. If EFRs are transgressed in a river basin(determined from the outflow cell) solely due to non-BP withdrawals, only rainfed BPs are assumed to becultivated there.

Finally, scenario WM assumes that in addition tothe EFR setup, advanced water management strategiesare applied on both food cropland and BPs. They cor-respond to practices such as mulching, local runoffcollection for supplemental irrigation during dryspells, modified irrigation thresholds, and soil man-agement practices (see also the LPJmL model anddescription in Jägermeyr et al (2016)).

For each of the water management scenarios, weconsider BP variants with different assumptions onthe carbon sequestration demand (seq), carbon con-version efficiency (ceff), and the maximum BP irriga-tion fraction (irrfrac). ceff defines, how much of thecarbon from the harvested biomass can be perma-nently removed from the carbon cycle (50% or 70%).The remaining carbon would eventually be trans-ported back to the atmosphere and thus not perma-nently removed. A BECCS life-cycle assessment bySmith and Torn (2013) reveals overall conversion effi-ciencies of 47%, while capture rates of CCS processestypically achieve 85%–90% (Gough and Vaughan2015). Technological change is likely to improve theefficiencies by reducing losses over time, which moti-vates our ambitious level of carbon conversion effi-ciency for the whole BECCS process-chain of 70%.The maximum BP irrigation fraction (irrfrac) indicatesthe maximum level of BP irrigation (1.0—all BPs canpotentially be irrigated; 0.33—at most a third of theBPs can be irrigated, roughly representing circum-stances where economic or other constraints to irriga-tion infrastructure apply; 0 for scenario RF).

In the basic parameter set, we consider the NEdemands of the regular emission pathway with nomitigation failure (seq=255 Gt C), a moderate car-bon conversion efficiency (ceff=50%) and a moder-ate irrigation fraction (irrfrac=0.33%). In theparameter sets TechUp and IrrExp, the parameters are

changed to ceff=70% and irrfrac=1.0, respectively.In order to account for increased NE demands causedby failed mitigation actions, we apply the setsTechUp355 and IrrExp355 which use the same para-meters for ceff and irrfrac as TechUp and IrrExp, but thesequestration demand is set to 355 Gt C (see table 1).

Potential area extent of BPsThe maximum land area that can be converted to BPs(fraction of 0.5° grid cell) was derived by excludingcurrent cropland (Frieler et al 2017, in year 2015 basedon HYDE 3.2 by Klein Goldewijk et al (2017)),secondary forest areas for industrial roundwoodproduction and urban build-up areas (Hurtt et al2016), intact forest landscapes (Potapov et al 2017),wetlands (Lehner and Döll 2004), and areas ofconservation interest. Areas of biodiversity concernare derived from a binary dataset developed in thisstudy, considering regions crucial to ecosystem func-tioning (see figure 1(a), a similar map is also used inWerner et al (2018)). Previous approaches usuallypreserved fractions of grid cells for conservation(Beringer et al 2011, Boysen et al 2016) rather thanexcluding entire cells, which can be interpreted as aland sparing approach. Here, a grid cell is excludedfrom conversion to BPs for reasons of biodiversityprotection if it is covered by the World Database onProtected Areas (UNEP-WCMC and IUCN 2015) or iflocated within Biodiversity Hotspots (Mittermeieret al 2011). In addition, we incorporated a catalogueon endemism richness, assuming plants as proxies forall floral and faunal species (Kier et al 2009), conser-ving all areas with an endemism richness above theglobal average (>21.66 endemic species km–2).Finally, a dataset on threatened species (mean value ofamphibians, birds, and mammals) was included(Pimm et al 2014), based on which we assume cells tobe protected where more than 3% of all species arecurrently threatened.

The global area potentially suitable for BPs accord-ing to our configuration sums up to 3286Mha(figure 1(b)). This would be more than twice the cur-rent cropland area. Large portions of this area, how-ever, can not sustain BPs with yields above theminimum yield threshold of 2.5 t C ha−1 yr−1 due toclimatic conditions, or are associated with too highland-use change (LUC) emissions due to the conver-sion of natural land to a BP (Houghton et al 2012, Har-per et al 2018). We only consider grid cells if the meanyield for the period from plantation start until 2099 isabove the harvest threshold. To calculate the LUCemissions as part of the carbon budget, we comparethe size of litter, soil and vegetation carbon poolsbefore and after the conversion to BPs and only con-sider sites where LUC emissions are at least two timescompensated for by the net sequestration amount,excluding areas where plantation of bioenergy wouldonly be marginally useful. To choose the most suitable

4

Environ. Res. Lett. 14 (2019) 084001

type of BP for each grid cell (see below for bioenergyfunctional types in LPJmL), fivemodel runs (assumingplantation on all potential areas with the same type ofBP—woody, irrigated woody, herbaceous, irrigatedherbaceous, no BP)were performed for scenarios IRR,EFR, and WM. These were used to determine thepotential yields and water demands for all grid cell

shares available for conversion to BPs in each simula-tion year. For RF, three such pre-runs (woody, herbac-eous, no BP) were sufficient, since irrigation isdisallowed.

The net yield (nY) for all four possible BP types(rainfed versus irrigated and woody versus herbac-eous) is given by the carbon conversion efficiency (ceff),

Figure 1. (a)Areas excluded from conversion to BPs due to biodiversity and conservation criteria: Biodiversity hotspots (Mittermeieret al 2011),WorldDatabase on Protected Areas (UNEP-WCMCand IUCN2015), endemic species (Kier et al 2009), threatened species(Pimm et al 2014) (b) potential BP fractional area (%) outside of regions covered by cropland and pastures, or regions protected forreasons of biodiversity (see a andMethods for detailed description). (c)Mean 2090–2099 fractional areas for rainfed (red) and irrigated(blue)BP assumed in scenarioWMandparameter set IrrExp; together with factors for not considering BPs in remaining potentiallyavailable areas shown in (b).

5

Environ. Res. Lett. 14 (2019) 084001

the yield of the respective bioenergy plant (beY), andthe potential timber yield from the initial land-useconversion (tY):

nY c beY tY0.475 , 1eff= +· · ( ) ( )

where ceff defines the percentage of the harvestedcarbon sequestered and thus extracted from theatmosphere; the factor 0.475 describes the averagecarbon content of dry biomass from Schlesinger andBernhardt (2013).

In every grid cell, the net yield is compared withthe associated LUC emissions (see figure S2). For mostregions, BPs reduce the natural carbon holding capa-city and thus have positive LUC emissions. In regionssuch as eastern Australia, the central/northern UnitedStates, or southern Africa, however, managed BPs canhave enhancing carbon sequestration effects, besidesthe yield. For the runs with EFR constraints, the mostproductive rainfed BP type is chosen if the whole basinor the current cell is already transgressing the EFRrequirements. Subsequently, all cells are rankedaccording to their yield/irrigation water ratio (irriga-tion water amount is set to 1 if it is a rainfed cell) andfrom this record, the cells are chosen consecutively(meaning the cell with the highest ratio—least waterper yield—is selected first) until the sequestration goalof the respective year is reached. Thereby, overall pro-ductivity in each grid cell determines both the type ofBP and irrigation, which in turn depend on the soiltype, climate conditions, and water availability. Thisresults in unique spatial patterns for each scenario (seefigure S3).

LPJmLmodelAll simulations were conducted with the process-based Dynamic Global Vegetation Model LPJmL(Schaphoff et al 2013, 2018), which has recently beenevaluated against various data sets from in situmeasurement sites, satellite observations, and agricul-tural yield statistics in Schaphoff et al (2018). Themodel considers 67420 land grid cells on a 0.5°×0.5°global grid. It simulates terrestrial carbon fluxes forestablishment, growth, and productivity of naturalvegetation (computed dynamically based on climaticconditions), agricultural crops, and pasture (Bondeauet al 2007), as well as water fluxes like ET, irrigation,and river routing (Gerten et al 2004, Rost et al 2008,Biemans et al 2011). For 12 crop functional typescalibrated to match national yield statistics (Fader et al2010) and a group of other annual and perennial crops,sowing dates are dynamically calculated (Waha et al2012), but herefixed after year 1999.

Themodel also considers two types of second-gen-eration bioenergy crops. Woody bioenergy crops areparameterized as willows or poplars for temperateregions and Eucalyptus for the tropics. Herbaceousbioenergy crops are parameterized as Miscanthus orswitchgrass. Herbaceous BPs are assumed to be har-vested once the above-ground carbon storage reaches

400 g m−1, but at least once a year. Bioenergy trees areharvested every eight years, with a maximum planta-tion life time of 40 years before total clearance andregrowth of saplings. The computed yields have beenevaluated against field data by Beringer et al (2011) andHeck et al (2016).

Dependent on the scenario, managed areas can berainfed or irrigated, which determines the source ofwater to fulfill the demand of the plants to be eitheronly precipitation water or precipitation and addi-tional water from local storage or main discharge ofthe respective grid cell and neighboring cells. The irri-gation module accounts for three irrigation techni-ques: surface, sprinkler, and drip (Jägermeyr et al2015), with different supply efficiencies. Water use forhousehold, industry, and livestock (HIL) (Flörke et al2013) is prescribed. Additionally, water managementstrategies such as mulching, water harvesting, andconservation tillage are represented for cropland and,newly in this study, for BPs as well, following (Jäger-meyr et al 2016) by adapting the parameters (reducedsoil evaporation of 50%, local storage capacity of200 mm, collected on 50% of the managed areas, irri-gation if soil moisture <40% of field capacity, andoptimized soil infiltration) for BPs.

We forced the LPJmLmodel withmonthly climatedata (1901–2100) from the PanClim dataset (Heinkeet al 2013) consistent with a 1.5 °C trajectory in 2100with a slight temperature overshoot; with soil texturedata (Nachtergaele et al 2009), and with land-use pat-terns (prescribed agriculture from Frieler et al (2017),based on HYDE 3.2 by Klein Goldewijk et al (2017),and BPs as per scenario). Since the target variables inthis study (freshwater withdrawals, BP area, carbonsequestration) are much more sensitive to the indivi-dual parameter setups than to the actual climate input(forcing LPJmL with output from other climate mod-els changed global BP water consumption by ±4%;data not shown), we force themodel with only one cli-mate model (MPI-ECHAM5). Simulations are per-formedwith an initial spinup of 5010 years of potentialnatural vegetation (recycling the first 30 years of cli-mate input) to bring global carbon pools to an equili-brium, followed by 316 years of transient spinup usinghistoric land-use patterns from1700 to 2015. The foodcrop land-use pattern from 2015 is kept constant forthe remainder of the 21st century. BP plantations areassumed to not be implemented before 2030.

The total annual water withdrawals in every gridcell are computed as the sum of applied irrigationwater as well as drainage and evaporative conveyancelosses and withdrawals for HIL.Water consumption iscomputed as the sum of applied irrigation water, eva-porative conveyance losses, and HIL consumptionminus return flows from applied irrigation water.Attribution of consumption and withdrawal to BPs isobtained through computing the cell-wise differencebetween withdrawals in the run with BPs and thereference simulationwithout.

6

Environ. Res. Lett. 14 (2019) 084001

Results

The projected total global freshwater withdrawals(2090–2099) exhibit a large range between 2619and 5998 km3 yr−1, with a BP contribution of387–3167 km3 yr−1 (see table 2 for tabled simulationdata). The baseline scenario without BPs reaches∼3000 km3 yr−1 for the same period. Adding BP withunrestrained withdrawals (IrrExp—IRR) almost dou-bles the total withdrawals compared to purely rainfedBPs. By respecting EFRs and applying improved watermanagement (EFR and WM), the total global waterwithdrawals can be kept below 4000 (3000) km3 yr−1

in IrrExp (TechUp). Note that despite non-negligiblewithdrawals for BPs in the order of 400 km3 yr−1 inscenario WM of basic and TechUp setups, the totalwithdrawalsmay even fall below those of the respectiveRF scenario (3011 km3 yr−1), because EFRs are takeninto account also for withdrawals of agriculturalirrigation and because water is assumed to be moreeffectively managed also on cropland. Total global(food) crop yields are not substantially changing forRF and IRR compared to a reference run without BP.They are reduced by 3.5% in EFR, while in WM thewater and soil management results in 8.4% highercrop yields than in RF.

We observe thatmost of the scenarios do not reachthe target sequestration, meaning that from a certainyear on, no more additional BP area is available thatfulfills the respective scenario requirements. The dedi-cated freshwater withdrawals for irrigation of BPs nee-ded to provide 255 Gt C of NEs range from416 km3 yr−1 (TechUp—WM) to 2388 km3 yr−1

(IrrExp—IRR).In the basic scenarios RF, IRR, EFR, and WM

(figure 2, bottom center), total NEs from BPs are notfulfilling the sequestration target of 255 Gt C. The RFscenario reaches 170 GtC (with no additional wateruse on top of the global non-BPwater use of currently3011 km3 yr−1). Irrigation of BPs (unconstrainedby EFRs) with 701 km3 yr−1 (2090–2099 mean)

Table 2.Globalmodel results for simulations included in figure 2.

Basic Unit RF IRR EFR WM

Sequestration Gt C 225 263 229 246

Net sequestration

(Seq − LUC)Gt C 170 217 181 195

Total BECCS yield Gt C 450 525 458 491

Rainfed BP yield Gt C 403 337 284 308

Total BP area Mha 1036 1416 1177 1247

Onlywoody

BP area

Mha 725 1047 881 927

Total withdrawals km3 yr−1 3011 3653 2739 2619

BPwithdrawals km3 yr−1 0 701 400 387

Total bluewater

consumption

km3 yr−1 1160 1782 1237 1144

BP bluewater

consumption

km3 yr−1 0 642 361 351

IrrExp Unit RF IRR EFR WM

Sequestration Gt C 225 275 258 278

Net sequestration

(Seq − LUC)Gt C 170 262 226 243

Total BECCS yield Gt C 450 550 517 556

Rainfed BP yield Gt C 403 58 121 118

Total BP area Mha 1036 1195 1164 1215

Onlywoody

BP area

Mha 725 1001 909 927

Total withdrawals km3 yr−1 3011 5280 3749 3895

BPwithdrawals km3 yr−1 0 2388 1474 1742

Total bluewater

consumption

km3 yr−1 1160 3358 2216 2313

BP bluewater

consumption

km3 yr−1 0 2239 1368 1553

TechUp Unit RF IRR EFR WM

Sequestration Gt C 318 326 320 326

Net sequestration

(Seq − LUC)Gt C 252 272 261 268

Total BECCS yield Gt C 455 466 457 466

Rainfed BP yield Gt C 388 282 270 275

Total BP area Mha 946 1158 1072 1097

Onlywoody

BP area

Mha 570 821 738 779

Total withdrawals km3 yr−1 3011 3612 2755 2654

BPwithdrawals km3 yr−1 0 638 417 416

Total bluewater

consumption

km3 yr−1 1160 1735 1258 1173

BP bluewater

consumption

km3 yr−1 0 587 383 378

IrrExp355 Unit RF IRR EFR WM

Sequestration Gt C 224 330 268 294

Net sequestration

(Seq − LUC)Gt C 170 321 235 259

Total BECCS yield Gt C 447 659 535 587

Rainfed BP yield Gt C 414 63 127 127

Total BP area Mha 1069 1377 1198 1262

Onlywoody

BP area

Mha 750 1105 920 937

Total withdrawals km3 yr−1 3010 5998 3768 3903

BPwithdrawals km3 yr−1 0 3167 1493 1744

Total bluewater

consumption

km3 yr−1 1160 4041 2227 2324

BP bluewater

consumption

km3 yr−1 0 2946 1379 1561

Table 2. (Continued.)

TechUp355 Unit RF IRR EFR WM

Sequestration Gt C 344 396 349 370

Net sequestration

(Seq − LUC)Gt C 277 337 289 309

Total BECCS yield Gt C 492 566 498 529

Rainfed BP yield Gt C 436 357 309 326

Total BP area Mha 1055 1396 1179 1237

Onlywoody

BP area

Mha 629 906 772 823

Total withdrawals km3 yr−1 3011 3731 2756 2707

BPwithdrawals km3 yr−1 0 775 415 473

Total bluewater

consumption

km3 yr−1 1160 1847 1254 1213

BP bluewater

consumption

km3 yr−1 0 706 378 419

7

Environ. Res. Lett. 14 (2019) 084001

increases this value to 217 GtC (IRR). With stringentenvironmental flow protection (EFR) the waterdemand is reduced to 400 km3 yr−1, whereby a totalsequestration of only 181 GtC is achievable. Addi-tional water management (WM) strategies slightlyincrease the sequestration to 195 GtC while stayingbelow the irrigation water demand of EFR(387 km3 yr−1).

To possibly increase the carbon sequestration, weconsidered either irrigation expansion or technologyupgrades. An increase of irrfrac from 0.33 to 1.0(figure 2, bottom right) enables scenario IRR to reachthe sequestration goal andWM to almost reach it (243GtC). These gains, however, come at the cost ofstrongly increased water withdrawals for the BPs. IRRmore than triples the demand for BP irrigation to2388 km3 yr−1, while in the WM scenario, more thanfour times more irrigation water is used(1742 km3 yr−1) compared to basic. In the EFR sce-nario, less water compared to WM is used(1474 km3 yr−1), however, for a lower sequestrationamount. In the TechUp setup (figure 2, bottom left),in which ceff is increased from 50% to 70%, the addi-tional carbon that can be sequestered from the rawyields is enough to fulfill the sequestration targetof 255 Gt C in all four scenarios (RF, IRR, EFR, WM).As a beneficial effect, the associated freshwaterwithdrawals for BP irrigation are comparable tothose of the basic setup (IRR, 638 km3 yr−1; EFR,417 km3 yr−1;WM, 416 km3 yr−1).

The higher sequestration demand of 355 Gt C,which could become necessary due to delayed or failedmitigation, was analyzed in the TechUp355 (top left)and IrrExp355 setups (top right). None of the scenar-ios, however, can deliver sequestrations that high. The

IRR scenarios come the closest, although they neglectthe EFRs (337 GtC / 775 km3 yr−1 for TechUp355 and321GtC / 3167 km3 yr−1 for IrrExp355).

For scenarios that reach the sequestration goal(figure 3) with restricted irrigation use (TechUP—RF,IRR, EFR, and WM) the majority of irrigated BPs issituated in higher latitudes, namely Canada, Scandina-via, and Russia (due to the preference for cells with alow water/productivity ratio), while areas of highestproductivity (figure S3) and highest LUC emissions(figure S2) are in the tropics. The biophysical limita-tions allow the productive growth of herbaceous bioe-nergy plants only in latitudes between −40° and 50°.Due to their plant physiology, woody bioenergy plantshave significantly lower yield productivities, but areable to grow in subpolar regions. The optimizationscheme also simulates plantation of bioenergy trees in

Figure 2.Total global carbon sequestration (2030–2100) and yearly freshwater withdrawals (mean 2090–2099) includingwithdrawalsfor agriculture andHIL for the four scenarios infive different parameter setups. Basic, ceff=50%and irrfrac=0.33; IrrExp, same butirrfrac=1.0; TechUp, same as basic but ceff=70%.The upper panel (‘more sequestration’) refers to an increased sequestration targetof 355 Gt C instead of 255 Gt C. RF—rainfed, IRR—unconstrainedwithdrawals, EFR—respective environmental flow requirements,WM—watermanagement (see also table 1).

Table 3. List of abbreviations.

BECCS Bioenergywith carbon capture and storage

BP Bioenergy plantation

ceff Carbon conversion efficiency

EFR Environmental flow requirement

ET Evapotranspiration

GMT Globalmean temperature

HIL Household, industry and livestock

irrfrac Irrigation fraction

LPJmL Lund-Potsdam-Jenamanaged land—a dynamic global

vegetationmodel

LUC Land-use change

NE Negative emission

NET Negative emission technology

seq Carbon sequestration

VMF Variablemonthly flowmethod (EFR allocationmethod)

8

Environ. Res. Lett. 14 (2019) 084001

the tropics, which are either chosen for their greaternet carbon sequestration or their lesser need forirrigation.

Discussion

This study was designed to estimate the biophysicalpotential and water requirements for BECCS if beingapplied as the primary NET for fulfilling the 1.5 °Ctarget. This approach to model BECCS is based on

explicit modeling of BPs (and the associated emissionsfrom land-use change in the process of plantationallocation) together with an assumed carbon conver-sion efficiency. We thereby adopt an Earth Systemperspective based on the planetʼs biophysical capacityand especially the trade-offs associatedwith freshwateravailability and management, rather than explicitlyaddressing economic feasibility. We have not consid-ered the logistics and economics of transport of solidbiofuels, nor the costs of CCS (e.g. Tauro et al (2018),

Figure 3.Mean 2090–2099 fractional areas for rainfed (red) and irrigated (blue)BPs—left panel—, andwater withdrawals forirrigation of BPs (computed as difference of total withdrawalsminuswithdrawals from food-crops-only reference run)—right panel—displayed for all scenarios that fulfill the sequestration target of 255 Gt C. Attributed bluewater withdrawals can be negative, if therespective scenariowithdraws less water than the reference run. This can happen for cells, where addition of BPs changes local ET-fluxes, or new upstream irrigation reduces discharges belowEFRs.

9

Environ. Res. Lett. 14 (2019) 084001

Strefler et al (2018)) as these were beyond the scope ofour research. We acknowledge that these issues will beimportant for the feasibility of strategies for BECCS,not least because the areas identified having the great-est potential for BPs are far from areas of greatestenergy demand. However, if these constraints wereconsidered additionally in a more comprehensivestudy, NE potentials may not necessarily be lower butthe BP area would most likely be simulated to shift toother regions (see Bonsch et al 2016).

The key finding is that NE demands necessary tolimit global warming to 1.5 °C cannot be met byBECCS alone due to freshwater limitations (and underthe land available for conversion assumed here),except under the most ambitious assumptions aboutconversion efficiency or water use. The only scenarionot relying on a high carbon conversion efficiency of70% is a scenario without respecting EFRs, which thuswould come at the cost of riverine ecosystems andoverall environmental sustainability. SafeguardingEFRs in turn, would largely limit the irrigation-sus-tainedNE potential. These results add new evidence tothe discussion that pathways towards higher water useefficiencies and carbon conversion efficiencies need tobe prioritized to meet targeted NEs. The projectedadditional freshwater withdrawals for achieving the1.5 °C target using BECCS as the primary NET(figure 2) are substantial (up to 2400 km3 yr−1

—mean2090–2099) and could thus reach the order of currentglobal water withdrawals. Correspondingly, the totalwater consumption across all sectors would rise toabove 3300 km3 yr−1 (figure S4), thereby possiblytransgressing the ‘planetary boundary’ for freshwateruse (currently set at a total human consumption of2800 or 4000 km3 yr−1, respectively; Gerten et al 2013,Steffen et al 2015), with associated detrimental effectsfor the Earth system. In comparison with previouswater consumption estimates for BPs, as for instancein Beringer et al (2011) (1481–3880 km3 yr−1), Bonschet al (2016) (3000–6000 km3 yr−1), or Jans et al (2018)(1500–5000 km3 yr−1), our global estimates exhibit asimilar to larger span while being somewhat moreconservative in absolute terms (351–2946 km3 yr−1)due to the large range of scenarios considered andother divergent assumptions such as on the potentiallocations for BPs in particular. Our study thus doesnot constrain the previous range but provides a sys-tematic exploration of underlying causes (water uselimitations, environmental constraints, managementoptions).

Thus, it is important to note for our study as wellthat the global amount of freshwater requirementsstrongly depends on the underlying assumptionsabout conversion efficiency, water management, andEFR protection. Most freshwater is simulated to beconsumed in the IrrExp scenario, whereas the thebasic and TechUp scenarios involve significantly lowerwater consumption. Naturally, among the water usescenarios of each parameter setup, IRR always leads to

highest water consumption, while EFR and WMshow lower values due to their strict water allocationscheme (EFR) and the constrained water use (WM),respectively.

Our results indicate that a targeted NE amount of255 Gt C (between 2030 and 2100) could be producedunder rainfed conditions only if high conversion effi-ciencies would apply (ceff�70%). Even under thiscondition, though, rainfed BPs would not provideenough biomass for possible higherNE demands up tothe here considered 355 Gt C, which may becomenecessary if climate change mitigation efforts fail orslow down. The basic setup cannot provide enoughNE to fulfill the sequestration demand of even thelower target (255 Gt C), suggesting that either irriga-tion expansion or highly efficient BECCS systemsexceeding 50% carbon conversion efficiency will beneeded (or a combination of both). It appears unlikelyto implement such high efficiencies at the global levelwithin the next decades due to multiple obstacles suchas a lack of socio-political acceptance, policy incen-tives, and technological readiness (Fuss et al 2014,Reiner 2016, Fridahl and Lehtveer 2018, Gough et al2018, Vaughan et al 2018).

In view of these technical and institutional chal-lenges, productivity improvements supported by irri-gation expansion come into focus for near-termsolutions. It is clear that additional water withdrawalsat the level presented here would be associated withsevere environmental degradation (at least in scenar-ios where EFRs are not respected) or increased waterstress (Rockström et al 2014, Hejazi et al 2015). Whilesuch obstacles require further systematic study, anysustainable implementation of BECCS requires ser-ious consideration of freshwater issues in the form ofrigid environmental protection, water legislation, andwatermanagement improvements.

In addition to the water requirements for irriga-tion, BPs need extensive land areas (for furtherdiscussion, see Boysen et al (2016), Heck et al (2016),Werner et al (2018)). In our study, themaximum addi-tional arable area for BPs under rainfed conditions isroughly 1000Mha. Irrigation makes more grid cells(200–400Mha) productive enough to cross the mini-mum yield threshold and compensate for LUC emis-sions (see figure S2). The yield threshold is the lowerlimit of what is considered economically feasibletoday, while both yields and the threshold may changein the future (even though they are already quite opti-mal parameters) due to e.g. genetic optimization andmanagement. The assumption that BPs would only beplanted if LUC emissions are at least twice compen-sated for by the net sequestration amount is strict, buteconomically justified. Conversely, irrigation makesBPs in many regions more productive, such that perunit of NEs, less land is needed. This can be under-stood as a trade-off between water and land, which hasbeen described before (Bonsch et al 2016, Jans et al2018).

10

Environ. Res. Lett. 14 (2019) 084001

However, large portions of the identified poten-tially available areas for BPs in this study are recrea-tional areas or wild remote landscapes which arealready in a state of increasing risk for biodiversity loss(Steffen et al 2015). Given that the scenarios suggestreplacement of, for example, larger fractions of borealforest in Scandinavia and northern Asia, which is unli-kely to occur in reality at such large scale, our estimatesappear to be on the conservative side. If those areaswould not be released for conversion, larger BP areas,or more intense irrigation, would have to take placeelsewhere to achieve a similar amount of NEs, prob-ably involving even stronger pressure on freshwatersystems there. Thus, we stress that the here simulatedspatial BP patterns are to be interpreted as biophysicalmaximum potentials derived under strict conserva-tion criteria, distributed and optimized globallyaccording to the water use efficiency. Further analysiscould evaluate the wider consequences of ecosystemchange (e.g. terrestrial and aquatic biodiversity lossthrough conversion of natural land to BPs), like Ost-berg et al (2018) provide for biospheric change underscenarios designed to sustain Paris mitigation efforts.Additionally, competition for water between irrigatedagriculture and BPs could be explicitly studied by, forexample, exploring scenarios where the irrigation ofcrops always has the highest priority.

In sum, we find that second-generation bioenergycombined with CCS alone can deliver sufficient NEsfor ambitious climate targets only under highly opti-mized conditions and with potentially detrimentalside effects on freshwater ecosystems. This first bench-mark quantification merits more detailed follow-upstudies, especially to analyze synergies and trade-offswith additional NETs operating in different domains,in a complex modeling framework. However, accord-ing to initial studies, other NETs would come alongwith environmental side-effects too, for example, withrespect to the area demand of afforestation or thewater demand of direct-air-capture (Smith et al 2016).

Conclusion

Despite the socio-political and technological barriersto the implementation of BECCS, bioenergy will mostlikely becomemore relevant as a substitution for fossilenergy with the need to convert large areas to BPs. Toincrease the yields and thus reduce the pressure onland, these plantations might have to be irrigated to asubstantial degree, potentially putting many fresh-water systems under severe additional pressure.Therefore, local water policies, such as for safeguard-ing EFRs, are important tools to sustain the integrity offreshwater ecosystems. We show that there is a trade-off between limiting irrigation on BPs to sustain EFRsand attaining levels of NEs likely required for limitingglobal warming to 1.5 °C. On-field water and soilmanagement can help reducing this water gap for BPs

and for agriculture. Nevertheless, a stringent and fastreductionofCO2 emissions is inevitable, because highercarbon sequestration demands would have profoundimpacts on freshwater systems and their ecologicalfunctions that are fundamental to life and societies.

Acknowledgments

The authors declare no conflict of interest. This studywas funded by the CE-Land+ project of the GermanResearch Foundationʼs priority program DFG SPP1689 on ‘Climate Engineering—Risks, Challenges andOpportunities?’, by the BMBF project BioCAP-CCS(Grant No. 01LS1620B) and with partial support fromthe University of Chicago Center for Robust Decision-making on Climate and Energy Policy (NSF Grant#SES-146364). We thank Sibyll Schaphoff, JensHeinke, Wolfgang Lucht, and Sebastian Ostberg forvaluable discussions, as well as two anonymousreviewers for their constructive comments.

ORCID iDs

Fabian Stenzel https://orcid.org/0000-0002-5109-0048Dieter Gerten https://orcid.org/0000-0002-6214-6991ConstanzeWerner https://orcid.org/0000-0003-1511-3086Jonas Jägermeyr https://orcid.org/0000-0002-8368-0018

References

BauerN et al 2018Global energy sector emission reductions andbioenergy use: overview of the bioenergy demand phase ofthe EMF-33model comparisonClim. Change 1–16

Beringer T, LuchtWand Schaphoff S 2011 Bioenergy productionpotential of global biomass plantations under environmentaland agricultural constraintsGCBBioenergy 3 299–312

BerndesG 2002 Bioenergy andwaterthe implications of large-scalebioenergy production forwater use and supplyGlob. Environ.Change 12 253–71

BiemansH,Haddeland I, Kabat P, Ludwig F,Hutjes RWA,Heinke J, von BlohWandGertenD2011 Impact of reservoirson river discharge and irrigationwater supply during the 20thcenturyWater Resour. Res. 47W03509

BondeauA et al 2007Modelling the role of agriculture for the 20thcentury global terrestrial carbon balanceGlob. Change Biol.13 679–706

BonschM et al 2016Trade-offs between land andwaterrequirements for large-scale bioenergy productionGCBBioenergy 8 11–24

Boysen LR, LuchtW,GertenD andHeckV 2016 Impacts devaluethe potential of large-scale terrestrial CO2 removal throughbiomass plantationsEnviron. Res. Lett. 11 095010

BoysenLR, LuchtW,GertenD,HeckV, LentonTMandSchellnhuberH J 2017The limits to global-warmingmitigationby terrestrial carbonremovalEarth’s Future5 463–74

Caldeira K, BalaG andCao L 2013The science of geoengineeringAnnu. Rev. Earth Planet. Sci. 41 231–56

FaderM, Rost S,Müller C, BondeauA andGertenD 2010Virtualwater content of temperate cereals andmaize: present andpotential future patterns J. Hydrol. 384 218–31

11

Environ. Res. Lett. 14 (2019) 084001

FlörkeM,Kynast E, Bärlund I, Eisner S,Wimmer F andAlcamo J2013Domestic and industrial water uses of the past 60 yearsas amirror of socio-economic development: a globalsimulation studyGlob. Environ. Change 23 144–56

FridahlM and LehtveerM 2018 Bioenergy with carbon captureand storage (BECCS): global potential, investmentpreferences, and deployment barriers Energy Res. Soc. Sci.42 155–65

Frieler K et al 2017Assessing the impacts of 1.5 °Cglobal warming-simulation protocol of the inter-sectoral impactmodelintercomparison project (ISIMIP2b)Geosci.Model Dev. 104321–45

Fuss S et al 2014 Betting on negative emissionsNat. Clim. Change 4850–3

GertenD,HoffH, Rockström J, Jägermeyr J, KummuMandPastor AV2013Towards a revised planetary boundary forconsumptive freshwater use: role of environmental flowrequirementsCurr. Opin. Environ. Sustain. 5 551–8

GertenD, Schaphoff S,HaberlandtU, LuchtW and Sitch S 2004Terrestrial vegetation andwater balancehydrologicalevaluation of a dynamic global vegetationmodel J. Hydrol.286 249–70

GoughC,Garcia-Freites S, Jones C,Mander S,Moore B, Pereira C,RöderM,VaughanN andWelfle A 2018Challenges to the useof BECCS as a keystone technology in pursuit of 1.5 °CGlob.Sustain. 1E5

GoughC andVaughanN2015 Synthesising existing knowledge onthe feasibility of BECCSAVOID2ReportWPD1a

Harper AB et al 2018 Land-use emissions play a critical role in land-basedmitigation for Paris climate targetsNat. Commun.9 2938

HeckV,GertenD, LuchtWandBoysen LR 2016 Is extensiveterrestrial carbon dioxide removal a ‘green’ formofgeoengineering? A globalmodelling studyGlob. Planet.Change 137 123–30

Heinke J, Ostberg S, Schaphoff S, Frieler K,Müller C, Gerten D,MeinshausenM and LuchtW 2013 A new climate datasetfor systematic assessments of climate change impactsas a function of global warmingGeosci. Model Dev. 61689–703

HejaziM I et al 2015 21st centuryUnited States emissionsmitigationcould increase water stressmore than the climate change it ismitigating Proc. Natl Acad. Sci. 112 10635–40

HoughtonRA,House J I, Pongratz J, van derWerf GR,DeFries R S,HansenMC, LeQuéréC andRamankuttyN 2012Carbonemissions from land use and land-cover changeBiogeosciences9 5125–42

Hurtt G et al 2016Harmonization of global land-use change andmanagement for the period 850–2100AGUFallMeetingAbstracts

Jägermeyr J, GertenD,Heinke J, Schaphoff S, KummuMandLuchtW2015Water savings potentials of irrigation systems:global simulation of processes and linkagesHydrol. EarthSyst. Sci. 19 3073–91

Jägermeyr J, GertenD, Schaphoff S,Heinke J, LuchtWandRockström J 2016 Integrated cropwatermanagementmightsustainably halve the global food gap Environ. Res. Lett. 11025002

Jans Y, Berndes G, Heinke J, LuchtW andGertenD 2018Biomass production in plantations: land constraintsincrease dependency on irrigation waterGCB Bioenergy 10628–44

Jägermeyr J, Pastor A, BiemansH andGertenD 2017Reconcilingirrigated food productionwith environmental flows forsustainable development goals implementationNat.Commun. 8 15900

KierG, KreftH, Lee TM, JetzW, Ibisch P L,Nowicki C,Mutke J andBarthlottW2009A global assessment of endemism andspecies richness across island andmainland regionsProc. NatlAcad. Sci. 106 9322–7

KleinGoldewijk K, Beusen A,Doelman J and Stehfest E 2017Newanthropogenic land use estimates for the holocene; HYDE3.2Earth Syst. Sci. DataDiscuss. 9 927–53

Lehner B andDöll P 2004Development and validation of a globaldatabase of lakes, reservoirs andwetlands J. Hydrol. 296 1–22

LentonTM2010The potential for land-based biological CO2

removal to lower future atmospheric CO2 concentrationCarbonManage. 1 145–60

Masson-Delmotte V 2018Mitigation pathways compatible with1.5 °C in the context of sustainable development IPCCSpecialReport on the Impacts of GlobalWarming of 1.5 °CAbove Pre-Industrial Levels andRelatedGlobal Greenhouse Gas EmissionPathways, in the Context of Strengthening the Global Responseto the Threat of Climate Change, Sustainable Development, andEfforts to Eradicate Poverty IPCC at press

Minasny B et al 2017 Soil carbon 4 permilleGeoderma 292 59–86Minx J C et al 2018Negative emissionspart 1: research landscape

and synthesisEnviron. Res. Lett. 13 063001MittermeierRA,TurnerWR,LarsenFW,BrooksTMandGasconC

2011Global biodiversity conservation:The critical role ofhotspotsBiodiversityHotspots (Berlin: Springer)pp3–22

Nachtergaele F, VanVelthuizenH,Verelst L, BatjesN,DijkshoornK, VanEngelenV, FischerG, Jones A,Montanarella L and PetriM2009Harmonizedworld soildatabaseReport ISRIC:Wageningen, TheNetherlands

Ostberg S, Boysen LR, Schhoff S, LuchtWandGertenD2018Thebiosphere under potential Paris outcomesEarth’s Future 623–39

Pastor AV, Ludwig F, BiemansH,HoffH andKabat P 2014Accounting for environmental flow requirements in globalwater assessmentsHydrol. Earth Syst. Sci. 18 5041–59

PimmSL, Jenkins CN,Abell R, Brooks TM,Gittleman J L,Joppa LN, Raven PH, Roberts CMand Sexton JO2014Thebiodiversity of species and their rates of extinction,distribution, and protection Science 344 1246752

PotapovP et al 2017The last frontiers ofwilderness: tracking loss ofintact forest landscapes from2000 to 2013Sci. Adv.3 e1600821

ReinerDM2016 Learning through a portfolio of carbon captureand storage demonstration projectsNat. Energy 1 15011

Rockström J et al 2014The unfoldingwater drama in theanthropocene: towards a resilience-based perspective onwater for global sustainability Ecohydrology 7 1249–61

Rockström J,GaffneyO, Rogelj J,MeinshausenM,NakicenovicN and SchellnhuberH J 2017A roadmap forrapid decarbonization Science 355 1269–71

Rogelj J, LudererG, Pietzcker RC, Kriegler E, SchaefferM,KreyV andRiahi K 2015Energy system transformations forlimiting end-of-century warming to below 1.5 °CNat. Clim.Change 5 519–27

Rogelj J, PoppA, CalvinKV, LudererG, Emmerling J, GernaatD,Fujimori S, Strefler J, HasegawaT andMarangoni G 2018Scenarios towards limiting globalmean temperature increasebelow 1.5°Nat. Clim. Change 8 325

Rost S, GertenD, BondeauA, LuchtW, Rohwer J and Schaphoff S2008Agricultural green and bluewater consumption and itsinfluence on the global water systemWater Resour. Res. 44W09405

Schaphoff S et al 2018 LPJmL4—a dynamic global vegetationmodelwithmanaged land: part II -model evaluationGeosci.ModelDev. Discuss. 2018 1–41

Schaphoff S,HeyderU,Ostberg S, GertenD,Heinke J and LuchtW2013Contribution of permafrost soils to the global carbonbudgetEnviron. Res. Lett. 8 014026

Schaphoff S et al 2018 LPJmL4—a dynamic global vegetationmodelwithmanaged land: part I -model descriptionGeosci.ModelDev. Discuss. 11 1343–75

Schewe J et al 2014Multimodel assessment of water scarcity underclimate change Proc. Natl Acad. Sci. 111 3245–50

SchlesingerWHandBernhardt E S 2013Biogeochemistry: AnAnalysis of Global Change (NewYork: Academic) p 120

Schleussner C-F, Rogelj J, SchaefferM, Lissner T, Licker R,Fischer EM,Knutti R, LevermannA, Frieler K andHareW2016 Science and policy characteristics of the paris agreementtemperature goalNat. Clim. Change 6 827–35

Smith L J andTornMS2013 Ecological limits to terrestrialbiological carbon dioxide removalClim. Change 118 89–103

12

Environ. Res. Lett. 14 (2019) 084001

Smith P,Davis S J, Creutzig F, Fuss S,Minx J, Gabrielle B, Kato E,JacksonRB, Cowie A andKriegler E 2016 Biophysical andeconomic limits to negative CO2 emissionsNat. Clim. Change6 42

SteffenW et al 2015 Planetary boundaries: guiding humandevelopment on a changing planet Science 347 1259855

Strefler J, BauerN, Kriegler E, PoppA,Giannousakis A andEdenhoferO 2018Between scylla and charybdis: delayedmitigation narrows the passage between large-scale CDR andhigh costs Environ. Res. Lett. 13 044015

SéférianR, RocherM,GuivarchC andColin J 2018Constraints onbiomass energy deployment inmitigation pathways: the caseof water scarcityEnviron. Res. Lett. 13 054011

TauroR,García CA, SkutschMandMaseraO2018The potentialfor sustainable biomass pellets inMexico: an analysis ofenergy potential, logistic costs andmarket demandRenew.Sustain. Energy Rev. 82 380–9

UNEP-WCMCand IUCN2015 Protected Planet: TheWorldDatabase on Protected Areas (WDPA) (Accessed: 18September 2017) (www.protectedplanet.net)

UNFCCCC2015Paris Agreement FCCCC/CP/2015/L. 9/Rev. 1UnitedNations Bonn

vanVuurenDP et al 2018Alternative pathways to the 1.5°C targetreduce the need for negative emission technologiesNat. Clim.Change 8 391

VaughanNE,GoughC,Mander S, Littleton EW,Welfle A,GernaatDEH J andVuurenDP v 2018 Evaluating the use ofbiomass energywith carbon capture and storage in lowemission scenarios Environ. Res. Lett. 13 044014

WadaY, VanBeek L andBierkensMF2011Modelling global waterstress of the recent past: on the relative importance of trendsinwater demand and climate variabilityHydrol. Earth Syst.Sci. 15 3785–805

WahaK, van Bussel LG J,Müller C andBondeauA2012Climate-driven simulation of global crop sowing datesGlob. Ecol.Biogeogr. 21 247–59

Werner C, SchmidtH-P, GertenD, LuchtWandKammannC2018Biogeochemical potential of biomass pyrolysis systems forlimiting global warming to 1.5 °CEnviron. Res. Lett. 13044036

Yamagata Y,HanasakiN, ItoA, Kinoshita T,MurakamiD andZhouQ2018 Estimatingwater-food-ecosystem trade-offs forthe global negative emission scenario (IPCC-RCP2.6)Sustain. Sci. 13 301–13

13

Environ. Res. Lett. 14 (2019) 084001

Supplementary Information667

2030 2040 2050 2060 2070 2080 2090 2100

02

46

8

Years

Ca

rbo

n r

em

ova

l [G

tC/y

r]

05

10

15

20

Ca

rbo

n r

em

ova

l [G

tCO

2/y

r]

255 GtC355 GtC

Figure S1: Amount of sequestration needed per year to stay within 1.5 ◦C warming (255 Gt C – black bars), afterRogelj et al. (2015), and to reach a higher sequestration demand of 355 Gt C (grey bars), obtained by linear up-scalingof the 255 Gt C curve.

−256

−128

−64

−32

−16

−8

−4

−2

−1

1

2

4

8

16

32

64

128

256

tC/ha

Figure S2: LPJmL-simulated LUC emissions for scenario WM and Basic parameter set, computed as the difference(2090–2099 average) relative to the reference run without BPs of the sum of the mean carbon content in soil, vegetationand litter pools, shown exemplarily for the TechUp parameter set and scenario WM.

18

−150 −100 −50 0 50 100 150

−60

−40

−20

020

4060

80

0 2 4 6 8 10 12

Woody BE tC/ha/yr

0 3 6 9 12 15 18 21 24

Grassy BE tC/ha/yr

a)

−150 −100 −50 0 50 100 150

−60

−40

−20

020

4060

80

0 2 4 6 8 10 12

Woody BE tC/ha/yr

0 3 6 9 12 15 18 21 24

Grassy BE tC/ha/yr

b)

−150 −100 −50 0 50 100 150

−60

−40

−20

020

4060

80

0 2 4 6 8 10 12

Woody BE tC/ha/yr

0 3 6 9 12 15 18 21 24

Grassy BE tC/ha/yr

c)

−150 −100 −50 0 50 100 150

−60

−40

−20

020

4060

80

0 2 4 6 8 10 12

Woody BE tC/ha/yr

0 3 6 9 12 15 18 21 24

Grassy BE tC/ha/yr

19

−150 −100 −50 0 50 100 150

−60

−40

−20

020

4060

80

0 2 4 6 8 10 12

Woody BP tC/ha/yr

0 3 6 9 12 15 18 21 24

Grassy BP tC/ha/yr

d)

−150 −100 −50 0 50 100 150

−60

−40

−20

020

4060

80

0 2 4 6 8 10 12

Woody BE tC/ha/yr

0 3 6 9 12 15 18 21 24

Grassy BE tC/ha/yr

e)

−150 −100 −50 0 50 100 150

−60

−40

−20

020

4060

80

0 2 4 6 8 10 12

Woody BE tC/ha/yr

0 3 6 9 12 15 18 21 24

Grassy BE tC/ha/yr

Figure S3: Simulated productivity and spatial distribution of woody and herbaceous BP types in the period2090–2099, for a) TechUp RF, b) TechUp IRR, c) TechUp EFR, d) TechUp WM, e) IrrExp IRR.

20

0 1000 2000 3000 4000

Basic − RF

IrrExp − RF

TechUp − RF

Basic − WM

Basic − EFR

TechUp − WM

TechUp − EFR

TechUp − IRR

Basic − IRR

IrrExp − EFR

IrrExp − WM

IrrExp − IRR

Freshwater Consumption [km^3/yr]

crops

households

livestock

industry

Bioenergy irrigation

goal reached

goal not reached

0

0

0

351

361

378

383

587

642

1368

1553

2239

Figure S4: Yearly (mean 2090–2099) freshwater consumption of bioenergy plantations, agriculture, households,industry and livestock for scenarios targeting a carbon sequestration of 255 GtC.

21