Name o he Deli e able
Ex-an e mul i-c i e ia
analysis o
supply chain
go e nance ini ia i es
a ge ing biodi e si y
p o ec ion conside ing
in e dependencies
Deli e able 7.4
2
Summa y
Wo k Package
7
Deli e able No
7.4
Dissemina ion Le el
Public
Type
R-documen , epo
Lead Pa ne
BASQUE CENTRE FOR CLIMATE CHANGE (BC3)
Due Da e
31s Augus 2025
Submission Da e
8 h Oc obe 2025
S a us
Final
Au ho s
Neus Escoba , Ja ie Ribal, Cla a F ezal, Da id Leclè e
O he con ibu ions
Sibylle Roue -Pollakis, Nelson Ke in Sinis e a-Solís, Neus
Sanjuán
Abou CLEVER
P ojec Numbe
101060765
P ojec Ti le
CLEVER: C ea ing le e age o enhance biodi e si y ou comes o
global biomass ade
Topic
HORIZON-CL6-2021-BIODIV-01-15
S a da e
1s Sep embe 2022
End da e
31s Augus 2025
Coo dina o
RHEINISCHE FRIEDRICH-WILHELMS-UNIVERSITAT BONN
This documen has been p epa ed in he amewo k o he p ojec CLEVER. Views and opinions
exp essed a e howe e hose o he au ho (s) only and do no necessa ily e lec hose o he
Eu opean Union o he Eu opean Resea ch Execu i e Agency. Nei he he Eu opean Union no
he g an ing au ho i y can be held esponsible o hem.
3
This p ojec has ecei ed unding om he Eu opean Resea ch
Execu i e Agency unde HORIZON Resea ch and Inno a ion Ac ions,
g an ag eemen no101060765; and om he UK Resea ch and
Inno a ion (UKRI) unde he UK go e nmen 's Ho izon Eu ope unding
gua an ee [g an numbe 10038491]
4
Con en s
0. EXECUTIVE SUMMARY ................................................................................................... 5
1. BACKGROUND AND SCIENTIFIC CONTRIBUTION ................................................................. 9
2. OBJECTIVES ...................................................................................................................... 11
3. METHODS ........................................................................................................................ 12
3.1. Policy selec ion and scena io de ini ion .................................................................... 12
3.2. Biodi e si y loss indica o s ....................................................................................... 17
3.3. SDG ela ed indica o s .............................................................................................. 18
3.3.1. En i onmen al impac s ........................................................................................................... 18
3.3.2. Socioeconomic impac s .......................................................................................................... 19
3.4. T ade-o analysis and in e dependencies be ween SDG indica o s ........................... 22
4. RESULTS AND DISCUSSION ............................................................................................... 24
4.1. Sus ainabili y ou comes ............................................................................................ 24
4.2. Eco-e iciency a io ................................................................................................... 29
4.3 Quan i ica ion o in e dependencies .......................................................................... 32
4.3.1. Spea man ank co ela ion esul s .......................................................................................... 32
4.3.2. Resul s om he P incipal Componen Analysis (PCA) ........................................................... 35
CONCLUSIONS ..................................................................................................................... 38
PROJECT OUTPUTS ACHIEVED .............................................................................................. 39
REFERENCES ........................................................................................................................ 40
5
0. EXECUTIVE SUMMARY
In esponse o he ala ming decline in biodi e si y linked o human socioeconomic de elopmen ,
CLEVER in es iga es he d i e s, mechanisms, and scale o biodi e si y loss ac oss ecosys ems.
A cen al objec i e is o e alua e how di e en policies and supply chain ini ia i es in luence
global biodi e si y, pa icula ly in ela ion o exis ing land conse a ion schemes and clima e
change mi iga ion policies. Recognizing he ole o in e na ional ade in accele a ing land
con e sion and shi ing impac s om consume s o p oduce s, he p ojec ocuses on h ee key
ag icul u al commodi ies wi h g owing in e na ional demand: ishmeal, imbe , and soy. Wi hin
WP6, he GLOBIOM modelling amewo k was enhanced o, on he one hand, e ine he
ep esen a ion o hese sec o s in he model o be e cap u e ma ke dynamics o de i a i e
p oduc s and in e en ion op ions ac oss ac o s; on he o he hand, o inco po a e high-
esolu ion impac indica o s ha link biodi e si y loss o speci ic d i e s, building on ad ances
om WP2. In WP7, he ex ended GLOBIOM amewo k has been applied o simula e combined
policy scena ios, linked o empi ical wo k in WP4 and WP5, and in o med by expe s akeholde
inpu . Deli e able D7.4 ocuses on he soy supply chains and assesses he speci ic policy
scena ios h ough a mul i-c i e ia lens o iden i y ade-o s be ween biodi e si y p o ec ion and
o he dimensions o sus ainabili y, as well as key in e dependencies. This deli e able combines
he esul s om D7.2 and D7.3 o iden i y ade-o s and ank al e na i e in e en ions a ec ing
he B azilian soy sec o acco ding o hei p ojec ed sus ainabili y pe o mance in 2050. The
esul s p o ide comp ehensi e insigh s in o he isks and bene i s o di e en in e en ions
along in e na ional soy supply chains, wi h a pa icula ocus on he EU and B azil, while also
conside ing global spillo e e ec s based on he concep ual amewo k o D5.2. O e all, hese
indings highligh he mos e ec i e app oaches o p omo ing sus ainabili y in B azil and
wo ldwide.
The scena io analysis explo es a ange o possible u u es o global soy ma ke s and biodi e si y,
including conse a ion e o s in B azil, due diligence policies, bo de adjus men a i s and
ade ag eemen s/dispu es, as well as global o egional die a y shi s and o he demand-side
measu es. The “business-as-usual” (BAU) scena io ep esen s a baseline unde Sha ed
Socioeconomic Pa hway 2 (SSP2) o “middle o he oad”, assuming no majo policy changes, o
assess impac s om cu en socioeconomic ends. The In eg a ed Ac ion Po olio (IAP)
scena io en isions s ong conse a ion and es o a ion measu es aligned wi h he Kunming-
Mon eal Global Biodi e si y F amewo k (KMGBF) goal o e e sing global biodi e si y loss by
2050, combined wi h sus ainable yield inc eases, shi s owa ds plan -based die s and educed
ood was e. The T .Dis. scena io in es iga es he long- e m consequences o he US–China ade
wa by capping China’s impo s o US soy-based p oduc s. The ZNLB a scena io assumes ze o
con e sion o na u al lands o ag icul u e in B azil a e 2020, explo ing he e ec s o s ic land
conse a ion on soy p oduc ion and ade. Se e al scena ios ocus on he impac s o EU policies,
such as he EUDR, which educes B azil’s soy expo s o he EU by 3% compa ed o BAU; and
EUM+EUDR, which ins ead assumes a 15% inc ease due o combined e ec s o he EU-
MERCOSUR ag eemen and he EUDR. Va ian s o his include EUM+EUDR_ZNLB a, whe e
s onge EU measu es igge enhanced conse a ion in B azil, and EUM+EUDR_WeakLUR,
whe e hey coincide wi h weakened land-use egula ion and educed en o cemen . Addi ional
policy expe imen s es al e na i e EU app oaches, such as he EUM+EUBBAM scena io,
eplacing he EUDR wi h a biodi e si y bo de adjus men mechanism ha axes high- isk
impo s based on biodi e si y impac ; and EUM+EUDemSide, whe e demand-side measu es—
6
like die a y shi s and educed ood was e—subs i u e he EUDR o mi iga e en i onmen al
impac s om EU consump ion.
T ade-o s among selec ed en i onmen al and economic indica o s a e i s ly assessed o he
di e en scena ios in ela ion o he BAU. Di e en se s o indica o s a e analysed oge he ,
combining a supply-o ien ed pe spec i e, i.e., o assess ade-o s among impac s gene a ed in
soy sou cing egions, and a demand-o ien ed pe spec i e, i.e., o assess ade-o s among
impac s a he global le el and in key soy consume egions. All scena ios esul in simila
soybean p oduc ion in B azil in 2050, anging be ween 210 and 230 M , wi h li le di e en ial
impac on global ood secu i y (SDG2) o economic g ow h in B azil (SDG8). Only he IAP
scena io, which combines global land conse a ion and es o a ion measu es wi h shi s owa d
plan -based die s, esul s in a ma ked educ ion in p oduc ion o abou 30% ( o 152 M ),
lowe ing he alue o soybean ou pu in B azil and globally, hough wi h minimal changes in
calo ie a ailabili y. Di e ences in soybean p oduc ion in luence o al na u al land con e sion, as
land-use policies de e mine how ag icul u al expansion in e ac s wi h o es and o he na u al
a eas. The Ze o Na u al Land Loss policy in B azil (ZNLB a) has he s onges e ec , d i ing
subs an ial o es and na u al land expansion —abou 67 Mha unde ZNL and
EUM+EUDR_ZNLB a, and a ound 40 Mha unde IAP— compa ed wi h he BAU. IAP addi ionally
esul s in abou 120 Mha o es o ed land in B azil, while o he scena ios main ain es o a ion
le els simila o BAU (a ound 18.5 Mha). O e all, hese esul s highligh how di e en policy
mixes in luence he balance be ween ag icul u al p oduc ion, land-use dynamics, and
en i onmen al sus ainabili y, including biodi e si y loss (SDG15).
Di e ences in en i onmen al impac s be ween scena ios a e mo e p onounced han di e ences
in socioeconomic ou comes. IAP educes o al emissions in B azil by nea ly 5 G CO₂eq h ough
dec eased consump ion o animal p oduc s, educed soy demand, and ag icul u al a ea sa ings.
The e is ne ca bon seques a ion o 4.5 G CO₂eq om land use change (LUC), pa icula ly in
he Ce ado, Ma a A lan ica, and Amazon. EUM+EUDR+ZNLB a and ZNL educe LUC emissions
by 319 M CO2eq (-97%), while he emaining scena ios ha e only mino impac on GHG
emissions (SDG13). IAP also educes eshwa e biodi e si y loss (-127%) and e es ial
biodi e si y loss (-47%), wi h ZNLB a and EUM+EUDR+ZNLB a showing mode a e educ ions
(16% and 32%, espec i ely). T ade dis up ion (T .Dis) sligh ly inc eases eshwa e biodi e si y
loss (+1.7%) due o highe soy p oduc ion and ela ed ag icul u al inpu s. To al i iga ion wa e
demand emains a 43 km3 in all scena ios excep in IAP (45 km3 o 4.6% highe han BAU) due
o an inc ease in he p oduc ion o i iga ed c ops. The EUDR scena ios p oduce ma ginal
changes (<1%) in ou pu a iables ela i e o he BAU, wi h small dec eases in GHG emissions
and biodi e si y impac s unde EUDR alone, bu sligh ly inc eased impac s unde EUM+EUDR
due o highe soy p oduc ion, p ima ily in he Ce ado. O e all, he Amazon biome exhibi s mo e
di e ences ac oss scena ios han o he egions, d i en by land ealloca ion and soy expansion.
Ou comes om he di e en scena ios ha e been combined in o a se o clima e and
biodi e si y Eco-e iciency Ra ios, which measu e he en i onmen al impac s om soy
p oduc ion o he g oss economic alue o soy ou pu . By including an economic me ic
( e enue) in he denomina o , he eco-e iciency a io assesses how e icien ly economic alue
is c ea ed while minimizing en i onmen al impac s, he eby linking en i onmen al pe o mance
di ec ly o economic p oduc i i y. In his way, highe alues indica e g ea e en i onmen al
impac pe USD gene a ed, meaning ha he economic ac i i y is less en i onmen ally e icien .
Lowe alues sugges highe en i onmen al e iciency, as less impac is gene a ed pe USD o
7
e enue. Compa ing hese me ics ac oss scena ios o e a gi en ime ho izon helps assess how
e ec i ely di e en policy mixes p omo e he decoupling o ag icul u al economic ac i i ies
om en i onmen al impac s. This in ol es achie ing en i onmen al p o ec ion a he lowes
possible economic cos by balancing inancial pe o mance wi h sus ainabili y goals (SDG8,
SDG12). I is calcula ed as he a io be ween he absolu e en i onmen al impac s (ei he AFOLU
GHG emissions o o al biodi e si y impac s om a ming) in a gi en scena io and ime ho izon,
ela i e he alue o soy p oduc ion (AbsEIIR), o B azil, he es o he wo ld (ROW), and
globally. The IAP scena io demons a es he mos e icien impac in ensi y pe o mance in
B azil, educing GHG emissions pe USD2000 gene a ed o -0.75 kgCO₂e/USD2000. The ZNLB a
scena ios pe o m mode a ely well (~1.3 kgCO₂e/USD2000), while he o he scena ios a e close
o he BAU (2.65 kgCO₂e/USD2000). The la e is only su passed by EUDR, EUM+EUDemSide and
EUM+EUBBAM, which dec ease soy g oss ou pu in B azil. In e ms o biodi e si y loss, IAP again
pe o ms bes in B azil, wi h a nega i e o al species ichness loss (-5.4E-9
PDF.y/MillionUSD2000) d i en by imp o ed e es ial biodi e si y h ough educed LUC.
Howe e , IAP shows sligh ly highe biodi e si y impac in ensi y in ROW and globally (5.6-5.8E-
8 PDF.y/MillionUSD2000) due o wa e and land s ess in egions such as No h Ame ica o he
es o Sou h Ame ica. Con e sely, IAP exhibi s he highes wa e s ess in ensi y in B azil (0.22
m³/USD2000) compa ed o o he scena ios (0.14–0.15 m³/USD2000), wi h global and ROW
alues a 0.36 m³/USD2000, highligh ing ade-o s be ween na ional e iciency gains and
ex e nalized en i onmen al p essu es.
A composi e Eco-E iciency a io (CsEIIR) has also been p oposed o ank scena ios, conside ing
wo di e en se s o weigh s o he o al biodi e si y loss, GHG emissions and wa e s ess. This
enables he in eg a ion o mul iple dimensions in o a single sco e o easie scena io anking, by
scaling indi idual indica o s o a common basis and applying weigh s ha e lec policy p io i ies,
in his case o biodi e si y conse a ion and en i onmen al p o ec ion. While his app oach
suppo s decision, i s esul s a e sensi i e o no maliza ion choices and weigh ing schemes,
which in oduces a deg ee o subjec i i y and simpli ica ion. When using he same weigh s o
he h ee impac s, IAP s ands ou as he mos eco-e icien scena io o B azil (0.3) bu he wo s
o he ROW (0.7) and globally (0.7). EUM+EUDR+ZNLB a and ZNLB a ha e simila sco es ac oss
egions (0.4), indica ing a good pe o mance and limi ed spillo e o en i onmen al impac s. In
con as , he emaining scena ios show he highes sco e (0.7) o lowes eco-e iciency in B azil,
wi h lowe sco es (a ound 0.3-0.5) o he ROW. When weigh ing only by biodi e si y and GHG
emissions, IAP is he mos eco-e icien acco ding o he B azilian CsEIIR (0), ollowed by
EUM+EUDR+ZNLB a (0.7) and ZNL (0.7). The emaining scena ios gene a e he highes sco es o
B azil (1), while EUM+EUDemSide shows he highes sco e o he ROW (0.7) and globally (0.9).
These igu es acili a e compa ison and enable a clea e classi ica ion o scena ios, al hough hey
omi he wa e s ess dimension, which p o ed less decisi e o e alua ing soy- ela ed policies.
I should be no ed ha o scena ios in ol ing b oad demand-side in e en ions
(EUM+EUDemSide and IAP), hese me ics should no be in e p e ed as spillo e e ec s om
B azil o he ROW, since hei p ima y objec i e is no he mi iga ion o de o es a ion- ela ed
impac s in B azil.
Finally, co ela ions be ween selec ed indica o s ha e been assessed h ough Spea man ank
co ela ion and P incipal Componen Analysis (PCA). The Spea man ank coe icien shows
posi i e co ela ions be ween soybean p oduc ion and en i onmen al impac s in B azil, i.e.,
wa e s ess, e es ial biodi e si y, and o al GHG emissions. F eshwa e biodi e si y loss and
8
e es ial biodi e si y loss a e s ongly co ela ed wi h GHG emissions, since LUC emissions
accoun o a la ge sha e o o al GHG emissions, which also a ec eshwa e biodi e si y
h ough clima e change. F om he demand side, he calo ies consumed in B azil and he soy
p oduc ion in he coun y a e posi i ely co ela ed wi h he o al wo ld GHG emissions and he
global e es ial biodi e si y loss. The esul s also show a s ong posi i e co ela ion be ween
he calo ies consumed as ood in he Eu opean Union (EUE) and Sou he n Asia (SAS) (incl. China),
while nega i e co ela ions a e obse ed be ween hese and he alue o he soy ma ke in
B azil. This indica es ha impo s in hese wo egions espond o inc eases in B azilian soy
p ices, al hough he esul s show ha EUE elies on B azilian soy impo s o a lesse ex en han
SAS o mee hei ood calo ie demand.
This deli e able compa es sus ainabili y indica o s and eco-e iciency o many policies and
in e en ions a ec ing soy ma ke s and supply chains, while p oposing quan i a i e e idence o
ank policy mixes acco ding o hei po en ial o mi iga e impac s and ade-o s be ween
biodi e si y loss (SDG15), clima e change mi iga ion (SDG13), wa e s ess (SDG6), economic
e u ns (SDG8), and ood a ailabili y (SDG2).
9
1. BACKGROUND AND SCIENTIFIC CONTRIBUTION
Socioeconomic de elopmen has led o he p og essi e inc ease o g eenhouse gas (GHG)
emissions, en i onmen al deg ada ion, and biodi e si y loss, especially since he indus ial
e olu ion (In an e-Ama e e al., 2025; Ma ques e al., 2019). The cu en a e o species
ex inc ion is es ima ed o be be ween 100 and 1000 imes as e han i would be wi hou
human in luence, which indica es ha an h opogenic ac i i ies could be igge ing he Six h
Mass Ex inc ion (Ceballos e al., 2015). Land use change (LUC) is he main cause o e es ial
biodi e si y impac s h ough habi a loss and deg ada ion, which signi ican ly impac bo h
species and ecosys ems (IPBES, 2019). This is pa icula ly ue in he opics (Ba low e al., 2018;
Socola e al., 2025), he mos biodi e se egion on he plane , whe e de o es a ion is la gely
d i en by ag icul u al commodi y p oduc ion (Maxwell e al., 2016; Pend ill e al., 2022). A he
same ime, clima e change inc easingly h ea ens ecosys ems unc ioning (Ca dinale e al.,
2012), and can accele a e he eaching o c i ical ipping poin s (Dakos e al., 2019). Land use
and LUC a e no only d i e s o biodi e si y loss, bu also play a cen al ole o many Sus ainable
De elopmen Goals (SDGs), including hose ela ed o ood secu i y, heal h, clean ene gy and
clima e ac ion. Thus, consis en policy-making equi es cap u ing hese in e linkages and
eedback e ec s.
In eg a ed Assessmen Models (IAMs) ha e p o en pa icula ly use ul o iden i y and
acknowledge ade-o s and le e age syne gies in he pu sui o he SDG Agenda and he Pa is
Ag eemen (Popp e al., 2017; Riahi e al., 2017; Rogelj e al., 2018). These models a e a
simpli ied ep esen a ion o he complex in e ac ions be ween human and en i onmen al
sys ems, pa icula ly in he con ex o global en i onmen al change and sus ainable
de elopmen . Beyond simula ing long- e m dynamics, IAMs allow he analysis o exogenous
in e en ions, including land p o ec ion and biodi e si y conse a ion (Leclè e e al., 2020;
Vee kamp e al., 2020). P e ious analyses ha e shown ha al e na i e Pa is-complian policy
measu es, such as a o es a ion, bio uel and bioene gy a ge s, and die a y shi s, can ha e
di e en e ec s on he GHG mi iga ion po en ial om Ag icul u e, Fo es y and O he Land Uses
(AFOLU) (F ank e al., 2019; Humpenöde e al., 2024; Rouhe e e al., 2024). Mos o hem
emphasize po en ial ade-o s h ough ma ke -media ed e ec s, including indi ec LUC and
GHG leakage, al hough only a ew quan i y biodi e si y impac s (Kozicka e al., 2023; Leclè e e
al., 2020; Read e al., 2022). In gene al, he abo e-men ioned s udies conclude ha a
combina ion o supply- and demand-side ini ia i es and policies a e needed o e ec i ely and
simul aneously p og ess owa ds conse a ion and b oade sus ainabili y goals.
CLEVER employs he pa ial equilib ium, ecu si e-dynamic model GLOBIOM, which is he land
use componen o he MESSAGE-GLOBIOM IAM and has a de ailed sub-na ional esolu ion o
simula e LUC ac oss ag icul u al and o es y sec o s, including mul iple en i onmen al
p essu es. The challenge o modelling long- e m biodi e si y impac s in CLEVER is wo old: on
he one hand, models need consis en quan i a i e indica o s ha ep esen he s a e o
biodi e si y h ough ime, and how i changes in esponse o local, egional and global p essu es,
such as LUC o clima e change. These indica o s should be compa ible wi h he empo al and
spa ial esolu ion o IAMs, mos o which a e global in scope. On he o he hand, modelle s need
o iden i y and be able o ep esen , in a s ylized way, he di e si y o policies, s a egies, and
socioeconomic de elopmen s ha po en ially a ec he dis ibu ion o land uses and associa ed
biodi e si y impac s a he global le el.
16
7. EU De o es a ion Regula ion + EU-Me cosu RTA + Ze o na u al land loss in B azil
(EUM+EUDR_ZNLB a) scena io: his scena io explo es he po en ial impac s o new EU
ade- ela ed policies (i.e., EU-Me cosu and EUDR) on B azil-EU soy ade and associa ed
biodi e si y impac s, unde he assump ion ha hey would os e inc eased land
conse a ion e o s in B azil. This is done by simul aneously implemen ing he EUDR+EUM
scena io and he ZNLB a scena io (de eloped in D7.2), which assumes ze o absolu e
con e sion om o es and non- o es na u al land o ag icul u al land in B azil a e 2020
(see D7.2). This scena io assumes ha in e na ional supply chains and he B azilian socie y
eac o he new EU policies by ambi ious supply chain aceabili y and land conse a ion
e o s (well beyond ha o cu en policies, o e en pu sued by he EUDR) in B azil. This
ep esen s he mani es a ion in B azil o a maximalis e sion o a ‘B ussels e ec ’, by which
EU ac ions lead o he adop ion o new, ambi ious s anda ds.
8. EU De o es a ion Regula ion + EU-Me cosu RTA + Weak Land Use Regula ion
(EUM+EUDR_WeakLUR) scena io: in con as o he EUM+EUDR_ZNLB a scena io, his
scena io explo es he po en ial impac o he assump ion ha he new EU ade- ela ed
policies (i.e., EU-Me cosu and EUDR) lead o weake supply chain and land conse a ion
e o s in B azil. This is done by simul aneously implemen ing he EUDR+EUM scena io and
weakening key policies included in SSP2 and all o he scena ios (see de ails on he Fo es
Code ep esen a ion in D6.5). F om 2030 onwa ds, his includes he emo al o he ASM and
a elaxa ion o he Fo es Code. F om 2030 onwa ds, an inc ease in land a ea dedica ed o
soy p oduc ion is allowed in he Amazon biome, and he en o cemen p obabili ies o illegal
de o es a ion con ol as pa o he Fo es Code a e educed by a ac o o 3 in bo h he
Amazon and Ce ado biomes. Howe e , as in all EUDR-based scena ios, i is assumed ha a
su icien sha e o B azil’s soy p oduc ion can comply wi h he EUDR equi emen s, and ha
any soy p oduced om newly de o es ed land in B azil is des ined o o he ma ke s. This
scena io p o ides insigh s in o he implica ions o weakening bo h public and p i a e land
conse a ion e o s in B azil as a po en ial ad e se consequence o unila e al in e en ions
om he EU, like he EUDR.
9. EU-Me cosu RTA + EU biodi e si y bo de adjus men mechanism (EUM+EUBBAM)
scena io: his scena io explo es he impac s o subs i u ing he EUDR by a biodi e si y
bo de adjus men mechanism applied o selec ed high- isk commodi ies. This is done by
implemen ing, om 2020 onwa ds, a ax on he olume o EU impo s o selec ed
commodi ies (soya bean, soya oil, soya cake, palm oil, bee ). The ax le el is p opo ional o
he di e ence be ween he expo ing egion and he EU in e ms o he ma ginal biodi e si y
impac o land occupa ion embedded in he supply o a commodi y. This measu e accoun s
no only o local di ec impac s bu also o emo e indi ec impac s, such as hose a ising
om eed p oduc ion in o he egions o li es ock. The ax a e inc eases o e ime,
eaching by 2050 a alue aligned wi h medium- ange es ima es o he global economic alue
o na u e (Cos anza e al., 2014). Fo soya beans expo ed om B azil o EU, his implies ax
alues o abou 2, 113 and 221 USD2000 pe by 2030, 2040 and 2050, espec i ely. This
scena io p o ides insigh s in o he implica ions o EU mobilizing an al e na i e ins umen
o he EUDR o educing he oo p in o i s impo s, based on a biodi e si y- ocused p ice-
based app oach.
10. EU-Me cosu RTA + EU demand-side measu es (EUM+EUDemSide) scena io: his scena io
explo es he impac s o subs i u ing he EUDR by demand-side sus ainabili y e o s simila
o hose assumed applied globally in he IAP scena io (was e educ ion, shi om animal-
17
based o plan -based p oduc s). This is done by implemen ing in he EU27 a p og essi e
ansi ion om 2030 onwa ds, o eaching by 2050 a 50% educ ion in ood was e and a
subs i u ion o 50% o he demand in animal-based p oduc s (mea and dai y) by plan -
based p oduc s. This scena io p o ides insigh s in o he implica ions o EU ocusing on
demand-side e o s as an al e na i e ins umen o he EUDR o educing he oo p in o
i s impo s.
As pa o an e o o imp o e he ep esen a ion o land use dynamics in B azil, key policy and
p i a e sec o in e en ions including he Fo es Code and he ASM we e pa ame e ized in he
model. As u he de ailed in appendix and in D6.5, he pa ame e iza ion o he Fo es Code
assumes es ic ions on illegal de o es a ion in he Amazon and Ce ado biomes om 2020
onwa ds, adjus ed by en o cemen p obabili ies, oge he wi h es o a ion e o s om 2030
onwa ds modelled as a con e sion o c opland and pas u e in o p o ec ed o es . Fo he ASM,
no expansion in he land a ea dedica ed o soy in he Amazon biome is assumed. These ea u es
apply bo h o he BAU scena io and all he in e en ion scena ios (excep o he
EUM+EUDR+WeakLURscena io whe e he ASM is emo ed om 2030 onwa ds).
3.2. Biodi e si y loss indica o s
This s udy uses ou comes om D7.2 and D7.3 in e ms o species ichness loss as PDF·y based
on LC-IMPACT endpoin CFs. These exp ess he po en ial damage o A eas o P o ec ion (AoPs),
in his case, ecosys em quali y, by dis inguishing se e al en i onmen al mechanisms ha a ec
bo h e es ial and eshwa e biodi e si y. Speci ically, in GLOBIOM, he ollowing
en i onmen al p essu es ha e been conside ed: land s ess, wa e s ess, clima e change, and
eshwa e eu ophica ion (see Table 2). O he d i e s a ailable in LC-IMPACT (i.e.,
pho ochemical ozone o ma ion, e es ial acidi ica ion, eshwa e and e es ial eco oxici y,
ma ine eu ophica ion) could no be linked o GLOBIOM a iables, because he ela ed
en i onmen al lows a e no co e ed in he model. Based on he wo k om D6.4, biodi e si y
loss associa ed wi h en i onmen al pollu ion downs eam he supply chain ( a ming, anspo )
has also been measu ed. Addi ionally, e ined indica o s om D2.4 ha e also been assessed,
which only cap u e species ichness loss associa ed wi h land s ess, as shown in Table 2. The
CFs o species ichness loss associa ed wi h land and wa e s ess and eshwa e
eu ophica ion we e implemen ed a he highes le el o esolu ion a ailable o B azil (ca. 50 x
50 km g id) in GLOBIOM, consis en wi h he de ini ion o Land Uni Iden i ica ion (LUID) uni s
o model ag icul u al supply; and hen agg ega ed a he ele an g anula i y (see Table 2);
whe eas clima e change-d i en CFs a e global in scope. O he socioeconomic indica o s canno
be epo ed a his le el because o a lack o subna ional modelling o key a iables such as
demand, ade, p ocessing, and p ices.
18
Table 2. Biodi e si y loss indica o s conside ed in GLOBIOM and unde lying d i e s.
Impac assessmen
me hod
Impac d i e
Spa ial g anula i y
o CF
Ha monisa ion needed o
GLOBIOM
LC-IMPACT (Ve ones
e al., 2020)
Clima e change
Wo ld
No applicable
Land s ess
Eco egions
based on sha e o eco egion
a ea in LUID
Wa e s ess
0.05° by 0.05°
based on sha e o wa e use
in LUID (CWATM da a)
F eshwa e
Eu ophica ion
F eshwa e
eco egions (FEOW)
based on sha e o FEOW
Fe ilise and manu e
applica ion in LUID (Po e e
al., 2012)
D2.4 (Oli ei a e al.,
2019, 2024)
Land s ess
Eco egions
based on sha e o eco egion
a ea in LUID
3.3. SDG ela ed indica o s
The ollowing indica o s we e measu ed h ough he ex-an e scena io assessmen , which can be
ela ed o speci ic SDG a ge s.
3.3.1. En i onmen al impac s
We ocus on indica o s e lec ing changes in land and wa e use, GHG emissions and biodi e si y
impac s. This leads o ou ypes o indica o s:
• Changes in land co e a eas (SDG15): a eas o land (in million ha) o ou land use
ca ego ies (c opland, pas u e, o es , and o he na u al land). The p ojec ed alues e lec
he ini ial land use and land co e o he yea 2000 and explici ly simula ed con e sions
be ween indi idual land uses a he subna ional le el (e.g., ca. 50 by 50 km in B azil, coa se
in he es o he wo ld – see D6.2) a each decadal ime s ep, accumula ed h oughou
epo ed ime ho izon. Resul s o each ime s ep a ise om ag icul u al p oduc ion changes
in esponse o he simula ed ag icul u al ma ke dynamics and LUC, subna ional land
endowmen s, and echnological changes (e.g., long- e m yield imp o emen s, adop ion o
al e na i e echnologies o i iga ion, e c.).
➔ Since ag icul u al land expansion is he mos impo an d i e o de o es a ion and
na u al co e loss globally (Wes e al., 2025), his indica o is ela ed o SDG15 Li e on
Land, e.g., Ta ge 15.2: By 2020, hal de o es a ion, es o e deg aded o es s and
subs an ially inc ease a o es a ion and e o es a ion globally.
• Changes in wa e use (SDG6): blue wa e wi hd awn o i iga ion (in km3). The p ojec ed
alues e lec he ini ial dis ibu ion o i iga ed ha es ed a ea o he yea 2000 and
ag icul u al p oduc ion changes in ela ion o egional ag icul u al ma ke dynamics,
subna ional su ace wa e , and g oundwa e endowmen s, and echnological changes (e.g.,
long- e m yield imp o emen s, adop ion o al e na i e echnologies o i iga ion, e c.).
19
➔ Globally, ag icul u al i iga ion accoun s o ∼70% o he o al eshwa e wi hd awal
and 80–90% o human wa e consump ion (Hoeks a & Mekonnen, 2012). This indica o
is ela ed o SDG6 Clean Wa e and Sani a ion, e.g., Ta ge 6.4: By 2030, subs an ially
inc ease wa e -use e iciency ac oss all sec o s and ensu e sus ainable wi hd awals and
supply o eshwa e o add ess wa e sca ci y.
• Changes in GHG emissions (SDG13): GHG emissions om ag icul u al ac i i ies (p ima ily
N2O emissions om c opland and pas u es, and CH4 om en e ic e men a ion in uminan
animals, manu e managemen and ice cul i a ion) and LUC (CO2 emissions h ough changes
in abo e-g ound ca bon s ocks). The p ojec ed alues e lec he ini ial dis ibu ion o he
yea 2000 and subsequen dynamics in ag icul u al ac i i ies, as well as c op and li es ock
managemen sys ems ( ha di e in emission in ensi y). Emissions om c op p oduc ion,
li es ock p oduc ion, and LUC a e epo ed as sepa a e ca ego ies.
➔ This indica o is di ec ly ela ed o SDG13 Clima e Ac ion, e.g., Ta ge 13.2: In eg a e
clima e change measu es in o na ional policies, s a egies and planning; Indica o
13.2.2: To al g eenhouse gas emissions pe yea .
• Changes in biodi e si y impac s om local p oduc ion (SDG15): species ichness loss om
ag icul u al ac i i ies in PDF·y. The p ojec ed alues ep esen o biodi e si y impac s on
eshwa e and e es ial ecosys ems a he endpoin le el (Ve ones e al., 2020),
associa ed wi h he simula ed en i onmen al p essu es (clima e change, land occupa ion,
land ans o ma ion, wa e s ess, and eshwa e eu ophica ion – see sec ion 3.2 and
Me hods in D6.3 o u he de ails).
➔ This indica o is di ec ly ela ed o SDG15 Li e on Land, e.g., Ta ge 15.1: Ensu e he
conse a ion, es o a ion and sus ainable use o e es ial and inland eshwa e
ecosys ems and hei se ices; Ta ge 15.5: Take u gen and signi ican ac ion o educe
he deg ada ion o na u al habi a s, hal he loss o biodi e si y and, by 2020, p o ec
and p e en he ex inc ion o h ea ened species.
3.3.2. Socioeconomic impac s
GLOBIOM quan i ies di e en socioeconomic indica o s ha co e a ious supply chain s eps,
om p oduc ion o consump ion, as well as a ious g oups o commodi ies: e.g., p ima y c op
p oduc s wi h a dis inc ion o soy s o he c ops, seconda y p oduc s – e.g., p o ein cakes and
ege able oils – and p ima y li es ock p oduc s al oge he . This deli e able speci ically ocuses
on he ollowing indica o s:
• Changes in ood a ailabili y (SDG2). Food a ailabili y is a consume -o ien ed me ic and one
o he componen s o ood secu i y ( oge he wi h ood access, u iliza ion, and s abili y). I
ep esen s he quan i ies o ood p oduc s a ailable o he consume in a gi en coun y,
including h ough domes ic supply and impo s. I is es ima ed as pe capi a daily a ailabili y
in e ms o ene gy con en (kilocalo ies pe capi a pe day, kcal/c/d), wi h a spli be ween
plan -based and animal-based p oduc s o assess die a y shi s.
➔ This indica o is ela ed o SDG2 Ze o Hunge , e.g., Ta ge 2.1: By 2030, end hunge and
ensu e access by all people, in pa icula he poo and people in ulne able si ua ions.
• Changes in p oduc ion and ne ade physical olumes (SDG2, SDG17). P oduc ion and ne
ade (de ined as he di e ence be ween expo s and impo s) a e ma ke balance
indica o s ele an o bo h p oduc ion (e.g., ag icul u al supply and des ina ion), ade (e.g.,
ne ade balance), and consump ion (e.g., domes ic demand as he sum o p oduc ion and
20
ne ade). They can be indica i e o egions wi h expo dependency o p oduc ion (e.g.,
posi i e ne ade ep esen ing a signi ican sha e o p oduc ion), o impo dependency o
consump ion (e.g., nega i e ne ade ep esen ing a signi ican sha e o consump ion). We
epo hem in physical olumes (e.g., housand ons o esh ma e ), and we dis inguish
li es ock inal p oduc s om c op inal p oduc s. Wi hin c op inal p oduc s, we u he
dis inguish soy, o he c ops, soy p o ein meals, o he cakes, soy oil, and o he ege able oils.
➔ This indica o can be used o measu e p og ess owa ds SDG2 Ze o Hunge , e.g., Ta ge
2.3: By 2030, double he ag icul u al p oduc i i y and incomes o small-scale ood
p oduce s; SDG17 Pa ne ships o he Goals, e.g., Ta ge 17.11: Signi ican ly inc ease
he expo s o de eloping coun ies, in pa icula wi h a iew o doubling he leas
de eloped coun ies’ sha e o global expo s by 2020.
• Changes in he alue o p oduc ion (SDG2, SDG8). This me ic is a p oduc ion-o ien ed
indica o , de ined a he p oduc le el as he domes ic p oduc ion olume mul iplied by he
p oduce p ice o g oss e enue. This igu e ep esen s he o al po en ial income om
selling soybeans a ma ke p ices, bu i does no accoun o cos s. This indic o
complemen s he indica o abo e on physical p oduc ion olumes by in eg a ing he impac
o changes in ma ke p ices ha ela e o he sca ci y o a p oduc as demand and supply
co-e ol e. I is epo ed as p oduc ion alue o a ious p ima y p oduc agg ega es: soy,
o he c ops, and li es ock p oduc s.
➔ This indica o can be used o measu e p og ess owa ds SDG2 Ze o Hunge , e.g., Ta ge
2.3: By 2030, double he ag icul u al p oduc i i y and incomes o small-scale ood
p oduce s; and SDG8 Decen Wo k and Economic G ow h, e.g., Ta ge 8.1: Sus ain pe
capi a economic g ow h in acco dance wi h na ional ci cums ances; Ta ge 8.2: Achie e
highe le els o economic p oduc i i y h ough di e si ica ion, echnological upg ading
and inno a ion.
• Eco-e iciency Ra io (SDG8, SDG12). This indica o was p oposed in D6.4, ep esen ing he
en i onmen al p oduc i i y o a p oduc o se ice in economic e ms. I measu es
En i onmen al Impac In ensi y pe Re enue (EIIR) and can be de i ed om he abo e-
men ioned indica o s. Speci ically, i is calcula ed as he a io o he en i onmen al impac s
om soy p oduc ion o he g oss economic alue o soy ou pu . By including an economic
me ic ( e enue) in he denomina o , he eco-e iciency a io assesses how e icien ly
economic alue is c ea ed while minimizing en i onmen al impac s, he eby linking
en i onmen al pe o mance di ec ly o economic p oduc i i y. In his way, highe alues
indica e g ea e en i onmen al impac pe USD gene a ed, meaning ha he economic
ac i i y is less en i onmen ally e icien . Lowe alues sugges highe en i onmen al
e iciency, as less impac is gene a ed pe USD o e enue.
➔ This indica o can be used o measu e p og ess owa ds SDG8 Decen Wo k and
Economic G ow h, e.g., Ta ge 8.2: Achie e highe le els o economic p oduc i i y
h ough di e si ica ion, echnological upg ading and inno a ion; Ta ge 8.4: Imp o e
p og essi ely global esou ce e iciency in consump ion and p oduc ion and endea ou
o decouple economic g ow h om en i onmen al deg ada ion; and SDG12 Responsible
Consump ion and P oduc ion, e.g., Ta ge 12.2: By 2030, achie e he sus ainable
managemen and e icien use o na u al esou ces.
The EIIR is calcula ed in wo di e en ways, as Absolu e Eco-e iciency o AbsEIIR (a) o as
Composi e Eco-e iciency o all impac s o CsEIIR (b). Whe eas a) di ec ly gi es he speci ic
21
en i onmen al impac pe uni o (g oss) e enue (eq. 1), b) p o ides he no malized eco-
e iciency impac sco e ac oss all impac s (eq. 2). The speci ic de ini ions and in e p e a ion o
he wo indica o s a e as ollows:
• AbsEIIR (Absolu e En i onmen al Impac In ensi y pe Re enue) shows he en i onmen al
cos pe uni o g oss economic ou pu — o example, how much g eenhouse gas, wa e
use, o biodi e si y loss is caused o each uni o e enue. Lowe alues mean be e eco-
e iciency (less impac pe dolla ea ned).
• CsEIIR (Composi e En i onmen al Impac In ensi y pe Re enue) combines se e al
en i onmen al impac s in o one no malized sco e so ha di e en ypes o impac s can be
compa ed on he same scale. The close he sco e is o 0, he mo e eco-e icien (be e );
he close o 1, he less eco-e icien (wo se).
Fo CsEIIR, he AbsEIIR alues o wa e sca ci y, o al GHG emissions, and o al biodi e si y loss
( e es ial plus eshwa e species ichness loss) a e no malized acco ding o eq. 3 (Min–Max
no maliza ion), exp essed ela i e o he maximum alue (as a ac ion o he maximum impac ),
whe e a lowe alue (close o 0) indica es be e pe o mance (highe eco-e iciency), and a
highe alue (close o 1) indica es wo se pe o mance (lowe eco-e iciency). In eco-e iciency
analysis, Min–Max no maliza ion p o ides a simple, in e p e able, and bounded indica o ha
di ec ly suppo s scena io compa ison and policy communica ion.
Finally, he no malized me ics a e used o calcula e a CsEIIR pe scena io (eq. 2), conside ing
wo di e en se s o weigh s, i.e., assuming equal weigh o each o he h ee impac s o
neglec ing wa e sca ci y, since soy is mos ly ain ed and he di e se scena io deli e mino
di e ences (eq. 2). I should be no ed ha , in his case, impac s on biodi e si y a e also caused
by wa e s ess and GHG emissions h ough he chain o causal e ec s, bu hey cap u e
di e en ca ego ies o impac , ei he a he midpoin o a he endpoin le els. In his way,
scena ios wi h a nega i e AbsEIIR ha e lowe CsEIIR (close o 0). While AbsEIIR is use ul o
measu e and compa e eco-e iciency o a single indica o ac oss scena ios, No EIIR enables
compa ison o he o e all sus ainabili y pe o mance o he policy scena ios. Bo h AbsEIIR and
CsEIIR a e calcula ed o B azil, ROW, and globally o unde s and po en ial spillo e e ec s. In
his con ex , spillo e s e e o cases whe e impac s —whe he posi i e o nega i e— sp ead o
o he coun ies beyond B azil as he a ge egion. Fo he scena ios ha in ol e b oad demand-
side in e en ions (EUM+EUDemSide and IAP), hese me ics canno be in e p e ed as spillo e s
in he ROW s B azil, since hei p ima y aim is no he mi iga ion o de o es a ion- ela ed
impac s in B azil.
𝐴𝑏𝑠𝐸𝐼𝐼𝑅𝑖,𝑠 = 𝐼𝑖,𝑠 / 𝑉
𝑠 (eq. 1)
𝐶𝑠𝐸𝐼𝐼𝑅𝑠= ∑(𝑁𝑜𝑟𝐸𝐼𝐼𝑅𝑖,𝑠 × 𝑤𝑖)
𝑖 (eq. 2)
𝑁𝑜𝑟𝐸𝐼𝐼𝑅𝑖,𝑠 = 𝐴𝑏𝑠𝐸𝐼𝐼𝑅𝑖,𝑠 − min(𝐴𝑏𝑠𝐸𝐼𝐼𝑅𝑖,𝑠)
max(𝐴𝑏𝑠𝐸𝐼𝐼𝑅𝑖,𝑠) − min(𝐴𝑏𝑠𝐸𝐼𝐼𝑅𝑖,𝑠) (eq. 3)
Whe e i is he speci ic en i onmen al indica o (GHG emissions, o al biodi e si y loss, wa e
sca ci y); s is he speci ic scena io; I is he impac alue; V he g oss soy p oduc ion alue; and w
is he weigh assigned o each indica o i.
22
3.4. T ade-o analysis and in e dependencies be ween SDG indica o s
To assess in e dependencies among indica o s and quan i y he magni ude o he obse ed
co ela ions, he Spea man ank co ela ion coe icien was es ima ed (Mye s & Si ois, 2005). I
is a non-pa ame ic measu e o ank co ela ion, which assesses how well he ela ionship
be ween wo a iables (in his case, pai s o indica o s) can be desc ibed by a mono onic
unc ion (a ela ionship ha consis en ly inc eases o dec eases, bu no necessa ily linea ly).
The Spea man ank co ela ion coe icien is obus o small sample sizes, skewed dis ibu ions,
o o dinal- ype da a. In his case, we use i o assess co ela ions be ween scena ios in 2050, he
yea when all in e en ions a e ully implemen ed, esul ing in a sample o 50 obse a ions pe
indica o o impac s in B azil (10 scena ios × 5 biomes, excl. Caa inga, wi hou soy p oduc ion)
and 70 obse a ions o global impac s (10 scena ios × 7 egions). P incipal Componen Analysis
(Jolli e & Cadima, 2016) is also applied, which is a s a is ical echnique used o educe he
dimensionali y o a da ase by ans o ming co ela ed a iables in o a smalle numbe o
unco ela ed a iables called p incipal componen s, which cap u e he maximum a iance in he
da a wi h minimal in o ma ion loss. Fu he mo e, ade-o s be ween en i onmen al and
economic pe o mance indica o s we e sys ema ically assessed by no malizing he impac and
alues quan i ying a unique sus ainabili y me ic ac oss scena ios o ul ima ely ank hem
acco ding o hei no malized EIIR (see eq. 1-3).
Di e en combina ions o a iables ha e been conside ed o assess ade-o s and co ela ions
based on GLOBIOM ou comes o B azil and he globe, as indica ed in Table 3.
Table 3. G oups o indica o s selec ed o assessing co ela ions and ade-o s ac oss scena ios o he yea 2050.
Regional
scope
Co ela ions
Indica o s
Desc ip ion
B azil
Be ween soy ou pu
and en i onmen al
impac s
Soy p oduc ion (1000 )
Soy ou pu in B azil
GHG emissions
(M CO2eq/y )
GHG emissions om LUC, o al c op
and li es ock p oduc ion ac i i ies
ac oss biomes
Wa e consump ion (km3)
I iga ion wa e consumed in c op
p oduc ion ac oss biomes
Te es ial biodi e si y
loss wi h LC Impac
(PDF.yea )
Impac s on e es ial species ichness
based on LC Impac CFs, d i en by
bo h land occupa ion and land
ans o ma ion ac oss biomes
F eshwa e biodi e si y
loss wi h LC Impac
(PDF.yea )
Impac s on eshwa e species
ichness based on LC Impac CFs,
d i en by clima e change, wa e s ess
and eshwa e eu ophica ion
Te es ial biodi e si y
loss wi h CLEVER CFs om
D6.3 (PDF.yea )
Impac s on e es ial species ichness
based on CLEVER CFs, d i en by bo h
land occupa ion and land
ans o ma ion ac oss biomes
23
B azil
Be ween soy a eas,
land co e changes,
LUC emissions and
e es ial biodi e si y
Soy a ea (1000 ha)
Soy a eas ac oss biomes
Pas u e a ea (1000 ha)
Pas u e a eas ac oss biomes
Fo es a ea (1000 ha)
Fo es land a eas ac oss biomes
Na u al land a ea (1000
ha)
O he na u al land a eas ac oss
biomes
Res o ed a ea (1000 ha)
Res o ed a eas ac oss biomes
LUC emissions
(M CO2eq/y )
Ne GHG emissions om LUC ac oss
biomes
Te es ial biodi e si y
loss wi h LC Impac
(PDF.yea )
Impac s on e es ial species ichness
based on LC Impac CFs, d i en by
bo h land occupa ion and land
ans o ma ion ac oss biomes
Wo ld
Be ween soy
p oduc ion, ma ke
alue, calo ie
a ailabili y in B azil,
EUE and SAS,
biodi e si y loss and
GHG emissions
Soy p oduc ion (1000 )
Ou pu o soy in B azil, in physical
quan i ies
Soy p oduc ion alue
(million USD, cons an
2000 p ices)
Ou pu o soy in B azil, in mone a y
alue
Food calo ie a ailabili y in
B azil (kcal/cap/d)
Calo ies associa ed wi h he
consump ion o plan - and animal-
based p oduc s in coun y’s ood
demand
Food calo ie a ailabili y in
Eu opean Union (EUE)
(kcal/cap/d)
Calo ies associa ed wi h he
consump ion o plan - and animal-
based p oduc s in coun ies’ ood
demand
Food calo ie a ailabili y in
Sou he n Asia (SAS), incl.
China (kcal/cap/d)
Calo ies associa ed wi h he
consump ion o plan - and animal-
based p oduc s in coun ies’ ood
demand
Te es ial biodi e si y
loss wi h LC Impac
(PDF.yea )
Impac s on global e es ial species
ichness based on LC Impac CFs,
d i en by bo h land occupa ion and
land ans o ma ion ac oss wo ld
egions, incl. B azil
F eshwa e biodi e si y
loss wi h LC Impac
(PDF.yea )
Impac s on global eshwa e species
ichness based on LC Impac CFs,
d i en by clima e change, wa e s ess
and eshwa e eu ophica ion ac oss
wo ld egions, incl. B azil
GHG emissions
(M CO2eq/y )
GHG emissions om LUC, o al c op
and li es ock p oduc ion ac i i ies
ac oss wo ld egions, incl. B azil
24
4. RESULTS AND DISCUSSION
4.1. Sus ainabili y ou comes
Socioeconomic and en i onmen al sus ainabili y ou comes om he scena ios in Table 1 a e
i s ly ep esen ed as ela i e changes o he BAU in 2050, since his is he las yea o he pe iod
when all policies and cons ain s a e implemen ed. E o ! Re e ence sou ce no ound. shows
esul s o he selec ed indica o s o B azil as a whole and ac oss unde lying biomes, o
unde s and ade-o s om a supply-o ien ed pe spec i e, i.e., among impac s gene a ed in soy
sou cing egions. Figu e 2 shows ade-o s among selec ed indica o s o he globe and key
speci ic egions, om a demand-o ien ed pe spec i e. As can be seen in E o ! Re e ence sou ce
no ound., all he scena ios lead o simila soybean p oduc ion ou pu in B azil, be ween 210
and 230 M in 2025 (Fig. 1a). Only IAP causes a signi ican educ ion, by a ound 30% (152 M ).
EUM+EUDR and EUM+EUDR_WeakLUR sligh ly os e soybean p oduc ion (up o 3%), while
EUDR, EUM+EUBBAM, EUM+EUDR_ZNLB a and EUM+EUDemSide cause ma ginal dec eases in
p oduc ion (up o 3.1%). This has implica ions o o al na u al land con e sion (as he sum o
o es and o he na u al land), since soy p oduc ion in e ac s wi h o he land uses, go e ned by
he di e en policies. As expec ed, he ZNL policy in B azil plays a big ole ac oss ZNLB a,
EUM+EUDR+ZNLB a, and IAP scena ios, leading o expansion in o es and o he na u al land
a eas by 67 Mha in B azil (40 Mha in IAP) ela i e o he BAU, up o a o al o 248 Mha in 2050.
The o he scena ios ha e ba ely any e ec on his indica o . IAP leads o addi ional +120 Mha o
es o ed land in B azil, while he o he scena ios ha e same es o ed a eas as BAU (18.5 Mha in
2050).
As o AFOLU emissions esul ing om he sum o GHG emissions om LUC, c op and li es ock
p oduc ion, he IAP scena io causes a dec ease in o al GHG emissions ela i e o BAU o 3.5 G
CO2eq globally and 5 G in B azil, wi h ne GHG sa ings o 4.2 G CO2eq in he la e . As indica ed
in D7.2, his is mainly h ough he educ ion in ood consump ion o animal p oduc s, which
leads o a sha p dec ease in li es ock p oduc ion as well as in he demand o soy cake and
associa ed soy p oduc ion in IAP. This lowe s GHG emissions om LUC and li es ock ac oss
egions, especially in B azil, whe e he e is ne ca bon seques a ion (4.5 G CO2eq) om LUC
h ough educ ion in pas u eland a eas and na u al a eas clea ing, also as a esul o he
inc eased land conse a ion and es o a ion e o s. These GHG sa ings om LUC occu mainly
in Ce ado (-1.4 G ), Ma a A lan ica (-1.3 G ) and he Amazon (-1.1 G ). In he BAU, LUC emissions
each 328 M CO2eq in 2050, mainly in he Amazon (271 M ) (Fig. 1b). The ZNLB a and
EUM+EUDR+ZNLB a scena ios cause a 97% emissions educ ion in B azil, while he o he
scena ios ha e only ma ginal e ec s in e ms o AFOLU emissions in he coun y – anging om
-2.2% in EUM+EUDR+WeakLUR o +0.42% in EUM+EUBBAM.
As a d i e o eshwa e biodi e si y loss (F eshW.BD) acco ding he LC Impac me hod,
a o emen ioned dec eases in o al GHG emissions in IAP lead o a no able dec ease in species
loss ela i e o BAU (-127%). ZNLB a, and EUM+EUDR+ZNLB a dec ease F eshW.BD loss by abou
16%, due o limi ed na u al land loss and associa ed GHG emissions. Whe eas he impac s o
EUDR alone a e mino , ade dis up ion (T .Dis.) inc eases F eshW.BD loss ela i e o BAU by
1.7%, due o he o e all ne inc ease in soy p oduc ion in B azil (4.85% o +10.7 M ), which
en ails g ea e ag icul u al inpu s consump ion, leading o highe eu ophica ion and wa e
s ess. EUM+EUDR_WeakLUR, EUM+EUDemSide and EUM+EUBBAM dec ease F eshW.BD
25
impac s in B azil by 3.3%, 1.7% and 0.7%, espec i ely, mainly due o lowe eshwa e
ecosys ems impac s in Ce ado and Ma a A lan ica.
Figu e 1. Spide cha s o sus ainabili y ou comes in B azil and he soy p oducing biomes. Resul s show ela i e
changes (%) o selec ed indica o s ac oss scena ios in 2050, ela i e o he Business-as-Usual (BAU) scena io.
32
beha iou wi h IAP and T .Dis. showing sligh ly lowe CsEIIRs (0.5). EUM+EUDemSide shows he
highes sco e o he wo ld (0.9), mainly due o eshwa e biodi e si y loss in ROWexEUE. These
la es igu es acili a e compa ison and enable a clea e classi ica ion o scena ios, al hough a
he cos o losing he dimension o wa e s ess, less decisi e in compa ing scena ios o soy-
ela ed policies.
Table 6. Composi e En i onmen al Impac In ensi y a io (CsEIIR) ac oss scena ios (dimensionless), conside ing
wo weigh ing schemes: same weigh o he h ee impac s, o same weigh o o al biodi e si y loss and GHG
emissions.
CsEIIR_BRA
w1
CsEIIR_ROW
w1
CsEIIR_WORLD
w1
CsEIIR_BRA
w2
CsEIIR_ROW
w2
CsEIIR_WORLD
w2
IAP
0.3
0.7
0.7
0.0
0.5
0.5
EUM+EUDR+
ZNLB a
0.4
0.4
0.4
0.7
0.6
0.5
ZNLB a
0.4
0.4
0.3
0.7
0.6
0.5
EUM+EUDR+
WeakLUR
0.7
0.4
0.4
1.0
0.6
0.7
T .Dis.
0.7
0.3
0.3
1.0
0.5
0.5
EUM+EUDR
0.7
0.4
0.4
1.0
0.6
0.7
BAU
0.7
0.4
0.5
1.0
0.6
0.7
EUDR
0.7
0.4
0.5
1.0
0.6
0.7
EUM+EUDemSide
0.7
0.5
0.6
1.0
0.7
0.9
EUM+EUBBAM
0.7
0.4
0.5
1.0
0.6
0.7
4.3 Quan i ica ion o in e dependencies
4.3.1. Spea man ank co ela ion esul s
The Spea man ank coe icien cap u es how well one a iable inc eases/dec eases as he o he
inc eases, ega dless o linea i y. The esul s in Figu e 2 con i m he posi i e co ela ion be ween
he selec ed ou pu a iables ac oss he eigh scena ios (see Table 3). All impac s a e posi i ely
co ela ed wi h he soybean p oduc ion (in onnes) in B azil, especially wa e s ess (0.77),
e es ial biodi e si y wi h LC Impac CFs (Te .BD, 0.55) and o al GHG emissions (0.54). As
expec ed, F eshW.BD is s ongly co ela ed wi h GHG emissions (0.80), and he wo Te .BD
indica o s (CLEVER s. LC Impac ) a e s ongly co ela ed (0.81), as bo h a e de e mined by he
same land use and LUC e ec s. F eshW.BD is co ela ed wi h Te .BD since he la e is pa ly
caused by LUC, which ep esen s a la ge sha e o GHG emissions ha also cause F eshW.BD
h ough clima e change.
33
Figu e 2. Spea man ank co ela ion index o en i onmen al ou pu indica o s o B azil.
Figu e 3. Spea man ank co ela ion index o land use a eas and en i onmen al indica o s o B azil.
A mo e in-dep h analysis o he co ela ions be ween land use a eas, LUC- ela ed GHG
emissions, and e es ial biodi e si y loss in Figu e 3 shows bo h nega i e and posi i e
co ela ions. This could be expec ed since some uses expand a he cos o o he in he di e en
scena ios. Fo ins ance, i is obse ed ha an inc ease in he o es land a ea in B azil weakly
co ela es wi h bo h o he na u al land (0.32) and es o ed a eas (0.29) in a posi i e ashion, bu
34
shows a s ong nega i e co ela ion be ween soy (-0.88) and o he c op a eas (-0.90), and a
ela i ely weake nega i e co ela ion wi h Te .BD impac s (-0.85) and LUC emissions (-0.61).
O he na u al land a eas only show a signi ican co ela ion wi h es o ed a eas (-0.52). Since,
by de ini ion, es o ed a eas in he model expand a he expense o ag icul u al and pas u e
lands, his ou come can be explained by wo mechanisms: (a) in some cases, he ag icul u al
land being es o ed migh ha e o he wise been abandoned, and (b) in o he cases, con e ing
ag icul u al land back o o es may indi ec ly d i e he con e sion o o he na u al a eas in o
new ag icul u al land. Finally, Figu e 3 indica es ha e es ial biodi e si y loss is s ongly and
posi i ely co ela ed wi h he a ea o o he c ops (0.87) and soy (0.82), whe eas LUC emissions
show a highe co ela ion wi h he a ea o o he c ops (0.72) han wi h soy (0.59).
The same analysis is done o a iables ela ed o global consump ion o soy (Figu e 4). As
expec ed, he calo ies consumed in B azil a e s ongly co ela ed wi h he soy p oduc ion in he
coun y (0.83), and bo h a e posi i ely co ela ed wi h he o al wo ld GHG emissions and he
global Te .BD loss. The esul s also show a s ong posi i e co ela ion be ween he calo ies
consumed as ood in EUE and SAS (0.66), while hese a e only weakly co ela ed wi h he global
Te .BD loss. Nega i e co ela ions a e obse ed be ween he calo ies consumed in SAS and EUE,
and he alue o he soy ma ke in B azil (-0.81 and -0.36, espec i ely), which indica es ha
impo s in hese wo egions espond o inc eases in B azilian soy p ices. The nega i e
co ela ion coe icien is howe e lowe (highe nega i e co ela ion) be ween he calo ies
consumed in EUE and he soy ma ke alue in he ROW (-0.88), which indica es ha EUE elies
less on B azilian soy impo s han SAS o mee hei ood calo ie demand. Acco dingly, calo ie
a ailabili y in SAS is nega i ely co ela ed wi h p oduc ion o soy in B azil (-0.53, s. -0.27 o
EUR). O he signi ican posi i e co ela ions exis , showing ha inc eased soy p oduc ion in
B azil is ela ed o inc eased GHG emissions (0.37) and Te .BD impac s (0.48) globally, while
F eshW.BD is nega i ely co ela ed wi h B azilian soy p oduc ion (-0.32). Global GHG emissions
a e posi i ely co ela ed wi h global Te .BD (0.83).
Figu e 4. Spea man ank co ela ion index o selec ed en i onmen al and economic indica o s o key wo ld
egions and he wo ld. LCFE is he eshwa e biodi e si y impac and LCTE is he e es ial biodi e si y impac
wi h LC Impac me hod. ROW: es o he wo ld (excep B azil).
35
4.3.2. Resul s om he P incipal Componen Analysis (PCA)
PCA educes he dimensionali y o he da a by summa izing he main co ela ion pa e ns in o
a ew p incipal componen s (PCs), iden i ying he axes ha bes explain he o e all a iabili y in
he esul s. Fo he analysis o he impac s in B azil, disagg ega ed pe biomes, he sc ee plo
e eals ha wo axes explain mo e han 71% o he a iabili y. The PCA in Figu e 5 displays he
i s p incipal componen (PC1) in x-axis and second p incipal componen (PC2) in he y-axis,
whe e each poin ep esen s one o he 50 alues (10 scena ios x 5 biomes). We obse e s ong
posi i e co ela ions be ween GHG emissions (EMIS), Te .BD wi h CLEVER CFs, and Te .BD wi h
LC IMPACT CFs. In he PCA, wa e s ess, soy p oduc ion, and F eshW.BD load posi i ely on PC1
bu nega i ely on PC2, g ouping oge he in he bo om- igh quad an , which means hey end
o inc ease oge he . In he op- igh quad an we ind he biomes in which he scena ios lead
o high GHG emissions and Te .BD impac s, mainly he Amazon. In he bo om- igh , we ind
scena ios-biome combina ions ha lead o high soy p oduc ion and high-wa e s ess and
F eshW.BD, mainly in Ce ado and Ma a A lan ica biomes. Figu e 5 shows ha he Amazon,
Pan anal and Pampa a e simila ly a ec ed in e ms o o al impac s and co ela ions, as a e he
Ce ado and Ma a A lan ica.
Figu e 5. P incipal Componen Analysis (PCA) biplo showing he ela ionships among impac s ac oss B azilian
biomes and hei con ibu ions o he i s wo p incipal componen s (PC1 and PC2). A ows indica e he di ec ion
and s eng h o a iable loadings, while poin s ep esen indi idual obse a ions o he biomes p ojec ed in o
he educed dimensional space.
PCA esul s in Figu e 6 show a complemen a y poin o iew, wi hou assessing co ela ions by
biome, only by scena io. A he B azilian le el, he e is a s ong co ela ion be ween GHG
emissions, soy p oduc ion, and F eshW.BD, while he e is also a posi i e co ela ion be ween
Te .BD wi h CLEVER CFs and Te .BD wi h LC Impac CFs. The scena ios beha e e y simila ly,
excep o IAP and EUM+EUDR+ZNLB a. As indica ed abo e, IAP inc eases wa e s ess and
36
dec eases GHG emissions, while he o he scena ios show Te .BD impac s simila o BAU (SSP2)
wi h he wo se s o CFs. F om he demand side, Figu e 7 shows s ong co ela ions be ween he
p oduc ion o soy in B azil, calo ies consumed in B azil, he global Te .BD impac s and he alue
o he soy ma ke in ROW, while wo ld GHG emissions a e nega i ely co ela ed. Mo eo e ,
he e a e s ong co ela ions be ween he calo ies consumed in EUE, SAS and he ma ke alue
o B azilian soy, as indica ed abo e.
Finally, he assessmen o co ela ions be ween biomes and scena ios shows ha he biome is
a much mo e decisi e ac o in de e mining he di e en impac me ics o B azil han he
scena io i sel . The obse a ions o he scena ios appea a he g ouped in he biplo , while he
biomes appea away om each o he , excep o Amazon, Pampa, and Ma a A lan ica (Figu e
8). Only IAP one he one hand, and EUM+EUDR+ZNLB a and ZNLB a on he o he , show a
di e en beha iou , as highligh ed h ough he esul s sec ion.
Figu e 6. P incipal Componen Analysis (PCA) biplo showing he ela ionships among impac s in B azil as a whole
and hei con ibu ions o he i s wo p incipal componen s (PC1 and PC2). A ows indica e he di ec ion and
s eng h o a iable loadings, while poin s ep esen indi idual obse a ions o he scena ios p ojec ed in o he
educed dimensional space. SSP2 e e s o BAU.
37
Figu e 7. P incipal Componen Analysis (PCA) biplo showing he ela ionships among impac s in wo ld egions
and hei con ibu ions o he i s wo p incipal componen s (PC1 and PC2). A ows indica e he di ec ion and
s eng h o a iable loadings, while poin s ep esen indi idual obse a ions o he scena ios p ojec ed in o he
educed dimensional space. SSP2 e e s o BAU.
Figu e 8. P incipal Componen Analysis (PCA) biplo showing he ela ionships among scena ios and biomes
and hei con ibu ions o he i s wo p incipal componen s (PC1 and PC2). Poin s ep esen indi idual
obse a ions o he scena ios-biomes p ojec ed in o he educed dimensional space. SSP2 e e s o BAU.
38
CONCLUSIONS
This deli e able e alua es sus ainabili y and eco-e iciency o policy in e en ions in B azil’s soy
sec o , using quan i a i e da a o ank policy combina ions – based on e idence and expe
opinions – acco ding o hei po en ial o add ess biodi e si y loss, clima e change, wa e s ess,
economic e u ns, and ood a ailabili y. D7.4 combines esul s om D7.2 and D7.3 o assess
2050 scena ios h ough a mul i-c i e ia lens, highligh ing ade-o s, in e dependencies, and he
mos e ec i e s a egies o p omo ing sus ainabili y in B azil and beyond. The assessmen o
en i onmen al and economic indica o s ac oss policy mix scena ios in ela ion o a BAU scena io
shows ade-o s among impac s gene a ed in soy sou cing egions, as well as among impac s a
he global le el and in key soy consume egions.
While mos scena ios main ain soybean ou pu in B azil be ween 210–230 M in 2025, IAP
educes p oduc ion by 30% (152 M ) h ough conse a ion and es o a ion measu es and
die a y shi s, leading o majo bene i s in e ms o GHG emissions and biodi e si y (SDG13,
SDG15). IAP educes GHG emissions and gene a es ne ca bon seques a ion o 4.5 G CO₂eq,
mainly in he Ce ado, Ma a A lan ica, and Amazon. I also deli e s signi ican biodi e si y gains,
wi h eshwa e and e es ial species losses educed by 127% and 47% espec i ely, bu 4.6%
highe wa e demand in ag icul u e han BAU. ZNLB a and EUM+EUDR+ZNLB a scena ios deli e
mo e mode a e imp o emen s ela i e o BAU, while T .Dis. inc eases eshwa e biodi e si y
loss sligh ly.
F om an eco-e iciency pe spec i e, IAP again s ands ou , showing nega i e GHG emissions pe
USD2000 gene a ed in B azil and he s onges pe o mance in biodi e si y impac in ensi y,
hough a he cos o highe wa e s ess. ZNL and EUM+EUDR+ZNLB a o e in e media e
imp o emen s wi h limi ed spillo e s, while o he scena ios pe o m close o BAU. Composi e
eco-e iciency a ios con i m IAP as he mos e icien scena io o B azil, while o he es o
he wo ld, ou comes a e mo e balanced, wi h EUM+EUDR+ZNLB a and ZNLB a also pe o ming
well. Impo an ly, he weigh ing o indica o s o he composi e indica o a ec s ankings: when
biodi e si y and GHG emissions a e emphasized, all scena ios con e ge o simila sco es, wi h
IAP emaining he bes op ion o B azil. In e es ingly, EUM+EUDemSide and EUM+EUBBAM
ha e he wo s eco-e iciency in he ROW due o ebound e ec s d i en by p ice changes in soy
and soy cake ma ke s. T ade-o s be ween impac s and biomes mus be ca e ully conside ed in
designing e ec i e soy- ela ed policies, wi h he Amazon being he egion mos sensi i e o he
di e en policy combina ions e alua ed and unde lying LUC.
The Spea man ank coe icien shows posi i e co ela ions be ween soybean p oduc ion and
en i onmen al impac s in B azil. F eshwa e and e es ial biodi e si y loss a e s ongly
co ela ed wi h GHG emissions, gi en he la ge con ibu ion ha LUC makes o o al GHG
emissions, which also a ec eshwa e biodi e si y h ough clima e change. F om he demand
side, he calo ies consumed in B azil and he soy p oduc ion in he coun y a e posi i ely
co ela ed wi h he o al wo ld GHG emissions and he global e es ial biodi e si y loss. The
esul s also show a s ong posi i e co ela ion be ween he calo ies consumed as ood in EUE
and SAS, while nega i e co ela ions a e obse ed be ween hese and he alue o he soy
ma ke in B azil. This indica es ha impo s in hese wo egions espond o inc eases in B azilian
soy p ices, al hough he esul s show ha EUE elies less on B azilian soy han SAS o mee hei
ood calo ie demand.
39
Absolu e eco-e iciency a ios, such as kilog ams o CO₂ pe uni o economic ou pu , p o ide a
anspa en and benchma kable measu e o en i onmen al in ensi y ha can be compa ed
ac oss ime, egions, and sec o s. Howe e , hey ea each impac dimension sepa a ely,
making i di icul o e alua e ade-o s among indica o s like GHG emissions, biodi e si y loss,
and wa e s ess. A no malized and weigh ed composi e indica o add esses his by scaling
indi idual indica o s o a common basis and applying weigh s ha e lec policy p io i ies,
enabling he in eg a ion o mul iple dimensions in o a single sco e o easie scena io anking.
While his app oach suppo s sys ema ic decision-making and highligh s ade-o s, i s esul s
a e sensi i e o no maliza ion choices and weigh ing schemes, which in oduces a deg ee o
subjec i i y. Compa ing hese me ics ac oss scena ios o e a gi en ime ho izon helps assess
how e ec i ely di e en policy mixes p omo e he decoupling o ag icul u al economic ac i i ies
om en i onmen al impac s.
PROJECT OUTPUTS ACHIEVED
• Quan i ica ion o syne gies and ade-o s be ween biodi e si y p o ec ion and clima e
mi iga ion o he go e nance scena ios ou lined in D7.3.
• Analysis o ade-o s and co ela ions be ween en i onmen al and economic indica o s
de ined o he soy sec o in WP6
• Ranking o policy mix scena ios acco ding o hei no malized eco-e iciency, conside ing
spillo e e ec s and ade-o s
40
REFERENCES
Alcamo, J. (2008). Chap e Six The SAS App oach: Combining Quali a i e and Quan i a i e
Knowledge in En i onmen al Scena ios. In De elopmen s in In eg a ed En i onmen al
Assessmen (Vol. 2, pp. 123–150). Else ie . h ps://doi.o g/10.1016/S1574-
101X(08)00406-7
Ba low, J., F ança, F., Ga dne , T. A., Hicks, C. C., Lennox, G. D., Be engue , E., Cas ello, L.,
Economo, E. P., Fe ei a, J., Guéna d, B., Gon ijo Leal, C., Isaac, V., Lees, A. C., Pa , C.
L., Wilson, S. K., Young, P. J., & G aham, N. A. J. (2018). The u u e o hype di e se
opical ecosys ems. Na u e, 559(7715), 517–526. h ps://doi.o g/10.1038/s41586-
018-0301-1
Bö ne , J., Schulz, D., Wunde , S., & P a , A. (2020). The E ec i eness o Fo es Conse a ion
Policies and P og ams. Annual Re iew o Resou ce Economics, 12(1), 45–64.
h ps://doi.o g/10.1146/annu e - esou ce-110119-025703
Ca dinale, B. J., Du y, J. E., Gonzalez, A., Hoope , D. U., Pe ings, C., Venail, P., Na wani, A.,
Mace, G. M., Tilman, D., Wa dle, D. A., Kinzig, A. P., Daily, G. C., Lo eau, M., G ace, J.
B., La igaude ie, A., S i as a a, D. S., & Naeem, S. (2012). Biodi e si y loss and i s
impac on humani y. Na u e, 486(7401), 59–67. h ps://doi.o g/10.1038/na u e11148
Ceballos, G., Eh lich, P. R., Ba nosky, A. D., Ga cía, A., P ingle, R. M., & Palme , T. M. (2015).
Accele a ed mode n human–induced species losses: En e ing he six h mass
ex inc ion. Science Ad ances, 1(5), e1400253. h ps://doi.o g/10.1126/sciad .1400253
Chaudha y, A., & B ooks, T. M. (2018). Land Use In ensi y-Speci ic Global Cha ac e iza ion
Fac o s o Assess P oduc Biodi e si y Foo p in s. En i onmen al Science & Technology,
52(9), 5094–5104. h ps://doi.o g/10.1021/acs.es .7b05570
41
Chaudha y, A., & Kas ne , T. (2016). Land use biodi e si y impac s embodied in in e na ional
ood ade. Global En i onmen al Change, 38, 195–204.
h ps://doi.o g/10.1016/j.gloen cha.2016.03.013
Cos anza, R., De G oo , R., Su on, P., Van De Ploeg, S., Ande son, S. J., Kubiszewski, I., Fa be ,
S., & Tu ne , R. K. (2014). Changes in he global alue o ecosys em se ices. Global
En i onmen al Change, 26, 152–158. h ps://doi.o g/10.1016/j.gloen cha.2014.04.002
Dakos, V., Ma hews, B., Hend y, A. P., Le ine, J., Loeuille, N., No be g, J., Nosil, P., Sche e ,
M., & De Mees e , L. (2019). Ecosys em ipping poin s in an e ol ing wo ld. Na u e
Ecology & E olu ion, 3(3), 355–362. h ps://doi.o g/10.1038/s41559-019-0797-2
Dockendo , C., Fuss, S., Ag a, R., Guye, V., He e a, D., & K axne , F. (2022). Commi ed o
es o ing opical o es s: An o e iew o B azil’s and Indonesia’s es o a ion a ge s
and policies. En i onmen al Resea ch Le e s, 17(9), 093002.
h ps://doi.o g/10.1088/1748-9326/ac8ab2
Escoba , N., Tizado, E. J., zu E mgassen, E. K. H. J., Lö g en, P., Bö ne , J., & Goda , J. (2020).
Spa ially-explici oo p in s o ag icul u al commodi ies: Mapping ca bon emissions
embodied in B azil’s soy expo s. Global En i onmen al Change, 62, 102067.
h ps://doi.o g/10.1016/j.gloen cha.2020.102067
F ank, S., Ha lík, P., S eh es , E., Van Meijl, H., Wi zke, P., Pé ez-Domínguez, I., Van Dijk, M.,
Doelman, J. C., Fellmann, T., Koopman, J. F. L., Tabeau, A., & Valin, H. (2019).
Ag icul u al non-CO2 emission educ ion po en ial in he con ex o he 1.5 °C a ge .
Na u e Clima e Change, 9(1), 66–72. h ps://doi.o g/10.1038/s41558-018-0358-8
