www.ecog aphy.o g
ECOGRAPHY
Ecog aphy
1311
––––––––––––––––––––––––––––––––––––––––
© 2021 The Au ho s. Ecog aphy published by John Wiley & Sons L d on behal o No dic Socie y Oikos
This is an open access a icle unde he e ms o he C ea i e Commons
A ibu ion License, which pe mi s use, dis ibu ion and ep oduc ion in any
medium, p o ided he o iginal wo k is p ope ly ci ed.
Subjec Edi o : Flá ia Regina
Capello o Cos a
Edi o -in-Chie : Miguel A aújo
Accep ed 17 May 2021
44: 1311–1324, 2021
doi: 10.1111/ecog.05598
44 1311–1324
Fo es unde s o ies play a i al ole in ecosys em unc ioning and he p o ision o
ecosys em se ices. Howe e , he ex en o which en i onmen al condi ions d i e
dominan ecological s a egies in o es unde s o ies a he con inen al scale emains
unde s udied. He e, we used ~29 500 o es ege a ion plo s sampled ac oss Eu ope
and classi ied in o 25 o es ypes o explo e he ela i e ole o mac oclima e, soil pH
and ee canopy co e in d i ing abundance-weigh ed pa e ns in he lea economic
spec um (LES) and plan size spec um (PSS) o o es unde s o ies (sh ub and he b
laye s). We calcula ed LES using speci ic lea a ea (SLA) and lea d y ma e con en
(LDMC) and PSS using plan heigh and seed mass o ascula plan species ound
in he unde s o ies. We ound ha o es unde s o ies had mo e conse a i e lea eco-
nomics in a eas wi h mo e ex eme mean annual empe a u es (mainly Fennoscandia
and he Medi e anean Basin), mo e ex eme soil pH and unde mo e open canopies.
Wa m and summe -d y egions a ound he Medi e anean Basin and a eas o A lan ic
Eu ope also had alle unde s o ies wi h hea ie seeds han con inen al empe a e o
bo eal a eas. Unde s o ies o b oadlea ed deciduous o es s, such as Fagus o es s on
non-acid soils, o a ine o es s, mo e commonly hos ed species wi h acquisi i e lea
economics. In con as , some coni e ous o es s, such as Pinus, La ix and Picea mi e
o es s, o Pinus syl es is ligh aiga and scle ophyllous o es s, mo e commonly hos ed
species wi h conse a i e lea economics. Ou indings highligh he impo ance o
mac oclima e and soil ac o s in d i ing ai a ia ion o unde s o y communi ies a
he con inen al scale and he media o e ec o canopy co e on hese ela ionships.
We also p o ide he i s maps and analyses o LES and PSS o o es unde s o ies
ac oss Eu ope and gi e e idence ha he unde s o ies o Eu opean o es ypes a e di -
e en ly posi ioned along majo axes o ai a ia ion.
The lea economic and plan size spec a o Eu opean o es
unde s o y ege a ion
Josep Padullés Cubino, Idoia Biu un, Gianma ia Bona i, Ta iana B asla skaya, Xa ie Fon , U e Jand ,
Flo ian Jansen, Vale ijus Rašoma ičius, Željko Šk o c, Wol gang Willne and Milan Chy ý
J. Padullés Cubino (h ps://o cid.o g/0000-0002-2283-5004) ✉ ([email p o ec ed]) and M. Chy ý (h ps://o cid.o g/0000-0002-8122-3075), Dep
o Bo any and Zoology, Facul y o Science, Masa yk Uni ., B no, Czech Republic. – I. Biu un (h ps://o cid.o g/0000-0002-1454-0433), Dep o Plan
Biology and Ecology, Uni . o he Basque Coun y UPV/EHU, Bilbao, Spain. – G. Bona i (h ps://o cid.o g/0000-0002-5574-6067), Facul y o Science
and Technology, F ee Uni . o Bozen-Bolzano, Bolzano, I aly. – T. B asla skaya (h ps://o cid.o g/0000-0001-7081-0533), Cen e o Fo es Ecology and
P oduc i i y, Russian Academy o Sciences, Moscow, Russia. – X. Fon (h ps://o cid.o g/0000-0002-7253-8905), Dep o E olu iona y Biology, Ecology and
En i onmen al Sciences, Uni . o Ba celona, Ba celona, Spain. – U. Jand (h ps://o cid.o g/0000-0002-3177-3669), Dep o Geobo any and Bo anical
Ga den, Ma in Lu he Uni . Halle-Wi enbe g, Halle, Ge many; Ge man Cen e o In eg a i e Biodi e si y Resea ch (iDi ) Halle-Jena-Leipzig, Leipzig,
Ge many. – F. Jansen (h ps://o cid.o g/0000-0002-0331-5185), Facul y o Ag icul u al and En i onmen al Sciences, Uni . o Ros ock, Ros ock, Ge many. –
V. Rašoma ičius (h ps://o cid.o g/0000-0003-1314-4356), Na u e Resea ch Cen e, Ins . o Bo any, Vilnius, Li huania. – Ž. Šk o c (h ps://o cid.
o g/0000-0002-2848-1454), Facul y o Fo es y, Uni . o Zag eb, Zag eb, C oa ia. – W. Willne (h ps://o cid.o g/0000-0003-1591-8386), Dep o
Bo any and Biodi e si y Resea ch, Uni . o Vienna, Vienna, Aus ia.
Resea ch
1312
Keywo ds: en i onmen al il e ing, Eu opean Vege a ion A chi e, unc ional biogeog aphy, plan unc ional ai s, TRY
da abase, ege a ion-plo da a
In oduc ion
De e mining he mechanisms d i ing plan communi y
unc ions has been a cen al goal in ege a ion ecology du -
ing he las decades (Acke ly and Co nwell 2007, Violle e al.
2007). These mechanisms ha e usually been in es iga ed
using plan unc ional ai s, de ined as any mo phological,
physical o phenological ea u es o indi iduals ha a ec
hei i ness (Violle e al. 2007). Plan ai s a e known o
p o ide deepe insigh s in o ecosys em unc ioning han
app oaches based only on plan species iden i y (Díaz and
Cabido 2001, McGill e al. 2006). As such, plan unc ional
ai s se e as indica o s o di e en communi y assembly
p ocesses and ha e impo an implica ions o unde s anding
global, egional and local plan species dis ibu ions.
Resea che s ha e globally iden i ied wo main plan
ai spec a a he species le el ha ep esen he ade-o s
be ween esou ce economics and he size o plan s and hei
o gans (W igh e al. 2004, Moles e al. 2009, Adle e al.
2014, Reich 2014, Díaz e al. 2016). On he one hand, he
lea economic spec um (LES) uns om species wi h an
acquisi i e esou ce-use s a egy (i.e. apid esou ce cap u e
and a high ela i e g ow h a e) o hose wi h a mo e con-
se a i e esou ce-use s a egy (i.e. long-li ed lea es and a
low pho osyn he ic a e) (W igh e al. 2004, Reich 2014).
On he o he hand, he plan size spec um (PSS) uns om
sho e species wi h smalle o gans o alle species wi h
la ge o gans. In empe a e zones, alle species end o ha e
la ge canopies, la ge lea es and a deepe and longe oo
sys em, which allow hem g ea e access o ligh , wa e and
soil nu ien s (Co nelissen 1999, Poo e e al. 2005, Padilla
and Pugnai e 2007, Díaz e al. 2016). Howe e , he cos o
main aining non-pho osyn he ic suppo issues inc eases
wi h heigh (King 1990). La ge plan s also end o ha e
hea ie seeds and seed size is known o a ec plan egene a-
ion, o example, h ough seed p oduc ion (Jakobsson and
E iksson 2000), seed su i al in he soil (Thompson e al.
1993) o seed dispe sal dis ance (Thomson e al. 2011).
Scaling up om indi idual plan species ai s o commu-
ni ies has con i med ha hal o he global ai a ia ion a
he communi y le el also e lec s he ade-o s in LES and
PSS (B uelheide e al. 2018). These communi y ade-o s a e
weakly associa ed wi h mac oclima ic and coa se-g ain soil
condi ions globally, al hough wi h di e ences ac oss habi a s,
sugges ing ha ai combina ions a e mainly il e ed by mo e
egional- and local-scale p ocesses (B uelheide e al. 2018).
Indeed, s udies conduc ed a egional scales ha e ound
a iables ela ed o clima e ha shness (minimum/maximum
empe a u e and p ecipi a ion) and seasonali y ( empe a u e
and p ecipi a ion seasonali y) o be he p ima y mac o- il e s
ac ing on he unc ional ai s uc u e o ee communi ies
(Swenson and Weise 2010, Shiono e al. 2015). Local soil
p ope ies, such as soil e ili y o pH, a e also impo an
de e minan s o lea , oo and seed ai s in ee species o
empe a e o es s (Simpson e al. 2016). The e o e, assessing
he ex en o which clima e and soil con ol he a ia ion in
key plan unc ional ai s a la ge spa ial scales can p o ide
new insigh s o he in e p e a ion o ai a ia ion wi hin
plan communi ies.
In Eu ope, o es s cu en ly ep esen o e 40% o he
land su ace (Eu opean En i onmen Agency 2016) and
mos o hei species occu in he unde s o y (i.e. in he
space benea h he o es canopy and abo e he o es loo ).
F om an ecological pe spec i e, ma u e o es unde s o ies
can be seen as highly selec i e en i onmen s adap ed o s a-
ble en i onmen al condi ions whe e shade ole ance a ec s
plan s’ abili y o cope wi h o he s esso s (Ve heyen e al.
2003, Su e al. 2019). T ai a ia ion ac oss Eu opean o es
unde s o ies, like ac oss hei ee laye s, is s ongly shaped
by mac oclima e (Chelli e al. 2019, Vannes e e al. 2019,
Maes e al. 2020) and soil nu ien a ailabili y a egional
scales (Gilliam 2006, Hedwall and B une 2016, Chelli e al.
2019). Fu he mo e, ee canopy co e and s uc u e can also
pa ially bu e clima e a iabili y (Zellwege e al. 2020) and
in luence pa e ns o ligh a ailabili y (Dahlg en 2006), hus
a ec ing unde s o y ai s ela ed o esou ce acquisi ion and
compe i ion. Al hough some la ge-scale s udies ha e exam-
ined ai –en i onmen ela ionships ei he o he woody
laye o o es communi ies (Swenson and Weise 2010,
Shiono e al. 2015, Simpson e al. 2016, Wieczynski e al.
2019) o ac oss all ege a ion laye s (B uelheide e al. 2018),
no s udies ha e assessed he combined e ec o clima e, soil
condi ions and ee canopy co e on ai a ia ion o o es
unde s o ies a he con inen al scale o examined how his
a ia ion di e s om andom expec a ion among o es ypes.
Explo ing hese ela ionships and pa e ns is essen ial because
unde s o y species may espond di e en ly o en i onmen al
luc ua ions han species in he woody laye (Šímo á e al.
2018) and play a i al ole in ecosys em unc ioning (e.g. li -
e decomposi ion and nu ien cycling; Gilliam 2007) and
he p o ision o ecosys em se ices (e.g. habi a p o isioning,
ee egene a ion and pollina ion; Nilsson and Wa dle 2005).
In his s udy, we used ~29 500 o es ege a ion plo s
sampled ac oss Eu ope and classi ied in o 25 di e en o es
ypes o add ess he ela i e ole o mac oclima ic and soil
ac o s and ee canopy co e in d i ing abundance-weigh ed
pa e ns in LES and PSS o o es unde s o ies. These wo
mos ly independen p incipal unc ional ai spec a we e
compu ed om ou plan unc ional ai s: speci ic lea
a ea (SLA) and lea d y ma e con en (LDMC) o LES
and plan heigh and seed mass o PSS. Based on p e ious
egional s udies om Eu opean empe a e and bo eal o es s
(Pe ing e al. 2018, Chelli e al. 2019, Vannes e e al. 2019,
Maes e al. 2020), we expec ed en i onmen al ha shness
1313
(i.e. mo e ex eme empe a u es and soil pH, inc eased cli-
ma ic seasonali y and educed p ecipi a ion) o be associ-
a ed wi h a mo e conse a i e esou ce-use s a egy o o es
unde s o y plan s, and o a lowe a ia ion in lea esou ce-
use s a egies. We expec ed his because a eas wi h ha she
en i onmen al condi ions a e o en poo e in esou ces such
as wa e o nu ien s. Simila ly, we expec ed dense canopy
o a o mo e conse a i e unde s o ies due o limi ed ligh
a ailabili y. Fu he mo e, we expec ed empe a u e and p e-
cipi a ion o be posi i ely ela ed and clima ic seasonali y
nega i ely ela ed o he size o o es unde s o ies and o
g ea e a ia ion in unde s o y plan sizes because s onge
abio ic cons ain s can limi he g ow h o plan s and hei
o gans. We also expec ed canopy openness o a o alle spe-
cies wi h hea ie seeds and p omo e g ea e a ia ion o hese
ai s in he unde s o y because inc eased ligh a ailabili y
educes compe i ion o his esou ce and allows species
o g ow alle and ha e a wide ange o sizes. None heless,
we also expec ed ee canopy co e o media e he e ec o
mac oclima e and soil condi ions on ai a ia ion in he
unde s o y since abio ic cons ain s, pa icula ly low em-
pe a u es and limi ed soil nu ien and wa e a ailabili y, can
also limi he o ma ion o dense canopies (Tang e al. 2019),
and in u n, a ec he unc ion o he unde s o y. Among
o es ypes, we expec ed ha hose in mo e esou ce-limi ed
en i onmen s, such as o es s in mi es, bogs o on acidic
soils, would p esen mo e conse a i e unde s o ies. In con-
as , hose in mo e esou ce- ich and humid en i onmen s,
such as a ine o ipa ian o es s, would show mo e acquisi-
i e unde s o ies (W igh e al. 2004, Reich 2014). We also
expec ed he mophilous o es s a sou he n la i udes o con-
ain la ge (i.e. alle species wi h hea ie seeds) unde s o ies.
Speci ically, we aimed a : 1) explo ing geog aphical
pa e ns in LES and PSS o o es unde s o ies ac oss
Eu ope; 2) de e mining he ela i e con ibu ion o cu en
mac oclima ic ac o s, soil pH and ee canopy co e in
explaining abundance-weigh ed pa e ns in LES and PSS
o Eu opean o es unde s o ies, and; 3) assessing how
unde s o y communi ies o Eu opean o es ypes di e in
hei LES and PSS.
Ma e ial and me hods
Vege a ion da a
We e ie ed geo e e enced ege a ion-plo eco ds om he
Eu opean Vege a ion A chi e (EVA; Chy ý e al. 2016). This
da abase con ains mo e han 1.7 million ege a ion plo s
sampled ac oss Eu ope (see Suppo ing in o ma ion o an
o e iew o con ibu ing da abases). We selec ed plo s wi h
a eas anging be ween 100 and 1000 m2, he mos common
sizes o o es plo s in he da abase, and emo ed plo s
sampled be o e 1970. We also emo ed plo s wi h he loca ion
unce ain y o coo dina es la ge han 10 km. Howe e , we
e ained plo s whe e his in o ma ion was no a ailable
because some coun ies we e ep esen ed only by such da a.
We classi ied ege a ion plo s in o di e en o es ypes
acco ding o he EUNIS classi ica ion o Eu opean habi-
a s using he expe sys em EUNIS-ESy e . 2020-06-08
(Chy ý e al. 2020) un in he JUICE 7.1 so wa e (Tichý
2002). We hen e ained ege a ion plo s belonging o na -
u al o es ypes and con aining a leas i e ascula plan
species (n = 123 682). To educe he disp opo iona ely high
sampling densi y in some a eas, we conduc ed, sepa a ely o
each o es ype, he e ogenei y-cons ained andom (HCR)
esampling (Lengyel e al. 2011) wi h he R package ‘ eg-
clus ’ (De Cáce es e al. 2010). The HCR esampling p o-
cedu e maximizes he mean and minimizes he a iance o
he composi ional dissimila i y be ween pai s o plo s. To
pe o m he HCR esampling, we ini ially assigned each
ege a ion plo o a geog aphical g id o 1° × 1° and cal-
cula ed he median numbe o plo s pe g id cell. Fo hose
g id cells whe e he numbe o plo s exceeded he median,
we pe o med HCR by calcula ing he composi ional dis-
simila i y be ween ege a ion plo s co-occu ing in he same
g id cell wi h he u no e componen o he Sø ensen’s index
(Baselga 2010) and unning 1000 i e a ions. Fo hose g id
cells whe e he numbe o plo s was lowe han he median,
we e ained all plo s. We hen disca ded o es ypes wi h less
han 100 plo s o gua an ee an op imum ep esen a ion o
he ege a ion in each o es ype (Suppo ing in o ma ion).
The o al inal numbe o plo s included in ou s udy was 29
668 (see Suppo ing in o ma ion o he spa ial dis ibu ion
o plo s in he s udy a ea).
We s anda dized he species names acco ding o The Plan
Lis using he R package ‘Taxons and’ (Cayuela e al. 2017).
We disca ded om subsequen analyses he species p esen
in < 1% o he plo s in all o es ypes whe e hey had been
eco ded. We disca ded hese a e species because mos o
hem had limi ed ai da a a ailabili y in he da abases and
hei e ec on he calcula ion o communi y weigh ed means
(CWM) and a iances (CWV) was p esumably negligible
(below). Fo each plo , we ob ained he ela i e abundance o
all ascula plan species.
Plan unc ional ai s
We ob ained da a on speci ic lea a ea (SLA), lea d y-ma e
con en (LDMC), plan heigh and seed mass om he TRY
da abase (<www. y-db.o g/>; Ka ge e al. 2020; accessed
on July 2020) (see Suppo ing in o ma ion o speci ic e e -
ences). SLA and LDMC a e ela ed o LES, wi h lea es wi h
high SLA and low LDMC being mo e acquisi i e (i.e. cheap
o build and e u ning a high pho osyn he ic e enue o a
sho ime) and lea es wi h low SLA and high LDMC being
mo e conse a i e (i.e. equi ing a highe ini ial in es men
o biomass and yielding a low pho osyn he ic income o e
many yea s) (W igh e al. 2004, Moles e al. 2009, Reich
2014). Plan heigh and seed mass a e posi i ely ela ed o
PSS (Díaz e al. 2016, B uelheide e al. 2018). These ai s
a e amongs he mos widely a ailable in TRY (Ka ge
e al. 2020).
1314
P e idophy es do no p oduce seeds, bu hey can be an
essen ial componen o Eu opean o es unde s o ies, pa -
icula ly in humid en i onmen s. To e ain p e idophy es in
ou analysis, we used he lowes seed mass alue ound ac oss
all spe ma ophy es included in he s udy as he seed mass o
all p e idophy es. We also epea ed he analysis excluding all
p e idophy es o es o he e ec o his g oup o species on
he esul s.
We also ob ained TRY da a on lea a ea, lea leng h, lea
N con en , lea C con en and lea P con en . These ai s,
oge he wi h plan axonomy (species, genus and amily),
we e used o es ima e missing ai alues (Suppo ing
in o ma ion) o plan heigh , seed mass, SLA and LDMC
using he Bayesian hie a chical ma ix ac o iza ion (BHPMF)
gap- illing me hod (Sch od e al. 2015). Al hough lea
N and P con en s a e also commonly associa ed wi h LES
(W igh e al. 2004, Reich 2014), we did no use hese
ai s o quan i y LES in ou s udy because a la ge ac ion
o ou species had no TRY da a o hese ai s (Suppo ing
in o ma ion). Be o e pe o ming BHPMF, we excluded TRY
ai eco ds wi h a dis ance > 4 s anda d de ia ions om he
mean o species o a oid po en ial ou lie e ec s (Díaz e al.
2016) and loge- ans o med all ai a iables. To assess he
gap- illing quali y, we used he p edic ion unce ain ies
p o ided by BHPMF o each impu a ion and emo ed all
impu a ions wi h a coe icien o a ia ion > 1 (Fazayeli e al.
2014, B uelheide e al. 2018). Fo he axa ha we e eco ded
a he genus le el only, we calcula ed es ima es o genus
means. We emo ed 68 axa ou o he o al o 2037 included
in ou da ase due o ai da a incomple eness.
In his s udy, we de ined o es unde s o y as he laye
o ege a ion g owing benea h he ee canopy and abo e
he o es loo . We ob ained he g ow h o m o all axa
om TRY, BIEN (Mai ne 2020) and specialized li e a u e
and emo ed all species classi ied as ‘ ees’ om ou lis o
unde s o y axa. We also emo ed ee seedlings in o es
unde s o ies because hey could no be dis inguished om
adul ees in ou da ase . We also emo ed species conside ed
‘c yp ic ees’ (i.e. usually wi h a sh ubby o m) in he
Medi e anean egion o Eu ope (Médail e al. 2019). As a
esul , we gene a ed a lis o o es unde s o y axa wi h 1802
species. We also epea ed he analysis excluding all sh ubs
(i espec i ely o hei size) and lianas, independen ly and
join ly and including as pa o he unde s o y ‘c yp ic ees’,
o explo e he e ec o hese species g oups on ou esul s.
En i onmen al da a
We ob ained cu en clima ic da a o mean annual em-
pe a u e (°C), empe a u e seasonali y (s anda d de ia ion
× 100), o al annual p ecipi a ion (mm) and p ecipi a ion
seasonali y (coe icien o a ia ion) a a esolu ion o 30 a c
seconds om he Wo ldClim e . 2.1 da abase (<www.wo ld-
clim.o g>; Fick and Hijmans 2017). We also ob ained he
Global A idi y Index (he ea e ‘A idi y index’) a a esolu ion
o 30 a c seconds om T abucco and Zome (2019). This
a iable e lec s mois u e a ailabili y o he po en ial g ow h
o e e ence ege a ion, excluding he impac o soil media -
ing wa e uno e en s. A idi y index alues inc ease owa ds
mo e humid condi ions and dec ease wi h mo e a id condi-
ions. Finally, we also ob ained da a o soil pH (a 15 cm
dep h) a a esolu ion o ~ 250 m ( escaled o 30 a c sec-
onds) om he SoilG ids da abase (<h ps://soilg ids.o g/>;
Hengl e al. 2017).
To ex ac en i onmen al da a o each plo , we delimi ed a
ci cula bu e wi h a ixed adius o 2.5 km a ound each plo
and calcula ed he mean alue o all cells wi hin he bu e
zone. The bu e ing app oach smoo hed po en ially ex eme
alues o en i onmen al a iables om poin ex ac ions and
educed biases de i ed om he inaccu acy in he coo dina es.
T ee canopy co e
We de i ed he p opo ion o ee canopy co e in each plo
using he Jennings–Fische ’s o mula (Jennings e al. 2009,
Fische 2015), which combines he pe cen age co e o all
ees in he ee laye in o a single alue ha does no exceed
100%. We p o ide he code o implemen he Jennings–
Fische ’s o mula in R in he Suppo ing in o ma ion.
Co ela ions among en i onmen al a iables and ee canopy
co e can also be ound in he Suppo ing in o ma ion.
S a is ical analyses
We pe o med p incipal componen analysis (PCA) on SLA,
LDMC, plan heigh and seed mass o plan species wi h
Va imax o a ion applied o he i s wo axes wi h cen e ed
and s anda dized da a. The i s p incipal componen
explained 38% o he a ia ion and de ined a g adien anging
om mo e acquisi i e o conse a i e lea es (Suppo ing
in o ma ion). In con as , he second p incipal componen
explained 30% o he a ia ion and de ined a g adien
anging om sho e unde s o y plan s wi h ligh e seeds
o alle unde s o y plan s wi h hea ie seeds (Suppo ing
in o ma ion). We labeled he i s and second ac o s as lea
economic spec um (LES) and plan size spec um (PSS),
espec i ely.
Fo each plo , we calcula ed he communi y weigh ed
mean (CWM; Ga nie e al. 2004, Enquis e al. 2015) o
bo h LES and PSS ac o s using he unc comp unc ion in R
package ‘FD’ (Lalibe é and Legend e 2010). Fu he mo e,
we also calcula ed he communi y weigh ed a iance (CWV;
Ga nie e al. 2004, Enquis e al. 2015) wi h cus omiza ion
o he unc comp unc ion in he same package. We hen cal-
cula ed he s anda dized e ec size (SES) o CWM (CWM.
ses) and CWV (CWV.ses) o assess how plo -le el means and
wi hin-plo ai a iances depa ed om andom expec a ion.
We ob ained CWM.ses and CWV.ses by andomizing 999
imes LES and PSS alues ac oss all species in ou da ase . In
each un, we main ained species abundances in he plo s and
shu led ai alues. We calcula ed SES alues as (obse ed
alue – expec ed alue)/s anda d de ia ion o he expec ed
alue. S anda dized e ec sizes o CWM.ses o LES and PSS
< −1.96 indica e ha o es unde s o ies a e signi ican ly
1315
mo e acquisi i e and smalle han andom, espec i ely. In
con as , s anda dized sizes o CWM.ses o LES and PSS >
1.96 indica e ha o es unde s o ies a e signi ican ly mo e
conse a i e and la ge han andom, espec i ely.
To map CWM.ses o LES and PSS o o es unde s o y
communi ies ac oss Eu ope comp ehensi ely, we c ea ed
a g id o 1° × 1°. We eclassi ied all o es ege a ion plo s
inside each g id cell as ollows: −1 i he plo had signi ican ly
lowe LES (i.e. mo e acquisi i e species) and PSS (i.e. smalle
species) han andom, 0 i he plo had LES and PSS ha did
no di e om andom and 1 i he plo had signi ican ly
highe LES (i.e. mo e conse a i e species) and PSS (i.e.
la ge species) han andom. Then, we calcula ed he o e all
alue o each g id cell as he a e age o all plo s included in
he cell. We mapped only g id cells wi h a minimum o i e
plo s o educe biases associa ed wi h e y low sample sizes
and, a he same ime, maximize he ep esen a ion o o es
unde s o ies ac oss he con inen . The obse ed spa ial pa -
e ns we e obus despi e di e ences in sampling in ensi y
ac oss he con inen (Suppo ing in o ma ion). We epea ed
his same app oach o map CWV.ses o LES and PSS. In his
case, we eclassi ied all o es ege a ion plo s inside each g id
cell as ollows: −1 i he plo had signi ican ly lowe a ia ion
in LES and PSS han andom, 0 i he plo had a a ia ion
o LES and PSS ha did no di e om andom and 1 i
he plo had a signi ican ly highe a ia ion in LES and PSS
han andom.
To es o he ela i e in luence o en i onmen al
a iables and ee canopy co e on plo -le el CWM.ses and
CWV.ses, we used boos ed eg ession ees (BRTs; Eli h e al.
2008) implemen ed in he R package ‘dismo’ (Hijmans e al.
2017). We used BRTs wi h Gaussian dis ibu ion a he
han adi ional eg ession because o he abili y o BRTs
o handle complex non-linea ela ionships and accoun o
collinea i y. We ob ained he op imum numbe o eg ession
ees using a 10- old c oss- alida ion p ocedu e (Suppo ing
in o ma ion). We calcula ed pseudo-R2 o each model
based on he co ela ion be ween obse ed and p edic ed
alues. We ob ained he con ibu ion sco e (in %) o each
en i onmen al a iable as a measu e o i s ela i e impo ance
in BRTs models. We used pa ial dependence plo s o isualize
he shape o he ela ionships be ween esponse and p edic o
a iables. Fu he mo e, we es ed o spa ial au oco ela ion
in he esiduals o each model using Mo an’s I s a is ics o
dis ance classes de ined using S u ge’s ule. Because we we e
in e es ed in spa ial au oco ela ion a sho spa ial dis ances,
we andomly selec ed 200 plo s and calcula ed he spa ial
co elog am o all plo s loca ed wi hin 2° (~225 km) o he
a ge plo (Suppo ing in o ma ion). In he main ex , we
ocus on he esul s o CWM.ses, bu we also e e o he
esul s o CWV.ses in he Suppo ing in o ma ion.
We addi ionally used s uc u al equa ion models (SEMs;
G ace 2006) o quan i y he causal ela ionships be ween
en i onmen al a iables, ee canopy co e and LES and
PSS o o es unde s o ies. We i ed SEMs using a piecewise
app oach o o e come he limi a ions o adi ional a iance–
co a iance SEMs (Le check 2016). Because ou da a we e
spa ially s uc u ed, we implemen ed SEMs using spa ial
simul aneous au o eg essi e (SAR; Kissling and Ca l 2007)
models, which supplemen o dina y leas squa es (OLS)
eg essions wi h a spa ial weigh ma ix ha accoun s o
spa ial au oco ela ion in model esiduals. Spa ial weigh
ma ices we e de ined by successi ely i ing a SAR model
and es ing se e al dis ances be ween neighbo s, anging
om 100 o 200 km away om a gi en g id cell a in e als
o 50 km. We inally selec ed he model ha minimized
spa ial au oco ela ion among he i s dis ance classes and
had he lowes Akaike in o ma ion c i e ion (AIC) (i.e. 100
km; Suppo ing in o ma ion). We included quad a ic e ms
calcula ed as [x − mean(x)]2 (Mau eaud e al. 2019) o mean
annual empe a u e and soil pH in ou SEMs o accoun o
he non-linea e ec o hese ac o s on ou esponse a iables
(Suppo ing in o ma ion). We also emo ed he A idi y index
om his pa o he analysis because i was highly co ela ed
wi h annual p ecipi a ion (Suppo ing in o ma ion) and
no ably inc eased he le els o mul icollinea i y in ou
models when measu ed using a iance in la ion ac o s (VIF
> 2.5; Legend e and Legend e 2012). We epo Nagelke ke
pseudo-R-squa ed (R2) o inal SAR models as a measu e o
he coe icien o de e mina ion. We an SEMs wi h he R
package ‘piecewiseSEM’ (Le check 2016) and SAR models
wi h he R package ‘spa ial eg’ (Bi and and Pi as 2015).
We es ablished signi icance a α < 0.05 and pe o med all
he analyses in R e . 3.5.3 (<www. -p ojec .o g>).
Resul s
Spa ial pa e ns o LES and PSS o Eu opean o es
unde s o ies
Fo es unde s o ies consis ing o species wi h a conse a i e
esou ce use occu ed mainly in Fennoscandia and he
Medi e anean Basin (Fig. 1). In con as , communi ies o
species wi h mo e acquisi i e lea es ended o occu in he
empe a e zone o Eu ope. Talle plan s wi h hea ie seeds
ended o occu in he Medi e anean Basin, F ance and
I eland, while plan s signi ican ly sho e han unde andom
expec a ion a ely domina ed o es unde s o ies, occu ing
spo adically a la i udes highe han 50°N. The a iance in
LES ended o be highe in he cen al pa s o Eu ope. In
con as , he a iance in PSS was pa icula ly high along
he A lan ic seasho e o F ance, England, I eland, Spain and
Po ugal (Suppo ing in o ma ion). Resul s conside ing only
species occu ences in communi ies ended o maximize hese
pa e ns (Suppo ing in o ma ion).
When we emo ed all sh ubs and lianas om he
analysis, he s ong pa e ns owa ds mo e conse a i e
o es unde s o ies in Fennoscandia and, o a lowe ex en ,
he Medi e anean Basin ended o dilu e (Suppo ing
in o ma ion). We also ound ha o es unde s o ies wi h
plan s o seeds signi ican ly la ge han unde andom
expec a ions dec eased in se e al a eas in he Medi e anean
Basin, F ance and I eland when we emo ed sh ubs and
1316
lianas. Resul s did no di e subs an ially when we emo ed
e ns (Suppo ing in o ma ion).
En i onmen al d i e s o LES and PSS o Eu opean
o es unde s o ies
The BRTs explained 36% and 43% o he a ia ion in
CWM.ses o LES and PSS, espec i ely. The mos impo an
a iables o LES included mean annual empe a u e
(34.8%), he p opo ion (%) o ee canopy co e (20.4%)
and empe a u e seasonali y (14.2%) (Fig. 2; Suppo ing
in o ma ion). LES showed a nega i e quad a ic ela ionship
wi h mean annual empe a u e and was nega i ely linea ly
ela ed o he % o canopy co e and empe a u e seasonali y.
The mos impo an a iables o PSS included mean annual
empe a u e (58.6%) and, o a lowe ex en , he % o
canopy co e (12.2%) and p ecipi a ion seasonali y (8.7%)
(Fig. 2; Suppo ing in o ma ion). PSS showed a posi i e
linea ela ionship wi h mean annual empe a u e, al hough
i ended o s abilize a high and low empe a u es. PSS also
showed a posi i e and nega i e linea ela ionship wi h he
% o canopy co e and p ecipi a ion seasonali y, espec i ely.
The BRT models also explained 18% and 28% o he
a ia ion in CWV.ses o LES and PSS, espec i ely. The
a iances in LES we e bes p edic ed by mean annual
empe a u e and empe a u e and p ecipi a ion seasonali y.
In con as , he a iances in PSS we e bes p edic ed by mean
annual empe a u e, empe a u e seasonali y and he % o
canopy co e (Suppo ing in o ma ion).
Causal ela ionships
Ou SEMs sugges ed mul iple causal ela ionships be ween
en i onmen al a iables, ee canopy co e and LES and
PSS o Eu opean o es unde s o ies (Fig. 3; Suppo ing
in o ma ion). Mean annual empe a u e and soil pH had
di ec and indi ec e ec s h ough ee canopy co e on
LES and PSS. P ecipi a ion seasonali y also had di ec and
indi ec e ec s on PSS, bu only an indi ec e ec on LES.
In con as , empe a u e seasonali y also had di ec and
indi ec e ec s on LES bu only an indi ec e ec on PSS.
Annual p ecipi a ion only had a signi ican di ec e ec on
PSS. Mo e ex eme mean annual empe a u es and soil pH
and highe clima ic seasonali y induced mo e open canopies,
which in u n we e associa ed wi h mo e conse a i e and
smalle (i.e. sho e plan s wi h ligh e seeds) unde s o ies.
T ee canopy co e and mean annual empe a u e had
he s onges di ec e ec s on LES, while mean annual
empe a u e disp opo iona ely had he s onges di ec
e ec on PSS.
Cha ac e iza ion o Eu opean o es ypes based on
LES and PSS o hei unde s o ies
In gene al, o es ypes wi h a highe p opo ion o mo e con-
se a i e plan unde s o ies han unde andom expec a ion
we e coni e ous o scle ophyllous (Fig. 4a, 5). Coni e ous o -
es s we e mainly ep esen ed by Pinus and La ix mi e o es ,
Medi e anean lowland o submon ane Pinus o es , Picea
Figu e 1. Spa ial pa e ns in he (a) lea economic spec um (LES) and (b) plan size spec um (PSS) o Eu opean o es unde s o y
ege a ion calcula ed as CWM.ses. Values o −1 indica e ha all ege a ion plo s in he gi en cell had (a) mo e acquisi i e lea es o (b)
smalle species han unde andom expec a ion. Values o 1 indica e ha all ege a ion plo s in he gi en cell had (a) mo e conse a i e
lea es o (b) la ge species han unde andom expec a ion. Values o 0 indica e ha plo s in he gi en cell did no di e om he andom
expec a ion o LES and PSS. Only g id cells wi h a leas i e plo s con aining a leas i e species we e mapped.
1317
mi e o es , Pinus syl es is ligh aiga and empe a e subalpine
La ix, Pinus cemb a and Pinus uncina a o es s. Scle ophyllous
o es s we e ep esen ed by Medi e anean e e g een Que cus
o es s. Deciduous o es ypes wi h he highes p opo ion
o plo s wi h a mo e conse a i e esou ce-use s a egy han
andom included b oadlea ed mi e o es on acid pea . In
con as , o es ypes wi h a mo e acquisi i e esou ce-use
s a egy co esponded o a ine o es , Fagus o es on non-
acid soil, Ca pinus and Que cus mesic deciduous o es , da k
aiga and empe a e moun ain Picea and Abies o es s.
Fo es ypes wi h a highe p opo ion o la ge (i.e. alle
and wi h hea ie seeds) unde s o y plan s han unde an-
dom expec a ion we e gene ally deciduous. They included
Medi e anean and Maca onesian ipa ian o es , a ine o -
es o empe a e and submedi e anean he mophilous decid-
uous o es (Fig. 4a, 5). Mo eo e , Medi e anean e e g een
Que cus o es s showed he highes p opo ion o la ge plan s
in unde s o y communi ies han unde andom expec a ion.
We ound no majo pa e ns in he a iance o LES among
o es ypes. S ill, we ound a endency o b oadlea ed decid-
uous and e e g een o es s o be mo e a iable in PSS han
coni e ous o es s (Suppo ing in o ma ion).
When we conside ed ‘c yp ic ees’ as pa o he unde -
s o y, he p opo ion o Medi e anean and submedi e anean
o es plo s wi h mo e conse a i e and alle unde s o ies
sligh ly inc eased, bu he o e all obse ed pa e ns emained
iden ical (Suppo ing in o ma ion). The lis o species wi h
he highes con ibu ions o LES and PSS o each o es ype
can be ound in he Suppo ing in o ma ion.
Discussion
Geog aphical pa e ns and d i e s o LES and PSS in
o es unde s o ies
Ou s udy quan i ied he unc ional ai a ia ion in he lea
economic (LES) and plan size (PSS) spec a o o es unde -
s o ies and ela ed he obse ed pa e ns o he unde lying
e ec s o mac oclima e, soil a iables and ee canopy co e
ac oss Eu ope. Clima e, pa icula ly mean annual empe a-
u e, exe ed he g ea es e ec on bo h LES and PSS. This
esul is in line wi h ecen s udies conduc ed in o es ecosys-
ems a mo e egional scales (Pe ing e al. 2018, Chelli e al.
2019, Vannes e e al. 2019, Maes e al. 2020). I con i ms he
ole o clima e as a p ima y mac o- il e shaping communi y
mean plan ai s and hei a ia ion a he con inen al scale.
I also shows ha he e ec o clima e and soil cha ac e is ics
Figu e 2. E ec s o he h ee op- anked p edic o a iables on CWM.ses o he (a–c) lea economic spec um (LES) and (d– ) plan size
spec um (PSS) o Eu opean o es unde s o y ege a ion (pa ial dependence plo s). The in eg a ed impo ance sco es a e shown in
pa en heses. LES de ines a g adien anging om mo e acquisi i e o conse a i e lea es o o es unde s o ies. PSS de ines a g adien
anging om smalle plan s wi h ligh e seeds o la ge plan s wi h hea ie seeds in he unde s o y. Smoo hed e sions o he i ed unc ions
( ed dashed cu es) we e calcula ed using local polynomial eg ession. Ve ical icks on he x-axis indica e deciles o he esponse a iable.
No e han he y-axis ep esen s he i ed unc ion and has a ze o mean o e he da a dis ibu ion. The e ec s o lowe - anked p edic o s can
be ound in he Suppo ing in o ma ion.
1318
on ai a ia ion o Eu opean o es unde s o ies is pa ly
media ed by ee canopy co e .
Fo es unde s o ies loca ed in a eas wi h he lowes and
highes empe a u es we e cha ac e ized by mo e conse a-
i e lea s a egies (i.e. high LDMC and low SLA). These
a eas we e mainly loca ed in Fennoscandia and a ound he
Medi e anean Basin, pa icula ly in he eas e n Ibe ian
Peninsula, sou he n I aly and Ana olia. The p e ailing
en i onmen s o hese egions a e s ess ul o he plan s:
Fennoscandia due o low empe a u e and sho g owing sea-
son and he Medi e anean due o summe d ough . Species
in hese s essed en i onmen s ocus on esou ce conse a ion
h ough hicke lea es wi h a longe li e span (Wes oby e al.
2002, W igh e al. 2004, Reich 2014). The obse ed en-
dency owa ds mo e conse a i e o es unde s o ies a colde
no he n la i udes was p ima ily d i en by E icaceae dwa
sh ubs such as Vaccinium spp., Calluna ulga is o Ledum
spp. (Suppo ing in o ma ion), which ha e adap a ions o
nu ien -poo habi a s, including ough e e g een lea es o
e icoid myco hiza (Cai ney and Meha g 2003). In wa me
and d ie egions, p oducing conse a i e e e g een lea es is
p obably a wa e -conse ing a he han nu ien -conse ing
s a egy. Acco dingly, species wi h he highes con ibu ions
o mo e conse a i e esou ce-use s a egies o unde s o ies
in wa m and d y a eas included se e al Medi e anean e e -
g een sh ubs ha h i e in d y condi ions, such as Cis us spp.,
E ica spp. o Rosma inus spp. In con as , o es unde s o ies
wi h mo e esou ce-acquisi i e s a egies occu ed in em-
pe a e and mo e mesic a eas o cen al and eas e n Eu ope.
They we e domina ed by species wi h mo e inexpensi e, hin-
ne and sho -li ed lea es ha pe o m well in nu ien - ich
en i onmen s (e.g. Oxalis ace osella, Galium spp. o Lac uca
mu alis; Suppo ing in o ma ion).
Plan unde s o ies in wa me egions we e no only mo e
conse a i e bu also composed o alle species wi h hea ie
seeds. Mo eo e , hey showed a la ge a ia ion in plan sizes
han hose in mo e empe a e egions (Moles e al. 2009, De
F enne e al. 2013, Vannes e e al. 2019). Howe e , we ound
his pa e n o be connec ed o he ela i ely high abundance
o sh ubs and lianas in he unde s o ies o hese egions. These
Figu e 3. S uc u al equa ion models (SEMs) explaining CWM.ses o he (a) lea economic spec um (LES) and (b) plan size spec um
(PSS) o Eu opean o es unde s o ies. Black and ed a ows ep esen posi i e and nega i e ela ionships, espec i ely. Pa h hickness
e lec s he s eng h o he ela ionship (i.e. alues o s anda dized β coe icien s). A ows o non-signi ican pa hs (p ≥ 0.05) a e semi-
anspa en . S anda dized β coe icien s and Nagelke ke pseudo-R-squa ed alues (R2) o he dependen a iables a e also shown. ‘x2’
indica es i he a iable was ans o med ollowing: [x − mean(x)]2. To al e ec s o p edic o a iables can be ound in he Suppo ing
in o ma ion.
1319
Figu e 4. (a) Mean CWM.ses o he lea economic spec um (LES) and plan size spec um (PSS) o o es unde s o y communi ies in
Eu opean o es ypes. We also show (b) he a e age posi ion o Eu opean o es ypes along wo majo axes o clima e a ia ion (‘mean
annual empe a u e’ and ‘annual p ecipi a ion’) o a be e in e p e a ion o (a). The size o he poin s in (b) e lec s he a e age p opo ion
(%) o ee canopy co e in each o es ype. In bo h plo s, ba s ep esen he 25% and 75% quan iles. B oadlea ed deciduous o es s (T1)
a e colo ed wi h wa m ed colo s, scle ophyllous o es s (T2) a e da k g ey and coni e ous o es s (T3) a e cold blue.
Figu e 5. P opo ion o plo s om o es ypes wi h unde s o ies con aining plan species wi h (a) mo e acquisi i e (b own) and conse a i e
(blue) lea economics han unde andom expec a ion, and wi h (b) smalle (b own) and la ge (blue) plan and seed sizes han unde
andom expec a ion, based on CWM.ses o he lea economic spec um (LES) and plan size spec um (PSS), espec i ely. The p opo ion
o plo s wi h andom LES and PSS a e shown in ligh g ey.
