Local opog aphic and edaphic ac o s la gely p edic sh ub
enc oachmen in Medi e anean d ylands
Alice Nunes
a
,MelanieKöbel
a
, Ped o Pinho
a,b
, Paula Ma os
a
, Edoa do A.C. Cos an ini
c
, C is ina Soa es
a
,
F ancesco de Bello
d
,O íliaCo eia
a
, C is ina B anquinho
a,
⁎
a
Cen e o Ecology, E olu ion and En i onmen al Changes, Faculdade de Ciências, Uni e sidade de Lisboa, Campo G ande, C2, Piso 5, 1749-016 Lisboa, Po ugal
b
CERENA - Cen o de Recu sos Na u ais e Ambien e, Uni e sidade de Lisboa, Ins i u o Supe io Técnico, A . Ro isco Pais, 1049-001 Lisboa, Po ugal
c
CREA-AA, Resea ch Cen e o Ag icul u e and En i onmen , Flo ence, I aly
d
Ins i u e o Bo any, Czech Academy o Sciences, CZ-37982 T ebon, Czech Republic
HIGHLIGHTS
•Sh ub enc oachmen is a wo ldwide
phenomenon ha a ec s many ecosys-
em se ices.
•I scausesneed obebe e unde s ood
and quan ified in he Medi e anean
Basin.
•We s udied i s d i e s, and impac on
plan unc ional ai s in Holm-oak
woodlands.
•We ound ha opo-edaphic ac o s, no
clima e, la gely p edic sh ub enc oach-
men .
•We help o p edic enc oachmen unde
clima e change and o imp o e manage-
men ac ions.
GRAPHICAL ABSTRACT
abs ac a icle in o
A icle his o y:
Recei ed 15 July 2018
Recei ed in e ised o m 29 No embe 2018
Accep ed 30 No embe 2018
A ailable online 1 Decembe 2018
Sh ub enc oachmen influences se e al ecosys em se ices in d ylands wo ldwide. Ye , commonly used s a e-
gies o educe enc oachmen show a low medium- e m success, calling o a be e unde s anding o i s causes.
P e ious wo ks iden ified mul iple d i e s esponsible o his phenomenon, including an h opogenic and en i-
onmen al causes. Howe e , he ela i e e ec o clima e, opog aphy and edaphic ac o s on sh ub enc oach-
men is no ully unde s ood no has been p ope ly quan ified in Medi e anean Basin d ylands. Also,
unde s anding how hese d i e s lead o changes in plan communi ies' unc ional ai s associa ed o sh ub en-
c oachmen is c ucial, conside ing ai s influence ecosys em p ocesses and associa ed ecosys em se ices. He e,
we s udied he unde s o y o a Medi e anean d yland ecosys em composed o sa anna-like Holm-oak wood-
lands, along a egional clima ic g adien . We specifically assessed (i) how clima ic, opog aphic and edaphic ac-
o s influence unde s o y ela i e sh ub co e (RSC) and (ii) hei di ec and indi ec e ec s ( ia RSC) on plan
unc ional ai s. We s udied he mean and di e si y o 12 unc ional ai s ela ed o plan egene a ion, es ab-
lishmen , and dispe sal, a he communi y-le el. We ound ha , unde simila low-in ensi y land use, opo-
g aphic and edaphic ac o s, namely slope a ia ions and soil C:N a io, we e he mos impo an p edic o s o
sh ub enc oachmen , de e mining communi ies' unc ional cha ac e is ics. Clima e, namely summe p ecipi a-
ion, had a much lesse influence. Ou model explained 52% o he a ia ion in ela i e sh ub co e . Clima e
had a s onge e ec on a se o unc ional ai s weakly in ol ed in sh ub enc oachmen , ela ed o flowe ing
Keywo ds:
A idi y
Clima e change
Holm-oak
Func ional ai s
S uc u al equa ion model
Sus ainable land managemen
Science o he To al En i onmen 657 (2019) 310–318
⁎Co esponding au ho .
E-mail add esses: amanunes@ c.ul.p (A. Nunes), mkoba is a@ c.ul.p (M. Köbel), paplopes@ c.ul.p (P. Pinho), psma os@ c.ul.p (P. Ma os), edoa do.cos an ini@c ea.go .i
(E.A.C. Cos an ini), cmdsoa es@ c.ul.p (C. Soa es), adebello@p .jcu.cz (F. de Bello), odga o@ c.ul.p (O. Co eia), cmb anquinho@ c.ul.p (C. B anquinho).
h ps://doi.o g/10.1016/j.sci o en .2018.11.475
0048-9697/© 2018 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
Con en s lis s a ailable a ScienceDi ec
Science o he To al En i onmen
jou nal homepage: www.else ie .com/loca e/sci o en
and dispe sal s a egies. We show ha sh ub enc oachmen is la gely p edic ed by opo-edaphic ac o s in Med-
i e anean d ylands subjec o con en ional low-in ensi y land use. Hence, managemen s a egies o educe en-
c oachmen need o ake hese d i e s in o accoun o e ficien o ecas ing and highe cos -e ec i eness. Ou
esul s sugges ha clima e change migh no g ea ly impac sh ub enc oachmen in he Medi e anean Basin,
bu may a ec unc ional s uc u e and educe unc ional di e si y o plan communi ies, hus a ec ing ecosys-
em unc ioning.
© 2018 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://
c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
1. In oduc ion
Sh ub enc oachmen , defined as he inc ease in densi y, co e and
biomass o na i e woody species (Van Auken, 2009), has been in en-
si ely deba ed in ecen li e a u e (Eld idge e al., 2011;Sala and
Maes e, 2014;Eld idge and Soli e es, 2015;Maes e e al., 2016). I
has been epo ed mos ly o d yland ecosys ems, e.g. in No h
Ame ica (Van Auken, 2009), Aus alia (Eld idge and Soli e es, 2015),
Sou h A ica (Roques e al., 2001) and in he Medi e anean Basin
(Cas o and F ei as, 2009;Maes e e al., 2009;Caldei a e al., 2015).
Sh ub enc oachmen esul s om bo h na u al and human-induced ac-
o s ha a e di ficul o disen angle, leading o appa en ly con adic o y
esul s. Long- e m o e g azing has been o en conside ed a cause o
sh ub enc oachmen in g asslands in wes e n Uni ed S a es o in sou h-
e n A ica d ylands (Schlesinge e al., 1990;Roques e al., 2001;MEA,
2005;Van Auken, 2009). Con as ingly, in he Medi e anean Basin,
sh ub enc oachmen is seen as a consequence o g azing exclusion
(Cas o and F ei as, 2009) o o he abandonmen o he land, ceasing
ac i e sh ub emo al (Maes e and Co ina, 2005), i.e. as an expec ed
ou come o a seconda y succession. I has been also associa ed o pas
sh ub clea ing, educed fi e equency, inc eases in CO
2
, N deposi ion
and long- e m clima e change (A che , 2010).
A a global scale, mean annual p ecipi a ion se s an uppe limi o
woody co e (sh ubs o ees) (Sanka an e al., 2005;Hi o a e al.,
2011). Howe e , locally, opog aphy and soil cha ac e is ics limi his
po en ial (Colgan e al., 2012;Sala and Maes e, 2014), pa icula ly in
wa e -limi ed ecosys ems (Gómez-Plaza e al., 2001), because hey
la gely de e mine wa e a ailabili y o plan s. Finally, bio ic-abio ic in-
e ac ions may also influence sh ub co e h ough posi i e eed-backs.
Sh ub coloniza ion may lead o a mo e he e ogeneous dis ibu ion o
soil wa e and nu ien s, hus p omo ing u he sh ub enc oachmen
(Schlesinge e al., 1990). In sho , sh ub enc oachmen depends on
mul iple ac o s, a ying wi h he his o ical managemen (e.g. g azing
p essu e), en i onmen al condi ions (e.g. clima e, soil p ope ies, o-
pog aphy) and he scale o analysis (e.g. landscape o pa ch le el)
(Eld idge e al., 2011;Eld idge and Soli e es, 2015).
Sh ub enc oachmen has been desc ibed as a land deg ada ion p o-
cess (Schlesinge e al., 1990;MEA, 2005), mos ly om he poin o iew
o li es ock p oduc ion, a p ima y use o d ylands. Howe e , i may im-
p o e ca bon seques a ion (Da yan o e al., 2013), soil e ili y and N
mine aliza ion a e (Maes e e al., 2009;Gómez-Rey e al., 2013), ha -
ing a beneficial e ec on ecosys em unc ioning a leas un il a ce ain
h eshold o ela i e woody co e (41–60%) is eached (Soli e es
e al., 2014;López-Díaz e al., 2015). Se e al p e ious wo ks suppo
he idea ha he consequences o sh ub enc oachmen a e con ex -
dependen , a ying e.g. wi h p ecipi a ion (Jackson e al., 2002;Knapp
e al., 2008;Anadón e al., 2014). Recen wo ks sugges ha he e ec
o sh ub enc oachmen on ecosys em unc ioning la gely depends on
he unc ional ai s o he woody and he baceous species in ol ed
(Ri es e al. 2011;Rolo e al., 2012;Soli e es e al., 2014;Maes e
e al., 2016), which la gely influence ecosys em p ocesses (Díaz e al.,
2007;de Bello e al., 2010). Sh ub enc oachmen should in ol e
changes in communi y unc ional ai s ha co- a y among sh ub and
he baceous g ow h- o ms, such as plan heigh , li e-cycle, oo dep h,
and specificlea a ea(Valencia e al., 2015), eflec ing ade-o s in e-
sou ce alloca ion pa e ns (Díaz e al., 2016).
P e ious s udies add essed he e ec s o mul iple ac o s on sh ub
enc oachmen , mainly om ou side o he Medi e anean Basin
(Roques e al., 2001;D'odo ico e al., 2012;Iannone e al., 2015). Some
o hem p edic i s inc ease unde a clima e change scena io o in-
c eased a idi y (D'odo ico e al., 2012), like he one o ecas ed o Med-
i e anean Basin d ylands (Huang e al., 2016). Howe e , o ou
knowledge, none has quan ified join ly he ela i e e ec o clima e, o-
pog aphy and edaphic ac o s on sh ub enc oachmen and associa ed
changes in communi y unc ional ai s in Medi e anean Basin d y-
lands. We p opose o o e come his knowledge gap by s udying he
plan communi y o sa anna-like Holm-oak woodlands along a spa ial
clima ic g adien , comp ising si es wi h a ying opo-edaphic cha ac-
e is ics. Hence, we adop ed a ‘space- o - ime subs i u ion’, assessing
changes along space o in e changes o e ime (Blois e al., 2013).
Holm-oak woodlands consis o sca e ed ees wi h an unde s o y o
semi-na u al g asslands and sh ubland pa ches, and ha e a adi ional
low in ensi y sil o-pas o al use. These he e ogeneous sa anna-like eco-
sys ems a e called mon ados in Po ugal o dehesas in Spain, and suppo
a ema kably ich biodi e si y, being e y impo an o he economy o
u al a eas (Pe ei a and Da Fonseca, 2003). Specifically, we assessed
(i) how clima ic, opog aphic and edaphic ac o s influence ela i e
sh ub co e and (ii) hei di ec and indi ec e ec s ( ia ela i e sh ub
co e ) on plan unc ional ai s. We s udied he mean and di e si y o
12 unc ional ai s ela ed o plan es ablishmen , pe sis ence, egene -
a ion, and dispe sal, a he communi y le el. To es di ec and indi ec
e ec s o en i onmen al d i e s on he plan communi y we buil an a
p io i causal model based on ou esul s and on obse a ionso p e ious
s udies, using s uc u al equa ion modelling (SEM model) (G ace e al.,
2012). Unde simila low-in ensi y land use, clima e and opo-edaphic
a iables a e expec ed o exe a s ong con ol on ela i e sh ub
co e (Sala and Maes e, 2014) a a egional and local scale, espec i ely.
Ou fi s hypo hesis is ha local opo-edaphic ac o s a e as impo an
as clima e, as d i e s o sh ub enc oachmen . G ass and sh ub-
domina ed communi ies a e expec ed o di e conside ably in unc-
ional ai s' means and di e si y. Because unc ional ai s a e linked
o ecosys em p ocesses (e.g. p ima y p oduc i i y, decomposi ion) (de
Bello e al., 2010), his in o ma ion can p o ide clues abou he conse-
quences o sh ub enc oachmen o ecosys em unc ioning. In addi ion,
i is impo an o disen angle which unc ional consequences a e mos ly
d i en by local en i onmen al ac o s, om hose due o egional fil e s,
i.e. clima ic a iables. Ou second hypo hesis is ha he communi y e-
sponse o local en i onmen al fil e s ia changes in ela i e sh ub co e
(i.e. indi ec ly) will a ec a di e en se o unc ional ai s han ha due
o egional clima ic fil e s.
2. Me hods
2.1. S udy si es and en i onmen al a iables
The s udy was de eloped in Medi e anean d ylands domina ed by
Holm-oak woodlands (Que cus ilex L. subsp. o undi olia) in sou hwes -
e n Ibe ian Peninsula. Field sampling was ca ied ou in 54 si es selec ed
along a egional clima ic g adien . A s a ified andom selec ion o si es
along a mac oclima ic g adien was made (Fig. S1), based on he Uni ed
Na ions a idi y index (AI) (Middle on and Thomas, 1992). The AI ep e-
sen s he a io o mean annual p ecipi a ion o annual po en ial
311A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318
e apo anspi a ion, and he lowe he index alue, he highe he a id-
i y. The global a idi y da abase (h p://www.cgia -csi.o g/da a/global-
a idi y-and-pe -da abase)(T abucco and Zome , 2009) was used o e-
ie e AI da a o he pe iod 1950–2000. Along he s udy a ea AI anged
om 0.42 o 0.56. The selec ion o he sampling si es was made in o de
o ensu e ha hey had no ag icul u al ac i i ies o e he las fi e yea s,
and mode a e o low g azing in ensi y (see Nunes e al., 2017 o mo e
de ails abou sampling si e selec ion). We sampled he unde s o y o
Holm-oak woodlands wi h a spa se ee co e (b40 ees/ha, on a e -
age) (Ama al e al., 1997). Sh ub co e anged om 0% o 87% (a e age
16%) o e all sampling si es. The mos common species in sh ub pa ches
we e Cis us ladani e ,Cis us sal i olius and La andula s oechas,and
in g assland a eas we e B achypodium dis achyon,Vulpia genicula a,
Gaudinia agilis,Leon odon a axacoides,Tolpis ba ba a,and
Chamaemelum mix um.
Sampling si es we e cha ac e ized by a se o 19 clima e a iables ex-
ac ed om Wo ldclim da abase wi h a 1 km
2
esolu ion (Hijmans
e al., 2005) (Table S1). Also, h ee opog aphic a iables we e com-
pu ed o each si e based on digi al ele a ion models wi h 10 m esolu-
ion using A cGIS 10.1 (ESRI, 2010): local slope alues, he s anda d
de ia ion o he slope wi hin a 250 m bu e a ound he sampling si e
cen oid, and a opog aphic we ness index (TWI) (Sø ensen e al.,
2005), o quan i y opog aphic con ol on hyd ological p ocesses
(Table S1). A highe TWI alue indica es a highe po en ial o uno .
Soil samples we e collec ed a each si e om he uppe 10 cm (compos-
i e sample o 5 subsamples). Soil ex u e (% o sand, clay and sil ) and
nu ien con en (soil o ganic ma e , soil N, and C:N a io) we e de e -
mined using s anda d p ocedu es (Table S1). A e explo ing he co e-
la ion among he clima ic, opog aphic and edaphic desc ip o s a each
si e (Tables S2 and S3), we la e summa ized hem in o 8 a iables
showing a Pea son's co ela ion b0.70, o be used in subsequen anal-
ysis (Table 1).
2.2. Vege a ion sampling and communi y ai me ics
Unde s o y plan sampling was pe o med in Ap il–June 2012, a he
peak o s anding biomass. Plan co e was assessed using he poin –
in e cep me hod, a oiding d ainage and flooding a eas. Species co e
was es ima ed along six 20 m ansec s es ablished a each si e, in poin s
spaced e e y 50 cm ( u he de ail on he sampling scheme may be
ound in Nunes e al., 2014). The yea in which he sampling
ook place had a low a e age annual p ecipi a ion (370 ± 61 mm)
when compa ed o he 50-yea mean o he sampling si es al oge he
(561 ± 27 mm). We used he dominan species a aining ≥80% o he
ela i e co e o cha ac e ize communi ies' unc ional ai s (Pakeman
and Ques ed, 2007) (95 species o e all, and 30 ± 9 pe si e). Plan
ai selec ion was based on plan s a egies ela ed o es ablishmen ,
pe sis ence, egene a ion, and dispe sal, o aling 12 unc ional ai s
(Table S4). T ai da a was ob ained h ough di ec measu emen s in
he field ollowing s anda d p o ocols (Pé ez-Ha guindeguy e al.,
2013), o e ie ed om a ious bibliog aphic sou ces (Table S4). We
calcula ed unc ional s uc u e and di e si y h ough he communi y-
weigh ed-mean (CWM) (Ga nie e al., 2007) and unc ional dispe sion
(FDis) (Lalibe e and Legend e, 2010), espec i ely. The CWM is calcu-
la ed as he a e age ai alue in a communi y weigh ed by he ela i e
abundance o he species ca ying each alue (Ga nie e al., 2007). FDis
co esponds o he weigh ed mean dis ance o indi idual species om
he weigh ed cen oid o all species in a mul idimensional ai space,
whe e weigh s a e species ela i e abundances (Lalibe e and
Legend e, 2010). I eflec s he deg ee o unc ional dissimila i y wi hin
he communi y (Lalibe e and Legend e, 2010), and a highe FDis is ex-
pec ed o lead o inc eased complemen a i y in esou ce use be ween
species and hus o inc eased ecosys em unc ioning (Tilman e al.,
1997). Mean ai alues pe species we e used o compu e unc ional
me ics. P io o analysis, con inuous ai s we e log ans o med. To
deal wi h con inuous, o dinal, and ca ego ical a iables, as well as
wi h missing alues, he Gowe dis ance was used in calcula ions. All
calcula ions we e done using he dbFD unc ion o he FD package
(Lalibe é e al., 2014)inR(RCo eTeam,2015).
2.3. S a is ical analyses
We fi s analysed ela i e sh ub co e h ough linea eg ession
using he selec ed clima ic (a idi y index, summe p ecipi a ion, win e
empe a u e), opog aphic (slope s anda d de ia ion, opog aphic we -
ness index) and edaphic (soil o ganic ma e , C:N a io, and sand con-
en ) p edic o s (Table 1), as well as wo-way in e ac ions among
hem. We also included a quad a ic e m o a idi y, o accoun o non-
linea esponses (G oss e al., 2013). Co ela ion among he p edic o s
was below Pea son's 0.7 in all cases (Tables S2 and S3). We used a
mul i-model in e ence app oach and he bes model was selec ed
based on he Akaike in o ma ion c i e ion (AIC), a e ensu ing no sig-
nifican mul icollinea i y h ough a iance infla ion ac o s (VIF). A e -
wa ds, we buil eg ession models o he unc ional ai me ics using
he same p edic o s and p ocedu e. As we ound s ong co ela ions be-
ween some ai unc ional me ics (Tables S5 and S6), we conduc ed
wo p incipal componen analyses (PCA), based on co ela ion ma ices,
using sepa a ely CWM and FD alues o all ai s measu ed. This allowed
us o summa ize ai a ia ion, and iden i y he mos in o ma i e axes
o unc ional specializa ion (Díaz e al., 2016). We hen used he wo
fi s PCA componen s o each analysis as a measu e o he CWM and
he FD o each communi y, i.e., as esponse a iables (see Valencia
e al., 2015 o a simila app oach). Then, we used a a iance decompo-
si ion analysis based on he bes models selec ed o ob ain he pe cen -
age o a iance explained by clima ic and opo-edaphic p edic o s and
hei in e ac ions (Dubuis e al., 2013). Finally, o iden i y he mecha-
nisms ha con ol sh ub enc oachmen , and o es di ec and indi ec
e ec s (i.e., ia changes in ela i e sh ub co e ) o clima ic and opo-
edaphic ac o s on he unc ional ai s o he plan communi y, we
used s uc u al equa ion modelling (SEM). This app oach is well sui ed
o s udying hypo heses abou p ocesses wi h complex causal connec-
ions (G ace e al., 2012). I allows sepa a ing he di ec and indi ec e -
ec s o he p edic o s included in a model and es ima ing he s eng hs
o mul iple e ec s. We es ablished an a p io i model based on ou cu -
en knowledge and on he bes eg ession models p e iously buil o
each esponse a iable. We hypo hesized ha (i) clima ic and opo-
edaphic a iables would di ec ly a ec ela i e sh ub co e and he
unc ional cha ac e is ics o he plan communi y and (ii) changes in
Table 1
En i onmen al a iables selec ed, hei uni s and ange in he s udy a ea.
En i onmen al
a iables
Desc ip ion Range and
uni s
Clima ic
A idi y index Ra io o mean annual p ecipi a ion o annual
po en ial e apo anspi a ion, o he pe iod
1950–2000
0.42–0.56
(uni less)
Summe
p ecipi a ion
P ecipi a ion o d ies qua e 17–34 mm
Win e
empe a u e
Mean empe a u e o he coldes qua e 9.4–11.3 °C
Topog aphic
Slope s anda d
de ia ion (slope
SD)
S anda d de ia ion o he slope wi hin a 250 m
bu e a ound he sampling si e (based on
digi al ele a ion models, 10 m esolu ion)
1.16–14.69°
Topog aphic
we ness index
(TWI)
Ln [(upslope a ea)/local slope], used o
quan i y opog aphic con ol on hyd ological
p ocesses
6.32–14.44
(uni less)
Edaphic
SOM Soil o ganic ma e con en 2.3–15.6%
C:N Soil ca bon/ni ogen a io 6.8–18.1
(uni less)
Sand Soil sand con en 17.9–71.4%
312 A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318
unc ional me ics would be in e - ela ed and la gely media ed by
changes in ela i e sh ub co e . Al hough we admi ha o he a p io i
model s uc u es could be conside ed, and ha assuming cause-e ec
ela ionships om obse a ional s udies has i s pi alls, he me ics o
goodness-o -fi o ou model, and ecological e idences d awn om
ou s and o he s udies, clea ly suppo he plausibili y o he ela ions
we p opose among a iables. Be o e modelling, we examined he dis i-
bu ions o all ou endogenous a iables, and es ed hei no mali y. The
fi s componen o he PCA on CWM alues was no malized (z-sco e)
be o e analyses. To es he o e all fi o SEM models we used he chi-
squa e s a is ic and i s significance ( he model is ejec ed i p- alue b
0.05), he compa a i e fi index (CFI N0.9), he oo mean squa e
e o o app oxima ion index (RMSEA b0.10) and he s anda dized
oo mean squa e esidual (SRMR b0.10). S anda dized pa h coe fi-
cien s es ima ed by maximum likelihood we e used o measu e he di-
ec and indi ec e ec s o he p edic o s (G ace e al., 2012). These
coe ficien s a e in e p e ed as he size o an e ec ha one a iable ex-
e s upon ano he . SEM models we e fi using he La aan package
(Rosseel, 2012). All he analyses we e pe o med unde R s a is ical en-
i onmen (R Co e Team, 2015). All he p edic o s used in modelling
we e s anda dized and no malized (z-sco e) be o e analyses.
3. Resul s
3.1. Rela i e sh ub co e
Va ia ion in ela i e sh ub co e (RSC) was mos ly explained by
opo-edaphic a iables (64% o he adj.
2
= 0.52), inc easing wi h
slope s anda d de ia ion and soil C:N a io, which showed addi i e e -
ec s (Table 2,Fig. 1).Lowe a idi y(highe alueso hea idi y
index) and lowe summe p ecipi a ion also con ibu ed o highe
RSC, al hough clima ic p edic o s explained a small p opo ion o i s
a ia ion along he clima ic g adien ei he pe se (15%), o oge he
wi h opo-edaphic ac o s (21%) (Table 2).
3.2. Func ional ai me ics
The PCA o he CWMs o all ai s seg ega ed wo main PCA compo-
nen s, which accoun ed o 57% o he o al a iance ound in he da a
(Fig. 2A). The fi s componen , he ea e called ‘CWM-PCA1’,explained
46% o he a iance and desc ibed a g adien o RSC, being nega i ely
co ela ed wi h i . Communi ies wi h lowe RSC showed a highe p o-
po ion o annuals (g aminoids and ose es), mo e anemocho y dis-
pe sal and highe SLA; communi ies wi h highe RSC showed highe
plan heigh and maximum oo dep h, mo e pe ennial species and
mo e ba ocho y dispe sal. This was shown by he s ong co ela ions
ound be ween CWM alues o hese ai s and he CWM-PCA1
(Fig. 2A, Table 3). The second componen , he ea e called ‘CWM-
PCA2’, explained 10% o he a iance. The ai s mo e s ongly co ela ed
wi h CWM-PCA2 we e flowe ing ai s (onse and du a ion), g ow h-
o m (e ec and p os a e), dispe sal s a egy (ec ozoocho y) and N-
fixing abili y (Fig. 2A, Table 3). Communi y FDis alues we e explained
by he wo main PCA componen s, accoun ing o 62% o he o al a i-
ance ound in he da a (Fig. 2B). The fi s componen , he ea e ‘FDis-
PCA1’, accoun ed o 42% o he a iance and sepa a ed communi ies
based on li e-cycle, oo dep h, heigh and SLA ai s, which showed
he s onges co ela ions wi h his axis (Fig. 2B, Table 3). The second
componen , he ea e ‘FDis-PCA2’, explained 20% o he a iance and
seg ega ed communi ies acco ding o flowe ing ai s (onse and du a-
ion), seed mass and g ow h- o m, all posi i ely co ela ed wi h his
axis (Fig. 2B, Table 3). The fi s componen s o bo h PCAs, i.e. CWM-
PCA1 and FDis-PCA1, we e mos ly explained by opo-edaphic ac o s
(64% and 76% o he o al a iance explained by he models, espec-
i ely), while he second PCA componen s, i.e., CWM-PCA2 and FDis-
PCA2 we e bes p edic ed by clima ic a iables (100% and 61% o he
o al a iance explained by he models, espec i ely) (Table 2).
3.3. S uc u al equa ion modelling (SEM)
Ou a p io y SEMs we e well fi ed o ou da a, as indica ed by
goodness-o -fi s a is ics (Fig. 3). They explained 54% o he a ia ion
Table 2
Summa y o he bes models selec ed o each esponse a iable (see Tables S7 and S8 o u he de ail). The sign o he coe ficien s o he selec ed p edic o s in each model is indica ed.
The p opo ion o a iance explained by opo-edaphic and clima ic p edic o s (and by bo h) was calcula ed using a a iance decomposi ion analysis based on each model.
Response
a iables
P edic o s P opo ion o he explained
a iance (%)
Topo-edaphic Clima ic Topo-edaphic × clima ic
Slope
SD
Soil
C/N
Slope SD × soil
C/N
Sand Soil C/N ×
Sand
A idi y Summe
p ecip.
Win e
emp.
Slope SD ×
summe
p ecip.
Ad.
R
2
Topo-edaphic Bo h Clima ic
Rela i e sh ub
co e
(+) (+) (−)(−) 0.52 64 21 15
CWM PCA 1 (−)(−) (+) (+) 0.55 64 14 22
CWM PCA 2 (−)(−) 0.20 0 0 100
FDis PCA 1 (+) (+) (−) (+) 0.42 76 17 7
FDis PCA 2 (−) (+) (−)(−) (+) 0.36 39 0 61
A idi y ep esen s – he alue o he a idi y index, so ha highe a idi y alues co espond o d ie condi ions, o acili a e esul s in e p e a ion.
The highes alues o he p opo ion o he explained a iance by di e en p edic o s o each esponse a iable a e highligh ed in bold.
Fig. 1. Rela i e sh ub co e a ia ion wi h slope s anda d de ia ion (slope SD) and soil C:N
a io. Planes ep esen he p edic ed alues o a linea eg ession fi ed o bo h a iables.
The colo s o he p edic ed planes change om blue (low alues o ela i e sh ub co e )
o ed (high alues o ela i e sh ub co e ).
313A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318
in RSC, which was mainly d i en by opo-edaphic ac o s. Slope s an-
da d de ia ion and soil C:N a io had a s ong di ec posi i e e ec on
RSC (Fig. 3). On he con a y, summe p ecipi a ion had a mode a e neg-
a i e di ec e ec on RSC. Changes in RSC la gely de e mined CWM-
PCA1 a ia ion, explaining 98% o i s a iance h ough a di ec nega i e
e ec (Fig. 3), showing ha he e ec s o opo-edaphic a iables and o
summe p ecipi a ion on CWM-PCA1 (Fig. S2) we e media ed by
changes in RSC. The CWM-PCA2 was ela ed only wi h clima ic ac o s,
al hough hey we e only able o explain 21% o i s a ia ion; summe
p ecipi a ion and win e empe a u e had a di ec s ong and mode a e
nega i e e ec on CWM-PCA2, espec i ely (Fig. 3). Topo-edaphic ac-
o s we e nega i ely ela ed o FDis-PCA1, explaining 64% o i s a ia-
ion; abou 3% o his e ec was di ec , h ough he in e ac ion
be ween slope s anda d de ia ion and soil C:N a io (al hough indi id-
ually hese ac o s had a ma ginal posi i e e ec ), and 97% indi ec ly,
ia CWM-PCA1 (Fig. 3). FDis-PCA2 was ela ed o clima ic ac o s and
also o CWM-PCA1, which join ly explained 38% o i s a ia ion; abou
87% o his e ec s was di ec , d i en by a ma ginal posi i e influence
o summe p ecipi a ion and by a nega i e e ec o a idi y, and 13%
was due o an indi ec posi i e e ec o CWM-PCA1 (Fig. 3).
4. Discussion
The causes o sh ub enc oachmen in Medi e anean Basin d ylands
need o be be e unde s ood and quan ified in o de o imp o e en-
c oachmen o ecas ing and he success o emo al p og ammes. He e,
Rela e sh ub co e
A
Rela e sh ub co e
B
Fig. 2. P incipal componen analysis (PCA) o A) communi y-weigh ed mean (CWM) and B) unc ional dispe sion (FDis) ai alues. Vec o s ep esen ai s desc ibed in Table S4.
Abb e ia ions: The = he ophy e; Phan = phane ophy e; Cham = chamaephy e; G am = g aminoid; Rose = ose e; E ec = e ec ; SLA = specific lea a ea; Anem =
anemocho y; Ba o = ba ocho y; Heigh = heigh ; Rdeph = oo dep h; Smass = seed mass; Slong = seed pe sis ence; Dispe s = dispe sal s a egy; Lcycle = li e cycle; Bflow =
onse o flowe ing; Dflow = du a ion o flowe ing. S udy si es a e ep esen ed by poin s in a g ay scale indica ing he class o ela i e sh ub co e (%) a each si e (see legend). See
Suppo ing in o ma ion Table S5 o co ela ions among CWM alues and Table S6 o co ela ions among FDis alues.
314 A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318

we show ha , unde simila low-in ensi y land use, sh ub enc oach-
men is la gely p edic ed by local opog aphic and edaphic ac o s,
and influenced o a much lesse ex en by clima e in he s udied com-
muni ies. Hence, in eg a ing opo-edaphic ac o s in o managemen
s a egies o p edic and e e sh ub enc oachmen may con ibu e o
imp o e hei cos -e ec i eness and sus ainabili y in he long un.
Also, acco ding o ou esul s, clima e change p edic ions o inc easing
a idi y o he Medi e anean Basin a e no expec ed o g ea ly impac
sh ub enc oachmen , bu may a ec pa icula unc ional ai s o he
plan communi y, he eby a ec ing ecosys em unc ioning.
4.1. Main d i e s o sh ub enc oachmen and implica ions o managemen
Ou findings pa ially ma ch ou fi s hypo hesis, as bo h opo-
edaphic and clima ic a iables influenced sh ub enc oachmen , join ly
accoun ing o 52% o he a ia ion in ela i e sh ub co e . Topo-
edaphic ac o s explained mos o he a ia ion in ela i e sh ub co e
(64% o he
2
), which inc eased owa ds a eas wi h highe slope s an-
da d de ia ion, and a eas wi h highe soil C:N a io. In d ylands, wa e
limi a ions cons ain ege a ion co e , a o ing uno and e osion in
slopes. Hence, opog aphy la gely con ols wa e and nu ien flow
pa hs om upslope o lowlands, whe e hey end o accumula e
(Gómez-Plaza e al., 2001). This leads o coa se - ex u ed soils on
hillslopes, wi h less abili y o e ain wa e a uppe soil dep hs, a o ing
deep- oo ed species such as sh ubs. Con e sely, plain a eas end o ha e
fine - ex u ed soils, and shallowe wa e p ofiles, a o ing shallow-
oo ed species such as mos he baceous species (Sala e al., 1997;
McAuli e, 2003).
Soil C:N a io, which showed no significan co ela ion wi h slope
s anda d de ia ion, nei he wi h a idi y (Table S3), was also an impo -
an p edic o o sh ub enc oachmen . The soil C:N a io eflec s he dy-
namic in e ac ion be ween slow-modi ying soil ea u es such as pa icle
size dis ibu ion and mine alogy, and local si e cha ac e is ics like hy-
d ology, and ege a ion. A highe soil C:N a io in sh ub-enc oached
a eas may indica e a lowe decomposi ion a e, p obably due o lowe
wa e a ailabili y. Ye , his assump ion would need o he di ec indica-
o s o soil o ganic ca bon u no e (e.g. mic obial biomass) o be con-
fi med. Addi ionally, a high soil C:N in sh ub-domina ed a eas may
also be a esul o highe inpu s o abo e and belowg ound li e om
sh ubs, which ha e a usually highe C:N a io and lignin con en , ha
ake longe o decompose, han ha om he baceous species (Bo and
Beni es, 2005). These condi ions may c ea e a posi i e eed-back loop
p omo ing u he sh ub enc oachmen .
One o he s a egies mos used o deal wi h enc oachmen is me-
chanical sh ub emo al, along wi h he use o he bicides o p esc ibed
bu ning (A che , 2010). Ye , hese echniques o en ail i s goals in he
long un (Rango e al., 2005;A che and P edick, 2014), calling o a be -
e unde s anding o he causes o sh ub enc oachmen unde di e en
con ex s. Conside ing he impo ance o opog aphic and edaphic ac-
o s as p edic o s o sh ub enc oachmen in Medi e anean Basin d y-
lands, we sugges ha combining sh ub clea ing wi h o he locally
applied echniques ha ake in o accoun opo-edaphic ac o s, in ol -
ing he manipula ion o wa e and nu ien flow pa hways and accumu-
la ion, migh con ibu e o hei longe - e m e ec i eness. The e a e
Table 3
Pea son co ela ion coe ficien s be ween communi y-weigh ed-means (CWM) and he
wo fi s componen s o he espec i e p incipal componen analysis (CWM-PCA1 and
CWM-PCA2); and be ween unc ional dispe sion (FDis) and he wo fi s componen s o
he espec i e p incipal componen analysis (FDis-PCA1 and FDis-PCA2).
T ai s Ca ego y CWM FDis
PCA1 PCA2 PCA1 PCA2
Li e cycle −0.99⁎⁎⁎ −0.02 0.95⁎⁎⁎ 0.04
Li e- o m The ophy e 0.96⁎⁎⁎ 0.10 0.78⁎⁎⁎ 0.16
Hemic yp oph 0.01 −0.35⁎⁎
Geophy e 0.04 0.05
Chamaephy e −0.34⁎0.34⁎
Phane ophy e −0.98⁎⁎⁎ −0.04
G ow h- o m Bulb 0.04 0.05 0.16 0.68⁎⁎⁎
E ec 0.44⁎⁎
−0.37⁎⁎
G aminoid 0.73⁎⁎⁎ 0.24
P os a e 0.26 −0.74⁎⁎⁎
Rose e 0.62⁎⁎⁎ 0.19
Sh ub −0.99⁎⁎⁎ 0.03
Max. heigh −0.94⁎⁎⁎ 0.10 0.87⁎⁎⁎ −0.05
SLA 0.94⁎⁎⁎ 0.06 0.84⁎⁎⁎
−0.28⁎
Onse flowe . −0.37⁎⁎ 0.67⁎⁎⁎ −0.20 0.79⁎⁎⁎
Du a ion flowe . 0.08 −0.52⁎⁎⁎ −0.08 0.81⁎⁎⁎
Dispe s. s a egy Anemocho y 0.95⁎⁎⁎ 0.23 0.79⁎⁎⁎ 0.23
Ba ocho y −0.95⁎⁎⁎ −0.15
Ec ozoocho y 0.16 −0.57⁎⁎⁎
Endozoocho y −0.15 −0.08
Seed mass −0.24 0.04 −0.26 0.65⁎⁎⁎
Seed pe sis ence −0.79⁎⁎⁎ 0.07 0.73⁎⁎⁎ 0.08
N-fixing abili y −0.32⁎
−0.48⁎⁎⁎ 0.13 0.04
Max. oo dep h −0.95⁎⁎⁎ 0.00 0.89⁎⁎⁎ 0.17
Significan co ela ions a e highligh ed in bold.
⁎pb0.05.
⁎⁎ pb0.01.
⁎⁎⁎ pb0.001.
Fig. 3. S uc u al equa ion model o explain ela i e sh ub co e and he main axis o unc ional specializa ion o he plan communi y ega ding unc ional s uc u e (CWM-PCA1 and
CWM-PCA2) and unc ional dispe sion (FDis-PCA1 and FDis-PCA2). O e all goodness-o -fi s a is ics: χ
2(23)
= 26.668, p = 0.270, oo mean squa e e o o app oxima ion (RMSEA) =
0.054 (0.00–0.121), compa a i e fi index (CFI) = 0.989, s anda dized oo mean squa e esidual (SRMR) = 0.056. A ow wid hs a e p opo ional o he s anda dized pa h
coe ficien s, which a e p esen ed. The R
2
nex o esponse a iables indica es he p opo ion o a iance explained. *p b0.05; **p b0.01; ***p b0.001. A idi y = -A idi y index.
315A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318
many o such echniques (e.g. c oss-slope ba ie s like e aces, s one
bunds, wa e -ha es ing s uc u es, manu e sp eading) p oposed o
sus ainable land managemen in d ylands which show some e idence
o success in inc easing wa e and nu ien a ailabili y o plan s
(Schwilch e al., 2012), al hough i s e ec i eness in dealing wi h
sh ub enc oachmen emains o be es ed. A sus ainable land manage-
men in d ylands, and imp o ed s a egies o p omo e he eco e y o
deg aded land e.g. h ough soil and wa e conse a ion, is among he
aims o he UN sus ainable de elopmen goals, and ou findings con ib-
u e o app oach hese goals.
Topo-edaphic and clima ic a iables join ly explained 52% o he a -
ia ion in ela i e sh ub co e , sugges ing ha i may be influenced by
o he ac o s, such as bio ic in e ac ions o pas land managemen . In
Holm-oak woodlands, plain a eas may be mo e g azed by li es ock
(pa icula ly cows) han slopes, because hey ha e in gene al mo e p o-
duc i e soils, p o iding highe quan i y and quali y o o age (Bailey
e al., 1996). In addi ion, occasional mechanical sh ub clea ing ha
may ha e occu ed in he pas , is easie o pe o m in fla e a eas
han in slopes. Hence, al hough a simila low-in ensi y land-use was
one o he c i e ia used o selec sampling si es, hese ac o s may
ha e pa ially con ibu ed o a highe sh ub co e ound in a eas wi h
highe slope a ia ion.
Clima e was a weak p edic o o sh ub enc oachmen . In ligh o his
esul , clima e change o ecas s o inc eased a idi y o Medi e anean
Basin d ylands (Huang e al., 2016) a e no expec ed o g ea ly a ec
sh ub enc oachmen . S ill, ela i e sh ub co e dec eased wi h inc eas-
ing summe p ecipi a ion ac oss he s udied communi ies. This may be
because p ecipi a ion du ing he wa me season (summe ) usually
leads o low pe cola ion o wa e in o he soil, hus a o ing
shallowe - oo ed he baceous species in de imen o sh ubs (Sala
e al., 1997). P e ious s udies based on manipula i e expe imen s
ound inc eased woody plan enc oachmen wi h highe p ecipi a ion
in ensi y (sho e e en s wi h highe p ecipi a ion), bu no wi h he
o al p ecipi a ion amoun (Kulma iski and Bea d, 2013). The au ho s
a gue ha a highe p ecipi a ion in ensi y can push soil wa e deepe
in o he soil, in sh ub oo ing dep h, hus a o ing woody species o e
g asses. The in ensi y and equency o ex eme p ecipi a ion e en s
and in a and in e -annual p ecipi a ion fluc ua ions a e likely o in-
c ease unde a clima e change scena io (IPCC, 2014). Howe e , as we
did no assess a ia ions in p ecipi a ion dis ibu ion in ou obse a-
ional field s udy, we canno confi m no disca d i s po en ial e ec on
sh ub enc oachmen , based on ou findings. Also, since ou s udy was
de eloped wi hin a egional clima ic g adien whe e he a idi y index
anged om 0.42 o 0.56, u he s udies along b oade clima ic g adi-
en s comp ising mo e ex eme a idi y le els would be needed o ully
assess he po en ial e ec o clima e change on sh ub enc oachmen .
4.2. Consequences o sh ub enc oachmen o communi y unc ional ai s
Changes in ela i e sh ub co e explained mos o he a ia ion in
he unc ional s uc u e and dispe sion o he plan communi ies.
Sh ub enc oachmen was associa ed wi h changes in CWMs o ai s di -
e en ia ing pe ennial alle sh ubs wi h low SLA and la ge seeds, om
he baceous sho -li ed and mainly anemocho ous species wi h high
SLA. These CWM a ia ions may be seen as axes o unc ional specializa-
ion om a esou ce-conse a i e o a esou ce-acquisi i e s a egy o
cope e.g. wi h wa e limi a ions (Díaz e al., 2016). In e es ingly, unc-
ional dispe sion o he same ai s, pa icula ly o li e-cycle, oo
dep h, heigh and SLA, inc eased wi h sh ub enc oachmen . This is
p obably due o he co-occu ence o he baceous and sh ub species in
a eas wi h highe ela i e sh ub co e , inc easing unc ional di e si y,
when compa ed o g asslands de oid o sh ubs. In addi ion, o he au-
ho s ha e ound ha a high le els o sh ub co e , only he baceous spe-
cies wi h con as ing ai alues a e p esen (Rolo e al., 2016), which
would con ibu e o a highe unc ional dispe sion wi h inc easing
sh ub co e . The e ec o opo-edaphic a iables and, o a lesse ex en ,
o clima ic a iables, in explaining a ia ion along CWM-PCA1 and FDis-
PCA1 axes was i ually all due o changes in ela i e sh ub co e . This
indica es ha main changes in plan communi y unc ional ai s we e
mos ly d i en indi ec ly by local opo-edaphic fil e s, h ough changes
in ela i e sh ub co e .
4.3. E ec s o clima e on communi y unc ional ai s
Clima e a ec ed mos ly a se o ai s no closely ela ed o changes
in ela i e sh ub co e , desc ibed by he second componen s o CWM
and FDis PCAs. P ecipi a ion and empe a u e a ec many plan ai s
(Moles e al., 2014), and can influence he iming o phenophases
(Peñuelas e al., 2004). Highe summe p ecipi a ion and highe win e
empe a u e led o an ea lie flowe ing and longe flowe ing du a ion.
This was p obably due o an a enua ion o wa e and empe a u e lim-
i a ions, espec i ely, which may p o ide a o able condi ions o lon-
ge g ow h and flowe ing pe iods (C immins e al., 2013;Ramos e al.,
2015). P e ious s udies epo ed an ea lie onse o flowe ing igge ed
by highe mean annual empe a u es (Mille -Rushing and P imack,
2008), suppo ing ou esul s. Mo e a o able clima ic condi ions also
inc eased he CWM o p os a e and e ec g ow h- o ms, leading o a
highe FDis o g ow h- o m, and inc eased he CWM o ec ozoocho ous
and N-fixing species. A enua ion o wa e and empe a u e limi a ions
in summe and win e , espec i ely, may allow a highe ep oduc i e
success o hese species. N-fixing species, in pa icula , seem o be
highly ulne able o d ough du ing hei ep oduc i e phase, which
may sho en he du a ion o ep oduc i e de elopmen , educing seed
numbe and weigh (Da yan o e al., 2015).
Highe a idi y le els and lowe summe p ecipi a ion we e associ-
a ed wi h a lowe FDis o flowe ing ai s and o seed mass. A lowe
FDis is expec ed o eflec a lowe complemen a y in esou ce use be-
ween species, sugges ing a educ ion in ecosys em unc ioning
(Mouillo e al., 2011;Valencia e al., 2015) and esilience o en i on-
men al change (Volai e e al., 2014). We should also no ice ha ou
s udy was conduc ed in a ela i ely d y yea , which may ha e led o a
lowe unc ional dispe sion o ege a i e ai s such as SLA and heigh ,
pa icula ly o annual species (Ca mona e al., 2015), possibly
p e en ing he de ec ion o changes on he FDis o hose ai s along
he spa ial a idi y g adien .
5. Concluding ema ks
The causes o sh ub enc oachmen need o be be e unde s ood and
quan ified in he Medi e anean Basin, o imp o e enc oachmen o e-
cas ing and he cos -e ec i eness and sus ainabili y o managemen
s a egies, and ou wo k con ibu es o fill his knowledge gap. We
show ha , unde simila low-in ensi y land use, and despi e he likely
influence o o he ac o s no add essed in his s udy (e.g. pas land
managemen ), opog aphic and edaphic ac o s we e he bes p edic-
o s o sh ub enc oachmen , influenced o a much lesse ex en by cli-
ma e, join ly explaining 52% o he a ia ion in ela i e sh ub co e .
Thus, we sugges ha moni o ing and managemen ac ions o educe
sh ub enc oachmen ha ake in o accoun opo-edaphic ac o s, in-
ol ing, o ins ance, he manipula ion o wa e and nu ien flow pa h-
ways and accumula ion, a e mo e likely o sus ainably succeed, han
sh ub emo al alone.
Clima e was a weak p edic o o sh ub enc oachmen , a ec ing
mos ly a se o unc ional ai s no so di ec ly in ol ed in sh ub coloni-
za ion. In ligh o his esul , clima e change p edic ions o inc eased
a idi y in he Medi e anean Basin a e no expec ed o g ea ly impac
sh ub enc oachmen . Ye , hey may a ec he unc ional s uc u e and
di e si y o plan communi ies, comp omising ecosys em unc ioning.
This wo k p o ides impo an ad ances o be e p edic sh ub en-
c oachmen now and unde a clima e change scena io, and o help de-
fine mo e cos -e ec i e and sus ainable s a egies o deal wi h i in
Medi e anean ecosys ems.
316 A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318
Acknowledgemen s
We kindly acknowledge he help o Susana Tápia du ing field and
labo a o y wo k and And eia Fe ei a du ing field wo k. This esea ch
was unded by EAA G an s - P og ama AdaPT and by FCT-MEC h ough:
p ojec PTDC/AAG-GLO/0045/2014; A.N. by con ac SFRH/BD/51407/
2011, P.P. by con ac SFRH/BPD/75425/2010. The p epa a ion o he pa-
pe was suppo ed by Ni oPo ugal p ojec [Eu opean Union's Ho izon
2020 esea ch and inno a ion p og amme unde g an ag eemen No
692331]. Ni oPo ugal is a co-finance o he p ojec Towa ds INMS,
implemen ed by UN En i onmen .
Da a accessibili y
Da a a ailable om D yad Digi al Reposi o y.
Appendix A. Supplemen a y da a
Supplemen a y da a o his a icle can be ound online a h ps://doi.
o g/10.1016/j.sci o en .2018.11.475.
Re e ences
Ama al, M.R., Bo ges, C., San os, G., 1997. Mon ados de sob o e azinho: es a egias pa a o
séc. XXI. Cong esos Fo es ales.
Anadón, J.D., Sala, O.E., Tu ne , B.L., Benne , E.M., 2014. E ec o woody-plan enc oach-
men on li es ock p oduc ion in No h and Sou h Ame ica. P oc. Na l. Acad. Sci. 111
(35), 12948–12953.
A che , S.R., 2010. Rangeland conse a ion and sh ub enc oachmen : new pe spec i es
on an old p oblem. Wild Rangelands: Conse ing Wildli e While Main aining Li e-
s ock in Semi-a id Ecosys ems, pp. 53–97.
A che , S.R., P edick, K.I., 2014. An ecosys em se ices pe spec i e on b ush managemen :
esea ch p io i ies o compe ing land-use objec i es. J. Ecol. 102 (6), 1394–1407.
Bailey, D.W., G oss, J.E., Laca, E.A., Ri enhouse, L.R., Coughenou , M.B., Swi , D.M., Sims,
P.L., 1996. Mechanisms ha esul in la ge he bi o e g azing dis ibu ion pa e ns.
J. Range Manag. 49, 386–400.
Blois, J.L., Williams, J.W., Fi zpa ick, M.C., Jackson, S.T., Fe ie , S., 2013. Space can subs i-
u e o ime in p edic ing clima e-change e ec s on biodi e si y. P oc. Na l. Acad. Sci.
110, 9374–9379.
Bo , A., Beni es, J., 2005. The Impo ance o Soil O ganic Ma e : Key o D ough - esis an
Soil and Sus ained Food P oduc ion. Food & Ag icul u e O g.
Caldei a, M.C., Lecom e, X., Da id, T.S., Pin o, J.G., Bugalho, M.N., We ne , C., 2015. Syne gy
o ex eme d ough and sh ub in asion educe ecosys em unc ioning and esilience
in wa e -limi ed clima es. Sci. Rep. 5, 15110.
Ca mona, C.P., Mason, N.W., Azcá a e, F.M., Peco, B., 2015. In e -annual fluc ua ions in
ain all shi he unc ional s uc u e o Medi e anean g asslands ac oss g adien s
o p oduc i i y and dis u bance. J. Veg. Sci. 26 (3), 538–551.
Cas o, H., F ei as, H., 2009. Abo e-g ound biomass and p oduc i i y in he Mon ado:
om he baceous o sh ub domina ed communi ies. J. A id En i on. 73, 506–511.
Colgan, M., Asne , G., Le ick, S.R., Ma in, R., Chadwick, O., 2012. Topo-edaphic con ols
o e woody plan biomass in Sou h A ican sa annas. Biogeosciences 9, 1809–1821.
C immins, T.M., C immins, M.A., Be elsen, C.D., 2013. Sp ing and summe pa e ns in
flowe ing onse , du a ion, and cons ancy ac oss a wa e -limi ed g adien . Am.
J. Bo . 100, 1137–1147.
Da yan o, S., Eld idge, D.J., Th oop, H.L., 2013. Managing semi-a id woodlands o ca bon
s o age: g azing and sh ub e ec s on abo e-and belowg ound ca bon. Ag ic. Ecosys .
En i on. 169, 1–11.
Da yan o, S., Wang, L., Jacin he, P.-A., 2015. Global syn hesis o d ough e ec s on ood le-
gume p oduc ion. PLoS One 10, e0127401.
de Bello, F., La o el, S., Diaz, S., Ha ing on, R., Co nelissen, J.H.C., Ba dge , R.D., Be g, M.P.,
Cip io i, P., Feld, C.K., He ing, D., da Sil a, P.M., Po s, S.G., Sandin, L., Sousa, J.P.,
S o key, J., Wa dle, D.A., Ha ison, P.A., 2010. Towa ds an assessmen o mul iple eco-
sys em p ocesses and se ices ia unc ional ai s. Biodi e s. Conse . 19,
2873–2893.
Díaz, S., La o el, S., de Bello, F., Que ie , F., G igulis, K., Robson, M., 2007. Inco po a ing
plan unc ional di e si y e ec s in ecosys em se ice assessmen s. P oc. Na l. Acad.
Sci. U. S. A. 104, 20684–20689.
Díaz, S., Ka ge, J., Co nelissen, J.H.C., W igh , I.J., La o el, S., D ay, S., Reu, B., Kleye , M.,
Wi h, C., Colin P en ice, I., Ga nie , E., Bönisch, G., Wes oby, M., Poo e , H., Reich,
P.B., Moles, A.T., Dickie, J., Gillison, A.N., Zanne, A.E., Cha e, J., Joseph W igh , S.,
She eme 'e , S.N., Jac el, H., Ba alo o, C., Ce abolini, B., Pie ce, S., Shipley, B., Ki kup,
D., Casano es, F., Joswig, J.S., Gün he , A., Falczuk, V., Rüge , N., Mahecha, M.D.,
Go né, L.D., 2016. The global spec um o plan o m and unc ion. Na u e 529
(7585), 167.
D'odo ico, P., Okin, G.S., Bes elmeye , B.T., 2012. A syn he ic e iew o eedbacks and
d i e s o sh ub enc oachmen in a id g asslands. Ecohyd ology 5, 520–530.
Dubuis, A., Rossie , L., Po ie , J., Pellissie , L., Vi oz, P., Guisan, A., 2013. P edic ing cu en
and u u e spa ial communi y pa e ns o plan unc ional ai s. Ecog aphy 36,
1158–1168.
Eld idge, D.J., Soli e es, S., 2015. A e sh ubs eally a sign o declining ecosys em unc ion?
Disen angling he my hs and u hs o woody enc oachmen in Aus alia. Aus . J. Bo .
62, 594–608.
Eld idge, D.J., Bowke , M.A., Maes e, F.T., Roge , E., Reynolds, J.F., Whi o d, W.G., 2011.
Impac s o sh ub enc oachmen on ecosys em s uc u e and unc ioning: owa ds a
global syn hesis. Ecol. Le . 14, 709–722.
ESRI, 2010. A cMap . 10.1. ESRI.
Ga nie , E., La o el, S., Ansque , P., Cas o, H., C uz, P., Dolezal, J., E iksson, O., Fo unel, C.,
F ei as, H., Golode s, C., 2007. Assessing he e ec s o land-use change on plan ai s,
communi ies and ecosys em unc ioning in g asslands: a s anda dized me hodology
and lessons om an applica ion o 11 Eu opean si es. Ann. Bo . 99, 967–985.
Gómez-Plaza, A., Ma ınez-Mena, M., Albaladejo, J., Cas illo, V., 2001. Fac o s egula ing
spa ial dis ibu ion o soil wa e con en in small semia id ca chmen s. J. Hyd ol.
253, 211–226.
Gómez-Rey, M.X., Madei a, M., Gonzalez-P ie o, S.J., Cou inho, J., 2013. Soil C and N dy-
namics in a Medi e anean oak woodland wi h sh ub enc oachmen . Plan Soil 371,
339–354.
G ace, J.B., Schoolmas e , D.R., Gun enspe gen, G.R., Li le, A.M., Mi chell, B.R., Mille , K.M.,
Schweige , E.W., 2012. Guidelines o a g aph- heo e ic implemen a ion o s uc u al
equa ion modeling. Ecosphe e 3, 1–44.
G oss, N., Boe ge , L., So iano-Mo ales, S.I., Le Bagousse-Pingue , Y., Que o, J.L., Ga cia-
Gomez, M., Valencia-Gomez, E., Maes e, F.T., 2013. Unco e ing mul iscale e ec s o
a idi y and bio ic in e ac ions on he unc ional s uc u e o Medi e anean
sh ublands. J. Ecol. 101, 637–649.
Hijmans, R.J., Came on, S.E., Pa a, J.L., Jones, P.G., Ja is, A., 2005. Ve y high esolu ion in-
e pola ed clima e su aces o global land a eas. In . J. Clima ol. 25, 1965–1978.
Hi o a, M., Holmg en, M., Van Nes, E.H., Sche e , M., 2011. Global esilience o opical
o es and sa anna o c i ical ansi ions. Science 334, 232–235.
Huang, J., Yu, H., Guan, X., Wang, G., Guo, R., 2016. Accele a ed d yland expansion unde
clima e change. Na . Clim. Chang. 6 (2), 166.
Iannone, B.V., Heneghan, L., Rijal, D., Wise, D.H., 2015. Below-g ound causes and conse-
quences o woodland sh ub in asions: a no el pai ed-poin amewo k e eals new
insigh s. J. Appl. Ecol. 52, 78–88.
IPCC, 2014. In: Co e W i ing Team, Pachau i, R.K., Meye , L.A. (Eds.), Clima e Change
2014: Syn hesis Repo . Con ibu ion o Wo king G oups I, II and III o he Fi h As-
sessmen Repo o he In e go e nmen al Panel on Clima e Change. IPCC, Gene a,
Swi ze land (151 pp.).
Jackson, R.B., Banne , J.L., Jobbágy, E.G., Pockman, W.T., Wall, D.H., 2002. Ecosys em ca bon
loss wi h woody plan in asion o g asslands. Na u e 418 (6898), 623.
Knapp, A.K., B iggs, J.M., Collins, S.L., A che , S.R., B e -Ha e, M.S., Ewe s, B.E., ... Clea y,
M.B., 2008. Sh ub enc oachmen in No h Ame ican g asslands: shi s in g ow h
o m dominance apidly al e s con ol o ecosys em ca bon inpu s. Glob. Chang.
Biol. 14 (3), 615–623.
Kulma iski, A., Bea d, K.H., 2013. Woody plan enc oachmen acili a ed by inc eased p e-
cipi a ion in ensi y. Na . Clim. Chang. 3 (9), 833.
Lalibe e, E., Legend e, P., 2010. A dis ance-based amewo k o measu ing unc ional di-
e si y om mul iple ai s. Ecology 91, 299–305.
Lalibe é, E., Legend e, P., Shipley, B., 2014. FD: Measu ing Func ional Di e si y F om Mul-
iple T ai s, and O he Tools o Func ional Ecology. R Package Ve sion 1.0-12.
López-Díaz, M.L., Rolo, V., Bení ez, R., Mo eno, G., 2015. Sh ub enc oachmen o Ibe ian
dehesas: implica ions on o al o age p oduc i i y. Ag o o . Sys . 89 (4), 587–598.
Maes e, F.T., Co ina, J., 2005. Remnan sh ubs in Medi e anean semi-a id s eppes: e -
ec s o sh ub size, abio ic ac o s and species iden i y on unde s o ey ichness and
occu ence. Ac a Oecol. 27 (3), 161–169.
Maes e, F.T., Bowke , M.A., Puche, M.D., Belén Hinojosa, M., Ma ínez, I., Ga cía-Palacios,
P., Cas illo, A.P., Soli e es, S., Luzu iaga, A.L., Sánchez, A.M., 2009. Sh ub enc oachmen
can e e se dese ifica ion in semi-a id Medi e anean g asslands. Ecol. Le . 12,
930–941.
Maes e, F.T., Eld idge, D.J., Soli e es, S., 2016. A mul i ace ed iew on he impac s o
sh ub enc oachmen . Appl. Veg. Sci. 19, 369–370.
McAuli e, J., 2003. The a mosphe e–biosphe e in e ace: he impo ance o soils in a id
and semi-a id en i onmen s. Changing P ecipi a ion Regimes in Te es ial Ecosys-
ems: A No h Ame ican Pe spec i e. Uni e si y o A izona P ess, Tucson, A iz,
pp. 9–27.
MEA, 2005. Ecosys ems and Human Well-being: Dese ifica ion Syn hesis. Wo ld Re-
sou ces Ins i u e.
Middle on, N., Thomas, D., 1992. Wo ld A las o Dese ifica ion: Uni ed Na ions En i on-
men al P og amme. A nold.
Mille -Rushing, A.J., P imack, R.B., 2008. Global wa ming and flowe ing imes in Tho eau's
Conco d: a communi y pe spec i e. Ecology 89, 332–341.
Moles, A.T., Pe kins, S.E., La an, S.W., Flo es-Mo eno, H., Awas hy, M., Tindall, M.L., Sack,
L., Pi man, A., Ka ge, J., Aa ssen, L.W., 2014. Which is a be e p edic o o plan ai s:
empe a u e o p ecipi a ion? J. Veg. Sci. 25, 1167–1180.
Mouillo , D., Villege , S., Sche e -Lo enzen, M., Mason, N.W.H., 2011. Func ional s uc u e
o biological communi ies p edic s ecosys em mul i unc ionali y. PLoS One 6.
Nunes, A., Tápia, S., Pinho, P., Co eia, O., B anquinho, C., 2014. Ad an ages o he poin -
in e cep me hod o assessing unc ional di e si y in semi-a id a eas. iFo es
Biogeosci. Fo . 884.
Nunes, A., Köbel, M., Pinho, P., Ma os, P., de Bello, F., Co eia, O., B anquinho, C., 2017.
Which plan ai s espond o a idi y? A c i ical s ep o assess unc ional di e si y
in Medi e anean d ylands. Ag ic. Fo . Me eo ol. 239, 176–184.
Pakeman, R.J., Ques ed, H.M., 2007. Sampling plan unc ional ai s: wha p opo ion o
he species need o be measu ed? Appl. Veg. Sci. 10, 91–96.
Peñuelas, J., Filella, I., Zhang, X., Llo ens, L., Ogaya, R., Llo e , F., Comas, P., Es ia e, M.,
Te adas, J., 2004. Complex spa io empo al phenological shi s as a esponse o ain-
all changes. New Phy ol. 161, 837–846.
317A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318
Pe ei a, P.M., Da Fonseca, M.P., 2003. Na u e s. nu u e: he making o he mon ado eco-
sys em. Conse . Ecol. 7 (3).
Pé ez-Ha guindeguy, N., Díaz, S., Ga nie , E., La o el, S., Poo e , H., Jau eguibe y, P., B e -
Ha e, M., Co nwell, W., C aine, J., Gu ich, D., 2013. New handbook o s anda dised
measu emen o plan unc ional ai s wo ldwide. Aus . J. Bo . 61, 167–234.
R Co e Team, 2015. R: A Language and En i onmen o S a is ical Compu ing. URL. h p://
www.R-p ojec .o g (Vienna, Aus ia).
Ramos, A., Pe ei a, M.J., Soa es, A., do Rosá io, L., Ma os, P., Nunes, A., B anquinho, C.,
Pinho, P., 2015. Seasonal pa e ns o Medi e anean e e g een woodlands (Mon ado)
a e explained by long- e m p ecipi a ion. Ag ic. Fo . Me eo ol. 202, 44–50.
Rango, A., Huenneke, L., Buonopane, M., He ick, J., Ha s ad, K., 2005. Using his o ic da a
o assess e ec i eness o sh ub emo al in sou he n New Mexico. J. A id En i on. 62,
75–91.
Ri es , D., Rolo, V., López-Díaz, L., Mo eno, G., 2011. Sh ub enc oachmen in Medi e a-
nean sil opas o al sys ems: Re ama sphae oca pa and Cis us ladani e induce con-
as ing e ec s on pas u e and Que cus ilex p oduc ion. Ag ic. Ecosys . En i on. 141
(3-4), 447–454.
Rolo, V., López-Díaz, M.L., Mo eno, G., 2012. Sh ubs a ec soil nu ien s a ailabili y wi h
con as ing consequences o pas u e unde s o y and ee o e s o y p oduc ion and
nu ien s a us in Medi e anean g azed open woodlands. Nu . Cycl. Ag oecosys .
93 (1), 89–102.
Rolo, V., Ri es , D., Lo en e, M., Ka ge, J., Mo eno, G., 2016. Taxonomic and unc ional di-
e si y in Medi e anean pas u es: insigh s on he biodi e si y–p oduc i i y ade-
o . J. Appl. Ecol. 53 (5), 1575–1584.
Roques, K., O'conno , T., Wa kinson, A., 2001. Dynamics o sh ub enc oachmen in an
A ican sa anna: ela i e influences o fi e, he bi o y, ain all and densi y depen-
dence. J. Appl. Ecol. 38, 268–280.
Rosseel, Y., 2012. la aan: an R package o s uc u al equa ion modeling. J. S a . So w. 48,
1–36.
Sala, O.E., Maes e, F.T., 2014. G ass–woodland ansi ions: de e minan s and conse-
quences o ecosys em unc ioning and p o isioning o se ices. J. Ecol. 102,
1357–1362.
Sala, O., Lauen o h, W., Golluscio, R., 1997. Plan unc ional ypes in empe a e semi-a id
egions. Plan Func ional Types: Thei Rele ance o Ecosys em P ope ies and Global
Change, pp. 217–233.
Sanka an, M., Hanan, N.P., Scholes, R.J., Ra nam, J., Augus ine, D.J., Cade, B.S., ... A do, J.,
2005. De e minan s o woody co e in A ican sa annas. Na u e 438 (7069), 846.
Schlesinge , W.H., Reynolds, J.F., Cunningham, G.L., Huenneke, L., Ja ell, W., Vi ginia, R.,
Whi o d, W., 1990. Biological eedbacks in global dese ifica ion. Science 247,
1043–1048.
Schwilch, G., Hessel, R., Ve zand oo , S. (Eds.), 2012. Desi e o G eene Land. Op ions o
Sus ainable Land Managemen in D ylands. Uni e si y o Be n - CDE, Al e a -
Wageningen UR, ISRIC - Wo ld Soil In o ma ion and CTA - Technical Cen e o Ag i-
cul u al and Ru al Coope a ion, Be n, Swi ze land, and Wageningen, The
Ne he lands.
Soli e es, S., Maes e, F.T., Eld idge, D.J., Delgado-Baque izo, M., Que o, J.L., Bowke , M.A.,
Galla do, A., 2014. Plan di e si y and ecosys em mul i unc ionali y peak a in e me-
dia e le els o woody co e in global d ylands. Glob. Ecol. Biogeog . 23 (12),
1408–1416.
Sø ensen, R., Zinko, U., Seibe , J., 2005. On he calcula ion o he opog aphic we ness
index: e alua ion o di e en me hods based on field obse a ions. Hyd ol. Ea h
Sys . Sci. Discuss. 2, 1807–1834.
Tilman, D., Knops, J., Wedin, D., Reich, P., Ri chie, M., Siemann, E., 1997. The influence o
unc ional di e si y and composi ion on ecosys em p ocesses. Science 277,
1300–1302.
T abucco, A., Zome , R., 2009. Global a idi y index (global-a idi y) and global po en ial
e apo- anspi a ion (global-PET) geospa ial da abase. CGIAR Conso ium o Spa ial
In o ma ion Published online, a ailable om he CGIAR-CSI GeoPo al a :. h p://
www.csi.cgia .o g/.
Valencia, E., Maes e, F.T., Le Bagousse-Pingue , Y., Luis Que o, J., Tamme, R., Boe ge , L.,
Ga cia-Gomez, M., G oss, N., 2015. Func ional di e si y enhances he esis ance o
ecosys em mul i unc ionali y o a idi y in Medi e anean d ylands. New Phy ol.
206, 660–671.
Van Auken, O., 2009. Causes and consequences o woody plan enc oachmen in o wes -
e n No h Ame ican g asslands. J. En i on. Manag. 90, 2931–2942.
Volai e, F., Ba kaoui, K., No on, M., 2014. Designing esilien and sus ainable g asslands
o a d ie u u e: adap i e s a egies, unc ional ai s and bio ic in e ac ions. Eu .
J. Ag on. 52, 81–89.
318 A. Nunes e al. / Science o he To al En i onmen 657 (2019) 310–318