Con en s lis s a ailable a ScienceDi ec
Fo es Ecology and Managemen
jou nal homepage: www.else ie .com/loca e/ o eco
Na ional-scale assessmen o o es si e p oduc i i y in Spain
Daniel Mo eno-Fe nández
a,b,⁎
, Juan Gab iel Ál a ez-González
c
, Roque Rod íguez-Soallei o
d
,
Ma ía Pasalodos-Ta o
a
, Isabel Cañellas
a
, Fe nando Mon es
a
, Emilio Díaz-Va ela
e
,
Ma iola Sánchez-González
a
, Felipe C ecen e-Campo
, Ped o Ál a ez-Ál a ez
g
,
Ma cos Ba io-An a
g
, Césa Pé ez-C uzado
c
a
INIA-CIFOR, C a. A Co uña km 7.5, E-28040 Mad id, Spain
b
MONTES (School o Fo es Enginee ing and Na u al Resou ces), Uni e sidad Poli écnica de Mad id, 28040 Mad id, Spain
c
Depa men o Ag o o es y Enginee ing, Uni o Sus ainable Fo es Managemen (UXFS), School o Enginee ing, Uni e si y o San iago de Compos ela E-27002 Lugo,
Spain
d
Depa men o Fo es P oduc ion and Enginee ing P ojec s, Uni o Sus ainable Fo es Managemen (UXFS), School o Enginee ing, Uni e si y o San iago de Compos ela,
E-27002 Lugo, Spain
e
Depa men o Plan P oduc ion and P ojec Enginee ing, ECOAGRASOC Resea ch G oup, School o Enginee ing, Uni e si y o San iago de Compos ela, E-27002 Lugo,
Spain
CERNA, Illas Cies 52, E-27003 Lugo Spain
g
GIS-Fo es Resea ch G oup, Depa men o O ganisms and Sys ems Biology, Poly echnic School o Mie es, E-33600 Mie es, As u ias, Spain
ARTICLE INFO
Keywo ds:
Une en-aged s ands
Heigh -diame e cu es
Sus ainable o es managemen
In e na ional o es policy
Guide cu es
ABSTRACT
Sus ainable p oduc ion o wood is one o he main se ices p o ided by o es sys ems. Si e p oduc i i y in he
case o o es s is o en e alua ed h ough he si e quali y. Howe e , mos o he wo ks add essing he si e quali y
ha e been done a local o egional scale. In his wo k, we aim o de elop si e quali y models o i e dominan
species in Spanish o es s (Fagus syl a ica,Pinus pinas e a lan ica,Que cus py enaica,Pinus nig a,Pinus syl es is)
and c ea e si e quali y maps a a na ional-scale om hese models. Fi s , we de elop si e quali y models using
si e o m (heigh -diame e ela ionship) as he e e ence index and he Spanish Na ional Fo es In en o y as
da ase . Then, we i spa ial addi i e models en e ing physiog aphic and clima ic a iables in o de o p edic he
si e quali y o e he whole coun y. Addi ionally, we plo si e o m maps o he i e species in o de o desc ibe
spa ial pa e n in si e quali y a a na ional scale. Al i ude and aspec appea ed o be undamen al a iables in he
assessmen o si e quali y. The accu acy o he spa ial addi i e models anged om 38.2% o 47.9%. The co -
espondence be ween he p edic ed and obse ed maps o si e quali ies is clea . Ou esul s p o ide a ool which
could be used by o es manage s in land use planning as well as in o es policy decision-making a a na ional
scale. We sugges ha his me hod could be used in o he coun ies and ha he maps could be expanded o he
Eu opean scale o assessing he way in which si e quali y a ies ac oss Eu ope always conside ing ha he
ela ionships be ween o es p oduc i i y and en i onmen al a iables could a y among biogeoclima ic zones.
1. In oduc ion
Si e quali y has long been used in o es y as a p oxy o si e p o-
duc i i y. This a iable is no only o in e es o p edic ing g ow h and
yield o o es s ands (Ál a ez-González e al., 2005; Clu e e al., 1983;
Diéguez-A anda e al., 2005) bu also o s udies on ecological di e si y
(F anklin e al., 1989), o es s uc u e (La son e al., 2008) and o es
dis u bances (Wei e al., 2003) among o he s. Si e quali y has adi-
ionally been exp essed as he ela ionship be ween dominan heigh
and age (si e index) o e en aged o es s. Howe e , his de ini ion is
di icul o apply in o es s wi h an une en dis ibu ion o ages o in
mixed-species o es s, whe e he heigh -diame e ela ionship (si e
o m) has p o ed o be a good measu e o si e p oduc i i y (Huang and
Ti us, 1993; McLin ock and Bick o d, 1957; S ou and Shumway, 1982;
Vanclay, 1994). Si e o m can be also use ul o de e mine he si e
quali y o e en-aged s ands whe e he age is unknown, as occu s in he
Na ional Fo es In en o ies (NFIs) o some coun ies (Tomppo e al.,
2010).
h ps://doi.o g/10.1016/j. o eco.2018.03.016
Recei ed 11 Janua y 2018; Recei ed in e ised o m 6 Ma ch 2018; Accep ed 9 Ma ch 2018
Abb e ia ions: AIC, Akaike’s In o ma ion C i e ion; B-R, Be alan y-Richa ds model; dbh, diame e a b eas heigh ; EF, expansion ac o ; EFT, Eu opean Fo es
Type; H-II, Hoss eld II model; NFI, Na ional Fo es In en o y
⁎
Co esponding au ho a : INIA-CIFOR, C a. A Co uña km 7.5, E-28040 Mad id, Spain.
E-mail add ess: [email p o ec ed] (D. Mo eno-Fe nández).
Fo es Ecology and Managemen 417 (2018) 197–207
A ailable online 16 Ma ch 2018
0378-1127/ © 2018 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY license
(h p://c ea i ecommons.o g/licenses/BY/4.0/).
T
Si e o m, as well as si e index, can be es ima ed ia di ec and in-
di ec me hods. Di ec me hods a e based on he ela ionship be ween
he heigh and diame e , in he case o si e o m, and be ween heigh
and age, in he case o si e index. Meanwhile indi ec me hods ela e
physiog aphic, clima ic, edaphic a iables and unde s o y o si e
quali y (B a o and Mon e o, 2001; Pacheco Ma ques, 1991). Al e -
na i e base models ha e been employed o de elop diame e -heigh
ela ionships. Meye ’s ma hema ical exp ession ha e been used o
i ing si e o m cu es in Picea ubens Sa g. (McLin ock and Bick o d,
1957) and in six ha dwood species (S ou and Shumway, 1982) while
Huang and Ti us (1993) selec ed he Be alan y-Richa ds model as he
base heigh -diame e unc ion.
Di e en s a is ical me hodologies ha e been used o de e mine he
in luence o en i onmen al a iables on he si e quali y, such as linea
models (Chen e al., 2002; Pacheco Ma ques, 1991; Seyna e e al.,
2005), disc iminan analysis (B a o-O iedo and Mon e o, 2005; Bueis
e al., 2016) o eg ession ees (Ál a ez-Ál a ez e al., 2011). The
g oup o a iables selec ed o si e quali y models and hose a iables
ound as signi ican ly associa ed o si e quali y ha e a ied among
species and s udy si es. B a o-O iedo e al. (2011) s a ed ha a ia ions
in si e quali y o Medi e anean Pinus pinas e Ai s ands a e mainly
explained by clima e a iables. O he au ho s, howe e , only con-
side ed edaphic and physiog aphic a iables in si e quali y models
(B a o e al., 2011; Bueis e al., 2016; Seyna e e al., 2005). Aspec ,
la i ude, con en o nu ien s, soil mois u e, soil ex u e, pH o soil
dep h appea ed o be ela ed o si e quali y (B a o and Mon e o, 2001;
Bueis e al., 2016; Seyna e e al., 2005). Ne e heless, he e ec s o a
a iable on he si e quali y o a gi en species may change among e-
gions (Chen e al., 2002). Addi ionally, some o hese ela ionships
migh no be linea as in he case o he si e quali y-al i ude (Seyna e
e al., 2005). In hese cases, linea models canno iden i y complex
ela ionships among a iables. The e o e, app oaches, such as addi i e
models, desc ibing nonlinea and complex ela ionships be ween he
si e quali y and he p edic o s can be e y use ul (Has ie and Tibshi ani,
1989; Wood, 2006).
Mos o he s udies assessing he quali y si e ha e been ca ied ou a
local o egional scales (Bueis e al., 2016; Seyna e e al., 2005)
whe eas he knowledge o o es p oduc i i y a la ge geog aphic
scales is sca ce (Chen e al., 2002). In his ega d, some au ho s a i m
ha he ep esen a i eness a a coa se scale o o es p oduc i i y om
es ic ed a ea da a is ques ionable (Cha u e al., 2010). In his sense,
NFIs p o ide he b oades sou ce o knowledge on he s a us o he
o es s a na ional le el in many coun ies (Ba ba i e al., 2014). In ac ,
NFIs ha e been used o es ima e p oduc ion (Cha u e al., 2010),
abo eg ound biomass (A i abile and Camia, 2018), ca bon s o age
(Woodall e al., 2008) and moni o ing species dis ibu ion (He nández
e al., 2014; Mo eno-Fe nández e al., 2016) o assessmen o biodi-
e si y (Ande sson and Ös lund, 2004) and abio ic damages (Jalkanen
and Ma ila, 2000). Addi ionally, he NFIs ha e been used as da ase s in
scien i ic s udies o model ec ui men (Lexe ød, 2005), ee biomass
(Ruiz-Peinado e al., 2011), ee mo ali y (Ruiz-Beni o e al., 2013),
deadwood olume (C ecen e-Campo e al., 2016) as well as si e quali y
(Adame e al., 2006).
In e na ional ins i u ions and p ocesses, howe e , a o using
Table 1
Numbe o plo s o he Na ional Fo es In en o y (N), numbe o plo s wi h basal a ea o he species in pa en hesis la ge han 90% (N G ≥ 90%), mean basal a ea (G
in m
2
ha
−1
), mean dominan heigh (H
0
in m) and mean dominan diame e (D
0
in cm) in he selec ed Eu opean Fo es Types. Minimum and maximum alues in
b acke s.
Eu opean Fo es Type N N G ≥ 90% G H
0
D
0
2.7 A lan ic ma i ime pine o es (Pinus pinas e a lan ica) 3410 1568 18.0
(0.4–85.0)
15.2
(2.0–29.3)
30.5
(7.5–71.9)
7.1 Sou h wes e n Eu opean moun ainous beech o es (Fagus syl a ica) 4449 1915 25.5
(0.4–64.2)
18.8
(3.7–36.5)
40.9
(7.5–158.8)
8.3 Py enean oak o es (Que cus py enaica) 5528 2889 10.9
(0.4–72.5)
9.9
(0.4–23.3)
24.2
(7.5–134.4)
10.2 Medi e anean and Ana olian black pine o es (Pinus nig a) 8352 3212 15.1
(0.4–80.4)
10.4
(2.6–33.3)
26.5
(7.5–100.3)
10.4 Medi e anean and Ana olian Sco s pine o es (Pinus syl es is) 10,919 5171 22.7
(0.4–83.8)
12.1
(2.0–30.5)
29.9
(7.5–64.8)
Table 2
Base model, si e speci ic pa ame e and si e speci ic pa ame e o each si e o m amily o cu es.
Species Base model Pa ame e Equa ion
Es ima e P (> | |)
P. pinas e H-II
a
b = 0.13035 < 2 10
−16
= +
+
SFI 1.3
D e
D
Hb Do e D
0
2
0
01.3 (0)
2
F. syl a ica H-II a = 2.5473 < 2 10
−16
= +
+
SFI 1.3
D e
aD
Ha
D e
D
0
2
0
01.3
0
0
2
Q. py enaica H-II a = 2.0539 < 2 10
−16
= +
+
SFI 1.3
D e
aD
Ha
D e
D
0
2
0
01.3
0
0
2
P. nig a B-R
b
b = 0.05405 < 2 10
−16
= +SFI H1.3 ( 1.3) ebD e
ebD
c
010
10
c = 2.2316 < 2 10
−16
P. syl es is H-II b = 0.1386 < 2 10
−16
= +
+
SFI 1.3
D e
D
Hb D e D
0
2
0
01.3 (00)
2
a
H-II: Hoss eld-II.
b
B-R: Be alan y-Richa d.
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
198
in e na ional a he han na ional c i e ia (FAO, 2015; Gable e al.,
2012) o moni o sus ainable o es managemen p ac ices and o es
s a e a in e na ional scales (La sson, 2001). In his con ex , Ba ba i
e al. (2014) and Ba ba i e al. (2007a) ha e p oposed a classi ica ion o
ca ego ize he o es o he Pan-Eu opean egion in o 14 Eu opean
Fo es Types (EFTs) ca ego ies (ca ego y le el) and he ca ego ies in o
78 EFTs ( ype le el) acco ding o ecological sound uni s. The EFTs
classi y o es ea u es by agg ega ing and a e aging da a om NFIs
in o ecologically homogeneous s a a o Eu opean ele ance, i.e. he 14
EFTs ca ego ies (Ba ba i e al., 2014). The e o e, epo ing o es p o-
duc i i y h ough an index independen o he o es s uc u e and
ollowing EFTs classi ica ion would con ibu e o s anda dize assess-
men esou ces in Eu ope o e en ab oad.
In his wo k, we aim o de elop si e quali y models a a na ional-
scale using si e o ms o i e species, and display he esul s spa ially.
In he i s s ep, we i si e o m models (diame e -heigh cu es) o he
Fig. 1. Si e o m amily o cu es o he Eu opean Fo es Types s udied.
Table 3
p- alues o he model componen s and pe cen age o de iance explained by he spa ial addi i e model o each Eu opean Fo es Type.
Species In e cep (X
i
,Y
i
) (al i ude) (aspec ) (slope) De iance (%)
P. pinas e < 0.0001 < 0.0001 < 0.0001 n.s.
a
n.s. 39.6
F. syl a ica < 0.0001 < 0.0001 < 0.0001 < 0.0007 0.0400 37.3
Q. py enaica < 0.0001 < 0.0001 < 0.0001 0.0192 n.s. 32.5
P. nig a < 0.0001 < 0.0001 < 0.0001 < 0.0001 n.s. 48.6
P. syl es is < 0.0001 < 0.0001 < 0.0001 < 0.0001 n.s. 42.0
a
n.s. = non-signi ican .
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
199
i e species using NFI as da ase . We hen model he si e o m using
spa ial addi i e models including en i onmen al a iables as p e-
dic o s. Finally, we used hese spa ial addi i e models o c ea e si e
quali y maps by species a a na ional-scale.
2. Ma e ial and me hods
2.1. Species s udied and Na ional Fo es In en o y da ase
We used he EFT classi ica ion o selec he a ge species. We se-
lec ed i e species widely dis ibu ed in Spain and which o m genuine,
monodominan o es acco ding o he EFT classi ica ion (Ba ba i e al.,
2014, 2007b): Pinus pinas e Ai . ssp a lan ica (P. pinas e , he ea e ),
Fagus syl a ica L., Que cus py enaica Willd., Pinus nig a A n. and Pinus
syl es is L. (Table 1). He ea e , each EFT is named a e he dominan
species p esen . All o he species a e ound mainly in mon ane a eas
wi h he excep ion o P. pinas e which is qui e common a sea le el
(Ruiz de la To e, 2006).
The Thi d Spanish NFI, conduc ed be ween 1997 and 2007, was
used o model and p edic he si e quali y in Spanish o es s. The NFI
plo s we e only es ablished in woodland a eas, acco ding o he FAO
de ini ion (FAO, 2001) (i.e. none in non- o es ed a eas) wi h an in-
ensi y o one sampling poin e e y 1 km
2
(1 × 1 km g id). A each
sampling poin , diame e and heigh o he ees we e measu ed in ou
concen ic ci cula subplo s wi h inc easing adii om 5 o 25 m. In he
5 m adius subplo , ees wi h a diame e a b eas heigh (dbh)
≥7.5 cm we e measu ed. In he 10 m adius subplo , ees wi h dbh
≥12.5 cm we e measu ed. In he 15 m adius subplo , ees wi h dbh
≥22.5 cm we e measu ed. Finally, in he 25 m adius subplo , ees
wi h dbh ≥42.5 cm we e measu ed. The a ibu es o he ees mea-
su ed in each concen ic subplo we e expanded o pe -hec a e alues
by conside ing a di e en expansion ac o (EF) o each subplo size,
which is EF
k
= 10000/a
k
whe e a
k
is he a ea o each subplo wi h size k
(k= 5, 10, 15 and 25). Thus, EF
k
co esponds o he ees pe hec a e
ep esen ed by e e y ee measu ed in each subplo wi h size k. (Albe di
Asensio e al., 2010; He nández e al., 2014).
2.2. Si e o m models
Si e quali y models we e de eloped o each EFT using he NFI plo s
wi h basal a ea dominance o he main species g ea e han 90%
(Table 1). Si e quali y was es ima ed using s and le el heigh -diame e
ela ionships (Huang and Ti us, 1993; S ou and Shumway, 1982),
which is e e ed o as si e o m (Vanclay, 1994) when using dominan
diame e (D
0
, cm) and dominan heigh (H
0
, m). D
0
and H
0
we e
compu ed o each NFI plo as he a i hme ic mean o he dbh (cm) and
ee heigh (h, m) espec i ely o he 100 ees pe hec a e wi h he
mean la ges dbh. As we ha e only one measu emen o hese a iables
o each sample plo , he si e o m models we e de eloped in wo s eps
by using he guide-cu e me hod (Clu e e al., 1983). In he i s s ep a
model was i ed o he pai s o D
0
and H
0
, he eby ob aining he
a e age H
0
/D
0
ela ionship o each EFT. In he second s ep, he amily
o si e o m cu es we e gene a ed making each o he model pa a-
me e s dependen on he si e o m, which is he alue o H
0
a a gi en
e e ence alue o D
0
. Thus, each base model gene a es as many si e
o m amily cu es as pa ame e s has in he base model o mula ion.
Two well-known base models used in he de elopmen o si e quali y
models we e conside ed in he i ing p ocess, namely, he Hoss eld II
model (H-II, Eq. (1)) and he Be alan y-Richa ds model (B-R, Eq. (2)),
and all possible esul an amilies o si e o m amily cu es we e
gene a ed o each EFT. The base models we e adap ed in o de o a -
ain p edic ions o H
0
=1.3 o D
0
= 0. The inal model o each EFT
was selec ed acco ding o he pe o mance o he si e o m cu es o e
he da a by conside ing as c i e ia he polymo phism o he esul ing
cu es as well as he p esence o mul iple asymp o es. The e e ence D
0
was selec ed o imp o e he accu acy o p edic ions, educing he
p edic ion bias associa ed wi h s ands whe e D
0
di e s g ea ly om he
e e ence D
0
(Weiski el e al., 2011). Fu he mo e, he ange o he
diame e s obse ed in he di e en EFTs and he e o o he model in
di e en egions o he alidi y ange o he model we e also aken in o
accoun .
= +
+
HD
a b D
1.3 ( · )
00
2
02
(1)
= +H a b D1.3 ·(1 exp( · ))c
0 0
(2)
whe e H
0
is he dominan heigh (m), D
0
is he dominan diame e (cm)
and a,band ca e model pa ame e s o be es ima ed.
A e selec ing he bes models o each EFT, we es ima ed he si e
o m alue in he NFI plo s. The si e o m p edic ions o each EFT we e
hen di ided in o ou si e quali y classes (ac ual si e quali y) using he
minimum, he i s qua ile, he median, he hi d qua ile and he
maximum o he si e o m da ase as b eakpoin s. Class A ep esen s he
highes si e quali y class, class B and C a e he high- and low-in e -
media e si e quali y classes espec i ely and, inally, class D is he
lowes si e quali y class.
2.3. Spa ial analysis
In ecology, as well as in o he ields such as o es y, non-linea
ela ionships be ween a iables a e common and, he e o e, linea
modelling o en pe o ms poo ly (Fa away, 2006). Hence, app oaches,
such as addi i e models (Has ie and Tibshi ani, 1989; Wood, 2006),
which desc ibe complex ela ionships be ween he esponse and he
p edic o s a e especially use ul. Thus, we i ed he ollowing spa ial
addi i e models o p edic he o m index o each o he i e EFT in
Spain:
= + + +
=
SFI X Y x( , ) ( )
i i i
j i
J
ij i
(3)
whe e SFI is he si e o m index, α ep esen s he in e cep o he model,
X Y( , )
i i
is a spa ial smoo h unc ion o accoun o he spa ial pa e n o
he SFI and elimina e he spa ial co ela ion. X
i
and Y
i
a e he co-
o dina es in me e s (UTM, da um ED50 zone 30N) o he i- h plo . (x
ij
)
a e smoo h unc ions ( om j= 1 o J) o physiog aphic (al i ude in m,
slope in deg ees, and aspec in deg ees) and clima ic a iables o be
included in he model. Physiog aphic a iables we e ex ac ed om a
digi al ele a ion model. We ob ained he clima ic da a om he
200 × 200 m Spanish clima ic g id (Gonzalo, 2010). This g id p o ides
bo h ain all ( o al, summe , win e , sp ing and au umn ain all) and
empe a u e a iables. Howe e , empe a u e a iables a e highly
Table 4
Model selec ion and Akaike’s In o ma ion C i e ion o each Eu opean Fo es s Type. In bold he selec ed model.
Model Model e ms P. pinas e F. syl a ica Q. py enaica P. nig a P. syl es is
(1) α7766.51 10254.05 14607.96 15852.38 27476.10
(2) (1) + (X
i
,Y
i
) 7332.67 9805.29 13951.03 14592.00 25631.34
(3) (2) + (al i ude)7151.53 9720.40 13906.19 14335.19 25333.93
(4) (3) + (aspec ) 7146.83 9699.32 13896.34 14301.22 25312.09
(5) (5) + (slope) 7146.00 9692.03 13892.92 14302.42 25311.93
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
200
Fig. 2. Es ima ed cen e ed smoo h unc ions and 95% con idence in e als o he physiog aphic a iables o he si e o m spa ial addi i e models o he Eu opean
Fo es Types s udied.
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
201
co ela ed o al i ude and we e no conside ed in he analysis. The se-
lec ion o he a iables was ca ied ou acco ding o Akaike’s In-
o ma ion C i e ion (AIC) using a o wa d s epwise p ocedu e se ing a
educ ion o i e poin s o AIC as a signi ican h eshold alue. All he
a iables we e en e ed as smoo h unc ions. Finally, ε
i
is he e o e m
o he model.
We ep esen ed he smoo h unc ions using hin pla e eg ession
splines (Wood, 2003). These splines keep he basis and he penal y o
he ull hin pla e splines (Duchon, 1977) bu he basis is unca ed o
ob ain low ank smoo he s. This educes he compu a ional equi e-
men s o he smoo hing splines and a oids he p oblems o he kno
placemen o he eg ession splines (Wood, 2003). We plo ed semi-
a iog ams o he esiduals o each model o check i he spa ial co -
ela ion had been elimina ed. Addi ionally, we c ea ed semi a iog am
en elopes a e 99 pe mu a ions unde he assump ion o no spa ial
co ela ion (Augus in e al., 2009).
Once models we e i ed, we classi ied he o m index p edic ed by
he spa ial addi i e model in o he ou si e quali y classes (p edic ed si e
quali y) using he same cu o s as he si e o m models. The accu acy
pe cen age was hen calcula ed as he a io o he plo s co ec ly clas-
si ied o he o al numbe o plo s.
2.4. Si e quali y maps
We used he spa ial addi i e models i ed in he p e ious sec ion o
p edic he o m index and he si e quali y in all he plo s o he NFI
whe e ees o he a ge species we e p esen (second column in
Table 1).
We used he ollowing R packages (R Co e Team, 2017) o ca y ou
he s a is ical analysis: “nls”, “mgc ” (Wood, 2011) and “geoR”. The
maps we e p oduced in A cGis 10.2.2. (ESRI, 2014).
3. Resul s
The pa ame e es ima ion o he inal si e o m models and he si e-
speci ic pa ame e s o each EFT a e shown in Table 2. H-II was he
model selec ed o all he EFTs wi h he excep ion o P. nig a, o which
B-R was selec ed. The si e o m amily o cu es ep esen s easonably
well he da a o he coun ywide condi ions co e ed in he NFI da a o
each EFT (Fig. 1).
The pe cen age o de iance explained by he spa ial addi i e models
anges om 32.5 o 48.6% (Table 3). The s uc u e o he spa ial ad-
di i e models a ies om one EFT o ano he a e he o wa d s epwise
a iable selec ion p ocedu e (Tables 3 and 4). The spa ial smoo h
unc ion ( (X
i
, Y
i
)) and he smoo h unc ion o al i ude ( (al i ude)) a e
included in all he models. In addi ion, o enhance he pe o mance o
he model, he spa ial e m elimina es he spa ial co ela ion o he
esiduals (see S1 Fig.).
The ela ionship be ween he al i ude and he si e o m index a ies
among he i e EFT. In he case o P. pinas e , a species well ep esen ed
om sea le el up o al i udes sligh ly abo e 1000 m o ele a ion, (al-
i ude) eaches alues a lowe al i udes and dec eases p og essi ely a
highe al i udes (Fig. 2), wi h he la ges alues o he si e o m index
expec ed o be obse ed a lowe al i udes. In he case o mo e mon ane
species (P. syl es is,P. nig a and F. syl a ica), he 95% con idence in-
e als o he mean widen a lowe ele a ions, and so he si e o m
shows no signi ican di e ences o ele a ions below han 700–800 m.
F om ha al i ude upwa ds, (al i ude) dec eases o hese h ee species,
bu he con idence in e al widens again o high ele a ions, wi h a
h eshold a a ound 1500 m o beech and 1800 m o P. nig a and P.
syl es is. In he case o Q. py enaica, a species well ep esen ed a al-
i udes be ween 400 and 1600 m, he pa e n is di e en , and (al i ude)
shows a maximum pla eau o ele a ions be ween 500 and 800 m; he
con idence in e als also widening o lowe and highe ele a ions.
Addi ionally, we ound a signi ican ela ionship be ween (aspec ) and
si e o m index o all he species wi h he excep ion o P. pinas e .Fig. 2
indica es ha he highes alues o he si e o m index a e expec ed o be
ound a no he n exposu es (aspec anging om 300° o 360° and om 0
o 60°) in F. syl a ica,Q. py enaica,P. nig a and P. syl es is EFT. E en so,
he di e ence be ween he maximum and minimum alues o (aspec ) is
g ea e in he cases o F. syl a ica and P. nig a, whe eas aspec is less in-
luen ial o P. syl es is and Q. py enaica. The a iable selec ion p ocedu e
e eals a weak associa ion be ween si e o m index and slope in F. syl a ica
(p- alue = 0.0400). In his ega d, (slope) shows a bell-shaped pa e n
peaking a 22°. Finally, none o he clima ic a iables signi ican ly educed
he AIC (less han i e poin s).
The la ges educ ions in AIC o he i e EFTs spa ial models appea
a e applying he spa ial smoo hing. This means ha he spa ial
s uc u e accoun s o a la ge pe cen age o a iance (Table 4). The
second mos impo an a iable in AIC educ ion is (al i ude). Finally,
(aspec ) in F. syl a ica,Q. py enaica,P. nig a and P. syl es is models and
(slope) in F. syl a ica play mino oles. Hence, he es ima ed cen e ed
smoo h unc ions o aspec and slope a e qui e close o ze o and he
ange o a ia ion o hese es ima ed cen e ed smoo h unc ions a e
na owe han ha o al i ude leading o s onge ela ionships be ween
(al i ude) and si e quali y (Fig. 2).
The accu acy o he si e quali y anges om 38.2% in he case o Q.
py enaica o 47.9% o P. nig a (Table 5). The si e o m p edic ions using
he spa ial addi i e models mainly classi ied he plo s in o in e media e
quali y si e classes (B and C) whe eas he numbe o plo s classi ied in o
he highes and lowes quali y classes (A and D, espec i ely) was
lowe . This poin s o he di icul y o p edic ing cases o p oduc i i y
le els well abo e o below he a e age.
The obse ed si e quali y classes a e shown in Fig. 3. The p edic ed
si e quali y maps sugges ha mos o he s ands o he i e EFTs s udied
a e loca ed in in e media e si e quali y a eas (Fig. 4 and Table 5). The
dis ibu ion o he quali y classes p edic ed shows mo e clea ly he
g adien s ha depend mainly on ele a ions and he aspec . O e all, he
co espondence be ween he p edic ed and obse ed maps o si e qua-
li ies is clea (Figs. 3 and 4), e en hough he model ails o p edic
classes A and B, which a e in ac obse ed o a eas whe e o he classes
a e mo e equen .
Table 5
Con usion ma ix o each Eu opean Fo es Type. Accu acy pe cen age is shown
in b acke s.
Species Si e o m model
(obse ed si e
quali y)
Spa ial addi i e model (p edic ed si e
quali y)
A B C D
Pinus pinas e
a lan ica
(45.4%)
A 139 121 44 1
B 37 145 114 10
C 11 109 159 26
D 2 59 133 112
Fagus syl a ica
(42.7%)
A 135 270 73 1
B 34 274 157 13
C 13 193 220 53
D 2 79 209 189
Que cus py enaica
(38.2%)
A 179 428 113 0
B 50 373 296 3
C 19 255 409 39
D 11 153 417 142
Pinus nig a (47.9%) A 418 309 72 4
B 120 388 253 42
C 31 277 406 89
D 6 117 354 326
Pinus syl es is
(45.2%)
A 654 464 161 13
B 197 659 400 36
C 47 468 664 113
D 11 225 696 360
A = high quali y si e, B = in e media e-high quali y si e, C = in e media e-low
quali y si e, D = low quali y si e.
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
202
Fig. 3. Obse ed si e quali y o he Eu opean Fo es Types s udied. Da ke colo s indica e low al i udes whe eas ligh in ensi ies deno e high al i udes.
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
203
Fig. 4. P edic ed si e quali y o he Eu opean Fo es Types s udied. Da ke colo s indica e low al i udes whe eas ligh in ensi ies deno e high al i udes.
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
204
4. Discussion
We es ima ed he si e quali y in he i e mos impo an EFTs in
Spain exp essed as he dominan heigh and dominan diame e e-
la ionship and hen c ea ed si e quali y maps using spa ial addi i e
models. O he au ho s ound highe pe cen ages o accu acy o si e
index in Medi e anean species: om 64% o 71% in P. syl es is (B a o
and Mon e o, 2001; Bueis e al., 2016) and om 61 o 75% in Pinus
pinea L. (B a o-O iedo and Mon e o, 2005; B a o e al., 2011). How-
e e , hese s udies we e pe o med a ine scales, local o egional,
whe eas ou aim was o es ima e he si e quali y a a na ional scale. In
his ega d, i has been shown ha he ela ionships be ween en i on-
men al a iables and si e index a e s onge a ine scales (Chen e al.,
2002).
Ou esul s a e in conco dance wi h he au oecology o he species,
con i ming he sui abili y o ou app oach om a biological poin o
iew. The nega i e ela ionship be ween P. pinas e si e quali y and
al i ude has p e iously been epo ed in Medi e aean a eas (B a o-
O iedo e al., 2011) in o he s udies unde A lan ic condi ions (Ál a ez-
Ál a ez e al., 2011; Eimil-F aga e al., 2014). This species is dis ibu ed
ac oss a eas o A lan ic clima e wi h i s cen al habi a in e ms o
ele a ion anging om 0 o 1000 m, eaching he maximum le els o
si e p oduc i i y a 513 m asl (Ál a ez-Ál a ez e al., 2011). The e o e
is clea ly di e en ia ed om he o he wo pine species s udied, which
a e genuinely mon ane (Gandullo and Sánchez-Paloma es, 1994). In
dis ibu ion models, al i ude o o he ela ed a iables such as em-
pe a u e usually show a posi i e o in e se U-shaped ela ionship wi h
he occu ence o mon ane species (He nández e al., 2014; Mo eno-
Fe nández e al., 2016). Tha was obse ed o si e p oduc i i y in P.
ubens (Chen e al., 2002) and, in ou s udy case, in Q. py enaica. This
species is known o g ow a a ange o ele a ions be ween 400 and
1400 m, wi h i s op imum habi a in no he n a eas a lowe ele a ions
han in he sou he n ones (Díaz-Ma o o e al., 2007; Sánchez-Paloma es
e al., 2008). An in e se U-shaped ela ionship may ha e been expec ed
o he o he h ee species, he cen al habi a s o which a e also wi hin
ele a ion anges classi iable as mon ane (Gandullo and Sánchez-
Paloma es, 1994). The occu ence o F. syl a ica a low ele a ions in
Spain is mainly es ic ed o humid si es on he slopes o no he n
Moun ain anges, whe e he species is no es ic ed by d ough s and
shows high p oduc i i y a es a an ele a ion ange o 400–800 m
(Sánchez-Paloma es e al., 2004). Some s ands showing poo p o-
duc i i y a ele a ions below he cen al habi a o hese species and
loca ed in d ye a eas p esen a mixed composi ion and he e o e, as we
only used plo s in which he main species o he EFT was dominan ,
hese s ands we e no conside ed in he s udy. Thus, in he plo s used in
his s udy, he species g ow unde condi ions which a e su icien ly
a o able o o m monospeci ic s ands o ha e been a o ed by o es
managemen . Ce ain ac o s associa ed wi h sys ems a high ele a-
ions, such as sho e g owing pe iods (Benis on, 2003), may explain
he gene al nega i e associa ion.
The esul s ob ained as ega ds he in luence o aspec on he si e
o m p o ide new insigh s in o he dependence o s and p oduc i i y on
si e pa ame e s, in addi ion o hose epo ed in s udies o species dis-
ibu ion o egene a ion niches (Gómez-Apa icio e al., 2006;
He nández e al., 2014). A clea e ec o aspec on s and p oduc i i y
was pa icula ly e iden in he case o F. syl a ica, a shade ole an
species ha g ows be e in no he n exposu es wi h mo e humid
condi ions (Ruiz de la To e, 2006). He nández e al. (2014) epo ed
an inc ease in he p esence o F. syl a ica on no h- acing exposu es in
no he n Spain o e he las 40 yea s. Such a clea posi i e e ec o
no he n exposu es on si e o m was also obse ed o P. nig a, a species
ha is conside ed o ha e an in e media e shade ole ance, a leas in
he case o he subspecies salzmannii (Ruiz de la To e, 2006), al hough
egene a ion in his species has been ound o be mo e ole an o high
le els o i adiances han se e al b oadlea ed species (Gómez-Apa icio
e al., 2006). The ole o aspec on si e quali y was ound o be smalle
o Q. py enaica and P. syl es is, e en hough he si e o m a no he n
exposu es was equen ly ound o be signi ican ly g ea e . Cañellas
e al. (2000) ha e al eady highligh ed he ac ha he in luence o
aspec on si e index in P. syl es is si e index is expec ed o be g ea e o
s ands wi h Medi e anean condi ions. P. pinas e is a ligh -demanding
species (Ruiz de la To e, 2006) and he subspecies a lan ica equi es
humid condi ions o a good de elopmen . I g ows, as men ioned
abo e, mainly a ela i ely low al i udes close o he coas (Ál a ez-
González e al., 2005), in luenced by he A lan ic clima e wi h mild
summe empe a u es (Elena Roselló, 1997); he e o e, i is no su -
p ising ha aspec had no s a is ically signi ican in luence on si e
quali y in his species.
O he local a iables, such as soil a iables, a e expec ed o in lu-
ence he si e index (Ál a ez-Ál a ez e al., 2011; B a o e al., 2011;
Bueis e al., 2016; Chen e al., 2002). Soil a iables, howe e , show
g ea spa ial a ia ions (see Vande linden e al. (2003) o soil wa e
holding in sou he n Spain). Ga he ing soil da a, such as soil ex u e, soil
dep h, o ganic ma e o nu ien s con en , o he whole coun y would
be e y use ul o p edic ing si e quali y bu i would be highly ime-
consuming and expensi e (McB a ney e al., 2006). The unmeasu ed
a iables could in ac explain he lowe abili y o he model o classi y
co ec ly some o he A and D p oduc i i y class plo s. Ne e heless, he
e ec o unmeasu ed en i onmen al a iables was pa ially accoun ed
o by he spa ial smoo h unc ion (Mo eno-Fe nández e al., 2018).
Addi ionally, he spa ial smoo h unc ion elimina ed he spa ial co e-
la ion minimizing he ype I e o a es (Do mann e al., 2007). Fu -
he mo e, i we conside he s ong ain all g adien s in Spain (Muñoz-
Díaz and Rod igo, 2004), he lack o signi ican ela ionships be ween
ain all a iables and si e quali y would be unexpec ed. Howe e ,
physiog aphical a iables and he spa ial smoo h unc ion may, in ac ,
be abso bing he e ec o he ain all.
I should also be men ioned ha bo h F. syl a ica and Q. py enaica
a e able o sp ou and hus can o m coppice o es (Ruiz de la To e,
2006). Hence, coppicing managemen sys ems ha e adi ionally been
hose mos equen ly employed in Spanish s ands o Q. py enaica
(Adame e al., 2008). I is known ha he g ow h pa e n o a gi en
species di e s om coppice o high o es , wi h he o me showing
as e ini ial heigh g ow h al hough his g ow h is less sus ained han
in high o es (Ciancio e al., 2006; Haneca e al., 2005). Howe e , he
e ec o s and s uc u e on si e o m is no clea . Fu he esea ch is
needed o add ess his aspec and i deemed necessa y, s and s uc u e
can be in eg a ed in o si e quali y models (Adame e al., 2008). Fu -
he mo e, wi hin he dis ibu ion a ea o each species in Spain, se e al
di e en subspecies can be ound. This is pa icula ly e iden in he
case o P. nig a which p esen s high in aspeci ic a iabili y wi hin
na u al s ands (Ruiz de la To e, 2006). Mo eo e , he use o Co sican
and Aus ian p o enances in e o es a ion has esul ed in s ill u he
a iabili y. This can lead o he a ibu ion o di e en si e quali ies o
each subspecies a he same loca ion when conside ing dominan heigh
and age as si e quali y c i e ia (Mo eno-Fe nández e al., 2014). How-
e e , since he ac ha he P. nig a subspecies appea a ely in close
p oximi y o each o he , spa ial smoo hing can cap u e he in luence o
he subspecies on si e o m. In addi ion, he use o he dominan dia-
me e ins ead o age can abso b pa o he e ec o he subspecies
di e en ia ion i he subspecies only di e in g ow h a es.
Ou esul s p o ide a ool which could be used by o es manage s in
land use planning as well as in o es policy decision-making a a na-
ional scale. Fo ins ance, he p oposed si e- o m models may be used
o es ima ing si e quali y in e en- o une en-aged s ands o he s udied
species, and he maps o la ge-scale o es planning. Addi ionally, his
me hodology ul ills he cu en demands o epo ing o es in-
o ma ion acco ding o in e na ional c i e ia and p o ides a iable
app oach o s anda dizing he me hod used o es ima e si e quali y,
he eby p oducing in e na ionally compa able da a. This me hodology
could be used in o he coun ies and he maps could be ex ended o he
whole o Eu ope. Thus, i would be possible o assess he way in which
D. Mo eno-Fe nández e al. Fo es Ecology and Managemen 417 (2018) 197–207
205
