ORIGINAL RESEARCH ARTICLE
published: 05 Ma ch 2015
doi: 10.3389/ pls.2015.00104
Modeling o bud b eak o Sco s pine in no he n Finland in
1908–2014
Hannu Salminen* and Ris o Jalkanen
Na u al Resou ces Ins i u e Finland (Luke), Ro aniemi, Finland
Edi ed by:
Heikki Hänninen, Uni e si y o
Helsinki, Finland
Re iewed by:
C is ina Nabais, Uni e si y o
Coimb a, Po ugal
Ignacio Ga cía-González, Uni e sidade
de San iago de Compos ela, Spain
*Co espondence:
Hannu Salminen, Na u al Resou ces
Ins i u e Finland (Luke), E elä an a 55,
Ro aniemi, FI-96300, Finland
e-mail: hannu.salminen@luke.fi
Bud b eak and heigh -g ow h o Sco s pine (Pinus syl es is L.) in he no he n bo eal
zone in Lapland, Finland, was ollowed h ough he en i e g owing seasons in he pe iods
2001–2003 and 2008–2010 in sapling s ands in wo di e en loca ions in no he n Finland
se some 250 km apa along a la i udinal ansec . Field measu emen s con inued a
he sou he n si e also in 2011–2013. Ai empe a u e was eco ded hou ly a he si es. A
simple op imiza ion algo i hm (GA) was used o adjus pa ame e s o he models p edic ing
he iming o bud b eak o Sco s pine in o de o minimize he di e ence be ween
obse ed and p edic ed da es. The models gi ing he bes pe o mance and cen u y-
long daily empe a u es we e used o econs uc bud-b eak ime se ies.The empe a u e
obse a ions we e eco ded o he pe iod 1908–2014 in Sodankylä, which is loca ed in-
be ween he sapling s ands in he no h–sou h di ec ion and o he pe iod 1877–2014 in
Ka asjok, which is in No way abou 145 km no h–wes om he no he nmos s and o
his s udy. On a e age buds began o ex end in he beginning o May in he sou he nmos
s and and in mid-May in he no he nmos s ands, and he a ia ion be ween yea s was
in he ange o 3 weeks. A simple day-leng h- igge ed (fixed da e) model p edic ed mos
accu a ely he da e o bud b eak; oo mean squa e e o (RMSE) was 2 and 4 days in he
no he n and sou he n si e, espec i ely.The econs uc ed bud-b eak se ies indica ed ha
based on empe a u e obse a ions om Sodankylä, g ow h onse o Sco s pine has clea ly
ad anced since he 1960s, hough i cu en ly ma ches ha o he ea ly 1920s and ea ly
1950s. The empe a u e eco d om Ka asjok indica ed a simila a ia ion, hough he e
was a weak linea end ad ancing bud b eak by abou 3–4 days o e a 100-yea pe iod.
Keywo ds: Pinus syl es is L., bud bu s , phenology, clima e, high la i ude, ee line
INTRODUCTION
Bud b eak in ees can be seen om di e en pe spec i es. The
iming o bud b eak in ees is influenced by bo h gene ic (Beuke ,
1994) and clima ic ac o s (Chuine, 2000;Heide, 2003;Linkos-
alo e al., 2006). Due o he la e , bud b eak can be linked
o changes in sp ing empe a u es, and bud b eak and o he
phenophases such as flowe ing ha e been used as a bioindica-
o o clima e change. In ega d o ee g ow h, he iming o
bud b eak plays an impo an ole in compe i ion o sca ce
esou ces; ea ly g ow h onse adds leng h o an o he wise sho
g owing season, al hough i also exposes he new g ow h o
os damage o en b ough on by he a ying sp ing empe -
a u es. In gene al, bud b eak has been o in e es because i
can be isually assessed and is he e o e ela i ely simple o
obse e.
Acco ding o Hänninen (1990), any model o bud b eak
can be p esen ed using h ee submodels; he es b eak, he
de elopmen o g ow h compe ence, and he ac ual bud de elop-
men o bu s ing. The mos salien ac o in bud-b eak models
is he accumula ion o empe a u e in sp ing, i.e., o cing.
Fo cing is ei he a empe a u e sum o an equa ion d i en by
empe a u e accumula ion. Accumula ion is e ec i e a e i s p e-
equisi e, g ow h compe ence, is me . The simple app oach
uses day-leng h o d i e g ow h compe ence, while he mo e
sophis ica ed one inco po a es he e ec o chilling o bo h (see
Chuine, 2000). The de elopmen o g ow h compe ence and
he ac ual on ogene ic de elopmen can be modeled as sequen-
ial o simul aneous p ocesses (Hänninen, 1990). The downside
o bud-b eak models is ha hey a e usually d i en by daily
empe a u e da a, and a ailable con inuous daily empe a u e
se ies om no he n Finland a e usually ela i ely ecen , ypi-
cally s a ing om he ea ly 1960s. In his s udy, obse a ions o
p o incial empe a u e co e ing a pe iod o o e 100 yea s we e
applied.
Recen epo s on he how he phenology o ees is impac ed
by clima e change a e qui e unanimous; phenological e en s ha e
ad anced and he g owing season has become longe o e ecen
decades (e.g., Chmielewski and Rö ze , 2001;Doi and Ka ano,
2008;Kaye and Wagne , 2014) o e en o e he las cen u y
(Linkosalo e al., 2009). Empi ical da a on phenology exis , bu
he se ies a e usually sho , las ing 7–30 yea s [see a e iew by
Linde holm (2006)] and wi hou econs uc ion o inco po a ing
da a on flowe ing o lea bud bu s o b oad-lea ed ee species.
No way sp uce (Picea abies (L.) H. Ka s .) and Sco s pine (Pinus
syl es is L.), he wo main coni e species in he no he n bo eal
zone in Fennoscandia, ha e a di e en s a egy in hei g ow h
phenology. Suscep ible o la e os , No way sp uce delays i s flush
o he leade bud beyond he beginning o he accumula ion o he
www. on ie sin.o g Ma ch 2015 |Volume 6 |A icle 104 |1
Salminen and Jalkanen Bud b eak o Sco s pine
empe a u e sum. The e o e he final leng h o he leade shoo is
mainly de e mined by he cu en -yea empe a u e. In con as ,
he final leade -shoo leng h in Sco s pine is p e-de e mined by
he p e ious-yea summe empe a u e (Lanne , 1968;Jun ila,
1986;Salminen and Jalkanen, 2005). As he empe a u e o he
cu en season canno be o ecas ed, i is a he isky o pine o
be de e mined o g ow much in a cold season. To minimize he
isk o ea ly os damage on he leade shoo and bud, pine’s
s a egy is o flush as quickly as he day leng h and empe a u e
de elopmen alone will allow. Ne e heless, i is e y necessa y
o bo h species o p og ess h ough hei phenological phases,
including bud b eak, in a clea ly sho e ime in no he n han
mo e sou he n la i udes simply o su i e in he ha sh no he n
ee-line condi ions.
To ou knowledge, bud b eak o Sco s pine a high la i-
udes has no been s udied p e iously using a ime-scale o
se e al decades. The aims o his s udy a e (1) o calib a e
p e iously published bud-b eak models (Hänninen, 1990) wi h
field obse a ions and me eo ological da a, (2) o es he di -
e en ypes o bud-b eak models, and (3) o econs uc a
100-yea -long bud b eak ime-se ies o selec ed si es using a ail-
able empe a u e da a in o de o es ablish how econs uc ed
bud-b eak da es in no he n Finland eflec he supposed clima e
change.
MATERIALS AND METHODS
FIELD OBSERVATIONS
Bud b eak was ollowed al oge he in fi e di e en Sco s pine
sapling plo s ep esen ing wo di e en loca ions in he no he n
bo eal zone in Lapland, Finland, h ough nine g owing seasons
in 2001–2003, 2008–2010, and 2011–2013 (Table 1). The num-
be o obse a ion yea s was six in one and nine in he o he
si e, and he same ees we e ollowed o h ee consecu i e
yea s. Thus he empi ical ma e ial consis s o fi e di e en ee
g oups. The heigh and age o he sample ees in he begin-
ning o he s udy pe iod we e 1.2–3.0 me e s and 8–20 yea s
(Table 1). Fo accu a e measu emen o g ow h accumula ion,
a pe manen pin was inse ed h ough he main s em o he
p e ious-yea shoo . Heigh g ow h o he leade shoo o he
s em was measu ed usually a leas once a week. The bud b eak
o a plo was defined o ake place when mo e han hal o
he sample ees had gained a leas 1 mm o leng h and he
mean inc emen o a plo was a leas 1% o he o al heigh
g ow h o ha season. I hose c i e ia we e ulfilled be ween
he once-a-week measu emen in e als, which was usually he
case, linea in e pola ion was used o es ima e he da e o bud
b eak.
Ai empe a u e was eco ded hou ly a he si e ei he du -
ing he whole yea o h oughou he g owing season using a
Tiny ag Ul a da a logge (Gemini Da a Logge s L d.) equipped
wi h an ex e nal senso ( ange –40 o +125◦C and esolu-
ion 0.4◦Ca +25◦C). Me eo ological measu emen s aken a
3-h in e als ( ans o med in o mean daily alues) we e also
a ailable om he nea es o ficial clima e s a ions (Table 2).
The da a we e calib a ed using mon hly mean empe a u e
di e ences be ween he s udied s and and he espec i e cli-
ma e s a ion. I onsi e measu emen s we e missing, as was he
case ou side he g owing season in some yea s, espec i e cal-
ib a ed alues om he nea es o ficial clima e s a ion we e
used.
Table 1 |Field obse a ions.
S and Plo Yea s Loca ion Heigh abo e
see le el (m)
Si e ype Mean heigh o
sample ees (cm)
Numbe o
sample ees
(1) Laanila, Ina i 1 2001–2003 N 68◦30E27
◦28215 D yish hea h 169 5
2 2008–2010 N 68◦30E27
◦28215 D yish hea h 178 101
(2) Van auskoski, Ro aniemi 1 2001–2003 N 66◦22E26
◦43150 D yish hea h 149 15
2 2008–2010 N 66◦22E26
◦43150 D yish hea h 188 5
3 2011–2013 N 66◦22E26
◦44140 Mesic hea h 202 10
1Due o mo ali y, h ee new sample ees we e measu ed in 2010.
Table 2 |Loca ion o he me eo ological s a ions.
S a ion S a ion
numbe
Owne Loca ion Heigh abo e
see le el (m)
Daily empe a u e
da a
Apukka, Ro aniemi, Finland 7502 Finnish Me eo ological Ins i u e N 66◦35E26
◦01106 1961–2014
Sodankylä, Finland 7501 Finnish Me eo ological Ins i u e N 67◦22E26
◦39179 1908–2014
I alo, Ina i, Finland 9601 Finnish Me eo ological Ins i u e N 68◦40E27
◦34 123 1958–2014
Ka asjok, No way 97250 No wegian Me eo ological Ins i u e N 69◦47E25
◦48 155 1876–2004
Ka asjok, No way 97251 No wegian Me eo ological Ins i u e N 69◦46E25
◦50 131 2004–2014
F on ie s in Plan Science |Func ional Plan Ecology Ma ch 2015 |Volume 6 |A icle 104 |2
Salminen and Jalkanen Bud b eak o Sco s pine
METHODS
The basic idea o model calib a ion was simple: a selec ion o
bud-b eak models we e d i en by daily empe a u e da a and
esul s we e compa ed o ac ual obse a ions. Fi s , he daily em-
pe a u e da a o he calib a ion pe iods we e impo ed and he
bud-b eak models we e implemen ed in Mic oso Excel (2010).
Second, we used an e olu iona y sol ing me hod (gene ic op i-
miza ion algo i hm) add-in included wi h Excel o selec hose
model pa ame e s ha esul ed in he smalles oo mean squa e
e o (RMSE) compu ed om modeled and obse ed bud-b eak
da es. Bud b eak was defined as he da e when mo e han hal
o sample ees had hei leade -shoo bud elonga ed a leas
1 mm and he mean inc emen o he plo was a leas 1% o
he o al heigh g ow h o ha season. The ollowing op ions
we e se o he e olu iona y me hod: con e gence 0.0001, mu a-
ion a e 0.1, popula ion size 250, and maximum ime wi hou
imp o emen 60 s. Op imiza ion was es ic ed by se ing he
allowed bounds o each pa ame e (Table 3). The op imiza-
ion algo i hm sc eened o he lowes RMSE wi hin he gi en
pa ame e ange. Each model was op imized sepa a ely, and he
numbe o pa ame e s depended on he model in ques ion. Fo
example, when fi ing he day-leng h- igge ed model, op imiza-
ion adjus ed he pa ame e s o he o cing model as well as he
day-leng h h eshold o o cing in o de o minimize RMSE. A
compa ison be ween he models was based on he ou come o
he op imiza ion: he lowes RMSE and he espec i e pa ame e
alues.
Bud-b eak models gene ally include componen s o es b eak,
a ainmen o g ow h compe ence and he ac ual on ogene ic
de elopmen (Hänninen, 1990,1995).Thefi s modelsca ch
he e ec o g ow h condi ions on de elopmen om ull do -
mancy o g ow h compe ence, and he la e desc ibes how
ac ual o cing e en ually esul s as a bud b eak. The de el-
opmen om es o g ow h compe ence is usually exp essed
as a unc ion o accumula ed chilling o pho ope iod o bo h.
The e a e a ious ways in which he de elopmen o g ow h
compe ence can be inco po a ed in bud-b eak models (see Hän-
ninen, 1990;Chuine, 2000). We chose o es he ou models
by Hänninen (1990), which desc ibe how accumula ed chilling
a ec s he po en ial a e o o cing. The models a e deno ed
acco ding o he o iginal a icle by IA,I
B, II, III, and IV
(Hänninen, 1990;Table 2). As a compa ison, we applied a
simple model whe e a fixed day o yea eleases he accumu-
la ion o o cing uni s (FUs). Tha model is deno ed by DL
(day-leng h).
Besides modeling he on ogene ic compe ence, o cing can also
be calcula ed using di e en ypes o equa ions ha a e usually
d i en by empe a u e. In his s udy, he FU equa ion based on
da a by Sa as (1974) and o mula ed by Hänninen (1990, Eq. 8)
was applied (Eq. 1).
m c ( )=⎧
⎨
⎩
0FU day−1,T( )≤0◦C
a·FU day−1
1+e−b·◦C−1·(T( )−c◦C)T( )>0◦C
(1)
whe e m c( ) is he po en ial a e o o cing in FUs pe day, and
T( ) is he p e ailing ai empe a u e.
The o iginal pa ame e s (Table 3) o his sigmoidal model a e
based on pooled da a o se e al bo eal species (Sa as, 1974;
Hänninen and K ame , 2007). The e o e, we used he o iginal
pa ame e s as s a ing alues and eleased hem o be selec ed by
he op imiza ion algo i hm wi hin a p edefined ange (Table 3).
The models gi ing he bes pe o mance in RMSE we e used o
econs uc a bud-b eak da e when d i en by he longes a ailable
daily empe a u e eco d measu ed since 1908 a he Sodankylä
me eo ological s a ion o he Finnish Me eo ological Ins i u e in
no he n Finland. As a compa ison, ano he ime se ies was con-
s uc ed based on empe a u e eco ds om Ka asjok, no he n
No way a ailable since 1877. The long daily empe a u e da a
we e impo ed o he same Excel implemen a ion o he bud-b eak
models as cons uc ed o he calib a ion phase. The s a is ical
significance o he end and au oco ela ion o he econs uc ed
bud-b eak se ies we e assessed wi h he ime-se ies analysis p o-
cedu e AUTOREG, which is included wi h he SAS s a is ical
so wa e (SAS Use ’s Guide, 2002).
RESULTS
BUD-BREAK DATES
The obse ed bud-b eak da es du ing he pe iods 2001–2003 and
2008–2013 showed la ge a ia ion in g ow h onse (Table 4). E en
du ing he sho 3-yea pe iod ha each plo was measu ed, he
ange o bud-b eak da es wi hin he same plo co e ed one week
Table 3 |The bounds o he model pa ame e s.
Model Op ion De aul Minimum Maximum
Fo cing Fo cing equi emen (FUc i )∗50 0 250
ain Eq. 1∗28.361 15 40
bin Eq. 1∗0.185 0.01 1.0
cin Eq. 1∗18.431 5 30
Chilling Chilling equi emen (CUc i )∗20 0 120
Res pe iod s a da e∗(fixed) Sep embe 1s Sep embe 1s Sep embe 1s
Day-leng h Quiescence pe iod s a da e Ma ch 15 h Janua y 1s May 31 h
Fo a de ailed pa ame e desc ip ion, see Eq. 1 and Hänninen (1990). *de aul alues a e aken om Hänninen (1990).
www. on ie sin.o g Ma ch 2015 |Volume 6 |A icle 104 |3
Salminen and Jalkanen Bud b eak o Sco s pine
Table 4 |The obse ed bud-b eak da es.
S and Plo Yea s Bud-b eak da es
(1) Laanila, Ina i 1 2001–2003 17.05.2001, 01.05.2002, and 13.05.2003
2 2008–2010 26.05.2008, 14.05.2009, and 18.05.2010
(2) Van auskoski, Ro aniemi 1 2001–2003 06.05.2001, 30.04.2002, and 09.05.2003
2 2008–2010 21.05.2008, 10.05.2009, and 13.05.2010
3 2011–2013 11.05.2011, 12.05.2012, and 14.05.2013
Bud b eak was defined as he da e when mo e han hal o sample ees had hei leade -shoo bud elonga ed by a leas 1 mm and he mean inc emen o he plo
was a leas 1%o he o al heigh g ow h o ha season.
Table 5 |Pa ame e s o he bud-b eak models and hei pe o mance (RMSE) a e op imiza ion.
S and (numbe o obse a ions) Model Da e o CU-c i e ia FU-c i e ia Pa ame e s o Eq. 1 RMSE
(1) Laanila, Ina i (n=6) DL Ma ch 30 h 124.2 a=19.8; b=0.73; c=6.1 1.58
IA15.4 127.9 a=31.9; b=0.53; c=7.8 1.58
(2) Van auskoski, Ro aniemi (n=9) DL Ma ch 7 h 196.2 a=21.5; b=0.47; c=6.0 3.12
IA21.7 112.4 a=15.8; b=0.95; c=6.2 3.45
on a e age, he la ges ange being 16 days. T ees in Laanila s a ed
hei g ow h a ew days la e han hose g owing abou 300 km
sou h in Van auskoski.
MODEL CALIBRATION
A simple day-o -yea - igge ed model (DL) ga e equal pe o -
mance o he simples chilling model (Model IA, Eq. 2) used
in connec ion wi h on ogene ic compe ence. Bo h o hese a e
sequen ial; o cing may s a a e he s a e o quiescence is
a ained. Model DL and Model IA esul ed in an RMSE o 1.6
and 3.1–3.4 o S ands 1 and 2, espec i ely, (Table 5). The o he
ou models o on ogene ic compe ence add essed by Hänninen
(1990) we e in e io o IA, so hei esul s a e no epo ed. In
Laanila in 2009, bud b eak ook place 3 days be o e p edic ed
(Figu e 1). In Van auskoski, he yea s 2003 and 2012 inc eased
he RMSE alue he mos , pa ly due o ea lie - han-p edic ed bud
b eak (Figu e 2).
Mchl ( )=
⎧
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎩
0CU day−1,T( )≤−3.4 ◦C
0.159 CU day−1◦C−1·−3.4 ◦C<T( )≤3.5 ◦C
T( )+0.506 CU day−1,
−0.159 CU day−1◦C−1·3.5 ◦C<T( )≤10.4 ◦C
T( )+1.621 CU day−1,
0CU day−1,T( )>10.4 ◦C
(2)
whe e Mchl( ) is he a e o chilling in chilling uni s (CU) pe day,
and T( ) is he p e ailing ai empe a u e.
The pa ame e s o he FU equa ion (Eq. 1) we e changed as a
esul o op imiza ion. The de aul alues (a=28.361, b=0.185,
FIGURE 1 |P edic ed and obse ed bud b eak in plo s 1 and 2 in
Laanila, Model IAwi h a chilling componen and Model DL wi h
day-leng h- igge ed o cing.
c=18.431) a e based on pooled da a o se e al bo eal species
(Sa as, 1974) and hus a e no species-specific. The pa ame e
alues ob ained in his s udy a e qui e di e en compa ed o he
de aul alues. Pa ame e b ha defines he slope o he empe a-
u e esponse cu e ecei ed a high alue compa ed o he de aul
alue, which esul ed in a much s eepe cu e (Figu e 3). Pa am-
e e cdefines he shape o he cu e, i.e., he empe a u e a he
u ning poin o he esponse cu e. The pa ame e alues a e
dependen on each o he and he change in he slope has an e ec
also on he shape.
BUD-BREAK RECONSTRUCTION
Model IA(includes a chilling e ec ) and he day-o -yea Model
DL we e used o econs uc bud-b eak da es based on he daily
F on ie s in Plan Science |Func ional Plan Ecology Ma ch 2015 |Volume 6 |A icle 104 |4
Salminen and Jalkanen Bud b eak o Sco s pine
FIGURE 2 |P edic ed and obse ed bud b eak in plo s 1, 2, and 3 in
Van auskoski, Model IAwi h a chilling componen and Model DL wi h
day-leng h- igge ed o cing.
FIGURE 3 |Fo cing uni (FU) equa ion esponse o empe a u e on
h ee di e en pa ame e se s; de aul s (dashed line), pa ame e s
sol ed o Laanila (solid line), and pa ame e s sol ed o Van auskoski
(do ed line).
empe a u e eco d om Sodankylä and Ka asjok since he yea
1908 and 1877, espec i ely. The di e ence be ween Ap il and May
mean empe a u es in each o he si es du ing he model calib a-
ion yea s was used o scale Sodankylä and Ka asjok empe a u e
eco ds. Hence, when using Sodankylä da a wi h pa ame e s om
Laanila, daily empe a u e alues we e educed by 1.4◦C and when
using pa ame e s omVan auskoski inc eased by 0.7◦C. Likewise,
when bud b eak in Laanila and Van auskoski we e econs uc ed
wi h da a om Ka asjok, empe a u e was educed by 0.4◦C and
inc eased by 1.7◦C, espec i ely. Thus, he o m o he econ-
s uc ed bud-b eak da es emained in ac bu hei le el was in
line wi h he mean empe a u es o he calib a ion da a. Due
o he missing obse a ions in Ap il–May o empe a u es in
Sodankylä and Ka asjok, he espec i e yea s 1918 and 1877–1889
we e omi ed om he esul s (Figu es 4 and 5).
In Laanila, Model IA(includes a chilling-e ec ) p edic ed
la e bud b eak han he Model DL wi h empe a u e da a om
Sodankylä da a, while da a om Ka asjok he o de was e e sed
(Table 6). The di e ence be ween he Sodankylä and Ka asjok
empe a u e eco ds is s ongly a ec ed by he adjus men s based
on he Ap il–May empe a u es. Hence, he absolu e alues
p oduced based on hese wo empe a u e eco ds canno be
di ec ly compa ed. Ins ead, he ocus should be on ends and
a ia ion.
A ime-se ies analysis e ealed ha none o he cons uc ed
se ies included s a is ically significan au oco ela ion. A possible
linea end in bud-b eak da es was analyzed by fi ing an au o e-
g essi e model o he da a. Se ies based on empe a u e eco ds
om Sodankylä did no include a significan linea end. Tem-
pe a u e eco ds om Ka asjok ga e sligh ly con adic o y esul s.
Model IAbased on pa ame e s o Van auskoski p oduced a se ies
wi h no s a is ically significan end, bu he h ee o he se ies,
Model DL o Van auskoski and Models IAand DL o Laanila,
included a end ha was s a is ically significan acco ding o he
maximum likelihood (ML) es ima es bu insignifican based on
o dina y leas squa es (OLS) es ima es. The o me a e known o
o e es ima e esidual a iance while he la e unde es ima es i
(SAS Use ’s Guide, 2002). The e o e, he conclusion is ha he e
is a linea end ha is bo de line s a is ical significan (p=0.05).
Coe ficien s o he ime- a iable in he ML es ima ion in he
Ka asjok da a anged om –0.028 o –0.045, which co esponds
o a 3–4 day ad ance in bud b eak in he las 107 yea s.
DISCUSSION
Cla k e al. (2014) poin ou se e al sou ces o e o in phenolog-
ical modeling. Ou s udy is no ee o hem. We se a minimum
equi emen o measu ed heigh g ow h a e bud b eak was
obse ed and applied linea in e pola ion when u ning weekly
obse a ions in o a p ecise da e. The weekly measu emen in e al
is a comp omise be ween cos s and da a p ecision. Fu he mo e,
he chilling and o cing models we e d i en by daily mean em-
pe a u e da a, which fla ens ou he e ec o minimum and
maximum alues, while we did no pay a en ion o sola adia ion,
soil empe a u e, g ound os , o snow mel . These simplifica-
ions we e due o he ela i ely sho and incomple e da a eco ds
a ailable.
Compa ison o he modeled and obse ed bud-b eak da es
e ealed ha due o he defini ion o bud b eak, he models pe -
o med be e when he empe a u e change om cool o wa m
was clea and s eep. In con as , when he empe a u e change
was slow, he models ga e la ge RMSE alues. A ypical example
is seen in 2008 in he sou he n loca ion, when hal o he ees
began o g ow ollowed by a b eak o o e a week, a e which he
es o he ees also b oke. Acco ding o he chosen c i e ia, he
ime when he la e g oup o ees s a ed o elonga e was selec ed
as he bud-b eak da e. Daily obse a ions and longe con inuous
field measu emen s would enable mo e eliable calib a ion o he
models.
Bud-b eak models by Hänninen (1990) we e calib a ed wi h
field obse a ions om wo di e en loca ions in no he n
Finland. Empi ical ma e ial consis s o fi e di e en ee g oups,
each ollowed o h ee consecu i e yea s. These sample da a we e
ed in o an op imiza ion p ocedu e ha selec ed pa ame e s o
chilling and o cing models and which ga e he lowes RMSE.
www. on ie sin.o g Ma ch 2015 |Volume 6 |A icle 104 |5

Salminen and Jalkanen Bud b eak o Sco s pine
FIGURE 4 |Recons uc ed bud b eak using adjus ed empe a u e eco ds om Sodankylä and pa ame e s o wo models, Model IAwi h a chilling
componen and Model DL wi h day-leng h- igge ed o cing, sol ed o Laanila (A) andVan auskoski (B).Yea 1918 was omi ed due o he missing
empe a u e obse a ions.
FIGURE 5 |Recons uc ed bud b eak using adjus ed empe a u e eco ds om Ka asjok and pa ame e s o wo models, Model IAwi h a chilling
componen and Model DL wi h day-leng h- igge ed o cing, sol ed o Laanila (A) andVan auskoski (B).Yea s 1877–1889 we e omi ed due o missing
empe a u e obse a ions.
Table 6 |Mean bud-b eak da es (day o yea ) wi h wo empe a u e
eco ds, Sodankylä and Ka asjok, and wo models (IAand DL)
pa ame e ized o Laanila and Van auskoski.
Laanila,
Model IA
Laanila,
Model DL
Van auskoski,
Model IA
Van auskoski,
Model DL
Sodankylä 144 142 138 135
Ka asjok 146 155 137 150
Bo h pu e nume ical pa ame e s and fixed-day-o -yea , when
applicable, we e analyzed simul aneously. Ou app oach exem-
plifies how young ees o oday would ha e eac ed o he chilling
and o cing empe a u es o he pas clima e. The size o he da a
does no jus i y ex ensi e gene aliza ion and a he han ocusing
on he pa ame e s o he models, he emphasis is on he ends
shown by he econs uc ed bud-b eak se ies. The e a e isible
cyclic changes bu no clea long- e m ends ha can be dis in-
guished in he se ies based on Sodankylä empe a u e da a; he
bud-b eak da es o he 1920s and he 1950s a e as ea ly as hose
seen in he beginning o he 2000s. This is no consis en wi h he
esul s o ecen a icles in he field o phenology, which ha e usu-
ally confi med a long- e m empe a u e ise (e.g., Linkosalo e al.,
2009). This s udy is based on obse a ions om no he n Finland
and ocus on ea ly sp ing empe a u e, which may explain why
findings a e no he same when analyzing sou he n loca ions and
F on ie s in Plan Science |Func ional Plan Ecology Ma ch 2015 |Volume 6 |A icle 104 |6
Salminen and Jalkanen Bud b eak o Sco s pine
empe a u e la e in he g owing season. On he o he hand, da a
om Ka asjok indica ed ha he e is a linea end, al hough i s
s a is ical significance was no e y clea . Al hough obse a ions
om Ka asjok and Sodankylä a e simila o a la ge deg ee, heAp il
and May mean empe a u es om Ka asjok in pa icula ha e a
small inc easing end o e he cen u y ha makes a di e ence.
Recen findings, o example, on empe a u e de elopmen based
on dend oclima ological s udies o pine in he uppe Fennoscan-
dia and NW Russia (McCa oll e al., 2013;Lindholm e al., 2014)
a e in line wi h ou esul s on he mal de elopmen .
Ou esul s o a end owa ds ea lie bud bu s in pine a e no
en i ely in keeping wi h esul s o downy bi ch (Be ula pubescens
Eh h.) om he same a ea, which had p edic ed he po en ial
g owing season would inc ease conside ably (Bennie e al., 2010).
Simila ly Heikinheimo and Lappalainen (1997) and Pudas e al.
(2008) see ha sp ings seasons a high la i udes ha e ad anced.
The eason o he dispa i y cen e s on he leng h o he ime
span; pu simply, one is unable o sugges sp ing ad ance wi h
jus da a co e ing 7–15 yea s. E en gi en he leng h o ou econ-
s uc ion pe iod, he maximum ins umen al pe iod since 1908
in Sodankylä may be oo sho , since i canno indica e low
equencies in he se ies a all.
ACKNOWLEDGMENTS
We acknowledge Ta mo Aal o and Pekka Nä hi o echnical assis-
ance. This wo k was pa ly unded by he EU p ojec PINE
(EVK2-2001-00182) and Academy o Finland (SA 138937).
REFERENCES
Bennie, J., Kubin, E., Wil shi e, A., Hun ley, B., and Bax e , R. (2010). P edic ing
spa ial and empo al pa e ns o bud-bu s and sp ing os isk in No h-Wes
Eu ope: he implica ions o local adap a ion o clima e. Glob. Change Biol. 16,
1503–1514. doi: 10.1111/j.1365-2486.2009.02095.x
Beuke , E. (1994). Adap a ion o clima ic changes o he iming o bud bu s in
popula ions o Pinus syl es is L. and Picea abies L. Ka s . T ee Physiol. 14, 961–
970. doi: 10.1093/ eephys/14.7-8-9.961
Chmielewski, F., and Rö ze , T. (2001). Response o ee phenology o clima e
change ac oss Eu ope. Ag ic. Fo . Me eo ol. 108, 101–112. doi: 10.1016/S0168-
1923(01)00233-7
Chuine, I. (2000). A unified model o budbu s o ees. J. Theo . Biol. 207, 337–347.
doi: 10.1006/j bi.2000.2178
Cla k, J.S.,Melillo, J.,Mohan,J., and Salk,C.(2014). The seasonal imingo wa ming
ha con ols onse o he g owing season. Glob. Change Biol. 20, 1136–1145. doi:
10.1111/gcb.12420
Doi, H., and Ka ano, I. (2008). Phenological imings o lea budbu s
wi h clima e change in Japan. Ag ic. Fo . Me eo ol. 148, 512–516. doi:
10.1016/j.ag o me .2007.10.002
Hänninen, H. (1990). Modelling bud do mancy elease in ees om cool and
empe a e egions. Ac a Fo . Fenn. 213, 1–47.
Hänninen, H. (1995). E ec s o clima e change on ees om cool and empe a e
egions: an ecophysiological app oach o modelling o bud bu s phenology. Can.
J. Bo . 73, 183–199. doi: 10.1139/b95-022
Hänninen, H., and K ame , K. (2007). A amewo k o modelling he annual
cycle o ees in bo eal and empe a e egions. Sil a Fenn. 41, 167–205. doi:
10.14214/s .313
Heide, O. M. (2003). High au umn empe a u e delays sp ing bud bu s in bo eal
ees, coun e balancing he e ec o clima ic wa ming. T ee Physiol. 23, 931–936.
doi: 10.1093/ eephys/23.13.931
Heikinheimo, M., and Lappalainen, H. (1997). Dependence o he flowe bud bu s
o some plan axa in Finland on e ec i e empe a u e sum: implica ions o
clima e wa ming. Ann. Bo . Fenn. 34, 229–243.
Jun ila, O. (1986). E ec s o empe a u e on shoo g ow h in no he n p o e-
nances o Pinus syl es is L. T ee Physiol. 1, 185–192. doi: 10.1093/ eephys/
1.2.185
Kaye, M., and Wagne , R. (2014). Eas e n deciduous ee seedlings ad ance sp ing
phenology in esponse o expe imen al wa ming, bu no we ing, ea men s.
Plan Ecol. 215, 543–554. doi: 10.1007/s11258-014-0322-2
Lanne , R. M. (1968).“Pa e ns o shoo de elopmen in Pinus and hei ela ionship
o g ow h po en ial,” in T ee Physiology and Yield Imp o emen , edsM.G.R.
Cannell and F. T. Las (London: Academic P ess), 223–243.
Linde holm, H. W. (2006). G owing season changes in he las cen u y. Ag ic. Fo .
Me eo ol. 137, 1–14. doi: 10.1016/j.ag o me .2006.03.006
Lindholm, M., Ogu so , M. G., Jalkanen, R., Gunna son, B. E., and Aal o, T.
(2014). Six empe a u e p oxies o Sco s pine om he in e io o no he n
Fennoscandia combined in h ee equency anges. J. Clima ol. 2014:578761.
doi: 10.1155/2014/578761
Linkosalo, T., Häkkinen, R., and Hänninen, H. (2006). Models o he sp ing phe-
nology o bo eal and empe a e ees: is he e some hing missing? T ee Physiol.
26, 1165–1172. doi: 10.1093/ eephys/26.9.1165
Linkosalo, T., Häkkinen, R., Te hi uo, J., Tuomen i a, H., and Ha i, P. (2009). The
ime se ies o flowe ing and lea bud bu s o bo eal ees (1846–2005) suppo
he di ec empe a u e obse a ions o clima ic wa ming. Ag ic. Fo . Me eo ol.
149, 453–461. doi: 10.1016/j.ag o me .2008.09.006
McCa oll, D., Loade , N., Jalkanen, R., Gagen, M., G udd, H., Gunna son, B.,
e al. (2013). A 1200-yea mul ip oxy eco d o ee g ow h and summe em-
pe a u e a he no he n pine o es limi o Eu ope. Holocene 23, 471–484. doi:
10.1177/0959683612467483
Mic oso Excel. (2010). Mic oso Excel. Redmond, WA: Compu e so wa e.
Pudas, E., Leppälä, M., Tol anen, A., Poikolainen, J., Venäläinen, A., and
Kubin, E. (2008). T ends in phenology o Be ula pubescens ac oss he bo eal
zone in Finland. In . J. Biome eo ol. 52, 251–259. doi: 10.1007/s00484-007-
0126-3
Salminen, H., and Jalkanen, R. (2005). Modelling he e ec o empe a u e on
heigh inc emen o Sco s pine a high la i udes. Sil a Fenn. 39, 497–508. doi:
10.14214/s .362
Sa as, R. (1974). In es iga ions on he annual cycle o de elopmen o o es ees.
Au umn do mancy and win e do mancy. Comm. Ins . Fo . Fenn. 84, 1–101.
SAS Use ’s Guide. (2002). SAS/ETS Use ’s Guide, Vol. 1, 2. Ca y, NC: SAS Publishing.
Con lic o In e es S a emen : The au ho s decla e ha he esea ch was conduc ed
in he absence o any comme cial o financial ela ionships ha could be cons ued
as a po en ial conflic o in e es .
Recei ed: 15 Sep embe 2014; accep ed: 08 Feb ua y 2015; published online: 05 Ma ch
2015.
Ci a ion: Salminen H and Jalkanen R (2015) Modeling o bud b eak o Sco s pine in
no he n Finland in 1908–2014. F on . Plan Sci. 6:104. doi: 10.3389/ pls.2015.00104
This a icle was submi ed o Func ional Plan Ecology, a sec ion o he jou nal F on ie s
in Plan Science.
Copy igh © 2015 Salminen and Jalkanen. This is an open-access a icle dis ibu ed
unde he e ms o he C ea i e Commons A ibu ion License (CC BY). The use, dis-
ibu ion o ep oduc ion in o he o ums is pe mi ed, p o ided he o iginal au ho (s)
o licenso a e c edi ed and ha he o iginal publica ion in his jou nal is ci ed, in
acco dance wi h accep ed academic p ac ice. No use, dis ibu ion o ep oduc ion is
pe mi ed which does no comply wi h hese e ms.
www. on ie sin.o g Ma ch 2015 |Volume 6 |A icle 104 |7