Analyses of the spring phenology of boreal trees and its response to climate change

Analyses of the spring phenology of boreal trees and its response to climate change

Tapio Linkosalo

!"##$%&&&

'%()

*

+ , ,%%

- ./ %%

0 ,1 2 + ,1 23

3 ,1 23 123

3 1 2*

3 124!5

Työn ohjaaja/ Supervisor: !

"# #

Esitarkastajat/ Reviewers: $ "%"#

%$## &

Vastaväittäjä/ Opponent: '' " ()!& #

* # " "+

*

6"789':;9:8'<<:<=/- 8 , %&&&

Contents

Summary ... 5

Yhteenveto ... 6

List of original articles ... 7

List of symbols ... 8

1 Introduction ... 9

')' ))))))))))))))))))))))))))))))))))))))))))))))))) 8 ')%6 ))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) '& ')#6))))))))))))))))))))))))))))))))))))))))))) '' ');! ))))))))))))))))))))))))))))))))))))))) '% ')9/ ))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) '# ')< )))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) ';

2 The data sets ... 15

%)' )))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) '9 %)% )))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) '>

3 The data combination methods ... 18

#)'6 ))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) '> #)% ))))))))))))))))))))))))))))))))))))))))))))) '8

#)#))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) '8 #);? ))))))))))))))))))))))))))))) %&

4 Removing outliers ... 21

5 Comparison of the phenological time series ... 22

9)'6 )))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) %% 9)%5 )))))))))))))))))))))))))))))))))))) %<

6 Phenological models ... 27

<)')- ))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) %$ <)% )))))))))))))))))))))))))))))))))))))))))))))))))))) %> <)# ))))))))))))))))))))))))))))))))))))))))))))))))))))))))) %8 <); ))))))))))))))))))))))))))))))))))))))))))))))))) #&

7 Model evaluation ... 31

$)'/ ))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) #'

$)%5 ))))))))))))))))))))))))))))))))))))))))))))))))))))) #%

$)#5 )))))))))))))))))))))))))))))))))))) #;

8 Simulation of the impact of climate change on phenology 35

>)' )))))))))))))))))))))))))))))))))))))))))))))))))))))) #9

>)% ))))))))))))))))))))))))))))))))))))))))) #<

>)# ))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) #$

>); )))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) #$

9 Discussion ... 39

8)'! )))))))))))))))))))))))))))))) #8 8)%6 )))))))))))))))))) ;' 8)#! )) ;' 8);5@? )))))))))))))))))))))))))))))))))))))))))) ;%

8)96 ))))))))))))))))))))))))))))))))))))))))))))))))))))))) ;%

8)< ))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) ;#

8)$! )))))))))))))))))))))))))))) ;;

8)>5 )))))))) ;;

8)8 )))))))))))))))))))))))))))))) ;9 8)'&5 )))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))) ;<

Acknowledgements ... 46

Literature cited ... 48

9

Summary

* %&&&) ) - ./ %%)6"789':;9:8#<%:<667'%#9:;;;8)

)A , , "- ) "- ) '>8<:'899) )7 : ) )

=6 '8$%'8$;B

@

@ )"

$ ) )

) )A =B ) ) - @)

=,B ) )

<

Yhteenveto

,, ,, + , + , )C , '>8<:'899 6 + + , 2 : , + , + + , ) ,,++ +:

, ++ +, )+:+ : +)6:

++ D ++D ,++, + )0 ,+++ ++D , D++ , ++D+ D, : , )

D +++ +++, : + =6 '8$% '8$;B + ,,++ + : + + : ) ! , + ) ,, )0++ : )

,+D : ) * , , ++ , +++++ ) ++ +++

=DB ++D , , D+ + +)7+,++D + ++ +,:

, ) . , , ++D )

+++ =+,B ,,:

+D ++ +:

)6 + ++ ++

+++, , +, + )

$

List of original articles

? E6( : )6 )

6 +4* /='889B!

2 =- B 5 '>8<:'899) 15$%':$%<)

6 * +4/='88<B F

@) 16<<':<<;)

6 * ='888B4 )%#33=;B%#$:%;9)

6 * =%&&&B ! ). # ) )

6 * 54+4/= )B/

-) 2 ) 6 A'888)

2

- ./)A)"?%;

7:&&&'; ) ;&"

.) G )

8

Listof symbols

symbol unit description

Acrit days Startingdateofdormancy

bj days averagedeviationoftheobservationseriesjfromthe referenceseries

cj days averagedeviationoftheobservationseriesjfromthe combinedtimeseries

dj days randomblockeffectofobservationseriesj Dcrit arbitrary dormancy(chilling)releasethresholdvalue ε^{ij days randomresidualerror}

fD arbitrary rateofdormancydevelopment fO arbitrary rateofontogeneticdevelopment

m days constant(averagelevelofallobservations)

n numberofyears

ni numberofobservationsinyeari nj numberofobservationsinseriesj

tcrit days thresholdvalueofsignalfromthelightclimate Ocrit arbitrary phenologicaleventthresholdvalue

O^*crit arbitrary phenologicaleventthresholdvalue

s days² squaresumofdeviationsofobservationseriesfromthe combinedtimeseries

SD arbitrary stageofdormancydevelopment SO arbitrary* stageofontogeneticdevelopment S^*O arbitrary stageofontogeneticdevelopment

t days time

tdr days startingpointofontogeneticdevelopment τ^{i days fixedeffectofyeari}

Ti days teststatisticoftheoutlierdetectionprocedure χij days observationinseriesj*inyeari.

χ'ij days adjustedobservationinseriesjinyeari χir days observationinyeariinthereferenceseriesr zi days valueofcombinedtimeseriesinyeari

ži days modelpredictedmomentofphenologicaleventinyeari

8

1 Introduction

1.1 The annual rhythm of boreal plants

? )7 ) =)'B¹)

? ? F =* '8<8 I '8<8B)

* )) ) =))6 '8$%B ) ) ? ? ) )

Figure 1. The annual cycle of boreal trees, with the timing of the bud burst of Betula sp. as an example.

The dotted line of dormancy indicates that the timing of dormancy release and start of ontogenetic development is unclear.

1

See chapter 6.2 for terminology of the annual rhythm.

'&

)

1.2 Studies of phenology

)'$#9 4 =6 '8$;B) ) '8<& 6 ? : )" ? - - "='8$%B ? )

) ) )6 ='8$;B

=B )4='8<#B : 5 ='88&B)I

!='8<$B?

) 6 = B ) =* '8<8B)I='89<'8<8B7='89$B

='8<;B ='8<>B )='88#'88#B ) C !)='8>8B 56='8>#B ) ='8<;B

? )

'' 1.3 Spring phenology and climate change

F ) =I '8$&6 '8$%'8$;*'8>#

'8>9 + '88<B) F '&&&&

:) @ =3'889B =*'8>;B) : ,

? =3+'8>$B )

A, =B ) : ) ? ) 6 = B =))!F '888B)

, ? ) =6'8><6 )'88$B)"

? , )5 @?

) ) ! ) ='88$B

'%

'8>''88& '^o5

=;^o5 B

'%J; ) '%K ;9^o7)

1.4 Modelling the impact of climate change

) ) ) +='88'B 6 0+ +=<%^o';(7%9^o%&(.B 5A₂ =3)'8>$B ? )3+)='889B *) 6 : )3='88;B + : =" '8>>B #7 L) +='88'B:.

+ F)*:

='88<B 6 :4)='88&B ) +='88'B)

6 @ )!)='8>8B '9 F , )+='889B ? !,+=<%^o;$(7#&^o9>(.B )

'#

)4)='88<B ? !,+

)

1.5 Phenological data sets

6

@) ) =0 )'889B =3)'88<!) '88$!F'888B) ="/'88'B) )

) A #3!%

) '>#&

'>8< ) '89& ) =')%B)

) : = ) '88;B

? )6 )

) ) ) )

';

2

? ) ? )*

: )A )

1.6 Aims

) 2

='B ! )

=%B ) )

=#B

)

=;B

)

=9B @ ? )

'9

2 The data sets

2.1 phenological data

'>8<:'899 #3!% =":

'8&9'8&<'8&$'8&>'8'&'8';'8';'8'8'8'8'8%&'8%' '8%''8%9'8%9/'8%$'8%$4'8%>'8#9'8#9'8#<

'8#$'8;''8;%'8;>'89%'89$B) '8<&:'8<9 ) = ' ) %B )

) - ) '>&

0+ +=<%^o';(7%9^o%&(.B '9 =)%B) '9

2,

*), *)- ) " *) - ) ")

? ) "

!=<&^o&<(7'8^o9$(.B -

=<8^o&<(7%$^o'%(.B3,+=<<^o;#(7%$^o%$(.B)6 ? )

- - -*) ) --4 - .) * - ) ) - - - )

'<

2 33 18 45

46 3 15 39 13 16

41 8

34 48 23

6 9

5 28,29 7

40 24,25 1

1730 47 22 11

4

49 12

1419 36,37,38

27 26

32 50 44

20,21

10 31

35 42 43

20o

60o

100km 66o

31^o Arctic circle

Jyväskylä

Figure 2. The sites of the phenological observations. The temperature data utilised was collected at the city of Jyväskylä. The numbers refer to Table 1.

'$

Table 1. The geographical location and number of observations for each observation site and species in the data utilised in the study

'>

2.2 Temperature records

60+ +=<%^o';(7%9^o%&(.B @

=)%B 0'>>#:0'8>' ! ) ) A =%B)

! '>>#:

'8&' )L '8'%:'8'<

)

3 The data combination methods

3.1 Straight averaging

) ) B "B =))B ) ) ? )?

MN=

B)

Table 2. Times of temperature observations in Jyväskylä (local times)

'8 3.2 Using a reference observation series

) ) ) ).

))

( )

4 4

M 5

LM LU

L M

=

∑

(1)

_j /

_ij /_ir

_j /) ) ? )

3.3 Iterative method

) 2 ) ) ) )

@ : @ ? =)) B ) - )

? =)#B) _j F=B) , ^[_LM^′ @ _ij=5B) 2_i

, =-B, @

%&

=.B) _j @

="B)I :, , j @ )

3.4 The mixed model of randomised block effects

,

=)) B

? : ) 2

4_LM = +τ_L +6_M +ε_LM (2)

_ij /

" = B

ti?_j

/_ij ) F ) @

@ ?)A "? t

i)

Figure 3. The flow chart of the data combination process. The symbols are: x_ij the original and the adjusted observation at site j in year i, c_j the average devia- tion of observations at site j from the combined time series, t_i the spatially averaged moment of phenological event in year i, and n_j the number of observations (sites) in year i.

%' ) : )

4 Removing outliers

7 )!

= B , )I : =:

) '88;B )

) )A )

: ) ,

= B ? ) ? 3='89#B ) _i 2

7

[ [

[ [

[ [

[ [

L

L Q L Q

L Q L

L L

L Q L

= −

−

−

−













max ^{( , ) ( ,} −⁾ ,

( , ) ( , )

( , ) ( , )

( , ) ( , )

1 1

2 1

1

(3)

, =_(i,1))))_(i,n)B ) ? ? )?

?

"* ='8$>B) ) )

- ) 0+ + '&

%%

#8 ) , )

2 '&'9

%>>'&$& ';;;='B) %&

- ! ) 7 ! 9$ )

5 Comparison of the phenological time series

5.1 Spatial comparison

: =.@)%B) ) : @ ) : ) : = B ) )) )¹

:

_j

=B )7 =O²O &);B) =);B = , ,

1

The underlying assumptions of data combination methods were poorly formulated in study III

%#

-12 -8 -4 0 4 8

59 60 61 62 63 64 65 66

/DWLWXGH^R

$YHUDJHGHYLDWLRQGD\V

Figure 5. The deviation of phenological event date at each observation site from that of the combined time series, averaged over years and species, as a function of latitude. The white circles show sites with elevation of less than 20m ASL, the black ones those above. The lines show the linear regression between the average deviations and the latitude (solid and dotted respectively).

Figure 4. The deviation of average date of phenological events at a specific site and of different species as a function of latitude. All observations of flowering and bud burst of Betula sp.

(black triangles and circles respectively) and Populus tremula (white triangles and circles respectively) are presented. The majority of regressions were similar (solid line), with flowering of Populus tremula (broken line) and Alnus incana (dotted line) deviating most.

-20 -15 -10 -5 0 5 10 15

59 60 61 62 63 64 65 66

/DWLWXGH^R

$YHUDJHGHYLDWLRQGD\V

%;

"B) :%&'& "

:''> ) : )

?) P %&

6* ) )6 ? ) ) )

² )

: )

? =)9B)?

=Q&)&%<B)

@ -

=)<B) - )) " , ) @ )

2

Note that there is a misprint in study III (p.242, bottom left) here

%9

-50 -40 -30 -20 -10 0 10 20 30 40

1895 1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955

<HDU

'LIIHUHQFHEHWZHHQHYHQWVGD\V

Figure 6. The annual average deviation of the phenological events from the bud burst of Betula sp. (shown as the zero-level). The events are: flowering of Alnus glutinosa, Alnus incana, Populus tremula, Betula sp. and Pinus sylvestris (white diamonds, white squares, white triangles, white circles and crosses respectively), and bud burst of Populus tremula (black triangles).

0.0 2.0 4.0 6.0 8.0

0 20 40 60

7LP HVSDQGD\V

506(GD\V

Figure 7. The root mean square error (RMSE) of the regression model between two phenological time series, as a function of the average temporal distance between the two. Note that the lower limit of RMSE increases with the time difference.

26

5.2Comparisonbetweenphenologicalevents

The range of the average occurrence of the phenological events studied was almost two months,fromtheflowering ofthe two species ofAlnus in lateApril to the flowering of Pinus sylvestris in mid-June. The time gap between the average dateof phenologicaleventsvariedfrom1.1 (budburst andflowering of Betula sp.) to 52 days (flowering of Alnus incana and flowering of Pinus sylvestris)(Table 3instudyIV).Thetimespanof thephenologicaleventsofthe combined time series between the years remained fairly constant for all species andphenomenainthisstudy(Fig3instudyIV).

Linear regression models were fitted between the phenological time series inorder to determine thepredictabilityofone series withanother. The fit of the regression models was estimated by root mean square error (RMSE) betweentheobservedandpredicteddates.

n z z

i i i

å

=

)2

( RMSE

(9)

wherezi isthevalueofthecombinedtimeseries inyear i,žithe modelpredicted momentof phenologicalevent inyear i,andnthe numberofyears.The RMSEs of the regression models were relatively uniform in magnitude, but showed a slight tendency to increase with the increasing time span between the average date of the events. This seems natural, as the further the occurrences of the eventsareapart,thelesssimilar aretheclimaticconditions prevailingthe events.

Thetwo series ofAlnusflowering hadthe largest RMSEs among the regression models when forecasting other phenological events. These two phenological phenomena takeplace quiteearlyin spring, andthus are probablydrivenbyless similarclimaticconditionsthantherest(Fig.7).

Theaveragetemporaldeviationofthe floweringandbudburstof Betula sp. was 1.1 days, and the RMSEs of the regression models between the two about 1.9 days. As the correlation between the events was high (0.971), the RMSEs ofthese two regression modelswas considered to be lergelydue to the inaccuracies in the phenological data, and the figure was utilised as a rough estimateofitsaccuracy(Fig.10).

%$

6 Phenological models

6.1. Definitions of dormancy

) ) ? ) )

='8<;BI='8<8B )) )6 ='8$;B 2 = 6 ? B = B )

6 ='89;B4='8<#)$9B*)='8>$B )+='88&B 3 ='88;B) @ )

) ) ) ) =))6 '8$;B)

))

%>

)

6.2 A few words about terminology

"

@ , )

6 ='8$;B " " "&

") )6 "

=56'8>#+'88&'88' 3'88;'889'88<B )

- " = B ) 6 ='8$;B 7 " "&&)-

=)>B)A

=6 '8$%B) : ) 6 ='8$; ) ;%B @ =)>B ) 6 : )

Figure 8. Rate of dormancy development (dotted line, scale to the left), rate of dormancy II development (Sarvas 1974, thin solid line, scale to the right) and rate of ontogenetic development (thick solid line, scale to the right) as a function of temperature.

0 0.2 0.4 0.6 0.8 1

-10 0 10 20 30 40 7HPSHUDWXUH^R&

'RU PDQ F\G HYHO RS PHQ W

FRHI ILFLH QWD UELWU DU\ XQ LWV

0 5 10 15 20 25

30 2

QWRJH QHWLF GHY HORSP HQW FRHII LFLH QWD UELWU DU\X QLWV

%8

! 6 ( =+ '88& '88' '889 3'88;'889'88<3+)'889*'88<5 )'88>'888+)'88>!F'888!F'888B ) 6 ='8$;B )+='88&B3='88;

'889'88<B8+='889B

") )" ,6 )) )

" "

" 1

1 ) ? :1 )I "

8 "

:) "

)

6.3 The chilling triggered model

) 6 ='8$% '8$;B)

@ )-

@ )6@

@ ) ? %_D 2

1_' + &_' + +

$ W

FULW

( )=

∫

_D

6 ='8$;B)-

#&

:#)9'&)%^o5

#)9^o5=)>B) ,_crit

) A%_{D crit}

) %_O

2

1₂ &₂ + +

W W

GU

=

∫

_O

6 ='8$%B )>) _dr )" %_O? )_crit=)8B)

6.4 The light climate triggered model

@ ) =!

'889A )'88$/)'88>B )

Wcrit

0 20 40 60 80 100 120 140 160

Sep Oct Nov Dec Jan Feb Mar Apr May Jun

6WD JHR IGRU PDQ F\

DUE LWUDU

\XQ LWVG

D\

0 20 40 60 80 100 120 140 160

6WD JHR IRQWR JHQH WLFG HYHOR SPHQ W DUE LWUDU

\XQLW VGD\

2crit

2crit

'crit

Figure 9. A comparison between the dormancy triggered model (dotted line) and the light climate triggered model (solid line) using 1955 as sample year. Dormancy accumulates until November, when the critical threshold value, Dcrit, is reached. This triggers ontogenetic development, which proceeds somewhat during a warm spell in December, starts again in April, and finally reaches the critical threshold value for bud burst, Ocrit, in May. Ontogenetic development according to the light climate model starts at the threshold date, tcrit, and also reaches the threshold value for bud burst, O*crit, in May.

31 Thus in this study another model, utilising the day length or some other annual featureof the light climate as the signaltriggering the ontogenetic development was used. It should be noted that in the boreal climate zone the winter is long andcold, andthe chilling requirement is usually met earlyin the winter (Sarvas 1974, Hanninen 1990, Heide 1993a, Leinonen 1996a). Assuming that the ontogeneticdevelopment is triggeredbyasignalfromthe light climate in spring does not denytheexistence ofthe chilling requirement, but rather suggests that the chilling is a necessary, but not a sufficient condition for ontogenetic developmentto commence.

The second model, referred to as the light climate triggered model, is thus a modification of Sarvas' model. In addition to chilling requirement, a regulatory mechanism related to the light conditions is assumed to hinder ontogenetic development until spring. This mechanism is operationalised as calendardate,i.e.ontogeneticdevelopmentinthemodelisassumedto beginat a constant, parametrised date, tcrit. The stage of ontogenetic development, S^*O, is once again described with a temperature sum type model, with the rate, fO, identicalto thatinthedormancymodel(Fig.8):

S^*(t) f (t)dt

t

t O O

crit

ò

= (6)

Bud burst takes place once the stage of ontogenetic development reaches a hresholdvalueO^*crit(Fig.9).

7 Modelevaluation 7.1Parameterestimation

The two phenological models (chapters 6.3 and 6.4) were fitted to all 7 combined time series (chapter 2. 1) byfinding a set of modelparameter values that minimise the root mean square error (RMSE) between the observed and predicteddates.

n z z

i i i

å

=

)2

( RMSE

(10)

where zi is the value of the combined time series in year i, ži the moment of a phenologicalevent inyear i predictedbythe model, and n the number of years.

An iterative optimisation algorithm (Hooke and Jeeves 1961) was utilised. The date parameters(startingdatefordormancyandontogeneticdevelopment)were

#%

'0#&0 )- =#B) :!=) '%B F)

,_crit

? ) =* '8$;B =4 ) '8$;B

=I '8>'B?=56'8>#+'88&'88' '8893'88;5)'88>'888B

=3 6+'8>9B)

) ") 4

='8<# B M N) ,_crit

M N )

7.2 Comparison of the two models

"- ) = " ,B = B

=)'&B) 4!6.

-=)'&B) 6 ) 4!6.

=#B) +( ='888B :: :

=. '88#B - =)%B

##

Table 3. Model parameter values and root mean square error of the model predictions

0 2 4 6 8 10 12

50 6( GD\

V

Flowering of

$OQ XVLQ FDQD

Flowering of

$OQ XVJ OXWLQ RVD

Flowering of

3RS XOXVW UHP XOD

Bud burst of

%HWX ODVS

Flowering of

%HWX ODVS

Bud burst of

3RS XOXVW UHP XOD

Flowering of

3LQ XVV\

OYHVWU LV

Figure 10. The root mean square error (RMSE) of the two phenological models with different phenological phenomena.

Diamonds indicate the chilling triggered model, and squares the light climate triggered model. The solid line shows the regression model RMSE between flowering and bud burst of Betula sp., which was considered to be an estimate of the phenological data accuracy.

#;

) ) ) F) )

7.3 Comparison of the results to other studies

); )

6 ='8$;B , "-) "

)*='88<B %>+='88&B 99& ) =#B ) 6 ='8$%B = ;B) + ='88&B 9& %&&

)

F3='88;B 6 (

= M @NB#

*) F

=; B) ) : @ )='889B - ) A ? ) 5)='88>B6 (=M @NB =M NB , ) @

=;B)

#9

)!

F 6 ( ) )),5)='888B : )6 F =3'88;'8895)'88>B)6 )

8 Simulation of the impact of climate change on phenology

8.1 The climate warming scenario

='88;B) 4 /55=6*!B=5:

) '88<B )A =

Table 4. A comparison between parameter values of phenological models. See text for the description of the models. Dcrit is the dormancy release threshold, and Ocrit the bud burst/ flowering threshold, independent of the symbols used in the original publications.

1Note that the threshold values presented are the originals divided by 24, as Sarvas calculated the units on an hourly, not a daily basis.

#<

B ) ');'=- B&)8;=!!B&)$'=0 B&)8;=6 7B)

5 =B0+ +8%:

&)9R5 S'&R5) 8%: =)):

B) '&R5 ? 6*! = ;)$R5 %'&&B) ,

? ? )

)6 )

8.2 Utilisation of the phenological models

=#B) ? )" %&

=>)'B) )

=)) B

#&0) : )

? 5 )6) '89$"='889B :9R5 ='88<B)

#$

:;)'^o5 - ) :>^o5

,-

, , =0 + B) :#)9^o5 =4'88#B)

8.3 Timing of bud burst

=)''B) <^o5 )I =)'%B)&

<^o5<)#

';)8,=)''B)

) ) %)>

'')',=)'#B) 8.4 Frost damage risk

" 2 =,, "B

=- ) "- ) B = 9)

- )

=)';-.B

=)';-.B) 9;K - )

$^o5

#>

Figure 12. Dependence of dormancy completion on mean annual temperature increase according to the chilling triggered model. Symbols as in Fig. 11.

Figure 11. Dependence of the average date of a phenological event on annual temperature increase according to the chilling triggered model.

Symbols: flowering of Alnus glutinosa (white diamonds), Alnus incana (white squares), Populus tremula (white triangles), Betula sp. (white circles) and Pinus sylvestris (crosses), and bud burst of Populus tremula (black triangles) and Betula sp. (black circles).

Figure 13. Dependence of the average date of a phenological event on annual temperature increase according to the light climate triggered model.

Symbols as in Fig. 11.

#8

@ )7

=)';-.B)

I =';LB)

, @ =)';"5B) ) ) )) ) ) "

?9&K 9^o5) ? )

9 Discussion

9.1 Methods for combining the phenological data

) ))

?)- ) =/55'88<B ="

/'88'B) : )

;&

)ORZHULQJRI$OQXVLQFDQD

0 20 40 60 80 100

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 9,0 10,0

0HDQDQQXDOZ DUPLQJ^R&

)URV WG DP DJH ULVN

A

)ORZHULQJRI$OQXVJOXWLQRVD

0 20 40 60 80 100

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 9,0 10,0

0HDQDQQXDOZ DUPLQJ^R&

)URV WG DP DJH ULVN

B

)ORZHULQJRI3RSXOXVWUHPXOD

0 20 40 60 80 100

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 9,0 10,0

0HDQDQQXDOZ DUPLQJ^R&

)URV WG DP DJH ULVN

C

%XGEXUVWRI%HWXOD VS

0 20 40 60 80 100

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 9,0 10,0

0HDQDQQXDOZ DUPLQJ^R&

)URV WG DP DJH ULVN

D

)ORZHULQJRI%HWXOD VS

0 20 40 60 80 100

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 9,0 10,0

0HDQDQQXDOZ DUPLQJ^R&

)URV WG DP DJH ULVN

E

%XGEXUVWRI3RSXOXVWUHPXOD

0 20 40 60 80 100

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 9,0 10,0

0HDQDQQXDOZ DUPLQJ^R&

)URV WG DP DJH ULVN

F

)ORZHULQJRI3LQXVV\OYHVWULV

0 20 40 60 80 100

0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 9,0 10,0

0HDQDQQXDOZ DUPLQJ^R&

)URV WGDP DJH ULVN

G

Figure 14. Frost damage risk, i.e. probability of occurrence of temperatures below a threshold value (see text) between the date of flowering/bud burst and 30 July, as a function of average annual increase in temperature. White bars refer to the chilling triggered model, black to the light climate triggered model.

;' =+'88'+:

)'88>B) ?) : )

9.2 Spatial and temporal regularities of boreal phenology

?99&

: : ) :

%&#& )/

? ) ?)6 F)

!) A )

@ ) )

9.3 Modelling dormancy and the onset of ontogenetic development - =6 '8$;+'88&'88#

*'88<B) : ) 6 : )

@

='88#!'889A )'88$/:

;%

)'88>B) )

9.4 Critique of the dormancy experiments

@

=@B

? T-

@ )A

? @ ) - 6 ='8$;B ?

@ )6?

: : :="U'88;B

?) =56'8>#*'88<B

@ ) 9.5 Signals from the light climate

)- =+ '889B ) ) 1 =0'8>&03'88&B

="U'88;B)

;#

:" ="U'88;B)

=6'8>$B ) ) =-4B =) '9B )

! )

9.6 The evolutionary argument

)6 =*F '88#B) ? ) A =* '8<8B)

)1

="U'88;B) =0'8>&/ )'8><

03'88&B) )

5 T . MN )6 F =?? B )

;;

)

9.7 Modelling the spring phenology of boreal trees

"

) ?) -

=+'888B )

) 6

@ =))

!'889A )'88$/)'88>B) =3'88;'8895)'888B 6 ='8$% '8$;B ) ) 9.8 Climate change, phenological timing and frost damage risk

) ) - ) 9&K =)';-.

B)"

,=)';"5B) , ? ? )A ? ) )) =) '9B )

= '88;3 '889B)A

;9

)

9.9 The recent lengthening of the growing season

6 '^o5 ;9^o7

;^o5)) =!)'88$B)6

5A₂

>'%

=3)'88<!)'88$!F'888B)

=8)&>)>

- ) " B

=%)>#)' B)

0.0 2.0 4.0 6.0 8.0

7HPSHUDWXUHUDQJH

R&

-15 -10 -5 0 5 10 15 20

7HPSHUDWXUH

& R

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 15. Diurnal temperature variation (afternoon observation minus daily minimum) and daily average temperature as a function of date (both 1883-1981 mean). The thick solid line shows the 14-day moving average with the thin line showing daily observations of the temperature variation. The smooth thin solid line shows the 14-day moving average of daily average temperature.

;<

9.10 Concluding remarks

)6 ? , ) ? ) @ )

) A =))I'8<84)'88;

'88$B ) )?

?) ? ) )

"

) )6 ? ? )

Acknowledgements

, 0 0 5) '88;:'88< @ )

;$

/ // !3- , 4 )

/ / ) '88%

4 ) -)4 +)

@ )6$ )

5, )A ) ,)A 2 V ) )/

4 )

- . * . / ) )

4

@ )W

;>

Literature cited

"I)'8>>)- 2) )& '2 )/!)*)5 )4)3,7))= )B3/ - )'%9:'9$)

"U5)*)'88;)* 26 )& 5 !)!) / 4)I) = )B .? )

;%8)

" )* )'8$>)A )I7C

#<9)

")*)/0)'88')L 6 2 )).) 92;&#:;&8)

"/)'889)":

@ )7)0))6)6)7)%&2':98)

"/)'88<)":

)7)0))6) '&='B2':<)

" ))'8&9)/F+ "'8&#) : )

" ))'8&<)/F+ "'8&;) : )

" ))'8&$)/F+ "'8&9) : )

" ))'8&>)/F+ "'8&<) : )

" ))'8'&)/F+ "'8&$) : )

" ))'8';)/F+ "'8&>) : ))$<7)%)

;8

" ))'8';)/F+ "'8&8) : ))$<7)#)

" ))'8'8)/F+ "'8'&) : ))$$7)<)

" ))'8'8)/F+ "'8'') : ))$$7)$)

" ))'8%&)/F+ "'8'#) : ))$>7)9)

" ))'8%')/F+ "'8';) : ))>&7)')

" ))'8%')/F+ "'8'9) : ))>&7)%)

" ))'8%9)/F+ "'8'<) : ))>&7);)

" ))'8%9)/F+ "'8'$) : ))>&7)9)

5!)L)4)64))'8>#) )0)).)%&289':8<#)

5 )4) / !) ) '88<) 6*! 2 ( ) L #%2%#9:%<&)

.") 4)0)'88#) )5 X7C;#<)

!)5 4)'88&)"=#

*)B) . )/ )<2;%8:;#>)

4).) 0) '8<>) 6 2 )/ )/)%'2'%;':'%;>)

+ 4) '888) 6 2 - )/ )'82<'#:

<'>)

9&

+ 4) * ) /) '88>) .

- )

/ )'>2$&$:$'%)

+)'88&)!

))))%'#;$)

+)'88')- T/5.)';2;;8:;9;)

+)'889).

2 )5)0)")$#=%B2'>#:'88)

+)/)'88<) 6 )&

/)4 0)!!)= )B/ 6 V )))%9;2'':%')

/)+4)'88')F ) / )>2%>':%>$)

A)!) '8$$) / 2 )/ )/)#82%9:#%)

A)!)'8>9)/

: )&/

)3Y)0A)7 0)= )B7:

/ Z)':%%)

A)!)'88#)- )/ )/)>>29#':9;&) A)!)'88#)- =#B@

)/ )/)>82'>$:'8')

4) '88;) 5 ))!) )5)'%%&8)

4)0 ))'8<')M- N )0) )5)!>2%'%:%%8)

/55'88<)5'889) )5 I L ) )

9' 0))*)L)!)5")) 7)3 )! 3)= )B)5 / 58:

;$)

0 )0 )3.)3 .)'889)5 7 )5)4 )92'>':'89)

0 A) '8>&) . %-)/ )/);>2#;$:

#9%)

0A)3Y)'88&).

% )6)0))4 92'89:%&;)

3 .) /) '889) - 7 2 ))4 )

#$2%'':%%>)

35)-)50))6)I)/)'88<) 5A₂ )7#>%2 ';<:';8)

3+6)'8>$)! +)0 '8>$)= )B 3+6)+)3 D!)'889)5

6 ).) )9='B2;%:9%)

3.)/)'89#)A , )0) )6) );>29#':9##)

33)'88;) 7 L)/5.)'$2#<$:#$$)

33)'889)/ . )/5.)'>28#:'&;) 33)'88<)/.

)/- I I 7 )%'&)

3 )6+4)'8>9) )& /)!))//) 3 )= )B5 ) ))'<$:'8#)

9%

* 0)0)'8$;)V ))")#>2'&'#:'&%#)

*L)).0)-)!L)5)-4)*)'8>$).::

2 )6%%=#B2#$':#$$)

*!)0)'8>;)I T ))7)'%;2

>%':>;%)

*)'88<) F 6 ))")$>2<>$:<8#)

*)'88<)- - )6)0)) 4 )''2'%%:'%>)

*)4)+)'88$)5 6 ))")$82'##:

'#>)

* 4)'8<8)- )&- I )I)=)B6)6).?)")%#)5:

/ )':'&)

*0))'8>#)A )A)<&2#;:#$)

*F.)'88#) )5*/ *%%$)

!F ) '88$) /+ I + 3 : "

/+ L+ !D

! /+) + +![)';$)

!F)/)'888)L ?.)7

#8$2<98)

!!)")5!)L)4)64))'8>8)- ")0)).)%<2

<8#:$&&)

9#

!4)")35)-)5)0) L)74)4)'88$) '8>''88') 7#><2<8>:$&%

70)/)'89$)/ )/))6))6)

$&29%<:9;;)

A 0).)0A)!F)'88$)*:

% L₁ :#:)/5/ )#>=9B29#<:

9;&)

/ 0) 3 ) + ) '88>) . 7 =B) / )'>2>'':>'<)

/ .)7 0)0A)'8><)"

% )/)")<2 8':89)

/!)'8%$)/F+ "'8'>

'8'8'8%&) : ))>&7)$) /!)'8%$)/F+ "'8%'

'8%%'8%#) : ))>&7)>) 4)!++)+)'88&)!

)&0F)=)B! )6:

5)'92<':$;)

4)'88#) /- )4 8 : 0 )9#)

4 ) + ) 3+ 6) '88<) : 5A₂ 6 ) /5.)'82%&8:%'<)

4!)'8%>)/F+ "'8%;'8%9 '8%<) : ))>&7)8)

4 !) '8#9) /F+ " '8'%) : ))$>7);)

9;

4 !) '8#9) /F+ " '8%$

'8%> '8%8 '8#&) : ) )>9 7)#)

4!)'8#<)/F+ "'>8<'>8$

'>8>'>88) : ))>97);) 4!)'8#$)/F+ "'8&&'8&'

'8&%) : ))>97)9)

4!)'8;')/F+ "'8#':'8#9) : ))>$7);)

4!)'8;%)/F+ "'8#<:'8;&) : ))>87)')

4!)'8;>)/F+ "'8;':'8;9) : ))8%7)')

4!)'89%)/F+ "'8;<:'89&) : ))8%7)#)

4!)'89$)/F+ "'89':'899) : ))'&&7)')

4 .))66)-)I-)4)'8$;) E4( E.( ) 682##':##%)

4 /) ) 0 A) '88;) 6 - ) / )';29;8:9<')

4 /) + ) 3 /) 0 0).) 4 ) '88$) F? -- )/5.)%&2''88:'%&;)

40))'8<#)! ) 6-))'%8%)62L)%';) 6 4)!)'89;)- ))4)// )92

'>#:%&;)

6 6) '8>$) / =5 B)& '#' 5)/)=)B7"6L 6 )'<)

99 6 4)'8$%)

))5) )))$<)#2':''&)

6 4)'8$;) ))5) )))

>;2':'&')

60))'8><) )0) I 7C$#>)

6 /)0) - 4).) 4 -)) " )3) )L) "

0))5FL)0)-)6)!))75))67) 6)") :6 )'88$)!?

)6

%$929&%:9&8

) /) 4 ) + 4) '88;) : 2 ) ) -) ./)''8%)

I 4))I-)4)66)-)'8>').

)0) )6))6)'&<='B28':8;)

I/))'89<)/ ))4)// )

$2'8':%';)

I/))'8<8) )&- I )I)=)B6)6).?)")%#) 5 / 98>)

)'8<;)- ))4)// )'92'>9:

%%;)

I 5)'8$&)5 , )6'<82'%<8:

'%$>)

I )I)'8<8)- )6)6).?)") 7)%#)5 / 98>)

I0)!)'8<$).

/ )/)%&=#B2$##:$;9)