Tree Physiology 27, 1019–1025
© 2007 Heron Publishing—Victoria, Canada
Temperature sum accumulation effects on within-population variation
and long-term trends in date of bud burst of European white birch
(Betula pendula)
MATTI ROUSI1,2 and JAAKKO HEINONEN3
1
Finnish Forest Research Institute, Punkaharju Research Unit, Finlandiantie 18, Fin-58450 Punkaharju, Finland
2
Corresponding author ([email protected])
3
Finnish Forest Research Institute, Joensuu Research Unit, P.O. Box 68, Fin-80101 Joensuu, Finland
Received June 13, 2006; accepted October 30, 2006; published online April 2, 2007
Summary Within-population variation in phenology of boreal trees indicates their adaptability to climatic variations. Although interannual variations in date of bud burst have been
widely discussed, little is known about within-population variation, the key determinants for this variation and the effects of
this variation on estimates of trends in bud burst date. Over a
period of nine years, we monitored timing of bud burst daily in
30 mature white birch (Betula pendula Roth) trees in a naturally regenerated stand. Our results revealed not only large
interannual variation but also considerable intraannual variation among individual trees in date of bud burst, the maximum
within-population variation being four weeks. Bud burst can be
accurately predicted by the date when a threshold value of temperature sum in spring is reached (base temperature +5 °C).
Based on this temperature sum and past temperature records,
we estimated the trend in date of bud burst. The linear trend estimate based on the years 1926–2005 is an advancement of
1.2 days per decade (95% confidence interval, ± 0.7 days),
which is much less than that predicted by time series based on
coarser time intervals. We conclude that, because of large
interannual differences, and large annual within-population
variations in bud burst, estimates of bud burst date based on
measurements made over a period of only a few decades are unreliable.
Keywords: climatic adaptability, global warming, trends for
bud burst.
Introduction
Plant phenological change is one of the most easily observed
responses to climatic change (Badeck et al. 2004). Ground observations suggest that, in England, the spring phenologies of
plants and animals have advanced by 4.5 days in the past decade (Fitter and Fitter 2002). For trees, observations across
several European sites indicate that the beginning of the growing season has advanced by 2.7 days per decade (Chmielewski
and Rötzer 2001) and observations in the International
Phenological Gardens in North Europe show an advancement
of 0.31 days year – 1 (see Menzel 2000). In North America, the
changes during the last 35 years have been less striking, about
–0.18 day year – 1 (Schwartz and Reiter 2000). In a long budburst time series there is, however, a possibility for several error components (e.g., see Häkkinen 1999a, and Chmielewski
and Rötzer 2001). Moreover, the results may be dependent on
length of time between observations because in variable spring
temperatures, bud development may not advance linearly with
time.
Advancement of spring phenology has also been noticed in
meteorological satellite observations. The Normalized Difference Vegetation Index (NDVI) exploits variation in spectral
properties of deciduous plants, and although data reflect the
radiative properties of the canopy at only a coarse scale, it is
considered appropriate for assessing vegetation phenology at
the ecosystem level (Stöckli and Vidale 2004). In the canopy
of boreal forests of Fennoscandia and western Russia, birch is
the most important deciduous species. In Finland, for example, the forests are mostly coniferous (82% of the growing
stock volumes), with birch comprising 15% of the growing
stock (Anon. 2005). Consequently, birch should be a good
species for comparisons of ground observations with NDVI
data, which suggest a stronger trend to earlier phenology than
do ground observations: 0.54 days year – 1 based on data for
1982–2001 (Stöckli and Vidale 2004) or 8 ± 3 days based on
data for 1981–1991 (Myneni et al. 1997).
To determine the date of bud burst at the stand level, it is
necessary to account for variation within populations; however, this variation is largely unknown. In pioneer species with
prolific pollen migration, such as birch, genetic adaptation can
be rapid, depending on the magnitude of variation within and
among present populations. Studies based on enzyme
(Rusanen et al. 2003) and nucleotide variations (Järvinen et al.
2003) suggest that most of the genetic variation in silver birch,
also known as European white birch (Betula pendula Roth), is
found within populations. Previous research has indicated
variable heritability (h 2 = 0.06–0.63) for timing of spring bud
1020
ROUSI AND HEINONEN
burst in European white birch, depending on study year and
population (Billington and Pelham 1991).
Several recent studies have focused on the construction of
phenological models for predicting tree bud burst (e.g.,
Hänninen 1990, 1995, Karlsson et al. 2003). Physiology based
models often incorporate chilling units and light climate triggers, which are assumed to limit the risk of premature growth
initiation (Hänninen 1990, 1995, Schaber and Badeck 2003).
Weather is unpredictable (Lorenz 1993) and has genotypespecific effects on trees, thus exact observations made in natural conditions are needed for comparison with time series
based on observations made at longer time intervals. Our earlier research (Rousi and Pusenius 2005) indicated large genetic variation in timing of bud burst among birch genotypes
(plus trees, selected for growth) originating from southern Finland. Therefore, to study within-population variation and
interannual differences, we took a random sample of 30 mature trees from a naturally regenerated birch forest and observed the timing of bud burst on a daily basis for nine years.
Specifically, we attempted to: (1) measure within-population
variation in bud burst; (2) assess the climatic determinants for
bud burst; and (3) compare the population-specific phenological trends to ground and satellite observations made at
longer intervals.
Material and methods
Experimental trees and observation
The birch forest selected for study is a 1-ha mixed stand of
B. pendula and B. pubescens ( Ehrh.) that regenerated naturally after logging in 1979 in Punkaharju, Finland (61°48′ N,
29°19′ E). Thirty B. pendula trees were randomly selected for
study. Three trees in the original selection were replaced in late
1997, because one tree had fallen after a storm and two trees
were suspected to be B. pendula × B. pubescens hybrids (in
1997, only 27 trees were observed). Mean tree height was
about 15 m at the beginning of the study. For bud-specific observations, the top of each tree was reached from a tractor
equipped with ladders and, in later years, from an aerial platform.
For the observations, we selected two branches from near
the top of each tree, one facing north and one facing south (in
1998 we observed only one branch per tree). Daily observations of bud burst started before any buds had opened. A bud
was determined as open when the protective bud scales had
separated and the emerging new leaf was clearly visible (see
Figure 1 in Rousi and Pusenius 2005). On each branch we observed 10 buds, starting from the tip of the branch. All observations, from 1997 to 2005, were made by the same persons. In
an earlier work (Rousi and Pusenius 2005), high within-tree
heterogeneity in the timing of bud burst was observed, due to
the exceptionally late burst of some individual buds. To remove the effect of these abherrant buds in the earlier study, the
tree (branch) was considered as open when 90% of the buds
were open. In the present study, we observed no such heterogeneity in mature experimental trees; thus, tree-specific bud
burst was considered complete when all the observed buds (20,
or 10 in 1998) of each tree were open.
Temperature
Temperatures were recorded at a height of 2 m with standard
meteorological screens located about 1 km from the experimental forest. Additional temperature measurements were
made at the crown level (years 1999–2005) with Tinytalk
units (TK-0040, Gemini Data Loggers, West Sussex, U.K.) attached to meteorological screens. Degree-day (dd) values
(dd5, dd0 and dd–2) indicate the sum of positive differences
between mean diurnal temperatures (on an hourly basis) and
the baseline temperatures (+5, 0 and –2 °C, respectively).
Additional temperature records for the years 1926–2005
(longest available dataset) from three weather stations of the
Finnish Meteorological Institute (FMI) in Punkaharju
(61°48′ N, 29°20′ E), Jyväskylä (62°24′ N, 25°40′ E) and Helsinki (60°10′ N, 24°56′ E) were used to estimate the long-term
trend in bud burst date. Because hourly values were not recorded at the weather stations at the beginning of the 20th century, we computed diurnal mean temperatures as the mean of
the minimum and 0600, 1200 and 1800 h readings. In 1999,
the Jyväskylä weather station was moved from the town to the
airport. Weather data from Jyväskylä airport were available
from 1950 and data for the overlapping years indicated that the
threshold date at the airport was consistently 2.2 days later
than at Jyväskylä, and that there was no trend in the difference.
Therefore, for the combined 1926–2005 time series from
Jyväskylä, we subtracted 2.2 days from the 2000–2005 airport
threshold dates.
Results
Temperature sum accumulation and bud burst
Branch direction (south versus north) had a negligible effect
(mean of 0.2 days) on bud burst. In 70% of cases, buds on the
north- and south-facing branches opened on the same day, and
in 14% of cases, buds on the south-facing branch opened earlier. Because the north versus south orientation of the branch
had no clear effect on bud burst date, all of the date observations for both branches on each tree were combined for further
calculations.
For all study trees, the mean date of complete bud burst was
May 9. The earliest occurrence of bud burst was in 2000
(April 26) and the latest was in 1999 (May 22; Table 1, Figure 1). Interannual differences in mean date of bud burst comprised the major part of the total variation, but inter-tree differences also varied considerably among study years (Table 3). In
1999, the first tree with all buds opened was observed on
April 21, the last on May 22. In some years, the inter-tree
phenological difference was only a couple of days (Figure 1).
Because of year-specific inter-tree variation, the date on which
the earliest 50% of the population had opened their buds sometimes differed considerably from the date when all the trees
had opened their buds. On average, the difference was six days
(May 3 versus May 9); however, in 1999, the difference was
TREE PHYSIOLOGY VOLUME 27, 2007
TRENDS IN DATE OF BUD BURST
Table 1. Date of bud burst (Date), degree day temperature sum at the
time of bud burst (dd) and the number of trees observed (n) at full bud
burst and at 50% bud burst in Punkaharju. The base temperatures for
dd sum calculations were +5, 0 and –2 °C (dd5, dd0 and dd–2, respectively). For example, if +5 °C was the base temperature, the mean
value for bud burst of all the trees was 44.8 dd.
Year
Date
dd5
dd0
dd–2
n
100% bud burst
1997
May 14
1998
May 11
1999
May 22
2000
April 26
2001
April 29
2002
May 1
2003
May 12
2004
May 5
2005
May 17
44.8
48.3
70.5
41.1
42.2
40.4
41.4
40.8
33.6
177.6
166.6
262.9
147.7
149.3
173.5
172.1
151.8
205.4
295.0
266.5
397.3
263.7
233.2
296.0
277.1
251.9
322.2
27
30
30
30
30
30
30
30
30
Mean
44.8
178.5
289.2
50% bud burst
1997
May 12
1998
May 6
1999
April 23
2000
April 24
2001
April 27
2002
April 25
2003
May 10
2004
May 3
2005
May 14
26.4
26.3
23.3
30.7
28.0
14.4
29.5
23.2
21.3
149.2
119.6
119.1
127.3
125.1
117.6
150.2
124.2
178.2
262.5
209.5
198.3
239.3
205.0
228.0
251.2
220.3
288.9
Mean
24.8
134.5
233.7
May 9
May 3
14
18
17
29
26
15
18
27
21
one month (April 23 versus May 22; Table 1).
The mean degree-day temperature sums for 100% bud burst
were 44.8 (dd5), 178.5 (dd0) and 289.2 (dd–2). When the date
of bud burst was predicted based on the mean date of nine experimental years, the maximum annual error was 13 days (in
1999 and 2000) and the mean error was 7.3 days (Table 2).
However, if the prediction was based on the temperature sum,
the error was considerably less: 2.2, 4 and 3.3 days for dd5,
dd0 and dd–2, respectively. When the temperature recordings
were made at the height of branches on which bud burst was
recorded, the temperature sums (for the comparable 7-year period) were higher (i.e., 56.8 versus 44.3 dd based on dd5 and
200.6 versus 180.4 dd based on dd0), but the deviations were
of similar magnitude (data not shown).
Based on dd5, the mean temperature sum for the earliest
50% of the trees to open their buds was 40% less compared to
the dd required for all the trees (Table 1). When the date of bud
burst was predicted for the earliest fraction of the population,
the mean error was one day (Table 2). If the temperature measurements were made at the branch level, the prediction was
even more accurate (0.7 day). In some years, bud burst of a
large part of the population (26–29 trees, years 2000, 2001,
2004) occurred on the same day, thus the date for 50% of buds
burst corresponded to the majority of the population (Table 1).
Temperature accumulations were calculated from January 1
1021
on. For dd5, the precision of the prediction was independent of
the starting day (within the range from the January 1 to
March 31) and was always better than the precision of the low
threshold temperature predictions. There was large variation
in autumn temperatures (September 1–October 31, 244 dd in
2001, and 143 dd in 2002). However, the temperature sums of
previous autumns (September or October, and combined)
were unrelated (P = 0.871–0.247, n = 9) to the timing of next
spring's bud burst.
Bud burst trends
The estimate of the linear trend in annual mean date of bud
burst at Punkaharju, based on observations from 1992 to 2005
(Figure 1), was –1.7 days per decade (P = 0.7). The standard
error of the estimate was 6.8 days and the 95% confidence interval of the trend, assuming random year effects, was
± 9.9 days per decade. For the median dates, the trend was –2.1
± 10.5 days per decade (P = 0.35) and the standard error of the
estimate was 7.2 days.
Because the date of bud burst correlated strongly with the
date when the temperature sum 24.8 dd was achieved (r = 0.94
(P < 0.001) for the mean date and r = 0.96 (P < 0.001) for the
median date, see Figure 2), we also used temperature records
Table 2. Annual and mean deviation (days) between the observed date
of bud burst and the mean date of bud burst (∆Date) and date of mean
degree day sum (∆dd), and the number of trees with open buds (No.)
at full bud burst and 50% bud burst in Punkaharju. Degree day sum
base temperatures were +5, 0 and –2 °C (∆dd5, ∆dd0 and ∆dd–2, respectively). For example, the mean dd5 for 50% bud burst (24.8; Table 1) was reached 1 day before the bud burst date in 1998, and on the
day of bud burst in 1997, 2000, 2001 and 2003.
Year
∆Date
100% bud burst
1997
5
1998
2
1999
13
2000
–13
2001
–10
2002
–8
2003
3
2004
–4
2005
8
Mean
7.3
50% bud burst
1997
9
1998
3
1999
–10
2000
–9
2001
–6
2002
–8
2003
7
2004
0
2005
11
Mean
7.0
∆dd5
∆dd0
∆dd–2
No.
0
0
3
–9
–1
–1
–1
–1
–4
–1
–2
15
–8
–4
–1
–1
–2
2
0
–3
12
–4
–5
0
–1
–3
2
27
30
30
30
30
30
30
30
30
2.2
0
1
–3
0
0
–3
0
–1
–2
1.1
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
4.0
1
–3
–3
–1
–1
–2
1
–1
6
2.1
3.3
2
–3
–5
0
–3
–1
1
–1
6
2.4
14
18
17
29
26
15
18
27
21
1022
ROUSI AND HEINONEN
Figure 1. Bud burst of birch in
Punkaharju. Data are daily observations from two stands. Left
panel: The data for 1992–1997
are from a replicated experiment
consisting of micropropagated
plantlets of plus trees (redrawn
from Rousi and Pusenius 2005).
Right panel: The data for
1997–2005 are from mature trees
in a natural stand. The first (䉭),
median (䊉,䊉), mean (䊏) and last
(䉬) trees to open all buds are
marked for each year. For the mature trees, the bud burst temperature sums (base temperature
+5 °C) and SD (bars) are shown,
and the median tree is underlined.
The median date and the mean
date are the same except for years
1999 and 2002 (䊉).
of the years 1926–2005 from three FMI weather stations in the
southern part of Finland to determine the corresponding date
(Figure 3). Despite the different number of diurnal temperature measurements, the correlation between the dates based on
FMI data and our own data gathered at Punkaharju was 0.97
(see Figure 2). The pair-wise correlations of time series of
dates of 24.8 dd from the three weather stations varied from
0.83 to 0.86 (see Figure 3). The trend estimate for the modeled
date of bud burst based on the years 1926–2005 was –1.2
± 0.7 days per decade. If the calendar dates were replaced by
time relative to the vernal equinox (Sagarin 2001), the corresponding values were –1.1 ± 0.7 days per decade. The standard
error of the estimate for long-term temperature dates
(6.8 days) was close to the corresponding value for the bud
burst dates (6.8 for the mean values and 7.2 days for median
values in 1992–2005). The results for the dates corresponding
to the temperature sum of 44.8 dd were similar.
Even if the correlation between date of bud burst and the
date of the attainment of the temperature threshold value was
high, the trend in the latter was not necessarily an unbiased es-
Table 3. Summary of variance component analyses of yearly variation
in date of bud burst (days) at the Punkaharju birch stand. The dependent variable is day of year for bud burst.
Component
Tree
Year
Tree × year
Estimate
2.254
55.795
8.123
P
0.007
0.047
< 0.001
95% Confidence interval
Lower bound
Upper bound
1.091
20.839
6.764
4.654
149.385
9.755
timate for the trend in the former. The estimate for the difference in trends in annual mean values and dates of the
temperature sum threshold value was 1.6 (P = 0.46) and the
confidence interval for the difference was ± 4.6 days per decade. For the annual medians, the corresponding estimate of
the difference was –1.3 (P = 0.59) and its confidence interval
was ± 4.9 days per decade. According to these results, our time
series was too short to assess the sign or amount of possible
bias.
Discussion
Within-population variation and threshold temperatures
Spring phenology of white birch trees during a 9-year period
was strongly dependent on spring temperature. The timing of
bud burst is considered an important determinant of tree productivity and resistance to abiotic and biotic hazards (see
Rousi and Pusenius 2005). Environmental signals, such as
chilling units and light signals, may prevent premature commencement of growth (Hänninen 1990, 1995, Heide 1993a,
1993b, Wielgolaski 1999, Linkosalo and Lechowicz 2006).
However, in boreal conditions, the chilling requirements of
trees seem to be met as early as late autumn or early winter
(see references in Linkosalo et al. 2006). In addition, specific
light signals may play no role in growth commencement of
birch, because the accuracy of our results was not improved
when the counting of temperature sums was started at later
dates. During the study years the weather was highly variable
throughout the winter and early spring, thus our finding that,
under present climatic conditions in boreal forests, the accumulation of temperature sum alone determines the onset of
bud burst supports previous studies (Häkkinen et al. 1998,
TREE PHYSIOLOGY VOLUME 27, 2007
TRENDS IN DATE OF BUD BURST
1023
Figure 2. Day of year when temperature sum threshold values of
24.8 dd (i.e., estimated temperature for bud burst of the median
tree) were reached according to
the Finnish Meteorological Institute weather stations at
Punkaharju, Helsinki and
Jyväskylä from 1992 to 2005. The
weather data were obtained from
four measurements per day.
Punkaharju (hourly) data is based
on hourly measurements at the
Finnish Forest Research Institute
weather station. Punkaharju
(buds) indicates the median dates
for bud burst in the experimental
forest.
Häkkinen 1999a,1999b, Hannerz 1999, Linkosalo 2000a,
2000b, Linkosalo et al. 2000). It has been suggested that high
autumn temperature delays the next spring bud burst of boreal
birches, counterbalancing the effect of climatic warming
(Heide 2003, Karlsson et al. 2003). However, despite large
variation in autumn temperatures, we found no evidence of af-
ter-effects of temperatures of the previous year.
The magnitude of within-population variation in birch bud
burst timing varied from year-to-year. In some years, full bud
burst was reached within a few days, whereas in other years it
took up to four weeks. Bud burst usually began when the temperature sum reached 10–20 dd (dd5). Half of the trees had
Figure 3. Day of year
when temperature sum
threshold values of 24.8
dd (i.e., estimated temperature for bud burst of the
median tree) were
reached in Jyväskylä,
Helsinki and Punkaharju
from 1926 to 2005.
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com
1024
ROUSI AND HEINONEN
open buds when the temperature sum reached 30 dd, and all of
the trees had open buds at 40–50 dd (Figure 1). The onemonth variation in timing of bud burst within the stand in 1999
can be explained by the spring weather pattern: almost all trees
had open buds on April 29 (34.2 dd). After that, a 3-week period of below-freezing temperatures nullified the effect of the
previous temperature sum, and another 37 dd were needed before all the buds of the three remaining trees had opened (Figure 1; but see Linkosalo et al. 2006). In southern boreal forests,
the rise in the temperature sum can be more than 10 dd per day,
further indicating not only a good match with our bud burst
data, but also a general need for detailed daily observations.
Previous experiments have suggested a genetic base for variation in timing of bud burst in birch trees (Rousi and Pusenius
2005), and our ongoing experiments with micropropagated
plantlets of the study material will allow us to estimate the
magnitude of heritable variation. Tree-to-tree variation in
phenology and the frequent and abundant flowering of birch
(Koski and Tallqvist 1978) suggest the possibility of rapid
genetic adaptation to changing environmental conditions (cf.
Davis et al. 2005).
For temperature sum calculations, a wide range of threshold
temperatures has been suggested. For example, 5 °C has been
frequently used (e.g., Cannell et al. 1985, Hunter and Lechowicz 1992, Kellomäki et al. 1995), but 0 °C has been suggested
based on phytotron experiments with small seedlings (Heide
1993b), and +2 to –1 °C based on a field experiment with
young birch plantlets (Rousi and Pusenius 2005). Our calculations indicate that use of a threshold temperature of +5 °C,
measured 2 m above ground or at the branch level (where night
temperatures were higher and day temperatures lower than at
2 m), allowed reliable predictions of the timing of bud burst in
natural birch populations. However, tree-to-tree variation
seemed to be more important than the exact place where temperature was measured, because when the mean temperature
sum at budburst was counted for the earliest half of the population, the error for bud burst prediction was much smaller (Table 2). Even one extreme tree can make a difference. In 2000,
the deviation from the mean annual temperature sum for bud
opening in all trees was 9 days (it took 9 days for the temperature sum to increase from 41.1 to 44.8 dd, see Table 1). However, for the first 50% of the stand, which in 2000 happened to
be the same date as for 97% of the trees, there was no error
(April 23 was 24.6 dd and next day was 30.7 dd, see also Table 1). Caution may also be needed when results from laboratory experiments are applied to natural conditions: Prozherina
et al. (2003) transferred (open-pollinated) seedlings of these
same genotypes indoors in late March, and in artificial conditions the temperature sum for bud burst was considerably
larger. Not only some feature of the artificial environment but
also ontogeny may play a part, because the best threshold
temperature to predict bud burst appears to be lower for
younger birch saplings (cf. Rousi and Pusenius 2005).
sum threshold for bud burst occurred (–1.2 ± 0.7 days per decade) was linear, and similar linear trends were found for the
dd to 50 and 100% bud burst (24.8 and 44.8 dd, respectively).
These findings are in good agreement with the results of
Tuomenvirta (2004), who found an almost linear increase in
mean monthly spring temperatures in Finland during the last
century.
The trend estimate based on daily bud burst observations in
Punkaharju was not statistically significant, indicating that
14 years (1992–2005) is too short a period for estimating
long-term changes in timing of bud burst. The linear trend parameter for this relatively short time period is, however, useful
for comparing different sources of information. For example,
in 1982–2001, the trend parameter for the date of the threshold value of the temperature sum was –1.8, which is one third
of the –5.4 value estimated by Stöckli and Vidale (2004) from
satellite observations (NDVI data). Correspondingly, the trend
parameter for the years 1981–1991 was –4.1 compared with a
value of –7.3 estimated by Myneni et al. (1997). More consistent results were expected, because birch is the major deciduous component of Euro-Asian boreal forests. One reason for
these inconsistencies might be the sensitivity of the short time
series to the method of assessing annual bud burst dates.
The standard error of the estimate for the long-term trend of
the temperature dates (6.8 days) is close to the corresponding
value for the bud burst dates (6.8 days for the mean dates and
7.2 for the median dates). If we assume a linear trend and
uncorrelated errors with a standard deviation of 6.8, we can
compute the effect of the length of the time series on the precision of the trend estimate. For example, if the estimate is based
on data for 30 consecutive years, the standard deviation of the
estimate is 1.4, and 32% of the estimates deviate more than
± 1.4 days per decade from the true value.
To conclude, not only large interannual variation, but also
annual within-population variation in bud burst date can be accurately predicted based on the accumulation of spring temperatures. According to our results, the long-term trends in
spring temperatures and phenology in Finland are small compared with the large interannual differences. For this reason, as
well as because of the large within-population variation, trend
estimates for bud burst date based on measurements of only a
few decades are unreliable.
Acknowledgments
We are indebted to Hanni Sikanen who climbed the trees every spring
day for 9 years to make observations and who also assisted at every
stage of the research. Saku Silvennoinen was heavily involved in daily
bud burst observations. Pentti Manninen was responsible for Punkaharju weather data collection. Heikki Hänninen, Tapio Linkosalo and
Seppo Ruotsalainen provided useful comments on earlier versions of
this paper. Joanna Goga checked the English.
References
Bud burst timing trends
The long-term yearly change in the date when the temperature
Anon. 2005. Finnish statistical yearbook of forestry. Finnish Forest
Research Institute, Helsinki, 424 p.
TREE PHYSIOLOGY VOLUME 27, 2007
TRENDS IN DATE OF BUD BURST
Badeck, F.-W., A. Bondeau, K. Böttcher, D. Doktor, W. Lucht,
J. Schaber and S. Sitch. 2004. Responses of spring phenology to
climate change. New Phytol. 162:295–309.
Billington, H.L. and J. Pelham. 1991. Genetic variation in the date of
bud burst in Scottish birch populations: implications for climate
change. Funct. Ecol. 5:403–409.
Cannell, M.G.R., M.B. Murray and L.J. Sheppard. 1985. Frost avoidance by selection for late budburst in Picea sitchensis. J. Appl.
Ecol. 22:931–941.
Chmielewski, F.-M. and T. Rötzer. 2001. Response of tree phenology
to climate change across Europe. Agric. For. Meteorol. 108:
101–112.
Davis, M.B., R.G. Shaw and J.R. Etterson. 2005. Evolutionary responses to changing climate. Ecology 86:1704–1714.
Fitter, A.H. and R.S.R. Fitter. 2002. Rapid changes in flowering time
in British plants. Science 296:1689–1691.
Hannerz, M. 1999. Evaluation of temperature models for predicting
bud burst in Norway spruce. Can. J. For. Res. 29:1–11.
Heide, O.M. 1993a. Dormancy release in beech buds (Fagus sylvatica) requires both chilling and long days. Physiol. Plant. 88:
187–191.
Heide, O.M. 1993b. Daylength and thermal time responses of budburst during dormancy release in some northern deciduous trees.
Physiol. Plant. 88:531–540.
Heide, O.M. 2003. High temperature delays spring bud burst in boreal
trees, counterbalancing the effect of climatic warming. Tree
Physiol. 23:931–936.
Hunter, A.F. and M.J. Lechowicz. 1992. Predicting the timing of budburst in temperate trees. J. Appl. Ecol. 29:597–604.
Häkkinen, R. 1999a. Analysis of bud-development theories based on
long-term phenological and air temperature time series: application
to Betula sp. leaves. Finnish Forest Research Institute, Research
Paper 754, Helsinki, 59 p.
Häkkinen, R. 1999b. Statistical evaluation of bud development theories: application to bud burst of Betula pendula leaves. Tree
Physiol. 19:613–618.
Häkkinen, R., T. Linkosalo and P. Hari. 1998. Effects of dormancy
and environmental factors on timing of bud burst in Betula pendula. Tree Physiol. 18:707–712.
Hänninen, H. 1990. Modeling bud dormancy release in trees from
cool and temperate regions. Acta For. Fenn. 213:1–47.
Hänninen, H. 1995. Effects of climatic-change on trees from cool and
temperate regions—an ecophysiological approach to modeling of
bud burst phenology. Can. J. Bot. 73:183–199.
Järvinen, P., J. Lemmetyinen, O. Savolainen and T. Sopanen. 2003.
DNA sequence variation in BpMADS2 gene in two populations of
Betula pendula. Mol. Ecol. 12:369–384.
Karlsson, P. S., H. Bylund, S. Neuvonen, S. Heino and M. Tjus. 2003.
Climatic control of budburst in the mountain birch at two areas in
northern Fennoscandia and possible responses to global change.
Ecography 27:617–625.
1025
Kellomäki, S., H. Hänninen and M. Kolström. 1995. Computations on
frost damage to Scots pine under climatic warming in boreal conditions. Ecol. Appl. 5:42–52.
Koski, V. and R. Tallqvist. 1978. Results of long-time measurements
of the quantity of flowering and seed crop of forest trees. Folia For.
364:1–60.
Linkosalo, T. 2000a. Analyses of the spring phenology of boreal trees
and its response to climate change. University of Helsinki Department of Forest Ecology, Publication 22, Helsinki, 55 p.
Linkosalo, T. 2000b. Mutual regularity of spring phenology of some
boreal tree species: predicting with other species and phenological
models. Can. J. For. Res. 30:667–673.
Linkosalo, T., T.R. Carter, R. Häkkinen and P. Hari. 2000. Predicting
spring phenology and frost damage risk of Betula spp. under climatic warming: a comparison of two models. Tree Physiol. 20:
1175–1182.
Linkosalo, T., R. Häkkinen and H. Hänninen. 2006. Models of the
spring phenology of boreal and temperate trees: is there something
missing? Tree Physiol. 26:1165–1172.
Linkosalo, T. and M.J. Lechowicz. 2006. Twilight far-red treatment
advances leaf bud burst of silver birch (Betula pendula). Tree
Physiol. 26:1249–1256.
Lorenz, E. 1993. The essence of chaos. UCL Press, London, 227 p.
Menzel, A. 2000. Trends in phenological phases in Europe between
1951 and 1996. Int. J. Biometeorol. 44:76–81.
Myneni, R.B., C.D. Keeling, C.J. Tucker, G. Asrar and R.R. Nemani.
1997. Increased plant growth in the northern high latitudes from
1981 to 1991. Nature 386:698–702.
Prozherina, N., V. Freiwald, M. Rousi and E. Oksanen. 2003. Interactive effect of springtime frost and elevated ozone on early growth,
foliar injuries and leaf structure of birch (Betula pendula). New
Phytol. 159:623–636.
Rousi, M. and J. Pusenius. 2005. Variation in phenology and growth
of silver birch (Betula pendula) clones. Tree Physiol. 25:201–210.
Rusanen, M., P. Vakkari and A. Blom. 2003. Genetic structure of Acer
platanoides and Betula pendula in northern Europe. Can. J. For.
Res. 33:1110–1115.
Sagarin, R. 2001. Phenology: false estimates of the advance of spring.
Nature 414:600.
Schaber, J. and F.-W. Badeck. 2003. Physiology-based models for
forest tree species in Germany. Int. J. Biometeorol. 47:193–201.
Schwartz, M.D. and B.E. Reiter. 2000. Changes in North American
spring. Int. J. Climatol. 20:929–932.
Stöckli, R. and P.L. Vidale. 2004. European plant phenology and climate as seen in a 20-year AVHRR land-surface parameter dataset.
Int. J. Remote Sens. 25:3303–3330.
Tuomenvirta, H. 2004. Reliable estimation of climatic variations in
Finland. Ph.D. Diss. University of Helsinki, Finland, 158 p.
Wielgolaski, F.E. 1999. Starting dates and basic temperatures in phenological observations of plants. Int. J. Biometeorol. 42: 158–168.
TREE PHYSIOLOGY ONLINE at http://heronpublishing.com