Content uploaded by Eugenija Baksiene
Author content
All content in this area was uploaded by Eugenija Baksiene on Mar 28, 2014
Content may be subject to copyright.
CLIMATE RESEARCH
Clim Res
Vol. 39: 209–219, 2009
doi: 10.3354/cr00813 Published September 10
1. INTRODUCTION
Phenological information is important in monitoring
all aspects of ecosystems in agricultural, medicine,
forestry, tourism and wildlife management (Beaubien &
Freeland 2000). As phenological rhythms are closely re-
lated to climate, phenological phases can play the role
of climate change indicators. Phenology is probably the
simplest and most cost-effective means of observing the
effects of changes in temperature; consequently, phe-
nology has become an important tool in the research of
global climatic change.
The origin of phenology can be traced back to ancient
society, but the oldest existing phenological record is
from Japan: the flowering of cherry has been observed
and recorded since 705 AD. In Europe, the oldest time
series, dating back to 1370, consists of the ripening of
grapes in France (Chuine et al. 2004). In the Baltic States,
the longest phenological data series was reconstructed
for rye harvest in Estonia and dates back to the 17th cen-
tury (Ahas 1999). The first phenological observations in
Lithuania were made at the end of the 18th century, but
the first phenological network was developed in 1959. In
Latvia, the first observations date to 1822, whereas the
phenological network has operated since 1926.
Numerous studies from around the world have shown
significant changes in the onset of phenological phases
during the 20th century (Ahas 1999, Ahas et al. 2002,
Bradley et al. 1999, Harrington et al. 1999, Menzel
2000, 2003, Bogaert et al. 2002, Root et al. 2003, Linder-
holm 2006, Zheng et al. 2006). For example, it was
found that in north and northeastern China and the
lower reaches of the Huaihe River, phenophases ad-
vanced 1.1 to 4.3 d per decade for early spring and 1.4
to 5.4 d per decade for late spring, but in the eastern
part of southwest China they were delayed by 2.9 to
6.9 d per decade for early spring and 2.4 to 6.2 d per
decade for late spring (Zheng et al. 2006). A study in
Wisconsin, USA, for the period 1976– 1998 showed that
several phenological events have been increasingly
early. The average regression for the 55 phenophases
studied was –0.12 d per year (Bradley et al. 1999).
© Inter-Research 2009 · www.int-res.com*Email: gunta.kalvane@lu.lv
Influence of climate change on phenological
phases in Latvia and Lithuania
Gunta Kalva
¯
ne1,*, Danuta Romanovskaja2, Agrita Briede1, Eugenija Bak$iene˙2
1University of Latvia, Faculty of Geography and Earth Sciences, Raina Blvd. 19, Rı¯ga 1059, Latvia
2Voke Branch of Lithuanian Institute of Agriculture, Zalioji a. 2, Vilnius 02232, Lithuania
ABSTRACT: To investigate the impact of recent climatic changes on plant development, this study
used phenological data of the volunteer networks in Latvia and Lithuania from the 1971– 2000 period.
The phenological calendar method was applied. Phenological seasons were described using data on
6 phenological phases at 10 stations. The growing season was described using birch Betula pendula
as an example. Correlation analysis, linear regression and non-parametric Mann-Kendall trend tests
were applied to establish the relationship between phenological phases and meteorological factors
(temperature, precipitation) and the North Atlantic Oscillation (NAO). The results indicate a statisti-
cally significant trend toward earlier onset of spring and summer phases. The study found that, unlike
the majority of trends observed in Europe, on average the onset of phenological autumn in Latvia and
Lithuania also started earlier. The observed trends in spring correlated well with temperatures in the
preceding months and with the NAO. A strong correlation (in 12 cases out of 20) was found between
spring phenophases and precipitation in February.
KEY WORDS: Phenology · Climate change · Phenological trends · Time series · Growing season
Resale or republication not permitted without written consent of the publisher
Contribution to CR Special 19 ‘European plant phenology’
OPENPEN
ACCESSCCESS
Clim Res 39: 209– 219, 2009
In Europe, based on International Phenological Gar-
dens data (1959– 1996), spring events such as leaf un-
folding have advanced an average of 6.3 d (–0.21 d per
year), while autumn events such as leaf colouring have
been delayed by 4.5 d (0.15 d per year). Combining
these, the average annual growing season has extended
by 10.8 d since the early 1960s (Menzel 2000). According
to Ahas et al. (2002), in the period 1951–1998, spring
phases have advanced by 4 wk in Western and Central
Europe and by up to 2 wk in Eastern Europe.
In Estonia, the onset of spring has advanced by 8 d
over the last 80 yr, with a faster change occurring over
the last 40 yr (Ahas 1999). In Lithuania, phenological
spring started 8 to 16 d earlier at the end of the 20th
century than in the late 70s (Romanovskaja & Bak$iene˙
2007). Over the last 50 yr, the beginning of spring and
summer phases in Latvia has shifted earlier by 4 d
(Grisule & Briede 2007).
The aim of the present study is to characterize the
variability of the phenological time series for the 1971–
2000 period and to assess the impact of climate change
on phenology in Latvia and Lithuania.
2. MATERIALS AND METHODS
2.1. Phenological database
The present study is based on volunteer-collected
phenological data in Latvia and Lithuania, at sites
located between 54° 40’ and 57° 23’ N and 21° 01’ and
27° 31’ E (Fig. 1).
With respect to plant development under different cli-
mate conditions, the territory of Lithuania is divided into
3 phenoclimatic regions. Phenological stations (Fig. 1)
were chosen to represent all phenoclimatic regions of
the country: Traku Voke (Aukstaiciai region), Silute
(Zemaiciai Upland) and Akademija, Papile and Ketur-
valakiai (West Zemaiciai Plain and Cen-
tral Lithuanian Plain; subdivided into 3
subregions).
The number of phenological stations in
Latvia varied from year to year and there
were gaps in the data series as well.
Based on the data quality and quantity,
we selected data from 5 phenological sta-
tions for the present study; at 2 of these
sites (Atasiene-Barkava and Liepaja-
Nica) the data series from nearby sta-
tions were combined. Unlike Lithuania,
Latvia’s territory is not divided into phe-
noclimatic regions, but into physiogeo-
graphic regions, which are distin-
guished by different geomorphological
and climate conditions.
Seasonal changes were studied using onset dates of
phenological seasons. According to V. Köppen classifi-
cation (McKnight & Hess 2004), Latvia and Lithuania
are located in the humid continental climate zone;
therefore, plant indicators denoting the beginning of
phenological phases common to both countries were
chosen. The choice of plants as season indicators
(Table 1) was made based on the division of indicator
plants commonly used in Europe and choosing those
most suitable for the climate zone of Lithuania and
Latvia (Kuliene˙ & Tomkus 1990).
210
Fig. 1. Location of phenological and meteorological stations in
Latvia and Lithuania used in the present study
Species Phase BBCH code
Phenological season
Spring Alnus incana Beginning of flowering 61
Corylus avellana
Summer Philadelphus coronarius Beginning of flowering 61
Syringa vulgaris
Autumn Betula pendula Leaf colouring 92
Acer platanoides
Growing season
Start Betula pendula Beginning of leafing 11
End Betula pendula Leaf colouring 92
Table 1. Indicators for the phenological and growing seasons. Bundesanstalt,
Bundessortenamt and Chemical Industry (BBCH) scale classifies the entire
development cycle of plants into 10 principal phases (0– 9) and 10 secondary
stages (0– 9) as a 2-digit number. For details see Meier (1997)
Kalva
¯
ne et al.: Climate change and phenological phases
The phenological dates refer to the day of the year
and phenological phases are in Bundesanstalt, Bun-
dessortenamt and Chemical Industry (BBCH) codes.
The extended BBCH-scale is a system for a uniform
coding of phenologically similar growth stages of all
mono- and dicotyledonous plant species. The entire
developmental cycle of the plants is subdivided into 10
clearly recognizable and distinguishable longer-last-
ing developmental phases (Meier 1997).
2.2. Meteorological database
Climate data (monthly temperature and precipita-
tion) for the years 1971–2000 were obtained from the
Latvian Environment, Geology and Meteorology
Agency and from the Lithuanian Hydrometeorological
Service.
Meteorological stations were selected at locations
close to the phenological stations with the exception of
2 sites, Atasiene-Barkava and Papile, located within
50 km northeast and south, respectively, from the clos-
est meteorological station (Fig. 1). We consider the
microclimate differences between these sites and the
meteorological stations to be small.
Monthly indices for the NAO were taken from the
NOAA Climate Prediction Centre web page (www.
cpc.noaa.gov/index.php).
2.3. Methods
Data were manually checked and outliers were
excluded from the data series if the difference in the
beginning of phenophases among the closest pheno-
logical stations was larger than 1 mo. Data quality was
verified by using the 3-sigma method: 3-sigma is a sta-
tistical boundary representing plus or minus 3 stan-
dard deviations from a measure of central tendency
(i.e. mean) for a group of values.
For the study of time series we used the following
analytical tools: regression analysis, correlation analy-
sis and non-parametric Mann-Kendall (M-K) test
analysis. Phenological data were statistically described
by calculating the mean, standard error and coefficient
of variation (Dospehov 1973).
The main statistical parameters were drawn from the
regression analysis. The mean temporal change (days
per year) of the studied variables was evaluated using
the regression slopes. The purpose of using the M-K test
(Libiseller & Grimvall 2002) for trend detection is to
determine the sign of all pairwise differences between
the consecutive elements of time series, while each of
them is compared with all previous values in the time
series. The M-K test is non-parametric and does not
require the data be distributed normally. It is not sensi-
tive to abrupt breaks due to non-homogeneous time se-
ries, which are characteristic of phenological records
(Libiseller & Grimvall 2002). The test statistic is near zero
when the number of positive and negative differences is
more or less equal. There is an increasing trend if the test
statistic is positive, i.e. the number of positive differences
is significantly higher than the number of negative dif-
ferences, and vice versa. The result was considered to be
statistically significant if the value of the normalised M-K
test statistic was >1.65 or <–1.65, with p < 0.05.
Correlation analysis was applied to investigate the
relationship between phenological time-series and
temperature, precipitation and NAO. Correlation coef-
ficients >0.36 or <–0.36 were considered as statistically
significant for the 30-yr period with p < 0.05.
3. RESULTS
In general, the character of changes in air tempera-
ture and precipitation for the 1971–2000 period was
similar for Latvia and Lithuania. Trend analysis of
spring (March–May) temperature changes according
to the M-K test demonstrated statistically significant
increases for all stations (M-K > 1.92, p < 0.05). At the
same time, examining monthly temperature data
showed significant increases only in April (M-K > 1.97,
p < 0.05). The mean temperature of autumn 1971– 2000
changed neither in Latvia nor Lithuania. Precipitation
in Latvia and Lithuania had a high spatial variability
influenced by peculiarities of local landscape, topogra-
phy and distance from the Baltic Sea. The highest
totals of monthly precipitation within the studied
period were measured in August at the stations lying
near the Baltic Sea coast —291 mm in Silute (1978) and
281 mm in Liepaja (1972) — while in August 1996 and
1997 at most of the stations located in the remote in-
land areas, zero or very low precipitation (<10 mm)
was recorded. Results show a statistically significant
increase in the amount of precipitation in Latvia and
Lithuania only in February (9 of 10 stations) and Janu-
ary (7 of 10 stations).
3.1. Patterns of the phenological seasons
According to phenological observation data of the
years 1971–2000, the spring season indicators Euro-
pean hazel Corylus avellana and grey alder Alnus in-
cana flowered in Lithuania and Latvia from the 3rd
decad (10 d period) of March through to the 1st decad
of April (Fig. 2).
In both Baltic countries, European hazel started
flowering earlier than grey alder. In Lithuania, which is
211
Clim Res 39: 209– 219, 2009
located just south of Latvia, European hazel started to
flower 6 d earlier on average, and this phenological
phase spread across the country within 8 d (20 to
27 March) (Fig. 2). The same process in Latvia took
twice as long as in Lithuania (19 d, from 21 to 23 March
in coastal areas until 8 April in Dagda, the easternmost
station in Latvia). The flowering of grey alder, another
spring indicator, started in both countries only 1 to 2 d
after hazel and spread over the entire territory of both
Lithuania and Latvia within 2 wk (21 to 24 March in
coastal areas to 5 April in eastern Latvia). It was found
that flowering of these spring season indicators started
earlier on average in localities closer to the Baltic Sea:
4 to 6 d earlier at Silute (Lithuania) and 5 to 8 d earlier
at Liepaja-Nica, Pope (Latvia). Annual variations in
the dates of the beginning of hazel and alder flower-
ing were very high in both Baltic countries (CV > 20 %
in some localities). The earliest flowering of hazel
and alder began in 1989– 1990 (January – February)
(Fig. 3). The latest flowering of these plants was re-
corded in the second half of April. Late flowering was
more frequent in the middle of the 1990s. Since the dif-
ferences between the earliest and average dates are
twice as great as those between the latest and average
dates, negative anomalies are more common. From
comparisons of spring, summer and autumn phenolog-
ical phases for the selected species, it is evident that
spring phases have larger variations. According to our
212
Corylus avellana y = –0.741x + 1556.1
R
2 = 0.1472
Alnus incana y = – 0.576x + 1230.4
R
2 = 0.1004
20
40
60
80
100
120
140
1971 1975 1979 1983 1987 19 91 19 95 19 99
Year
DOY
Stations in Lithuania
Stations in Latvia
Avg Corylus avellana
Avg Alnus incana
Fig. 3. Corylus avellana and Alnus incana. Trends in the beginning of flowering of European hazel C. avellana and grey alder
A. incana (average data from 10 stations) 1971– 2000 in Latvia and Lithuania. DOY: day of year
01 Jan 31 Jan 02 Mar 01 Apr 01 May 31 May 30 Jun 30 Jul 29 Aug 28 Sep 28 Oct
Latvia Lithuania
Corylus avellana
Alnus incana
Syringa vulgaris
Betula pendula
Philadelphus coronaruis
Acer platanoides
Fig. 2. Phenological calendar for the period 1971–2000 for Latvia and Lithuania. y-axis represents the sequence of phases.
Phenophases are the beginning of flowering for spring and the beginning of leaf colouring for autumn
Kalva
¯
ne et al.: Climate change and phenological phases
data, the beginning of phenological spring spread from
SW to NE by, on average, 4.5 d per 100 km.
The beginning of flowering of the summer indicators
sweet mock orange/false jasmine Philadelphus coro-
narius and common lilac Syringa vulgaris did not coin-
cide. In both Baltic countries, lilac started flowering
3 wk earlier on average than sweet mock orange.
Flowering of the spring season indicators (hazel and
alder) and flowering of the summer season indicators
(lilac and mock orange) started earlier in Lithuania
than in Latvia (Fig. 2). Common lilac started to flower
in Lithuania, on average, from 17 to 22 May but in
Latvia from 24 to 30 May when this phenophase had
already spread over the whole territory of Lithuania.
The flowering of common lilac and sweet mock orange
began earlier in the southern and central parts of the
countries. In both countries, the flowering of sweet
mock orange and common lilac started 4 to 6 d later in
localities closer to the Baltic Sea. The flowering of com-
mon lilac spread over the whole territory of both Latvia
and Lithuania within a 5 d period. The beginning of
summer spread from south to north with a phenologi-
cal gradient of 3.7 d per 100 km.
Annual variation in the dates of the beginning of com-
mon lilac flowering was low (<10 %): 4.7 to 5.9 % in
Lithuania and 5.6 to 7.0% in Latvia. Mean deviations of
the earliest and latest dates from the average were al-
most equal but they were smaller than for species that
started to flower early in spring. The earliest and latest
dates of the beginning of common lilac flowering in
Latvia and Lithuania occurred in different years; the ear-
liest flowering at all phenological stations in Latvia was
recorded in 1990, while in Lithuania it occurred 1 decade
later in 2000. All of the latest dates of the beginning of
common lilac flowering in Lithuania were in 1980, while
in Latvia the latest dates were recorded in 1980 only in 2
locations (Atasiene-Barkava and Dagda); in other loca-
tions the latest dates were recorded in 1987 (Fig. 4).
Autumn, as indicated by the colouring of the leaves
of Norway maple Acer platanoides and common birch
Betula pendula, started in September; the dates of the
recorded phenological phases were earlier in Lithuania.
In Latvia, Norway maple and common birch leaves
started colouring during the second half of September,
but in Lithuania this occurred in the beginning of Sep-
tember, except at Papile˙. Papile˙ is located in the northern
part of Lithuania, and the colouring of Norway maple
leaves started on 24 August and 20 August for birch
leaves. The average latest colouring of Norway maple
was recorded at Dobele on 26 September, whilst the av-
erage latest common birch leaf colouring was at Ketur-
valakiai on 5 October. Thus the duration of phenological
phases of autumn indicators over the territories of both
Baltic countries was about 4 wk, i.e. considerably longer
than the periods of spring and summer events. The
largest deviations from the average (earliest and latest
dates) were recorded in different years (Fig. 5).
3.2. Trends in phenological seasons
The data for 1971–2000 show a tendency of annual
advancement of the phenological phases for summer,
autumn and spring indicators (Table 2). However, the
213
1971 1975 1979 1983 1987 19 91 19 95 19 99
Year
DOY
Stations in Lithuania
Stations in Latvia
Avg Philadelphus coronarius
Avg Syringa vulgaris Philadelphus coronarius y = –0.2587x + 677.88
R
2 = 0.1253
Syringa vulgaris y = –0.3723x + 881.39
R
2 = 0.1807
100
120
140
160
180
200
Fig. 4. Philadelphus coronarius and Syringa vulgaris. Trends in the beginning of flowering of sweet mock orange P. coronarius
and common lilac S. vulgaris (average data from 10 stations) for the period 1971– 2000 in Latvia and Lithuania. DOY: day of year
Clim Res 39: 209– 219, 2009214
Stations in Lithuania
Stations in Latvia
Avg Acer platanoides
Avg Betula pendula
1971 1975 1979 1983 1987 1991 1995 1999
Year
DOY
Acer platanoides y = –0.1036x + 465.51
R
2 = 0.0345
Betula pendula y = –0.2579x + 771.21
R
2 = 0.2022
200
220
240
260
280
300
Fig. 5. Acer platanoides and Betula pendula. Trends in the beginning of leaf colouring of Norway maple Acer platanoides and
common birch Betula pendula (average data from 10 stations) for the period 1971–2000 in Latvia and Lithuania. DOY: day of year
Season Station Year (N) Onset of phenological CV (%) Slope Mann-Kendall test
and species phases (DOY) M-K p
Spring
Corylus avellana
Latvia Atasiene Barkava 23 91 ± 4 21.30 –1.11 –1.14 0.13
Dagda 15 98 ± 4.8 18.80 –0.63 0.10 0.46
Dobele 30 87 ± 3.4 21.30 –0.76 –1.58 0.06
Liepaja Nica 23 80 ± 4.5 27.10 –0.95 –1.51 0.07
Pope 27 83 ± 4 24.90 –1.06 –1.69 0.05
Lithuania Traku Voke 27 82 ± 3.2 20.60 –0.91 –1.82 0.04
Akademija 28 83 ± 3.6 22.80 – 0.68 –1.68 0.05
Silute 29 78 ± 4 27.40 –0.69 –1.22 0.11
Papile 30 86 ± 3.4 21.40 –0.12 0.63 0.27
Keturvalakiai 28 80 ± 3.1 20.60 –0.48 – 0.36 0.36
Alnus incana
Latvia Atasiene Barkava 16 89 ± 5.6 25.10 –1.34 –1.76 0.04
Dagda 22 95 ± 3.2 15.90 –0.52 0.14 0.44
Dobele 26 87 ± 3.4 20.20 –0.49 – 0.55 0.29
Liepaja Nica 23 82 ± 4.3 25.10 –1.47 –2.70 0.00
Pope 26 83 ± 4.1 25.50 –1.00 –1.57 0.06
Lithuania Traku Voke 27 82 ± 3.1 19.80 –0.85 –1.65 0.05
Akademija 26 86 ± 3.4 20.10 – 0.69 –1.63 0.05
Silute 26 80 ± 4.1 25.90 –0.52 – 0.82 0.21
Papile 29 90 ± 3.3 20.00 0.17 1.32 0.09
Keturvalakiai 15 92 ± 2.9 12.30 –0.10 0.20 0.42
Summer
Syringa vulgaris
Latvia Atasiene Barkava 27 145 ± 1.6 5.90 – 0.27 –1.61 0.05
Dagda 22 145 ± 1.7 5.60 – 0.46 –1.87 0.03
Dobele 27 144 ± 1.6 5.70 – 0.34 –1.55 0.06
Liepaja Nica 22 145 ± 2.2 7.00 –0.44 – 1.73 0.04
Pope 26 150 ± 1.7 5.90 – 0.17 –1.20 0.12
Table 2. Statistical parameters of the phenological seasons and phases in Latvia and Lithuania (1971–2000). Data for ‘Onset of
phenological phases’ are mean ± SE. DOY: day of year. Significant values (M-K >1.65 or <–1.65, with p < 0.05) are in bold
Kalva
¯
ne et al.: Climate change and phenological phases
advancement of dates of the beginning of phenologi-
cal phases is more marked in plants starting to flower
in the phenological summer; in 11 of 20 cases the
trend is statistically significant (M-K <–1.65, p < 0.05,)
and the mean slope for phenological summer is –0.30
(–0.11 to – 0.52). The beginning of flowering of sum-
mer indicators advanced by 3 to 15 d during the study
period.
During the last 30 yr, the flowering of European
hazel in Lithuania advanced by 3.5 to 27.4 d (–0.1 to
–0.9 d per year) and by 18.9 to 33.3 d (–0.6 to – 1.1 d per
year) in Latvia; alder has advanced by 3 to 25.5 d in
Lithuania and 14.6 to 44 d in Latvia.
The present study also confirms that Norway maple
and birch leaf colouring in the latest years tended to
have an earlier timing; in 8 of 20 cases the trend was
statistically significant and the mean slope for pheno-
logical autumn was –0.15 (– 0.8 to 0.4) (Table 2).
During the last 10 yr of the study period, the onset of
phenological autumn occurred earlier than average in
50% of cases. From 1990 onwards, 9 out of 10 years
saw spring phases recorded earlier than the 30-yr
average. The earliest summer onset values were
recorded in 6 of the last 10 yr of the study period.
During all periods there was little influence of the
distance from the sea on the trends in onset of pheno-
logical phases in summer, spring or autumn.
Table 3 shows recorded changes in phenological
phases over 15 yr periods: 1971–1985 and 1986–2000.
According to mean values, all species’ phases ob-
served during 1986–2000 had an earlier onset time
than in the period 1971–1985.
215
Season Station Year (N) Onset of phenological CV (%) Slope Mann-Kendall test
and species phases (DOY) M-K p
Summer
Syringa vulgaris
Lithuania Traku Voke 27 137 ± 1.5 5.70 – 0.40 –1.91 0.03
Akademija 28 138 ± 1.5 5.90 – 0.36 – 1.96 0.03
Silute 30 138 ± 1.4 5.40 – 0.40 –2.34 0.01
Papile 30 142 ± 1.2 4.70 – 0.16 – 0.65 0.26
Keturvalakiai 30 141 ± 1.4 5.50 –0.38 – 2.01 0.02
Philadelphus coronarius
Latvia Atasiene Barkava 27 169 ± 1.7 5.20 – 0.45 –2.79 0.00
Dagda 17 164 ± 2 5.10 – 0.52 – 1.08 0.14
Dobele 19 165 ± 2.1 5.50 – 0.20 – 0.77 0.22
Liepaja Nica 23 173 ± 2.1 5.90 0.39 1.61 0.05
Pope 23 176 ± 1.4 3.80 – 0.25 –1.25 0.11
Lithuania Traku Voke 30 159 ± 1.6 5.40 – 0.45 –2.40 0.01
Akademija 30 159 ± 1.6 5.40 – 0.41 – 2.13 0.02
Silute 30 159 ± 1.7 5.80 – 0.50 –2.85 0.00
Papile 30 164 ± 1 3.50 – 0.12 – 0.86 0.20
Keturvalakiai 30 160 ± 1.4 4.70 –0.11 – 0.32 0.37
Autumn
Acer platanoides
Latvia Atasiene Barkava 27 262 ± 1.8 3.50 0.31 1.42 0.08
Dagda 19 267 ± 1.7 2.80 – 0.63 –2.04 0.02
Dobele 12 269 ± 3.3 4.30 0.21 – 0.21 0.42
Liepaja Nica 23 262 ± 1.1 2.00 0.04 0.43 0.34
Pope 24 267 ± 1.3 2.40 – 0.03 – 0.55 0.29
Lithuania Traku Voke 24 256 ± 1.4 2.70 – 0.39 –2.07 0.02
Akademija 25 255 ± 2.8 5.50 – 0.57 – 1.83 0.03
Silute 18 262 ± 1.6 2.60 0.34 1.10 0.14
Papile 30 236 ± 2.5 5.70 – 0.56 –2.11 0.02
Keturvalakiai 27 273 ± 2.4 4.70 0.18 1.24 0.11
Betula pendula
Latvia Atasiene Barkava 27 265 ± 1.8 3.50 0.40 1.54 0.06
Dagda 19 276 ± 1.7 2.80 – 0.27 –2.22 0.01
Dobele 12 278 ± 3.3 4.30 0.05 0.27 0.39
Liepaja Nica 23 254 ± 1.1 2.00 –0.24 – 1.39 0.08
Pope 24 260 ± 1.3 2.40 0.33 1.67 0.05
Lithuania Traku Voke 22 253 ± 1.5 2.70 – 0.47 –1.67 0.05
Akademija 24 250 ± 3 5.90 –0.53 – 2.22 0.01
Silute 18 247 ± 2.5 4.20 – 0.80 –2.31 0.01
Papile 30 232 ± 2.6 6.20 – 0.38 – 0.75 0.23
Keturvalakiai 28 278 ± 1.6 3.00 0.04 –0.04 0.48
Table 2. (continued)
Clim Res 39: 209– 219, 2009
3.3. Growing season
Within the study period, the average duration of the
growing season, defined as the time between leaf
onset and leaf colouring for birch Betula pendula, was
138 d. The shortest average growing season on record
was 112 d in Papile (Lithuania), and the longest 157 d
in Dagda (Latvia) and Keturvalakiai (Lithuania).
The average growing season for birch started on
1 May in the southern part of Lithuania and on 7 May
in Pope, the northernmost phenological station. The
end dates of the growing season varied from 20 August
(Papile) to 5 October (Dagda, Keturvalakiai). The earli-
est dates of leafing (the beginning of the growing sea-
son) were recorded in 1990, when the growing season
for birch started on average 18 d earlier (Fig. 6). The
latest dates were recorded in 1980 (11 d later then
average).
The shortest growing season at an individual station,
Papile (Lithuania), was only 84 d in 1981 and the
longest was 187 d in Keturvalakiai (Lithuania) in 1989.
The shortest growing season (123 d) on average (data
from 10 stations) was observed in 1979 and is in agree-
ment with phenological observations for all European
countries. The longest growing season (153 d) on aver-
age was observed in 1998.
As seen in Fig. 6, the trend in birch leafing — the
beginning of the season — was statistically significant
and negative (r = –0.27 to – 0.55), which is confirmed by
the results of the Mann-Kendall test (M-K ≤–1.65). The
trend for the end date of the growing season was neg-
ative, yet it was statistically significant only for 4 out of
10 stations.
Within the study period, the beginning of the grow-
ing season advanced by 6.4 to 15.4 d (on average –4.0 d
per decade). For the most part, early dates have pre-
vailed since the end of the 1980s. It is also noteworthy
that the trend of the end date of the growing season
was negative for 6 stations (phenological autumn has
been observed 8.2 to 23.9 d earlier); leaf colouring,
which marked the end of the period, occurred on aver-
age 2 d earlier over the 31-yr period. Examining the
long-term time series of the beginning and end dates of
the growing season showed that the growing season
shifted to earlier dates in both spring and autumn.
Overall, the length of the season was extended by an
average of 7 d, due to an earlier start.
3.4. Influencing factors
Three climate parameters —temperature, precipita-
tion and the NAODec-Feb index — were chosen for the
study of influencing factors on phenological phases.
The strongest correlation was observed between the
average air temperature of the preceding month and
the onset of the phenological phase (Table 4). The cor-
relation coefficient between temperature and the onset
of the spring and summer phases varied from r = –0.55
to –0.82. The value of the associated correlation coeffi-
cient of the autumn phase was positive (with individual
exceptions) and varied from r = –0.11 to 0.30 with the
average temperatures of the previous 3 mo.
Concerning rainfall data, they showed a negative
correlation with the spring phases (raverage = – 0.26).
February precipitation correlated well (r = ~–0.50) with
the flowering of grey alder and hazel in March.
216
Phase (code) and plant 1971–1985 1986–2000
Beginning of flowering (BBCH61)
Alnus incana 01.04 22.03
Corylus avellana 01.04 19.03
Syringa vulgaris 25.05 19.05
Philadelphus coronarius 15.06 11.06
Leaf colouring (BBCH92)
Betula pendula 18.09 14.09
Acer platanoides 18.09 16.09
Table 3. Phenological phase means over 15-yr periods (average
data from 10 stations). Data are dates (dd.mm)
133
104
100
110
120
130
1971 1976 1981 1986 1991 1996
Year
DOY
271
249
245
255
265
275
1971 1976 1981 1986 1991 1996
123
153
110
120
130
140
150
1971 1976 1981 1986 1991 1996
x = –0.2402x + 262.65
R
2 = 0.185
y = –0.4326x + 128.25
R
2 = 0.2949
BEL
y = 0.2103x + 134.5
R
2 = 0.0657
Fig. 6. Betula pendula. Trends in the average (data from 10 stations) beginning (B), end (E) and length (L) of the growing season
for common birch B. pendula for the period 1971– 2000. DOY: day of year
Kalva
¯
ne et al.: Climate change and phenological phases
Overall, the NAO correlated well
with the phenological anomalies. In
analysing the relationship between the
NAO and the onset of the phases, it was
noticed that the connection with the
spring and summer phases was
stronger than that with autumn phases.
The NAOFeb relationship was slightly
stronger (raverage = –0.41 to –0.53) in
comparison with NAODec-Feb, where r
varied from –0.02 to –0.52.
4. DISCUSSION
As established in the majority of stud-
ies undertaken in Europe, the Latvian
and Lithuanian data analysis for the
1971–2000 period shows that a signifi-
cant negative trend existed in the onset
dates of spring and summer phases. In
addition, the spring phases had a wider
variation in onset dates than summer or
autumn phases and spring deviated
considerably (up to 40 d) from the long-
term average.
Data show that the earliest spring
flowering plants were very sensitive to
climate change. Unusually warm win-
ters influenced large negative devia-
tions from the average as in 1989 (the
highest January temperatures for the
1971–2000 period at all studied sta-
tions) and 1990 (the highest February
and March temperatures for the 1971–
2000 period at all studied stations). In
those years the flowering of hazel and
grey alder was recorded in January and
February, i.e. ~42 d earlier than the ob-
served average recorded. The present
study shows that phenophase onset
times changed considerably over the
last decade of the study period. For
example, since the 1990 spring phase, 9
out of 10 years have seen an earlier
onset than the long-term average.
In contrast to what has been recorded
in the most of Europe, the phenological
autumn tended to start earlier in Latvia
and Lithuania. This is consistent with the
findings of studies by Chmielewski &
Rötzer (2001) and Shutova et al. (2006),
who carried out their research in the
International Phenological Gardens
and in Kola Peninsula, respectively. Shu-
217
Phase (code) and Month Correlation with temperature Correlation with precipitation Correlation with NAO
species ravg rprev r avg. rprev ravg rDec – Feb
T0T1T2T3P0P1P2P3Dec Jan Feb
Beginning of flowering (BBCH61)
Alnus incana March – 0.64 –0.73 – 0.67 –0.54 0.00 –0.26 – 0.23 – 0.40 – 0.04 – 0.06 – 0.35 –0.01 –0.34 –0.41
(–0.76, (–0.80, (–0.81, (– 0.54, (– 0.30, (–0.67, (– 0.70, (– 0.73, (– 0.52, (–0.22, (– 0.52, (– 0.20, (–0.42, (–0.56,
–0.28) –0.62) –0.42) – 0.23) 0.38) 0.17) 0.12) –0.05) 0.39) 0.18) –0.02) 0.15) –0.24) 0.07)
Corylus avelana March –0.67 – 0.70 – 0.70 – 0.56 – 0.09 –0.27 – 0.23 – 0.38 – 0.03 –0.16 –0.41 –0.03 –0.33 –0.53
(–0.78, (–0.78, (–0.82, (– 0.72, (– 0.27, (–0.65, (– 0.39, (– 0.58, (– 0.47, (–0.40, (– 0.52, (– 0.08, (–0.37, (–0.66,
–0.47) –0.56) –0.45) – 0.38) 0.11) –0.02 –0.06) –0.09) 0.25) 0.01) –0.26) 0.05) –0.15) –0.33)
Syringa vulgaris May – 0.51 –0.59 –0.68 –0.41 – 0.29 –0.23 0.19 0.04 – 0.33 – 0.18 – 0.32 –0.02 –0.44 –0.25
(–0.62, (–0.68, (–0.76, (– 0.50, (– 0.36, (–0.55, (– 0.03, (– 0.15, (– 0.51, (–0.26, (– 0.43, (– 0.11, (–0.53, (–0.40,
–0.37) –0.49) –0.58) – 0.32) – 0.18) –0.01) 0.47) 0.19) – 0.14) – 0.08) –0.19) 0.09) – 0.30) –0.14)
Philadelphus June –0.48 – 0.33 – 0.52 – 0.52 –0.33 0.03 0.08 0.21 0.09 –0.29 –0.27 – 0.04 – 0.32 –0.24
coronaries (–0.68, (– 0.74, (–0.79, (– 0.73, (–0.52, (–0.17, (–0.16, (0.00, (–0.24, (–0.61, (–0.40, (– 0.27, (– 0.52, (–0.46,
–0.22) –0.13) –0.32) – 0.05) 0.07) 0.45) 0.40) 0.40) 0.34) 0.17) –0.03) 0.16) – 0.10) –0.04)
Leaf colouring (BBCH92)
Betula pendula September 0.12 0.06 0.23 0.12 – 0.07 –0.08 –0.03 –0.09 – 0.04 – 0.03 – 0.12 – 0.12 – 0.10 –0.04
(–0.30, (–0.49, (–0.25, (– 0.40, (– 0.43, (–0.34, (– 0.38, (– 0.42, (– 0.37, (– 0.25, (– 0.49, (– 0.54, (–0.46, (–0.35,
0.37) 0.51) 0.59) 0.38) 0.13) 0.44) 0.41) 0.11) 0.66) 0.24) 0.22) 0.29) 0.18) 0.25)
Acer platanoides September 0.05 0.15 0.22 0.05 –0.13 –0.07 0.01 – 0.03 – 0.19 0.10 –0.11 –0.09 –0.14 0.02
(–0.32, (–0.49, (–0.14, (– 0.44, (– 0.39, (–0.38, (– 0.47, (– 0.30, (– 0.46, (–0.07, (– 0.36, (– 0.43, (–0.43, (–0.40,
0.23) 0.48) 0.52) 0.23) 0.19) 0.49) 0.66) 0.38) 0.37) 0.68) 0.07) 0.12) 0.15) 0.25)
Table 4. Correlation coefficients (r) between start dates of phenological phases and temperature, precipitation and the North Atlantic Oscillation (NAODec – Feb) (average data
from 10 stations; minimum, maximum values for individual stations in parentheses); statistically significant correlation coefficients (> 0.36) with p-values < 0.05 indicated in
bold; ravg: correlation between phase and average data from 3 previous months; rpr ev —T
0, T1, T2, T3: relationship between phase and temperature in current month, previous
month, 2 months ago, 3 months ago; rprev —P
0, P1, P2, P3: relationship between phase and precipitation in current month, previous month, 2 months ago, 3 months ago
Clim Res 39: 209– 219, 2009
tova et al. (2006) suggested that earlier yellowing in the
Kola Peninsula was closely associated with a trend to-
wards an earlier decrease in autumn air temperature
within the study period, indicating that this environmen-
tal factor was important for autumn phenophases.
Chmielewski & Rötzer (2001) noted that growing sea-
sons in the Baltic region ended 0.1 d per 10 yr earlier, ex-
plained by Estonian phenologists (Ahas & Aasa 2006) as
the influence of the Baltic Sea. Bukantis (1995) found
that the most obvious influence of the Baltic Sea on cli-
mate, depending on land relief, was in the coastal belt
about 30 to 100 km wide.
Accordingly, analysing the long-term temperature
time series in Europe (Klein Tank & Können 2003)
found an increase in the number of cold days in
autumn during the period 1946–1999. We suggest that
the observed trend towards earlier autumn phenologi-
cal phases was associated with the increase in cold
days that cannot be derived from average monthly
temperatures (i.e. the extreme temperatures values
are not shown in the monthly mean values).
The length of the growing season (defined here as the
period between birch leafing and colouring) in the stud-
ied areas increased by 7 d, mainly through an earlier
onset of the spring phase. Similar to the observations in
Europe (Chmielewski & Rotzer 2002), the earliest
dates of leafing (the beginning of the growing season) in
Latvia and Lithuania were recorded in 1990, when
the growing season for birch started 20 d earlier, on av-
erage. The shortest growing season was observed in
1979, which was a cold year across the whole of Europe.
To provide a comparison with the work of other
authors (D’Odorico et al. 2002, Menzel 2003, Aasa et
al. 2004, Jaagus 2006), we examined the correlation
between the phenological data and the average air
temperature and precipitation of the present and the 3
preceding months at each station.
We found a strong correlation between the onset of a
phase and the average air temperature of the preced-
ing month. The strongest correlation among the spring
phases was observed between the onset of the phase in
March and the average air temperature in February. It
should be noted that another very significant relation-
ship was found between March air temperature and
the onset of the phase in March. During the summer
phase the strongest correlations were between the
onset of lilac flowering in May and April air tempera-
ture and between the onset of sweet mock orange
flowering in June and May air temperature. The rela-
tionship between the spring and summer phases is
inversely related because a higher air temperature of
the preceding month causes earlier onset of the phase.
The relationships between precipitation and pheno-
logical phases were generally weak. Among them the
strongest negative correlations were found between
the amount of precipitation in February and flowering
of Corylus avellana (5 of 10 coefficients were statisti-
cally significant) and Alnus incana (7 of 10 coefficients
were statistically significant) in March. This might be
associated with the character of snow accumulation,
which may delay the onset of the growing season.
Overall, the NAO correlated rather well with the
phenological anomalies of particular seasons. At the
same time we found stronger correlation with NAO in
winter months in Latvia, which could be explained by
the dominance of northern Europe circulation over
Latvia and central Europe circulation over Lithuania
(Jaak et al. unpubl. data).
The results indicate that, compared to summer, phe-
nological spring more often deviates from the long-
term average. For example, the 1989–1990 phenologi-
cal spring started 43 d earlier and larger NAO positive
values were detected during those years. The 1990
phenological summer, the greatest negative deviation,
started 2 wk earlier. In turn, both negative deviations,
earlier onset of the spring phase and summer phase
observed after the 1980s, correlated well with positive
NAO values. Positive deviations, i.e. later onset of
phase, recorded in the 1980s related to negative values
of the NAO.
No correlations with NAO were observed in the
autumn phases, with the exception of the colouring of
birch leaves (Latvian data) and the NAO of the 3 pre-
ceding months. One possible explanation could be the
weakening of the NAO and other types of circulation
during the autumn period (Jaagus 2006). Concerning
circulation characteristics and types in autumn, Drave-
niece (2007) verified that beginning in early autumn
(end of August) the equator–pole temperature gradi-
ent gradually gets stronger and, as a result, the west-
erly circulation starts strengthening and a transition to
a winter circulation occurs.
Further studies on the relationship between pheno-
phases and the changes in temperature and precipita-
tion should be carried out by using daily extremes of
temperature and precipitation values, which could
provide new insights in phenological variation on a
and temporal spatial scale.
Acknowledgements. Financial support from the Agency for
International Science and Technology Development Pro-
grammes in Lithuania and European Social Fund (Project No.
2004/0001/VPD1/ESF/PIAA/04/NP/3.2.3.2./0001/0001/0063)
is gratefully acknowledged.
LITERATURE CITED
Aasa A, Jaagus J, Ahas R, Sepp M (2004) The influence of
atmospheric circulation on plant phenological phases in
central and eastern Europe. Int J Climatol 24:1551–1564
218
➤
Kalva
¯
ne et al.: Climate change and phenological phases
Ahas R (1999) Long-term phyto-, ornitho- and ichthyopheno-
logical time-series analysis in Estonia. Int J Biometeorol
42:119– 124
Ahas R, Aasa A (2006) The effects of climate change on the
phenology of selected Estonian plant, bird and fish popu-
lations. Int J Biometeorol 51:17– 26
Ahas R, Aasa A, Menzel A, Fedotova VG, Scheifinger H (2002)
Changes in European spring phenology. Int J Climatol 22:
1727–1738
Beaubien EG, Freeland HJ (2000) Spring phenology trends in
Alberta, Canada: links to ocean temperature. Int J Bio-
meteorol 44:53–59
Bogaert J, Zhou L, Tucker CJ, Myneni RB, Ceulemans R (2002)
Evidence for a persistent and extensive greening trend in
Eurasia inferred from satellite vegetation index data.
J Geophys Res 107(D11):4119, doi:10.1029/2001JD001075
Bradley NL, Leopold AC, Ross J, Huffaker W (1999) Pheno-
logical changes reflects climate change in Wisconsin. Proc
Natl Acad Sci USA 96:9701– 9704
Bukantis A (1995) Extreme winters in the Baltic Sea area.
Geografijos Metra$tis 28:178– 194
Chmielewski FM, Rötzer T (2001) Response of tree phenology
to climate change across Europe. Agric Meteorol 108:
101–112
Chmielewski FM, Rötzer T (2002) Annual and spatial variabil-
ity of the beginning of growing season in Europe in rela-
tion to air temperature changes. Clim Res 19:257– 264
Chuine I, Yiou P, Viovy N, Seguin B, Daux V, Le Roy Ladurie
E (2004) Grape ripening as a past climate indicator.
Nature 432:289–290
D’Odorico P, Yoo JC, Jaeger S (2002) Changing seasons: an
effect of the North Atlantic Oscillation? J Clim 15:435 –445
Dospehov BA (1973) Meтoдикa пoлeвoгo oпытa (Field work
methodology). Kolos, Moscow (in Russian)
Draveniece A (2007) Oceanic and continental air masses over
Latvia. Latv Vegetacija 14:1–135 (in Latvian with English
Abstract)
Grisule G, Briede A (2007) Phenological time series in Latvia
as climate change indicators. In: Klavins M (ed) Climate
change in Latvia. University of Latvia, Riga
Harrington R, Woiwod I, Sparks T (1999) Climate change and
trophic interactions. Trends Ecol Evol 14:146– 150
Jaagus J (2006) Climatic changes in Estonia during the sec-
ond half of the 20th century in relationship with changes
in large-scale atmospheric circulation. Theor Appl Clima-
tol 83:77– 88
Klein Tank AMG, Können GP (2003) Trends in indices of daily
temperature and precipitation extremes in Europe,
1946– 1999. J Clim 16:3665 –3680
Kuliene˙ L, Tomkus J (1990) Bendroji fenolo‘
gija (General phe-
nology). Mokslas, Vilnius (in Lithuanian with English
Abstract)
Libiseller C, Grimvall A (2002) Performance of partial Mann
Kendall tests for trend detection in the presence of covari-
ates. Environmetrics 13:71– 84
Linderholm HW (2006) Growing season changes in the last
century. Agric Meteorol 137:1–14
McKnight TL, Hess D (2004) Physical geography: a landscape
appreciation. Prentice Hall, Upper Saddle River, NJ
Meier U (ed) (1997) Growth stages of mono- and dicotyledo-
nous plants. BBCH Monograph. Federal Biological Re-
search Centre for Agriculture and Forestry, Blackwell
Wissenschafts-Verlag, Berlin
Menzel A (2000) Trends in phenological phases in Europe be-
tween 1951 and 1996. Int J Biometeorol 44:76–81
Menzel A (2003) Plant phenological anomalies in Germany
and their relation to air temperature and NAO. Clim
Change 57:243– 263
Romanovskaja D, Bak$iene˙ E (2007) Influence of a thermal
mode on the seasonal phenological phenomena in Lithua-
nia. Ekologija 1:15– 20
Root TL, Price JT, Hall KR, Schneider SH, Rosenzweig C
(2003) Fingerprints of global warming on wild animals and
plants. Nature 421:57– 60
Shutova E, Wielgolaski F, Karlsen S, Makarova O and others
(2006) Growing seasons of Nordic mountain birch in north-
ernmost Europe as indicated by long-term field studies and
analyses of satellite images. Int J Biometeorol 51:155–166
219
Submitted: June 30, 2008; Accepted: March 19, 2009 Proofs received from author(s): August 3, 2009
➤
➤➤
➤
➤➤
➤
➤➤
➤
➤
➤➤
➤
➤➤
➤
➤
➤
➤➤
➤
➤➤