免费文献传递   相关文献

Response of Chestnut Flowering in Beijing to Photosynthetically Active Radiation Variation and Change in Recent Fifty Years

近五十年北京板栗始花物候对光合有效辐射变化的响应



全 文 :近五十年北京板栗始花物候对光合有效辐射变化的响应∗
郭  梁1ꎬ2ꎬ 胡  波3ꎬ 戴君虎4ꎬ 许建初1ꎬ5∗∗
(1 中国科学院昆明植物研究所山地生态系统研究中心ꎬ 云南 昆明  650201ꎻ 2 中国科学院大学ꎬ 北京  100049ꎻ
3 中国科学院大气物理研究所大气边界层物理与大气化学国家重点实验室ꎬ 北京  100029ꎻ 4 中国科学院
地理科学与资源研究所ꎬ 北京  100101ꎻ 5 世界农用林业中心东亚分部ꎬ 云南 昆明  650201)
摘要: 与全球范围内气候变暖对植物物候影响研究相比ꎬ 其他气候因素 (如光合有效辐射 PAR等) 对物
候影响报道较少ꎬ 果树花期物候对光合有效辐射变化响应的研究更是未见报道ꎮ 本研究以 1963-2008年间
北京板栗始花物候资料及相应的日光合有效辐射数据为基础ꎬ 利用偏最小二乘回归法确定了 PAR 影响板
栗始花物候的两个关键阶段ꎬ 进而分析了两阶段内 PAR、 温度及相对湿度变化对板栗花期的具体影响ꎮ
结果表明ꎬ 北京过去 50年两相关阶段内 PAR呈显著下降趋势ꎬ 其中 9 月 24 日至次年 2 月 5 日间 PAR 下
降对板栗花期提前具有促进作用ꎬ 可解释 12%的花期提前趋势ꎻ 2月 6 日至次年 5 月 31 日间 PAR 下降促
使花期延迟ꎬ 但未达显著水平 (P>0􀆰 1)ꎮ 板栗花期提前主要与 2 月 6 日至次年 5 月 31 日间温度升高有
关ꎬ 其间温度变化可解释 41%的花期提前趋势ꎻ 其次是相对湿度ꎬ PAR变化对花期影响较小ꎮ 鉴于 PAR、
温度及相对湿度间的互作效应ꎬ PAR和相对湿度对花期物候的影响可由温度效应加以解释ꎮ
关键词: 板栗ꎻ 始花ꎻ 温度ꎻ 相对湿度ꎻ 光合有效辐射ꎻ 偏最小二乘回归法
中图分类号: Q 948            文献标识码: A                文章编号: 2095-0845(2014)04-523-10
Response of Chestnut Flowering in Beijing to Photosynthetically
Active Radiation Variation and Change in Recent Fifty Years
GUO Liang1ꎬ2ꎬ HU Bo3ꎬ DAI Jun ̄Hu4ꎬ XU Jian ̄Chu1ꎬ5∗∗
(1 Centre for Mountain Ecosystem Studiesꎬ Kunming Institute of Botanyꎬ Chinese Academy of Sciencesꎬ Kunming 650201ꎬ Chinaꎻ
2 University of Chinese Academy of Sciencesꎬ Beijing 100049ꎬ Chinaꎻ 3 State Key Laboratory of Atmospheric Boundary Layer
Physics and Atmospheric Chemistryꎬ Institute of Atmospheric Physicsꎬ Chinese Academy of Sciencesꎬ Beijing 100029ꎬ Chinaꎻ
4 Institute of Geographical Sciences and Natural Resources Researchꎬ Chinese Academy of Sciencesꎬ Beijing
100101ꎬ Chinaꎻ 5 World Agroforestry Centreꎬ East Asia Nodeꎬ Kunming 650201ꎬ China)
Abstract: Climate warming has affected plant phenology throughout the worldꎬ but few studies have evaluated plant
phenology response to other climate factors ( e􀆰 g. photosynthetically active radiation ̄PAR). In particularꎬ the re ̄
sponse of fruit flowering to PAR variation has not been explored yet. Long ̄term (1963-2008) of chestnut (Castanea
mollissima Blume) first flowering dates from Beijingꎬ China were related with daily PAR for the 12 monthsꎬ using
Partial Least Squares (PLS) regression analysis. Two relevant phases were identifiedꎬ during which mean PARꎬ
temperatureꎬ and relative humidity (RH) were correlated with flowering datesꎬ respectively.
PAR during the both relevant periods decreased significantly in Beijing over the past 50 years. Reduced PAR
during 24 September ̄5 February showed an advance impact on chestnut floweringꎬ and could explain 12% of ad ̄
vance trend in flowering timing. Deceased PAR during 6 February ̄31 May had a delayed effect on tree floweringꎬ but
植 物 分 类 与 资 源 学 报  2014ꎬ 36 (4): 523~532
Plant Diversity and Resources                                    DOI: 10.7677 / ynzwyj201413175

∗∗
Funding: The National Natural Science Foundation of China (NSFC)’s project on “Response of Rhododendron arboreum Smith to climate
change in Eastern Himalaya” (31270524) and another Key Project of NSFC (41030101)
Author for correspondenceꎻ E ̄mail: jxu@mail􀆰 kib􀆰 ac􀆰 cn
Received date: 2013-09-06ꎬ Accepted date: 2013-11-19
作者简介: 郭  梁 (1984-) 男ꎬ 博士ꎬ 从事气候变化对作物物候及产量影响研究ꎮ E ̄mail: guoliang@mail􀆰 kib􀆰 ac􀆰 cn
it was not significant enough to reject the null hypothesis of no impact over time. Advanced flowering of chestnut was
mainly determined by increasing temperature between 6 February and 31 May which could explain 41% of flowering
trend. Relative humidity variation during this period played secondly important role on tree flowering. Considering the
interaction among these three climate factorsꎬ the impacts of PAR and RH on flowering timing could partially be at ̄
tributed to the effects of temperature variation.
Key words: Chestnutꎻ First floweringꎻ Temperatureꎻ Relative humidityꎻ Photosynthetically active radiationꎻ Partial
Least Squares regression
  Long ̄term phenological observations at species
level and specific sites can provide direct evidence
of the impacts of climatic variation and change
(Rosenzweig et al.ꎬ 2008). While the majority of
published studies have focused on phenological re ̄
sponses to global warming and interpreted phenologi ̄
cal advance as the consequence of increasing tem ̄
perature (Esterlla et al.ꎬ 2007ꎻ Peñuelas and Filel ̄
laꎬ 2001ꎻ Vitasse et al.ꎬ 2011)ꎬ relatively few re ̄
ports are available on the response of plants to other
climatic factor variations. One such change is a de ̄
crease in the levels of photosynthetically active radi ̄
ation (PAR) reaching the Earth’s surface.
PAR is the visible portion of the electromagnet ̄
ic spectrum (400-700 nm) of solar radiationꎬ cov ̄
ering both energy and photon terms ( Hu et al.ꎬ
2007). PAR plays an important role in plant
growthꎬ developmentꎬ biomass productionꎬ and
flowering responses (Fausey et al.ꎬ 2005ꎻ Gregoriou
et al.ꎬ 2007ꎻ Hicklentonꎬ 1987ꎻ Nathanꎬ 1984ꎻ
Olesen and Grevsenꎬ 1997ꎻ Wu et al.ꎬ 2010). It is
also vital to understand the Earth’s climate system
and climate change (Hu et al.ꎬ 2010). Advanced or
delayed flowering events have been reported recently
mainly through manual control irradiation amounts
(Fausey et al.ꎬ 2005ꎻ Hicklentonꎬ 1987ꎻ Saifuddin
et al.ꎬ 2010). Responses of plant phenology to PAR
variation in nature are reported less frequently.
Located on the North China Plainꎬ Beijing has
the longest and most abundant records of phenology
in China (Zhang et al.ꎬ 2005). Compared to Ameri ̄
can and European zones at similar latitudeꎬ climate
variation and change in Beijing are substantially
greaterꎬ leading to greater variation in the timing of
phenological events and providing valuable informa ̄
tion for elucidating climate responses of species (Lu
et al.ꎬ 2006). Detailed and long ̄term meteorological
observations in Beijing further provide an unparallel
opportunity to assess impacts of PAR changes on
plant phenology. Beijing currently has a total popula ̄
tion near to 21 million. Rapid economic growth and
urbanization progress have caused severe air pollu ̄
tion ( e􀆰 g. hazeꎬ sand and dust storm) in and a ̄
round Beijing. Increased aerosol mass loading in the
atmosphere likely affects both the quantity and quali ̄
ty of solar radiation received at earth surface. De ̄
creasing PAR during 1958-2005 has been reported
in Beijing (Hu et al.ꎬ 2010). How will plants re ̄
spond to reduced PAR? What will happen for plant
phenology with decreasing PAR? More scientific at ̄
tention should be paid to such research to close the
knowledge gap.
Response of fruit tree phenology to PAR varia ̄
tionꎬ to our knowledgeꎬ has never been reported. In
Beijingꎬ there is a long history of chestnut cultiva ̄
tionꎬ and chestnut production generates tens of mil ̄
lions of dollars in revenue for farmers. For chestnutꎬ
flowering phenology has vital impacts on pollinationꎬ
fruit set and production (Legave et al.ꎬ 2008). Our
analysis was based on an observational record of
chestnut first flowering from Beijingꎬ China. A novel
method (Partial Least Squares regression) was used
to correlate the phenology dates to PAR variation at
daily resolution. The objective of the present study is
to identify the relevant periods during which PAR
can influence chestnut first floweringꎬ and to com ̄
prehensively evaluate responses of chestnut flowering
to PAR variation during the relevant periods. The
relative importance of other climatic factors (temper ̄
atureꎬ relative humidity) for explaining variation in
425                                  植 物 分 类 与 资 源 学 报                            第 36卷
chestnut first flowering has also been evaluated in
the present study.
1  Materials and methods
1􀆰 1  Phenology data
In Beijingꎬ species ̄level phenological observa ̄
tions of plants were mainly conducted at the Summer
Palace (40°01′Nꎬ 116°20′Eꎬ 50 m a􀆰 s􀆰 l.)ꎬ a for ̄
mer royal gardens with a long history of more than
150 years. Phenological data of chestnut (Castanea
mollissima Blume) at the Summer Palace during
1963-2008 were acquired from the Chinese Pheno ̄
logical Observation Network (CPON) ̄a nationwide
system of monitoring stations that has conducted
standardizedꎬ systematic and comprehensive pheno ̄
logical observations of plants and animals across Chi ̄
na since 1963. Details of the phenological observa ̄
tion method have been described by Wan and Liu
(1979) and Lu et al. (2006). In this analysisꎬ first
flowering phenology was usedꎬ and it is defined
when 10% of flowers are openꎬ corresponding to
stage 61 on the BBCH (‘Biologische Bundesanstalt
Bundessortenamt und Chemische Industrie’ ) scale
(Meier et al.ꎬ 1994).
1􀆰 2  Climate data
Since direct photosynthetically active radiation
(PAR) measurement has not been routinely conduc ̄
ted in Chinaꎬ daily PAR values in Beijing during
1963-2008 were calculated according to procedures
used by Hu et al. (2010). Required inputs for these
calculations were daily broadband solar radiationꎬ
daily extraterrestrial solar irradianceꎬ daytime lengthꎬ
and average of the cosine of the solar zenith angles
from sunrise to sunset. They were all from the China
Meteorological Administration ( CMA) No. 54511
site ( Beijing Meteorological Station) which is only
2􀆰 5 km from the Summer Palaceꎬ so that climatic
data recorded there should closely mirror conditions
at the observation site. To ensure the emergence of
recognizable response patterns between PAR and
chestnut first flowering in subsequent statistical ana ̄
lysesꎬ we subjected daily PAR to an 11 ̄day running
mean (Guo et al.ꎬ 2013ꎻ Luedeling and Gassnerꎬ
2012ꎻ Luedeling et al.ꎬ 2013).
To evaluate impacts of other climate factors on
tree flowering datesꎬ daily mean temperature and daily
relative humidity data at Beijing Meteorological Station
during 1963-2008 were also used in this analysis.
1􀆰 3   Identification of relevant periods influen ̄
cing chestnut first flowering
Partial Least Squares (PLS) regression was used
to analyze the response of chestnut first flowering to
variation of daily PAR during all 365 days of the
yearꎬ based on data for 1963-2008. Dependent vari ̄
ables were the first flowering dates (expressed in day
of the year)ꎬ while independent variables were daily
PAR values for 365 days preceding chestnut first
flowering. PLS regressionꎬ a procedure commonly
used in chemometrics (Wold et al.ꎬ 2001) and hy ̄
perspectral remote sensing (Luedeling et al.ꎬ 2009)ꎬ
is a regression technique that can be used reliably in
situationsꎬ where independent variables are highly
auto ̄correlated and where the number of independ ̄
ent variables exceeds the number of observations.
Recent studies have proved that PLS regression can
effectively be used to analyze relationships between
phenology and climate variation (Guo et al.ꎬ 2013ꎻ
Luedeling and Gassnerꎬ 2012ꎻ Luedeling et al.ꎬ
2013ꎻ Ranjitkar et al.ꎬ 2013ꎻ Yu et al.ꎬ 2010).
The two major outputs of PLS analysis are the
variable importance in the projection ( VIP ) and
standardized model coefficients. The VIP values re ̄
flect the importance of all independent variables for
explaining variation in the dependent variablesꎬ with
0􀆰 8 often used as threshold for determining impor ̄
tance (Woldꎬ 1995). The standardized model coef ̄
ficients indicate the strength and direction of the im ̄
pact of each variable in the PLS model (Luedeling et
al.ꎬ 2013). Periods with VIP scores greater than
0􀆰 8 and high absolute values of model coefficients
represent the relevant phases influencing chestnut
bloom timing (Guo et al.ꎬ 2013). Positive coeffi ̄
cients indicate that positive deviation of PAR during
the respective period are related to delayed first flow ̄
5254期      GUO Liang et al.: Response of Chestnut Flowering in Beijing to Photosynthetically Active Radiation 􀆺     
eringꎬ while negative coefficients imply an advanced
impact on flowering.
1􀆰 4   PAR temporal trend during the relevant
periods and its impacts on chestnut flowering
Linear regression was used to analyze the recent
50 years’ PAR temporal trends and the relationship
between chestnut first flowering phenology and mean
PAR during the relevant periods confirmed above.
Results were assessed for significance using analysis
of variance. To investigate the relationship between
flowering timing and mean PAR during these two rel ̄
evant periodsꎬ instead of using common multiple re ̄
gressionꎬ a novel three dimensional PAR response
surface was constructed in this analysisꎬ which was
interpolated using the Kriging technique in the R
package ‘fields’ (Furrer et al.ꎬ 2012).
1􀆰 5  Impacts of other climate factors during the
relevant periods on chestnut flowering
Temporal trends of temperature and relative hu ̄
midity during the relevant periodsꎬ and the relation ̄
ships between chestnut first flowering and mean tem ̄
perature and relatively humidity during above phases
were analyzed using linear regressionꎬ respectively.
Results were evaluated for significance using analysis
of variance.
All analyses were implemented in the R 2􀆰 15􀆰 2
programming language (R Development Core Teamꎬ
2012). All procedures used in this study are includ ̄
ed in the package ‘ chillR’ ( Luedelingꎬ 2013)ꎬ
which depends heavily on the ‘pls’ package (Mevik
et al.ꎬ 2011).
2  Results
2􀆰 1  Relevant periods for chestnut first flower ̄
ing to variation of daily PAR
Between 1963 and 2008ꎬ the average first flow ̄
ering date of chestnut at Beijing Summer Palace was
2 June. The 365 daily PAR values between the pre ̄
vious July and June were used as independent varia ̄
bles in the PLS regressionꎬ while dependent varia ̄
bles were chestnut first flowering datesꎬ expressed in
day of the year. Based on the VIP and standardized
model coefficients of the PLS regressionꎬ two periods
during which PAR showed significant correlations with
chestnut first flowering dates were identified (Fig􀆰 1).
The impacts of variation of daily PAR on flow ̄
ering before autumn (July ̄September) were not con ̄
sidered since a long time span from first flowering of
chestnut (2 June) for the next year existedꎬ and the
response of tree flower buds to climate factors com ̄
monly starts from the autumn of previous year. Be ̄
tween 24 September and 5 Februaryꎬ model coeffi ̄
cients were mostly positive and VIP values mostly
exceeded 0􀆰 8 ( the threshold for variable impor ̄
tance)ꎬ indicating that higher PAR should delay
first flowering of chestnut. Howeverꎬ this period also
included two spells with negative model coefficients
indicating that the direction of PAR effect varied
throughout this period. The lower absolute values of
negative model coefficients meant a relatively smaller
impact on chestnut flowering compared with the en ̄
tire delayed effect of higher PAR. Taking a broader
view at model coefficients and VIP scores during this
periodꎬ we interpreted the entire period (24 Septem ̄
ber to 5 February) as one relevant phase for the tim ̄
ing of chestnut first flowering.
Between 6 February and 31 Mayꎬ most model
coefficients were negative and VIP values were al ̄
most always importantꎬ justifying consideration of the
entire period as a homogeneous blockꎬ during which
increasing PAR were correlated with advanced chest ̄
nut floweringꎬ while decreasing PAR linked to de ̄
layed flowering.
2􀆰 2  PAR temporal trend during the relevant pe ̄
riods and its impacts on chestnut flowering
During 1963-2008ꎬ PAR in Beijing displayed
significant decreasing trends in the two relevant peri ̄
ods confirmed aboveꎬ especially in autumn and win ̄
ter (P<0􀆰 01) (Fig􀆰 2). Decreased mean PAR du ̄
ring 24 September - 5 February advanced chestnut
first flowering by 0􀆰 75 day per one unit decrease of
mean PAR (P<0􀆰 05). The correlation between mean
PAR in later ̄winter and spring (6 February to 31
May) and chestnut flowering was not significant (P>
625                                  植 物 分 类 与 资 源 学 报                            第 36卷
0􀆰 1)ꎬ although delayed impacts of decreased PAR
on tree bloom were identified in Fig􀆰 1 and Fig􀆰 2.
Plotting first flowering dates of chestnut as a
function of mean PAR during the both relevant peri ̄
ods clearly showed that chestnut bloom dates were
primarily determined by variation of mean PAR in
autumn and winter (24 September to 5 February).
Reduced autumn and winter PAR could advance first
flowering of chestnut. The limited influence of varia ̄
tion of PAR in later ̄winter and spring was indicated
by almost vertical contour lines in Fig􀆰 2 (right part).
2􀆰 3  Impacts of other climate factors during the
relevant periods on chestnut flowering
In Beijingꎬ mean temperature during the two
relevant periods showed significant increase trendsꎬ
by 0􀆰 5 and 0􀆰 7 ℃ per decade (P<0􀆰 01)ꎬ respec ̄
tivelyꎬ between 1963 and 2008 ( Fig􀆰 3). The ad ̄
vance in blossoming was significantly related to mean
temperatures between 6 February and 31 May. First
flowering advanced by 2􀆰 5 days per 1 ℃ rise of mean
temperature during this period (P<0􀆰 01). Tempera ̄
tures between 24 September and 5 February were
not significantly related to first flowering (P>0􀆰 1).
Rough 41% of advance trend of chestnut flowering
can be explained by temperature variation between 6
February and 31 May (P<0􀆰 01).
Fig􀆰 1  Results of PLS regression correlating first flowering dates for chestnut at Beijing Summer Palace with 11 ̄day running means of
daily PAR from the previous July to June. Blue bars in the top panel indicate VIP values greater than 0􀆰 8ꎬ the threshold for variable
importance. In the middle panelꎬ red color means the model coefficients are negative (and important)ꎬ while the green color indicates
positive (and important) relationships between flowering and PAR. The black line in the bottom figure stands for the daily mean
PARꎬ while the greyꎬ green and red areas represent the standard deviation of PAR for each day of the year
7254期      GUO Liang et al.: Response of Chestnut Flowering in Beijing to Photosynthetically Active Radiation 􀆺     
Fig􀆰 2  Trends of PAR and relationships between chestnut flowering and mean PAR in the relevant periods from 1963 to 2008 at Beijing
Summer Palace. Panels in the first and second column show trends of PAR over time and relationships between the timing of flowering and
mean PAR during relevant periods. Trends are significant with ∗P<0􀆰 1ꎬ ∗∗P<0􀆰 05ꎬ ∗∗∗P<0􀆰 01. The right panel displays first flowering
dates of chestnut as a function of mean PAR during 24 September to 5 February (x ̄axis) and 6 February to 31 May (y ̄axis) . Variation in
color reflects variation in first flowering datesꎬ while black dots indicate phenological observations between 1963 and 2008
Fig􀆰 3  Trends of temperature and relative humidity (RH)ꎬ and relationships between chestnut first flowering and mean temperature and
RH in the relevant periods from 1963 to 2008 at Beijing Summer Palace. The top row is the analysis for temperatureꎬ and the bottom row for
RH analysis. Trends are significant with ∗P<0􀆰 1ꎬ ∗∗P<0􀆰 05ꎬ ∗∗∗P<0􀆰 01
825                                  植 物 分 类 与 资 源 学 报                            第 36卷
    Mean relative humidity (RH) during 24 Sep ̄
tember - 5 February and 6 February - 31 May has
decreased significantly since 1963 (P<0􀆰 05). Neg ̄
ligible impact of variation of mean RH during the for ̄
mer period on tree flowering was clear in Fig􀆰 3 (P>
0􀆰 1 and R2 =0􀆰 01)ꎬ while decreased mean RH du ̄
ring 6 February - 31 May advanced chestnut flower ̄
ing significantly ( P < 0􀆰 01). Rough 18% of ad ̄
vanced chestnut flowering can be attributed to varia ̄
tion of mean RH during this period.
3  Discussion
3􀆰 1  Long ̄term trend of PAR in Beijing
Based on an all ̄weather reconstruction modelꎬ
Hu et al. (2010) evaluated the inter ̄annual varia ̄
tions and long ̄term trend of PAR in Beijing. During
1958-2005ꎬ significant decrease in PAR had been
observed (0􀆰 12 mol􀅰m-2􀅰a-1) which was similar to
previous studies in Beijing (Che et al.ꎬ 2005). The
decreasing trend of PAR was sharp in spring and
summerꎬ but slight in autumn and winter. In present
researchꎬ PAR during the relevant periods showed
significant decreasing trend between 1963 and 2008ꎬ
mainly occurring in autumnꎬ winter and spring (24
September - 31 May).
In the cold winter of Beijingꎬ large amounts of
fossil fuels are burned for heating which has been
exacerbated recently by rapid population and eco ̄
nomic growthꎬ and urbanization progressꎬ causing an
increase in aerosol emission. These aerosol particles
can absorb and scatter solar radiation in the atmos ̄
phere and likely result in the decrease of PAR in au ̄
tumn and winter. In springꎬ the decrease of PAR
seems to be caused by frequently occurred sand
storms which have assailed Beijing even since Yuan
Dynasty of China (1271-1368 A􀆰 D.) (Songꎬ 2002).
Howeverꎬ sand storms have been weakened and air
quality has been improved since 1990 due to the im ̄
provement of land surface ecological environment in
and around Beijing (Hu et al.ꎬ 2010)ꎬ especially
the construction of the Three ̄North Shelter Forest
Program initiated in 1978 (Yang et al.ꎬ 1998). In
our analysisꎬ fluctuant but whole ascending PAR in
spring after 1990 gave certain supports to the reports
of reduced sand storm threat to Beijing ( Fig􀆰 2).
Howeverꎬ in recent ten yearsꎬ haze and dust storms
with average particulate matter diameter of 2􀆰 5 μm
(PM2􀆰 5) have occurred frequently in Beijing and
have aroused worldwide attention. They also reduce
PAR in spring to some extent. To prevent threat from
hazeꎬ sand and dust stormsꎬ China should take more
vigorous action to reduce discharge of greenhouse
gasꎬ and issue a series of relative policies in afforest ̄
ationꎬ water conservationꎬ energy restructuringꎬ air
pollution controlꎬ and so on.
3􀆰 2  PAR change impacts on tree flowering
PAR is an important climatic factor for plant
growthꎬ developmentꎬ productivityꎬ and flowering
responses ( Fausey et al.ꎬ 2005ꎻ Gregoriou et al.ꎬ
2007ꎻ Hicklentonꎬ 1987ꎻ Nathanꎬ 1984ꎻ Olesen
and Grevsenꎬ 1997ꎻ Wu et al.ꎬ 2010). In natural
selectionꎬ plants have evolved to induce leafing and
flowering to coincide with seasonal peaks of irradia ̄
tionꎬ maximizing PAR interception and absorption
(De Camacaro et al.ꎬ 2002ꎻ Foggo and Warringtonꎬ
1989ꎻ Mitchell and Woodwardꎬ 1988ꎻ Williams ̄Lin ̄
eraꎬ 2003ꎻ Wright and Van Schaikꎬ 1994). Many
studies have been conducted to evaluate the response
of plant flowering to PAR variation through artificial ̄
ly increased and decreased irradiation. Low PAR can
suppress flower bud development ( Turner and E ̄
wingꎬ 1988 )ꎬ and cause incomplete or delayed
plant flowering (Fausey et al.ꎬ 2005ꎻ Saifuddin et
al.ꎬ 2010)ꎬ while supplemental PAR in spring can
advance flowering of several floricultural crops (Hick ̄
lentonꎬ 1987). Advanced flowering phase caused by
increased PAR has also been reported in an oak for ̄
est phenology study (Sierra et al.ꎬ 1996).
Limited attention is paid to explore how PAR
variations are related to plant flowering based on di ̄
rect ground observations rather than manual experi ̄
ments. Long ̄term phenological data and detailed me ̄
teorological records in Beijing provide direct evi ̄
dence of the impacts of PAR variation and change on
9254期      GUO Liang et al.: Response of Chestnut Flowering in Beijing to Photosynthetically Active Radiation 􀆺     
tree flowering. In the present studyꎬ the relevant pe ̄
riods for chestnut flowering to PAR variation were
confirmed using PLS analysisꎬ and it is the first time
to report different impacts of PAR during different
periods on tree flowering in the scientific community.
In autumn and winter (24 September to 5 Februar ̄
y)ꎬ there was a positive relationship between PAR
and tree first floweringꎬ and reduced PAR during
this period has significantly advanced chestnut flow ̄
ering in the past 50 years (P<0􀆰 05). In later ̄winter
and spring (6 February to 31 May)ꎬ negative corre ̄
lation between PAR and flowering phenology exis ̄
tedꎬ howeverꎬ decreased PAR has not showed sig ̄
nificant impact on chestnut flowering (P>0􀆰 1). On ̄
ly taking impacts of solar irradiation on tree flowering
into considerationꎬ chestnut bloom responses were
relatively dominated by an advancing effect of re ̄
duced PAR in autumn and winter.
3􀆰 3   Different climate factor impacts on tree
flowering
Temperature and PAR are believed to be inter ̄
active to determine plant flowering since increased
irradiance could cause thermal effect to some extent
(Berningerꎬ 1994ꎻ Erwinꎬ 2006ꎻ Oh et al.ꎬ 2009).
The photo ̄thermal ratio ( PTR) modelꎬ which was
defined as the ratio of radiant energy to thermal en ̄
ergy ( temperature)ꎬ has been used recently to de ̄
scribe the combined effects of temperature and irra ̄
diance on plant growth and development (Niu et al.ꎬ
2000). Other climatic factors have also been tested
to clarify which is critical to plant phenology varia ̄
tion. Recentlyꎬ a flowering phenology research con ̄
ducted in beech and oak forests has indicated that
the most important factors for flowering there were
soil moistureꎬ and relative humidityꎬ followed by
temperatureꎬ numbers of daylight hoursꎬ and PAR
(Sierra et al.ꎬ 1996). Howeverꎬ most of current
knowledge about plant phenologyꎬ including numer ̄
ous experimental studies ( Menzel and Fabianꎬ
1999ꎻ Price and Waserꎬ 1998)ꎬ indicates that ob ̄
served phenology changes are mostly due to increas ̄
ing temperatures (Borchert et al.ꎬ 2005ꎻ Esterlla et
al.ꎬ 2007ꎻ Peñuelas and Filellaꎬ 2001ꎻ Zheng et
al.ꎬ 2006).
In our analysisꎬ rough 41% of advance trend of
chestnut flowering can be explained by temperature
variation during 6 February - 31 May (P < 0􀆰 01)ꎬ
while decreased relative humidity during this period
can explain 18% of tree flowering trend. Comparative ̄
lyꎬ only 12% of flowering trend can be attributed to
variation of PAR during 24 September - 5 February.
Thusꎬ our results were in line with most reports that
plant phenology are mostly determined by temperature.
There seems to be some correlation and interac ̄
tion among temperatureꎬ PARꎬ and relative humidi ̄
tyꎬ while the latter two can both be explained
through temperature effects. Based on long ̄term re ̄
cords of chestnut first flowering and daily tempera ̄
ture data between 1963 and 2008 in Beijingꎬ Guo et
al. (2013) confirmed the relevant periods for flo ̄
wering to daily temperature variation which were al ̄
most identical to these periods in this analysis. This
indicated similar impacts of temperature and PAR on
tree flowering. Decreased PAR in autumn and winter
seems to function as reduced temperatureꎬ while the
latter benefits the chilling accumulation of plantsꎬ
and leads to earlier dormancy breaking and flower
sprouting. The advanced impact on flowering of de ̄
creasing PAR in autumn and winter actually oc ̄
curred in this analysisꎬ probably due to general cor ̄
relation between radiation and warmth. Between 6
February and 31 Mayꎬ decreased relative humidity
can link to increased temperature in the same period.
Thusꎬ the advance effect of relative humidity varia ̄
tion in spring could also be explained by the corre ̄
sponding temperature increase.
4  Conclusions
Response of fruit flowering to PAR variation has
been evaluated in this study for the first timeꎬ based
on long ̄term phenological and meteorological records
in Beijingꎬ China. Two relevant periods were identi ̄
fied during which PAR had completely opposite im ̄
pacts on tree flowering. In autumn and winter (24
035                                  植 物 分 类 与 资 源 学 报                            第 36卷
September to 5 February )ꎬ observed decreasing
PAR advanced chestnut flowering significantlyꎬ
while reduced PAR in later ̄winter and spring ( 6
February to 31 May) imposed negligible effect on
flowering. Comparativelyꎬ the most important climate
factor for chestnut flowering was temperatureꎬ fol ̄
lowed by relatively humidity and PAR. Impacts of
PAR and relative humidity on tree flowering could
partially be explained by temperature effects.
Acknowledgements: This research further support was sup ̄
plied by the Consultative Group on International Agricultural
Research Program 6: Forestsꎬ Trees and Agroforestryꎬ and
Research Program 7: Climate Changeꎬ Agriculture and Food
Security. We greatly appreciate the staff in the Institute of Ge ̄
ographic Sciences and Natural Resource Research at the Chi ̄
nese Academy of Sciences for organizingꎬ collecting and pub ̄
lishing phenology data across China over several decades. The
history meteorological database used in this study was ob ̄
tained from the National Meteorological Centre of China Mete ̄
orological Administration (CMA)ꎬ which was highly appreci ̄
ated by the authors. We also greatly thank Eike Luedelingꎬ a
famous phenologistꎬ giving detailed methodology guides in the
data analysis.
References:
Berninger Eꎬ 1994. Development rate of young greenhouse rose plants
(Rosa hybrida) rooted from cuttings in relation to temperature
and irradiance [J] . Scientia Horticulturaeꎬ 58 (3): 235—251
Borchert Rꎬ Robertson Kꎬ Schwartz M et al.ꎬ 2005. Phenology of tem ̄
perate trees in tropical climates [J] . International Journal of Bi ̄
ometeorologyꎬ 50 (1): 57—65
De Camacaro MEPꎬ Camacaro GJꎬ Hadley P et al.ꎬ 2002. Pattern of
growth and development of the strawberry cultivars Elsantaꎬ Bo ̄
leroꎬ and Everest [J] . Journal of the American Society for Horti ̄
cultural Scienceꎬ 127 (6): 901—907
Che HZꎬ Shi GYꎬ Zhang XY et al.ꎬ 2005. Analysis of 40 years of so ̄
lar radiation data from Chinaꎬ 1961-2000 [ J] . Geophysical Re ̄
search Lettersꎬ 32 (6): L06803
Erwin Jꎬ 2006. Factors affecting flowering in ornamental plants [A].
In: Anderson N (ed.)ꎬ Flower Breeding and Genetics SE ̄1 [M].
Netherlands: Springerꎬ 7—48
Esterlla Nꎬ Sparks THꎬ Menzel Aꎬ 2007. Trends and temperature re ̄
sponse in the phenology of crops in Germany [J] . Global Change
Biologyꎬ 13 (8): 1737—1747
Fausey BAꎬ Heins RDꎬ Cameron ACꎬ 2005. Daily light integral af ̄
fects flowering and quality of greenhouse ̄grown Achilleaꎬ Gauraꎬ
and Lavandula [J] . HortScienceꎬ 40 (1): 114—118
Foggo MNꎬ Warrington IJꎬ 1989. The influence of photosynthetically
active radiation and vernalization on flowering of Deschampsia
flexuosa ( L.) Trin. ( Poaceae) [ J] . Functional Ecologyꎬ 3:
561—567
Furrer Rꎬ Nychka Dꎬ Sain Sꎬ 2012. Fields: Tools for spatial data. R
package version 6􀆰 7 [ OL ]. http: / / cran􀆰 r ̄project􀆰 org / web /
packages / fields /
Gregoriou Kꎬ Pontikis Kꎬ Vemmos Sꎬ 2007. Effects of reduced irradi ̄
ance on leaf morphologyꎬ photosynthetic capacityꎬ and fruit yield
in olive ( Olea europaea L.) [ J] . Photosyntheticaꎬ 45 ( 2):
172—181
Guo Lꎬ Dai Jꎬ Ranjitkar S et al.ꎬ 2013. Response of chestnut phenol ̄
ogy in China to climate variation and change [ J] . Agricultural
and Forest Meteorologyꎬ 180: 164—172
Hicklenton PRꎬ 1987. Flowering of Gypsophila paniculata cv. Bristol
Fairy in relation to irradiance [ J] . Acta Horticulturaeꎬ 205:
103—112
Hu Bꎬ Wang Yꎬ Liu Gꎬ 2007. Measurements and estimations of pho ̄
tosynthetically active radiation in Beijing [ J] . Atmospheric Re ̄
searchꎬ 85 (3): 361—371
Hu Bꎬ Wang Yꎬ Liu Gꎬ 2010. Long ̄term trends in photosynthetically
active radiation in Beijing [ J] . Advances in Atmospheric Sci ̄
encesꎬ 27: 1380—1388
Legave JMꎬ Farrera Iꎬ Almeras T et al.ꎬ 2008. Selecting models of
apple flowering time and understanding how global warming has
had an impact on this trait [ J] . Journal of Horticultural Science
and Biotechnologyꎬ 83 (1): 76—84
Lu Pꎬ Yu Qꎬ Liu J et al.ꎬ 2006. Advance of tree ̄flowering dates in
response to urban climate change [ J] . Agricultural and Forest
Meteorologyꎬ 138 (1): 120—131
Luedeling Eꎬ 2013. chillR: Statistical methods for phenology analysis
in temperate fruit trees. R package version 0􀆰 54 [ OL]. ht ̄
tp: / / cran􀆰 r ̄project􀆰 org / web / packages / chillR /
Luedeling Eꎬ Gassner Aꎬ 2012. Partial Least Squares Regression for
analyzing walnut phenology in California [ J] . Agricultural and
Forest Meteorologyꎬ 158 ̄159: 43—52
Luedeling Eꎬ Hale Aꎬ Zhang M et al.ꎬ 2009. Remote sensing of spi ̄
der mite damage in California peach orchards [ J] . International
Journal of Applied Earth Observation and Geoinformationꎬ 11
(4): 244—255
Luedeling Eꎬ Kunz Aꎬ Blanke MMꎬ 2013. Identification of chilling
and heat requirements of cherry trees—a statistical approach
[J] . International Journal of Biometeorologyꎬ 57: 679—689
Meier UHꎬ Graf Hꎬ Hack H et al.ꎬ 1994. Phänologische entwick ̄
lungsstadien des kernobstes (Malus domestica Borkh. Und Pyrus
communis L.)ꎬ des Steinobstes ( Prunus ̄Arten)ꎬ der Johannis ̄
beere ( Ribes ̄Arten) und der Erdbeere ( Fragaria × ananassa
Duch.) [J] . Nachrichtenbl Deut Pflanzenschutzdꎬ 46: 141—153
Menzel Aꎬ Fabian Pꎬ 1999. Growing season extended in Europe [J] .
1354期      GUO Liang et al.: Response of Chestnut Flowering in Beijing to Photosynthetically Active Radiation 􀆺     
Natureꎬ 397 (6721): 659
Mevik BHꎬ Wehrens Rꎬ Liland Kꎬ 2011. PLS: Partial least squares
and principal component regression. R package version 2􀆰 3􀆰 0
[OL]. http: / / cran􀆰 r ̄project􀆰 org / web / packages / pls /
Mitchell PLꎬ Woodward FIꎬ 1988. Responses of three woodland herbs
to reduced photosynthetically active radiation and low red to far ̄
red ratio in shade [J] . Journal of Ecologyꎬ 76: 807—825
Nathan KKꎬ 1984. A note on the relationship between photosyntheti ̄
cally active radiation and light intensity [J] . Archives for Meteor ̄
ologyꎬ Geophysicsꎬ and Bioclimatologyꎬ Series Bꎬ 35 (3): 233—
238
Niu Gꎬ Heins RDꎬ Cameron AC et al.ꎬ 2000. Day and night tempera ̄
turesꎬ daily light integralꎬ and CO2 enrichment affect growth and
flower development of pansy (Viola × wittrockiana) [ J] . Journal
of the American Society for Horticultural Scienceꎬ 125 (4): 436—
441
Oh Wꎬ Cheon IHꎬ Kim KS et al.ꎬ 2009. Photosynthetic daily light in ̄
tegral influences flowering time and crop characteristics of Cyc ̄
lamen persicum [J] . HortScienceꎬ 44 (2): 341—344
Olesen JEꎬ Grevsen Kꎬ 1997. Effects of temperature and irradiance on
vegetative growth of cauliflower ( Brassica oleracea L. botrytis)
and broccoli (Brassica oleracea L. italica) [J] . Journal of Exper ̄
imental Botanyꎬ 48 (8): 1591—1598
Peñuelas Jꎬ Filella Iꎬ 2001. Responses to a warming world [J] . Sci ̄
enceꎬ 294 (5543): 793—795
Price MVꎬ Waser NMꎬ 1998. Effects of experimental warming on
plant reproductive phenology in a subalpine meadow [ J] . Ecolo ̄
gyꎬ 79 (4): 1261—1271
R Development Core Teamꎬ 2012. R: A language and environment for
statistical computing. R Foundation for Statistical Computingꎬ
Viennaꎬ Austria. ISBN 3 ̄900051 ̄07 ̄0 [OL]. http: / / www􀆰 r ̄
project􀆰 org /
Ranjitkar Sꎬ Luedeling Eꎬ Shrestha KK et al.ꎬ 2013. Flowering phe ̄
nology of tree rhododendron along an elevation gradient in two
sites in the Eastern Himalayas [J] . International Journal of Bio ̄
meteorologyꎬ 57 (2): 225—240
Rosenzweig Cꎬ Karoly Dꎬ Vicarelli M et al.ꎬ 2008. Attributing physi ̄
cal and biological impacts to anthropogenic climate change [ J] .
Natureꎬ 453 (7193): 353—357
Saifuddin Mꎬ Hossain AMBSꎬ Normaniza Oꎬ 2010. Growth of Bougain ̄
villea glabra [J]. Asian Journal of Plant Sciencesꎬ 9 (1): 20—27
Sierra GGꎬ Merino APꎬ Herrero EAꎬ 1996. Phenology of Hyacinthoides
non ̄scripta (L.) chouardꎬ Melittis melissophyllum L. and Symphy ̄
tum tuberosum L. in two deciduous forests in the Cantabrian
mountainsꎬ Northwest Spain [J] . Vegetatioꎬ 122 (1): 69—82
Song Yꎬ 2002. Cause of Beijing sandstorm and its control [J] . Urban
Environment and Urban Ecologyꎬ 15 (6): 26—28
Turner ADꎬ Ewing EEꎬ 1988. Effects of photoperiodꎬ night tempera ̄
tureꎬ and irradiance on flower production in the potato [J] . Po ̄
tato Researchꎬ 31 (2): 257—268
Vitasse Yꎬ François Cꎬ Delpierre N et al.ꎬ 2011. Assessing the effects
of climate change on the phenology of European temperate trees
[J] . Agricultural and Forest Meteorologyꎬ 151 (7): 969—980
Wan MWꎬ Liu XZꎬ 1979. Method of Phenology Observation of China
[M]. Beijing: Science Pressꎬ Beijingꎬ 1—22
Williams ̄Linera Gꎬ 2003. Temporal and spatial phenological variation
of understory shrubs in a tropical montane cloud forest [J] . Bio ̄
tropicaꎬ 35 (1): 28—36
Wold Sꎬ 1995. PLS for multivariate linear modeling [A]. In: van der
Waterbeemd H ( ed.)ꎬ Chemometric Methods in Molecular De ̄
sign: Methods and Principles in Medicinal Chemistry [M]. Ger ̄
many: Verlag ̄Chemieꎬ Weinheimꎬ 195—218
Wold Sꎬ Sjöström Mꎬ Eriksson Lꎬ 2001. PLS ̄regression: a basic tool
of chemometrics [ J] . Chemometrics and Intelligent Laboratory
Systemsꎬ 58 (2): 109—130
Wright SJꎬ Van Schaik CPꎬ 1994. Light and the phenology of tropical
trees [J] . American Naturalistꎬ 143: 192—199
Wu Gꎬ Leeuw Jꎬ Skidmore A et al.ꎬ 2010. Comparison of extrapola ̄
tion and interpolation methods for estimating daily photosynthet ̄
ically active radiation (PAR) [J] . Geo ̄spatial Information Sci ̄
enceꎬ 13 (4): 235—242
Yang Dꎬ Fang Xꎬ Li Xꎬ 1998. Analysis on the variation trend of sand ̄
storm in Northern China [J] . Quarterly Journal of Applied Mete ̄
orologyꎬ 9: 352—358
Yu Hꎬ Luedeling Eꎬ Xu Jꎬ 2010. Winter and spring warming result in
delayed spring phenology on the Tibetan Plateau [J] . Proceed ̄
ings of the National Academy of Sciences of the United States of
Americaꎬ 107 (51): 22151—22156
Zhang Xꎬ Ge Qꎬ Zheng J et al.ꎬ 2005. Responses of spring phenology
to climate changes in Beijing in last 150 years [ J] . Chinese
Journal of Agrometeorologyꎬ 26 (3): 263—267
Zheng Jꎬ Ge Qꎬ Hao Z et al.ꎬ 2006. Spring phenophases in recent
decades over Eastern China and its possible link to climate chan ̄
ges [J] . Climatic Changeꎬ 77 (3 ̄4): 449—462
235                                  植 物 分 类 与 资 源 学 报                            第 36卷