免费文献传递   相关文献

Flower Color Polymorphism in Rhododendron cyanocarpum (Ericaceae), an Endangered Alpine Species Endemic to NW Yunnan, China

滇西北特有植物蓝果杜鹃的花色多态性研究



全 文 :滇西北特有植物蓝果杜鹃的花色多态性研究∗
马永鹏1ꎬ 吴之坤1ꎬ 张长芹1ꎬ2ꎬ 孙卫邦1∗∗
(1 中国科学院昆明植物研究所植物园ꎬ 昆明  650201ꎻ 2 云南特色木本花卉工程技术中心ꎬ 昆明  650000)
摘要: 蓝果杜鹃 (Rhododendron cyanocarpum) 为大理苍山特有的濒危植物ꎬ 有粉色和白色两种花冠类型ꎮ
为了探讨该物种花色多态性的意义ꎬ 本研究调查了粉色花和白色花植株在已知的各居群的分布频率、 花冠
的反射光谱及其它的花部特征、 有效传粉者及其访花频率与结实情况ꎮ 结果表明: 粉色花植株在所有调
查的居群中占优势 (77%~100%)ꎮ 粉色花的花冠反射光谱在 430 nm 和 650 nm 有两个峰ꎬ 而白色花只在
430 nm有一个反射峰ꎮ 同时ꎬ 花特征如: 花柱与柱头颜色、 花冠长度、 花萼长度、 花梗长度以及雌雄蕊
最短距离ꎬ 两种花冠存在显著差异ꎮ 另外ꎬ 尽管熊蜂作为这两种花冠的主要传粉者ꎬ 但粉红花的访花
频率以及自然条件下的结实情况显著高于白色花ꎮ 本研究结果推测粉红色花可能受到了稳定性选择的
作用ꎮ
关键词: 蓝果杜鹃ꎻ 熊蜂ꎻ 花色多态性ꎻ 访花频率
中图分类号: Q 944ꎬ Q 948          文献标志码: A          文章编号: 2095-0845(2015)01-021-08
Flower Color Polymorphism in Rhododendron cyanocarpum
(Ericaceae)ꎬ an Endangered Alpine Species
Endemic to NW Yunnanꎬ China
MA Yong ̄peng1ꎬ WU Zhi ̄kun1ꎬ ZHANG Chang ̄qin1ꎬ2ꎬ SUN Wei ̄bang1∗∗
(1 Kunming Botanical Gardenꎬ Kunming Institute of Botanyꎬ Chinese Academy of Sciencesꎬ Kunming 650201ꎬ Chinaꎻ
2 Yunnan Characteristics of Wood Flowers Engineering and Technical Research Centerꎬ Kunming 650000ꎬ China)
Abstract: Rhododendron cyanocarpum is a narrow endemic species with pink and white floral color. In the present
studyꎬ to investigate the significance of petal color morphsꎬ we examined color morph frequenciesꎬ petal color reflec ̄
tance and other associated floral charactersꎬ effective pollinatorsꎬ visitation frequenciesꎬ and fruit production in the
field. In all surveyed known populationsꎬ plants with pink color morph dominanted and comprised 77%-100% of in ̄
dividuals. Two peaks at 430 nm and 650 nm were found in the petal color reflectance of pink flowersꎬ and only one
peak at 430 nm was found in the reflectance spectrum of white flowers. In additionꎬ color morphs were also associat ̄
ed with colors of style and stigmaꎬ lengths of corollaꎬ calyx and pedicelꎬ the closest distance between stigma and sta ̄
menꎬ but not with style length and nectar production. Moreoverꎬ higher visitation frequency of their shared pollina ̄
tors (bumblebees) and fruit production were observed of pink flowers than white flowers. Despite a briefly temporal
and spacial studyꎬ we suggest that color morph frequenciesꎬ visit frequencies of bumblebees and fruit productionꎬ all
favor to be stablilizing selection for the pink color morph.
Key words: Rhododendron cyanocarpumꎻ Bumblebeesꎻ Floral color morphsꎻ Visit frequency
植 物 分 类 与 资 源 学 报  2015ꎬ 37 (1): 21~28
Plant Diversity and Resources                                    DOI: 10.7677 / ynzwyj201514068

∗∗
Funding: The National Natural Science Foundation of China (31200247ꎬ U1302262)ꎻ The Natural Science Foundation of Yunnan Province
(2012FB180)
Author for correspondenceꎻ E ̄mail: wbsun@mail􀆰 kib􀆰 ac􀆰 cn
Received date: 2014-04-28ꎬ Accepted date: 2014-07-10
作者简介: 马永鹏 (1982-) 男ꎬ 副研究员ꎬ 主要从事重要观赏植物类群的自然杂交与新品种选育方面的研究ꎮ
E ̄mail: mayongpeng@mail􀆰 kib􀆰 ac􀆰 cn
  Flowers are detected and discriminated by pol ̄
linators according to specific signals such as colorꎬ
sizeꎬ shapeꎬ odor and other characteristics. Of theseꎬ
flower color is considered to be associated with polli ̄
nator attraction. Therefore the maintenance of floral
color polymorphism within natural populations is an
interesting evolutionary puzzle ( Frey et al.ꎬ 2011ꎻ
Rausherꎬ 2008). It is well established that pollina ̄
tors are agents of directional selection on floral color
(e􀆰 g. Stanton et al.ꎬ 1989ꎻ Campbell et al.ꎬ 1997ꎻ
Emms and Arnoldꎬ 2000ꎻ Wesselingh and Arnoldꎬ
2000ꎻ Jones and Reithelꎬ 2001ꎻ Streisfeld and
Kohnꎬ 2005ꎻ Frey et al.ꎬ 2011). Henceꎬ adaptation
to different pollinators was thought to drive diver ̄
gence in flower color and morphology (e􀆰 g. Frey et
al.ꎬ 2011). Otherwiseꎬ if floral color morphs sharing
the same pollinatorsꎬ then the color morph most at ̄
tractive or frequent should be favoredꎬ thereby lead ̄
ing to stabilized flower color variation if selection is
consistent ( e􀆰 g. Waser and Priceꎬ 1981ꎻ Smithson
and Macnairꎬ 1996ꎬ 1997ꎻ Subramaniam and
Rausherꎬ 2000ꎻ Tremblay and Ackermanꎬ 2003).
Moreoverꎬ it has been well documented that bumble ̄
bees showed a preference for a common floral color
by frequency ̄dependent selection when foraging on
two colors of artificial flowers (Realꎬ 1990ꎻ Smith ̄
son and Macnairꎬ 1996).
Compared with interspecific variationsꎬ intraspe ̄
cific variations of flower color are relatively uncom ̄
mon (Matsumura et al.ꎬ 2006). To explain this dis ̄
connectꎬ several authors have proposed alternativeꎬ
but not mutually exclusiveꎬ hypotheses where selec ̄
tion through pollinators ( direct selection on color)
or other environmental factors ( indirect selection on
color) maintains the color polymorphism (e􀆰 g. Jonesꎬ
1996ꎻ Gigord et al.ꎬ 2001ꎻ Medel et al.ꎬ 2003ꎻ
Tremblay and Ackermanꎬ 2007ꎻ Frey et al.ꎬ 2011).
On the basis of field observationꎬ we found pet ̄
al color polymorphism in Rhododendron cyanocarpumꎬ
an alpine species with narrowly distribution and en ̄
demic to NW Yunnanꎬ China (Fig􀆰 1Aꎬ C). In the
present studyꎬ we investigated color morph frequen ̄
cies across all known populationsꎬ compared floral
characters and flower color reflectanceꎬ effective pol ̄
linators and their visitation frequenciesꎬ and fruit
production among treatmentsꎬ with the aim of better
understanding the maintenance of floral color pat ̄
terns in R􀆰 cyanocarpum.
1  Materials and methods
1􀆰 1  Plant species and study sites
R􀆰 cyanocarpum is an endangered evergreen shrub
belonging to Subsect. Thomsonii Sleumer in Subge ̄
nus. Hymenanthes (Chamberlainꎬ 1982ꎻ Wang and
Xieꎬ 2004). It is endemic to mountain slopes above
3 000 m in the Cangshan Mountains around Daliꎬ
NW Yunnan (Chamberlainꎬ 1982ꎻ Wuꎬ 1986). It
has umbelliform or racemose inflorescences with 5-9
flowers. Flowers exhibited color polymorphism and
petal color (pink or white) was consistent within in ̄
dividuals (Fig􀆰 1Aꎬ C). The flowering period is a ̄
bout late March to May. According to three consecu ̄
tive years’ observationꎬ flowering time was not diver ̄
gent for both pink and white flowering individuals.
For a single flowerꎬ both of pink and white color mor ̄
phs flowered approximately 8 - 10 days. The flowers
have characteristics of apically porose anthers and ad ̄
hesive pollen. The stigma surface usually began to
become wet at the same time as corolla openingꎬ and
remained sticky throughout anther dehiscence.
1􀆰 2  Frequencies of petal color morphs among
populations
We examined all known five populations of R􀆰 cy ̄
anocarpum covering both east and west slopes of the
Cangshan Mountains in Aprilꎬ 2009. In Huadianba
(HDB) and Ganchaiqing (GCQ)ꎬ we randomly chose
20 plots with 15 m × 15 m and carefully examined the
color morphs of all flowering individuals. Due to re ̄
stricted distributionsꎬ the other three populationsꎬ i􀆰 e.
Xiaohuadian (XHD)ꎬ Guogaishan (GGS) and Dian ̄
shita (DST)ꎬ 12 plots with 15 m×15 m were chosen
and the color morphs of all flowering individuals
were carefully examined in each plot ( Table S1).
Thenꎬ we calculated proportion of each color morph
22                                  植 物 分 类 与 资 源 学 报                            第 37卷
among total examined individuals per plot.
1􀆰 3  Petal color analysis
To assess light reflection patterns of both pink
and white petals at different wavelengthsꎬ we ob ̄
tained spectral data from petals using a S2000 minia ̄
ture fiber optic spectrometer with a PX ̄2 pulsed xen ̄
on lamp (Ocean Opticsꎬ Dunedinꎬ FLꎬ USA). All
measurements were carried out in the range from 250
nm to 750 nmꎬ using 0􀆰 30 nm increments. Differ ̄
ences in the reflectance pattern have been found in
certain plants (Frey et al.ꎬ 2011)ꎬ but preliminary
testing for changes in the reflectance pattern along
the same petal did not show any differencesꎬ and
therefore only one measurement per petal was taken.
In total 20 petals from different flowersꎬ obtained
from ten individuals for each color morph were ana ̄
lyzed to assess their potential differences.
1􀆰 4  Floral characters of color morphs
To determine whether color morph was associat ̄
ed with other floral traitsꎬ nine floral characters were
observed at HDBꎬ i􀆰 e. style colorꎬ stigma colorꎬ
flower number per inflorescenceꎬ corolla lengthꎬ
style lengthꎬ calyx lengthꎬ pedicel lengthꎬ nectar
volume and the closest distance between stigma and
stamen. Flowers were randomly selected from 4 pink
and 4 white individuals for above mentioned meas ̄
urements. To obtain accurate volume of nectarꎬ we
selected flower buds from 4 pink and 4 white individ ̄
uals and bagged to exclude visitors and prevent evap ̄
oration (Ng and Corlettꎬ 2000)ꎬ and then measured
it directly from calibrated capillary tube for a volume
reading when they were newly open (Table 2).
Fig􀆰 1  Floral colors (Aꎬ C) and associated reflectance spectrums (Bꎬ D) in R􀆰 cyanocarpum
321期      MA Yong ̄peng et al.: Flower Color Polymorphism in Rhododendron cyanocarpum (Ericaceae)ꎬ an 􀆺     
1􀆰 5  Visit frequencies of the effective pollinators
to different color morphs
Effective pollinators have been well documented
and bumblebees were the mainly effective pollinators
to R􀆰 cyanocarpum (Ma et al.ꎬ 2010ꎻ Ma et al.ꎬ 2014).
To investigate visit frequencies of pollinators to each
color morphꎬ we observed pollinator activity in a 5 m
× 5 m quadrate that only contained one pink flowering
individual and one white flowering individualꎬ in
Aprilꎬ 2008. To make the same floral displayꎬ each
plant was left with 300 flowers and we recorded the
number of pollinators’ visit to the selected individuals
(from 10 ∶ 00 to 14 ∶ 00) in four successively sunny
days. We then calculated the number of visits to se ̄
lected individuals per hour as visit frequency. For sta ̄
tistical needꎬ four plots were conducted to newly con ̄
structed quadrats under similar weather conditions
(Sun et al.ꎬ 2005). In 2009ꎬ we used the same
method and floral display sizeꎬ and three replications
were conducted in four successively sunny days.
1􀆰 6  Pollination treatments
To examine reproductive success under narual
condition and whether reproductive barrier might oc ̄
cur between the two floral color morphsꎬ six treat ̄
ments were manipulated from each of 12 inflores ̄
cences in 4 pink and 4 white flowering individuals at
HDB. Totally 76ꎬ 78ꎬ 67 and 92 flowersꎬ prior to
anther dehiscenceꎬ were selectedꎬ then flowers were
covered with nylon nets after four pollination treat ̄
ments: (1) white × whiteꎬ (2) white × pink♀ꎬ
(3) pink × white♀ and (4) pink × pinkꎬ with
pollen donor from different plants at least 10 m away.
Other two treatmentsꎬ 67 white flowers and 75 pink
flowers in each of 12 inflorescences were randomly
selected from 4 pink and 4 white individuals as con ̄
trol ( 5)ꎬ ( 6). Fruits produced by these flowers
were counted in October when they were full ̄sized
and after fruit abortion had occurred ( Ma et al.ꎬ
2010). Reproductive success was assessed on the
basis of fruit set.
1􀆰 7  Data analysis
All data were examined for normal distribution
with a one ̄sample K ̄S test (Ma et al.ꎬ 2012). For
frequencies of each color morph in examined popula ̄
tionsꎬ data were arcsine ̄transformed before compari ̄
son to normalize the data. One ̄way anova analysis
was used to compare frequencies of color morphsꎬ
floral charactersꎬ visit frequencies and fruit sets a ̄
mong treatments. Two ̄way anova was used to assess
the effects of population and petal color on propor ̄
tions of each color morph. All means and standard
deviations were calculated for all quantitative varia ̄
bles using SPSS 11􀆰 5 for Windows ( SPSSꎬ Chica ̄
goꎬ ILꎬ USA).
2  Results
2􀆰 1  Frequencies of color morphs among populations
Among the observed 1 711 flowering plantsꎬ the
white flowering individuals only accounted for 7% in
total ( Table S1). Pink and white flowering plants
were asymmetrically mixed in four populations. The
white morph plants accounted for 13%ꎬ 4%ꎬ 3%ꎬ
11%ꎬ and pink morph plants for 77%ꎬ 96%ꎬ 97%ꎬ
89% in HDBꎬ XHDꎬ GCQ and DSTꎬ respectively
(Fig􀆰 2). In GGSꎬ we did not find white color morph
and the pink color morph hence accounted for 100%.
Proportion was significant difference between color
morphs in each of five examined populations (Table
1). For the same color morphꎬ observed proportions
also varied among populations (Table 1)ꎬ with the
Fig􀆰 2  Proportions (mean ± s􀆰 e.) of each color morph
in the examined five populations
42                                  植 物 分 类 与 资 源 学 报                            第 37卷
lowest proportion of white color morph in GGS and
highest in HDB (Fig􀆰 2).
Table 1  Two ̄way anova table for the effect of population and petal
color on proportions of each color morph in R􀆰 cyanocarpum
Source of variation d􀆰 f. MS F
Population 4 0􀆰 168 10􀆰 412∗∗
Petal color 1 56􀆰 209 3490􀆰 235∗∗
Population × petal color 4 0􀆰 548 34􀆰 01∗∗
Error 142 0􀆰 016
Total 152
∗∗ Significant effect (P < 0􀆰 01)
2􀆰 2  Petal color reflectance
The reflectance spectrum of petals showed ex ̄
tremely low variation between plants ( gray area in
Fig􀆰 1Bꎬ D)ꎬ all clearly showing a marked peak in
the reflectance spectrum at 430 nm. Howeverꎬ an ̄
other peak reflectance spectrum was also examined
at 650 nm only in pink flowers.
2􀆰 3  Floral characters of color morphs
Pink color morph had red style and red stigmaꎬ
but style was lighter green and stigma was pink in
white color morph. In the observed flower numbers
per inflorescenceꎬ style length and nectar produc ̄
tionꎬ no difference existed between two color mor ̄
phs. Corollaꎬ calyx and pedicel in white color morph
were significantly longer than in pink color morph.
Howeverꎬ for the distance between stigma and sta ̄
menꎬ white color morph was significantly shorter
than pink color morph (see Table 2 for details) .
2􀆰 4  Visit frequencies of the effective pollinators
to different color morphs
The visit frequencies of bumblebees were 1􀆰 02
± 0􀆰 16 (mean ± s􀆰 d.) and 0􀆰 73 ± 0􀆰 07 in 2008 and
2009 to pink flowersꎬ and 0􀆰 53 ± 0􀆰 18 and 0􀆰 50 ±
0􀆰 03 to white flowers in 2008 and 2009. The differ ̄
ence is significant in both years (Fig􀆰 3).
2􀆰 5  Pollination treatments
Fruit set in pink flowers (61%) was significantly
higher than white flowers (74%) under natural con ̄
dition (Fig􀆰 4). For other four treatmentsꎬ fruit sets
Table 2  Floral characters of the two color morphsꎬ with the same letter indicating no difference
Color
patterns
Style
color
Stigma
color
Flower
number1
(n= 40)
Corolla
length
(n= 20)
Style
length
(n= 20)
Calyx
length
(n= 20)
Pedicel
length
(n= 20)
Closest
distance2
(n= 20)
Nectar
volume
(n= 22)
Pink Red Red 6􀆰 00±0􀆰 99a 4􀆰 58±0􀆰 55a 3􀆰 96±0􀆰 42a 0􀆰 52±0􀆰 09a 1􀆰 27±0􀆰 10a 0􀆰 81±0􀆰 08a 0􀆰 076±0􀆰 018a
White Lighter ̄green Pink 6􀆰 05±0􀆰 88a 4􀆰 95±0􀆰 18b 3􀆰 96±0􀆰 14a 0􀆰 68±0􀆰 09b 1􀆰 81±0􀆰 14b 0􀆰 74±0􀆰 06b 0􀆰 086±0􀆰 017a
1. Flower number per inflorescenceꎻ 2. The closest distance between stigma and stamen
Fig􀆰 3  Visit frequencies (mean ± SE) of bumblebees
in each color morph in 2008 and 2009ꎬ with the
same letter indicating no difference
Fig􀆰 4  Fruit sets (mean ± SE) among the six treatments (1ꎬ white
× whiteꎻ 2ꎬ white × pink♀ꎻ 3ꎬ pink × white♀ꎻ 4ꎬ pink ×
pinkꎻ 5 and 6ꎬ natural conditions of white and pink flowers)ꎬ
with the same letter indicating no difference
521期      MA Yong ̄peng et al.: Flower Color Polymorphism in Rhododendron cyanocarpum (Ericaceae)ꎬ an 􀆺     
( all>87%) were higher than natural condition treat ̄
mentsꎬ but are not significantly different within these
hand cross pollination treatments (Fig􀆰 4).
3  Discussion
Few reports have been involved in intraspecific
variations of flower color mixed in the same popula ̄
tion but see Aquilegia caerulea ( Millerꎬ 1981 )ꎬ
Dactylorhiza sambucina (Gigord et al.ꎬ 2001ꎻ Pel ̄
legrino et al.ꎬ 2005 )ꎬ Limonium wrightii ( Mat ̄
sumura et al.ꎬ 2006) and Lepanthes rupestris (Trem ̄
blay and Ackermanꎬ 2007 ). R􀆰 cyanocarpum had
pink and white flowersꎬ and the proportions of two
color morphs varied among populationsꎬ nevertheless
high proportion of pink color morphs was consistent
in all surveyed populations. In GGSꎬ no white color
morph was foundꎬ which can be explained mainly for
such two reasons. Firstꎬ considering the low propor ̄
tion of white color morph in other four observed pop ̄
ulationsꎬ it seemed that GGS was only composed of
pink flowering individuals. Secondꎬ white color mor ̄
ph existed but was not flowering yet during our ob ̄
servationꎬ hence we could have missed the white
color morph. In observed populationsꎬ these two col ̄
or morph individuals gathered in the same areaꎬ and
once a white flowering individual was foundꎬ often
with many pink flowering individuals closely sur ̄
rounding. Soꎬ environmental factors such as pHꎬ
water availability might not contribute to color differ ̄
entiation.
Floral color variation was often associated with
other floral characters (Freyꎬ 2007). Our results in ̄
deed proved such traits as colors of style and stigmaꎬ
length of corollaꎬ calyx and pedicel. For the closest
distance between stigma and stamenꎬ autogamy can
occur easier in white color morph than in red oneꎬ
which will need to be confirmed in the future studies.
Abundant nectar generally attracted more pollinators
(see Mitchellꎬ 2004ꎬ for a review). No significant
difference of nectar production showed flower color
may be the main reason for visit frequencies of bum ̄
blebees.
Floral color offered a visual cue that stimulated
pollinator sensory systems and that selectively attrac ̄
ted certain types of pollinators ( Stebbinsꎬ 1974ꎻ
Melendez ̄Ackerman and Campbellꎬ 1998ꎻ Hodges
et al.ꎬ 2002). Our results showed both color morphs
shared the same pollinators and floral color variation
was not associated with pollinators’ divergency. For
the same floral displayꎬ pink color morph is capable
of attracting more visit frequencies of pollinatorsꎬ
thereby sucessfully leading to higher fruit set under
natural condition. It should be pointed out that bird
might be another pollinator in the late flowering of
R􀆰 cyanocarpum on the basis of personal information
from local people in the study area. If soꎬ the loss of
petal color reflectance peak which are attractive to
bird by the peak reflectance between 600 and 700
nm in white flowers might indicate the loss of bird
pollination potential (Shrestha et al.ꎬ 2013).
Divergency of floral color may due to adaptive
selection or non ̄adaptive mechanismsꎬ such as ran ̄
dom mutation ( Epperson and Cleggꎬ 1987ꎻ Brad ̄
shaw and Schemskeꎬ 2003)ꎬ genetic drift (Wilson
and Thomsonꎬ 1996) and indirect response to selec ̄
tion on pleiotropically related floral traits (Tremblay
and Ackermanꎬ 2007ꎻ Coberly and Rausherꎬ 2008).
For R􀆰 cyanocarpumꎬ we do not understand which
mechanism contributed to the formation of such floral
color morphs. Howeverꎬ once floral polymorphism
existedꎬ which are frequently assumed to be main ̄
tained via natrual selection (Tremblay and Ackermanꎬ
2007). In the present studyꎬ effective pollinators are
both bumblebeesꎬ hence hypothesis that pollinator
divergency maintains floral polymorphism can be re ̄
jected. Howeverꎬ even initial visit frequencies of
both two color morphs are the sameꎬ higher frequen ̄
cy of pink color morph (92􀆰 9%) favored to be fre ̄
quency ̄dependent selection by bumblebeesꎬ as a
reault of learning ( Smithson and Macnairꎬ 1996).
In additionꎬ fruit set also favored to pink color mor ̄
ph under natural condition. Thereforeꎬ reasons men ̄
tioned aboveꎬ all favor to be stablilizing selection for
the pink color morph. Howeverꎬ selection can be
62                                  植 物 分 类 与 资 源 学 报                            第 37卷
complex in space and timeꎬ other floral charactersꎬ
such as the closest distance between stigma and sta ̄
menꎬ may partially facilitate reproduction by autoga ̄
my in the futureꎬ thereby maitaining white color
morph within and among populations.
Acknowledgements: We sincerely thank T􀆰 W. Yang for
field assistance and Prof. L􀆰 Z. Wang for insect identification
from Yunnan University.
References:
Bradshaw HDꎬ Schemske DWꎬ 2003. Allele substitution at a flower
colour locus produces a pollinator shift in monkey flowers [ J] .
Natureꎬ 426: 176—178
Campbell DRꎬ Waser NMꎬ Melendez ̄Ackerman EJꎬ 1997. Analyzing
pollinator ̄mediated selection in a plant hybrid zone: humming ̄
bird visitation patterns on three spatial scales [ J] . American
Naturalistꎬ 149: 295—315
Chamberlain DFꎬ 1982. A revision of Rhododendron II. Subgenus
Hyemanthes [J] . Notes on Royal Botanic Garden Edinburgꎬ 39:
209—486
Coberly LCꎬ Rausher MDꎬ 2008. Pleiotropic effects of an allele pro ̄
ducing white flowers in Ipomoea purpurea [ J] . Evolutionꎬ 62:
1076—1085
Emms SKꎬ Arnold MLꎬ 2000. Site ̄to ̄site differences in pollinator vis ̄
itation patterns in a Louisiana iris hybrid zone [J] . Oikosꎬ 91:
568—578
Epperson BKꎬ Clegg MTꎬ 1987. Instability at a flower color locus in
the morning glory [J] . Journal of Heredityꎬ 78: 346—352
Frey FMꎬ 2007. Phenotypic integration and the potential for independ ̄
ent color evolution in a polymorphic spring ephemeral [ J] .
American Journal of Botanyꎬ 94: 437—444
Frey FMꎬ Dunton Jꎬ Garland Kꎬ 2011. Floral color variation and asso ̄
ciations with fitness ̄related traits in Malva moschata (Malvace ̄
ae) [J] . Plant Species Biologyꎬ 26: 235—243
Gigord LDBꎬ Macnair MRꎬ Smithson Aꎬ 2001. Negative frequency ̄
dependent selection maintains a dramatic flower color polymor ̄
phism in the rewardless orchid Dactylorhiza sambucina (L.) [J] .
Proceedings of the National Academy of Sciencesꎬ USAꎬ 8:
6253—6255
Hodges SAꎬ Whittall JBꎬ Fulton M et al.ꎬ 2002. Genetics of floral
traits influencing reproductive isolation between Aquilegia formo ̄
sa and Aquilegia pubescens [J] . American Naturalistꎬ 59: 51—
60
Jones KNꎬ 1996. Fertility selection on a discrete floral color polymor ̄
phism in Clarkia (Onagraceae) [J] . Evolutionꎬ 50: 71—79
Jones KNꎬ Reithel JSꎬ 2001. Pollinator ̄mediated selection on a flower
color polymorphism in experimental populations of Antirrhinum
( Scrophulariaceae) [J] . American Journal of Botanyꎬ 88: 447—
454
Ma YPꎬ Zhang CQꎬ Zhang JL et al.ꎬ 2010. Natural hybridization be ̄
tween Rhododendron delavayi and R􀆰 cyanocarpum (Ericaceae)ꎬ
from morphologicalꎬ molecular and reproductive evidence [ J] .
Journal of Integrative Plant Biologyꎬ 52: 844—851
Ma YPꎬ Wu ZKꎬ Tian XL et al.ꎬ 2012. Growth discrepancy between
filament and style facilitates autonomous self ̄fertilization in
Hedychium yunnanense [J] . Plant Ecology and Evolutionꎬ 145:
185—189
Ma YPꎬ Wu ZKꎬ Dong K et al.ꎬ 2014. Pollination biology of Rhodo ̄
dendron cyanocarpum ( Ericaceae)ꎬ an alpine species endemic
to NW Yunnanꎬ China [ J] . Journal of Systematics and Evolu ̄
tion (acceptedꎬ DOI: JSE12114)
Matsumura Sꎬ Yokoyama Jꎬ Tateishi Y et al.ꎬ 2006. Intraspecific var ̄
iation of flower colour and its distribution within a sea lavenderꎬ
Limonium wrightii (Plumbaginaceae)ꎬ in the northwestern Pacif ̄
ic Islands [J] . Journal of Plant Researchꎬ 119: 625—632
Medel Rꎬ Botto ̄Mahan Cꎬ Kalin ̄Arroyo Mꎬ 2003. Pollinator ̄media ̄
ted selection on the nectar guide phenotype in the Andean mon ̄
key flowerꎬ Mimulus luteus [J] . Ecologyꎬ 84: 1721—1732
Melendez ̄Ackerman Eꎬ Campbell DRꎬ 1998. Adaptive significance of
flower color and inter ̄trait correlations in an Ipomopsis hybrid
zone [J] . Evolutionꎬ 52: 1293—1303
Miller RBꎬ 1981. Hawkmoths and the geographic patterns of floral var ̄
iation in Aquilegia caerulea [J] . Evolutionꎬ 35: 763—774
Mitchell RJꎬ 2004. Heritability of nectar traits: why do we know so
little [J]? Ecologyꎬ 85: 1527—1533
Ng SCꎬ Corlett RTꎬ 2000. Comparative reproductive biology of six
species of Rhododendron (Ericaceae) in Hong Kongꎬ South Chi ̄
na [J] . Canadian Journal of Botanyꎬ 78: 221—229
Pellegrino Gꎬ Caimi Dꎬ Noce ME et al.ꎬ 2005. Effects of local density
and flower colour polymorphism on pollination and reproduction in
the rewardless orchid Dactylorhiza sambucina ( L.) Soo [ J ] .
Plant Systematics and Evolutionꎬ 251: 119—129
Real LAꎬ 1990. Predator switching and the interpretation of animal
choice behavior: the case for constrained optimization [A]. / /
Hughes RN eds.ꎬ Behavioural Mechanisms of Food Selection
[M]. Springer ̄Verlagꎬ Heidelbergꎬ 1—21
Rausher MDꎬ 2008. Evolutionary transitions in floral color [J] . Inter ̄
national Journal of Plant Sciencesꎬ 169: 7—21
Shrestha Mꎬ Dyer AGꎬ Boyd ̄Gerny S et al.ꎬ 2013. Shades of red:
bird ̄pollinated flowers target the specific colour discrimination
abilities of avian vision [J] . New Phytologistꎬ 198: 301—310
Smithson Aꎬ Macnair MRꎬ 1996. Frequency ̄dependent selection by
pollinators: mechanisms and consequences with regard to behav ̄
iour of bumblebees Bombus terrestris ( L) ( Hymenoptera: Api ̄
dae) [J] . Journal of Evolutionary Biologyꎬ 9: 571—588
Smithson Aꎬ Macnair MRꎬ 1997. Negative frequency ̄dependent selec ̄
tion by pollinators on artificial flowers without rewards [J] . Evo ̄
721期      MA Yong ̄peng et al.: Flower Color Polymorphism in Rhododendron cyanocarpum (Ericaceae)ꎬ an 􀆺     
lutionꎬ 51: 715—723
Stanton MLꎬ Snow AAꎬ Handel SN et al.ꎬ 1989. The impact of a
flower color polymorphism on mating patterns in experimental pop ̄
ulations of wild radish (Raphanus raphanistrum L.) [ J] . Evolu ̄
tionꎬ 43: 335—346
Stebbins GLꎬ 1974. Flowering Plants: Evolution Above the Species Level
[M]. Belknap Pressꎬ Cambridgeꎬ MA: Belknap Press Harvard
Streisfeld MAꎬ Kohn JRꎬ 2005. Contrasting patterns of floral and mo ̄
lecular variation across a cline in Mimulus aurantiacus [ J] . Evo ̄
lutionꎬ 59: 2548—2559
Subramaniam Bꎬ Rausher MDꎬ 2000. Balancing selection on a floral
polymorphism [J] . Evolutionꎬ 54: 691—695
Sun SGꎬ Liao Kꎬ Xia J et al.ꎬ 2005. Floral colour change in Pedicu ̄
laris monbeigiana (Orobanchaceae) [ J] . Plant Systematics and
Evolutionꎬ 255: 77—85
Tremblay RLꎬ Ackerman JDꎬ 2003. The genetic structure of orchid
populations and its evolutionary importance [J] . Lankesterianaꎬ
7: 87—92
Tremblay RLꎬ Ackerman JDꎬ 2007. Floral color patterns in a tropical
orchid: are they associated with reproductive success? [ J ] .
Plant Species Biologyꎬ 22: 95—105
Wang Sꎬ Xie Yꎬ 2004. China Species Red List 1 [M]. Beijing: High ̄
er Education Press
Waser NMꎬ Price MVꎬ 1981. Pollinator choice and stabilizing selec ̄
tion for flower color in Delphinium nelsonii [J] . Evolutionꎬ 35:
376—390
Wesselingh RAꎬ Arnold MLꎬ 2000. Pollinator behaviour and the evo ̄
lution of Louisiana iris hybrid zones [J] . Journal of Evolutionary
Biologyꎬ 13: 171—180
Wilson Pꎬ Thomson JDꎬ 1996. How do flowers diverge? [ A]. / /
Lloyd Dꎬ Barrett SCH eds.ꎬ Floral Biology [M]. New York:
Chapman & Hallꎬ 88—111
Wu CYꎬ 1986. Flora of Yunanica [M]. Tomus 4. Beijing: Science
Press
Table S1  The investigated populationsꎬ mean and stand deviations (SD) of
flowering individuals of the two color morphs
Population Location Plot number
Pink
Mean SD n1
White
Mean SD n
HDB N25°52′ꎬ E99°59′ 20 19􀆰 35 5􀆰 17 387 3􀆰 3  1􀆰 81 66
GGS N25°51′ꎬ E100°02′ 12 10􀆰 92 5􀆰 12 131 0   0   0
XHD N25°51′ꎬ E100°02′ 12 27􀆰 67 21􀆰 72 1589 0􀆰 9  1􀆰 06 9
GCQ N25°52′ꎬ E99°58′ 20 29􀆰 05 12􀆰 96 581 1􀆰 2  1􀆰 19 24
DST N25°40′ꎬ E100°06′ 12 16􀆰 17 5􀆰 31 158 1􀆰 92 1􀆰 51 23
Total 76 20􀆰 63 10􀆰 06 2846 1􀆰 46 1􀆰 31 122
1. Number of investigated plants
82                                  植 物 分 类 与 资 源 学 报                            第 37卷