全 文 ：二型花柱植物滇丁香二态和单态种群间花部形态变异模式
*
周 伟1，2，李德铢1，2，王 红1＊＊
(1 中国科学院昆明植物研究所东亚植物多样性与生物地理学重点实验室，昆明 650201;
2 中国科学院昆明植物研究所中国西南野生生物种质资源库，昆明 650201)
摘要:花部形态特征在植物交配系统的演化与维持过程中起关键作用，交配方式的转变将可能伴随着相应
花部形态的调整。为探寻交配系统与花形态变异之间的联系，本研究对二型花柱植物滇丁香 (Luculia pin-
ceana)的二态和单态种群的花部形态特征进行比较分析。结果表明花部形态在两种不同花型构造的种群
间分化程度显著大于花型构造相同的种群间变异。相对于二态种群而言，长花柱的单态种群中雌雄异位距
离缩小，且个体间雌雄生殖器官在空间上的非法重叠程度增高，花部形态的调整能够同时促进自交和同型
异交。在异型花柱物种内，异交为主的二型花柱种群向单态种群转变过程中自交比率预期上升，本研究表
明花部形态的变异模式与交配系统转变相一致。
关键词:滇丁香;花部形态变异;二型花柱;二态;单态
中图分类号:Q 944 文献标志码:A 文章编号:2095－0845(2015)05－513－09
Patterns of Floral Variation between Dimorphic and Monomorphic
Populations in Distylous Luculia pinceana (Ｒubiaceae)*
ZHOU Wei1，2，LI De-zhu1，2，WANG Hong1＊＊
(1 Key Laboratory for Plant Biodiversity and Biogeography of East Asia，Kunming Institute of Botany，Chinese
Academy of Sciences，Kunming 650201，China;2 Germplasm Bank of Wild Species in Southwest China，
Kunming Institute of Botany，Chinese Academy of Sciences，Kunming 650201，China)
Abstract:Flower morphology plays an important role in the evolution and maintenance of plant mating systems，in-
cluding disassortative mating of heterostylous species． The transition of mating patterns may be associated with the re-
modification of intraspecific flower morphology． To determine the functional relationship between floral variation and
transition of mating patterns，we conducted a series of morphometric analyses in a distylous species Luculia pin-
ceana，which possesses dimorphic and monomorphic populations． Our results indicate that floral variation was higher
between different types of populations than between populations of the same type． Compared to dimorphic popula-
tions，some floral characters，reduced stigma-anther separation within flowers and increased overlap of stigmas and
anthers (illegitimate spatial matching of sexual organs)among individuals in populations containing only the long-
styled morph may have been modified to increase both selfing and intra-morph crossing． The observed patterns of flo-
ral variation between dimorphic and monomorphic populations coincide with the transition of mating patterns from
disassortative mating to selfing and /or intra-morph crossing．
Key words:Luculia pinceana;Distyly;Floral morphological variation;Dimorphic;Monomorphic
植 物 分 类 与 资 源 学 报 2015，37 (5):513～521
Plant Diversity and Ｒesources DOI:10．7677 /ynzwyj201515011
＊
＊＊
Funding:The National Key Basic Ｒesearch Program of China (2014CB954100)，the Key Ｒesearch Program of the Chinese Academic of
Sciences (KJZD-EW-L07) ，the National Natural Science Foundation of China (31200289，31570384，31320103919) ，and the
Natural Science Foundation of Yunnan Province (2012FB182)
Author for correspondence;E-mail:wanghong@mail. kib. ac. cn
Ｒeceived date:2015－01－22，Accepted date:2015－03－25
作者简介:周伟 (1980－)男，博士，主要从事植物繁殖生态学研究。E-mail:zhouwei@mail. kib. ac. cn
Flowers may be morphologically constant within
species because pollinator mediated stabilizing selec-
tion may reduce the variation in floral traits (Steb-
bins，1970;Feinsinger，1983;Gong and Huang，
2009)． Intraspecific variation in floral morphology
may constitute an obstacle for sexual reproduction，
especially for animal-pollinated plants (Cresswell，
2000;Herrera，2001)． Nevertheless，floral morpho-
logical variation among conspecifics has been repor-
ted at different hierarchical levels，including among
flowers within individuals，among individuals，and
among populations (Cresswell，1998;Galen，1999;
Williams and Conner，2001;Herrera，2005)． Theo-
retically，variation in floral traits may correlate with
the proportion of mixed mating types (i. e． disassor-
tative mating，assortative mating，selfing，and cross-
ing)(Barrett，2003)． However，it is difficult to in-
vestigate the evolutionary relationship between floral
variation patterns and changes in mating patterns di-
rectly，since the differentiation of floral parameters
is hard to quantify and its effect on mating compo-
nents could be too weak to detect．
Heterostylous species are an ideal system to in-
vestigate the functional relationship between floral
variation and mating patterns (Hodgins and Barrett，
2006，2008)． In these species，the reciprocal ar-
rangement of sex organs is recognized as a floral syn-
drome that increases the precision of pollen transfer
between the corresponding morphs，resulting in dis-
assortative mating in floral polymorphic populations
(Lloyd and Webb，1992)． Previous studies of het-
erostylous species with morph-biased populations in-
dicate that the components of mating patterns showed
variegated variation across the geographical distribu-
tion and were strongly associated with floral morpho-
logy variation (Barrett and Hodgins，2006;Hodgins
and Barrett，2006，2008;Brys et al．，2007;Valle-
jo-Marin and Barrett，2009)． Ｒecently，monomor-
phic populations have been reported in some hetero-
stylous species (Naiki and Nagamasu，2004;Hod-
gins and Barrett，2008;Zhou et al．，2012)． In such
monomorphic populations，disassortative mating is
impossible，and therefore the sexual reproductive
mating strategy is predicted to consist of selfing and /
or intra-morph crossing． A comparative study between
polymorphic and monomorphic populations can thus
provide an opportunity to shed light on the evolutio-
nary paradigm of mating strategy transition and floral
morphological differentiation．
Luculia pinceana Hook． (Ｒubiaceae)is a geo-
graphically restricted shrub，pollinated by both long-
and short-tongued animals (Ma et al．，2009)． The
populations of the species are either distylous，con-
taining long- and short-styled morphs (hereafter L-
and S-morphs，pin and thrum flower) ，or are mono-
morphic，containing only the L-morph (i. e． L-mon-
omorphic population) ，which possess a particular
geographic distribution (Zhou et al．，2012)． In this
study，morphometric data are used to analyze the
floral differentiation between dimorphic and mono-
morphic populations of L. pinceana． Specifically，we
address the following two questions: (1)What pat-
terns of floral trait variation are existing between
monomorphic and dimorphic populations? (2) Is
variation in floral morphology correlated with the
transition of mating patterns from disassortative mat-
ing to selfing and /or intra-morph crossing between
the two types of populations?
1 Materials and methods
1. 1 Study species and populations
Luculia pinceana is a perennial shrub，restrict-
ed to southwest China and northern Vietnam and
Myanmar at altitudes between 350 and 1 800 m (Luo
et al．，1999;Zhou et al．，2012)． The species begin
flowering mid-July at its southern range margin of
Sapa (Lào Cai，Vietnam)and continue flowering
until December at higher elevations in northwest
Yunnan (China)． Each corymb has five to 12 flow-
ers (mean flower per inflorescence = 7. 8，N = 48，
SD= 2. 4)that vary in color from white to pale red
among individuals or populations． Each flower lasts
five to eight days and is fragrant． The corolla is sym-
petalous with five lobes，and five epipetalous sta-
415 植 物 分 类 与 资 源 学 报 第 37卷
mens attach to the corolla-tube． Pollinators of L. pin-
ceana are mainly long-tongued nectar feeding insects
(bumblebees，moths and butterflies) and pollen-
collecting bees (Apis florae)． In this study，we sam-
pled six populations of L. pinceana in Yunnan． Of
these populations，three were dimorphic (JCW:24°
47. 40 N，104°30. 75 E，1 798 m above sea level;
MN:22°57. 16 N，104°19. 26 E，1 418 m;LP:
25°58. 92 N，098°48. 39 E，1 763 m)and three
were long-styled monomorphic (WLL:23° 15. 76
N，103° 58. 06 E，1 780 m;PD:22° 57. 39 N，
103°42. 10 E，1 614 m;JP:22°53. 52 N，103°
13. 20 E，1 475 m)． Each population was separated
from all others by at least 50 km，with the greatest
amount of separation of nearly 400 km．
1. 2 Floral measurements
In each population，we obtained a random sam-
ple of flowering stems and classified them according
to style morph． Only fresh flowers with mature sex
organs in which styles were fully mature were sam-
pled and preserved in FAA (10% formalin，glacial
acetic acid，and 70% ethanol)for subsequent meas-
urements． For each flower，we measured stigma hei-
ght (SH)and anther height (AH)from the base of
the ovary to the midpoints of the stigma and anther，
respectively (Fig. 1)． For stigma length (SL)and
anther length (AL)，we measured from one end of
the stigma lobe or anther to their other endpoints．
Stigma-anther separation (SAS)was measured from
the tip of the anther to the base of the stigma for L-
morph flower and reverse for S-morph flower (Fig. 1)．
All measurements were made using a dissecting mi-
croscope or digital calipers．
1. 3 Statistical analysis
To quantify floral morph ratios in dimorphic
populations，we assessed whether floral morphs oc-
curred at equal frequency，as is typical of many di-
stylous species，by using goodness-of-fit G-tests． Poo-
led G-tests were calculated to determine whether
pooled morph ratios differed significantly from 1 ∶ 1．
We also calculated Gheterogeneity statistics to test for het-
erogeneous morph ratios among populations． Varia-
tion in floral morphological parameters between the
three types of flowers，one type from monomorphic
population and two types from dimorphic population，
were analyzed in a one-way ANOVA followed by a
Tukey-Kramer test． Each type exceeded 120 flowers，
which came from three populations at the same pro-
portion and more than 30 individuals were sampled
for each population． The stigma-anther separation
between monomorphic and dimorphic populations was
compared using the same method．
Fig. 1 The floral diagrams of (A)S-morph (thrum)and (B)
L-morph (pin)flowers in Luculia pinceana． SH:stigma height;
AH:anther height;AL:anther length;SL:stigma length;
SAS:stigma-anther separation
To examine potential differentiation in floral
morphology between the two types of populations，we
used the reciprocity index (Ｒ)proposed by Sanchez
et al． (2008)． Eighteen Ｒ values were calculated
(three from natural dimorphic populations，six from
dimorphic populations constructed using two floral
morphs from different natural dimorphic populations，
and the remaining nine from populations constructed
with floral morphs from both the monomorphic and
dimorphic populations)and compared． To determine
if the morphological variation between population
types was associated with the functional change and
the transition of mating patterns，the degree of recip-
rocal herkogamy was evaluated by calculating the
spatial matching of organ i (stigma or anther)to or-
5155期 ZHOU Wei et al．:Patterns of Floral Variation between Dimorphic and Monomorphic Populations …
gan j (stigma or anther)as follows (Lau and Bos-
que，2003;Kalman et al．，2007):spatial matching
(ij)=(Aij /Ai)×100，where Aij is the area of over-
lap between the distribution of organ i and organ j，
and Ai is the area of the distribution of organ i． The
length of divided part is 0. 5 mm． All analyses were
performed in Ｒ version 2. 51 (Ｒ Development Core
Team，2010)．
2 Ｒesults
Analysis of floral morphs showed that individu-
als of L. pinceana exhibited either short-styled or
long-styled flowers． The co-existence of both morphs
in the same individual was not observed in any of the
populations． Field observations revealed there were
two types of populations in L. pinceana，monomor-
phic and dimorphic． In dimorphic populations，stig-
mas and anthers were reciprocally placed in each flo-
ral morph． Distribution of anther height was bimod-
al，and the average distance between the S- and L-
morph exceeded 5 mm (Fig. 2A－C)． Alternatively，
stigma height showed a continuous distribution in
population JCW and LP，but the degree of overlap
between floral morphs within the same population
was not significant，especially in population MN
(Fig. 2A－C)． In monomorphic populations，the stig-
mas were located above the anthers in all flowers
(Fig. 2D－F)，suggesting these populations are mo-
nomorphic only for the L-morph． The floral morph
ratio was 1∶1. 11 (N=80)in population MN，1∶0. 52
(N= 70)in population JCW and 1 ∶ 1. 54 (N = 56)
in population LP． The G-tests indicated that the flo-
ral morph ratios of all dimorphic populations do not
deviate from 1 ∶ 1 (MN:G = 0. 200;P = 0. 655;
JCW:G= 3. 457;P = 0. 063;LP:G = 1. 286;P =
0. 257)． There was also no significant deviation from
a 1 ∶ 1 ratio when the data were pooled across three
populations (Gpooled = 1. 53，P＞＞0. 05)． No signifi-
cant heterogeneity was detected among population
morph ratios (Gheterogeneity = 3. 62，P＞＞0. 05)．
In dimorphic populations， significant differ-
ences were found between morphs for anther height，
stigma height，and stigma length (Fig. 3)． However，
there was no significant difference between morphs
Fig. 2 Flowers of Luculia pinceana ranked by style length to illustrate the reciprocal correspondence of stigma and anther positions
in L-and S-morphs． The flowers are sampled from two extreme of stigma height． Positions of stigma are indicated by crosses
and those of anthers by rectangles． A，B and C represent dimorphic populations JCW，LP and MN;D，E
and F represent L-monomorphic populations PD，JP and WLL
615 植 物 分 类 与 资 源 学 报 第 37卷
Fig. 3 Mean value of floral traits of three types of flowers in Luculia
pinceana． PFM represent L-morph flowers of L-monomorphic popula-
tions (WLL，PD and JP);PFD and TFD represent L- and S-morph
flowers of dimorphic population (MN，JCW and LP) ，respectively．
Columns with the same letters do not differ significantly (Turkey test，
P ＜ 0. 01)． Vertical lines show standard deviations
for anther length or between the stigma height of L-
morph and anther height of S-morph． The stigma
length in S-morph (6. 54 ± 0. 73 mm)was nearly 1. 3
times greater than that of L-morph (4. 88 ± 0. 71 mm)．
Stigma and anther lengths of the L-monomorphic
populations were statistically similar to those of the
dimorphic populations，but stigma and anther posi-
tions were all higher in the L-monomorphic popula-
tions compared to the pin flowers in the dimorphic
populations．
The indices of reciprocity (Ｒ) for dimorphic
populations were all less than 0. 025 (MN:0. 007;
LP:0. 017;JCW:0. 021)(Fig. 4)． When pin flowers
from L-monomorphic populations matched to thrum
flowers from dimorphic populations，nine putative
dimorphic populations were created，and their resul-
ting Ｒ values were all greater than 0. 1 (MN-WLL:
0. 105;MN-PD:0. 112;MN-JP:0. 100;LP-WLL:
Fig. 4 Ｒeciprocity index (Ｒ)in populations of Luculia pinceana． The values were calculated from three natural dimorphic
populations and fifteen constructed populations using two floral morphs from different populations of
monomorphic or dimorphic (see Materials and Methods for details)
7155期 ZHOU Wei et al．:Patterns of Floral Variation between Dimorphic and Monomorphic Populations …
0. 107;LP-PD:0. 115;LP-JP:0. 105;JCW-WLL:
0. 121;JCW-PD:0. 128;JCW-JP:0. 116)(Fig. 4)．
Alternatively，the Ｒ values of another six putative
dimorphic populations，which were composed of L-
morph from one dimorphic population and S-morph
flowers from another two dimorphic populations，were
all less than 0. 025 (MN-JCW:0. 010;MN-LP:
0. 024;LP-JCW:0. 015;LP-MN:0. 023;JCW-LP:
0. 012;JCW-MN:0. 013)． These results indicate that
L-morph flowers in the monomorphic populations
were significantly differentiated from those of the di-
morphic populations，but there was no significant di-
fferentiation between the dimorphic populations．
In dimorphic populations，there was higher spa-
tial matching between corresponding legitimate sexual
organs than between illegitimate ones (Fig. 5A－D)．
Overlap between S-morph stigmas and their corre-
sponding legitimate organs，i. e． L-morph anthers，was
nearly five times greater than their overlap with their
illegitimate organs，i. e． S-morph anthers (55. 22%
and 11. 38%，respectively)． Overlap between L-mor-
ph stigmas and their corresponding legitimate organs，
i. e． S-morph anthers，was nearly seven times greater
than their overlap with L-morph anthers (73. 25%
and 11. 99%，respectively)． Similarly，pin anthers
matching to S-morph stigmas (100%)was almost
eight times greater than their matching to L-morph
stigmas (12. 02%) ，and the matching of S-morph
anthers to legitimate L-morph stigmas (71. 28%)
was over three times their matching to illegitimate
S-morph stigmas (20. 32%)． In monomorphic popu-
lations，illegitimate spatial matching of sexual organs
(Fig. 5E － F)was greater than that in dimorphic
populations． The matching of stigmas to anthers was
Fig. 5 Special matching of the sexual organs of Luculia pinceana flowers． Abscissas show the distance from the distal margin of the ovary taken
as arbitrary reference point． The ordinates indicate the percentage of flowers in which at least part of their reproductive organs were present in
each of 0. 5 mm segments along the abscissa． Unfilled triangles pin stigmas;filled triangles pin anthers;unfilled circles thrum stigmas;filled
circles thrum anthers． A－D represents dimorphic population JCW;E and F represent pin-monomorphic population JP and PD，respectively
815 植 物 分 类 与 资 源 学 报 第 37卷
22. 96% and 31. 13% in populations JP and PD，re-
spectively，which was two to three times greater than
population JCW． In contrast，the matching of anthers
to stigmas in the same monomorphic populations was
31. 19% and 33. 66%，which were three times grea-
ter than that of the dimorphic population JCW． The
distribution of stigmas and anthers of the dimorphic
population type was more concentrated compared to
that of the monomorphic population type．
In L-monomorphic populations WLL，PD，and
JP，the means of stigma-anther separation were 2. 53
± 0. 98 mm(N=56)，2. 76 ± 0. 77 mm(N=60) ，and
2. 88 ± 0. 72 mm(N=63) ，respectively． For pin flow-
ers of dimorphic populations MN，JCW，and LP，the
means of stigma-anther separation were 3. 85 ± 1. 13
mm(N=60) ，4. 03 ± 1. 01 mm(N=60) ，and 4. 13 ±
1. 18 mm (N=60) ，respectively． There were no sig-
nificant difference within each type of population in
anther-stigma separation，but anther-stigma separa-
tion was significantly different between the two types
of populations．
3 Discussion
In heterostylous species，sex-organ reciprocity
serves to promote cross-pollen transfer between floral
morphs and the floral polymorphism maintained by
negative frequency-dependent selection (Charles-
worth and Charlesworth，1979;Lloyd and Webb，
1992)． In L. pinceana，the stigma and anthers never
occupied the same position within a given flower and
the homostyly individual was not present in this and
previous survey (Zhou et al．，2012)，thus the spe-
cies was convinced to be a true distylous species．
Quantified reciprocal herkogamy described by legiti-
mate and illegitimate matching of sexual organs fur-
ther confirmed the functional syndrome of distyly in
this species．
Sanchez et al． (2008)proposed the reciprocity
index (Ｒ value)to evaluate the matching fidelity of
stigma and anther between morphs，which can re-
flect the role of floral traits on promoting disassorta-
tive mating． The reciprocity indices in three natural
dimorphic populations were all less than 0. 025
(0. 007～ 0. 021) ，which were similar to the values
resulting from matching L-and S-morph flowers from
different dimorphic populations (0. 010 ～ 0. 024)．
These results indicate that these floral morphological
characters are very efficient in promoting inter-morph
pollination and that there was no significant differen-
tiation in floral morphology between dimorphic popu-
lations． However，the Ｒ values generated from the
two types of populations all exceed 0. 1 (0. 100 ～
0. 128) (Fig. 4) ，and the variability was similar to
the Ｒ values generated by matching different dimor-
phic populations． As suggested by Sanchez et al．
(2008) ，an Ｒ value of 0. 1 or greater represents no
functional significance considering with heterostyly．
Thus，we suggest that floral morphology has experi-
enced significant differentiation between dimorphic
and monomorphic populations，but that it is stable
within the same type of population．
Previous studies have proposed that reduced
stigma-anther separation may increase the degree of
self-fertilization (Barrett and Shore，1987;Motten
and Stone，2000)and provide reproductive assur-
ance (Mal and Lovett-Doust，2005)in heterostylous
species． In addition，the increased illegitimate spa-
tial matching of sexual organs at the population level
suggests that more intra-morph crossing could be
possible (Lau and Bosque，2003;Kalman et al．，
2007)． For L. pinceana，the reciprocity values indi-
cate that disassortative mating was a reliable mating
system in natural dimorphic populations，this pattern
was confirmed in a natural population of the species
using microsatellite markers and paternity analysis
(Zhou et al．，2015)． However，the mating compo-
nent characterized by prevailing disassortative mating
in the species may shift to selfing and /or intra-mor-
ph crossing in L-monomorphic populations since the
absence of the inter-morph mating pair (S-morph)．
Such a mating pattern transition should be accompa-
nied with modifications of floral morphology． Our re-
sults confirmed this prediction:first，the stigma-an-
ther separations were significantly reduced in all
9155期 ZHOU Wei et al．:Patterns of Floral Variation between Dimorphic and Monomorphic Populations …
three L-monomorphic populations compared to that
in the dimorphic populations，and，second，the de-
gree of illegitimate spatial matching of sexual organs
in L-monomorphic populations was significantly high-
er than that in dimorphic populations． These results
indicate that the floral syndrome of disassortative
mating was breakdown and that both selfing and in-
tra-morph crossing were facilitated by modifications
to floral morphology in pin-monomorphic population．
Future studies of the pollen loads deposited by parti-
cular pollinator would be required to determine the
mating components of selfing and crossing in L-mon-
omorphic populations． The shift in mating patterns
from outcrossing to selfing accompanied by a reduc-
tion in herkogamy was also reported in the mid-
styled morph of Eichhornia paniculata (Vallejo-Ma-
rin and Barrett，2009);however，our results also
indicated that intra-morph crossing was increased in
L. pinaeana． This difference suggests that，besides
selfing，animal-mediated pollination may also have
an important effect on the reproduction of individuals
in monomorphic populations． Sakai and Wright (2008)
have reported that heterostylous and monomorphic
Psychotria species were visited by the same bee spe-
cies，but that visit frequencies were lower in the
monomorphic species． Although we could not dis-
criminate the proportion of the two mating types re-
spectively，the similar pollinator spectrum (com-
prised Apis florae，moths and butterflies)were ob-
served in both the monomorphic and dimorphic po-
pulations of L. pinceana． In L. pinceana，the central
monomorphic and the eastern dimorphic area share a
similar latitude compared with the latitude of the
western dimorphic area． Considering the similar cli-
matic conditions and the contiguous distribution of
the central and eastern populations，it is therefore
quite possible that the monomorphic populations
were more likely to be connected with the eastern di-
morphic populations than western populations． How-
ever，this association was not compatible with the
previous molecular phylogeographic analysis which
inferred that the centrally located lineage composed
of long-styled morph-biased and monomorphic popu-
lations were derived from the western dimorphic line-
age by founder event after the last glacial maximum
(Zhou et al．，2012)．
Floral morphology plays an important role in the
evolution and maintenance of many mating systems，
including the disassortative mating of heterostylous
species． Ｒesults from our comparative study of floral
morphology between dimorphic and monomorphic po-
pulations of L. pinceana indicate that patterns of vari-
ation in floral morphology coincide with the transition
of mating patterns． These results demonstrate that
different mating patterns could select for different
floral morphologies，and that reciprocal herkogamy
was not the only means to increase the proficiency of
cross-pollination in distylous species (Lloyd and
Webb，1992)．
Acknowledgements:The authors thank Zong-Xin Ｒen，Hai-
Dong Li，Yan-Hui Zhao，Ming-Ying Zhang，Jie Liu，Li-Na
Dong，and Shu Zhang (Kunming Institute of Botany，Chinese
Academy of Sciences)for help with fieldwork．
Ｒeferences:
Barrett SCH，2003． Mating strategies in plants:the outcrossing-selfing
paradigm and beyond ［J］． Philosophical Transactions of the Ｒoy-
al Society of London，358:991—1004
Barrett SCH，Hodgins KA，2006． Floral design and the evolution of a-
symmetrical mating systems ［A］/ / Harder LD，Barrett SCH
eds．，Ecology and Evolution of Flowers ［M］． Oxford:Oxford U-
niversity Press，239—254
Barrett SCH，Shore JS，1987． Variation and evolution of breeding sys-
tems in the Turnera ulmifolia L． complex (Turneraceae)［J］． E-
volution，41:340—354
Brys Ｒ，Jacquemyn H，Hermy M，2007． Impact of mate availability，
population size，and spatial aggregation of morphs on sexual re-
production in a distylous，aquatic plant ［J］． American Journal
of Botany，94:119—127
Charlesworth D，Charlesworth B，1979． A model for the evolution of
distyly ［J］． The American Naturalist，114:467—498
Cresswell JE，1998． Stabilizing selection and the structural variability
of flowers within species ［J］． Annals of Botany，81:463—473
Cresswell JE，2000． Manipulation of female architecture in flowers re-
veals a narrow optimum for pollen deposition ［J］． Ecology，81:
3244—3249
Feinsinger P，1983． Coevolution and pollination ［A］/ / Futuyma
025 植 物 分 类 与 资 源 学 报 第 37卷
DJ，Slatkin M eds．，Coevolution ［M］． Sunderland:Sinauer，
282—310
Galen C，1999． Why do flowers vary?The functional ecology of varia-
tion in flower size and form within natural plant populations ［J］．
Bioscience，49:631—640
Gong YB，Huang SQ，2009． Floral symmetry:pollinator-mediated
stabilizing selection on flower size in bilateral species ［J］． Pro-
ceedings of the Ｒoyal Society B:Biological Sciences，276:
4013—4020
Herrera J，2001． The variability of organs differentially involved in
pollination，and correlations of traits in Genisteae (Leguminosae:
Papilionoideae)［J］． Annals of Botany，88:1027—1037
Herrera J，2005． Flower size variation in Ｒosmarinus officinalis:indi-
viduals，populations and habitats ［J］． Annals of Botany，95:
431—437
Hodgins KA，Barrett SCH，2006． Mating patterns and demography in
the tristylous daffodil Narcissus triandrus ［J］． Heredity，96:
262—270
Hodgins KA，Barrett SCH，2008． Asymmetrical mating patterns and
the evolution of biased style-morph ratios in a tristylous daffodil
［J］． Genetics Ｒesearch，90:3—15
Kalman K，Medvegy A，Penzes Z et al．，2007． Morph-specific varia-
tion of floral traits associated with reciprocal herkogamy in natural
populations of Primula vulgaris and Primula veris ［J］． Plant
Systematics and Evolotion，268:15—27
Lau P，Bosque C，2003． Pollen flow in the distylous Palicourea
fendleri (Ｒubiaceae) :an experimental test of the disassortative
pollen flow hypothesis ［J］． Oecologia，135:593—600
Lloyd DG，Webb CJ，1992． The selection of heterostyly ［A］/ / Bar-
rett SCH ed．，Evolution and Function of Heterostyly ［M］． Ber-
lin:Springer-Verlag，179—207
Luo XＲ，Gao WZ，Chen WQ et al．，1999． Ｒubiaceae ［A］/ / Wu
ZY ed．，Flora Ｒepublicae Popularis Sinicae ［M］． Beijing:Sci-
ence Press
Ma H，Wang Y，Li ZH et al．，2009． A study on the breeding system
of Luculia pinceana ［J］． Forest Ｒesearch，22:373—378
Mal TK，Lovett-Doust J，2005． Phenotypic plasticity in vegetative and
reproductive traits in an invasive weed，Lythrum salicaria (Lyth-
raceae) ，in response to soil moisture ［J］． American Journal of
Botany，92:819—825
Motten AF，Stone JL，2000． Heritability of stigma position and the
effect of stigma-anther separation on outcrossing in a predominant-
ly self-fertilizing weed，Datura stramonium (Solanaceae) ［J］．
American Journal of Botany，87:339—347
Naiki A，Nagamasu H，2004． Correlation between distyly and ploidy
level in Damnacanthus (Ｒubiaceae) ［J］． American Journal of
Botany，91:664—671
Ｒ Development Core Team，2008． Ｒ:A Language and Environment
for Statistical Computing ［OL］． Available from:http:/ /www．r-
project．org．
Sakai S，Wright SJ，2008． Ｒeproductive ecology of 21 coexisting Psy-
chotria species (Ｒubiaceae) :when is heterostyly lost ［J］． Bio-
logical Journal of the Linnean Society，93:125—134
Sanchez JM，Ferrero V，Navarro L，2008． A new approach to the
quantification of degree of reciprocity in distylous (sensu lato)
plant populations ［J］． Annals of Botany，102:463—472
Stebbins GL，1970． Adaptive radiation of reproductive characteristics
in angiosperms，I:pollination mechanisms ［J］． Annual Ｒeview
of Ecology and Systematics，1:307—326
Vallejo-Marin M，Barrett SCH，2009． Modification of flower architec-
ture during early stages in the evolution of self-fertilization ［J］．
Annals of Botany，103:951—962
Williams JL，Conner JK，2001． Sources of phenotypic variation in flo-
ral traits in wild radish，Ｒaphanus raphanistrum (Brassicaceae)
［J］． American Journal of Botany，88:1577—1581
Zhou W，Barrett SCH，Wang H et al．，2012． Loss of floral polymor-
phism in heterostylous species Luculia pinceana (Ｒubiaceae) :a
molecular phylogeographic perspective ［J］． Molecular Ecology，
21:4631—4645
Zhou W，Barrett SCH，Wang H et al．，2015． Ｒeciprocal herkogamy
promotes disassortative mating in a distylous species with in-
tramorph compatibility ［J］． New Phytologist，206:1503—1512
1255期 ZHOU Wei et al．:Patterns of Floral Variation between Dimorphic and Monomorphic Populations …