全 文 ：Infection of the Anther_smut Microbotryum violaceum in Populations of
Silene dioica:Variation in Floral Morphology , Patterns of Spore
Deposition and Pathogen_Pollinator Mediated Selection
LIU Deng_Yi1 ＊ , Ulla CARLSSON2
(1.Biodiversity Research Center , Anhui Normal University , Wuhu 241000 , China;
2.Department of Ecology and Environmental Science , University of Umeá , S_901 87 Umeá , Sweden)
Abstract:　The anther_smut Microbotryum violaceum (Brandenburger and Schvinn)G.Deml.and Oberw.)
causes a systemic infection of its host Silene dioica (L.)Clairv., resulting in sterility and production of telios-
pores(dispersal propagules)in flowers.These spores are transmitted to healthy plants mainly by flower visi-
tors.The behavioral responses of flower visitors to a variation in floral characters , are not only likely to affect
rates of pollen export import , but also the rate of spore deposition and probability of disease.In a transplanta-
tion experiment , using plants from four different populations , we tested for correlation between variation in fe-
male floral morphology and patterns of spore and pollen deposition , and a resulting risk of disease.The source
populations in this experiment were located on four islands in Skeppsvik archipelago in northern Sweden , and
represented a gradient of disease incidence from completely healthy (Island 1), low incidence (Island 2)to
high incidences(Islands 3 and 4)of disease.Fifty plants from each population were transplanted to the center
of the population on Island 4.There were significant differences among the transplants in floral characters , i.
e.corolla size , style length and ovule number.Plants from the non_diseased population had larger flowers and
longer styles than plants from the highly diseased populations.Numbers of pollen grains and spores deposited
on flowers were strongly and positively correlated.We found that plants originating from the non_diseased pop-
ulation captured approximately 4 times more pollen and 9 times more spores per flower than the individuals from
the resident population(Island 4 , population 4).The incidences of disease among plants from the four popula-
tions differed significantly , and was 37%, 20%, 18% and 0 for populations 1 , 2 , 3 and 4 respectively.In
a survey of ten populations we found a significant negative correlation between the mean style length(positively
correlated with corolla size and ovule number)among healthy plants and incidence of disease in these popula-
tions.The potentiality for pathogen_pollinator mediated selection on floral characters and consequences for gene
flow between populations of Silene dioica are discussed.
Key words:　Silene dioica;floral morphology;anther_smut infection;patterns of spore deposition;pathogen_
pollinator mediated selection
　　The dynamics of disease transmission in plant patho-
gen systems are often poorly documented , but are crucial
for understanding host_pathogen interactions[ 1-6] .In order
to predict the patterns of disease transmission in a particu-
lar host population , we first need information about how
inoculum sources are distributed in time and space.For
air_borne diseases , transmission is often assumed to be a
simple function of the number of healthy , susceptible in-
dividuals , the number of diseased individuals and a trans-
mission coefficient
[ 7-10] .However , for vector_borne fun-
gal diseases , e.g.where flower visitors transmit spores ,
transmission is a function of the host , pathogen and vector
populations
[ 11-14] .Variations in both abundance and be-
haviour of the vectors are likely to influence transmission
efficiencies.
Effects of the pollinator transmitted anther_smut dis-
ease , Microbotryum violaceum , on host phenology , floral
morphology and flower visitation rates have received con-
siderable attention
[ 15-20] .However , only limited informa-
tion is available about the patterns of spore deposition on
recipient healthy plants in natural populations
[ 9 , 21-23] .
Successful infections through floral inoculations have been
demonstrated repeatedly
[ 12 , 13, 15 ,24 , 25] .Infection is systemic
and leads to completely sterile flowers , where normal re-
productive parts are replaced by stamens filled with te-
liospores
[ 15] .Flower visitors , such as bumble bees and
several lepidopteran species , have often been found to
carry Microbotryum spores and pollen of host species si-
c multaneously
[ 17 , 21 , 26] .The behavioral responses by flower
visitors to a variation in floral characters are likely to af-
fect not only the rates of pollen export import , but also
the rate of spore deposition.For example , plants with
many flowers may be visited preferentially by flower visi-
tors , and thus experience a higher risk of becoming in-
fected compared with plants with smaller floral displa-
ys
[ 22] .Since the infection sterilises the plants , and
is systemic and perennial , infected plants are effectively
removed from the gene pool.In this host_pathogen
Received:2001-03-29　Accepted:2001-07-17
Supported by The Swedish Natural Science Research Counci l.
＊Author for correspondence.
植　物　学　报　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　
Acta Botanica Sinica　2002 , 44(1):88-96
system , there is therefore a considerable potential for a
pathogen pollinator mediated selection on floral charac-
ters.However , there is virtually no information available
about how variation in floral characters such as corolla
size , nectar production , stigmatic surface that influence
spore deposition patterns.Among females , style and stig-
ma morphology are of particular interest , since stigmas are
the most likely site for the pathogen to penetrate the
host
[ 15 ,27] .
In this study , we measured variation in a number of
floral traits in 14 populations of Silene dioica , with differ-
ent disease incidence.In a transplantation experiment ,
we used plants from four different populations , and tested
for correlation between variation in female floral morpholo-
gy and patterns of spore and pollen deposition , and the
resulting risk of becoming diseased.The following ques-
tions were addressed:1)Are the patterns of deposition of
spores and pollen on female stigmas strongly correlated ?
2)How is floral morphology related to spore deposition ?
And 3)how is spore deposition related to the development
of disease?
1　Materials and Methods
1.1　Plant species
Silene dioica (L.)Clairv.is a dioecious perennial
herbaceous plant widespread in northern Fennoscandia.It
is abundant in the deciduous phase of primary succession
along the shores of the Gulf of Bothnia and in meadows ,
but also in more fertile deciduous and coniferous forests.
Flowering occurs in early June , and males begin flowering
a few days before females.The main pollinators of S.di-
oica in this area are the bumble bees , Bombus hortorum ,
B .Lucorum , B.hypnorum , and B .pascuorum ssp.
sparreanus.In this study , we did not attempt to quantify
activity or relative abundance of different bumble bee spe-
cies.
S .dioica has a great variation in flower morpholo-
gy , particularly among females.Corolla sizes (diameter
range 14-34 mm), the lengths of styles(range 3-14
mm)and the morphology of the styles vary (styles are
sometimes referred to as stigmas or stigmatic lobes
[ 15] ,
however , we will use the term style , even if this includes
a variable area of receptive stigmatic tissue).In individu-
als with long styles , these can be curly or straight , short
styles can be club_formed or thin.The long styles(11-
14 mm)usually protrude out of the floral tube , while
short ones (3-8 mm)do not.Furthermore , the styles
vary in papillation , both in the density and length of indi-
vidual papilla.Although floral infection may be the major
pathway , seedlings growing adjacent to diseased plants
can also become infected
[ 15 , 24, 25] .
Infection is easily detected among flowering plants by
the presence of purple_brown teliospores in flowers.
Among vegetative plants , no macroscopically unambiguous
method is available.
1.2　Study sites
We studied plants from the Skeppsvik archipelago at
the River Savaran on the Gulf of Bothnia (in Savar par-
ish , Vasterbotten Province , Sweden 63°44′-48′N , 20°
31′-33′E)[ 18 , 28] .This archipelago is subject to land
uplift and primary succession is in strikingly different
phases on different islands.Populations of S .dioica are
present on most of the approximately 100 islands in the
archipelago , the exceptions being the very young and
small islands.Approximately 70% of the populations are
infected by Microbotryum violaceum
[ 18] .The estimation of
ages of islands and populations shown in Table 1 is based
on data given by Carlsson et al
[ 28] .
1.3　Field survey
Four populations of S .dioica were chosen for de-
tailed field measurement in June 1998.These populations
were located on four islands in the archipelago and repre-
sent a gradient of disease incidence from completely
healthy (Island 1), low incidence(Island 2)to high in-
cidence(Islands 3 and 4).Characteristics of these popu-
lations are given in Table 1.Between 8 to 12 June(peak
flowering)each population was visited , and 30 healthy
female plants were chosen at random from each popula-
tion.Each female plant was tagged , and style length and
corolla diameter of open flowers were measured.Styles
from one flower with corolla just closing were sampled
from each individual and placed in a vial , and one cap-
sule on each individual was marked.We carefully chose
to sample and mark flowers and capsules that were in the
same developmental stage.
Table 1　Characteristics of the four island populations of Silene di-
oica
Island 1 2 3 4
Population size(nos.) 4 000 4 000 16 000 4 900
Population age(years) 31±15 39±14 96±15 111±15
Density
(individual m2) 53.5±8.0 56.5±5.2 34.6±4.5 18.9±2.6
Disease (%) 0 9.2±4.4 39.6±9.2 59.4±5.2
Flower number＊＊＊ 9.7±0.2 4.8±0.6 3.6±0.8 3.9±0.5
Corolla 22.3±0.5 21.0±0.5 20.2±0.5 19.5±0.5
diameter (mm)＊＊＊
Style length(mm)＊＊＊ 9.8±0.2 8.7±0.3 8.2±0.3 7.4±0.3
Seeds caps＊＊ 213±8.1 192±8.3 139±22.4 141.1±9.8
＊＊, P<0.01;＊＊＊, P<0.001(ANOVA).For estimating population
size , density and disease incidence , n=5m2 plots , for plant characters n=
30 , from each population.Mean±SE given.
The styles were immediately stored in a freezer.In
order to count the number of spores and pollen present on
the style accurately , we first treated the samples by ace-
tolysis.The styles were transferred to a test tube , 2mL of
Hac were added , and the sample was washed and centri-
fuged.After decantation , 4.5 mL of acetaldehyde and 0.
5 mL concentrated sulphuric acid were added , and the
test tubes were put in water and heated for 2 h.When the
colour turned dark brown , the oxidization process was
complete.After decantation , the samples were washed
with Hac and distilled water.Since pollen and spores are
not affected by the acetolysis , we were able to obtain a
pure sample containing only spores and pollen.After ace-
tolysis , 1 mL of distilled waterwas added to each sample ,
LIU Deng_Yi et al:Infection of the Anther_smut Microbotryum violaceum in Populations of Silene dioica 89　
which was then agitated on a vortex for 1 min before
counting.Spores and pollen were counted using a haema-
cytometer and the counts per chamber were translated to
counts per millilitre and then to number of spores or pol-
len per sample.From each sample , four different counts
were made and the mean of these counts was used in the
statistical analyses.Spore and pollen number were log_
transformed in the statistical analyses to increase the ho-
mogeneity of variance.
Mature capsules were collected between 9 July and
23 July.We collected all capsules , noting herbivore dam-
age and flowers that failed to develop mature capsules.
Fruit set was invariably high in all populations(>95%), and therefore we only considered seed set and
seed production in the statistical analyses.From each in-
dividual , seeds were counted from a maximum of six cap-
sules.
During peak flowering in 1999 , the length of styles(mm)and corolla diameter were measured on females(n=40-60)in 10 different populations of S .dioica in
the Skeppsvik archipelago.In the same populations , we
also measured corolla diameter and noted flower numbers
on male plants(n =30).
1.4　Field experiment
On 14-18 June 1999 , 50 female plants from each
of the four islands were transplanted to Island 4 , in a
completely randomised block design.The plants (n =
200)were carefully excavated and planted in 25 randomly
positioned experimental stations (1 station =1 block)(grid size=10 m×10 m)in the center of the population
on Island 4.Only females with one flowering shoot were
chosen to minimize size variation.These plants carried a
variable number of flower buds , but none had fully
opened flowers at the start of the experiment.Two females
from each population were planted at each station(total
number of plants per station=8)in a circular plot;each
plant was 16 cm away from its nearest neighbour , and 20
cm from the center of the plot.One female in each pair
was randomly allocated to serve as a control.From the
other female , flowers were sampled at different intervals
during the flowering season.Each plant was randomly po-
sitioned in the circle plot.Each female had 1-3 expand-
ed flower buds(the tip of the corolla visible)at the time
of planting.The first flowers started to open on 18 June
and on 20 June all plants had at least one flower fully
opened and exposed to pollinators (mean±SE=1.6±
0.2 , range 1-3 , n=200).Flower size and style length
were measured for all individuals on 20 June.No differ-
ences between control plants and plants used for sampling
were detected in analyses of these variables (ANOVA ,
corolla diameter F1 , 198 =0.16 , P =0.69 , and style
length F1 ,198=0.01 , P =0.91).
Numbers of flowers and capsules on each plant were
counted four times during the experimental period of 14
June-16 July.The numbers of open flowers decreased
during the season;by 26 June an average of 0.4(±0.04), and by 2 July , 0.2(±0.08), flowers per
plant were open.At none of the censused times were
there any significant differences between plants from the
four different populations in numbers of opened flowers
per plant(ANOVA , P >0.05).From the sampling pla-
nts , flowers just closing(1 per individual.)were collect-
ed in vials , on 24 June and 28 June corresponding to
flowers open during peak and late flowering , respectively.
The styles were stored in a freezer and the ovules per
flower were counted.Mature capsules were collected re-
peatedly during July as they gradually matured.The num-
bers of spores and pollen on collected styles were counted
using the method described above.
At each station , the number of healthy flowers(male
and female)and diseased flowers were counted within a 1
m radius(=3.14m2)from the center of the station.The
proportion of diseased flowers varied in these plots , with
an average of 73.4%(±4.1), range 0-92%, n =
25.
1.5　Nectar sampling
Nectar was sampled using 5 μL microcaps (Drum-
mond Science Company)from 60 female plants on Island
4 on 25 June.Newly opened flowers were sampled early
in the morning (7:00-8:30).On each female plant , 1-3 flowers were probed and nectar volume , style length
and corolla diameter measured.
1.6　Temperature treatment
Production of healthy flowers could be induced in
diseased plants by increasing the temperature to above
27 ℃ during initiation and early growth of flowering
shoots
[ 22] .We took advantage of this to see whether par-
ticular female morphologies were over_represented among
infected individuals.Fifteen naturally diseased females
were carefully dug up from Island 3 and 4 , and planted in
pots in the greenhouse on 11 July 1998.On 9November ,
they were transferred to a growth chamber where tempera-
ture was programmed to be in the range of 27 ℃ to 30℃.The light regime was 20 h light and 4 h dark.Five
healthy females were included as controls.However , only
14 of the 30 diseased plants and none of the control plants
came into flower.We measured the style length for all the
flowering plants.
1.7　Greenhouse experiment
The 100 control female plants used in the field ex-
periment were carefully excavated and transferred into a
greenhouse on 7October 1999.They were planted in pots
with fertilized soil and stored outdoors until 7 November ,
when they were placed in a heated greenhouse (15 ℃ to
19 ℃)under long day conditions(15 h light , 9 h dark).
All flowering plants were checked for the presence of dis-
ease.On healthy plants from each population , the length
of papillae on styles were measured under a dissecting mi-
croscope.
2　Results
2.1　Field measurements_variation in floral morphology
One_way ANOVA revealed significant differences in
corolla diameters and in style lengths between female pla-
90　 植物学报　Acta Botanica Sinica　Vol.44　No.1　2002
nts on the four islands(Table 1)(plants from Islands 1
and 4 were significantly different at P <0.05 , Newman_
Keuls test).Females from the different islands also dif-
fered significantly in seed production per capsule and in
flower number (Table 1).Style length was significantly
correlated with corolla diameter(r=0.53 , P<0.01 , n=118).
Among plants on Island 1 , none of the flowers had
spores , while 53%, 98%, and 100% of the female
flowers had spores present at peak flowering on Islands 2 ,
3 and 4 respectively.Average number of pollen grains per
flower differed among the four island populations(ANO-
VA , F3 , 106 =5.4 , P <0.002), as did numbers of
spores per flower between plants from Islands 2 , 3 , and 4(ANOVA , F 2 ,78 =20.4 , P <0.001)(Fig 1).Among
plants on Islands 3 and 4 , numbers of pollen grains and
spores per flower were significantly correlated(r =0.42
and 0.43 respectively , n=30 , P<0.05), while there
was no correlation on Island 2(r=-0.03 , P<0.05).
Fig.1.　The numbers of pollen and spores per flower in female pla-
nts of Silene dioica from four different islands with different inci-
dence of disease.
Mean±SE given , n=30.Samples taken 26 June 1998.
2.2　Field survey
There was a significant difference in style lengths(ANOVA , F 9, 504=17.4 , P <0.000)and corolla diam-
eters(F9 , 504 =14.1 , P <0.000)among females from
different populations.Mean style lengths(and corolla di-
ameters , since these variables were highly correlated (P<0.01)in eight of the ten populations)tended to de-
crease as the estimated size of the source populations and
incidence of disease increased(Fig.2).A partial correla-
tion analysis , controlled for population age[ 28] , showed
that disease incidence explained most of the variation in
mean style lengths (disease , r=-0.749 , P =0.003 ,
r
2 = 0.56;population size (log_transformed) r =
-0.495 , P =0.1 , r 2 =0.24).No significant differ-
ences were detected between populations in the corolla di-
ameter of male flowers (ANOVA , F8 ,261 =1.3 , P =
0.27), and no detectable trends were found in relation to
disease incidence(r=0.01 , P >0.05 , n=10).
2.3　Field experiment
In the tansplantation experiment of 1999 , we first
tested whether there were spatial patterns (block effects)
among the 25 stations.No significant differences were
found between stations in style length , corolla diameter or
in pollen or spore numbers per flower in the ANOVA.For
numbers of pollen grains there was a marginal block effect(P=0.06), which indicates a spatial variation in pollen
abundance.
Fig.2.　The mean style length of female Silene dioica from 14 dif-
ferent populations in the Skeppsvik archipelago in relation to popula-
tion size(a)and incidence of disease(b).
Samples taken 20 June-6 July 1999.Mean±SE given , n=40-
60.
Both the style length and corolla diameter differed ,
however , between plants from the four populations(F 3 ,196=17.1 , P<0.001;F3 , 196=16 , P <0.001 respective-
ly , Table 2 and Fig.3), and in the same directions as
found in the 1998 study.Plants from the four populations
also differed in numbers of ovules per flower (F3 , 96 =
11.3 , P <0.001)but not in total number of flowers per
individual(F 3 ,81=1.4 , P=0.2)(Table 2).We found
small but significant correlation between style length and
corolla diameter(r =0.42 , P <0.01 , n =200)and
ovule numbers(r=0.27 , P<0.01 , n =180).Plants
from the four populations differed in numbers of spores
and pollen grains per flower (Table 3), and there was a
large effect of sampling time on the numbers of pollen
grains(Table 3 , Fig.4).When studying the frequency
distribution of spore and pollen numbers per flower , there
was a clear difference between plants from populations 1
and 2 , and plants from the other two populations(Fig.
5a , 5b).　　To test whether the higher number of spores in flow-
ers with long style was an effect of a larger physical area
available for pollen and spore deposition , we performed
an ANCOVA with style length as a covariate.These
LIU Deng_Yi et al:Infection of the Anther_smut Microbotryum violaceum in Populations of Silene dioica 91　
Table 2　Characteristics of plants in the transplantation experiment
Population 1 Population 2 Population 3 Population 4
Flower number individualn.s. 2.0±0.2a 1.9±0.2a 1.5±0.2a 1.6±0.2a
Ovule number flower＊＊＊ 263.2±5a 240.3±4a 197.3±5b 207.8±4b
Corolla diameter (mm)＊＊＊ 24.2±0.4a 20.8±4bc 21.9±0.4b 20.2±0.4c
Style length(mm)＊＊＊ 9.6±0.3a 8.0±2b 7.7±0.3b 7.6±0.2b
Length of papi llae(mm)＊＊＊ 0.28±0.01a 0.23±0.01b 0.20±0.01b 0.21±0.01b
Seeds capsule ＊ 137±11a 118±10a 95±11a , b 92±11b
Seed set(%)n.s. 52.2±3.9a 48.3±4.2a 47.7±4.6a 44.5±5.2a
Seed number individual ＊ 260±26a 206±24a 150±27b 134±26b
Pollen number flower＊＊＊1 5 346±396a 3 916±400b 2 389±448c 1 444±405c
Spore number flower＊＊＊1 27 313±233a 16 398±2 357b 9 370±2 635c 3 077±2 384c
ANOVA on pooled sample f rom both sample periods.Mean±S.E.given , n , 50 from each population.Different letters represent a statistically difference at P
<0.05:＊, P<0.05;＊＊＊, P<0.00;n.s., P>0.05 in one_way ANOVA.
Table 3　ANOVAs for number of pollen grains per flower(a)and
number of spores per flower(b) on female Silene dioica from four
different populations in a transplantation experiment
Number of pol len per
flower(a)
Number of spores
per flower (b)
F_ratio P F_ratio P
Population 24.6 <0.000 61.2 <0.000
Time 187.8 <0.000 5.6 <0.020
Population×Time 1.8 n.s. 5.6 <0.001
Analyses performed on log_transformed data.DF=3 , 159(Population), 1 ,
159(Time)and 3 ,159(Population×Time).Flower samples taken 24 June
and 28 June 1999.
analyses showed that there was still a significant effect of
population , even when the effect of variation in style
length was removed (F3 , 92 =19.5 , P <0.000 1 (pol-
len)and F 3, 93=35.2 , P<0.000 1(spores)).
There was an overall correlation between spore and
pollen number per flower (r=0.73 , P <0.01 ,
n =130).In a stepwise regression model involving the
floral characteristics measured , ovule number explained
most of the variation in both spore number(r=0.53 , P<0.000 , r2 =28), and pollen number per flower(r=
0.45 , P<0.000 , r2=30).
Nectar volume(μL)per flower was significantly cor-
related with style length (r=0.44 , P <0.01 , n=60)
and corolla diameter(r=0.4 , P <0.01 , n =60).
2.4　Temperature treatment
Healthy flowers developed in the diseased plants
when kept under conditions of high temperature.Howev-
er , only 50% of the plants flowered.Among these , the
style length was on average 37% greater (t_test , P <
0.005)than that among healthy plants from the same dis-
eased populations (Fig.6).Unfortunately , our control
plants(healthy plants exposed to the same treatment)did
not flower during this experiment.Although the healthy
flowers from these diseased plants looked perfectly nor-
mal , and when pollinated , developed fruits with viable
seeds , we cannot rule out an effect of the temperature it-
self on flower morphology.
2.5　Greenhouse study
Only 30 of the control plants(n=25 for each popu-
lation)flowered in the greenhouse , of which six plants
were diseased and the fractions of diseased flowering for
Fig.3.　The frequency distribution(%)of style length(mm)in
experimentally transplanted plants of Silene dioica with different in-
cidence of disease.
n =50 from each population.Measurements made on 20 June
1998.
each of the populations were 37%(n =8)(Population
1), 20%(n =6)(Population 2), 18%(n =11)(Population 3)and 0(n=5)(Population 4)respective-
ly.
92　 植物学报　Acta Botanica Sinica　Vol.44　No.1　2002
Fig.4.　The numbers of pollen and spores per flower on experi-
mentally transplanted female plants of Silene dioica , originating from
four populations (1 , 2 , 3 and 4)with different incidence of dis-
ease.
All plants transplanted to the population on Island number four.a)
samples taken at peak flowering 24 June , b)samples taken 28 June
1999.Mean±SE given , n=25.
The length of papillae on style differed significantly
among populations(ANOVA F3 , 56=9.0 , P <0.000 1)(Table 2).
3　Discussion
In both Silene dioica and S .alba(Miller)Krause ,
males produce more flowers than females and flower for a
longer period of time
[ 15 , 19 , 25] .Furthermore , bumble bees
visit males of S .dioica preferentially over females[ 29] .A
common prediction based on this observation is that male
plants are more likely to be visited by bees carrying spores
and thus experience a higher risk of becoming diseased
than female plants
[ 22] .No consistent differences in infec-
tion of males and females of S.dioica or S.alba have
been found
[ 15 , 16 , 25] , although at very low incidences of
disease , males may be more likely to become diseased
than females
[ 28] .Alexander[ 17 , 22] , however , found that
males of S .alba , with large floral displays , had a higher
probability of becoming diseased than those having fewer
flowers , presumably because the larger males received
spores on a larger number or higher proportion of the flow-
ers.Furthermore , male plants flowering early in the sea-
son were more likely to become diseased than those flow-
ering later
[ 22 , 33, 34] .This pattern may be attributed to an
increased likelihood of successful over_wintering of the
pathogen in the root of the host , if spores germinate and
penetrate the host early in the season.In contrast to the
results for males , Alexander[ 22] found that among female
plants , a variation in phenology or flower number did not
explain the probability of disease.
Fig.5.　The frequency distribution(%)of spore number(a)and
pollen number(b)per flower on experimentally transplanted females
of Silene dioica originating from four populations with different inci-
dence of disease.
All plants transplanted to the population on Island 4.Samples taken
at peak flowering 24 June , 1999.
In our study of female S.dioica , numbers of pollen
and spores deposited on flowers were strongly and posi-
tively correlated in the experiment , and moderately corre-
lated in the 1998 field survey of the two highly diseased
populations.The temporal decline in spore and pollen
deposition observed in the experiment corresponds to find-
ings in other studies
[ 17 , 21] .Female flowers varied in mor-
phology between populations and moderate positive pheno-
typic correlations were found between different floral char-
acters.Plants from the population on Island 1 had signifi-
cantly longer style and larger corolla compared to the other
plants.Pollen and spore deposition showed only moderate
correlation with the floral characters measured , and no
LIU Deng_Yi et al:Infection of the Anther_smut Microbotryum violaceum in Populations of Silene dioica 93　
Fig.6.　 The frequency distribution (%) of style length from
healthy plants and diseased plants from population number 3 and 4.
Diseased plants grown under high temperature and long day condi-
tions in greenhouse November 1998 , causing the induction of normal
healthy flowers.
single character explained the great variations of the ob-
served characters.
In the transplantation experiment , plants originating
from the non_diseased population on Island 1(Population
1)captured approximately 4 times more pollen(averaged
over the two sampling periods), and 9 times more spores
per flower , than the individuals from the resident popula-
tion(Population 4).The frequency distribution showed
that among plants from Population 1 , a few flowers cap-
tured more than 100 000 spores , while among plants from
Population 4 , most captured less than 10 000 and none
more than 20 000.The resulting incidences of disease
among control plants from the four populations differed
significantly , being 37%, 20%, 18% and 0 for plants
from populations 1 , 2 , 3 and 4 respectively.Although
only 30% of the control plants came into f lower , we have
no reason to suspect that the proportion of diseased plants
among the remaining non_flowering plants should be dif-
ferent from that recorded above.Long_term tagging of both
healthy and diseased female plants has shown very similar
flowering patterns for the two groups of plants
[ 18] .
One obvious interpretation of the observed pattern of
disease is that a higher deposition of spores results in a
higher probability of disease.The resident Population 4
plants would thus be more likely to escape disease by hav-
ing a low ability to capture spores present on the visiting
bumble bees.However , spore deposition patterns and
population origin were confounded in this experiment.The
results may also indicate that immigrants to an island , on
which there is diseased Silene population , are more likely
to become diseased than the resident healthy plants , be-
cause they may have lower frequencies of genotypes resis-
tant to the particular isolates of the pathogen present in
the new population.Furthermore , since plants from the
non_diseased population showed the highest incidence of
disease , we cannot rule out that the specific background
in the source population (disease present or absent)of
immigrant plants may also be of importance.The natural
rate of newly infected plants in population 4 is very low ,
being in the order of 0 -2% per year when estimated
during a five year period
[ 18] .
Since we did not attempt to study the activity and
behaviour of bumble bees in the experimental plots , we
have no detailed knowledge about the proximate mecha-
nisms behind the observed patterns of spore and pollen
deposition.Long style and a high density of papillae may
increase the efficiency of pollen spore capture at each av-
erage visit by a flower visitor.Bumble bees may also pref-
erentially visit plants with large flower corolla
[ 30 , 31] .The
observation that style length and flower size are positively
correlated with nectar volume may also influence pollen
and spore deposition in two ways.Large nectar volume
may result in bumble bees staying longer in a single visit ,
or in higher frequencies of multiple visits , compared with
plants with small nectar volume.Both factors have been
found to increase pollen deposition
[ 32] .
The temperature treatment experiment suggested that
the fraction of plants in a population that became dis-
eased , had longer style compared with the fraction that re-
mained healthy.This indicates that the pathogen may be
an important selective factor on flower morphology.Fur-
thermore , in the survey of ten populations , style length
was significantly and negatively correlated with the inci-
dence of disease.However , at present , we know very lit-
tle about the basis for the variation , and to what degree
these characters show a plastic response to a variation in
the environment.A crossing_experiment is under way to
examine the potential additive genetic variance in these
traits and to determine the degree to which flower charac-
ters may be under selective pressure by the pathogen.
In our experimental study there was no evidence that
plants with low pollen capturing capacity were at a disad-
vantage.There were no differences between the popula-
tions in seed set even though pollen numbers per flower
varied four_folds.This suggests that even if plants form
populations 3 and 4 are less successful in capturing pol-
len , there is no apparent fitness cost in terms of a de-
creased proportion of offspring successfully matured.In
Skeppsvik archipelago , S .dioica seeds are transported
by water , resulting in a rather high potential for dispersal
between islands and populations.This study suggests that
a)immigrants to a diseased population may be more likely
to become infected than resident healthy plants , b)high
fecundity immigrants(large flowers )may have a higher
risk of infection , and c)if the variation in flower size and
fecundity is at least partly genetically determined , there is
a considerable potential for pathogen pollinator_mediated
selection on floral characters , and influences by the
pathogen on gene flow.
To further evaluate the influence of the pathogen on
the genetic structure of host populations and gene flow ,
we also need to carry out an inverse transplantation exper-
iment in addition to the crossing_experiments.Do plants
with small flowers and a low pollen capturing ability(originating from highly diseased populations)have a high
risk of pollen limited reproductive success in populations
where the pathogen is absent ?Most non_diseased popula-
tions in this archipelago are found on young small islands ,
94　 植物学报　Acta Botanica Sinica　Vol.44　No.1　2002
where host population size is small and stochastic varia-
tions in sex ratios and pollinator abundance increase the
risk of pollen limited reproductive success
[ 18] .We predict
that variation in pollen capturing ability under these cir-
cumstances , and in the absence of the pathogen , will be
rather closely and positively correlated with reproductive
fitness.
References:
[ 1 ] 　Burdon J J.Diseases and Plant Population Biology.Lon-
don:Cambridge University Press , 1987.[ 2 ] 　Lipsitch L , Herre E A , NowakM A.Host population struc-
ture and the evolution of virulence:a “ law of diminishing
returns” .Evolution , 1995 , 49:743-748.[ 3 ] 　Liu D_Y(刘登义).Pathogenic fungi and natural plant pop-
ulations I.Effects of pathogenic fungi on natural plant popu-
lations.Acta Ecol Sin(生态学报), 1994 , 14 (Suppl):
125-130.(in Chinese with English abstract)[ 4 ] 　Liu D_Y(刘登义).Pathogenic fungi and natural plant pop-
ulations Ⅱ.Pathogenic fungi and plant population biology.
Acta Ecol Sin(生态学报), 1996 , 16:660-663.(in Chi-
nese with English abstract)
[ 5 ] 　Liu D_Y(刘登义).Pathogenic fungi and natural plant pop-
ulations Ⅲ.Pathogenic fungi_host plant coevolution and
population models for vector dispersed disease.Acta Ecol
Sin(生态学报), 1997 , 17:105-108.(in Chinese with
English abstract)[ 6 ] 　Wennstrom A , Ericson L.The effect and transmission of
one isolate of the rust Puccinia minussensis on five clones of
Lactuca sibirica.Oecologia , 1994 , 97:407-411.[ 7 ] 　Anderson R M , May R M.The invasion , persistence and
spread of infectious disease within animal and plant commu-
nities.Phil Trans R Soc Lond B , 1986 , 314:533-570.
[ 8 ] 　May R M.Population biology and population genetics of
plant_pathogen associations.Burdon J J , Leather S R ,
Pests , Pathogens and Plant Communities.New York:
Blackwell , 1990.[ 9 ] 　Jarosz A M , Burdon J J.Host_pathogen interactions in natu-
ral populations of Linum marginale and Melampsora lini.
Oecologia , 1992 , 89:53-61.
[ 10] 　Jarosz A M , Davelos A L.Effects of disease in wild popula-
tions and the evolution of pathogen aggressiveness.New
Phytologist , 1995 , 129:371-387.
[ 11] 　Ingvarsson P.The effects of vector_borne fungal pathogens
on natural plant populations.Lund Universiyt.Introductory
Paper Press of Lund University , 1991.[ 12] 　Garcia_Guzman G , Burdon J J , Ash J E , Cunningham R B.
Regional and local patterns in the spatial distribution of the
flower_infecting smut fungus Sporisorium amphilophis in nat-
ural populations of its host Bothriochloa macra.New Phy-
tol , 1996 , 132:459-469.[ 13] 　Garcia_Guzman G , Burdon J J.Impact of the flower smut
Ustilago cynodontis (Ustilagiaceae)on the performance of
the clonal grass Cynodon dactylon (Gramineae).Amer J
Bot , 1997 , 84:1565-1571.[ 14] 　Giles B E , Goudet J.Genetic differentiation in Silene dioica
metapopulations:estimation of spatiotemporal effects in suc-
cessional plant species.Amer Nat , 1997 , 149:507-526.[ 15] 　Baker H G.Infection of species of Melandrium by Ustilago
violacea (pers.)Fuckel and the transmission of the result-
ant disease.Ann Bot , 1947 , 9:43.[ 16] 　Lee J A.Variation in the infection of Silene dioica (L.)
clairv.By Ustilago violacea (Pers.)Fuckel in northwest
England.New Phytol , 1981 , 87:81-89.[ 17] 　Alexander H M.Epidemiology of anther_smut infection of
Silene alba by Ustilago violacea:patterns of spore deposi-
tion and disease incidence.J Ecol , 1990 , 78:166-179.[ 18] 　Carlsson U , Elmqvist T.Epidemiology of the anther_smut
disease Microbotryum violaceum and numeric regulation of
populations of Silene dioica.Oecologia , 1992 , 102:253-
262.[ 19] 　Carlsson_Graner U.Anther smut disease in Silene dioica:
variation in susceptibility among genotypes and populations ,
and patterns of disease within populations. Evolution ,
1997 , 51:1416-1426.[ 20] 　Jennersten O.Insect dispersal of fungal disease:effects of
Ustilago infection on pollinator attraction in Viscaria vulgar-
is.Oikos , 1998 , 51:163-170.[ 21] 　Liu D_Y(刘登义).Effects of an interspecific competition
gradient on the interactions between Trifolium repens and its
pathogenic rust fungus Uromyces trifolii_repentis.Acta Phy-
toecol Sin(植物生态学报), 2001 , 25:344-350.[ 22] 　Alexander H M.An experimental field study of anther_smut
disease of Silene alba caused by Ustilago violacea:genotyp-
ic variation and disease incidence.Evolution , 1989 , 43:
835-847.[ 23] 　Jarosz A M , Burdon J J , Muller W J.Long_term effects of
disease epidemics.J Ecol , 1989 , 26:725-733.[ 24] 　Hassan A , MacDonald J A.Ustilago violacea on Silene di-
oica.Trans Br Mycol Soc , 1971 , 56:541-461.
[ 25] 　Alexander H M , Antonovics J.Disease spread and popula-
tion dynamics of anther_smut infection of Silene alba caused
by the fungus Ustilago violacea.J Ecol , 1988 ,76:91 -
104.
[ 26] 　Kevan P G , Parmelee J A.Insect_flower relationships for
the transmission of the smut Ustilago violacea by flower vis-
iting insects in the High Arctic.Greenhouse , Garden ,
Grass , 1972 , 11:6-13.[ 27] 　Ebert D , Mangin K L.The influence of host demography on
the evolution of virulence of a microsporidian gut parasite.
Evolution , 1997 , 51:1828-1837.
[ 28] 　Carlsson U, Elmqvist T , Wennstrom A , Ericson L.Infec-
tion by pathogens and population age of host plants.J Ecol ,
1990 , 78:1094-1105.
[ 29] 　Kay Q O N , Lack A J , Bamber F C , Davies C R.Differ-
ences between sexes in floral morphology , nectar production
and insect visits in a dioecious species Silene dioica.New
Phytol , 1984 , 98:515-529.
[ 30] 　Galen C , Newport M E A.Bumble bee behavior and selec-
tion on flower size in the sky pilot Polemonium viscosum.
Oecologia , 1987 , 74:20-23.
[ 31] 　Bell G.On the function of flowers.Proc R Soc Lond R ,
1985 , 224:223-265.[ 32] 　Thompson J D.Pollen transport and deposition by bumble
bees in Erythronium:influences of floral nectar and bee
grooming.J Ecol , 1986 , 74:329-341.[ 33] 　Alexander H M.Pollination limitation in a population of
Silene alba infected by the anther_smut fungus Ustilago vio-
lacea .J Ecol , 1987 , 75:771-780.[ 34] 　Alexander H M , Maltby A.Anther_smut infection of Silene
alba caused by Ustilago violacea :factors determining fungal
reproduction.Oecologia , 1990 , 84:249-253.
LIU Deng_Yi et al:Infection of the Anther_smut Microbotryum violaceum in Populations of Silene dioica 95　
麦瓶草种群中的花药黑粉菌感染:花的形态学变异 、菌孢子
散布式样及病原菌-传媒调节性选择
刘登义1＊ 　Ulla CARLSSON2
(1.安徽师范大学生物多样性研究中心 , 芜湖 241000;
2.Department of Ecology and Environmental Science , University of Umeá , S_901 87 Umeá , Sweden)
摘要:　花药黑粉菌(Microbotryum violaceum)可系统侵染其寄主植物麦瓶草(Silene dioica), 使其不育而代之以在植物
花中布满菌孢子。这些菌孢子主要由花传媒昆虫带到健康植株。花传媒昆虫对花部特征变异的行为反应不仅可
能影响花粉的输入 输出率 ,而且影响菌孢子的着落率和植株的发病与否。为研究 S .dioica 雌株花部性状特征与
花粉传布 、菌孢子着落及由此而导致的植株染病之间的相关性 , 用采自 4 个不同 S .dioica 种群的植物进行了移栽
实验。该 4个种群均位于瑞典北部的 Skeppsvik 群岛 , 代表了从健康(岛 1 , 种群 1), 低度发病(岛 2 ,种群 2)到高度发
病(岛 3、4 ,种群 3 、4)的发病梯度。从上述4 个不同种群中各采 50 个植株移栽至岛 4的中部。来自健康种群的植株
较来自高发病种群的植株具有较大的花 ,较长的花柱。研究发现 ,着落在花上的花粉粒数和菌孢子数呈强正相关。
来自健康种群的植株每朵花上着落的花粉粒和菌孢子数分别是高发病种群(种群 4)植株的 4倍和 9 倍 ,导致来自 4
个不同种群的植株的发病率存在着显著差异 ,种群1 、2、3、4 的发病率分别为 37%、20%、18%、0。在涉及 10 个种群
的田间调查研究中 ,发现种群中健康植株的平均花柱长度(与花冠大小 , 胚珠数目正相关)与植株的发病率显著负
相关。讨论了病原体_传媒调节的植物花部性状特征选择潜势及由其导致的麦瓶草 S.dioica 种群间的基因漂移。
关键词:　麦瓶草;花形态学;花药黑粉菌侵染;菌孢子着落式样;病原体_传媒调节性选择
中图分类号:Q945.8　　　文献标识码:A　　　文章编号:0577-7496(2002)01-0088-09
收稿日期:2001-03-29　接收日期:2001-07-17
基金项目:瑞典自然科学基金项目。
＊通讯作者。
(责任编辑:崔金钟)
96　 植物学报　Acta Botanica Sinica　Vol.44　No.1　2002