• bumblebees;
  • conservation;
  • microsatellites;
  • population density;
  • Primula sieboldii


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    We assessed the effects of population density and the spatial arrangement of genetically compatible mates on the seed set and pollen flow of a heterostylous, bumblebee-pollinated perennial, Primula sieboldii E. Morren (Primulaceae), by using an experimental population under natural pollination conditions.
  • 2
    We also examined the intermorph differences in the pollen dispersal distance and the frequency of self- and intramorph fertilization.
  • 3
    Seed set was significantly correlated with the number of opposite-morph flowers within 2 m in short-styled genets, and within 3 m in long-styled genets.
  • 4
    Mean pollen dispersal distance within the experimental population was 5.4 m, which was slightly shorter than that in a low-density population measured in a previous study (7.2 m).
  • 5
    The proportions of seedlings sired by the same morph or selfed were 10.2% and 3.0%, respectively, in long-styled mothers, compared with 1.9% and 0.0% among short-styled mothers. However, these differences were not statistically significant and no marked differences in pollen dispersal distances were observed between the two morphs.
  • 6
    Our results suggest that the presence of opposite-morph genets within several metres is essential for success of seed reproduction in P. sieboldii, irrespective of population density, and that stochastic deviation of the morph ratio in small populations is therefore likely to cause considerable reduction in seed set.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Population spatial structure strongly affects the success of seed reproduction in plant species (Kunin 1993; Groom 1998; Richards et al. 1999). One of the most important spatial components that influences both gene flow by pollen (hereafter ‘pollen flow’) and seed set (Morris 1993; Karron et al. 1995b; Kunin 1997; Richards et al. 1999) is population density, with seed production tending to be lower in lower-density populations (Silander 1978; Klinkhamer et al. 1989; Allison 1990). Major factors affecting seed production in zoophilous outbreeding plants include changes in pollinator behaviour in response to population structure (Rathcke 1983; Ågren 1996; Kearns & Inouye 1998), local deviations in sex ratio or lack of genetically compatible mates (Wyatt & Hellwig 1979; Goodell et al. 1997; Richards et al. 1999; Wang et al. 2005; Stehlik et al. 2006) and inbreeding depression (Barrett & Kohn 1991; Ellstrand & Elam 1993).

In general, plant density influences the visitation frequency of pollinators per plant, which tends to be low in low-density populations (Kunin 1993, 1997; Ohashi & Yahara 1998). Other aspects of pollinator behaviour also are affected by plant density, and the pattern of pollen dispersal in populations of differing densities cannot be inferred straightforwardly from their densities. Because pollinators tend to visit adjacent plant patches in sequence, the distance that pollinators move (Morris 1993; Karron et al. 1995b; Cresswell 1997) and, thus, the pollen dispersal distance (Karron et al. 1995a) tends to be longer in lower density populations. Plant density also influences the pattern of pollinator movement between patches (Cresswell 1997) and the number of flowers probed within a patch (Ohashi & Yahara 2002). Optimal foraging strategy predicts that the longer the distance between patches, the more flowers are probed within a patch (marginal value theorem; Charnov 1976) and that pollinators may overlook some patches, visiting only a portion of the patches in low-density populations. This visitation behaviour reduces pollen flow between patches in low-density populations and enhances it in high-density populations (Richards et al. 1999). In addition, pollen carryover would tend to be less in low-density populations because of greater loss of pollen by grooming and wing beat during long flights between patches.

Pollinator behaviour is affected by flower density and phenotype and, although this is irrespective of plant genotype per se, the density of genetically compatible mates and the interaction with ramet density are expected to have considerable influence on the seed set of self-incompatible, outcrossing plants (Wyatt & Hellwig 1979; Aizen 1997; Richards et al. 1999; Ishihama et al. 2003; Wang et al. 2005; Stehlik et al. 2006). Heterostylous breeding systems represent a typical case. In a previous study, we measured seed set and pollen flow of an endangered heterostylous and bumblebee-pollinated herb, Primula sieboldii E. Morren (Primulaceae) (Ishihama et al. 2003). The measurement was performed in an experimental population with low flowering-ramet density, in which small patches with different mate densities were linearly arranged at about 15-m intervals. We found that the presence or absence of the opposite morph within a patch had a significant effect on seed set, and that most of the pollen flow was within a patch (76.9%).

Because components of the population spatial structure and environmental factors often mutually covary in natural populations (Aizen 1997; Kunin 1997), we used an experimental population, in which spatial structure parameters can be arbitrarily controlled. In combination, we used highly polymorphic microsatellite markers to enable reliable paternity assignment. Even though the combined use of such markers and an experimental population can be very effective in answering diverse questions in evolution and ecology, such studies are crucially lacking for plants (Smithson & Macnair 2003).

Our objective was to measure the pollen dispersal distance and effect of compatible-mate density on seed reproduction in a population with high flowering-ramet density, as contrasted with those in the low-density population examined in a previous study (Ishihama et al. 2003), addressing the following questions. (i) What is the spatial scale at which the local opposite-morph density strongly affects seed set in a high ramet density population? (ii) How does the pattern of pollen flow in the high-density population differ from that in the low-density population? (iii) Is there any difference in the pollen flow pattern between long-styled and short-styled morphs?

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

study species

Primula sieboldii is a clonally growing, heterostylous perennial. It is listed as vulnerable on Japan's national red list (Environment Agency of Japan 2000). It flowers in spring from May to June and is pollinated mainly by queen bumblebees (Washitani et al. 1994a). A flowering ramet has five flowers on average, and the longevity of a flower is about 1 week. The flowering period of a ramet is about 2 weeks. Seeds of the species have no special mechanism for primary dispersal and are scattered from ripened capsules within close vicinity of the parent plants, although secondary dispersal by water or other agents may contribute to rare long-distance dispersal (Kitamoto et al. 2005a).

field experiment

We performed the experiment at a site with a relatively large natural population (> 1000 flowering ramets) of P. sieboldii at Monbetsu, Hokkaido Prefecture (42°32.3′ N, 142°2.2′ E) in 2001. The mean annual precipitation and air temperature at the nearest meteorological station were 945.2 mm and 7.0 °C, respectively.

Because the species has a narrow, deep tubular corolla with a small opening (< 2 mm in diameter; Washitani et al. 1994b), inexperienced bees often miss the opening and fail to feed from the flower (F. Ishihama, unpublished data). Therefore, to observe natural patterns of pollen flow, we performed our experiments near a natural population where a natural level of visitation by queen bumblebees well conditioned to P. sieboldii was assured.

We established the experimental plants along a line about 45 m in length (Fig. 1). The nearest genet of wild P. sieboldii was 8 m away from the experimental site and was a long-styled morph; the second closest patch was 35 m away and contained both morphs (one genet for each). Because of site limitations, the experimental array was bent orthogonally at one point. This configuration was, however, unlikely to affect the behaviour of the pollinators moving along the line of plants on their foraging trips, because a bush of bamboo grass and a low hill discouraged passing directly from one end to the other.


Figure 1. Spatial structure of the experimental population of Primula sieboldii. All patches consist of different clones. Number of flowering ramets per patch was varied among patches at random. Samples for the paternity analysis were collected from the clones indicated by the arrows. Molecular analysis showed that the same genet was present in two patches (*) and that two patches (†) consisted of two different genets.

Download figure to PowerPoint

spatial structure of the experimental population

The experimental population consisted of 32 small patches arranged in line at about 1.5-m spacing (Fig. 1), with a flowering ramet density about 15 times that of the low-density population in the previous study (Ishihama et al. 2003). We designed the population so that opposite-morph density was highest in the centre of the population and lowest at either end. One end consisted of nine patches of long-styled genets in succession and the other of nine patches of short-styled genets, separated by 14 patches in which both morphs alternated. The number of flowering ramets per patch was varied randomly to evaluate the effect of patch size on seed reproduction. The number and morph of flowering ramets in each patch are given in Table 1.

Table 1.  Number and morph of flowering ramets in each patch in the experimental population, ordered by the position in the experimental population, beginning with the left end (Fig. 1)
Patch nameMorphNumber of flowering ramets
  1. S, short-styled; L, long-styled.

  2. Molecular analysis showed that the same genet was present in two patches(*) and that two patches (†) consisted of two different genets. See text for details.

S1S 1
S2S 1
S4S 4
S5S 1
S6S 1
S7*S 6
S8S 2
S9S 2
S10S 2
L1L 2
S11S 4
L2L 2
S12S 2
L3L 3
S13S 5
L4L 6
S14S 1
L5L 1
S15S 1
L6L 2
S16S 3
L7L 1
L9L 2
L10L 3
L11L 3
L12L 2
L13L 3
L14L 3
L15L 1
L16L 2

We allocated different genets to individual patches, so that we could specify a pollen donor of seeds through paternity analysis. Because microsatellite markers for this species were still under development at the time when the experimental population was established, we distinguished individual genets according to flower appearance (size, colour and shape; Washitani et al. 1991, 1994b) and leaf morphology, which varied considerably from genet to genet. After the experiment, we collected mature leaves from all flowering ramets used in the experiment, and certified the genotypes of microsatellites (see the section on paternity analysis for details of the genetic analysis). As a result, two short-styled patches were found to consist of a common genet, and one patch for each morph to contain two genets (Table 1). Patches containing two genets did not cause any problems in terms of paternity assignment. Although a common genet in two different patches could potentially have caused problems in paternity analysis, none of the seedlings examined was sired by this genet. Nevertheless, the morphs were assigned correctly to all patches, and there were no problems in controlling the local opposite-morph density, which was essential to our study.

field survey and sample collection

We surveyed flowering phenologies of all flowers in the experimental array every 2 days throughout the species’ flowering season (from 24 May to 11 June). We marked each flower at the pedicel with two small strips of coloured vinyl chloride tape, coloured to represent the date of bloom and fall of the corolla. We confirmed sufficient pollinator service by the presence of bumblebee claw marks (Washitani et al. 1994a) on the petals of all flowers.

In June 2001, we collected leaves for DNA extraction from genets in the experimental population, from all flowering genets growing naturally within 100 m of the experimental population and from some flowering genets growing naturally at greater distances. We mapped the coordinates of all 236 wild genets collected using differential GPS (Trimble Japan, Tokyo, Japan) and the measurement error was corrected using electronic control point data supplied by the Geographical Survey Institute of Japan.

stigmatic pollen load

To check whether seed production is limited by pollen, we measured the stigmatic pollen load. We collected stigmas from every other genet of each morph in the experimental array (five flowering ramets per genet and five flowers per ramet, when possible) at the time when the corolla would fall if lightly touched. At this stage, fertilization of ovules was complete, but ungerminated pollen and exines of germinated pollen remained on the stigma. Stigmas were sealed with transparent nail enamel on glass slides for pollen counting in the laboratory. We collected stigmas of 85 flowers from eight short-styled genets and 83 flowers from eight long-styled genets in total. We could not measure the seed set of some of these flowers, however, because they were damaged as a result of herbivores or fungal attack or they swelled with rain and burst before maturity. Therefore, we used 45 stigmas from five short-styled genets and 42 stigmas from seven long-styled genets in later analysis.

The number of pollen grains was counted under a fluorescence microscope (BX50, Olympus, Tokyo, Japan). As pollen from the long-styled and short-styled morphs can be differentiated based on size, the donor morph of each pollen grain was distinguished by determining the maximum diameter of each pollen grain. We determined the threshold diameter for morph identification from the distribution of the widest diameter of each morph according to Stone & Thomson (1994). We obtained the size distribution for each morph by measuring a randomly chosen genet in the Hidaka region (17.4 ± 2.1 µm (mean ± 1 SD) and 13.3 ± 1.4 µm for short-styled and long-styled morphs, respectively; J. Nishihiro, unpublished data). The threshold diameter that was determined was 15.3 µm.

seed set

We collected all fruits of all individuals from late July to early August of 2001 when they had fully matured, and the numbers of intact seeds were counted in the laboratory. We omitted fruits that were damaged by herbivores or fungi (31% of the fruits) from the following analysis. The average number of fruits analysed per patch was 11.0 (ranging from 4 to 40).

paternity analysis

Sample preparation

For paternity analysis, seeds from one or two fruits per genet were germinated in the same way as described in Ishihama et al. (2003). Whenever possible, we chose 25 seedlings per ramet arbitrarily from all seedlings from a fruit. In total, we analysed 395 seedlings from 18 fruits (11 fruits from 10 ramets of eight long-styled genets and seven fruits from six ramets of five short-styled genets) for their paternity.

DNA extraction and microsatellite analysis

We conducted genetic analysis using 10 microsatellite markers developed for P. sieboldii (Ueno et al. 2003, 2006; Kitamoto et al. 2005b). The characteristics of the microsatellite primer pairs used for the analysis are given in Table 2. We conducted DNA extraction and determined the genotypes at microsatellite loci in the same way as described in Ishihama et al. (2003).

Table 2.  Characteristics of microsatellites used for paternity analysis, based on 236 clones
Locus A * H O H E PIC § Exclusion probabilityDDBJ accession no.Reference
  • *

    No. of alleles detected.

  • Observed heterozygosity.

  • Expected heterozygosity.

  • §

    Polymorphism information content (Botstein et al. 1980).

  • Paternity exclusion probability of multiple loci (Weir 1996).

ga0188  80.610.640.570.37AB187573 Ueno et al. (2006)
ga0212 110.710.720.680.49AB088770 Ueno et al. (2003)
ga0235  60.860.800.760.59AB088772 Ueno et al. (2003)
ga0666  90.740.740.700.52AB187574 Ueno et al. (2006)
ga1000 100.740.730.680.49AB187575 Ueno et al. (2006)
ga1116  50.700.770.730.54AB187576 Ueno et al. (2006)
ga1277  80.920.810.780.61AB088775 Ueno et al. (2003)
Pri0141  80.750.750.710.53AB189733 Kitamoto et al. (2005b)
Pri0146 110.750.790.760.60AB189732 Kitamoto et al. (2005b)
PS-5  50.570.550.500.31AF299261 Isagi et al. (2001)
Mean 8.10.750.750.690.9993  
Paternity assignment

We assigned seedling paternity by the maximum likelihood method according to the microsatellite genotypes by using Cervus 2.0 software (Marshall et al. 1998). We included genets of both the opposite and the same morphs for the maternal genet and the maternal genet itself as paternal candidate for the analysis, considering the possibility of selfing and intramorph fertilization implied by the previous study (Washitani et al. 1994b).

The simulation parameters used to evaluate the level of confidence in parentage of the most likely candidate parent were: simulation iteration = 10 000, strict confidence level = 95%, relaxed confidence level = 80%. The number of candidate parents (NCP) has significant impact on the error rate of paternity assignment (Oddou-Muratorio et al. 2003). To estimate this parameter, we used the program PATRI (Nielsen et al. 2001), which gives an unbiased estimate of NCP when there is a uniform distribution of male reproductive success (Oddou-Muratorio et al. 2003). Rate of typing error can also affect the confidence of paternity assignment. However, a simulation scoring error rate of zero seems to give the best results in comparison with higher values, even when the actual error rate is non-null (Oddou-Muratorio et al. 2003). When the simulation scoring error rate = 0, the estimate of NCP was unbiased; the incidence of type II error (failing to assign paternal parent even though the true father is included in the sampled individuals) was lowest in comparison with other values and the incidence of type I error (assigning to a wrong paternal candidate) was relatively independent of the simulation error rate (Oddou-Muratorio et al. 2003). We therefore only report results obtained using a simulation scoring error rate = 0. Because PATRI allows no missing data, we used only seedlings without missing genotypes for later analysis.

statistical analysis

The effect of opposite-morph density on seed set

To detect the scale at which the local opposite-morph density affects the seed set, we constructed a correlogram. We calculated correlation coefficients between the average number of seeds per fruit of each genet and the effective number of opposite-morph flowers within a given radius from a focal genet. We calculated the effective number of opposite-morph flowers by considering the flowering phenology as follows:

  • image

where Fir denotes the effective number of opposite-morph flowers for genet i within radius r, xij denotes the number of flowers of genet i on day j, yijr denotes the number of flowers of the opposite-morph within the radius r from the genet i on day j, and n denotes the total number of flowering days.

In addition, correlations between seed set and pollen load (number of opposite, intramorph and total pollen grains) were tested to check if seed set was limited by pollen. Because there was no significant effect of patch size on seed set in the preliminary analysis, we do not show this result.

Intermorph pattern of mating

To test if the mating patterns differed between the two floral morphs, we performed randomization tests. For the test statistics, we used differences between the two floral morphs in the ratios of seedlings sired by legitimate, self, intramorph and illegitimate (sum of self and intramorph) fathers. In the randomization process, fruits were randomized between the two morphs. These processes were iterated 2000 times. Bootstrap estimates and confidence intervals of the rates of self, intramorph and illegitimate mating were also estimated by bootstrapping fruits within a morph and seedlings within a fruit.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

seed set

There was a trend for the number of seeds per fruit to be higher in the middle of the experimental population, where local opposite-morph density was higher, than at the ends of the population (Fig. 2). The correlation coefficients between the number of seeds per fruit and the effective number of opposite-morph flowers within a given radius were highest at a radius of 2 m and were similar for the two morphs (Fig. 3). However, in the short-styled morph, the correlation coefficient at 2-m radius was remarkably higher than at any other radii, and significantly larger than zero only at this radius, whereas in the long-styled morph the correlation coefficient at 3-m radius was as high as that at 2-m radius, and significantly larger than zero at both these radii.


Figure 2. Relationship between spatial arrangement and seed set in the Primula sieboldii experimental population. Horizontal axis corresponds to the spatial arrangement in the experimental population. Error bars are SE. L5 set no seeds.

Download figure to PowerPoint


Figure 3. Scale of effect of opposite-morph density in Primula sieboldii. Vertical axis is the squared correlation coefficient between the number of seeds per fruit and the effective number of opposite-morph flowers (Fir) within the radius indicated on the horizontal axis. Values with asterisks are statistically significant (modified Fisher's z-transformation, P < 0.05; P-value was corrected for multiple tests by the Bonferroni method).

Download figure to PowerPoint

stigmatic pollen load

There were significant positive correlations between the number of seeds per fruit and opposite-morph pollen grains on stigmas in both morphs (Fig. 4). There was no significant relationship between seed set and the number of intramorph pollen grains or the sum of pollen grains of both morphs (data not shown).


Figure 4. Relationship between pollen deposition and seed set in Primula sieboldii. There were significant correlations between the number of seeds per fruit and opposite-morph pollen grains on stigma in both morphs (long-styled: r2 = 0.904, P = 0.028; short-styled: r2 = 0.936, P = 0.0158).

Download figure to PowerPoint

paternity analysis

Intermorph pattern of mating

The number of paternal candidates estimated by PATRI was 275, and thus the proportion of sampled paternal candidates was 85.8%. Using this value and a zero scoring error rate in the Cervus software, paternities for 95.2% of 336 seedlings analysed were successfully assigned with a 95% confidence level. There were no further candidates assigned with an 80% confidence level. Of the assigned fathers, 68.4% were within the experimental population.

In the middle part of the experimental population, where the local opposite-morph density was high, a large proportion of seedlings were sired by the opposite-morph genet for both morphs (Fig. 5). However, near the end of the experimental population, where opposite-morph density was low, most seedlings of short-styled genets were sired by pollen from outside the experimental population (Fig. 5). Although the rates of self and intramorph mating seem to be higher in the long-styled morph than in the short-styled morph (Fig. 5), the differences were not significant (Table 3).


Figure 5. Composition of fathers in each fruit of short-styled (upper) and long-styled morph (lower and right) in Primula sieboldii. Pieces of the same pattern in a pie chart indicate the pollen flow from different fathers in the same category. Numbers against the pie charts indicate the number of seedlings analysed in each fruit.

Download figure to PowerPoint

Table 3.  Mating patterns in distylous Primula sieboldii based on microsatellite markers. Values are the mean and 95% confidence intervals (in parentheses) of rate of each mating type in the long-styled and short-styled morphs estimated by the bootstrapping method. Illegitimate is the sum of selfing and intramorph mating. Significance of the differences between morphs was tested by randomization test
 Long-styledShort-styled P-value
Selfing0.051 (0–0.18)0 (0–0)0.77
Intramorph0.099 (0–0.23)0.033 (0–0.079)0.44
Illegitimate0.15 (0.023–0.32)0.033 (0–0.079)0.26
Pollen dispersal distances

Average pollen dispersal distance was 4.4 m when pollen flow from outside the experimental population was neglected (Fig. 6), but increased to 15.1 m when pollen flow from outside the experimental population was included. Although the distribution was very skewed, a large part (78.6%) of the external pollen flow was from the two closest long-styled genets, located within 35 m of the experimental population. The remaining external pollen flow was from a number of pollen donors, each of which contributed to only 1–5 seedlings. Maximum pollen dispersal distance was 396 m. The average pollen dispersal distance only differed significantly between two morphs when pollen flow from outside the experimental population was included (Table 4). The observed increase for the short styled morph seems to have been caused by the presence of a wild long-styled genet near the experimental population (Fig. 1).


Figure 6. Distribution of the pollen dispersal distance in Primula sieboldii. Average pollen dispersal distance within the experimental population was 4.4 m. If the pollen from outside the population was included, it was 15.1 m.

Download figure to PowerPoint

Table 4.  Pollen dispersal in distylous Primula sieboldii. Values are the mean and standard deviation of pollen dispersal distance (m). Significant differences in pollen dispersal distances were only evident when the pollen flow from outside was included (P < 0.05, Mann–Whitney U-test)
Within the experimental population 5.0 ± 5.9 2.9 ± 4.2 4.4 ± 5.5
Including pollen flow from outside13.8 ± 40.416.5 ± 19.215.1 ± 32.0


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

effects of opposite-morph density

In either morph, the effective number of opposite-morph flowers within a 2-m radius had the strongest effect on the number of seeds per fruit (Fig. 3). The significant positive correlation between the number of seeds per fruit and the number of opposite-morph pollen grains on the stigma (Fig. 4) indicated possible pollen limitation and confirmed the importance of local density of the opposite morph, despite the high global density. These results are similar to those of our previous study (Ishihama et al. 2003) in which our experimental population had a low global density.

Paternity analysis showed that the most frequent pollen dispersal distance was within 2 m (Fig. 6), matching the neighbourhood radius at which the correlation between seed set and the number of effective opposite-morph flowers was highest (Fig. 3). This neighbourhood radius is comparable with the estimated value of 5 m from the seed set pattern in a natural population (Nishihiro et al. 2000). The shorter distance in the present experimental population may be explained by the higher genet density than that in the previous study.

population density and pollen dispersal pattern

The mean pollen dispersal distance in the present high-density experimental population was 4.4 m (Fig. 3), and the proportion of pollen flow from the adjacent opposite-morph was 54.7%. By contrast, in the low-density experimental population in the previous study (Ishihama et al. 2003), pollen dispersal distance was 7.3 m and the proportion of pollen flow from the adjacent opposite-morph within the same patch was 76.9%.

In the low-density population, small patches (composed of up to three genets) were linearly arranged at about 15-m intervals, and genets within the same patch were separated by 1 m. Local opposite-morph density was controlled by including both morphs in a patch or by isolating a short-styled genet within a patch.

Because of the different arrangements, simple comparison of pollen dispersal distance between high- and low-density populations is inappropriate. To take this difference into consideration, comparison with the expected dispersal distance from the nearest opposite-morph would be effective. The expected values are 2.7 m and 10.9 m for the high- and low-density populations, respectively. Although these expectations explain the observed distances relatively well, some discrepancies are noteworthy. The observed value is somewhat less than the expected value in a low-density population, but a little larger in a high-density population. In addition, the proportion of pollen flow from the adjacent opposite-morph was much higher in the low-density population than in the present high-density population.

Behavioural responses of pollinators to ramet density may partly explain the difference. In contrast to the sequential visits in high-density populations with continuous patch distribution, pollinators tend to visit only a portion of the patches in low-density populations where patches are discontinuously distributed, especially when, as here, patches are small (2–11 flowering stems), reducing between-patch pollen flow. Higher pollen loss during long flights between patches in populations of low ramet density may also reduce the frequency of interpatch pollen flow and, thus, dispersal distance. A similar pollen flow pattern was reported by Richards et al. (1999), who found that pollen flow between small patches of Silene alba was less frequent in low-density (80-m spacing) experimental arrays than in high-density arrays (20-m spacing). Changes in the number of flowers probed within a patch due to differential density (Charnov 1976) are less likely to explain the difference, because the patches were generally small in both our studies.

mating patterns in the two floral morphs

There were no significant differences in the rate of self- and intramorph fertilization between long-styled and short-styled genets (Table 3). However, the estimated proportion of illegitimate mating was above zero in the long-styled morph (Table 3). In addition, higher self- and/or intramorph compatibility was suggested from the seed set pattern in a natural population of P. sieboldii (Nishihiro et al. 2000). Higher intramorph and self-compatibility in the long-styled morph has previously been reported in some Primula species (Crosby 1949; Wedderburn & Richards 1990), although there is no systematic pattern in the intermorphic difference in compatibility among Primula species (Wedderburn & Richards 1990). The non-significance of our results might have been caused by the small sample sizes, a frequent problem in genetic analyses due to their high cost. Because higher levels of assortative mating within either morph could lead to fixation of that morph (Baker et al. 2000), further study is needed on the mating pattern and fitness of the progenies from intramorph mating to clarify the mechanisms maintaining distyly in P. sieboldii.

suggestion for conservation

Even in populations of high global density, local opposite-morph density still had a considerable effect on seed set (Fig. 3). The lack of opposite-morphs at a very local scale (within several metres) can lead to serious reduction in seed set.

For the successful seed reproduction of P. sieboldii, not only is maintenance of ramet density, which is required to ensure a certain level of pollinator visitation, of importance, but so too is consideration of the local quality of mates. Genet density needs to be high enough to ensure an adequate opposite-morph density at a local scale. The demographic consequences of reduced seed set as a result of limited mate availability are still unclear in this long-lived perennial species, and further studies on its effect on population persistence are needed.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Financial support was provided by grants from the JSPS Research for the Future Program (00L0162), Japan Ministry of Education, Science and Culture Grants-in-Aid for Scientific Research (B2-15370008), and the Environmental Research and Technology Development Fund from the Japan Ministry of the Environment. We thank Dr A. Takenaka for comments on an early version of the manuscript and Dr J. Nishihiro, Dr M. Ajima-Nishihiro and Ms M. Nagai for their assistance with the field work.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Ågren, J. (1996) Population size, pollinator limitation, and seed set in the self-incompatible herb Lythrum salicaria. Ecology, 77, 17791790.
  • Aizen, M.A. (1997) Influence of local floral density and sex ratio on pollen receipt and seed output: empirical and experimental results in dichogamous Alstroemeria aurea (Alstroemeriaceae). Oecologia, 111, 404412.
  • Allison, T.D. (1990) Pollen production and plant density affect pollination and seed production in Taxus canadensis. Ecology, 71, 516552.
  • Baker, A.M., Thompson, J.D. & Barrett, S.C.H. (2000) Evolution and maintenance of stigma-height dimorphism in Narcissus. II. Fitness comparisons between style morphs. Heredity, 84, 514524.
  • Barrett, S.C.H. & Kohn, J.R. (1991) Genetics and evolutionary consequences of small population size in plants: implication for conservation. Genetics and Conservation of Rare Plants (eds D.A. Falk & K.E. Holsinger), pp. 330. Oxford University Press, New York.
  • Botstein, D., White, R.L., Skolnick, M. & Davis, R.W. (1980) Construction of a genetic linkage map in man using restriction fragment length polymorphisms. American Journal of Human Genetics, 32, 314331.
  • Charnov, E.L. (1976) Optimal foraging, the marginal value theorem. Theoretical Population Biology, 9, 129136.
  • Cresswell, J.E. (1997) Spatial heterogeneity, foraging behaviour and pollinator-mediated gene flow: bumblebee movements in variously aggregated rows of oil-seed rape. Oikos, 78, 546556.
  • Crosby, J. (1949) Selection of an unfavorable gene-complex. Evolution, 3, 212230.
  • Ellstrand, N.C. & Elam, D.R. (1993) Population genetic consequences of small population size: implications for plant conservation. Annual Review of Ecology and Systematics, 24, 217242.
  • Environment Agency of Japan (2000) Threatened Wildlife of Japan: Red Data Book, 2nd edn. Japan Wildlife Research Center, Tokyo (in Japanese).
  • Goodell, K., Elam, N.C., Nason, J.D. & Ellstrand, N.C. (1997) Gene flow among small populations of a self-incompatible plant: an interaction between demography and genetics. American Journal of Botany, 84, 13621371.
  • Groom, M.J. (1998) Allee effects limit population viability of an annual plant. American Naturalist, 151, 487496.
  • Isagi, Y., Honjo, M. & Washitani, I. (2001) Development of microsatellite markers for Primula sieboldii using degenerate oligonucleotide-primed PCR-amplified DNA. Molecular Ecology Notes, 1, 2224.
  • Ishihama, F., Nakano, C., Ueno, S., Ajima, M., Tsumura, Y. & Washitani, I. (2003) Seed set and gene flow patterns in an experimental population of an endangered heterostylous herb with controlled local opposite-morph density. Functional Ecology, 17, 680689.
  • Karron, J.D., Thumser, N.N., Tucker, R. & Hessenauer, A.J. (1995a) The influence of population density on outcrossing rates in Mimulus ringens. Heredity, 75, 175180.
  • Karron, J.D., Tucker, R., Thumser, N.N. & Reinartz, J.A. (1995b) Comparison of pollinator flight movements and gene dispersal patterns in Mimulus ringens. Heredity, 75, 612617.
  • Kearns, C.A. & Inouye, D.W. (1998) Endangered mutualisms: conservation of plant–pollinator interactions. Annual Review of Ecology and Systematics, 29, 83112.
  • Kitamoto, N., Honjo, M., Ueno, S., Takenaka, A., Tsumura, Y., Washitani, I. & Ohsawa, R. (2005a) Spatial genetic structure among and within populations of Primula sieboldii growing beside separate streams. Molecular Ecology, 14, 149157.
  • Kitamoto, N., Ueno, S., Tsumura, Y., Washitani, I. & Ohsawa, R. (2005b) Development of microsatellite makers in Primula sieboldii E. Morren, a threatened herb. Japanese Journal of Conservation Ecology, 10, 4751 (in Japanese with English summary).
  • Klinkhamer, P.G.L., de Jong, T.J. & de Bruyn, G.J. (1989) Plant size and pollinator visitation in Cynoglossum officinale. Oikos, 54, 201204.
  • Kunin, W.E. (1993) Sex and the single mustard: population density and pollinator behavior effects on seed-set. Ecology, 74, 21452160.
  • Kunin, W.E. (1997) Population size and density effects in pollination: pollinator foraging and plant reproductive success in experimental arrays of Brassica kaber. Journal of Ecology, 85, 225234.
  • Marshall, T.C., Slate, J., Kruuk, L. & Pemberton, J.M. (1998) Statistical confidence for likelihood-based paternity inference in natural populations. Molecular Ecology, 7, 639655.
  • Morris, W.F. (1993) Predicting the consequences of plant spacing and biased movement for pollen dispersal by honey bees. Ecology, 74, 493500.
  • Nielsen, R., Mattila, D.K., Clapham, P.J. & Palsboll, P.J. (2001) Statistical approaches to paternity analysis in natural populations and applications to the North Atlantic humpback whale. Genetics, 157, 16731682.
  • Nishihiro, J., Washitani, I., Thomson, J.D. & Thomson, B.A. (2000) Patterns and consequences of stigma height variation in a natural population of a distylous plant, Primula sieboldii. Functional Ecology, 14, 502512.
  • Oddou-Muratorio, S., Houot, M.-L., Demesure-Musch, B. & Austerlitz, F. (2003) Pollen flow in the wild service tree, Sorbus torminalis (L.) Crantz. I. Evaluating the paternity analysis procedure in continuous populations. Molecular Ecology, 12, 34273439.
  • Ohashi, K. & Yahara, T. (1998) Effects of variation in flower number on pollinator visits in Cirsium purpuratum (Asteraceae). American Journal of Botany, 85, 219224.
  • Ohashi, K. & Yahara, T. (2002) Visit larger displays but probe proportionally fewer flowers: counterintuitive behaviour of nectar-collecting bumble bees achieves an ideal free distribution. Functional Ecology, 16, 492503.
  • Rathcke, B.J. (1983) Competition and facilitation among plants for pollination. Pollination Biology (ed. L.A. Real), pp. 305329. Academic Press, New York.
  • Richards, C.M., Church, S. & McCauley, D.E. (1999) The influence of population size and isolation on gene flow by pollen in Silene alba. Evolution, 53, 6373.
  • Silander, J.A. Jr (1978) Density-dependent control of reproductive success in Cassia biflora. Biotropica, 10, 292296.
  • Smithson, A. & Macnair, M.R. (2003) Have recent developments in molecular techniques led to greater insight into evolutionary and ecological processes in plant populations? Genes in the Environment (eds R.S. Hails, J.E. Beringer & H.C.J. Godfray), pp. 112129. Blackwell, Oxford.
  • Stehlik, I., Caspersen, J.P. & Barrett, S.C.H. (2006) Spatial ecology of mating success in a sexually polymorphic plant. Proceedings of the Royal Society of London Series B, 237, 387394.
  • Stone, J.L. & Thomson, J.D. (1994) The evolution of distyly: pollen transfer in artificial flowers. Evolution, 48, 15951606.
  • Ueno, S., Kitamoto, N., Ohsawa, R., Tsumura, Y. & Washitani, I. (2006) Nine additional microsatellite markers for Primula sieboldii E. Morren. Conservation Genetics, 6, 10631064.
  • Ueno, S., Tsumura, Y. & Washitani, I. (2003) Development of microsatellite markers in Primula sieboldii E. Morren, a threatened Japanese perennial herb. Conservation Genetics, 4, 809811.
  • Wang, Y., Wang, Q.-F., Guo, Y.-H. & Barrett, S.C.H. (2005) Reproductive consequences of interactions between clonal growth and sexual reproduction in Nymphoides peltata: a distylous aquatic plant. New Phytologist, 165, 329336.
  • Washitani, I., Kato, M., Nishihiro, J. & Suzuki, K. (1994a) Importance of queen bumble bees as pollinators facilitating inter-morph crossing in Primula sieboldii. Plant Species Biology, 9, 169176.
  • Washitani, I., Namai, H., Osawa, R. & Niwa, M. (1991) Species biology of Primula sieboldii for the conservation of its lowland-habitat population. I. Inter-clonal variation in the flowering phenology, pollen load and female fertility component. Plant Species Biology, 6, 2737.
  • Washitani, I., Osawa, R., Namai, H. & Niwa, M. (1994b) Patterns of female fertility in heterostylous Primula sieboldii under severe pollinator limitation. Journal of Ecology, 82, 571579.
  • Wedderburn, F. & Richards, A.J. (1990) Variation in within-morph incompatibility inhibition sites in heteromorphic Primula L. New Phytologist, 116, 149162.
  • Weir, B.S. (1996) Genetic Data Analysis II. Sinauer Associates, Sunderland, MA.
  • Wyatt, R. & Hellwig, R.L. (1979) Factors determining fruit set in heterostylous bluets, Houstonia caerulea (Rubiaceae). Systematic Botany, 4, 103114.