Author for correspondence: Philip M. Gilmartin Tel: +44 (0)113 3432902 Fax: +44 (0)113 3433144 Email: firstname.lastname@example.org
• Heterostyly in Primula is characterized by the development of long-styled pin and short-styled thrum flowers, with anthers midway down the corolla tube in pin flowers, and at its mouth in thrum flowers. Other differences include pollen size and stigmatic papillae length. Several linked genes at the S locus control these differences.
• In this study we have analyzed pin and thrum flowers through the temporal development of heteromorphy.
• These studies indicate that the S locus linked genes that orchestrate heteromorphic flower development act in coordination, but with different temporal and spatial dynamics.
• Style length is differentiated by longer style cells in pin than thrum. However, our studies on cell shape and size within the corolla tube show that a different mechanism mediates the dissimilar elevation of anthers between pin and thrum types. These studies have also revealed that upper corolla tube cells in thrum flowers are wider than those in pin flowers. This results in a larger corolla tube mouth in thrum flowers and represents a new and previously undocumented heteromorphic variation between pin and thrum flowers.
Primula plants have heteromorphic flowers and a sporophytically controlled, diallelic incompatibility system that prevents or reduces both self-fertilization and intramorph fertilization. The two forms of flower – pin and thrum – develop on separate plants. Thrum plants have flowers with anthers visible at the mouth of the corolla tube, and the stigma approximately half way down the corolla tube on a short style and with short stigmatic papillae. Pin plants have flowers with the stigma presented at the mouth of the corolla tube, long stigmatic papillae, and anthers approximately half way down inside the corolla tube. The visible differences in flower morphology are shown in Fig. 1. Heteromorphy is common in Primula, with 91% of all species showing this phenomenon (Richards, 1993), but it is not limited to Primula, and has been documented in a number of species (reviewed in Ganders, 1979; Barrett et al., 2000) with examples found in 28 different families (Barrett et al., 2000). A number of these examples have attracted specific attention, including Linum (for example Dulberger, 1973) and Turnera (for example Truyens et al., 2005), but the genus which has been most widely studied is Primula.
Observations on the significance of the reciprocal arrangement of reproductive organs in promoting out-crossing in Primula were documented in the middle of the 19th century (Darwin, 1862; Hildebrand, 1863). Subsequent to this, other associated differences between the two different floral morphs were reported in different species of Primula. These differences include: (1) pollen size – the pollen produced by pin plants is smaller than that produced by thrum plants (Darwin, 1862; Ornduff, 1979; Heslop-Harrison et al., 1981); (2) pollen number – anthers from pin plants produce twice as much pollen as anthers from thrum plants (Piper & Charlesworth, 1986); (3) style cell length and stigmatic papillae length – the presence of longer cells in the style and longer stigmatic papillae in pin flowers as compared to thrum flowers has been observed (Ganders, 1979; Heslop-Harrison et al., 1981). It has been proposed that stigmatic papillae length is a correlate of style cell length (Richards, 1993), and a mutation in P. sinensis was found that reduces style cell length as well as shortening the length of the stigmatic papillae (Mather, 1950). Other studies on pin and thrum styles from P. sinensis (Stirling, 1932) revealed that, in addition to the presence of longer cells in pin styles, there were also more cells in the pin style than in the thrum style. The differences in style length, pollen size and anther position between pin and thrum flowers are controlled by genes designated G, P and A, respectively, which are tightly linked and collectively known as the S locus.
In combination, these differences in flower form promote intermorph crosses (Darwin, 1862). Pin plants are homozygous for the s allele (ss), and thrum plants heterozygous with a dominant S allele (Ss). Before the genetic basis for this observation had been established, the significance of this breeding system was recognised and documented by Darwin (Darwin, 1862), who observed that pin-thrum and thrum-pin crosses yielded equal numbers of pin and thrum progeny. This ratio is what would be expected from crosses between ss and Ss parents. Although illegitimate within-morph pollinations can be achieved, the seed set in such cases is dramatically reduced, and homozygous true-breeding thrum plants are rarely obtained from such forced crosses between thrum plants (Wedderburn & Richards, 1990), indicating the lethality of this genotype and suggesting the presence of an additional gene at the S locus that prevents SS homozygosity (Kurian & Richards, 1997). Other genes linked to the S locus, but not directly involved in the control of heteromorphy, include Hose in Hose (Ernst, 1931; Webster & Grant, 1990), sepaloid (Webster & Gilmartin, 2003), Staminoid Carpels, a possible new allele of Hose in Hose (Webster & Gilmartin, 2003) and Oak Leaf (Webster, 2005). In addition, the further genes magenta (mg) and maroon (ma) involved in the control of red and purple flower colour, have been localised some distance at either side of the S locus (Kurian, 1986; Richards, 1997).
As a prelude to the molecular genetic investigation of heteromorphic flower development in Primula it is important to fully define the developmental events that coordinate the formation of pin and thrum flowers. Knowledge of the processes and insight into the mechanisms, such as cell elongation and cell division, that result in the differential development and placement of structures within the two forms of flowers, is required to inform the molecular analysis of genes implicated in the control of this phenomenon. Furthermore, a definition of the timing of distinct events that lead to differential formation of pin and thrum flower forms will contribute to an interpretation of the coordination in gene expression once candidate genes at the S locus have been identified.
We have previously defined the early stages of development of both wild-type and mutant forms of P. vulgaris (Webster & Gilmartin, 2003) and found that Primula flowers are homomorphic during their early development. Here we have investigated pin and thrum flowers to reveal the temporal development of heteromorphy in P. vulgaris. We have analysed cell morphology in pin and thrum flowers in order to investigate the basis of differences in style and anther positions between them. A previously undocumented heteromorphic feature was discovered by comparing pin and thrum flowers of the same size and with the normal complement of five petals. The thrum corolla tube mouth is demonstrably wider than the pin corolla tube mouth.
These analyses not only underpin ongoing investigations identifying and characterizing genes located at the Primula S locus, but will contribute to their subsequent functional analysis. Once such genes are identified, they will provide appropriate tools for an investigation and interpretation of the evolutionary events which resulted in the establishment of heterostyly in Primula.
Materials and Methods
Origin of plants
Seed for Primula vulgaris L. var. Blue Jeans, an F1 hybrid derived from inbred parental lines by the Farmen Seed Company, Naples, were obtained from Thompson and Morgan (Suffolk, UK). Wild-type P. vulgaris (Primrose) was grown from wild flower seed. The red P. vulgaris used to provide sibling progeny for the measurement of corolla tube mouths of pin and thrum flowers was a commercial line obtained from a garden centre and crossed with wild-type Primrose (Primula vulgaris).
Maintenance of plants
The plants were pot-grown in Levington's plant protection compost and maintained in an unheated glasshouse under natural daylight. Shade during the warmer months was provided by applying ‘Coolglass’ to the outside of the glasshouse and green netting inside. During the summer, they were placed outside on hard-standing.
Scanning electron microscopy
Four developing buds of different sizes were examined, together with a mature flower from a pin plant and from a thrum plant of P. vulgaris var. Blue Jeans. Upper and lower corolla tube tissue was dissected from the middle region of the zone with scalpels and razor blades and using a ×20 hand lens. All tissue was prepared for SEM using an Emscope SP 2000 Sputter Cryo system. Samples were mounted on a copper stub using a thin layer of 2% aqueous methyl cellulose. Following mounting, the stub was screwed on to the freezing rod and immediately plunged into liquid nitrogen under dry argon gas and subsequently transferred to the microscope cold stage where its temperature was raised to −60°C until no ice crystals remained. The sample was then transferred to a sputter chamber where the sample was sputter coated with gold under vacuum. Specimens were transferred to a Phillips 501B SEM cold stage and maintained at −153 to −155°C during observation. Photographs of the images were taken on FP4 film and processed in Ilford ID11 developer.
Photographs of whole developing buds were taken with a Zeiss Tessovar macro photography system using an Olympus OM2 camera with dedicated off-camera flash and Kodak Royal 100ASA colour print film. Colour photographs of corolla tubes were taken using an OM10 SLR hand-held camera fitted with a wide-angle lens set on Macro and with additional close-up lenses where necessary using Activa 400ASA film. A total of 92 developing flowers from 24 different plants (both P. Vulgaris var. Blue Jeans and wild-type P. vulgaris) were used for the analysis of developing flowers and data is presented for Blue Jeans.
Measurement of cell length and width
A total of 300 cells, 15 from each SEM photographic image, were measured with a ruler and their mean cell length was calculated. Measurements in millimetres were converted to micrometres based on the size bars on the images, and means and standard deviations calculated using Microsoft Excel. Results were plotted with error bars shown as standard deviations.
Comparison of pin and thrum lower corolla tube cells
Two flowers of the same or similar size from both pin and thrum plants of P. vulgaris L. var. Blue Jeans were used. Corolla diameter and corolla tube height from the base of the calyx to the top of the corolla tube was measured and recorded. The flowers were split in two and the inner cell layers were scraped away from the middle of the lower corolla tube using a razor blade. This central section of the lower corolla tube was then mounted on a slide and examined under a light microscope fitted with a digital camera.
The lower corolla tube cells of both pin and thrum flowers were photographed alongside a graticule for scale. Cell lengths were measured in millimetres from printed images using a ruler and the lengths subsequently converted to micrometres. Mean and standard deviations were calculated and graphs plotted using Microsoft Excel.
Measurement of pin and thrum corolla mouths
The diameter of the face of the flower and corolla tube mouths was measured in pin and thrum flowers from sibling plants, each with the normal complement of five petals. The data was collected from both of two different lines, with all measurements made to within 0.5 mm. The plants used were Primula vulgaris var. Blue Jeans (60 flowers: 30 pin and 30 thrum) and a red cultivar derived from a cross between a red commercial P. vulgaris and wild primrose (50 flowers: 25 pin and 25 thrum).
Comparison of development between pin and thrum flowers
As Primula flowers were found to be homomorphic during early ontogeny (Webster & Gilmartin, 2003), development of the pin and thrum heteromorphic features of Primula were investigated in them as they developed to maturity. The heteromorphic features of mature flowers are shown in Fig. 1a–d. The developing flowers of P. vulgaris cv. Blue Jeans are shown in Fig. 2; observations of development in this cultivar are typical of other Primula cultivars (not shown). A comparison of pin and thrum flower development and the onset of heteromorphic features are shown in Fig. 2 with measurements on the length of the second floral whorl taken from the base of the ovary to the tip of the petal lobes. The first indication of differences between the pin and thrum morphs can be seen in flowers with a corolla of 2 mm. Comparison of a pin flower of this size (Fig. 2a), with a similar sized thrum flower (Fig. 2b), reveals that elongation of the pin style has begun to raise the stigma above the anthers in pin plants.
In flowers with a whorl 2 of less than 5 mm, the elongation of the pin style continues to be much more pronounced, as illustrated in Fig. 2c,d, and this difference becomes more evident in flowers with a 7 mm corolla (Fig. 2e,f) and 11 mm corolla (Fig. 2g,h). During the development of this aspect of heteromorphy in pin flowers, in similar sized thrum flowers the thrum stigma remains below the top of the anthers throughout these stages of development (Fig. 2b,d,f,h). There is limited growth of the corolla tube below the point of stamen attachment in pin and thrum flowers with a whorl 2 of less than 5 mm (Fig. 2a–d).
When whorl 2 of the Blue Jeans cultivar has reached approx. 7 mm, elongation of the corolla tube below the anthers can be observed in the thrum flower but not in the pin flower (Fig. 2e,f). Elongation of the corolla tube below the anthers was also observed in developing thrum flowers with a whorl 2 of 6 mm (Webster, 2005). It is not until whorl 2 has exceeded 11 mm in length that elongation of the corolla tube below the position of stamen attachment can be seen to have begun in the pin flower (Fig. 2g). Beyond this size, elongation of the corolla below the point of stamen attachment increases. Both the pin and the thrum flowers in Fig. 2i,j are about to open and the final position of the anthers – at the mouth of the flower in thrum and half way down the corolla tube in pin – can be observed. Similarly, the elongated style of the pin flower and the short style of the thrum flower can also be observed. Figure 2k,l illustrate the fully developed heteromorphic differences in mature pin and thrum flowers.
Analysis of upper and lower corolla tube cell size and shape in pin and thrum flowers
From analyses of Primula floral architecture, and following defined previously nomenclature (Sporne, 1974; Erbar, 1991), we divided the corolla tube into two domains. The region below the point of anther attachment is the lower corolla tube, and the region above the point of attachment the upper corolla tube (Fig. 1a,b). Beyond the mouth of the corolla tube is the petal lobe, which forms the face of the flower (Fig. 1c,d). Our observations highlighted differences in the timing of the elongation of upper and lower corolla tubes in pin and thrum plants (Fig. 2). Following these observations of temporal differences in corolla tube development, we investigated the size and shape of cells of the upper and lower corolla tube in pin and thrum plants throughout their development. It might be expected that the elevation of anthers to the mouth of the corolla in thrum flowers would be accompanied by an increase in cell length below the anthers as compared to the corresponding tissue in pin flowers following the precedent of longer cells in pin styles as compared to thrum styles (our data not shown, and Heslop-Harrison et al., 1981).
Scanning electron microscopy was used to analyse outer epidermal cells of the corolla tube in developing pin and thrum flowers of P. vulgaris var. Blue Jeans. Flowers ranging from 10 mm in length to maturity (Fig. 3) were analysed. The measurements represent sizes from the base of the flower to the tip of the sepals. The data for cell length, cell width and the ratio of cell length to cell width (n = 15) are presented in Fig. 4. In 11 mm buds, where whorl 2 was 7 mm long, the anthers in thrum flowers were higher than the anthers in pin flowers (Fig. 2e,f). However, analysis of corolla tube cells above and below the point of anther attachment in 10 to 15 mm flower buds did not reveal major differences between the cell size and shape, either in pin (Fig. 3a–d), or in thrum flowers (Fig. 3k–n). Graphical representations of cell length, cell width and cell length to width ratio confirmed this observation (Fig. 4).
In 10 mm flower buds, there was still evidence of cell division and associated growth in more than one direction (Fig. 3a,b,k,l). In 12 to 15 mm buds, recent cell divisions were observed by less deeply marked divisions between the cells, typically across the axis of corolla tube growth with cell elongation along the axis growth. However, in 16 to 18 mm buds, differences in cell shape were apparent (Fig. 3e,f,o,p). Although the upper and lower corolla tube cells in pin flowers were similar (Fig. 3e,f), the shape of the cells in the thrum upper corolla tube (Fig. 3o) took on a more globular appearance, in contrast to the columnar shaped cells in the lower corolla tube (Fig. 3p). These data are tabulated in Fig. 4.
At 19 mm, the thrum upper corolla tube cells are significantly shorter than their counterparts in the upper corolla tube of equivalent pin flowers (Fig. 4a). However, there is no significant increase in the length of thrum lower corolla tube cells and their mean cell length is less than the mean cell length in the corresponding pin lower corolla tube. Analysis of the ratio of cell length to cell width emphasised the difference, with the ratio falling below 2 for the thrum upper corolla cells in comparison to a ratio of 3 in the other three samples (Fig. 4c).
As development progressed towards maturity (Fig. 3g,h,q,r), as well as in mature flowers (Fig. 3i,j,s,t), the difference in shape between thrum upper (Fig. 3q,i) and lower (Fig. 3r,t) corolla tube cells became more obvious. A comparison of upper (Fig. 3g,i) and lower (Fig. 3h,j) pin corolla tube cells in flowers of the same size did not reveal obvious differences in cell shape.
As the flowers approached maturity, a comparison of cell length and cell width revealed a difference in length between the upper corolla tube cells in pin and thrum flowers (Fig. 4); thrum upper corolla tube cells are also significantly wider than in pin (Fig. 4b). There was no significant difference in the width of upper and lower corolla tube cells in pin flowers and lower corolla tube cells in thrum flowers. In mature pin flowers (20 mm), the lower corolla tube cells had a longer mean cell length than the upper corolla tube cells (Fig. 4a), but the difference was less marked than in thrum flowers. In thrum flowers (Fig. 4a), the lower corolla tube cells were significantly longer than the cells in the upper corolla tube.
Comparison of lower corolla tube cells in pin and thrum
In maturing 19 mm flowers, the lower corolla tube cells in thrum (185 ± 44 µm) were longer than in pin flowers (126 ± 32 µm); however, in mature flowers (20 mm) there was no significant difference in the length of these cells (pin, 250 ± 37 µm; thrum, 244 ± 79 µm). These observations did not reveal a difference in cell length that could account for the longer lower corolla tube in thrum flowers (Fig. 4). There is clearly a wider range of cell sizes in the thrum sample analysed, and it should be noted that the five longest cells in this analysis extended beyond the field of view; they may therefore underestimate the true mean. The standard deviation of the mean reflects the variability and range of individual measurements within this sample.
In order to further investigate the observation that cells in the lower corolla tubes of pin and thrum flowers are of similar length, we undertook an extended analysis using three mature pin flowers from a single plant and three mature thrum flowers from a single plant, of Primula vulgaris var. Blue Jeans. Care was taken to select flowers of exactly the same size in order to minimise any differences due to flower size. The three pairs of flowers were analysed separately, and as a pooled data set (Fig. 5a), results from these independent analyses are in close agreement with each other. The mean lower corolla tube cell length from the first pair of pin and thrum flowers were 137 µm and 169 µm, respectively; with the second pair 141 µm and 163 µm, and the third pair 160 µm and 181 µm. The combined means from 60 measurements of pin and thrum lower corolla tube epidermal cells are 146 µm and 171 µm, respectively. These data are shown in Fig. 5b, and reveal that the mean values from all pin flowers are lower than the mean values from equivalent thrum flowers. Error bars, based on standard deviations, overlapped, suggesting that there was no overall significant difference in the mean values. However, the individual cell length measurements from pin and thrum flowers were plotted in order of increasing length and this representation of the data (Fig. 5c) clearly showed that the thrum cells were typically longer than pin cells, but the differences in length were not sufficient to account for the longer lower corolla tube in thrums. This observation supports the conclusions made by Stirling (Stirling, 1932) following analyses of P. sinensis and P. veris, in which he calculated that there were more cells below the anther insertion point in the thrum flowers than in the pin flowers.
The diameter of the floral mouth is different in Pin and thrum flowers
Based on observations of cell length and width in different regions of the pin and thrum corolla, and the discovery that the upper corolla tube cells in thrum flowers had an average width of 33 µm as compared to those in pin plants with an average width of 20 µm, we investigated the effect of this differential cellular development on floral architecture. We reasoned that, if there were similar numbers of cells around the circumference of pin and thrum corolla tubes, the circumference, and therefore the diameter, of the mouth of the flower should be approximately 1.65-fold greater in thrum flowers than in pin flowers.
The mean diameter of the flower face in the P. vulgaris red cultivar (Fig. 6a) was 34.4 mm for both pin and thrum flowers (Fig. 6b). The mean flower diameter in the Primula vulgaris var. Blue Jeans was 37.0 mm (Fig. 6c). The corolla tube mouth diameters in both samples was greater in the thrum flowers than in the pin, with mean values of 2.0 ± 0.2 mm and 3.3 ± 0.5 mm, respectively, for pin and thrum flowers from progeny derived from red primrose and 2.9 ± 0.9 mm and 4.0 ± 0.9 mm from pin and thrum flowers of from P. vulgaris var. Blue Jeans. In both samples, the ratio of flower diameter to corolla mouth diameter was greater for pin than for thrum flowers of the same size and further emphasises the difference between the two floral morphs (Fig. 6b,c). The error bars (Fig. 6b,c), representing standard deviations, indicate that differences in flower mouth diameter are statistically significant between pin and thrum flowers of the red cultivar of the wild-type P. vulgaris. The commercial cultivar P. vulgaris var. Blue Jeans showed a similar, although slightly less dramatic, difference. This aspect of Primula heteromorphic floral architecture has not been documented previously.
The presence of pin and thrum flowers in Primula and the phenomenon of heteromorphic flower development has reputedly been known since Clusius, 1583 (van Dijk, 1943, quoted in Ornduff, 1993), and has been of scientific interest since Darwin, 1877. However, there is as yet no insight into the identity of the three genes G, P and A, or understanding of the cellular processes controlled by these genes that contribute to the differential development of pin and thrum flowers. This analysis provides new insight into the temporal development of heteromorphic flowers and in addition we revealed a previously unreported characteristic of heteromorphic flowers in P. vulgaris.
Definition of the upper and lower corolla tube
Previous ontogenetic studies on other species has revealed that, where stamens are attached to the corolla tube, the two parts of the tube above and below the anthers can be formed by two spatially and temporally separate processes (Erbar, 1991). In Vinca for example, the upper corolla tube is formed by post-genital fusion while the lower corolla tube forms by congenital fusion (Boke, 1948). This division of the corolla tube into two parts, the upper corolla tube as the part above the point of anther attachment and the lower corolla tube as the part below anther attachment, has also been employed by Sporne (Sporne, 1974). Studies in Lactuca sativa and other members of the Compositae with epipetalous stamens have revealed that both parts of the petal develop by intercalary growth (Sporn, 1974). The lower part of the corolla tube develops later than the upper part of the corolla tube and, as the lower corolla tube develops, the stamens are carried up to their final epipetalous position. Erbar also referred to the lower corolla tube as the stamen-corolla tube (Erbar, 1991). Expression analysis of the Primula vulgaris MADS box gene PvPlena (Cook, 2002), which is the presumed homologue of the Antirrhinum and Arabidopsis C function genes Plena (Carpenter & Coen, 1990; Bradley et al., 1993) and Agamous (Bowman et al. 1989; Yanofsky et al., 1990), revealed that this gene is expressed in the lower corolla tube, but not in the upper corolla tube above the anthers (Cook, 2002). This observation is in agreement with Erbar's definition of the stamen-corolla tube (Erbar, 1991).
As shown in Fig. 2, development of the upper corolla tube in Primula flowers proceeds ahead of expansion of the lower corolla tube in both pin and thrum flowers. However, when elongation of the lower corolla tube initiates, this growth is observed earlier in thrum than in pin flowers (Fig. 2e,f). Observation of a series of developing flowers from eight thrum plants (not shown) revealed that in all cases, growth of the lower corolla tube was evident by the time whorl 2 had reached approximately 10 mm in length. In contrast, observation of a series of developing flowers from 19 pin plants (not shown) revealed only two that exhibited any elongation of the lower corolla tube by the time that whorl 2 had reached 10 mm. This observation reveals some flexibility in the temporal aspect of growth, but it is the differential physical growth of the corolla tube above and below the attachment point of the stamens that results in different anther positions in the two morphs. Differential physical growth of the corolla tube above and below the anthers is therefore an integral feature of heteromorphy in Primula.
The temporal development of heteromorphic characteristics sheds light on the functioning of S locus genes
Despite considerable interest in the S locus for very many years (Ernst, 1936; Lewis & Jones, 1993; Richards, 1997), there have been no previous observations on the timing of events that lead to the final architecture of the mature Primula flower, and our data provides the first visual analysis of the developmental timing of heteromorphy. We have shown that observable differences between pin and thrum flowers are first seen in buds of 5 mm containing a corolla of 2 mm (Fig. 2a,b), when in pin flowers, the stigma is slightly elevated above the top of the anthers (Fig. 2a) and in the corresponding thrum flower the tip of the stigma remains below the top the anthers (Fig. 2b). The observation that pin style elongation is the first discernible step in heteromorphic development suggests that the visible effects mediated by the G locus may be the first to be implemented. Pin plants are homozygous for the recessive g allele, and therefore an initial key event in Primula heteromorphy may be the action of the dominant G allele in thrum flowers in inhibiting style elongation. Elevation of the thrum anthers as observed by growth of the lower corolla tube is not discernible until the flowers are at least 11 mm long and contain a corolla of 7 mm (Fig. 2e,f) indicating that since the visible effect of A activity is seen later than the visible effect of G activity it is possible that the timing of A activity may occur later than G activity. Pin plants are homozygous for the recessive a allele and do not show an elevation of anthers from the base of the corolla until much later than thrum flowers of the same size (Fig. 2e–j). Taken together, these observations suggest that A plays a dominant role in promoting anther elevation by growth of the lower corolla tube, while G plays a dominant role by suppression of growth of the style.
These observations provide the first indication of a differential timing of developmental events during heteromorphy and indicate that temporal control of S locus gene expression may be an integral component of pin and thrum flower development.
Heteromorphic features achieved by G and A are determined by different mechanisms
The mechanistic distinction where G suppresses style elongation and A promotes anther elevation raises the possibility that they may achieve their respective functions by distinct and separate mechanisms. From our observations of pin and thrum flowers (Fig. 1) and those of others (Darwin, 1877; Heslop-Harrison et al., 1981) it is apparent that the final length of the style in pin flowers is approximately twice the length of the style in thrum plants. Similarly, the proportions of upper and lower corolla tube lengths are approximately 1 : 1 in pin flowers and 1 : 3 in thrum flowers (Figs 1, 2). The developmental difference could in both cases be achieved either by increased cell elongation, or by increased cell division, or by a combination of both. Studies on cell size and shape in the styles (Heslop-Harrison et al., 1981) and corollas of pin and thrum flowers (this study) have indicated that the predominant mechanisms underpinning these developmental events are different. Previous observations on style cell length in Primula vulgaris (Heslop-Harrison et al., 1981) have shown that pin style cells were longer than the corresponding cells in thrum styles. Stylar cell length differences between pins and thrums can also be found in other Primula species (Richards, 1997; Al Wadi & Richards, 1993). The predominant determinant of different style lengths in pin and thrum plants is differential cell elongation, although this is not the sole determinant, as pin style cells are not twice the length of thrum style cells (data not shown, Heslop-Harrison et al., 1981).
The dominant G allele in thrum individuals must therefore play a role in the suppression of style cell elongation, reducing the average thrum style cell length. In contrast, anther height in pin and thrum flowers is not controlled primarily by differential cell elongation, but is mediated initially by different profiles of cell division above and below the points of anther attachment in the corolla tube of pin and thrum flowers. Despite clear differences in the position of the anthers in 10 mm pin and thrum flowers, both of which contain petals that are 7 mm in length (Fig. 2e,f), the lower corolla tube cell lengths are not significantly different (Fig. 4a). The finding that these differences only become significant as the flower approaches maturity suggests that the elevation of anthers in thrum flowers (Fig. 2e,f) must include an increased cell division below the point of anther attachment. Conversely, the corolla tube cell size above and below the point of anther attachment in pin flowers is essentially similar (Figs 3, 4) and is consistent with the position of the anthers in the middle of the corolla tube. Given this observation it is surprising that the mean lower corolla cell lengths in pin and thrum flowers are not significantly different (Fig. 4). A comparison of all cell sizes provides a possible explanation for this finding (Fig. 5). Although the mean cell lengths show small and consistent differences, the variability in cell size within an individual flower causes the spread of the samples. When all cell sizes are ordered and plotted against each other (Fig. 5c) a consistent, but small difference can be seen in the lengths of the lower corolla tube cells. These combined analyses indicate that the elevated anther position in thrum plants is mediated by an increase in cell division in the thrum corolla cells below the point of anther attachment.
Corolla tube mouth diameter, a new heteromorphic character
This analysis was undertaken in duplicate using both wild-type primrose hybrid flowers and a horticultural variety P. vulgaris cv. Blue Jeans. In both examples, the observed greater diameter of the flower mouth in thrum as compared to pin flowers indicated that the presence of this previously undocumented characteristic correlated with floral morph. The selection of equal numbers of pin and thrum flowers of similar sizes and with the normal complement of five petals enabled the elimination of differences that could be attributed to different flower sizes or to the presence of extra petals. The above difference is more pronounced when expressed as a measure of flower diameter to mouth diameter.
An examination of corolla tube cells from above the anthers in thrum and pin flowers using both light microscopy (not shown) and scanning electron microscopy (Fig. 3) showed the corolla tube cells above the anthers to be broader in mature thrum flowers than in mature pin flowers. This observation indicates that the increased width of cells in the thrum upper corolla tube leads to an increased diameter of the flower mouth in thrum as compared to pin flowers. However, quantification of this observation provides a further indication that the increased flower mouth diameter of thrum plants is due entirely to the increased cell widths in the upper corolla.
Given a wild primrose pin flower mouth diameter of 2.0 mm (Fig. 6b), the circumference of the opening can be calculated to be 6.3 mm. Similarly, the circumference of a 3.3 mm diameter opening in thrum flowers is 10.4 mm. Given a mean width of 21 µm in pin upper corolla cells, 300 cells would be needed to create this opening. A similar opening created from approximately 300 cells of 33 µm in diameter, the width of thrum upper corolla cells, would generate a mouth of circumference 9.9 mm – remarkably close to the experimental value of 10.4 mm. Similarly for the horticultural variety P. vulgaris Blue Jeans (Fig. 6c), the circumference of the pin flower mouth can be calculated as 9.1 mm, and for the corresponding thrum flower as 12.6 mm (Fig. 6c). The respective numbers of cells of 21 µm and 33 µm diameter (Fig. 6b) required to form openings of this size is approximately 430 and 380, respectively. Again these are remarkably similar numbers suggesting that flower mouth diameter in flowers with the normal complement of five petals is primarily dictated by the width, and not the number, of cells in the upper corolla. It is considered that this phenomenon may be a direct consequence of the presence of a dominant A allele at the S locus. This new difference between pin and thrum flowers may have previously gone unnoticed due to the varying sizes of flowers in bloom on a plant at any one time, with the earliest flowers to bloom always being larger than the later ones. We propose that this new heteromorphic feature is probably under the control of the A component of the S locus.
These studies have demonstrated the spatial and temporal functions of the dominant alleles of the S locus-linked genes A and G, and our analyses of cell size and shape have identified a new aspect of the Primula heteromorphic phenotype. Isolation and characterisation of the genes located at the Primula S locus, including G and A, will be required to provide a molecular explanation for these cellular and developmental observations. We are currently engaged in identifying DNA sequences and genes that are linked to the Primula S locus (Manfield et al., 2005), with the objective of establishing an integrated physical and genetic map to facilitate the identification and characterisation of genes that control heteromorphy in Primula.
We are grateful for funding from the Gatsby charitable foundation, which has supported our work on heteromorphic flower development in Primula. We would also like to thank The Bristol Naturalist Society for funding the scanning electron microscopy analyses and Alan Beckett for enabling Margaret Webster to undertake the SEM work at the SEM unit in the University of Bristol. We are also grateful to Robert Porter for his excellent technical assistance and help with the SEM in Bristol, and Adrian Hick for assistance with the photography of the P. vulgaris Blue Jeans flowers.