The effect of flower position on male and female reproductive success in a deceptively pollinated tropical orchid

In many plants, including orchids, differential fruit set along the inflorescence has been attributed to pollinator behaviour. For instance, the pollinator, moving up the inflorescence, becomes satiated with the resources and leaves before visiting the upper flowers. Consequently, the pollinators do not visit flowers as frequently higher up the inflorescence. Alternatively, flower size may vary along the inflorescence, making pollination ineffective as flowers decrease in size. I tested for the presence of differential pollination along the inflorescence in a pollinator-limited tropical epiphyte, Lepanthes rupestris Stimson, and determined the likely cause of the observed pattern. As this species has inflorescences with sequential flowering, pollinator behaviour, moving up the inflorescence as in synchronous multiflowering inflorescences, can be discounted as an explanation for differential fruit set. Fruit set is shown to be more frequent at the base of the inflorescence, but male reproductive success through pollinarium removal is basically independent of flower position. Moreover, cross-pollination by hand at variable flower positions along the inflorescence results in equal fruit set, suggesting that resources are not limiting and cannot explain the cause of differential fruit production along the inflorescence in natural populations. Furthermore, flower size is shown to diminish along the inflorescence, suggesting that the pollinator(s) may be ineffective at depositing the pollinarium in the smaller higher flowers. Consequently, pollinator behaviour and its interaction with flower size, and not resource limitation, is likely to be the main cause of differential fruit set along the inflorescence in L. rupestris .  © 2006 The Linnean Society of London, Botanical Journal of the Linnean Society , 2006, 151 , 405–410.     

Temporal and spatial variation in flower and fruit production in a food-deceptive orchid: a five-year study

Deceptive orchids are generally characterized by low levels of fruit set; however, there may be substantial variations in fruit set between sites and years. Within a single population, individual plants may also differ greatly in their reproductive output as a result of differences in inflorescence size or local density. In this study, we determined flower and fruit production over 5 years in two populations of the food-deceptive orchid, Orchis purpurea . All plants were monitored annually for survival and flowering at each site to determine whether flowering and fruiting induced costs. The number of flowers per inflorescence varied considerably from year to year (min: 36.6, max: 49.5). Average fruit set was low (7%) and varied considerably among years and populations. A considerable proportion of plants also failed to set any fruit. However, the probability of producing at least one fruit was not affected by inflorescence size or local density. The number of fruits was significantly related to inflorescence size, but proportional fruit set was not. Local density also did not affect the number of fruits, nor proportional fruit set. There was also no evidence that plants with large inflorescence size or high fruiting success had a larger probability of remaining vegetative the year after flowering than plants with small inflorescence size or low fruiting success. Our results suggest that pollinator-mediated selective forces on inflorescence size through female reproductive success alone are weak, most likely because of the low overall level of visitation and the resulting uncertainty of pollination at the individual level. Our results further demonstrate that investigation of patterns of fruit set over several years is needed to better understand the variability in female reproductive success that is typical of most plant–pollinator interactions.     

Fruit set in Styrax obassia (Styracaceae): the effect of light availability, display size, and local floral density

Maternal reproductive success was examined in Styrax obassia (Styracaceae), a bumble-bee pollinated mass-flowering tree in a cool-temperate deciduous forest in northern Japan. The effects of flower number on the success of individual flowers at three levels (inflorescence, individual, and population) were considered. During 1995 and 1996, variations in size, light availability to branches, floral display size, and fruit set were monitored in 37 out of 211 individual S. obassia trees in a 4-ha forest plot. In addition, the locations of the 211 trees in this plot were mapped and the number of inflorescences in each tree was counted. A multiple regression analysis showed that flower number per inflorescence and inflorescence number per individual had negative effects on fruit set, and inflorescence number of aggregated clumps of flowering trees, tree size, and light resource had positive effects on fruit set although significant level were marginal. It is concluded that pollinator attraction may occur not at the individual tree level, but at the level of a clump of flowering trees. It is also suggested that geitonogamy increased with inflorescence number of tree and inflorescence size and that resource limitation was related to the light condition and variation of tree size.     

A test of the effect of floral color change on pollination effectiveness using artificial inflorescences visited by bumblebees

Floral color change has been recognized as a pollination strategy, but its relative effectiveness has been evaluated insufficiently with respect to other floral traits. In this study, effects of floral color change on the visitation pattern of bumblebees were empirically assessed using artificial flowers. Four inflorescence types were postulated as strategies of flowering behavior: type 1 has no retention of old flowers, resulting in a small display size; type 2 retains old flowers without nectar production; type 3 retains old flowers with nectar; and type 4 retains color-changed old flowers without nectar. Effects of these treatments varied depending on both the total display size (single versus multiple inflorescences) and the pattern of flower-opening. In the single inflorescence experiment, a large floral display due to the retention of old flowers (types 2–4) enhanced pollinator attraction, and the number of flower visits per stay decreased with color change (type 4), suggesting a decrease in geitonogamous pollination. Type-4 plants also reduced the foraging time of bees in comparison with type-2 plants. In the multiple inflorescence experiment, the retention of old flowers did not contribute to pollinator attraction. When flowering occurred sequentially within inflorescences, type-4 plants successfully decreased the number of visits and the foraging time in comparison with type-2 plants. In contrast, floral color change did not influence the number of visits, and it extended the foraging time when flowering occurred simultaneously within inflorescences but the opening of inflorescences progressed sequentially within a plant. Therefore, the effectiveness of floral color change is highly susceptible to the display size and flowering pattern within plants, and this may limit the versatility of the color change strategy in nature.     

EFFECTS OF DISPLAY SIZE,FLOWERING PHENOLOGY,AND NECTAR AVAILABILITY ON EFFECTIVE VISITATION FREQUENCY IN COMPARETTIA FALCATA (ORCHIDACEAE)

Comparettia falcata is an epiphytic, neotropical orchid that produces nectar as a pollinator reward. In Puerto Rico, C. falcata is allogamous and pollinated by the endemic hummingbird Chlorostilbon maugaeus. Autogamous pollinations are possible, but may result in reduced fruit set. For the 1989 and 1990 flowering seasons, the probability of pollinarium removals and natural pollinations increased with individual inflorescence display size. However, the frequency of effective pollinator visits was independent of flowering phenology in both years. A positive correlation between inflorescence size and reproductive success occurred in 1990 but not in 1989. In 1990 plants produced longer spurs, a higher standing crop of nectar, and a more concentrated nectar than in 1989. There was no relation between nectar availability (= standing crop of nectar) and sugar concentration in either year. Nectar availability and sugar concentration did not vary among the first four flowers of an inflorescence in either season. Nectar availability was not a good predictor of effective visitation. Comparettia falcata has a higher natural fruit set than tropical deceptive orchids, suggesting that pollinator visitation may be enhanced by nectar reward. The small, dilute nectar volumes secreted by C. falcata may benefit the plant by increasing interplant pollinator movement and pollen dispersal.     

Flowering dynamics in Arum italicum (Araceae): relative role of inflorescence traits, flowering synchrony, and pollination context on fruit initiation

We studied the relative role of inflorescence traits, flowering synchrony, and pollination context for infructescence and fruit initiation in two Spanish populations of Arum italicum, a species in which inflorescences are the pollination unit. In this species, a specialized inflorescence organ, the appendix, is important for pollinator attraction. However, the short floral longevity and the production of mostly one inflorescence per plant make its pollination potentially dependent on strong flowering synchrony and on external factors not controlled by the plant (the pollination context). The flowering period in both sites lasted >3 mo. Day-to-day variation in simultaneous antheses was high, and 11-50% of antheses occurred on days during which no pollen donor was present. Inflorescence traits, flowering synchrony, and between-plant distance all influenced infructescence and fruit initiation, but their relative importance differed between sites. In one large population, infructescence initiation was positively related to inflorescence traits; in a smaller population infructescence initiation increased with the number of donor inflorescences. In both sites, percentage of fruits initiated per infructescence was dependent on a combination of inflorescence traits, flowering synchrony, and between-plant distance. Plants producing 2-4 inflorescences had higher probability of infructescence initiation and overlapped their antheses with more plants than single-inflorescence ones.     

Evidence for stabilising selection acting on flowering time in Arum maculatum (Araceae): the influence of phylogeny on adaptation

The relationship between flowering time and reproductive success was investigated in the fly-pollinated, monoecious perennial herb Arum maculatum L. (Araceae). This species temporarily traps its principle pollinator, a psychodid midge. Probability of fruit set was analysed in relation to early, peak and late periods of the flowering phenology of four British populations between 1992 and 1997. In three out of five cases, plants which flowered during early and late periods were significantly less likely to set fruit. In addition, one population showed a similar relationship for percentage fruit set of individual inflorescences, and seeds from peak-flowering plants were significantly heavier. There was no variation in number of female flowers per inflorescence over the flowering season. Probability of fruit set appears to be mediated by the likelihood of trapping psychodid midges that have previously been trapped and picked up pollen, an unlikely event during early and late flowering periods when few inflorescences are open. The majority of plants in all populations produce only one inflorescence which means that timing of flowering may be crucial to reproductive success. We interpret our findings as evidence that stabilising selection may be acting on some populations and/or during some years. The ultimate cause, however, can be related to the very short (12–18 h) female phase of each inflorescence, a phylogenetically conservative trait within the Araceae. Received: 19 August 1998 / 15 February 1999     

A ballistic pollen dispersal system influences pollination success and fruit‐set pattern in pollinator‐excluded environments for the endangered species Synaphea stenoloba (Proteaceae)

The critically endangered Synaphea stenoloba (Proteaceae) has numerous scentless flowers clustered in dense inflorescences and deploys a ballistic pollen ejection mechanism to release pollen. We examined the hypothesis that active pollen ejection and flowering patterns within an inflorescence influence the reproductive success (i.e. fruit formation) of individual flowers within or among inflorescences of S. stenoloba in a pollinator‐excluded environment. Our results showed that: (1) no pollen grains were observed deposited on the stigma of their own flower after the pollen ejection system was manually activated, indicating self‐pollination within an individual flower is improbable in S. stenoloba; (2) fruit set in the indoor open pollination treatment and the inflorescence‐closed pollination treatment indicated that S. stenoloba is self‐compatible and pollen ejection can potentially result in inter‐floral pollination success; (3) fruit set in the inflorescence‐closed pollination treatment was significantly lower than that of indoor open pollination, indicating within‐ and between‐flower pollination events in an inflorescence are most likely limited, with pollination between inflorescences providing the highest reproductive opportunity; and (4) analysis of the spatial distribution of cumulative fruit set on inflorescences showed that pollen could reach any flower within an inflorescence and there was no functional limitation on seed set among flowers located at various positions within the inflorescence. These data suggest that the pollen ejection mechanism in S. stenoloba can enhance inter‐plant pollination in pollinator‐excluded environments and may suggest adaptation to pollinator scarcity attributable to habitat disturbance or competition for pollinators in a diverse flora. © 2012 The Linnean Society of London, Botanical Journal of the Linnean Society, 2012, 170 , 59–68.     

Gender variation of sequential inflorescences in a monoecious plant Sagittaria trifolia (Alismataceae)

In protogynous plants, female flowers of early blooming plants are at a reproductive disadvantage because they cannot set fruit due to the lack of available pollen. To study this phenomenon, gender expression of the monoecious herb Sagittaria trifolia was investigated over the entire flowering season in two field and two cultivated populations in Hubei and Hunan Provinces, China. In racemes of S. trifolia, flowers open sequentially from bottom to top, with female flowers opening first followed by male flowers. This creates a temporal separation of sexes in the species. Under field conditions small plants are often male, with production of both male and female flowers increasing with plant size. Femaleness increased among sequential inflorescences since female flower production increased whereas male flower production did not. Seed production was greater in large inflorescences because they contain more female flowers, and the number of ovules increased in female flowers at basal positions within the raceme. A consistent pattern of high seed set was observed in flowers from both field and cultivated populations. About 1 % of unfertilized ovules resulted from no pollination and 2 % of the seeds produced were only partly developed due to resource limitation. In the first inflorescence of the six experimental populations, 6.7-40.0 % of individuals produced only male flowers, and female flowers of 1.9-6.5 % individuals were aborted. The occurrence of male flowers in early blooming inflorescences could be an adaptive strategy to conserve resources and enhance pollination of female flowers in protogynous S. trifolia.     

The effects of a bumble bee nectar robber on plant reproductive success and pollinator behavior

Interactions between a plant species (Corydalis caseana), a bumble bee nectar robber (Bombus occidentalis), and a bumble bee pollinator (B. appositus) were studied. There were no significant differences between naturally robbed and unrobbed flowers in fruit set or mean seed set per fruit. Plots of C. caseana plants were subjected to treatments of robbing and no robbing using commercially available colonies of B. occidentalis. Robbers did not pollinate the flowers. Pollinator behavior was observed to determine (1) the number of bees attracted to each plot, (2) the number of inflorescences visited in a plot, (3) the number of flowers visited on each inflorescence, and (4) the distance flown between inflorescences. There were no significant differences in the number of inflorescences visited per bee or the number of flowers visited per inflorescence per bee when robbed and unrobbed treatments were compared. Of the parameters measured, only distance flown between inflorescences differed in the robbed and the unrobbed treatments. Bees flew significantly further between inflorescences in the robbed plots than in the unrobbed plots. The results indicate that the nectar robbers have no negative effect on fruit set or seed set in C. caseana and that they may cause increased pollen flow distances by changing the behavior of the pollinator.     

The effect of protandry on siring success in Chamerion angustifolium (Onagraceae) with different inflorescence sizes

Abstract Protandry, a form of temporal separation of gender within hermaphroditic flowers, may reduce the magnitude of pollen lost to selfing (pollen discounting) and also serve to enhance pollen export and outcross siring success. Because pollen discounting is strongest when selfing occurs between flowers on the same plant, the advantage of protandry may be greatest in plants with large floral displays. We tested this hypothesis with enclosed, artificial populations of Chamerion angustifolium (Onagraceae) by experimentally manipulating protandry (producing uniformly adichogamous or mixed protandrous and adichogamous populations) and inflorescence size (two-, six-, or 10-flowered inflorescences) and measuring pollinator visitation, seed set, female outcrossing rate, and outcross siring success. Bees spent more time foraging on and visited more flowers of larger inflorescences than small. Female outcrossing rates did not vary among inflorescence size treatments. However, seed set per fruit decreased with increasing inflorescence size, likely as a result of increased abortion of selfed embryos, perhaps obscuring the magnitude of geitonogamous selfing. Protandrous plants had a marginally higher female outcrossing rate than adichogamous plants, but similar seed set. More importantly, protandrous plants had, on average, a twofold siring advantage relative to adichogamous plants. However, this siring advantage did not increase linearly with inflorescence size, suggesting that protandry acts to enhance siring success, but not exclusively by reducing between-flower interference.     

Influence of inflorescence size on sexual expression and female reproductive success in a monoecious species

Sex allocation theory forecasts that larger plant size may modify the balance in fitness gain in both genders, leading to uneven optimal male and female allocation. This reasoning can be applied to flowers and inflorescences, because the increase in flower or inflorescence size can differentially benefit different gender functions, and thus favour preferential allocation to specific floral structures. We investigated how inflorescence size influenced sexual expression and female reproductive success in the monoecious Tussilago farfara, by measuring patterns of biomass, and N and P allocation. Inflorescences of T.?farfara showed broad variation in sex expression and, according to expectations, allocation to different sexual structures showed an allometric pattern. Unexpectedly, two studied populations had a contrasting pattern of sex allocation with an increase in inflorescence size. In a shaded site, larger inflorescences were female-biased and had disproportionately more allocation to attraction structures; while in an open site, larger inflorescences were male-biased. Female reproductive success was higher in larger, showier inflorescences. Surprisingly, male flowers positively influenced female reproductive success. These allometric patterns were not easily interpretable as a result of pollen limitation when na?vely assuming an unequivocal relationship between structure and function for the inflorescence structures. In this and other Asteraceae, where inflorescences are the pollination unit, both male and female flowers can play a role in pollinator attraction.     

The effect of inflorescence display size and flower position on pollination success in two deceptive and one rewarding orchid

• Inflorescence display size and flower position on the inflorescence play important roles in plant reproduction, in the formation of fruits and are primarily linked to pollinator behaviour. We used three orchids to determine how visitation rates and choice of pollinator depend on number and position of the flowers along the inflorescence.
• We measured reproductive success in (1) natural conditions, (2) hand-pollination experiments and (3) an experimental design, by modifying composition of inflorescences in populations of two deceptive orchids, Orchis anthropophora and O. italica, and one rewarding orchid, Anacamptis coriophora subsp. fragrans.
• There were no differences in natural fruit production in relation to flower position on the inflorescence (i.e. upper versus lower part), suggesting no preference of pollinators for different parts of the inflorescence. Hand-pollination experiments highlighted low pollen limitation in A. coriophora subsp. fragrans but high limitation in O. italica and O. anthropophora. Reproductive success of deceptive orchids in experimental plots decreased significantly when flowers on the upper half of the inflorescence were removed leading to reduced floral display, while reproductive success of the nectariferous species did not differ significantly.
• Our data highlight that in the examined orchids there is no clear relationship between fruit formation and flower position along inflorescences. Thus we can affirm that, for orchids, the entire inflorescence plays a dominant role in insect attraction but the part of the flower spike does not influence the choice of the insect. This implies that all flowers have the same possibility of receiving visits from pollinators, and therefore each flower has the same opportunity to set fruit.

     

The structure and roles of sterile flowers in Viburnum macrocephalum f. keteleeri (Adoxaceae)

The formation and ecological roles of sterile flowers in flowering plants are interesting issues in floral biology and evolution. Here, we investigated the morphological and anatomical characteristics of both fertile and sterile flowers of Viburnum macrocephalum f. keteleeri, a self-incompatible and insect-pollinated shrub, during different developmental stages of flowers. In addition, pollinator visitation rates and fruit set were determined in intact inflorescences and those with sterile flowers removed. The results indicate that sterile and fertile flowers were developmentally similar during early developmental stages, and that development of the flower types diverged about 15 days before flowering. In addition, pollinator visitation rates, number of pollen grains on stigmas and fruit set were significantly higher in inflorescences with sterile flowers than those without sterile flowers. The results suggest that sterile flowers of this species evolved from fertile flowers under long-term selective pressure, and play a crucial role in enhancing reproductive success through effectively attracting pollinators to the plant and thus enhancing fruit set.     

Pollination limitation,costs of capsule production and the capsule-to-flower ratio in Dendrobium monophyllum F. Muell. (Orchidaceae)

Abstract In perennial plants, life-history theory suggests that natural selection should result in the optimization of fruit-to-flower ratios within the limits imposed by the trade-offs between resource allocation for present reproduction and future growth and reproduction. The tropical orchid Dendrobium monophyllum F. Muell., an epiphyte or lithophyte, offers no nectar rewards, is self-incompatible and has a capsule-to-flower ratio of about 1:14. The influence of pollination limitation and the costs of capsule production on capsule-to-flower ratios were assessed using experimental and field studies in which individual plants were observed for 3 years. Pollinators visited about 80% of flowers, and capsule production was significantly related to inflorescence size and pollinaria removal. About nine pollinator visits occurred per capsule. Pollinator visitation and capsule production did not vary significantly between years. The inflorescence size classes most successful in capsule production were also the most frequent in natural populations. The experimental supplementation of outcross pollen to flowers increased capsule set over controls by 45% within a year, but was limited to about 53%. A capsule-to-flower ratio of 1:2 in experimental plants significantly decreased the subsequent growth and flowering of individuals relative to controls. A capsule-to-flower ratio above 1:10 in naturally pollinated plants decreased flowering in the subsequent year. Thus, it is suggested than an increase in capsule production above 10% would not necessarily correlate with greater reproductive fitness because of the increased cost of capsule production. The capsule-to-flower ratio recorded in this study could be evolutionarily stable because of trade-offs between selection for pollinator attraction and the cost of capsule production. The production of surplus flowers appears to function in pollinator attraction and increases fitness through male function.     

INFLORESCENCE ARCHITECTURE: HOW FLOWER NUMBER,ARRANGEMENT, AND PHENOLOGY AFFECT POLLINATION AND FRUIT-SET

The adaptive significance of different types of inflorescences in flowering plants has been largely ignored. The few published studies investigating adaptive aspects of floral displays suggest that numbers of flowers and their arrangement in space and time determine levels of pollination and fruit-set in natural populations. The frequently conflicting demands placed on inflorescence architecture have led to an evolutionary compromise that maximizes the genetic contribution of an individual plant to the next generation. These conflicting demands include pollinator attraction vs. ovary competition, fruit dispersal vs. fruit predation, and reproductive vs. vegetative resource allocation. In most cases, the inflorescence size most successful in fruit production is also the most frequent in natural populations. In addition to quantity of offspring, inflorescence architecture affects, and in turn is affected by, the quality of offspring that result from selfing vs. outcrossing.     

Patterns of pollinator-generated fruit set inOrchis morio (Orchidaceae)

Reproductive success of a deceptive species,Orchis morio, was investigated in relation to floral display, vertical position of fruits and dispersion pattern. A caging experiment confirmed the necessity of the presence of insect vectors for fruit development. In two Czech populations studied in 1997, insect-pollinated plants had on average 27.3 and 38% fruit set, compared to 96.7% for flowers pollinated by hand. Floral display (number of flowers per spike) was expected to positively affect the probability of pollinator visit. However, fruit set (number of fruits per spike) did not increase linearly with increasing floral display, but rose sigmoidally and approached an asymptotic value. Floral display enhanced reproductive success (fruit/flower ratio) up to a certain number of flowers (11–14), above which the relationship started to decrease and only increased the probability of production of at least one fruit. We also studied the effect of flower position within an inflorescence on the probability of being pollinated and we counted the number of seeds per capsule. A rapid decline in fruit set from the bottom to the top of the inflorescence observed probably reflects the behaviour of pollinators. The number of seeds per one capsule varied from 550 to 12 270. In addition, individuals growing in large clumps or isolated had lower reproductive success than those growing in small clumps. This result implies the existence of an intraspecific competition for pollinators and on the other hand inability of isolated individuals to attract a large number of pollinators.     

FLORAL DISPLAY IN PHYLA INCISA: CONSEQUENCES FOR MALE AND FEMALE REPRODUCTIVE SUCCESS

The inflorescences of Phyla incisa consist of flowers in two phases: younger, nectar-containing flowers that have yellow corolla throats and older, nectar-lacking flowers that have dark purple corolla throats. Observations of pollinator visitation patterns to both natural and manipulated inflorescences were made to determine the role of each flower phase in pollinator attraction. The effect of older-phase flowers on male and female reproductive success was determined by comparing stigmatic pollen loads and estimates of pollen removal from inflorescences having different numbers of these flowers. The pollinators of Phyla selected larger inflorescences more often than expected based upon the size distribution of inflorescences available to them. Both younger- and older-phase flowers contributed to the attraction of pollinators, but the latter were less effective in this function. The presence of older-phase flowers significantly increased the visitation rate to inflorescences and the amount of pollen removed but had little effect on pollen deposition on stigmas. The lack of correspondence between pollen deposition and pollinator-visitation rate was not due to stigma saturation, since stigma loads varied greatly. The data indicate that the deposition of pollen on stigmas in this species is a relatively stochastic process, whereas pollen removal from inflorescences occurs at a much more regular rate. Old-phase flower retention appears to contribute to reproductive success through increased pollen donation when pollinator activity is high and may also increase the probability of seed set when pollinators are rare.     

Flowering phenology, display size, and fruit set in an understory dioecious shrub, Aucuba japonica (Cornaceae)

I investigated the effects of display size and flowering phenology on fruit set in Aucuba japonica, an understory dioecious shrub pollinated by opportunistic insects. Natural variations in display size, flowering phenology, and fruit set were monitored in 1997. A hand-pollination experiment was also conducted to check whether pollen limitation was a factor in fruit set in the field. Increases in floral display size did not affect fruit set; the proportion of flowers that set fruit was almost constant irrespective of the total number of flowers per inflorescence, the total number of inflorescences per plant, and the total number of flowers per plant. The hand-pollination experiment showed that fruit set was not pollen limited despite the low mating probability that resulted from the combination of dioecism and the species' dependence on opportunistic pollinators. This was due, in part, to the fact that female flowers did not have a predetermined period of receptivity, but instead remained receptive until they received pollen. In contrast, flowering phenology did affect fruit set. Fruit set was most abundant when male and female flowering was most abundant. This suggests there was some degree of pollen limitation during the part of the flowering season when male flowers were scarce.     

Floral display and pollination success in Senecio jacobaea (Asteraceae): Interactive effects of head and corymb size

The self-sterile Senecio jacobaea (Asteraceae) presents its rayed heads in large, compound inflorescences (corymbs). I examined the role of head and corymb size in pollinator attraction, and whether the positive effect of intact rays (if any) depends on the size of the corymb. Using female fertility as a measure of pollination success, I assessed the performance of stems representing four experimentally produced character combinations: (1) few heads without rays, (2) many heads without rays, (3) few heads with rays, and (4) many heads with rays. The proportion of flowers setting fruit was higher for intact stems (treatments 2, 4) than for stems on which the majority of the heads had been removed (treatments 1, 3), suggesting selection for maximum inflorescence production. By contrast, experimental removal of all rays had a relatively weak negative effect on fruit set, with few-headed stems (treatment 1) experiencing a greater reduction than stems with many heads (treatment 2). These results suggest that clusters of heads produce a synergistic effect on pollinator attraction, allowing plants to maintain high visitation rates even if there are drastic reductions in the basic attraction units. Hence, the number of heads and the attractiveness of the individual heads interacted in their effect on pollination success. Fruit set per flower differed greatly between sites and was positively correlated with plant density.