Increase of pollinator attraction by means of a visual signal in the sexually deceptive orchid, Ophrys heldreichii (Orchidaceae)

Orchids of the genus Ophrys are pollinated by males of solitary bees and wasps through sexual deception. Flowers mimic the odor of a receptive female and thus attract males that seek to copulate. Visual stimuli have been assumed so far to play only a minor role in male attraction. We investigated the role of the perigon as a potential visual signal in attracting pollinators in the orchid Ophrys heldreichii and its pollinator, the males of the long-horned bee Tetralonia berlandi (Apidae). In contrast to many other Ophrys species, O. heldreichii exhibits a large and bright pinkish perigon that appears visually conspicuous to a human observer. In a dual choice test we presented two flowers from a single plant and counted visitation rates. We then removed the perigon of one flower and retested the relative attractiveness of both flowers. For 292 male visits in ten trials we found a significant decrease of visitation rate for flowers with the perigon removed. In a second experiment we repeated the dual choice test using photos of the flowers. Males also significantly chose the picture of an intact flower over the picture of a modified flower where the perigon was digitally removed. From our data, we conclude that T. berlandi males respond to and are attracted by the bright pink perigon of the orchid in addition to other stimuli. A bright colorful perigon occurs almost only in the Ophrys holoserica-oestrifera group, a large sub-group of the genus. We hypothesize that this kind of visual signal is adaptive particularly in those Ophrys species where the targeted males patrol resourced-based encounter sites and strongly rely on their visual system while searching for their females.     

Notes on the pollination mechanisms ofMoraea inclinata andM. brevistyla (Iridaceae)

Individual flowers ofMoraea inclinata are nectariferous and last about six hours. They appear to be pollinated largely by bees in the familyHalictidae (Lasioglossum spp.,Nomia spp.,Zonalictus) and to a lesser extent by bees in the familyAnthophoridae (Amegilla). The mechanism of bee-pollination inM. inclinata is the Iris type; i.e., each flower consists of three pollination units (an outer tepal, a partly exserted anther, and the opposed style branch which terminates in a pair of petal-like crests). Bees rarely visit more than one pollination unit per flower. Transferral of pollen to the bee is passive and nototribic although all bees collected on the flowers were female and 55% of the bees carried pollen loads with 2–5 pollen taxa in their scopae.Moraea brevistyla flowers are nectariferous but lack scent and last two days. They are visited infrequently by bees and only one femaleLasioglossum spec. carried the pollen ofM. brevistyla. Unlike flowers ofM. inclinata those ofM. brevistyla deposit pollen only on the head and thorax. Bee-mediated autogamy in both species is avoided due to the erratic foraging patterns of the bees and the flexibility of each stigma lobe as the bee backs out of the flower. Approximately 2–4 flowers in the inflorescences of both species (6–8 flowers/infloresence) develop into capsules.     

Floral mimesis inThelymitra nuda (Orchidaceae)

Insect pollination occurred inThelymitra nuda R. Br. on sunny days when the ambient temperature exceeded 20 °C. The flower buds on a raceme opened subsynchronously displaying the brightly-colored, actinomorphic perianth and exposed the contrastingly-colored, scented and ornamented column. InT. nuda the staminodes and the filament of the fertile stamen are fused to one another producing an inflated hood over the anther. This staminodal hood is terminated in two non-ornamental, but brightly-colored, central lobes and two terete lateral lobes bearing approximately 400 white trichomes. Each trichome bears a double chain of 30–40 spherical, rugulose cells. Female bees, in the genusLasioglossum (Halictidae), were observed to land directly on the hood and curled their bodies around the four lobed tip of the staminodal complex. The bees attempted to forage on the lobes as if they were collecting pollen from fertile, poricidal stamens. These bees applied thoracic vibration to the yellowish central lobes and actively scraped the trichome clusters (Pseudopollen) with their forelegs. Bees carried the pollinaria ofT. nuda dorsally on their abdomens. Abdominal contact with the rostellum appeared to occur when the female bee disengaged herself from the staminodal hood. Observations made of bees on co-blooming flowers, and analyses of pollen loads collected by bees suggested that the orchid flower mimics the guild of blue-purple flowers that lack floral nectar but offer pollen in poricidal anthers. The models ofT. nuda are co-bloomingLiliaceae in the generaDichopogon andThysanotus. However, nectarless, buzz-pollinated flowers are also extensively distributed over the orchid’s range. More than 30% of the flowers in theT. nuda population had their pollinaria removed, suggesting a high capacity for cross-pollination in an orchid genus usually considered to be self-pollinated via mechanical autogamy. This study confirmed previous predictions that column modifications represented a trend towards pseudanthery.     

Deceptive pollination of the Lady's Slipper Cypripedium tibeticum (Orchidaceae)

To test whether the nectarless flowers of Cypripedium tibeticum attract pollinators through mimicry like the allied species C. macranthos var. rebunense, pollination biology of C. tibeticum was investigated in western China. Although C. tibeticum was also pollinated by bumble bee queens, i.e. Bombus lepidus , B. lucorum and B. hypnorum , no special, rewarding model plants were found in the habitat. Field experimentation confirmed that the flowers were self-compatible but insects were required to transfer orchid pollen to the stigma. Both Bombus queens and workers were visitors, but queens were much more frequent than workers and only queens were effective pollinators. Floral functional morphology analysis showed that it was large queens rather than small workers that fitted well with the flowers of C. tibeticum. With the faint sweet-fruity scent, the minor floral fragrance compound, ethyl acetate, probably plays a role in attracting bumble bees by food deception. The dark flowers with the inflated, trap-like labellum are hypothesized to mimic the nest site of queens. Therefore, bumble bee queens tend to be attracted by C. tibeticum through nest site mimic combined with food deception. Considering that the co-blooming flowers of C. flavum are pollinated by the Bombus workers, and C. smithii pollinated by a queen, we suggest that using the same bumblebees with different body sizes as the pollinators is the main reproductive isolation between interfertile C. tibeticum and C. flavum, while C. tibeticum and C. smithii tend to hybridize naturally.     

The reproductive biology and effective pollinators of the endangered beardtonguePenstemon penlandii (Scrophulariaceae)

Penland's beardtongue, a rare endemic plant of the Colorado Plateau, displays a mixed breeding system. Plants are partially self-compatible but set more fruits when cross-pollinated than when self-pollinated. Fruit production is significantly increased by pollinators. However, in two years of study there was no indication that fruit set was being limited by inadequate pollinator visitation. Pollinator effectiveness was judged by correlating bee behavior at the flowers with analysis of the pollen carried on bee bodies. The most important pollinators were native megachilid bees, particularly in the genusOsmia. The bees that pollinate Penland's beardtongue are essential to its reproduction and must be preserved along with this rare plant.     

Pollination ofEchinocereus fasciculatus andFerocactus wislizenii

One of the most common types of cactus flower in the southwestern United States is the large, colorful, cup-shaped flower.Echinocereus fasciculatus var.boyce-thompsonii in Arizona is a representative of this class of flowers. Its flowers are visited by three common types of insect visitors: medium-sized bees, small solitary bees, and beetles. All three types of visitors come into contact with the pollen, but only the mediumsized bees regularly touch the stigma in their visitations. The main effective pollinators are therefore the medium-sized bees (Megachile, etc.).Ferocactus wislizenii has a similar floral mechanism and is likewise pollinated mainly by medium-sized bees (Megachile, Lithurge, Diadasia, etc.).Pollination of North American Cacti, 1.     

Bee-pollination inHibbertia fasciculata (Dilleniaceae)

In direct contrast to mostHibbertia spp., the flowers ofH. fasciculata R. Br. ex D. C. bear only a single whorl of stamens and these stamens are arranged separately (not in typical bundles). The short filaments are appressed to the three carpels so that the inflated, porose and introrsive anthers form a centralized cluster obscuring the three ovaries. The three slender styles emerge at right angles from between the filaments. These styles curve upward and the stigmas form the three points of a triangle; each stigma is approximately one millimeter outside the centralized cluster of anthers. The flowers are nectarless and bear a bright yellow corolla. A pungent and unpleasant fragrance appears to be concentrated within the pollenkitt. When native bees attempt to forage for the pollen, within the cluster of anthers, the ventrally deposited loads of pollen, on the bees' abdomens, contact the outer triangle of stigmas. The major pollinators ofH. fasciculata are female bees in the polylectic genera,Lasioglossum (subgenusChilalictus, Halictidae) andLeioproctus (Colletidae). These bees carry an average of more than two pollen taxa when they are caught foraging onH. fasciculata. 78% of the 47 bees, captured onH. fasciculata carried the pollen from at least one sympatric taxon bearing nectariferous flowers (e.g., genera in theMyrtaceae, Compositae, andEpacridaceae). The pollination biology ofH. fasciculata is assessed in relation to the known radiation of bee-pollinated flowers in the genusHibbertia, and within theDilleniaceae s. l.     

Stamen movements in flowers ofOpuntia (Cactaceae) favour oligolectic pollinators

Opuntia brunneogemmia andO. viridirubra occur sympatrically in the Serra do Sudeste, Rio Grande do Sul, Brazil. Their flowers have 450–600 thigmonastic stamens and provide large amounts of pollen and nectar for bees. Bees of 41 species were registered at the flowers ofO. brunneogemmia and 30 at the flowers ofO. viridirubra. Females of three oligolectic species are the only effective pollinators:Ptilothrix fructifera (Anthophoridae),Lithurgus rufiventris (Megachilidae), andCephalocolletes rugata (Colletidae). During their visits inOpuntia-flowers, bees touch the filaments and stimulate the movement of the stamens to the centre of the flower. At the end of this movement, the anthers are densely packed around the style. As a consequence the pollen is presented in an easily accessible upper layer of anthers and various, nearly inaccessible lower layers. The lower layers contain about 80% of the pollen reward. Only females of the three oligolectic pollinators exploit the pollen from the lower layers and reach the nectar furrow. Therefore, through their stamen movements,Opuntia flowers hide most of their pollen from flower visitors but favour effectively pollinating, oligolectic bees.     

Solitary invasive orchid bee outperforms co-occurring native bees to promote fruit set of an invasive Solanum

Our understanding of the effects of introduced invasive pollinators on plants has been exclusively drawn from studies on introduced social bees. One might expect, however, that the impacts of introduced solitary bees, with much lower population densities and fewer foragers, would be small. Yet little is known about the potential effects of naturalized solitary bees on the environment. We took advantage of the recent naturalization of an orchid bee, Euglossa viridissima, in southern Florida to study the effects of this solitary bee on reproduction of Solanum torvum, an invasive shrub. Flowers of S. torvum require specialized buzz pollination. Through timed floral visitor watches and two pollination treatments (control and pollen supplementation) at three forest edge and three open area sites, we found that the fruit set of S. torvum was pollen limited at the open sites where the native bees dominate, but was not pollen limited at the forest sites where the invasive orchid bees dominate. The orchid bee’s pollination efficiency was nearly double that of the native halictid bees, and was also slightly higher than that of the native carpenter bee. Experiments using small and large mesh cages (to deny or allow E. viridissima access, respectively) at one forest site indicated that when the orchid bee was excluded, the flowers set one-quarter as many fruit as when the bee was allowed access. The orchid bee was the most important pollinator of the weed at the forest sites, which could pose additional challenges to the management of this weed in the fragmented, endangered tropical hardwood forests in the region. This specialized invasive mutualism may promote populations of both the orchid bee and this noxious weed. Invasive solitary bees, particularly species that are specialized pollinators, appear to have more importance than has previously been recognized. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Observations of the floral biology ofPrasophyllum odoratum (Orchidaceae,Spiranthoideae)

Prasophyllum odoratum is a vernal, nectariferous, terrestrial orchid that flowers profusely six to eight months following cyclical fires that disrupt sclerophyll woodlands. The morphology of the column and pollinarium is indicative of taxa placed within the subfam.Spiranthoideae. The orientation of the pollinaria to the stigma appears to prevent mechanical self-pollination. Both cross- and self-pollination appear to be effected by insects that forage within the brightly-colored, scented, non-resupinate flowers. Ants and drosophilid flies remove nectar, but do not appear to transport pollinaria between flowers. The primary pollinators are polytrophic flies in the fam.Syrphidae and opportunistic male bees in the genusLeioproctus (Colletidae). Approximately 52% of the flowers on a raceme set seed. The comparatively short floral tube ofP. odoratum reflects the dependence of this species on short-tongued insects to effect successful dispersal of pollinaria.     

The pollination biology ofHibbertia stricta (Dilleniaceae)

The exines of pollen grains ofHibbertia stricta (DC.)R. Br. exF. Muell. (Sect.Pleurandra) wear an oily, yellow pollen coat that stains positively for lipids. The pollen is collected by asocial bees, exclusively. The most common floral foragers are members of the genusLasioglossum (subgenusChilalictus;Halictidae) and they harvest pollen via thoracic vibration. As these bees cling to the inflated anthers their pollen smeared bodies come in contact with either of the two wet, nonpapillate stigmas. The stigmas respond positively to cytochemical tests for the presence of esterase immediately following expansion of the corolla, indicating the effective pollination period. The foraging patterns of the bees are narrowly to broadly polylectic. AsH. stricta flowers are nectarless, it is not surprising that bees bearing mixed pollen loads always carry the pollen of at least one nectariferous, coblooming plant. The pollination biology ofH. stricta is compared with otherHibbertia spp. and with pollen flowers in general.     

Pollination of the lady's slipper Cypripedium henryi Rolfe (Orchidaceae)

The pollination ecology of Cypripedium henryi Rolfe, a slipper orchid endemic to west China, was investigated, and its floral shape, size, colour, and scent were analysed. Examination of the breeding system suggests that the flowers are self-compatible, but need pollen vectors for successful reproduction. The flower is rewardless; over 15 insects belonging to Araneida, Hymenoptera, Diptera, Lepidoptera, and Coleoptera were recorded as flower visitors, but most only alighted or rested on the flower. In the total 32 h of observations over 2 years, female Lasioglossum bees were found to be the most frequent visitors and the only pollinators. They showed a high visitation frequency and, surprisingly, re-visited the same flowers frequently. Cypripedium henryi probably attracts pollinators visiting the flowers through general food deception (odour components, colour, false nectar guides), as well as special structures (slippery labellum, slippery staminode). Although three Lasioglossum species visited the flowers, only L. sauterum Fan et Ebmer was found with pollen. Lasioglossum flavohirtum Ebmer was large and climbed out from the entrance. Morphologically, L. sichuanense Fan et Ebmer could be considered as a potential pollinator, but the collected specimens were found to have no pollen of C. henryi on their bodies. It was speculated that the particular floral scent of C. henryi discouraged the entrance of L. sichuanense bees. Lasioglossum sauterum was matched morphologically to the flower, but not all of the visitations resulted in effective pollinations, as some flowers of C. henryi were frequently re-visited and the pollen mass had been taken away by bees on previous visitations. © 2008 The Linnean Society of London, Botanical Journal of the Linnean Society , 2008, 156 , 491–499.     

Lack of floral nectar reduces self-pollination in a fly-pollinated orchid

One explanation for the widespread absence of floral nectar in many orchids is that it causes pollinators to visit fewer flowers on a plant, and thus reduces self-pollination. This, in turn, could increase fitness by reducing inbreeding depression in progeny and promoting pollen export. The few previous investigations of this hypothesis have all involved bee-pollinated orchids and some have given contradictory results. We studied the effects of adding artificial nectar (sucrose solution) to the spurs of a non-rewarding long-proboscid fly-pollinated orchid, Disa pulchra. Addition of nectar significantly increased the number of flowers probed by flies (2.6-fold), the time spent on a flower (5.4-fold), the number of pollinia removed per inflorescence (4.8-fold) and the proportion of removed pollen involved in self-pollination (3.5-fold). The level of self-pollination increased dramatically with the number of flowers probed by flies. Experimental self-pollination resulted in fruits with only half as many viable seeds as those arising from cross-pollination. Pollinators were more likely to fly long distances (>40 cm) when departing from non-rewarding inflorescences than when departing from rewarding ones. These findings provide support for the idea that floral deception serves to reduce pollinator-mediated self-pollination.     

Effect of nectar supplementation on male and female components of pollination success in the deceptive orchid Dactylorhiza sambucina

Many orchids lack floral nectar rewards and therefore rely on deception to attract pollinators. To determine the effect that a mutation for nectar production would have on overall pollination success of the deceptive orchid Dactylorhiza sambucina, we recorded pollen deposition and removal in flowers of plants that had either been supplemented with an artificial nectar solution or left unmanipulated as controls. Nectar supplementation resulted in significant increases in the proportion of flowers pollinated, regardless of morph colour and the density of plants supplemented in the population. However, nectar supplementation had a significant positive effect on pollinaria removal only for the yellow morph in one experiment in which a low proportion of plants were supplemented. Thus a mutation for nectar production would have a positive effect on overall pollination success in D. sambucina, particularly the female component. The observed patterns are discussed in relation to other factors, such as cross-pollination and the reallocation of nectar resources for other plant functions, which are traditionally considered to shape the rewardless strategies of orchids.     

Reproductive biology of the Andean-disjunct genusRetanilla (Rhamnaceae)

Various aspects of the reproductive biology of the Andean-disjunct genusRetanilla (DC.) Brongn. were studied in populations ofR. trinervia andR. ephedra located in central Chile and ofR. patagonica in southern Argentina. Flowering extends from -austral- winter (early spring inR. patagonica) to early summer. The small, white, entomophilous flowers are incompletely protandrous and last four to five days. A weak pleasant odor along with nectar and pollen attract thirty-seven diurnal insect species, ten of which are probable pollinators: these include the honey bee, eight species of solitary bees pertaining toAnthophoridae (1 sp.),Colletidae (3 spp.) andHalictidae (4 spp.), and a nemestrinid fly. Although flower morphology and individual flower phenology do not fully prevent self-pollination (and geitonogamy can easily take place), the level of autogamy is low. Therefore, some self-incompatibility mechanism seems operative inRetanilla. Fruit set of open-pollinated flowers is extremely low, with a maximal value of c. 3% inR. patagonica. In this species, ripe fruits contain on average 1.37 viable seeds. Predispersal (maternal) reproductive success (percent ovules becoming viable seeds) is 2.3%. For medium to full-sized individuals this corresponds to c. 4600 potential offspring per year. As presently known,Retanilla is a reproductively uniform group, in which the Andean disjunction seems to have exerted no particular impact, which is consistent with the view that pollination generalization exerts some stabilizing influence on floral morphology and other reproductive traits.Retanilla expresses a basic rhamnacean set of traits (including protandry and self-incompatibility) showing also high pollen production and secondary pollen presentation. The latter two traits seem characteristic of theRetanilla-Trevoa clade, suggesting that a trend to increased male effort and, perhaps, dependence upon polleneating has evolved within the tribe Colletieae.     

Pollination by deceit,floral sex ratios and seed set in dioecious Rubus chamaemorus L.

Summary Male and female flowers of the dioecious perennial herb Rubus chamaemorus L. are similar in general appearance. However, female flowers are somewhat smaller, do not produce any pollen, and contain very small amounts of nectar. Syrphids and bumblebees, which are important pollinators of R. chamaemorus, showed a strong preference for male flowers. Male flowers were also less often rejected by flower visitors than were female flowers, and two different groups of syrphid species stayed longer in male than in female flowers. These observations suggest that female flowers of R. chamaemorus attract pollinators by deceit.Hand-pollination experiments indicated that pollen availability limited seed production of R. chamaemorus in female dominated habitats but not in areas with an equal floral sex ratio. We suggest that the relative importance of factors limiting female reproductive success is not constant, but is influenced by the floral sex ratio of the population. This should apply also to other dioecious species that show variable sex ratios on either a local or regional scale.     

The dimorphic heterostyly inAconogonon campanulatum (Polygonaceae)

Heterostyly is clearly confirmed inAconogonon campanulatum. This distylous species is dimorphic for tepals, styles, stigma surface, stamens, pollen grain size, and pollen sexine ornamentation. The floral shape is campanulate and thrum flowers are slightly larger than pin flowers. Small solitary bees were observed as flower visitors and probably effected pollination. The possible evolution of dioecy via heterostyly within the genusAconogonon is discussed.     

Anthecology ofSchrankia nuttallii (Mimosaceae) on the tallgrass prairie

Schrankia nuttalii flowers through late spring on the tallgrass prairie. Although each stem produces an average of 26 capitate inflorescences only 12% of those inflorescences will open each day to disperse and receive polyads. Each inflorescence may live up to 48 hours but anthers abscise by late afternoon on the first day and the filaments change color and lose their scent. The 78–93 florets comprising each inflorescence open synchronously before dawn or during early morning hours. First day inflorescences ofS. nuttallii are herkogamous and fragrant. They are nectarless. Bombyliid flies and male bees are infrequent floral foragers so the major pollinators include female bees representing five families;Anthophoridae, Apidae, Colletidae, Halictidae, andMegachilidae. All foraging insects ignore second day inflorescences although stigmas are still receptive. Although 97% of all bees collected onS. nuttallii carrySchrankia polyads in their scopae or corbiculae 59% also carry the pollen/pollinaria of one or more coblooming angiosperms. At least 98% of all bees carrying mixed pollen loads incorporate the pollen/pollinaria of one or more nectariferous taxa (e.g.Asclepias spp.,Asteraceae, Convolvulaceae, Delphinium spec., etc.). Species of halictid bees are more likely to carry pure loads ofS. nuttallii polyads (70%) than bees of the four remaining families. Due to the nectarless florets and high degree of polylectic foraging bee-pollination inS. nuttallii converges more closely with the pollination systems of some AustralianAcacia spp. than with most other xeric/tropical genera of mimosoids studied in the western hemisphere.     

Mechanisms and evolution of deceptive pollination in orchids

The orchid family is renowned for its enormous diversity of pollination mechanisms and unusually high occurrence of non-rewarding flowers compared to other plant families. The mechanisms of deception in orchids include generalized food deception, food-deceptive floral mimicry, brood-site imitation, shelter imitation, pseudoantagonism, rendezvous attraction and sexual deception. Generalized food deception is the most common mechanism (reported in 38 genera) followed by sexual deception (18 genera). Floral deception in orchids has been intensively studied since Darwin, but the evolution of non-rewarding flowers still presents a major puzzle for evolutionary biology. The two principal hypotheses as to how deception could increase fitness in plants are (i) reallocation of resources associated with reward production to flowering and seed production, and (ii) higher levels of cross-pollination due to pollinators visiting fewer flowers on non-rewarding plants, resulting in more outcrossed progeny and more efficient pollen export. Biologists have also tried to explain why deception is overrepresented in the orchid family. These explanations include: (i) efficient removal and deposition of pollinaria from orchid flowers in a single pollinator visit, thus obviating the need for rewards to entice multiple visits from pollinators; (ii) efficient transport of orchid pollen, thus requiring less reward-induced pollinator constancy; (iii) low-density populations in many orchids, thus limiting the learning of associations of floral phenotypes and rewards by pollinators; (iv) packaging of pollen in pollinaria with limited carry-over from flower to flower, thus increasing the risks of geitonogamous self-pollination when pollinators visit many flowers on rewarding plants. All of these general and orchid-specific hypotheses are difficult to reconcile with the well-established pattern for rewardlessness to result in low pollinator visitation rates and consequently low levels of fruit production. Arguments that deception evolves because rewards are costly are particularly problematic in that small amounts of nectar are unlikely to have a significant effect on the energy budget of orchids, and because reproduction in orchids is often severely pollen-, rather than resource-limited. Several recent experimental studies have shown that deception promotes cross-pollination, but it remains unknown whether actual outcrossing rates are generally higher in deceptive orchids. Our review of the literature shows that there is currently no evidence that deceptive orchids carry higher levels of genetic load (an indirect measure of outcrossing rate) than their rewarding counterparts. Cross-pollination does, however, result in dramatic increases in seed quality in almost all orchids and has the potential to increase pollen export (by reducing pollen discounting). We suggest that floral deception is particularly beneficial, because of its promotion of outcrossing, when pollinators are abundant, but that when pollinators are consistently rare, selection may favour a nectar reward or a shift to autopollination. Given that nectar-rewardlessness is likely to have been the ancestral condition in orchids and yet is evolutionarily labile, more attention will need to be given to explanations as to why deception constitutes an 'evolutionarily stable strategy'.     

Effects of variation in inflorescence size and floral rewards on the visitation rates of traplining pollinators ofAralia hispida

Summary In field experiments withAralia hispida inflorescences, the following variables were manipulated: number of umbels per inflorescence, number of flowers per umbel, and amounts of pollen and nectar per flower. Visitation rates by bumble bees, the principal pollinators, were then observed. In the reward-variation experiments, bees appeared to learn the positions of nectar-rich shoots, and visited them significantly more often than nectar-poor shoots. They did not respond to similar variation in pollen production. The nectar preferences developed slowly after the treatments were imposed, and bees continued to favor sites that had been occupied by nectar-rich shoots even after the treatments were discontinued. Visitation rate was approximately proportional to flower number, making it unlikely that increases in inflorescence size produced a disproportionate gain in male reproductive success (a necessary condition in certain models for the evolution of dioecy). For a fixed number of flowers per inflorescence, bees preferred inflorescences with more umbels. In pairwise choice tests of male-phase and female-phase umbels of various sizes, bees preferred male-phase umbels and larger umbels; the preference for male-phase umbels is stronger in bees that had previously fed on male-phase umbels.